Search is not available for this dataset
article stringlengths 4.36k 149k | summary stringlengths 32 3.35k | section_headings listlengths 1 91 | keywords listlengths 0 141 | year stringclasses 13 values | title stringlengths 20 281 |
|---|---|---|---|---|---|
Rift Valley fever ( RVF ) is a mosquito-borne viral zoonosis of ruminants and humans that causes outbreaks in Africa and the Arabian Peninsula with significant public health and economic consequences . Humans become infected through mosquito bites and contact with infected livestock . The virus is maintained between outbreaks through vertically infected eggs of the primary vectors of Aedes species which emerge following rains with extensive flooding . Infected female mosquitoes initiate transmission among nearby animals , which amplifies virus , thereby infecting more mosquitoes and moving the virus beyond the initial point of emergence . With each successive outbreak , RVF has been found to expand its geographic distribution to new areas , possibly driven by available vectors . The aim of the present study was to determine if RVF virus ( RVFV ) transmission risk in two different ecological zones in Kenya could be assessed by looking at the species composition , abundance and distribution of key primary and secondary vector species and the level of virus activity . Mosquitoes were trapped during short and long rainy seasons in 2014 and 2015 using CO2 baited CDC light traps in two counties which differ in RVF epidemic risk levels ( high risk Tana-River and low risk Isiolo ) , cryo-preserved in liquid nitrogen , transported to the laboratory , and identified to species . Mosquito pools were analyzed for virus infection using cell culture screening and molecular analysis . Over 69 , 000 mosquitoes were sampled and identified as 40 different species belonging to 6 genera ( Aedes , Anopheles , Mansonia , Culex , Aedeomyia , Coquillettidia ) . The presence and abundance of Aedes mcintoshi and Aedes ochraceus , the primary mosquito vectors associated with RVFV transmission in outbreaks , varied significantly between Tana-River and Isiolo . Ae . mcintoshi was abundant in Tana-River and Isiolo but notably , Aedes ochraceus found in relatively high numbers in Tana-River ( n = 1 , 290 ) , was totally absent in all Isiolo sites . Fourteen virus isolates including Sindbis , Bunyamwera , and West Nile fever viruses were isolated mostly from Ae . mcintoshi sampled in Tana-River . RVFV was not detected in any of the mosquitoes . This study presents the geographic distribution and abundance of arbovirus vectors in two Kenyan counties , which may assist with risk assessment for mosquito borne diseases .
Rift Valley fever virus ( RVFV ) , of the genus Phlebovirus , family Bunyaviridae is a mosquito-borne virus present in Africa and the Arabian Peninsula [1] . It causes disease of varying severity including hemorrhagic fever , encephalitis and mortalities in humans and abortions and death among ruminants . RVF outbreaks occur in many parts of Africa every 5 to 15 years during periods of heavy and persistent rainfall that often leads to flooding . Animals are mainly infected through bites of infected mosquitoes , while humans are typically exposed when they come in direct contact with infected bodily fluids or tissues of infected animals . Transmission to humans via mosquito bites is speculated to cause milder disease or asymptomatic infections [2 , 3] . Since being first identified in the 1930s , recurring RVF outbreaks have led to high morbidity and mortality in humans and livestock as well as significant economic loss in affected regions/countries [3] . The outbreaks that affected eastern Africa in 1997/98 and 2006/2007 were most widespread in Kenya , Tanzania , Somalia , Djibouti , Sudan and South Sudan and although the full impact of the outbreaks in terms of public health and economic loss for the entire region may not have been fully assessed , it is documented that Kenya suffered losses to the extent of US $ 32 million due to losses of animal herds , vaccination costs and trade bans/value chain ramifications[4] . During the 2006/2007 outbreak there were more than 150 reported human deaths due to RVFV and over 700 human cases , and there was strain on the already overstretched public health resources and facilities in the North-Eastern regions of Kenya [5] . Mosquitoes collectively referred to as floodwater Aedes have been classified as the primary vectors of RVF , maintaining the virus through transovarially infected drought , resistant eggs that survive in dry soils on low lying depressions on land over inter-epidemic periods that could be as long as 5 to 15 years . However , it is also suspected that inter-epidemic period may last for more than 15 years in some regions [6 , 7] . Flooding due to heavy persistent rainfall results in mass emergence of flood water Aedes mosquitoes . Vertically infected ( infected eggs ) that emerge initiate virus transmission to nearby animals , which could lead to an outbreak depending on continued precipitation and flooding of vector breeding habitats and elevated abundance of vectors [6] . Other mosquitoes in the Culex , ( generally referred to as secondary vectors ) and other potential secondary vectors such as Anopheles and Mansonia genera succeed the primary vector species taking over the flooded grounds to further support virus transmission in the later part of the outbreak period [8] . During investigations of the 2006/2007 RVF outbreak in Kenya , 10 mosquito species principally Ae . mcintoshi , Ae . ochraceus ( primary vectors ) , and a range of other secondary vector species , sampled in ecologically diverse affected regions ( including Garissa and Tana-River ) were found positive for RVFV [9] . Eleven national epizootics of RVF have occurred in Kenya between 1951and 2007; 8 ( 12% ) districts being affected in 1951 , 22 ( 32% ) in 1961–64 ( including Garissa , Tana River and Isiolo ) and 48% ( 33/69 ) in the 2006/2007 outbreak period [10] . Thus the geographic expansion of RVF is increasing with each successive outbreak and , apart from environmental drivers ( rainfall and temperature ) and the density and movement of livestock , the presence of competent vector species is very important for virus transmission to occur and to be established in any new area [6 , 11] . Transmission via infected mosquitoes remains crucial for the dissemination of RVFV between herds or flocks over short and long distances allowing for the emergence and dissemination of the disease throughout a region or a country preceded by the movement of infected animals [11] . The sensitivity and specificity of disease risk assessment and forecasting may be improved by characterizing more small scale and explicit factors that are associated with varying disease occurrences in certain regions within a country . To generate data that would improve assessment of disease risk and regional vulnerability , we investigated the composition and distribution of known vectors of RVFV in two counties , namely Tana-River and Isiolo , known to have different ecologic settings and different levels of disease activity . In 2006/2007 , Tana-River suffered a significantly higher impact RVF with 7 deaths out of 16 reported human cases compared to Isiolo with 0 deaths out of 7 probable cases and in addition , Tana-River and Isiolo have been classified as being at high and medium risk of RVF respectively , based on livestock infection data [5 , 12] . We also investigated presence of circulating arboviruses in the mosquito population .
This study was implemented in the Tana-River and Isiolo counties of Kenya , selected based on the differential impact of the RVF outbreak in 2006/2007 . Tana-River was more affected with16 human cases than Isiolo that only reported 7 probable cases [5 , 12] . Tana-River County; borders Garissa County to the west , covers 38 , 437 km2 and has a coastal strip of 35 km . The county is composed of three sub-counties; Bura , Galole and Garsen and has a population of 240 , 075 , according to the 2009 census distributed in 47 , 414 households . It is inhabited by a mixture of ethnic Orma and Somali communities that practice pastoral farming , with large herds of livestock , consisting mainly of cattle , sheep and goats . Riverine forest , woodland , grassland , bush lands , lakes , open river channels , sand dunes , mangroves and coastal waters are among the diverse ecologies broadly classified under the semi arid and semi humid ecological zones in Tana-River . The county is generally dry and prone to drought . Rainfall is erratic , with rainy seasons falling in March–May and October–December while mean annual rainfall amounts vary between400mm and 750mm . The mean annual temperature ranges between 30°C and 33°C . Tana-River has been classified as being at high risk for RVF outbreaks [12] and it suffered a significantly high impact with 7 deaths out of 16 reported probable cases during the 2006/2007 outbreak period [5] although these figures may be considered an underestimation as some cases may have been missed due to various reasons including poor access to health facilities and challenges of identifying cases . Isiolo county; is an expansive county ( 25 , 336 km2 ) inhabited by diverse ethnic communities . Although the population is predominantly Cushite communities ( Oromo-speaking Boran and Sakuye ) there are Turkana , Samburu , Meru , Somali and other immigrant communities from other parts of the country . Borana form the largest proportion and except for the Meru , the rest of the communities practice pastoralism . Isiolo has three ecological zones; semi-arid , arid and the very arid . The semi-arid zone makes 5% of the county with an annual rainfall of between 400–650mm . The relatively high rainfall here is due to the influence of mount Kenya and Nyambene Hills in the neighbouring Meru County . However , 95% of the county falls in the arid to very arid zone . Isiolo suffered RVF outbreak at a smaller scale than Tana-River and in the most recent RVF risk classification for Kenya by counties , Isiolo was classified as being at medium risk for RVF outbreaks [5 , 12] . In the scarce data available of RVF cases during outbreaks of 2006/2007 , Isiolo documented no deaths out of 7 probable cases [5] although again there may have been significant under- reporting . For this study , sampling in all sites was performed following long and short rains to target periods of possible vector activity and RVF transmission . Mosquitoes were trapped using CO2-baited CDC light traps ( John W . Hock Company-Model 512 ) twice every year at the selected study sites in Tana-River and Isiolo areas ( Fig 1 ) during the long rains ( April–June ) and short rains ( November–December ) between 2014 and 2015 , respectively . There were a total of seven sampling sites in each area cutting through a transect of all the sub-counties and ecological zones ( Fig 1 ) . These sites were selected along the major livestock movement routes used by nomadic herders in both regions and also represented the different ecozones in each of the two major sites . During each trapping period and in each site , ten traps were set at 1800 hrs and retrieved at 0600 hrs the following day for three consecutive sampling days in both seasons . Trapped mosquitoes were anesthetized using triethylamine ( Sigma-Aldrich-471283 ) for ten minutes , separated from other insects , placed into 15 ml labeled tubes , and transported to the laboratory in liquid nitrogen where they were stored at -80°C and subsequently morphologically identified to species level using available taxonomic keys [13–16] . Mosquitoes were grouped in pools of up to 25 mosquitoes belonging to the same species , sex , collection date and trap and stored to be homogenized and analyzed for viruses . Mosquito homogenates were prepared from identified mosquito pools for virus isolation and characterization following previously published standard procedure [17 , 18] . Homogenates were transferred to a 1 . 5 ml cryovial and stored at -80°C ready for testing . Homogenates were screened for viruses by inoculation of 50 μl of each pool into a monolayer of Vero E6 cells ( monkey kidney continuous cell line ) grown in 24 well plates following previously published standard virus isolation procedure [17 , 18] . Samples giving reproducible CPE were processed for molecular analysis to determine the identity of the virus isolate following previously published procedures and using available primer sets that flank conserved regions of African arbovirus species or families [17 , 18] . The PCR cycling conditions varied for each specific virus . The specific reactions were conducted using cycling condition for specific primers for virus genus ( alpha viruses , flaviviruses and orthobunya viruses ) and virus type ( RVFV , Bunyamwera , West Nile , Sindbis , Batai ) . The mosquito species diversity and density data were analyzed using R version 3 . 1 . 1 [19 , 20] . The differences in the proportions of the total captures for mosquito species between the areas ( Isiolo and Tana-River ) were evaluated using generalized linear models ( GLM ) . Quasi-poisson regression was used to test significant difference between the vector groups and individual species .
A total of 4 , 636 mosquito pools representing collections sampled from the different sites and sampling dates were screened for viruses , and 14 virus isolates were obtained from 14 mosquito pools . By conventional RT-PCR the isolates were shown to include seven SINV: five from Ae . mcintoshi ( Ghalamani , Tana-River ) ; one from Cx . pipiens ( Ghalamani , Tana-River ) ; one from a Culex species ( Kone , Tana-River ) ; one WNV isolated from Cx . vansomereni ( Kone , Tana-Riverone Orthobunyavirus isolated from Ae . mcintoshi ( Ghalamani , Tana-River ) . Two isolates from Ae . mcintoshi ( Ghalamani , Tana-River ) , one from Ae . furfurea ( Kone , Tana-River ) , and one from Ae . tricholabis ( Ghalamani , Tana-River ) remained unidentifiable despite attempts using available arbovirus primers . An Orthobunyavirus virus isolate was obtained from Cx . vansomereni from Ngarua , the only viral isolate from all Isiolo mosquito samples . Seven isolates ( five SINV , one WNV , one Orthobunyavirus ) were confirmed by PCR with respective primers ( S1 File ) .
The mosquito species sampled in the study sites revealed differences in species composition and abundance that could influence the differential epidemic impact of RVF in the two counties of Isiolo and Tana-River . The observed difference in the distribution of vectors between the two regions could be attributed to the ecology and habitats of the regions . Entomologic investigations performed during the RVF outbreak 2006/2007 in Kenya , together with previous field and laboratory studies incriminated a number of mosquito species as primary and secondary vectors of RVF through virus detection in wild caught specimens including Ae . mcintoshi , Ae . ochraceus , Ae . dentatus to name a few primary vectors and RVFV was also detected in wild caught Cx . pipiens , and Cx . poicilipes , secondary vectors of RVF virus including potential secondary vectors; Cx . univittatus and Cx . vansomereni , [6 , 9 , 21] . During the 2006/2007 RVFV outbreak the other important primary vector identified was Ae . ochraceus [9] which in the present study was found in significantly lower numbers compared to Ae . mcintoshi in Tana-River . However , the most striking observation was the total absence of Ae . ochraceus in all the Isiolo sites sampled . It is possible that this species , which was abundantly sampled in Garissa and Tana-River during the RVF outbreak in 2006/2007 and found commonly infected with RVF virus is yet to expand its geographic spread to Isiolo and possibly to other counties in Kenya . Indeed , recent population genetic studies conducted on representative samples of Ae . ochraceus sampled from various sites in north-eastern Kenya and West Africa affected by RVF outbreaks revealed that Ae . ochraceus constituted a recently introduced species to Kenya [22] . Our findings further corroborate this observation suggesting that the species is yet to colonise parts of Kenya and possibly the Eastern African region fully . Most RVF epidemics have been linked to Ae . mcintoshi as the key primary vector . Experimental studies have been conducted on Ae . mcintoshi whose significant role in RVF transmission have been suggested through virus isolation from samples collected from the wild . However , it was found to be an inefficient vector exhibiting a major salivary gland barrier with only 14% of the mosquitoes which developed disseminated infection transmitted virus by bite [23] . The importance of this mosquito as key RVF vectors could potentially be attributed to its abundance and feeding patterns in RVF prone regions . Apart from the multiple detections of RVFV in wild caught specimens during outbreaks [9 , 24] , there are no studies done to determine the efficiency of Ae . ochraceus in transmitting RVFV . We can speculate that the explosive outbreaks of RVF experienced in Tana-River and Garissa could be attributed to the possible role of Ae . ochraceus , supporting Ae . mcintoshi , since both species were found to be infected with the virus during the outbreak in 2006/2007[9] . Exploring the distribution and vectorial capacity of Ae . ochraceus in other high and medium risk zones could shed more light on this . The densities of other mosquito species which also transmit RVFV and other arboviruses , including Cx . pipiens and Cx . univittatus , were however , found in significantly greater numbers in Isiolo compared to Tana-River . These are among the Culex species from which the RVFV was isolated during previous outbreaks [9 , 25] . The efficiency of Cx . univittatus in transmitting RVFV has not been explored fully but experimental studies on a member of the complex , Cx . perexiguus has shown this to be an efficient vector of RVFV [26] . Although the species feed more readily on birds , it also feed opportunistically on humans , and where RVF virus transmission has been initiated by the primary vectors , secondary vectors such as Cx . poicilipes and Cx . pipiens , other Culex species including Cx . vansomereni and Cx . univittatus , can potentially play a role in transmitting the virus among humans and even to animals [27] . Thus , RVFV could spread widely in human populations in Isiolo , following initial transmission among livestock by floodwater Aedes , and would require vector control efforts targeting Cx . pipiens species to be put in place in order to break human to human transmission and reduce public health impact . Other potential secondary vectors like the Mansonia and Anopheles species were more predominant in Tana-River than in Isiolo including Ma . uniformis , Ma . africanus and An . squamosus , all of which were found infected with RVFV during the 2006/2007 RVF outbreak in Kenya [9] . This study has demonstrated clear differences in vector species composition and abundance in two counties of diverse ecologies and epidemic risks and we suggest that this approach could be used in determining risk levels for transmission and outbreaks of RVF to augment the other currently used ecological risk factors . Efforts to isolate and detect RVFV circulation among the sampled vectors both in Isiolo and Tana-River sites did not yield any isolate . This is in spite of low level circulation of the RVFV noted through monitoring of livestock migrating through these regions [28] . Previous efforts to isolate RVFV from vectors in the inter-epidemic period have not been successful [29] and we suggested that outside of outbreaks , RVFV circulates in vectors at levels that are below detection in terms of minimum infection rates ( infection rates per 1000 mosquitoes tested ) . During the 2006/2007 outbreak [9] , minimum infection rates in mosquito species sampled ranged between 0 . 8 and 2 . 5 , during which time RVFV was detected in 51 out of 1 , 038 mosquito pools of diverse species in Garissa county , thus during inter-epidemic period , RVFV isolation becomes very unlikely , requiring analysis of huge numbers of mosquitoes to be able to detect a single infected mosquito . However , other arboviruses were isolated from three virus genera , Alphavirus , Flavivirus and Orthobunyavirus . This study provides important information about the distribution of key vectors of RVF in Tana-River and Isiolo counties . Diversity and distribution of the vectors in two study sites could be one of the factors which could contribute differential patterns of RVF occurrence in the two regions . These findings support previous studies on RVFV and risk factors associated with RVF outbreaks in Kenya [2 , 6 , 30] . Most of the viruses isolated were obtained from mosquitoes sampled in Tana-River and mostly from Ae . mcintoshi even though they were not the most abundant species in the collection . It is not known why most virus isolates were found in Ae . mcintoshi species but we speculate that due to their feeding preference , they may be infected while feeding on animals/birds which may serve as reservoirs and which mainly converge in Tana-River probably due to availability of fodder , water and pasture . Tana also serves as a major stopover and roosting site for local and migratory birds using the East African flyway [31] . Only one isolate was obtained from a single mosquito pool in Isiolo compared to Tana-River where 13 virus isolates were obtained . This could be attributed to the ecological differences between the regions as well as the diversity of potential vector species . Previous studies have also reported low numbers of arbovirus cases in Isiolo relative to Tana-River [5 , 12] . In previous arbovirus surveys conducted in neighboring Garissa county , WNV , Semliki Forest virus and Ndumu virus , traditionally known to be mosquito-borne , were detected in ticks taken off cattle and other wild animals [32] suggesting that although these viruses have traditional reservoirs and vectors which have been documented , they may infect other animal species from which mosquitoes such as Ae . mcintoshi ( known to feed preferentially on livestock ) could acquire the virus making the network of arbovirus transmission even more complex and unconventional . Culex species from which WNV was isolated feed on diverse species including birds which are known reservoirs of SINV and WNV .
This study demonstrated a marked difference in species composition , abundance and distribution of primary and secondary mosquito vector sof RVFV in two counties of Kenya , classified as being at high and medium risk for RVF outbreaks , respectively . Some of the vectors are known to be involved in RVFV maintenance and transmission suggesting that presence/absence of key RVFV vector species or combination of species could define disease risk , distribution and expansion . The need for RVFV vector competence evaluation for key species is needed to improve the risk evaluation further . | Rift Valley fever ( RVF ) is a mosquito-borne disease caused by the Rift Valley fever virus ( RVFV ) transmitted by diverse species of mosquitoes broadly classified into primary vectors and secondary vectors . Primary vectors consist of floodwater Aedes ( e . g Ae . mcintoshi , Ae . ochraceus , Ae . sudanensis , Ae . dentatus etc ) , known to maintain the virus in their drought resistant eggs which are deposited on wet soils on low lying depressions on land , remaining viable in dry soil for variable number of years during dry periods . Following heavy persistent rains with flooding , such eggs hatch with a proportion already infected . Emerging infected adult female mosquitoes initiate transmission to nearby animals which serve as amplifiers , infecting more mosquitoes resulting in outbreaks . Another group of mosquito species , the secondary vectors , mainly from the Culex ( Culex pipiens and Culex poicilipes ) , and other potential vectors including , Culex univittatus , Anopheles and Mansonia species may take over such breeding sites , breed in abundance and incidentally propagate RVFV transmission . Outbreaks of RFV occur at varying intensities among livestock in different counties in Kenya , and counties are classified into high , medium and low risk zones . We assessed the species composition , distribution and abundance of primary and secondary vectors in two counties; Isiolo ( medium risk ) and Tana-River ( high risk ) . Striking difference in composition of primary vector species between Isiolo and Tana-River was observed suggesting that vector species composition in different regions could further be applied to assess risk of RVF outbreaks and intensity . We propose further evaluation of vector species surveillance as an additional risk assessment tool for RVFV and other mosquito borne viruses . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion",
"Conclusion"
] | [
"invertebrates",
"livestock",
"medicine",
"and",
"health",
"sciences",
"rift",
"valley",
"fever",
"virus",
"pathology",
"and",
"laboratory",
"medicine",
"viral",
"transmission",
"and",
"infection",
"pathogens",
"geographical",
"locations",
"microbiology",
"animals",
"vi... | 2017 | Distribution and abundance of key vectors of Rift Valley fever and other arboviruses in two ecologically distinct counties in Kenya |
In the presence of oxygen ( O2 ) the model bacterium Escherichia coli is able to conserve energy by aerobic respiration . Two major terminal oxidases are involved in this process - Cyo has a relatively low affinity for O2 but is able to pump protons and hence is energetically efficient; Cyd has a high affinity for O2 but does not pump protons . When E . coli encounters environments with different O2 availabilities , the expression of the genes encoding the alternative terminal oxidases , the cydAB and cyoABCDE operons , are regulated by two O2-responsive transcription factors , ArcA ( an indirect O2 sensor ) and FNR ( a direct O2 sensor ) . It has been suggested that O2-consumption by the terminal oxidases located at the cytoplasmic membrane significantly affects the activities of ArcA and FNR in the bacterial nucleoid . In this study , an agent-based modeling approach has been taken to spatially simulate the uptake and consumption of O2 by E . coli and the consequent modulation of ArcA and FNR activities based on experimental data obtained from highly controlled chemostat cultures . The molecules of O2 , transcription factors and terminal oxidases are treated as individual agents and their behaviors and interactions are imitated in a simulated 3-D E . coli cell . The model implies that there are two barriers that dampen the response of FNR to O2 , i . e . consumption of O2 at the membrane by the terminal oxidases and reaction of O2 with cytoplasmic FNR . Analysis of FNR variants suggested that the monomer-dimer transition is the key step in FNR-mediated repression of gene expression .
The bacterium Escherichia coli is a widely used model organism to study bacterial adaptation to environmental change . As an enteric bacterium , E . coli has to cope with an O2-starved niche in the host and an O2-rich environment when excreted . In order to exploit the energetic benefits that are conferred by aerobic respiration , E . coli has two major terminal oxidases: cytochrome bd-I ( Cyd ) and cytochrome bo′ ( Cyo ) that are encoded by the cydAB and cyoABCDE operons , respectively [1] , [2] . Cyd has a high affinity for O2 and is induced at low O2 concentrations ( micro-aerobic conditions ) , whereas Cyo has a relatively low affinity for O2 and is predominant at high O2 concentrations ( aerobic conditions ) [3] . These two terminal oxidases contribute differentially to energy conservation because Cyo is a proton pump , whereas Cyd is not [1] , [2]; however , the very high affinity of Cyd for O2 allows the bacterium to maintain aerobic respiration at nanomolar concentrations of O2 , thereby maintaining aerobic respiratory activity rather than other , less favorable , metabolic modes [4]–[6] . The transcription factors , ArcA and FNR , regulate cydAB and cyoABCDE expression in response to O2 supply [7] . FNR is an iron-sulfur protein that senses O2 in the cytoplasm [8] , [9] . In the absence of O2 the FNR iron-sulfur cluster is stable and the protein forms dimers that are competent for site-specific DNA-binding and regulation of gene expression [10] . The FNR iron-sulfur cluster reacts with O2 in such a way that the DNA-binding dimeric form of FNR is converted into a non-DNA-binding monomeric species [10] . Under anaerobic conditions , FNR acts as a global regulator in E . coli [11]–[13] , including the cydAB and cyoABCDE operons , which are repressed by FNR when the O2 supply is restricted [7] . Under aerobic conditions , repression of cydAB and cyoABCDE is relieved and Cyd and Cyo proteins are synthesized [3] . In contrast , ArcA responds to O2 availability indirectly via the membrane-bound sensor ArcB . In the absence of O2 ArcB responds to changes in the redox state of the electron transport chain and the presence of fermentation products by autophosphorylating [14]–[16] . Phosphorylated ArcB is then able to transfer phosphate to the cytoplasmic ArcA regulator ( ArcA∼P ) , which then undergoes oligomerization to form a tetra-phosphorylated octomer that is capable of binding at multiple sites in the E . coli genome [17] , [18] , including those in the promoter regions of cydAB and cyoABCDE to enhance synthesis of Cyd and inhibit production of Cyo [7] , [17] . Because the terminal oxidases ( Cyd and Cyo ) consume O2 at the cell membrane , a feedback loop is formed that links the activities of the oxidases to the regulatory activities of ArcA and FNR ( Figure 1 ) . These features of the system - combining direct and indirect O2 sensing with ArcA∼P and FNR repression of cyoABCDE , and ArcA∼P activation and FNR repression of cydAB - result in maximal Cyd production when the O2 supply is limited ( micro-aerobic conditions ) and maximal Cyo content when O2 is abundant ( aerobic conditions ) [3] . Although the cellular locations of the relevant genes ( cydAB and cyoABCDE ) , the regulators ( ArcBA and FNR ) and the oxidases ( Cyd and Cyo ) are likely to be fundamentally important in the regulation of this system , the potential significance of this spatial organization has not been investigated . Therefore , a detailed agent-based model was developed to simulate the interaction between O2 molecules and the electron transport chain components , Cyd and Cyo , and the regulators , FNR and ArcBA , to shed new light on individual events within local spatial regions that could prove to be important in regulating this core component of the E . coli respiratory process .
The dynamics of the system were investigated by running the simulation through two cycles of transitions from 0–217% AU . Figure 3a shows a top view of a 3-D E . coli cell at 0% AU ( steady-state anaerobic conditions ) . Under these conditions , the FNR molecules are present as dimers , all ArcB molecules are phosphorylated and the ArcA is octameric . The DNA binding sites for ArcA ( 120 in the model ) and FNR ( 350 in the model ) in the nucleoid are fully occupied . The number of ArcA sites was chosen from the data reported by Liu and De Wulf [18] . The model must include a mechanism for ArcA∼P to leave regulated promoters . Upon introduction of O2 into anaerobic steady-state chemostat cultures ∼5 min was required to inactivate ArcA-mediated transcription [15] . In the agent-based model presented here , each iteration represents 0 . 2 sec . Therefore , assuming that ArcA∼P leaving the 120 DNA sites is a first order process , then t½ is ∼45 sec , which is equivalent to ∼0 . 3% ArcA∼P leaving the DNA per iteration ( Table 3 ) . The number of FNR binding sites was based on ChIP-seq and ChIP-Chip measurements , which detected ∼220 FNR sites and a genome sequence analysis that predicted ∼450 FNR sites; thus a mid-range value of 350 was chosen [23]–[25] . Interaction with O2 causes FNR to dissociate from the DNA ( Table 3 ) . Under fully aerobic conditions ( 217% AU ) the FNR dimers are disassembled to monomers , and the different forms of ArcA coexist ( Figure 3b ) . The ArcA- and FNR- DNA binding sites in the nucleoid are mostly unoccupied due to the lower concentrations of FNR dimers and ArcA octamers . Examination of the system as it transits from 0% to 217% AU showed that the DNA-bound , transcriptionally active FNR was initially protected from inactivation by consumption of O2 at the cell membrane by the terminal oxidases and by reaction of O2 with the iron-sulfur clusters of FNR dimers in the bacterial cytoplasm - the progress of this simulation is shown in Video S1 . This new insight into the buffering of the FNR response could serve a useful biological purpose by preventing pre-mature switching off of anaerobic genes when the bacteria are exposed to low concentration O2 pulses in the environment . In the various niches occupied by E . coli , the bacterium can experience the full range of O2 concentrations from zero , in the anaerobic regions of a host alimentary tract , to full O2 saturation ( ∼200 µM , equivalent to ∼120 , 000 O2 molecules per cell ) , but fully aerobic metabolism is supported when the O2 supply exceeds 1 , 000 O2 molecules per cell . The profiles of five repetitive simulations for each agent in the model are presented in Figure 4 . From iteration 1 to 5000 and iteration 15000 to 20000 , O2 was supplied at a constant value of ∼6 , 500 molecules per cell such that the total number of O2 molecules entering the cell increased linearly; when the O2 supply was stopped ( 5000 to 15000 and 20000 to 30000 iterations ) no more O2 entered the cell and thus the number of O2 molecules that had entered the cell remained unchanged during these periods ( Figure 4a ) . When O2 became available to the cell ( from iteration 1 ) , the sensor ArcB was de-phosphorylated and started to de-phosphorylate ArcA . Consequently , the number of ArcA octamers bound at their cognate sites in the nucleoid decreased rapidly . The ArcA tetramers and dimers produced during de-phosphorylation of the ArcA octamer were transformed to inactive ( de-phosphorylated ) ArcA dimers , ( Figure 4d–f ) . Under aerobic conditions ( iteration 5000 ) all the ArcA was decomposed to inactive ArcA dimers . When the O2 supply was stopped ( from iteration 5001 ) , the number of inactive ArcA dimers decreased rapidly as shown in Figure 4f , being transformed into phosphorylated ArcA dimers , tetramers and octamers ( Figure 4c–e ) . Due to the phosphorylated ArcA dimers and tetramers combining to form ArcA octamers , their numbers dropped after initially increasing . The rate at which the ArcA octomers accumulated ( ArcA activation ) after O2 withdrawal was slower than the rate of ArcA inactivation ( Figures 4b and c ) . In this implementation of the modeled transition cycle , the numbers of ArcA octamers in the cytoplasm and bound to DNA did not reach that observed in the initial state before the second cycle of O2 supply began , indicating that a longer period is required to return to the fermentation state . The numbers of FNR dimer bound to binding sites and free FNR dimer ( cytoplasmic FNR dimer ) decreased when O2 was supplied to the system ( Figures 4g–h ) , but the rate was slower than that for ArcA inactivation , consistent with O2 consumption at the membrane , which can be sensed by ArcB to initiate inactivation of ArcA , but lowers the signal for inactivation of FNR . When O2 was removed from the system ( from iteration 5001 ) FNR was activated over a similar timeframe to ArcA ( Figures 4b and g ) , which was again consistent with previous observations [15] . As with ArcA , free FNR dimers and FNR monomers did not fully return to their initial states after O2 supply was withdrawn in the model , indicating that further iterations are required to reach steady-state ( Figure 4h–i ) . These results clearly indicate that the model is self-adaptive to the changes in O2 availability , and the reproducible responses prove the reliability and robustness of the model . The ArcBA system simulated in this model is based on a preliminary biological assumption , and the agent-based model presented here should prove a reliable and flexible platform for exploring the key components of the system and testing new experimental findings . In order to validate the model with biological measurements of FNR DNA-binding activity estimated using an FNR-dependent lacZ reporter , the ArcBA system agents were removed from the model by setting their agent numbers to zero . The ArcBA system is an indirect O2 sensor and does not consume O2 , hence the FNR system was not affected by withdrawing ArcBA from the model , but this simplification increased simulation speed . The O2 step length and other model parameters were estimated using the experimental data obtained at 31% AU . Using the estimated O2 step length at 31% AU and defining the step length of O2 molecule , , as 0 at 0% AU , a linear model , , was constructed to predict the step lengths of O2 at other AU levels , where k = 2 . 1 and represents the O2 concentration at different AU levels ( Table 4 ) . The O2 step lengths predicted by this model were used to validate the model at 85% , 115% and 217% AU , and the accuracy of the linear model was shown by the good correlation between the model and experimental data . Profiles of five repetitive simulations in which the simplified model was used to predict the numbers of active FNR dimers in steady-state cultures of bacteria grown at different AU values are presented in Figure 5 . At 31% AU , the model implied that FNR-mediated gene expression is unaffected compared to an anaerobic culture ( 0% AU ) , i . e . the number of FNR binding sites occupied in the nucleoid remained unchanged ( Figures 5a and e ) . Even at 85% AU , ∼80% of the FNR-binding sites remained occupied ( Figures 5b and f ) . It was only when the O2 supply was equivalent to >115% AU that occupation of the FNR-binding sites in the nucleoid decreased ( Figures 5 c , d , g and h ) . These outputs matched the FNR activities calculated from the measurements of an FNR-dependent reporter ( Table 5 ) and thus demonstrate the abilities of the model to simulate the general behavior of FNR dimers in steady-state cultures of E . coli . A second validation approach using two FNR variants that are compromised in their ability to undergo monomer-dimer transitions was adopted . The FNR variant FNR I151A can acquire an iron-sufur cluster in the absence of O2 , but subsequent dimerization is impaired [26] . The FNR D154A variant can also acquire an iron-sulfur cluster under anaerobic conditions , but does not form monomers in the presence of O2 [26] . To mimic the behavior of these two FNR variants the interaction radius for FNR dimer formation was changed in the model . Thus , the interaction distance for wild-type FNR monomers , which was initially set at 6 nm ( r3 , Table 3 ) was increased to 2000 nm for the FNR D154A variant , essentially fixing the protein as a dimer , or decreased to 2 . 5 nm for the FNR I151A variant , making this protein predominantly monomeric under anaerobic conditions . The results of simulations run under aerobic ( 217% aerobiosis ) and anaerobic conditions ( 0% aerobiosis ) suggested that under aerobic conditions wild-type FNR and FNR I151A should be unable to inhibit transcription from an FNR-repressed promoter ( i . e . the output from the reporter system is 100% ) , whereas FNR D154A should retain ∼50% activity ( Table 6 ) . Under anaerobic conditions , wild-type FNR was predicted to exhibit maximum repressive activity ( i . e . 0% reporter output ) , whereas FNR I151A and FNR D154A mediated slightly enhanced repression compared to the simulated aerobic conditions ( Table 6 ) . To test the accuracy of these predictions , the ability of wild-type FNR , FNR I151A and FNR D154A to repress transcription of a synthetic FNR-regulated promoter ( FFgalΔ4 ) under aerobic and anaerobic conditions was tested [27] . The choice of a synthetic FNR-repressed promoter was made to remove complications that might arise due to iron-sulfur cluster incorporation influencing the protein-protein interactions between FNR and RNA polymerase; in the reporter system chosen FNR simply occludes the promoter of the reporter gene and as such DNA-binding by FNR controls promoter activity . The experimental data obtained matched the general response of the FNR variants in the simulation , but not very precisely for FNR D154A , with the experimental data indicating more severe repression by FNR D154A under both aerobic and anaerobic conditions than predicted ( Table 6 ) . This suggested that the interaction radius ( r2 = 5 nm; Table 3 ) , which controls the binding of FNR to its DNA target required adjustment to enhance DNA-binding of the FNR D154A variant . Therefore , the simulations were rerun after adjusting r2 to 7 nm for all the FNR proteins considered here . The results of the simulations for both FNR variants now matched the experimental data well ( Table 6 ) . However , it was essential to ensure that the adjustment to r2 did not significantly influence the model output for wild-type FNR . Therefore , simulations of the behaviour of wild-type FNR at 31 , 85 , 115 and 217% aerobiosis were repeated using the adjusted r2 value of 7 nm . The model output was very similar to those obtained when r2 was at the initial value of 5 nm ( Table 7 ) . These analyses imply that for FNR D154A , which is essentially fixed in a dimeric state , the rate of binding to the target DNA governs transcriptional repression , but for wild-type FNR the upstream monomer-dimer transition is the primary determinant controlling the output from the reporter . The FNR switch has been the subject of several attempts to integrate extensive experimental data into coherent models that account for changes in FNR activity and target gene regulation in response to O2 availability [15] , [28]–[31] . These models have provided estimates of active and inactive FNR in E . coli cells exposed to different O2 concentrations and the dynamic behavior of the FNR switch . The ability of FNR to switch rapidly between active and inactive forms is essential for it to fulfill its physiological role as a global regulator and the models are able to capture this dynamic behavior . Thus , it is thought that the ‘futile’ cycling of FNR between inactive and active forms under aerobic conditions has evolved to facilitate rapid activation of FNR upon withdrawal of O2 and hence the physiological imperative for rapid activation has determined the structure of the FNR regulatory cycle [30] , [31] . However , it is less clear from these approaches how the system avoids undesirable switching between active and inactive states at low O2 availabilities ( micro-aerobic conditions , >0%–<100% AU ) . To achieve rapid FNR response times it has been suggested that minimizing the range of O2 concentrations that constitute a micro-aerobic environment , from the viewpoint of FNR , is advantageous [31] . Unlike previous models of the FNR switch , the agent-based model described here recognizes the importance of geometry and location in biology . This new approach reveals that spatial effects play a role in controlling the inactivation of FNR in low O2 environments . Consumption of O2 by terminal oxidases at the cytoplasmic membrane and reaction of O2 with the iron-sulfur clusters of FNR in the cytoplasm present two barriers to inactivation of FNR bound to DNA in the nucleoid , thereby minimizing exposure of FNR to micro-aerobic conditions by maintaining an essentially anaerobic cytoplasm for AU values up to ∼85% . It is suggested that this buffering of FNR response makes the regulatory system more robust by preventing large amplitude fluctuations in FNR activity when the bacteria are exposed to micro-aerobic conditions or experience environments in which they encounter short pulses of low O2 concentrations . Furthermore , investigation of FNR variants with altered oligomerization properties suggested that the monomer-dimer transition , mediated by iron-sulfur cluster acquisition , is the primary regulatory step in FNR-mediated repression of gene expression . It is expected that the current model will act as a foundation for future investigations , e . g . predicting the effects of adding or removing a class of agent to identify the significant regulatory components of the system .
Knowledge of the rate of O2 supply , , to the E . coli cells was required in order to simulate the response of the regulators of cydAB and cyoABCDE to different O2 availabilities . Therefore , un-inoculated chemostat vessels were used to measure dissolved O2 concentrations , , as a function of the percentage O2 in the input gas , Pi , in the absence of bacteria . This allowed the rate at which O2 dissolves in the culture medium to be calculated from the equation: , yielding = 5 . 898 µmol/L/min . The number of O2 molecules distributed to a single bacterial cell was then calculated from the following equation: ( where , NA is the Avogadro constant ( 6 . 022×1023 ) , Vcell is the volume of E . coli cell ( 0 . 3925 µm3 ) and as a constant for this equation , n ( 3 . 3×10−9 ) includes the unit transformations , min to sec ( 60−1 ) and µmol to mol ( 10−6 ) , and the time unit represented by an iteration ( 0 . 2 sec ) . In the model the individual agents ( Cyd , Cyo , ArcB , ArcA , FNR and O2 ) are able to move and interact within the confines of their respective locations in a 3-D-cylinder representing the E . coli cell . To control the velocity of agents , the maximal distances they can move in 3-D space during one iteration ( step length ) were pre-defined ( Table 4 ) . Thus , a step length is pre-defined in program header file ( . h ) and for each movement , this is multiplied by a randomly generated value within [0 , 1] to obtain a random moving distance , which in turn is directed towards a 3-D direction ( movement vector ) that was also randomly generated within defined spatial regions . An example is shown in Figure 6 to illustrate the movements of an O2 molecule when it enters the cell . Interactions between agents depend upon the biological rules governing their properties and being in close enough proximity to react . The interaction radius of an agent encapsulates the 3-D space within which reactions occur . As the interaction radii cannot be measured , they were first estimated on the basis of known biological properties . For the radii r1…4 , r12 and r13 ( Table 3 ) , arbitrary values were initially set at 31% AU , and the model was then trained to match the experimental result for the number of FNR dimers at 31% AU ( Table 5 ) . The modeled output of FNR dimer number at steady-state was compared with the experimental data , and the difference suggested re-adjustment of interaction radii . The adjusted radii were then tested against the FNR dimer numbers at 85% , 115% and 217% AU ( Table 5 ) during model validation , and the results indicate that the interaction radii values are capable of describing the behavior of the system . The interaction radii of Cyd and Cyo with O2 reflect their relative affinities for O2 ( i . e . Cyd has a high O2 affinity and thus reacts more readily , 7 nm interaction radius , than Cyo , which has a lower affinity for O2 , 3 nm interaction radius ) . As , thus far , no accurate biological data is available for ArcBA system , the radii r5…11 were arbitrarily defined and were refined by training the model to match current biological expectations . The rod-shaped E . coli cell was modeled as a cylinder ( 500 nm×2000 nm ) [32] with the nucleoid represented as a sphere with a diameter of 250 nm at the centre of the cell . The experimentally-based parameters and locations of the agents in their initial state are listed in Table 2 . As the number of ArcB molecules has not been determined experimentally , this value was arbitrarily assigned ( see above ) . The interaction rules for the agents are shown in Table 3 ( additional descriptions of an exemplar agent ( O2 ) and the rules for ArcBA and FNR are provided in , Table S1 and Text S1 ) . These rules , combined with the interaction radii , determine the final status of the system . The scale of the model is such that high performance computers are required to implement it , and the flexible agent-based supercomputing framework , FLAME ( http://www . flame . ac . uk ) acted as the framework to enable the simulation [33] , [34] . For more information on FLAME see Figure S2 and Text S2 . Plasmids encoding the FNR variants were constructed by site-directed mutagenesis ( Quikchange , Agilent ) of pGS196 , which contains a 5 . 65 kb fragment of wild-type fnr ligated into pBR322 [35] . The three isogenic plasmids pGS196 ( FNR ) , pGS2483 ( FNR I151A ) and pGS2405 ( FNR D154A ) were used to transform E . coli JRG4642 ( an fnr lac mutant strain ) containing a pRW50-based reporter plasmid carrying the lac-operon under the control of the FFgalΔ4 promoter [27] . β-Galactosidase assays were carried out as described previously on strains grown in LBK medium at pH 7 . 2 containing 20 mM glucose [36] , [37] . Cultures were grown either aerobically ( 25 ml culture in a 250 ml flask at 250 rpm agitation with 1∶100 inoculation ) or anaerobically ( statically in a fully sealed 17 ml tube with 1∶50 inoculation ) . Cultures ( three biological replicates ) were grown until mid-exponential phase ( OD600 = 0 . 35 ) before assaying for β-galactosidase activity . | The model bacterium Escherichia coli has a modular electron transport chain that allows it to successfully compete in environments with differing oxygen ( O2 ) availabilities . It has two well-characterized terminal oxidases , Cyd and Cyo . Cyd has a very high affinity for O2 , whereas Cyo has a lower affinity , but is energetically more efficient . Expression of the genes encoding Cyd and Cyo is controlled by two O2-responsive regulators , ArcBA and FNR . However , it is not clear how O2 molecules enter the E . coli cell and how the locations of the terminal oxidases and the regulators influence the system . An agent-based model is presented that simulates the interactions of O2 with the regulators and the oxidases in an E . coli cell . The model suggests that O2 consumption by the oxidases at the cytoplasmic membrane and by FNR in the cytoplasm protects FNR bound to DNA in the nucleoid from inactivation and that dimerization of FNR in response to O2 depletion is the key step in FNR-mediated repression . Thus , the focus of the agent-based model on spatial events provides information and new insight , allowing the effects of dysregulation of system components to be explored by facile addition or removal of agents . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Methods"
] | [
"systems",
"biology",
"computer",
"and",
"information",
"sciences",
"network",
"analysis",
"regulatory",
"networks",
"biology",
"and",
"life",
"sciences",
"computational",
"biology"
] | 2014 | Agent-Based Modeling of Oxygen-Responsive Transcription Factors in Escherichia coli |
Electrocorticography ( ECoG ) is becoming more prevalent due to improvements in fabrication and recording technology as well as its ease of implantation compared to intracortical electrophysiology , larger cortical coverage , and potential advantages for use in long term chronic implantation . Given the flexibility in the design of ECoG grids , which is only increasing , it remains an open question what geometry of the electrodes is optimal for an application . Conductive polymer , PEDOT:PSS , coated microelectrodes have an advantage that they can be made very small without losing low impedance . This makes them suitable for evaluating the required granularity of ECoG recording in humans and experimental animals . We used two-dimensional ( 2D ) micro-ECoG grids to record intra-operatively in humans and during acute implantations in mouse with separation distance between neighboring electrodes ( i . e . , pitch ) of 0 . 4 mm and 0 . 2/0 . 25 mm respectively . To assess the spatial properties of the signals , we used the average correlation between electrodes as a function of the pitch . In agreement with prior studies , we find a strong frequency dependence in the spatial scale of correlation . By applying independent component analysis ( ICA ) , we find that the spatial pattern of correlation is largely due to contributions from multiple spatially extended , time-locked sources present at any given time . Our analysis indicates the presence of spatially structured activity down to the sub-millimeter spatial scale in ECoG despite the effects of volume conduction , justifying the use of dense micro-ECoG grids .
Electrical recording from the brain surface , known as electrocorticography ( ECoG ) , is becoming more common due to technological advances that enable recording from large cortical surface area with high temporal resolution and far better spatial resolution than non-invasive EEG [1 , 2] . ECoG has also been used as an alternative to penetrating intracortical recording electrodes in brain-computer-interface ( BCI ) applications [3–9] due to its less invasive nature and long-term stability that are important features for driving drive BCI’s [3] . Electrodes designed for BCI will typically have more closely spaced electrodes to target specific cortical regions compared to clinical ECoG , in which large cortical coverage is important . Recently , high density ECoG grids have become more common , and many questions on the properties , uses , and design of these grids , e . g . , what is the optimal spacing for the electrodes [8–11] , remain to be answered . Recording hardware sets an upper limit on the number of channels that can be simultaneously recorded . This creates a tradeoff in designing ECoG electrode grids between coverage and resolution . Clinical grids are typically on the larger coverage side , with 1 centimeter being a typical pitch between electrodes . BCI and research applications have pushed for more resolution in order to place more electrodes near cortical regions of interest [3 , 6 , 8–10 , 12–17] . A challenge of scaling down the size of ECoG grids is that low impedance electrodes improve signal quality , but electrode impedances increase as the contact area decreases [18] . Combining fabrication techniques that allow for smaller , more closely spaced ECoG contacts with novel materials that can significantly reduce impedance makes very small contact sizes feasible . In the present study , we used electrodes coated with PEDOT:PSS on gold traces embedded in a parylene-C substrate [19 , 20] with contact diameter as small as 20 microns and pitches as low as 200 microns . Hereafter , we will refer to these ECoG electrode grids as micro-ECoG . Previous work has shown that recordings with micro-ECoG electrodes are more similar to intracortical recordings than to the recordings from larger clinical ECoG electrode grids [14] . The primary signal of interest in ECoG recordings is the lower frequency component ( less than ~200–500 Hz ) called the local field potential or LFP . LFP is an uncertain signal in that its precise physiological and spatial origins are poorly understood [21 , 22] . Much of the difficulty both in studying and using LFP is due to its lack of spatial specificity , that is , the potentials are an aggregate of nearby activity—in contrast to single- or multi-unit electrophysiological signals which are indicative of action potentials very near the recording site [23] . The spatial extent of LFP is itself an area of study [24–26] with a dependence on the geometry and activity of the region generating the signal . The spread of the potentials manifests itself in ECoG as similar signals being recorded on different electrodes . This feature of the potentials at different electrodes can be used to interpret the signals [27 , 28] or guide the design of electrode grids to optimally sample the cortical surface . To quantify the similarity between electrodes , previous studies examined the correlation or coherence of EEG , ECoG , and intracortical electrodes as a function of inter-electrode distance by averaging the correlation or coherence across many pairs separated by the same distance [8 , 13–15 , 28–31] . Most of these studies are in human or nonhuman primate , with some early investigations on smaller mammals , reptiles , and invertebrates . In ECoG recordings these studies have shown a consistent nearly monotonic decrease in the correlation as the electrode separation increases that exhibits a roughly exponential shape . Also consistent across the studies is a dependence of the correlation/coherence on the frequency band examined . It is expected that the correlation/coherence would tend to zero ( or bias level ) at large distances , and this can be seen in EEG and clinical ECoG [14 , 30 , 31] . On the other hand , the correlation should approach 1 as the separation approaches 0 . This is the case because the brain is a conductive medium , and for finite sources distant from the electrode in a volume conductor the potential will be the sum of all of the sources with amplitudes attenuated with distance . The distance at which the similarity will effectively approach 1 will depend on both the geometry of the sources and the properties of the medium . An example of this is discussed in [29 , 32] , where ECoG correlation between submillimeter-spaced electrodes is mostly close to 1 while correlation between even more closely spaced intracortical electrode pairs is frequently much lower than 1 . This sub-millimeter regime in ECoG is largely unexplored , and at the smallest distances in most previous studies the correlation or coherence is significantly below 1 , meaning there is still room to explore smaller electrode spacing . On the other hand , for practical purposes recording ECoG at the scale in which the neighboring pairs measure very close to the same signal is not optimal because this would mean a large amount of redundancy between channels . The spacing should be guided by the spatial extent of features of interest . It has been suggested that contacts should be less than 5 mm apart for adequate sampling of gamma band in human ECoG [12] , that for BMI applications subdural electrodes in humans be spaced 1 . 7 mm apart and in rat 0 . 6 mm apart [9] , and that by halving the spacing of electrodes from 3 . 5 to 1 . 68 mm implanted in minipig , more and separate response peaks could reliably be identified [10] . The optimal separation will depend on factors such as species , location , and the nature of the activity of interest , but in general it will be difficult experimentally to recognize the optimal spatial resolution for a specific application until it is exceeded . However , we expect there may be an approximate resolution to use as a rule of thumb for each species . We analyze the similarity of micro-ECoG with inter-electrode spacings down to 0 . 4 mm in human recordings and 0 . 2 mm in mouse . In agreement with past studies , we found that on average the signals were more similar for more closely spaced electrodes . With exceptions , higher frequency components of the signal showed a larger decrease in similarity with distance . We also investigated the nature of the pairwise correlation between electrodes across the electrode grid . For a group of closely spaced electrodes to be correlated to each other there must be parts of their signals that are common between each channel pair , and parts that are independent to each electrode . The relative size , distribution , and properties of these signals determines the correlation between each pair of electrodes across the whole array . There are several analyses that are tailored to finding common signals across multiple channels commonly used on electrophysiological data such as principal component analysis ( PCA ) , factor analysis , or independent component analysis ( ICA ) . We modeled the effect of the properties of the components on the correlation structure , and then used ICA on the data to identify and separate common signals and find how they are distributed across the grid . We found that there are smoothly distributed sources present in the data , and due to the linearity of the ICA decomposition , show that the spatially coherent ICA components account for much of the correlation structure in the data .
Human subjects ( n = 2 ) were implanted with a grid of 56 electrodes with 0 . 4 mm center-to-center distance referenced to another larger ( 3 mm diameter ) ECoG electrode a few centimeters away , and mice ( n = 2 ) were implanted with 32-electrode square grids with 0 . 2 mm or 0 . 25 mm spacing with a subcutaneous reference near the skull . After removing poor channels and potential artifacts , the signals were bandpass filtered into 6 different bands , and separated into non-overlapping 2 . 0 second windows ( 527 windows for subject 1 , 326 for subject 2 , 1 , 486 for mouse 1 , and 893 for mouse 2 ) . The Pearson correlation coefficient ( referred to as simply correlation ) was calculated for each window between all pairs of channels on the ECoG grids for each filter . After averaging correlations across equidistant electrode pairs as in [26] which we will call distance-averaged correlation ( DAC ) , we see that the correlation between pairs of channels decreases on average as the distance between the electrodes increases ( Fig 1 ) . The values of correlation in Fig 1 are averages of correlation calculated in 2 second segments of the data across all segments and channel pairs that share the same spacing . We find similar values to previous studies of the correlation as a function of electrode distance [14 , 15 , 30] . Correlation was analyzed separately for different frequency bands due to the 1/f nature of the signal power , and that the presence of distinct processes present in different bands are common in electrophysiology studies . Also , it is a well-known feature of ECoG that common activity in lower frequencies tends to appear over larger regions than high frequency activity . We find a similar trend in the correlation plots with some exceptions between adjacent frequency bands , and that between the two commonly used bands in electrophysiology studies , beta ( 15–30 Hz ) , and high gamma ( 70–110+ Hz ) , the difference is quite large as expected . The differences between the results in human and mouse is also large ( Fig 1B inset ) . In mouse the correlations fall below 0 . 5 within 1 . 5 mm while in human even the highest frequencies are correlated above 0 . 5 up to around 2 mm . The low frequencies in human are noticeably more correlated across distances of a few millimeters . The values within the 2 . 0 second time windows vary considerably but are concentrated near the mean values ( Fig 1C and 1D ) . Correlation is a measure of to what degree two signals are similar , but an alternative approach to view the similarity is to look for commonality among all the signals simultaneously rather than an aggregate of pair-wise comparisons . There are a few methods which are commonly used in electrophysiology for identifying common signals that are present on multiple channels: principal component analysis ( PCA ) , factor analysis , and independent component analysis ( ICA ) . All are built around the assumption that there exists a set of signals that are present in the data with various amplitudes across all of the channels . ICA and factor analysis were developed to find underlying signals while PCA was not . Factor analysis assumes the components are drawn from a Gaussian distribution ( across time samples , not channels ) , which does not describe the data , especially the sinusoidal signals obtained after bandpass filtering . We used ICA because we expect it to best find the underlying signals , and it is commonly used for this purpose . An important point in using ICA ( as well as PCA and FA ) is that the geometry of the recordings is not an input to the algorithm . The inputs are a set of time series ( in this case ) with no particular ordering , arrangement , or any other information relating the channels to one another . Therefore , an orderly spatial arrangement of the ICA results has been taken as an indication of the efficacy of ICA in separating sources , and is compelling in many cases . To explore the connection between the ICA/PCA decomposition of the data into components and the correlation we start with a model how the spatial extent of the components affects the DAC . Component weights are modeled as two-dimensional Gaussian distributions sampled on a square grid of “electrodes” . The resulting correlation vs . distance curves are well-approximated by Gaussians , and we find a direct correlation between the width of the component Gaussians and the standard deviation parameter of the fit of the correlation vs . distance curve as shown in Fig 2 . The relationship between the two is linear in the limit that there are many components that are sufficiently sampled by the grid of electrodes . The addition of uncorrelated noise to all channels decreases all correlation values by a factor , and the effect of having a reference signal added to every channel is to increase all correlation values . The effect of the noise is more apparent at small distances where even with the Gaussian components , the apparent y-intercept of the DAC drops as noise is added . On the other hand , the reference has the effect of raising the asymptotic value of the DAC for large distances . We also directly connected the components to the DAC through the weight matrices ( mixing matrices in ICA ) . Applying the ICA unmixing matrix ( the inverse of the mixing matrix ) to the data will decorrelate the data , and the unmixing matrix can be arbitrarily rescaled , therefore it is always possible for all of the components to have unit variance . This makes the ICA unmixing matrix a whitening , or sphering , matrix which is straightforwardly connected to the covariance matrix of the data because when multiplied by its transpose it must equal the covariance matrix . This allows a reduced , single-component covariance matrix to be calculated for each component of the mixing matrix separately , and the contribution of each component to the DAC can then be calculated . ICA is applied ( using the EEGlab implementation , see Methods ) to the same filtered 2 second windows as were used in the DAC calculation . The mixing matrices , when plotted in the arrangement of the electrodes , show readily apparent spatial patterns throughout the recordings as shown in Fig 3 . As a way to quantify the spatial patterns in the component weights in the mixing matrices , we fit the weights as they are laid out on the brain to a circular two-dimensional Gaussian function . The goodness-of-fit gives a rough assessment of the smoothness of the mapped weights and their spatial gradients ( see Fig 4 ) . Of the parameters of the fit , the one with we expect to have the most relevance to the correlation is the standard deviation , or width , of the Gaussian . Larger widths would correspond to larger correlated areas , and as a result , a higher correlation at larger distances . This effect can be seen when comparing the median width values for each frequency band independently . As shown In Fig 5 , the median Gaussian width decreases with frequency in agreement with the frequency dependence of the DAC , and that the components tend to be more Gaussian for lower frequencies . At the highest frequencies there is a marked decrease in the goodness of fit that may be due to lower signal-to-noise ratios expected as the 1/f decrease in the signal approaches the noise floor of the hardware . By comparing the contributions of each ICA component to its Gaussian fit we find if and how the DAC curves are influenced by the spatial distribution of the component weights . The spatial distribution of the weights must explain the DAC curves , but to determine whether the Gaussian fits have any significance for the DAC in real data , the contribution of the components is plotted as a function of the R2 values of their fits in Fig 6 . The higher R2 components account for a disproportionately large amount of the drop in the DAC with distance , and therefore the particular shape of the DAC is mostly attributable to the more Gaussian components . The correlation and variance are concentrated in more Gaussian components which shows that the larger , more significant components are generally roughly Gaussian ( Supplement Fig 1 ) . As a control , PCA is substituted for ICA , and because it can also be rescaled into a whitening matrix , all of the analyses can be carried out in the same manner for PCA as for ICA . PCA is not a source separation algorithm like ICA , but in the case where there are sources that account for most of the variation these sources will show up in the first principal components . This can be seen in the R2 histograms ( Supplement S2 Fig ) in where the components are concentrated near 0 , but there is also a smaller number of components very near 1 . 0 that account for much of the variance . These are the first few PCA components which are larger and more Gaussian . These more spatially Gaussian principal components account for much of the drop in correlation , so using PCA for comparison both shows the effectiveness of ICA in finding local sources , and that again , the more Gaussian components are more strongly tied to the drop in the DAC with distance . The location of the reference electrode can have a large impact on the correlation values as shown in Fig 2D [31 , 33–35] . The reference electrode subtracts the same signal from all of the recording channels and this will act to increase the correlation between any two channels if the reference is sufficiently uncorrelated with the signals . In mouse the reference electrode was placed subcutaneously and not on the skull , while in human the reference was multiple contacts within 10 cm of the of the micro-ECoG grid on the cortical surface . The latter are more likely to be active at the frequencies of interest , and even correlated with the unipolar signals measured at the grid . The reference electrode placement should always be taken into account when interpreting correlation or other measures of signal similarity , and that the relatively close reference used in the human subjects is not an ideal placement for studying DAC . Consequently , the DAC curves we obtained in Fig 1 should not be interpreted as the DAC corresponding to unipolar potentials ( potentials measured against a theoretical reference potential of zero ) which would be the ideal for studying spatial correlation across the brain . Using the methods in [36] we attempted to identify the reference signal in the human recordings , but a signal that matched the criteria was not found . Additionally , the reference will ideally be identified by ICA as a component with the same weight on every channel across the grid . In practice this is unlikely , but it may be identified in part and represented by components with relatively flat weights . In fact , this method was used in [37] . In our case it may be correlated with the unipolar surface potentials at the grid and could be mixed in with components of those . A common method in ECoG to remove the true reference is to use the common average reference ( CAR ) , at the expense of introducing a virtual reference which is also unknown . When ICA is applied after CAR , and fits are recalculated the distribution of R2 values is nearly unchanged ( supplemental S4 Fig ) . While the mean Gaussian widths are significantly different after CAR ( two-sample Kolmogorov-Smirnov test p > 0 . 05 ) , they follow the unreferenced values closely but are slightly larger ( 19 +/- 9% ) . This shows that the references that were used did not have a large effect on the components and their spatial properties . The effect of CAR on the correlation is shown in supplemental S3 Fig , and the large difference between the original and CAR correlations can be understood through the effect of the CAR matrix discussed in the Methods . The amount subtracted from each component is uniform across all the channels and is equal to the average weight . Therefore , the shapes of the components are unchanged , but they are shifted such that they mean weight of each component across channels is zero . Reference effects are removed in this way due to their representation as a uniform component across all channels . This shifting of the weights can be seen in the data through the offset term of the Gaussian fit becoming strictly negative after CAR .
The results of our study indicate high degree of variability of the DAC , both within any set of data , as Fig 1 shows , and between datasets due to external factors such as where on the cortex the electrodes are placed . There is large variation across the 2 . 0 second windows as shown in Fig 1 , that may reflect changes in ongoing activity . In fact , it has been shown that there are task-related changes in the DAC [13 , 15 , 38] , however we did not find there to be task- or state-related changes in the DAC in the human recordings . The particular curve of the DAC may change between time windows , recording epochs , subjects , and species , but a robust feature in our recordings and previous studies is the frequency dependence of the spatial correlation . This agrees with past studies that the responses in lower frequency bands are more spread out than in higher bands [13 , 17 , 26 , 32 , 39] and is evident in similar studies that used coherence instead of correlation , which is an inherently frequency dependent similarity metric . The coherence is plotted as a function of frequency , and in for ECoG data almost completely monotonically decreases with frequency [13–15] . The geometry of the ICA component weights offers a possible explanation of this frequency dependence . Two ways by which neighboring electrodes can be correlated are by volume conduction and coactivation of populations close to each electrode which produce distinct , but correlated potentials [40 , 41] . In many cases there is a degree of both which contributes to the correlation , but for large distances where volume conduction is assumed to be negligible the presence of correlation is used as an indication of connectivity [28] . On the other hand , at the submillimeter scale we expect volume conduction may play a larger role . The presence of a single-peaked , radially symmetric , and smooth ICA weights map is consistent with volume conduction of the potentials , and the Gaussian fits show that many of the components fit this description . Coactivating regions could also be described by this shape but are not limited to it; there could be distinct , separated peaks , plateaus , or checkerboard-patterned regions . The large and consistent difference in the DAC between human and mouse can also be explained by either larger coactivated areas of cortex or a larger effect of volume conduction in human cortex . The spread of a signal due to volume conduction in brain tissue would be the same in either species assuming they have similar conductivities , but human cortex has neurons with larger lateral spread of their dendritic and axonal trees and is roughly twice as thick as mouse cortex . The effect of volume conduction on deeper sources will spread the potentials they cause more widely across the cortical surface . Additionally , the size of functionally distinct cortical regions is larger in humans , and we are again left with the ambiguity between the two possible factors: the size of the correlated activity , and its spatial spread as in [26] . In our analysis the choice of ICA as the particular form of whitening and 2 dimensional Gaussians as the function used for fitting are not the only choices that could have been used , but they were chosen for simplicity and applicability to this analysis . As a fitting function , a 2-dimensional Gaussian was chosen for its simplicity and flexibility , and not due to any assumption that the components would take this particular form . The purpose of the fit is to identify smoothly varying component weights . The function is smooth on a small scale , with the only peak being the center , so that neighbor-to-neighbor oscillations in the weights will degrade the fit . It is also able to describe radially symmetric peaked distributions as well as flat linear gradients by being fit to a very wide Gaussian with its center far from the electrode grid . Alternatives more tailored to quantify the smoothness could be used such as taking the second spatial derivative of the component weights and finding smooth gradients and peaks by taking first spatial derivatives . These are harder to implement and interpret , and the high R2 values when using ICA and low ones from PCA show the fitting approach is able to describe the components while not being so flexible that it can be fit to any component weights . Another reason for using the Gaussian fits is that they provide a single parameter that characterizes the size of the regions that contain the components . The mean width of the fits of each frequency band decreases with increasing frequency , and this may be due to the effect that frequency has on the spatial spread of LFP . Additionally , the fits provide another method of removing distant volume conducted activity similar to the ICA approach used in [37] by using both the center location of the Gaussian along with the width to identify components of the signals that are far from the grid location . This kind of ICA-based method as an alternative to standard re-referencing schemes has been proposed in [42] . On the other hand , the DAC curves are not fit to Gaussians for the data despite the modeling results that showed that Gaussian components have Gaussian DACs . We expect that using the same model , but with other peaked , but not necessarily radially symmetric distributions , will still result in monotonically decreasing DAC curves with a different shape . Additionally , the effect of noise and reference will add a predictable modification to the curve but adds additional unknown parameters . These may be estimated but this is confounded by the unknown effect of the actual non-Gaussian shape of the components , and the fit becomes more difficult to implement and interpret . The component mixing matrix weights analysis requires any whitening matrix to separate the components , but ICA was chosen for this purpose . Commonly used whitening transformations are not intended to perform source separation , but ICA can be both a whitening and a blind source separation algorithm . PCA and factor analysis have been used for finding common sources in the data , but the assumptions about the data of ICA are more well suited to finding sources in electrophysiology , hence its popularity . Factor analysis is designed to find similar localized sources but is not easily modified to be a whitening transformation and its assumption of normally distributed components is incompatible with the sinusoidal nature of narrow bandpass filtered signals . There are drawbacks to calculating ICA in separate frequency bands , as any broadband processes or ones that span frequency ranges between or across multiple bands will not be as accurately identified or be recognized as part of the same component . Still , ICA was calculated by frequency band so as to be calculated on the exact same windows and signals as the correlation , and due to the 1/f power spectrum typical of electrophysiology . The first consideration is necessary specifically for linking the correlation and the component mixing matrix through the covariance matrix , while the second is a general problem in applying ICA to LFP . ICA may less accurately separate sources whose power is concentrated in higher frequencies due to the much larger power present in lower frequencies biasing ICA towards identifying sources concentrated in those . In addition , if PCA is applied as a pre-processing step , the components that contain some high frequency sources may even be discarded . ICA was applied only to small time windows in addition to narrow frequency bands . This has similar drawbacks in terms of the effectiveness of ICA because it limits the number of observations which ICA can use to identify source . For the same reason as before , consistency with the segments analyzed for correlation , the 2 second windows are needed . Also , this length of time may be appropriate because the components were found to vary even between adjacent windows . However , there is some consistency in the ICA mixing matrices across time—that is very similar components show up repeatedly , but not consistently . This suggests that the time scale of the duration of stable components may be 2 seconds or less , and perhaps ICA would be better suited to even shorter windows for this data in future work to avoid temporal fluctuations in source strengths which does not fit the assumption in ICA of time-invariant mixing matrices . The curve generated by averaging the correlation over many contacts offers some guidance as to how the signals will be related for a given electrode spacing , but it is more straightforward to choose a spacing when given a measure of the spatial extent of the activity . The two are linked , and as has been shown previously with the correlation , the frequency has a strong effect on the spread of potentials measured at the surface of the brain . Electrodes spaced less than a millimeter apart are more suited to higher frequencies or to smaller animals than humans , but even with very limited cortical coverage volume conduction still allows activity that is not directly under the grid to be recorded .
All human subject research was conducted in accordance with a study protocol that was reviewed and approved by the UC San Diego Health Institutional Review Board ( protocol #121090 ) . All animal work procedures were in accordance with a protocol approved by the Institutional Animal Care and Use Committees of UC San Diego ( protocol S07360 ) . Anesthetics used in mice were isoflurane and alpha-chloralose . Pentobarbital injection was used as the method of euthanasia . The details of the electrodes , their preparation , their implantation , and the recording setup are given in [18–20 , 43] . Subjects who were undergoing awake craniotomy surgeries were chosen for recording . The entire section of hardware up to the amplifiers had to be sterilized due to its proximity to the surgical field . The electronics underwent STERRAD sterilization , and it was important to ensure that the devices would remain intact after autoclave sterilization [43] . The electrode grid was placed over STG , with larger electrodes within a few centimeters of the grid used as reference electrodes . The ground electrode was placed in the scalp . Recording was sampled at 20 kS/s , with a built-in high pass filter at 0 . 1 Hz and low pass filter at 7500 Hz . ICR mice weighing 25–35 g were used in the experiments . The mice were placed on a heating pad and anesthetized with isoflurane . A femoral artery was catheterized for monitoring and injection , and a tracheotomy was performed for ventilation of the mice . After fixing the skull to a holder using dental acrylic , a craniotomy and durotomy were performed over the right whisker barrel and surrounding cortex . A well was formed around the craniotomy using dental acrylic , and the exposure was kept filled with artificial CSF until the electrode array was placed . Prior to recording the mice were administered pancuronium and artificially ventilated , and prior to stimulus trials the mice were switched to alpha-chloralose anesthesia . The exposure was dried prior to the electrode placement , and then covered with 0 . 7% agarose made with artificial CSF . The electrodes arrays used were arranged in square grids with either 0 . 2 or 0 . 25 mm spacing , and either 50 micron or variable diameter contact sizes . The reference electrode was silver-chloride ball placed between muscle tissue exposed for the craniotomy . Whisker flick stimuli were presented every 2 seconds , and recordings included both spontaneous epochs as well as periods with stimulation . All data was recorded with an Intan RHD2000 system , and the recordings were sampled at 20 kS/s with a high pass filter at 0 . 1 Hz and low pass filter at 7500 Hz . Channels that by visual inspection were highly contaminated with noise were assumed to be from damaged electrodes removed from further analysis . Data was then downsampled to 4 kS/s , and 6 FIR bandpass filters were applied , chosen such that they span about 0 . 6 octaves , have no overlap , not include 60 Hz , and roughly correspond to physiological bands ( theta , alpha , beta , gamma , high gamma ) : 6–9 Hz , 10–15 Hz , 20–30 Hz , 35–50 Hz , 70–110 Hz , and 130–200 Hz . Windows of time where the reference signal was more than 35 uV from zero , any one channel was outside +/- 4 mV , or the signal in the highest band was more than 20 times the RMS in that band were marked as potential artifacts and excluded along with 750 ms prior and 1 . 25 s after . Regions that were not removed in this way but were shorter than 6 seconds were also excluded . The data was segmented into continuous 2 s windows . For each window correlation was calculated using Pearson’s correlation coefficient for every possible pair of electrodes on the grid . Each channel pair also has an associated inter-electrode distance , and the correlation vs . distance plots are the result of averaging the pairwise correlations with all equally spaced pairs . For a subject the average correlation is calculated by pooling all correlations across time and averaging the values by distance . PCA and ICA decompose the data matrix , s , into linear combinations of components , z , with the transformation , mixing , or weight matrix , W , s=Wz The components are all uncorrelated with every other component . Therefore , the mixing matrix obtained from either ICA or PCA can be used to whiten the data—which is to linearly transform the data such that the covariance matrix of the transformed data is the identity matrix . PCA is commonly used as for whitening data , and for our purpose ICA can be defined such that it is a whitening transform due to the ambiguity in the scaling of the components . The components can be arbitrarily scaled so long as the weights are scaled inversely such that the original data is unchanged . We make use of this choice to conveniently express the covariance of the data as a function of the weights cov ( s ) =cov ( Wz ) =Wcov ( z ) WT=WIWT=WWT Using this result , the correlation matrix can be computed directly from the mixing matrix . For the model we can start with mixing matrices with each component’s weights being drawn from a two-dimensional Gaussian Wi , j=Aje− ( xj−xi ) 2+ ( yj−yi ) 22σj2 where i is the channel with position ( xi , yi ) , and j is the component with location ( xj , yj ) and standard deviation 𝜎j . For our model we “sample” the components on a 10x15 square grid with 150 components , one per channel , whose center positions are drawn from a uniform random distribution on the 2D space covered by the grid plus one fifth of the standard deviation of the component for which the center is being determined on either side . The amplitudes of each component are drawn from a uniform random distribution between 0 . 5 and 1 . 5 , sorted in descending order and then scaled by e-0 . 1 k , where k is 1 , 2 , 3… corresponding to the first , second , third , etc . amplitude . This is to mimic the trend of decreasing variance for components typically obtained from PCA and ICA . Once the mixing matrix is determined the covariance , Σ , is given as before by Σ=WWT The properties of products of Gaussians allows the covariance elements to be rewritten as Σi , j=∑k=1NWi , kWj , k=∑k=1N e− ( xk−xi ) 2+ ( xk−xj ) 2+ ( yk−yi ) 2+ ( yk−yj ) 22σ2 =∑k=1N e−2 ( xk−12 ( xi+xj ) ) 22σ2e−2 ( yk−12 ( yi+yj ) ) 22σ2e−12 ( xi−xj ) 2+12 ( yi−yj ) 22σ2 Separating the terms that involve the distance between channels i and j gives Σi , j=e−dij22 ( 2σ ) 2∑k=1Ne−2 ( xk−12 ( xi+xj ) ) 22σ2e−2 ( yk−12 ( yi+yj ) ) 22σ2=Fi , je−dij22 ( 2σ ) 2 Each element is the product of a Gaussian function of the distance between channels and Fi , j , a sum over 2D Gaussian functions of the component positions centered at the average location of two electrode positions , which given a fixed set of component locations , is a function of the two electrodes’ positions . For uniformly distributed component positions xk , Fi , j looks like a discrete approximation of the integral over xk of the 2D Gaussian . In the limit of an infinite number of components uniformly distributed across a sufficiently large area it becomes proportional to the integral Fi , j∝∫−∞∞∫−∞∞e−2 ( xk−12 ( xi+xj ) ) 22σ2e−2 ( yk−12 ( yi+yj ) ) 22σ2dxkdyk=F0 and given that this integral is not a function of the center position of the Gaussian , in this limit F is a constant regardless of the positions of channels i and j . In this limit the correlation matrix is exactly a Gaussian function of the distance between the channel pairs with a standard deviation square-root of 2 larger than the standard deviation of the components that generated it: ri , j=F0e−dij22 ( 2σ ) 2 ( F0 ) ( F0 ) =e−dij22 ( 2σ ) 2 For the above form of the correlation to be valid doesn’t require that F is near the limit of infinite components , rather that F is not a function of the electrode pair i and j . For the correlation to take the form above on average is an even weaker condition that F can be a weak function of the electrode pair in relation to the distance term so that the small factors multiplying the distance term will tend to cancel when averaged over equidistant pairs and multiple DAC’s . To include the effect of noise specific to each channel , and uncorrelated from the activity or noise on the other channels , a new component has to be created for each noise element because every component is required to be uncorrelated to all other components . This results in a diagonal matrix of weights which we model as each having the same amplitude , ϵ , across channels Wnoisei , j=ϵδij Where ẟij is the Kronecker delta , not the distance used previously . The effect of a reference electrode which is added to all channels can be modeled as a single component with a constant weight vector across all channels Wref=ρ so that the modified mixing matrix is the original mixing matrix with additional columns for the noise and reference W′= ( WWnoiseWref ) = ( ϵ0ρW0ϵρ⋱⋮ ) The modified covariance matrix is now given by Σi , j′=Σi , j+ϵδij+ ρ and its modified correlation matrix r’ is given by ri , j′=Σi , j+ϵδij+ρ ( Σi , i+ϵ+ρ ) ( Σj , j+ϵ+ρ ) ICA was applied using the runica ( ) function from EEGLab [44] to the same filtered and segmented 2 second windows that were used to compute the DAC . The built-in PCA option was used to apply PCA prior to ICA as a dimensionality reduction technique , and the ‘extended-ICA’ option was used . For human data with 56 channel grids the first 30 components were kept , and for mice with 32 channel grids the first 20 components were kept . In both cases the excluded components accounted for less than 5% of the variance in the data , and usually were close to 1% . Component mixing matrices were fit using least squares fitting function lsqcruvefit ( ) in MATLAB to a two-dimensional circular Gaussian function with 5 parameters Ae ( x−B ) 2+ ( y−C ) 22D2+E with lower and upper bounds such that A be positive , B and C to force the center to lie within 40 grid pitches on either side , D to fit to Gaussians that have standard deviations larger than a single grid pitch . The median width parameter D was chosen instead of the mean due to the distribution of values being skewed towards zero . In order to calculate the 95% confidence interval of the median a bootstrap with 5 , 000 resamples was used . Only widths corresponding to components with R2 over 0 . 7 to exclude components for which a Gaussian is not a good representation and the value of D may not be meaningful . The individual contribution of each component to the overall covariance matrix can be found by re-calculating the covariance using only the desired component . Any entry in the covariance matrix is a sum over weights corresponding to all components . Therefore , we define the reduced covariance corresponding to a single component , c , as just the terms involving that component . The sum of all of the reduced covariance matrices is therefore the full covariance matrix . The corresponding reduced correlation matrix cannot be calculated as usual ri , j=Σi , jΣi , iΣj , j because all of the reduced correlation matrices would be identity matrices . Rather the reduced correlation is calculated using the reduced covariance in the numerator , and the full covariance in the denominator , ri , jc=Σi , jcΣi , iΣj , j so that the sum of all components of the reduced correlation matrices is equal to the full correlation matrix . The reduced correlations can be averaged by distance in the same manner as the full one . The contribution of each component to the variance , correlation , and drop in the correlation can be calculated using the reduced covariance and correlation . The variances of the channels are the diagonal entries of the covariance matrix , and we will define the overall variance ( across channels ) as the trace of the covariance matrix . The variance across channels for a given component is then given by Varc=∑i=1NWi , cWi , c such that the overall variance is the sum over components , and the percentage of the variance explained by each component is the component variance divided by the overall variance . We also want to know the contribution of each component to the DAC . The contribution to the DAC is a similar measure to the contribution to the variance , but the contribution to the drop in the DAC is more relevant for explaining the shape of the correlation vs . distance curve . To calculate these , the DAC curve for each reduced correlation matrix is calculated in the same manner as for correlation matrices . The drop in the DAC due to each component is calculated by subtracting the zero-distance value of the DAC from all the values , and it retains the desired property that the sum over all components is equal to the drop in the full DAC . To summarize the amount each component contributes to the drop in the DAC into a single quantity , the percentage explained is calculated at each distance , and then averaged over all distances . With this , each component can be assigned a percentage of the total variance , total DAC , and total drop in the DAC . The common average reference can be computed with the matrix 1NJ Where N is the number of electrodes and J is the NxN matrix of ones . The product of this matrix with s computes the average signal and is subtracted from the original signal to yield the CAR version of the signal sCAR=s−s¯=s−1NJs=[I−1NJ]s=Cs The covariance of the signals after CAR is then cov ( sCAR ) =Ccov ( s ) CT The effect on the component-based representation is sCAR=Cs=CWz It is important to note that this modification of the weight matrix applies to the weight matrix obtained without the change of reference . When CAR is applied to the data the temporal components identified by a whitening algorithm such as PCA are not necessarily the same . | Electrocorticography ( ECoG ) is a type of electrophysiological monitoring that uses electrodes placed directly on the exposed surface of the brain . ECoG is a promising technique for studying the brain , and EcoG signals can be used to control brain-computer interfaces . Advances have made it possible to record simultaneously with an increasing number of smaller , and more closely spaced electrodes . However , a property of electrical recording from outside the brain is that common signals appear on different electrodes at different locations , and this affects decisions about how to best distribute a limited number of electrodes to maximize the information that can be gathered . Large spacing of electrodes around one centimeter apart on the brain’s surface has proven useful for clinical and research use , but how much benefit there is to recording from more locations in a smaller area remains to be answered . We found that we can explain the commonality between the different locations as the combination of different patterns of brain activity that are present at multiple electrode locations , and that signals recorded from very closely spaced electrodes , around a millimeter or less apart , are able to identify patterns that are at this small scale . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"engineering",
"and",
"technology",
"statistics",
"electronics",
"random",
"variables",
"electrophysiology",
"neuroscience",
"covariance",
"multivariate",
"analysis",
"mathematics",
"brain",
"mapping",
"membrane",
"electrophysiology",
... | 2019 | Correlation Structure in Micro-ECoG Recordings is Described by Spatially Coherent Components |
Recombination , complementation and competition profoundly influence virus evolution and epidemiology . Since viruses are intracellular parasites , the basic parameter determining the potential for such interactions is the multiplicity of cellular infection ( cellular MOI ) , i . e . the number of viral genome units that effectively infect a cell . The cellular MOI values that prevail in host organisms have rarely been investigated , and whether they remain constant or change widely during host invasion is totally unknown . Here , we fill this experimental gap by presenting the first detailed analysis of the dynamics of the cellular MOI during colonization of a host plant by a virus . Our results reveal ample variations between different leaf levels during the course of infection , with values starting close to 2 and increasing up to 13 before decreasing to initial levels in the latest infection stages . By revealing wide dynamic changes throughout a single infection , we here illustrate the existence of complex scenarios where the opportunity for recombination , complementation and competition among viral genomes changes greatly at different infection phases and at different locations within a multi-cellular host .
Intracellular interactions among co-infecting viral genomes play a central role in viral evolution and ecology as they determine three important phenomena: ( i ) competition and selection , ( ii ) re-association with other genetic backgrounds through recombination , and ( iii ) functional complementation of ( or by ) other genomes . The overall intensity of these phenomena depends on the probability of encounter of the countless variants of a viral population within the multitude of individual cells composing the host . The basic parameter determining the potential for such encounters is the multiplicity of cellular infection ( cellular MOI ) , i . e . the number of viral genomes ( number of genome units ) that enter and effectively replicate in individual cells . For example , a cellular MOI above 1 in a given cell corresponds to the co-infection of the same cell by several viral variants , favoring recombination , complementation , and intra-cellular competition; on the contrary , a cellular MOI of 1 will preclude these phenomena . Notably , complementation between viral genomes co-infecting individual cells has been investigated both theoretically and experimentally for the bacteriophage Φ6 , and has been demonstrated to be a predominant evolutionary force which directly depends on the MOI , as defined here [1]–[5] . More generally , complementation ( shared production of viral polymerase , movement proteins , suppressors of host defenses , structural proteins of the virion , etc . ) is undoubtedly frequent in viral populations and is at the basis of collective actions , which largely operate at the intra-cellular level . Empirical investigations on the cellular MOI are extremely scarce . In fact , the values for this parameter that prevail in nature remain elusive , and their putative dynamic changes during colonization of a host by a virus population have never been conclusively investigated . Formal MOI estimates have been established in only four systems: one bacteriophage [6] , [7] , one insect virus [8] , and two plant viruses [9] , [10] . For the bacteriophage and the insect virus , the MOI was considered as a single value calculated at one single time point . For plant viruses , both studies were limited to the initial onset of the host infection . Miyashita and collaborators [10] defined the number of virions infecting individual cells in a local lesion within a leaf immediately following the artificial inoculation of the virus in a single cell . González-Jara and collaborators [9] went a little further by analyzing the MOI both in the artificially inoculated leaf , as well as in the very first leaf where the virus appears through natural systemic movement . These empirical analyses provide important insights into the MOI , but at a very limited spatial and temporal scale during host invasion , thus leaving two remarkable lacunas . First , they cannot inform on whether MOI is constant and homogeneous throughout the entire host and infection process or , on the contrary , subject to ample dynamic changes in time and/or space . Such opposite situations could have totally different implications for viral population genetics ( further discussed later ) . Second , and consequently , the estimated values might not even approximate the average MOI that could be calculated from the entire host across the whole infection process , potentially yielding a totally biased view of the reality . The present study fills these important gaps by describing the first extensive spatio-temporal monitoring of the cellular MOI of a eukaryotic virus , the Cauliflower mosaic virus ( CaMV ) , from the onset of the systemic invasion until senescence of its host plant . CaMV is an aphid-transmitted double-stranded DNA virus which replicates through reverse transcription of a genomic RNA intermediate , and is thus expected to have a high mutation rate [11] , [12] . This virus has been shown to recombine extremely frequently [13] , indirectly indicating an elevated cellular MOI . Our analysis at different time points and at different leaf levels demonstrates the occurrence of important dynamic changes of the MOI throughout the infection cycle , starting close to 2 early in infection , peaking at 13 , and then decreasing to initial levels . Most importantly , we obtained similar MOI values under different experimental conditions of inoculum doses , plant growth , and inoculation methods - including natural inoculation by aphids - suggesting that our results are robust to experimental conditions , and thus faithfully illustrate what actually happens during a natural CaMV infection cycle .
Our aim was to evaluate the intensity with which the variants in a viral population can interact with each other at the cellular level , or , in other words , to assess how frequently these variants co-exist in individual cells . To this end , we estimated the cellular MOI and its putative dynamic changes during the invasion of turnip plants ( Brassica rapa ) by CaMV . Host plants were co-inoculated mechanically with VIT1 and VIT3 , two equi-competitive tagged CaMV variants , previously characterized in [14] , [15] , differing only in a 40-bp non-coding insert that allows their specific identification . In all experiments , the two variants were co-inoculated at the same time and location in order to mimic the situation where a mutant coexists with other genomes from its appearance . The principle of the procedure in all time-course analyses was as follows ( see full details in Materials and Methods ) . Six plants were inoculated in parallel and sampled at different time points , starting from the development of the first symptoms of systemic infection until flowering and senescence . At each sampling date a single mature leaf was sampled from the same leaf level in all plants . In each individual leaf two parameters were measured: ( i ) the ratio of the variants VIT1 and VIT3 , and ( ii ) the proportion of cells infected by both variants . From these data , we derived a maximum likelihood estimate of the average number of viral genomes infecting individual cells ( i . e . , the MOI ) at each leaf level . In fact , assuming that the monitored viral variants infect cells at random , the probability for a given variant to enter a cell directly depends on both its frequency within the corresponding leaf and the total number of viral genomes that enter each cell ( MOI ) . Given the known relative frequency of the variants VIT1/VIT3 within each analyzed leaf , we estimate the average MOI for which the likelihood to lead to the observed proportion of cells co-infected by the two variants is maximum . The full details and formulas for this maximum likelihood framework are given in the Materials and Methods . Preliminary experiments were designed to define the VIT1/VIT3 ratio to be used in the inoculum in order to obtain an intermediate proportion of cells co-infected by both variants ( when all cells contain both variants , it becomes impossible to estimate the MOI ) . The outcome of these preliminary experiments indicated a very high proportion of cells co-infected by both VIT1 and VIT3 , and ample variations of this proportion at different sampling dates . Because variations were also important between repeated plants at each sampling date , these preliminary trials were principally used to adjust and better control our sampling protocol , and are thus fully described in the Materials and Methods , and shown in Figure S1 and Table S1 . In order to remove irrelevant sources of variation as much as possible , we repeated the whole time-course experiment homogenizing parameters during plant growth and leaf sampling ( see Materials and Methods ) . In particular , the exact same leaf levels were collected in all six repeated plants , and all leaves were collected 13 days after their first appearance on the plant ( when the leaves were 13 days old ) . The results from this controlled repetition of the time-course monitoring of CaMV cellular MOI are shown in Figures 1 and 2 ( the full data set is provided in Table S2 ) . The VIT1/VIT3 ratio within infected leaves was close to that in the initial inoculum , and remained nearly constant throughout the experiment ( Figure 1 plain line ) . The slight differences in the VIT1 relative frequency at different time points were not statistically significant ( linear mixed-effects model; P = 0 . 112; F = 2 . 16; dfnum = 4; dfden = 20 ) . Moreover , the slope of the linear regression of VIT1 relative frequency versus time was not significantly different from 0 ( P = 0 . 078; F = 3 . 46; dfnum = 1; dfden = 23 ) , consistently with the equi-competitiveness of VIT1 and VIT3 in our experimental condition ( see Materials and Methods ) . In contrast , the proportion of cells infected by both variants varied significantly between leaf levels ( Figure 1 dotted line; linear mixed-effects model; P = 3 . 1×10−4; F = 8 . 71; dfnum = 4; dfden = 20 ) . In line with the preliminary results presented in Figure S1 , we found that the estimated MOI values ( Figure 2 ) followed a bell-shaped curve with a peak at approximately 13 genomes per cell ( in leaf level 21 ) , and minima of around 2 at the early symptoms appearance ( leaf level 6 ) and during flowering preceding plant senescence ( leaf level 43 ) . Variations between the six repeated plants were lower than in the preliminary experiment mentioned above , and the statistical analysis confirmed both a significant MOI increase from leaf level 6 to 21 ( Tukey HSD test; P = 0 . 027 ) , and a significant decrease from leaf level 21 to 43 ( Tukey HSD test; P = 0 . 048 ) . Because the leaves successively developing on the same plant were all analyzed at the same leaf-age , we conclude that they were infected by CaMV at a significantly different MOI . This conclusion was further confirmed by an alternative statistical approach where the MOI in each leaf-level was estimated within a full maximum likelihood framework ( described in the materials and methods ) which results are presented and discussed in detail in the Supporting Online Information ( Figure S2 ) . Our next goal was to test whether our results were specific to the experimental design , in particular to the mechanical inoculation process , which is commonly used in laboratories but does not correspond to the natural mode of inoculation of CaMV . Thus , we investigated how the MOI estimates varied in different experimental conditions ( Figure 3 , the full dataset is provided as Supporting online Information in Table S3 ) . These experimental conditions included changes ( i ) in the plant growing conditions , ( ii ) in the virus dose inoculated mechanically , and ( iii ) in the mode of inoculation ( including aphid transmission ) . The experimental design was similar to that in the time-course experiment described above except that , for practical reasons discussed later , only two leaf levels were sampled ( leaves 12 and 33 ) . Consequently , we could not investigate the effects of these treatments on the MOI dynamics , but we could nevertheless compare their respective values for these two leaf levels . Figure 3 shows that all conditions yielded values of the same order of magnitude as in the other experiments reported in Figure 2 ( and also in Figure S1 ) . A linear mixed-effects model ( with leaf level and treatment and their interaction as fixed effects , and plant as a random effect ) revealed that treatment did not affect MOI ( P = 0 . 99; F = 0 . 042; dfnum = 3; dfden = 18 ) , while leaf level did ( P = 0 . 016; F = 7 . 01; dfnum = 1; dfden = 18 ) . The interaction of leaf level and treatment was marginally significant ( P = 0 . 0488; F = 3 . 19; dfnum = 3; dfden = 18 ) . The results of the “leaf level” and “leaf level” x “treatment” interaction are driven by the two treatments shown in “red” and “yellow” in Figure 3 ( increased viral dose and different growing conditions respectively ) . They suggest that the MOI dynamics might be shifted to the “left” , i . e . occur faster , under these conditions . While fully testing this possibility would have required more time points in all four treatments , our results strongly suggest that our estimates are robust and most likely representative of MOI values in nature . This important conclusion is particularly supported by the condition where CaMV was inoculated by aphid vectors ( shown in green in Figure 3 ) , which is the only mode of transmission reported for this virus in nature .
We here report the first time-course analysis of the cellular multiplicity of infection of a virus invading a eukaryotic host , from the beginning to the end of the host infection process . Our experimental design , monitoring the MOI at different time points and in different locations within the host , was intended to accommodate the likely heterogeneous structure of viral populations in different organs and at different phases of the infection cycle , as suggested by previous studies both in animals [16] and plants [17] . The genetic markers used in CaMV VIT1 and VIT3 are both neutral [15] and highly stable: they are not deleted from the viral genome after at least three successive passages in host plants [14] . These properties enabled the monitoring of the MOI for a very long period ( over 80 days ) , without biases due either to the competitive exclusion of one variant by the other , or to increasing frequency of marker-deleted genomes within the population . The flipside of the use of these markers is that the search for cells co-infected by VIT1 and VIT3 is extremely tedious and time-consuming [18] , and this is precisely why we have limited the study of the robustness of our results to the experimental conditions to only two leaf levels ( Figure 3 ) . The rationale for choosing VIT1 and VIT3 markers rather than seemingly more amenable markers ( such as fluorescent protein genes ) allowing high throughput detection in single cells is fully explained in the Materials and Methods . We simply wish to mention here that VIT1 and VIT3 markers can be detected within infected cells for unlimited amounts of time . Upon replication , the CaMV forms characteristic and very stable electron-dense inclusion bodies ( “viral factories” ) , where hundreds or thousands of mature viral particles accumulate and remain sequestered indefinitely [19] . In consequence , once a CaMV variant has entered a cell and replicated , it likely remains detectable by our nested-PCR procedure until cell death . In preliminary experiments where plants were co-inoculated with both VIT1 and VIT3 at a 1∶1 ratio , we rapidly observed nearly 100% of the cells infected by both variants . This observation is extremely interesting because it indicates that the CaMV variants are not spatially segregated in contrast to most RNA plant viruses [10] , [20]–[22] . Together with the equi-competitiveness of VIT1 and VIT3 , this observation is consistent with the assumption that CaMV variants infect cells at random within a leaf . Thus , assuming that the number of genomes of a given viral variant entering a cell follows a Poisson distribution , which is at the basis of most statistical methods estimating the MOI [7] , [8] , [10] , appears appropriate in the case of CaMV . The cellular MOI in a given host/virus association depends a priori on two parameters: the number of viral infectious units available per cell ( viral load ) , and the maximal number of these units that can effectively co-infect the same cell . Variations in these parameters should influence cellular MOI values and explain the dynamics observed in CaMV-infected plants . It is reasonable to imagine that the viral load increases over time in the plant , with a concomitant increase in the multiple infections of cells , as more and more infected leaves develop and shed virus into the phloem . However , the decline in cellular MOI late in infection contradicts this prediction . Since VIT1 and VIT3 are equi-competitive [15] , this decline cannot be explained by the dominance of one variant over the other , as confirmed by the unchanged VIT1/VIT3 ratio throughout the experiment ( Figure 1 and S1A ) . One possibility would be a host developmental or physiological effect on the MOI , related to the previously described impairment of virus infection upon flowering [23] , or to the onset of a plant defense mechanism [24] , [25] , with a resulting drop in viral load . Another explanation of the observed MOI pattern would be a changing balance between benefits and costs of multiple infection of cells . The benefits are basically those derived from recombination [26]–[28] , and from cooperation among the genomes co-existing within the same cell ( i . e . collective action and mutualistic complementation ) . The costs of multiple infection arise from the competition for cell resources , and from the evolution of “cheater” genotypes , better adapted to this competition than to host exploitation in single infections [1] . The best studied example of the latter phenomenon is the recurrent observation of defective interfering particles ( DIPs ) appearing in virus populations [29]–[32] . The CaMV recombines at very high rates in turnip [13] , and cooperative behaviours in this virus exist at least during the transmission process [33]–[35] and the suppression of gene silencing [36] , [37] . One could thus hypothesize that an increasing cellular MOI could benefit CaMV during the invasion of a host , up to a value ( around 13 ) where the costs would overwhelm the benefits . For example , as indicated above , a high MOI value might increase the proportion of DIPs [7] up to a threshold were functional genomes can no longer sustain the growth of the viral population . The resulting crash of the virus load could therefore explain the MOI drop late in infection . A quantitative monitoring of the virus load within the vasculature of the plants , and an estimate of the frequency of DIPs therein , would support or disqualify these hypotheses . The MOI values and their dynamic changes reported here cannot be directly compared with the situations in other host/virus associations , because no equivalent information is available . The MOI estimate around 4 for a baculovirus infecting lepidopteran insects possibly represents an average over the complete infection process [8] . For the sake of comparison we calculated the equivalent average MOI for CaMV by compiling the full data sets from time-course experiments shown in Figures S1 and 2 , and found values of the same order of magnitude , 7 . 87±2 . 03 and 6 . 67±1 . 43 ( mean±SE ) respectively . Whether the value of 4 found in baculovirus-infected caterpillars resulted from a constant MOI throughout the infection cycle or represented the average of ample variations , as is the case here for CaMV , is not known . In two recent studies on plant viruses , the MOI was investigated in the artificially inoculated leaf . Very early after inoculation , the values found for the Tobacco mosaic virus ( TMV ) infecting Nicotiana benthamiana plants [9] , and for the Soil-born wheat mosaic virus ( SBMV ) infecting Chenopodium quinoa plants [10] were remarkably similar ( between 5 and 6 ) . We did not analyze the inoculated leaf in our study on CaMV , because the mechanical inoculation procedure does not reflect any natural process , and how this might or might not bias the viral infection of neighboring cells is hard to evaluate . The study on TMV [9] also reported the analysis of MOI values in the first systemically infected leaf , where the virus enters via its natural route ( the plant vascular system ) . In this leaf , the MOI of TMV was estimated to lie between 1 and 4 , very close to our estimate for CaMV in leaf level 6 ( mean = 2 . 73; SE = 1 . 73 ) which also represents the first systemically infected leaf level . Interestingly , the same authors assessed a putative time variation of the TMV MOI within this single leaf ( a question not tested here on CaMV ) , and they concluded that the TMV MOI can change through time . However , this conclusion was challenged in the discussion by Miyashita and Kishino [10] , thus leaving opened the basic question of a MOI change with time . On this important question , we here definitely demonstrate that dynamic changes of the MOI indeed occur with large amplitudes during the whole host infection by CaMV . Unfortunately , this remarkable phenomenon cannot be compared to the situation with TMV and SBMV , where the viral infection was not monitored in upper leaf-levels being systemically infected . A dynamic MOI similar to that described here for CaMV likely occurs in other systems , as suggested in HIV by the number of proviruses per cell indicating an elevated MOI [38] , and by the fluctuating rates of cell co-infection in cell cultures [39] . However , alternative scenarios are also possible since segregation and isolation of genetic variants in different cells of the same host has been repeatedly observed for several plant viruses [17] , [20]–[22] , [40]–[42] , suggesting more stringent limits to cellular co-infection , and thus to MOI values , at least within some specific cell types , organs , or tissues . At present , no theoretical predictions are available to fuel a discussion on the potentially different impact that a steady or a variable cellular MOI could have on the evolution of the corresponding viral populations . The few theoretical and experimental studies addressing specifically the role of MOI in the evolution of the phage Φ6 were considering low , intermediate , or high values , but always constant in a given viral line ( reviewed in [5] ) . While we here observe ample MOI variations during host infection by CaMV , we cannot control it , and a comparison with a constant MOI is thus far impossible in this system . In contrast , other virus-host models , like phage systems , would allow the experimental evolution of lines with constant or changing MOI , with various different patterns but similar average value , and the outcome on the evolution of the average fitness in each line would be extremely interesting . Beyond the within-host scale of virus evolution , a specific pattern of variable cellular MOI might have important implications also at a higher organization level , in a broader ecological context . For instance , in the specific case of CaMV , it is possible that populations evolve under different cellular MOI values depending on the vector species . This virus can indeed be transmitted by several aphid species [43] with different behaviors: colonizing the plant or not , feeding from lower or upper leaves , or from younger or older plants . Given the implications of the MOI for viral evolution and epidemiology , our results urgently call for a broader investigation of this important trait in a wide panel of natural virus/host associations , characterizing the values , their putative dynamic changes and the underlying mechanisms .
The two engineered CaMV variants , VIT1 and VIT3 , have been previously characterized in detail [14] . Both are infectious full-length clones of the CaMV Cabb-S isolate [44] harboring a 40-bp DNA insert used as a specific genetic marker that can be quantified in a mixed population [14] and specifically detected within single cells [18] . Such markers were demonstrated to be stably maintained within CaMV genomes over at least three successive passages in turnip host plants [14] . Co-infecting CaMV-VIT1 and -VIT3 proved equi-competitive during turnip plant invasion [15] . The virus particles used in the inoculum were purified from plants infected with each variant individually and quantified as previously described [45] . The inoculum was prepared by mixing purified virus particles and a convenient ratio of 4/1 ( VIT1/VIT3 ) was determined in preliminary experiments ( see below ) . For all time-course analyses of the MOI six healthy plantlets were mechanically inoculated in parallel with 400 ng of virus particles per plantlet as previously described [18] , except for conditions with a different viral dose or inoculation by aphids . When symptoms appeared on systemically infected leaves they were harvested and processed as described below . Unless otherwise indicated turnip plants ( Brassica rapa cv . “Just Right” ) were maintained in an insect-proof growth chamber under controlled conditions ( 24/15°C day/night with a photoperiod of 15/9 h day/night ) . The actual VIT1/VIT3 ratio in each sampled leaf was estimated from a pool of ∼3000 protoplasts per leaf , using real-time quantitative PCR ( PCR conditions and primer sequences are provided in Table S4 ) . A linear mixed-effects model , taking into account the repeated measures within each plant , was used to test for changes in VIT1 frequency between dates ( fixed effect ) within plants ( random effect ) ; it showed that VIT1 frequency was close to that in the mixed inoculum and varied only slightly ( if at all ) over time ( Figure 1 and Supporting online Information Figure S1 and Table S1 ) , confirming previous estimates of marker neutrality [15] . Thirty protoplasts from each sampled leaf were analyzed individually to determine the co-occurrence of VIT1 and VIT3 genomes and thus the frequency of cell infected by both variants . The region of the CaMV genome bearing the genetic markers was amplified from each isolated cell by single-cell nested-PCR , and VIT1 and VIT3 sequences were specifically identified in the amplicons by high resolution melting analysis exactly as described previously [18] . A linear mixed-effects model , taking into account the repeated measures within each plant , was used to test if the proportion of cells infected by both variants varied between leaf-levels ( fixed effect ) within plants ( random effect ) . Despite the tediousness of the single-cell detection of such markers [18] , we have altogether analyzed over 3400 individual cells ( Table S1 , S2 and S3 ) . The use of another type of markers , based on the insertion of genes encoding fluorescent proteins such as GFP ( green ) and RFP ( red ) into viral genomes , would have provided a straightforward high-throughput approach to visualize their presence within single cells , using for example epifluorescence microscopy ( on tissues or extracted protoplasts ) . However , in contrast to the VIT1 and VIT3 markers used here , such fluorescent markers have a number of drawbacks which limits their usefulness for studies such as that presented in this paper: ( i ) currently available fluorescent protein genes cannot be introduced in CaMV and in other viruses with an icosahedral shell , because of the limited size of the encapsidated genome [46] , [47]; ( ii ) GFP can diffuse autonomously from cell to cell in plants [48] , a phenomenon potentially misleading in identifying cells infected with a GFP-expressing virus; ( iii ) two Tobacco mosaic virus variants , respectively expressing GFP and RFP , proved differentially competitive in co-infected plants [9] , and we observed a similar phenomenon with Turnip mosaic virus ( unpublished results ) ; ( iv ) , these GFP or RFP markers are often rapidly deleted from the genomes of plant viruses [49] , [50] , a phenomenon incompatible with their monitoring throughout the infection process . The MOI was inferred with a maximum likelihood procedure from ( i ) the relative proportion of the two variants measured in each sampled leaf , and ( ii ) in the same leaf , the number of cells infected by both variants among the infected cells . Assuming that cell infections occur in a random and independent manner for both variants , the number of genomes of a given variant entering a cell follows a Poisson distribution with a parameter equal to the product between the cellular MOI ( λ ) and the relative frequency of this variant in the sampled leaf ( pi , for VIT1 ) . The null class of each Poisson distribution corresponds to the probability of not being infected by the corresponding variant . Thus , in the ith sampled leaf , the probability for a given infected cell to be co-infected by the two variants is , and , among the Ni infected cells observed within this leaf , the number of co-infected cells has a binomial distribution with parameters Ni and pc , i . The corresponding likelihood function is: , where ki is the observed number of cells infected by both variants within the ith sample . The MOI within each sample is then easily derived as the maximum likelihood estimate of λ . A linear mixed-effects model , taking into account the repeated measures within each plant , was used to test if the MOI varied between treatments and between dates ( fixed effects ) , within each plant ( random effect ) . The significance of MOI differences between specific levels of the factors was investigated using Tukey's HSD ( honest significant difference ) method . The above-described statistical approach was confronted to an alternative analysis , which consisted in working within a full maximum likelihood framework providing one MOI estimate at each date from all 6 replicates . This full maximum likelihood framework is derived from the likelihood function , with profile-likelihood confidence intervals . The MOI parameter ( λ ) was first held constant across all plants and leaf levels , and we used likelihood ratio tests to test whether allowing variation in λ across leaf levels ( dates ) significantly improved the likelihood of the model . We also similarly tested whether we had a plant effect , though we were much less interested by this factor which should be modeled as a random effect ( as indicated above ) . The outcome of both analyses are shown and discussed in the Supporting Online Information ( Figure S2 ) . All statistical procedures were implemented in the statistical software R [51] . As a first exploratory experiment , the plants were inoculated with a VIT1/VIT3 mixture at a 1/1 ratio and sampled twice , at early and later stages of the infection . The proportion of cells infected by both variants was around 30% in leaves collected 17 days post infection ( dpi ) , and reached nearly 100% in upper leaves collected 60 dpi ( not shown ) . This result interestingly suggested that cell co-infection increased with time , but that it could become frequent enough to “saturate” our experimental system when a 1/1 variant ratio was used in the inoculum: when both variants are detected in nearly all cells it becomes impossible to obtain an accurate MOI estimate with our method . In a second time-course experiment , we thus decided to use a 4/1 ratio for VIT1 and VIT3 . At 21 , 42 , 60 and 84 dpi , fully expanded leaves were collected near the apex of six plants infected in parallel . The results shown in Figure S1A indicate that the relative ratio of VIT1 and VIT3 was indeed close to 4/1 in infected leaves , and remained approximately constant throughout the experiment . Most interestingly , the average proportion of cells infected by both variants dramatically increased in successive sampling times but remained below saturation , suggesting both that the 4/1 ratio was appropriate and that important changes in the MOI may occur during the invasion of the host . The calculated average MOI values showed a dynamic pattern , starting at lows around 1 , sharply increasing up to 13 and then decreasing late in infection ( Figure S1B ) . Unfortunately , important variation between the six replicated plants at each sampling date resulted in too wide confidence intervals , and the statistical analysis failed to confirm the significance of the observed bell-shaped pattern ( the full data set is provided in Table S1 ) . In order to reduce to a minimum the variations between repeated plants , we very precisely adjusted the leaf-sampling protocol during time-course experiments . The development of every new leaf was periodically scrutinized in six plants infected in parallel , to record the dates of their first appearance in the center of the rosette , and to later estimate their respective age at the sampling time . Leaves were numbered so that the first true leaf ( above cotyledons ) was leaf level 1 . The mixture of CaMV VIT1/VIT3 purified virions ( ratio 4/1 ) was inoculated to leaf levels 3 and 4 , and the first leaf level showing systemic symptoms homogeneously distributed all over its surface was leaf level 6 . The induction of flowering was generally observed around 40 dpi , when leaf 30 appeared . Senescence of individual leaves started when they were approximately 35 days old , whatever the leaf level considered . At each of five time points , one identical leaf level was sampled in the six replicated plants . Selected leaf levels corresponding to the five time points were levels 6 , 12 , 21 , 33 and 43 . All leaves were sampled at the same age ( 13 days after their apparition on the plant ) to improve comparison among leaf levels . At this age , all cells within the leaf were likely infected as indicated by the high proportion of CaMV-positive cells found during PCR analysis of individual cells ( Table S1 and Table S2 in Supporting online Information ) . Moreover , 13 days old leaves had already gone through the physiological sink-to-source transition that stops import of photo-assimilates and viruses from the phloem [52] . Finally , to limit interference of the sampling process with plant development and systemic infection , several evenly distributed leaf discs ( 0 . 8 cm Ø ) , amounting solely 20% of the total leaf surface , were collected from each leaf . Protoplasts were extracted from each sampled leaf as previously described [18] , [53] . Four treatments were compared for their putative impact on MOI values . To limit potential sources of variation , the experiments were carried out in parallel with the previous experiment on the MOI dynamics and with the same batch of plantlets and inoculum . In all treatments , 6 turnip plants were co-inoculated with VIT1 and VIT3 and the leaves were sampled when they were 28 days old . Sampling was performed exactly as described above except that , for practical reasons , only two leaf levels were sampled ( leaf levels 12 and 33 ) . We reasoned that limiting this experiement to two sampling points could provide enough resolution to address the question of a possible MOI difference in different experimental conditions . In three treatments plants were kept in the same growth chamber as for the experiment shown in Figure 1 and 2 . The first treatment corresponded to a mechanical inoculation exactly as above , the second to the mechanical inoculation with a 4X dose , and the third to a more natural inoculation by aphid vectors . For the latter , 20 individuals of the aphid Myzus persicae ( Sulz . ) were fed on a plant co-infected by the two viral variants and then released on the fourth leaf of healthy plantlets as previously described [54] . Finally , in the fourth treatment plants were mechanically inoculated with a 1X dose but maintained in a greenhouse where they were exposed to approximately 16 hours sunlight and higher temperatures . Under these conditions the rate of leaf appearance was nearly identical to that in the growth chamber , but total biomass was multiplied by three and flowering started approximately one week earlier ( not shown ) . | Viruses are fast evolving organisms for which changes in fitness and virulence are driven by interactions between genomes such as recombination , functional complementation , and competition . Viruses being intra-cellular parasites , one basic parameter determines the potential for such interactions: the cellular multiplicity of infection ( cellular MOI ) , defined as the number of genome units actually penetrating and co-replicating within individual cells of the host . Despite its importance for virus evolution , this trait has scarcely been investigated . For example , there are only three point estimates for eukaryote-infecting viruses while the possibility that the cellular MOI may vary during the infection or across organs of a given host individual has never been conclusively addressed . By monitoring the cellular MOI in plants infected by the Cauliflower mosaic virus we found remarkably ample variations during the development of the infection process in successive leaf levels . Our results reveal that the opportunities for recombination , complementation and competition among viral genomes can greatly change at different infection phases and at different locations within a multi-cellular host . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"evolutionary",
"biology",
"virology",
"plant",
"biology"
] | 2010 | Dynamics of the Multiplicity of Cellular Infection in a Plant Virus |
We have investigated the immunogenicity in rabbits of native-like , soluble , recombinant SOSIP . 664 trimers based on the env genes of four isolates of human immunodeficiency virus type 1 ( HIV-1 ) ; specifically BG505 ( clade A ) , B41 ( clade B ) , CZA97 ( clade C ) and DU422 ( clade C ) . The various trimers were delivered either simultaneously ( as a mixture of clade A + B trimers ) or sequentially over a 73-week period . Autologous , Tier-2 neutralizing antibody ( NAb ) responses were generated to the clade A and clade B trimers in the bivalent mixture . When delivered as boosting immunogens to rabbits immunized with the clade A and/or clade B trimers , the clade C trimers also generated autologous Tier-2 NAb responses , the CZA97 trimers doing so more strongly and consistently than the DU422 trimers . The clade C trimers also cross-boosted the pre-existing NAb responses to clade A and B trimers . We observed heterologous Tier-2 NAb responses albeit inconsistently , and with limited overall breath . However , cross-neutralization of the clade A BG505 . T332N virus was consistently observed in rabbits immunized only with clade B trimers and then boosted with clade C trimers . The autologous NAbs induced by the BG505 , B41 and CZA97 trimers predominantly recognized specific holes in the glycan shields of the cognate virus . The shared location of some of these holes may account for the observed cross-boosting effects and the heterologous neutralization of the BG505 . T332N virus . These findings will guide the design of further experiments to determine whether and how multiple Env trimers can together induce more broadly neutralizing antibody responses .
Multiple recombinant , soluble envelope glycoprotein ( Env ) trimers from human immunodeficiency virus type 1 ( HIV-1 ) are currently being produced as immunogens for induction of neutralizing antibody ( NAb ) responses [1–5] . As NAbs act by recognizing native Env trimers on the HIV-1 surface , trimer-based immunogens intended to induce NAbs should mimic the native structure as closely as possible [6–12] . The SOSIP . 664 design of soluble native-like trimers is based on the natural cleavage of the gp140 precursor protein into its gp120 and gp41ECTO subunits and stabilization of the metastable trimer by engineered sequence changes [13] . Multiple native-like SOSIP . 664 trimers based on env sequences from clades A , B and C have now been described [7 , 9 , 10 , 14–16] . When tested as individual immunogens in rabbits , two such trimers ( BG505 , clade A; B41 , clade B ) induced consistently high titers of NAbs against the autologous viruses , which are classified as Tier-2 on the neutralization sensitivity spectrum [5 , 16] . While the generation of autologous Tier-2 NAbs is likely to be a necessary step in a vaccine-development program , it is clearly not sufficient [17–20] . The diversity of circulating HIV-1 strains is so extensive that , for immunogens to be practically useful , they must be able to induce broadly active NAbs ( bNAbs ) that can counter a wide range of viruses . The key unanswered question is how such bNAbs can be induced . Several approaches to the bNAb-induction problem have been proposed . One strategy involves the use of engineered Env proteins designed to engage bNAb-germline antibodies and provide a path towards affinity maturation [21–27] . Others are based on the use of sequential immunogens derived from infected individuals that developed bNAbs [24 , 26 , 28–32] . However , one technically straightforward method that requires further evaluation is the simple combination of genetically diverse trimers , such as those based on different clades . The availability and immunogenicity of the clade A BG505 and clade B B41 SOSIP . 664 trimers allow appropriate experiments to be devised and conducted in rabbits . Two obvious ways to combine trimers are to deliver them simultaneously or sequentially , and we have assessed both approaches . We then extended the study in time to assess the effect of heterologous boosting with a clade C trimer , either DU422 or CZA97 . 012 , both as SOSIP . 664 . Overall , the experiment allows us to address several questions relevant to the induction of Tier-2 NAb responses: Do trimers interfere with each other’s immunogenicity or do they generate independent or even reinforcing antibody responses ? Does cross boosting occur when a second or third trimer is given to animals previously immunized with a different trimer ? Can the use of more than one trimer increase the breadth of the NAb response at the Tier-2 level ? Taken together , our data suggest that the autologous Tier-2 NAb responses to different SOSIP . 664 trimers are generated independently , any interference effects being at most modest; that cross boosting can occur; and that while some cross-neutralizing Tier-2 NAbs can arise when more than one trimer is used , they are limited in frequency , breadth and magnitude . The mechanisms underlying several of these observations may be based on observations that the autologous NAb responses to the BG505 . T332N , B41 and CZA97 trimers target specific holes in the glycan shields of the corresponding viruses , some of which are shared between immunogens . The inference is that , if trimer cocktails are to be pursued , particularly for boosting responses initiated by germline bNAb-targeting trimers , their composition should not be chosen randomly , but rather based on knowledge of the antigenicity , structure and immunogenicity of each individual component . Moreover , knowledge of how individual trimers induce autologous Tier-2 NAb responses will aid in engineering of improved variants with greater immunogenicity .
The designs and in vitro properties of the BG505 and B41 SOSIP . 664 trimers have been described previously [10 , 15] as well as the D7324 epitope-tagged version of B41 SOSIP . 664 trimers , designated B41 SOSIP . 664-D7324 ( abbreviated to B41-D7324 ) [15] . All three of these trimers were produced in stably transfected Chinese Hamster Ovary ( CHO ) cell lines ( Flp-in CHO cell line from Thermo-Fisher ( Cat# R75807 ) ) . Unfortunately , because of a cell line-contamination problem that was not detected until after the first phase of the immunization experiment was completed , the B41-D7324 trimer preparation contained ~25% of co-purified BG505 SOSIP . 664 trimers . The suspected contamination was detected by liquid chromatography-mass spectrometry of the immunogen preparation . Briefly , the purified trimer samples were denatured , reduced , alkylated , deglycosylated with PNGaseF , and digested either by trypsin or GluC protease . The trimer composition was estimated from the relative intensities of the peptides unique to each construct [33] . Once the problem was known , the immunizations were continued with contaminant-free B41-D7324 trimers made from a different cell line ( see Fig 1 ) . For late boosts , SOSIP . 664 trimers derived from two different clade C viruses were used: CZA97 . 012 ( abbreviated to CZA97 ) transiently expressed in 293F cells ( derived from human embryonic kidney 293 cells , as FreeStyle 293-F cells from Thermo-Fisher ( Cat# R79007 ) , [34] ) and DU422 expressed in a stable CHO cell line [14] ( Fig 1 ) . All stable CHO cell lines were made by the same procedures described elsewhere for the BG505 SOSIP . 664 line , and have broadly comparable properties [35] . As only the SOSIP . 664 trimer design was used in these experiments , that descriptor is usually omitted from hereon . The BG505 , B41 , B41-D7324 and DU422 trimers were all purified by 2G12 affinity chromatography followed by size-exclusion chromatography ( SEC ) , while the CZA97 trimers were purified by PGT151 affinity chromatography , all as previously described [10 , 14 , 15 , 34] . Each trimer preparation was highly homogeneous and fully native-like ( >95% ) when analyzed by gel electrophoresis and negative-stain electron microscopy [10 , 14 , 15 , 34] . All of the immunogen preparations were prospectively or retrospectively verified by liquid chromatography mass spectrometry as described above . The authenticity of all the cell lines was confirmed by specific PCR amplification of a 421-bp segment spanning the gp120-gp41 junction in the respective env gene inserts . No other contamination was detected at the cell-line or purified-trimer stages . BG505-D7324 , B41-D7324 and DU422-D7324 trimers were used as ELISA antigens for determining serum antibody titers [10 , 14 , 15] . The BG505-D7324 and B41-D7324 trimers were expressed in CHO cells , the DU422-D7324 trimers in 293F cells , and each was purified by the same 2G12/SEC method used for non-tagged trimers . CZA97 SOSIP . 664-His ( CZA97-His ) trimers , also used for serology ELISA , were produced in 293F cells and purified by PGT151-affinity chromatography and SEC [34] . This study was approved and carried out in accordance with protocols provided to the Institutional Animal Care and Use Committee ( IACUC ) at Covance Research Products ( CRP ) Inc . ( Denver , PA ) , approval number C0014-15 . The rabbits were kept , immunized and bled at Covance in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals , and adhered to the Guide for the Care and Use of Laboratory Animals , National Research Council , 1996 . Rabbit immunizations and blood sampling were carried out under contract at CRP according to the schedule presented in Fig 1 , and essentially as described previously ( 5 ) . Female New Zealand White rabbits ( 5 per group ) were immunized intramuscularly with trimers at various doses ( see Results ) , formulated with 75 Units of Iscomatrix , a saponin-based adjuvant obtained from CSL Ltd . ( Parkville , Victoria , Australia ) via the International AIDS Vaccine Initiative [36] . Neutralizing antibodies in rabbit sera were detected and quantified with Env-pseudotyped viruses in the TZM-bl cell assay as described previously ( Tzm-bl cells are derived from the HeLa cell line and supplied by the NIH AIDS Reagents Program , Catalog Number 8129 [5 , 10 , 15] ) . For additional information on this assay and all supporting protocols see: http://www . hiv . lanl . gov/content/nab-reference-strains/html/home . htm . NAb assays were carried out at either Duke University Medical Center ( DUMC ) [37 , 38] or the Weill Cornell Medical College ( WCMC ) [5 , 15 , 37] . Env-pseudotyped viruses used at DUMC were made with the SG3Δenv backbone [39]; at WCMC , the NL-Luc-AM vector was used [40] . For BG505 neutralization , the autologous Env-pseudotyped viruses bore either full-length BG505 . T332N ( at WCMC ) or cytoplasmic-tail-deleted BG505 . T332NΔCT ( at DUMC ) envelope glycoproteins . When the two variants were directly compared no differences in neutralization sensitivity were observed . The CZA97 cl . 12 virus ( hereafter , CZA97 ) is autologous to the CZA97 SOSIP . 664 trimer . The DU422 . K295N . D386N ( hereafter , DU422 ) autologous virus contains the same two glycan-site knock-in mutations at positions 295 and 386 that were incorporated into the DU422 SOSIP . 664 trimer immunogen [14] . All three of the B41 , CZA97 and DU422 Env-pseudotyped viruses were based on full-length Env glycoproteins . Their Tier-2 classifications , which were determined at DUMC , have been described elsewhere [37 , 41] . In addition , sera from a subset of time points were tested at DUMC against two Tier-1A Env-pseudotyped viruses , MN . 3 ( clade B ) and MW965 . 27 ( clade C ) ; and against a panel of nine Tier-2 viruses: Ce703010217_B6 ( clade A ) , 246-F3_C10_2 ( clade AC ) , CNE55 ( CRF01_AE ) , TRO . 11 ( clade B ) , X1632-S2-B10 ( clade B ) , BJ0X002000 . 03 . 2 ( CRF07_BC ) , CH119 . 10 ( CRF07_BC ) , 25710–2 . 43 ( clade C ) and Ce1176_A3 ( clade C ) [42 , 43] . An amphotropic murine leukemia virus ( MLV ) Env-pseudotyped virus [44] was used as a negative control at both sites to determine non-specific inhibition of infection by rabbit sera . In each assay , all serum dilutions were tested in duplicate . Neutralization was defined as the reduction ( % ) of the infectivity obtained in the absence of serum . The serum dilution factors reducing infectivity by 50% were calculated from nonlinear regression fits of a sigmoid function ( with maximum constrained to ≤100% and minimum unconstrained ) to the normalized inhibition data by the use of Prism software ( Graphpad ) . For convenience , the resulting reciprocal titers or dilution factors [1/ ( IC50 [ppV] ) ] or reciprocal inhibitory dilution [1/ ( ID50 [ppV] ) ] are henceforth referred to simply as “titers” or “IC50 values” . We sometimes observed that neutralization curves plateaued at values of <100% ( sometimes < 50% ) , particularly with the B41 Env-pseudotyped virus; a mechanistic analysis of the variable plateaus will be described elsewhere . We recorded IC50 values only when the maximum percentage neutralization was > 50%; other values were recorded as IC50 < 20 . The titers against the MLV control virus were consistently < 30 . We therefore deemed titers >40 to represent robust neutralization . Mutant BG505 and B41 full-length env genes containing point substitutions , most of which introduced potential sites of N-linked glycosylation , were made as previously described ( 5 , 10 , 15 ) . Env-pseudotyped virus mutants based on these genes , together with the corresponding wild-type viruses , were used in TZM-bl cell neutralization assays , as described above . Viruses based on various clones of the MG505 virus are derived from the mother of the BG505 HIV-1-infected infant [45 , 46] . The amino-acid sequences of the MG505 cl . A2 and MG505 cl . H3 clones used in this study , and their relationship to the BG505 sequence , are shown in S1 Fig . The CZA97 cl . 12 and cl . 29 Env-pseudotyped viruses differ at several Env positions , as described in S1C Fig . Mutants of the CZA97 cl . 12 virus , which is autologous in sequence to the CZA97 SOSIP . 664 trimers , were made as described elsewhere [47] . The infectivities of the various mutant viruses were similar ( within 10-fold ) to those of the corresponding parental viruses . The mutant Env-pseudotyped viruses were tested in TZM-bl cell neutralization assays as described above . At a low dilution of serum ( 1/50 for BG505 and B41 at WCMC , or 1/60 for CZ97 at DUMC ) , the extent of neutralization was expressed as a percentage of that obtained with the wild type virus ( defined as 100% ) . In addition , sera were titrated to > 40 , 000-fold dilution , and the inhibition curves for wild-type and mutant viruses were compared . The trimer model was derived from the PDB 5ACO structure [48] . The glycan added at position 241 was modeled in accordance with the homologous glycan in JR-FL Env ( PDB 5FUU; [49] ) the glycan at 289 , for which there are no experimental data , was modeled as a Man3 structure . Modeling and visualization were performed by the UCSF Chimera method ( PMID 15264254 ) . Anti-trimer capture ELISAs with the BG505-D7324 , B41-D7324 and DU422-D7324 trimers were performed as described previously [10] . Briefly , the Env protein antigens were captured onto the solid phase by the sheep antibody D7324 ( Aalto Bio Reagents , Dublin , Ireland ) , the rabbit sera were titrated , and the Env-bound antibody detected . The CZA97-His trimers were captured by an anti-His antibody ( 6x-His Epitope Tag Antibody 4A12E4 , Thermo Scientific ) , but the other steps of the assay were identical to those in the D7324-based ELISA . Midpoint titers ( EC50 ) were derived from the titration curves and serve as measurements of the trimer-binding antibody response to the immunogens [10 , 15] . The peptide APTKAKRRVVQREKR corresponding to the epitope-tag on the B41-D7324 trimers was obtained from GenScript , as was a scrambled control peptide , KKQARRARPVREVKT ) . The peptides were used to coat Maxi-Sorp ( Nunc ) ELISA plate wells at 0 . 6 μg/ml . After standard washing and blocking procedures , rabbit sera were added at a dilution of 1/20 in PBS with 2% milk and bound antibodies were detected with a goat anti-rabbit IgG-HRP conjugate diluted 1/3000 ( BioRad ) . As positive controls , sheep Ab D7324 was titrated from 10 to 0 . 013 μg/ml , and a pool of IgG purified from HIV-1 infected humans ( HIVIG; supplied by the NIH AIDS Reagents Program , as #3957 , lot # 140406–23 ) was titrated from 1000 to 1 . 3 μg/ml . The detection antibodies were rabbit anti-sheep IgG-HRP ( Thermo ) and goat anti-human IgG-HRP ( BioRad ) , respectively . Groups were compared by two-tailed Mann-Whitney U tests . Two time points for the same rabbits were compared by Wilcoxon matched-pairs test; the matched pairs results are presented only when the pairing was efficient , according to correlation analyses within the test . Otherwise the results of Mann-Whitney U test are given . To determine whether titers were significantly above the cut-off value of a titer of 20 , we applied the Wilcoxon signed rank test . Correlations were analyzed non-parametrically as Spearman rank correlations and the concomitant significances were calculated . All statistical analyses were performed in Prism ( GraphPad ) . Group sizes of 5 for rabbit immunization studies limit the robustness of conclusions from group comparisons . Here , and in earlier papers [5 , 16] , we show that SOSIP trimer-induced autologous NAb titers in responding rabbits can span a >3-log range ( i . e . , <100 to >10 , 000 ) , some rabbits not responding at all [5 , 16] . Hence , the ranges of responses in groups often overlap .
The 8 groups of rabbits and the immunization regimens are summarized in Fig 1 . As noted in Methods , after the immunizations had been initiated , we discovered that a stable CHO cell line producing B41-D7324 trimers had become contaminated with cells from a BG505 SOSIP . 664 trimer-expressing line . The resulting immunogen preparation used until week 24 therefore contained both B41-D7324 and BG505 trimers in an approximate 4/1 ratio ( Fig 1 ) . B41-D7324 trimers used after week 24 were made from a non-contaminated line and verified to be authentic ( see Methods ) . The D7324-tag was not itself immunogenic and its presence did not markedly influence the immunogenicity of the B41 trimers ( see S1 Text and S1 Table ) . The same adjuvant , Iscomatrix at 75 U per dose , was used for all the immunizations , which were given intramuscularly . For monovalent immunizations , a trimer dose of 30 μg was always used . In all the groups , the rabbits were bled immediately before and 2 weeks after each immunization to obtain sera that were used in virus-neutralization and Ab-binding assays . The neutralization assays , performed at two different sites ( WCMC and DUMC ) used Env-pseudotyped viruses and the TZM-bl cell line ( see Methods ) . The principal endpoints were the autologous Tier-2 NAb responses to the immunogen trimers , but we also quantified heterologous NAb responses against both Tier-1 and Tier-2 viruses . In addition , binding titers of antibodies against the cognate SOSIP . 664 trimers were determined in an antigen capture ELISA . The initial phases ( weeks 0–24 ) of immunization groups 1 ( B41 ) and 3 ( BG505 ) , before heterologous boosting , serve as comparators for other groups ( Fig 1A ) . One BG505 trimer-immunized rabbit ( #5726 , group 3 ) developed a strong response after the second immunization ( a titer of 680 ) , but that is atypical in our general experience and no such strong early autologous response occurred in the B41 group . Much more consistent autologous NAb responses were raised to each trimer after the third immunization ( at week 20 ) that were poorly boosted by the fourth at week 24 and declined further over the next 12 weeks ( Fig 2A and 2B ) . The autologous NAb responses in the BG505-trimer-immunized group 3 ( Fig 2A ) at week 22 were in the range 4100–12 , 000 ( median 6000 ) . For group 1 animals that received only the B41 trimers ( Fig 1A ) , the corresponding autologous titer ranges in the four responders were 1100–10 , 000 ( median 4800 ) , while the fifth rabbit ( #5715 ) had a borderline response ( peak titer of 40 ) ( Fig 2B ) . Some moderate cross-neutralization of the BG505 . T332N virus occurred at this stage of the experiment ( Fig 2A ) . Three rabbits in the B41 trimer group ( #5713 , #5715 and #5717 ) did develop titers >40 ( 70–140 ) to BG505 . T332N at week 22 ( Fig 2A ) , and other rabbits in this group did so after they were later boosted with clade C trimers ( see below ) . However , the converse was not seen , in that rabbits immunized with BG505 trimers did not develop NAbs against the B41 virus ( Fig 2B ) . The rest period from the fourth immunization ( week 24 ) to the fifth at week 36 ( see below ) allowed us to look at how both autologous NAb responses diminished over time . We were unable to determine the half-lives of the titers because of the paucity of time points , and are aware that a second phase of titer decay might commence during the 10-week period from the assay at week 26 . However , we note that the autologous responses to the BG505 and B41 trimers in groups 1 and 3 dropped ~10-fold during this time . Responses in these and other groups to later boosting with clade C trimers are discussed further below and are shown in Fig 2C and 2D . We analyzed the effect of immunogen dose , and whether one co-administered trimer could interfere with the autologous NAb response to the other . These analyses are described in the SI ( see S3 Fig ) . In summary , we found no strong evidence that either increasing the trimer dose above the 30-μg standard , or the presence of a second trimer , affected the autologous Tier-2 NAb responses to the clade A and B trimers . There was some indication of a modest reduction in the BG505 . T332N response at 22 when B41 trimers were co-delivered , but not the converse ( S3 Fig ) . Further studies would be required to determine whether the former outcome is attributable to genuine interference , to a dose-reduction effect , or simply to the random variation associated with a group size of 5 rabbits . The more robust conclusion is that a bivalent mixture of clade A and clade B trimers usually elicited autologous NAb responses to both Tier-2 viruses . Groups 5 and 6 were also affected by the BG505 contamination ( ~20% ) of the initial B41-D7324 trimer preparations ( Fig 1C ) . Nonetheless , some useful information can be gleaned from the experiment in its full duration ( Fig 3 ) . The BG505 . T3332N NAb responses for group 5 were quite strong at week 22 ( median titer 1100 ) , which may reflect an amplification of responses by the standard BG505 trimer dose ( 30 μg ) at week 20 that were primed by the earlier exposure to lower doses ( ~6 μg at weeks 0 and 4 ) . The next two standard doses of BG505 trimer at weeks 24 and 36 boosted the BG505 . T332N NAb titer only modestly , to a median of 2100 at week 38 ( Fig 3A ) . There was also an increase in the BG505 . T332N NAb median titer ( from 85 to 200 ) between weeks 60 and 62 , in response to the B41-D7324 trimer boost at week 60 ( Fig 3A ) . This rise could reflect a new cross-reactive response to the clade B trimer or some cross-boosting of the earlier response to the clade A trimer ( see below ) . The B41 NAb response for group 5 at week 22 was unexpectedly strong in the three responding rabbits , considering that these animals had received only two doses of B41-D7324 trimer ( at weeks 0 and 4 ) and then a heterologous boost by BG505 trimers at week 20 ( Fig 3B ) . Thus , the B41 NAb titers for group 5 at week 22 ( median 920 ) were not significantly lower than those for group 1 ( median 2100 ) at the same time point ( p = 0 . 14 ) , even though the group 1 animals had received the standard three doses of B41 trimers by this time ( cf . Fig 2B ) . Usually , two doses of a trimer are not sufficient to induce a strong autologous NAb response ( e . g . , see group 1 at week 6 , after 2 doses of B41 trimers , Fig 2B ) . In summary , the effect of the week-20 BG505 trimer immunization on the B41 NAb response of the group-5 rabbits , through cross- boosting , was comparable to that of the third B41 trimer dose on the group 1 rabbits at week 20 . We discuss below the possible mechanisms underlying cross-boosting . The B41 NAb titers then declined until they were boosted again by the later B41-D7324 trimer doses at weeks 48 and 60 and , for the three responding rabbits , at week 73 ( Fig 3B ) . Note that two rabbits ( #5733 and #5737 ) did not develop B41 NAbs at any time-point; the autologous NAb epitope ( s ) on the B41 trimers were not immunogenic in these two animals because of unknown but possibly genetic influences . The BG505 . T332N NAb response in group-6 rabbits was consistently boosted by each immunization given during the 60-week duration of the experiment . This was the case even when B41-D7324 trimers were given at week 36 , which is therefore another example of cross- boosting ( Fig 3C ) . Thus , a prolonged trimer immunization regimen involving multiple boosts is feasible . A B41 NAb response in group 6 was seen in 4 of 5 rabbits only after the third exposure to B41-D7324 trimers at week 36 , and then declined during the later period of BG505-only immunizations ( Fig 3D ) . This pattern is consistent with that seen with group 1 ( B41 trimers only ) in the period up to week 22 , and from then until the clade C trimer boosts began at week 36 ( see Fig 2B ) . Thus , whereas immunizing with B41 trimers enhanced the BG505 . T332N NAb response in both groups 5 and 6 , the converse occurred only in group 5 at week 20 . Below , we discuss the possible mechanisms underlying these observations . We explored heterologous boosting with clade C trimers by immunizing group 1 with DU422 trimers and groups 2 , 3 , 4 and 8 with CZA97 trimers at weeks 36 , 48 and 60 , and selected animals also at week 73 ( Fig 1A and 1B ) . In each case , the trimer dose was 30 μg . If we nominally reset week-36 as the 0-time point for this particular study , the rabbits received three immunizations with clade C trimers at weeks 0 , 12 , 24 and , in some cases , a fourth at week 37 . This ad hoc extension to the rabbit experiment did not include a study of clade C trimers in naïve rabbits ( this experiment is now in progress ) . Hence , there is no comparator arm to judge whether the response to the DU422 and CZA97 trimers was influenced by prior exposure to the clade A and/or clade B trimers . After the second immunization at week 48 , autologous NAb responses developed in 2 of 5 rabbits immunized with DU422 trimers and were boosted by the third and fourth immunizations at weeks 60 and 73 ( titers of 110 , 190 and 84 for rabbit #5715 , and 210 , 350 and 110 for #5716 ) ; the other three animals were non-responders but #5713 , #5714 and the responder #5715 had cross-reactive titers of 60 at week 36 ( Fig 2C ) . The CZA97 trimers induced more consistent and higher-titer autologous responses , particularly after the third immunization . Overall , autologous NAb titers >100 were seen in 12 of the 20 CZA97 trimer recipients at the week-62 time-point ( Fig 2D ) . The median titer for these 12 responders was 1700 , and it was 580 for the entire set of 20 animals . We note that the responses to each of the clade C trimers after the second immunization were more frequent than we generally observe ( here and in other studies ) to two immunizations with other trimers ( see , for example , the week-6 responses to clade A and clade B trimers; Fig 2A and 2B ) . However , there are too many variables to determine what this means: the trimer genotype , the prior exposure to other trimers , and the spacing between the first and second immunizations ( 4 vs . 12 weeks ) could all have influenced the observed outcome . The study design allowed us to further investigate whether heterologous trimers from a different clade can cross-boost existing autologous NAb responses . The first immunization ( week 36 ) with clade C trimers ( DU422 for group 1 , otherwise CZA97 ) of rabbits previously administered clade A and/or B trimers cross-boosted the existing B41 ( groups 1 , 2 , 4 and 8 ) and BG505 . T332N ( groups 3 , 2 , 4 and 8 ) NAb titers ( measured 2 weeks post-boost , i . e . , at week 38 ) in most animals , although to varying degrees ( Fig 2A and 2B ) . For BG505 . T332N neutralization , the titers for group 3 were only marginally higher at week 38 than at week 36 ( median 1000 vs . 620 ) and likewise they were higher after the second clade C trimer boost at week 50 than at week 48 ( 490 vs . 330 ) ( Fig 4A ) . The corresponding effect of the clade C ( DU422 ) boost on the B41 NAb titers for group 1 was also weak for week 38 vs . week 36 , median 110 vs . 91; but significant for week 50 vs . week 48 , median 140 vs . 20; p = 0 . 048 ( Fig 4B ) . The clade C trimer boosts to groups 2 , 4 and 8 combined , which had previously received clade A and B immunogens simultaneously , also cross-boosted higher BG505 . T332N NAb titers at week 38 compared with the pre-boost level at week 36 ( 120 vs . 58 , p = 0 . 022 ) ; the same effect was seen at week 50 compared with week 48 ( p = 0 . 013 ) ( Fig 4C ) . In these groups , the B41 NAb titers rose only at week 38 compared with week 36 ( medians 240 and 70 , p = 0 . 066 ) ; they were not sustained between weeks 50 and 62 ( Fig 4D ) . No cross boosting of the B41 response was seen after week 48 . The diminishing effect of cross boosting over time may reflect the increasing period after the last exposure to the initial immunogen ( s ) , i . e . at week 24 . The week 60 and 73 immunizations with clade C trimers boosted the BG505 . T332N NAb titers for group 1 only , which was a completely heterologous response ( no exposure to BG505 trimers ) that is described more fully below ( Fig 5 ) . Overall , the evidence suggests that immunization with a clade C trimer was able to cross-boost the autologous NAb responses induced by prior exposure to the BG505 trimer or the B41 trimer . We also explored whether boosting with clade C trimers broadened the overall neutralization response . One notable observation was that NAbs against the BG505 . T332N virus were detected in 5 of 5 rabbits from group 1 at week 38 ( median titer 180 , p = 0 . 062 ) ( Figs 2A and 5 ) . These animals had been sequentially immunized with B41 and then DU422 trimers , but had never received BG505 trimers ( Fig 1A ) . Hence , they developed a heterologous response to the BG505 . T332N virus . The response was even stronger after the week-60 boost ( median titer of 530 at week 62 , p = 0 . 062 ) , and the two rabbits ( #5715–1 , #5716–1 ) that received a further DU422 trimer immunization at week 73 had BG505 . T332N titers of 290 and 640 at week 75 ( Figs 2A and 5 ) . Conversely , no NAbs against B41 were induced in the group 3 rabbits that were initially immunized with the BG505 trimer and then boosted with its clade C CZA97 counterpart ( Fig 2B ) . During the 60- or 73-week immunization period , the 15 rabbits in groups 2 , 4 and 8 received three different trimers from clades A , B and C ( Fig 1B ) . All 15 rabbits raised autologous NAb responses ( titer >100 ) to at least one of the trimer immunogens at various time points . Ten of the rabbits were triple responders , in that they raised moderate/strong NAb responses ( titers >100 ) to each of the corresponding autologous viruses ( Fig 2A , 2B and 2D; Table 1 ) . Sera from three double responders ( ( #5728–4 , #5748–8 and #5722–2 ) neutralized the BG505 . T332N and B41 viruses but not CZA97 , while sera from two rabbits ( #5752–8 and #5720–2 ) neutralized only BG505 . T332N ( Fig 2A , 2B and 2D; Table 1 ) . In groups 5 and 6 , after the sequential immunization with BG505 and B41 trimers , all 10 rabbits developed medium/strong NAb responses against BG505 . T332N ( peak titers 630–8200 ) . Six of these rabbits also raised NAbs against B41 ( peak titers 340–4700 ) , whereas the other four ( #5733–5 and #5737–5; #5740–6 and #5742–6 ) did not ( Fig 3 ) . We then analyzed the heterologous Tier-1 ( MN , clade B and MW965 . 26 , clade C ) NAb titers in all rabbits ( Fig 6 ) . We compared the Tier-1 NAb titers of “complete responder” rabbits with animals that responded less consistently . The “complete responder” subset included rabbits from groups 2 , 4 and 8 that developed autologous Tier-2 NAbs in response to all three trimers they received ( n = 10 ) , as well as rabbits from groups 5 and 6 that did so in response to both trimers they were given ( n = 6 ) ( see Table 1 ) . The Tier-1 NAb responses of these 16 “complete responder” rabbits ( median titers of 1400 against MN . 3 and 6200 against MW965 . 26 ) were indistinguishable from those of the remaining 9 rabbits from the same groups that developed Tier-2 NAbs less consistently ( median titers of 980 against MN . 3 and 6200 against MW965 . 26; p = 0 . 64 for MN . 3 and p = 0 . 89 for MW965 . 26 ) . Thus , the Tier-2 autologous and Tier-1 NAb responses induced by the various trimers do not track with one another . To supplement the above analyses , we plotted the Tier-1 NAb titers ( again MN . 3 and MW965 . 26 ) for all groups against the various autologous Tier-2 NAb titers at weeks 22 and 62 ( Fig 6 , S2 Fig , S2 Table ) . Spearman correlation analyses again showed that the Tier-1 and autologous Tier-2 NAb titers were consistently non-correlated . This conclusion is concordant with our previous findings , where we also showed that peptide-reactive , anti-V3 antibodies dominate the Tier-1 NAb response to BG505 SOSIP . 664 trimers , and that these responses do not correlate with BG505 neutralization [5] . Why are the autologous Tier-2 NAb responses to multiple trimers strong and consistent in some rabbits , but weak or absent in others ? As described above , the poor Tier-2 NAb responses in the latter subset of rabbits are not attributable to a global inability to respond to the trimers , as the Tier-1 NAb titers in these animals were not atypical . The explanation might be rooted in whether the antibody repertoires of individual animals can recognize the types of Tier-2 NAb epitopes that are described below . Thus , whereas all the rabbits may be capable of raising antibodies to Tier-1 NAb epitopes , some may be incapable of responding to Tier-2 NAb epitopes . Moreover , the different characteristics of the epitopes for Tier-1 and Tier-2 NAbs may explain why their titers are uncorrelated . We based our mapping strategy on the hypothesis that autologous Tier-2 NAbs recognize small holes in the glycan shield where a glycan is absent that is present in multiple HIV-1 strains [50] . Accordingly , we inserted N-linked glycans at specific sites on the BG505 . T332N and B41 Env-pseudotyped viruses to fill "glycan holes": For BG505 . T332N , these holes are at positions 130 , 241 and 289; the resulting BG505 . T332N virus mutants are designated BG505-Q130N , BG505-S241N , BG505-P291T ( inserting a glycan at position-289 ) and BG505-S241N+P291T ( double mutant containing glycans at positions-241 and -289 ) . We also used clones of the maternal virus MG505 , including the cl . A2-K241S mutant in which the lysine in the mother’s virus at residue-241 was changed to the serine found in the infant’s virus [46] ( S1 Fig ) . The converse substitution at residue 241 was used to make the BG505-S241K mutant . For mapping the B41 response , we identified potential glycan holes at positions 130 and 289 by inspecting the sequence in the structural context , and made the B41-N132T ( glycan at position-130 ) and B41-A291T ( glycan at position-289 ) mutants . For CZA97 , we used two clones that differ by the presence ( cl . 29 ) or absence ( cl . 12 ) of a glycan at position-411 , together with other nearby substitutions ( see Methods and S1 Fig ) . The neutralization sensitivities of the various viruses to sera from BG505 , B41 or CZA97 trimer-immunized rabbits were then assessed in the TZM-bl cell assay ( Fig 7 ) . We analyzed the autologous NAb response to the BG505 trimer in 30 different rabbits . The key results are based on the peak BG505 . T332N responses recorded in each rabbit , and represent relative neutralization at a single serum dilution of 1/50 ( Fig 7A ) . One serum that fully neutralized the mutants at the 1/50 dilution did so with reduced efficacy at higher dilutions when the samples were titrated; in those cases , the relative neutralization titers are given in S4 Fig and reinforce the conclusions drawn below . Additional analyses include how the specificity of the BG505 . T332N NAb response changed over time in some rabbits , and whether the immunization regimen influenced the specificity of the response ( S1 Text , S3A Table ) . Overall , the 30 rabbits fell into two major sub-groups based on whether the sera could neutralize BG505 . T332N virus mutants containing glycan knock-ins at positions 241 and/or 289 , or the S241K point substitution . Neutralization by 16 of the 30 sera was partially or completely eliminated when a glycan was present at both of these sites; the single glycan knock-in mutants and the double 241+289 mutant viruses yielded highly concordant data ( Fig 7A ) . In addition , reduced capacities to neutralize the glycan knock-in mutants were revealed when serum #5739–6 from week 62 was titrated ( S4 Fig ) . Our interpretation , supported by structural modeling , is that each of the knocked-in 241 and 289 glycans occludes the target epitope for prominent autologous NAb specificities present in these 17 sera ( Fig 8 ) . In general , the neutralization properties of the S241K mutant tracked those of the glycan knock-in mutants , and indicate that Ser-241 has an important influence on this NAb epitope ( Fig 7A ) . In contrast , the other 13 sera efficiently neutralized the 241 and 289 single and double glycan knock-in mutant viruses , and also the S241K mutant ( Fig 7A , S3A Table ) . It is possible that these sera target the same region of the trimer in a way that is unaffected by the knocked-in glycans or the presence of a Lys at residue-241 . Alternatively , they may target an entirely different and as yet unknown epitope ( s ) elsewhere on the trimer , in addition to or instead of one located in the 241/289 region . Those epitopes do not involve a glycan hole present on the BG505 . T332N virus at residue-130 because 29 of the 30 sera neutralized the Q130N glycan knock-in mutant with the same efficacy as the wild-type virus , and the decrease in neutralization seen with the other serum ( #5717–1 ) was only modest ( Fig 7A ) . Additional clues , and also complexities , are provided by neutralization assays with clones of the MG505 maternal virus and mutants thereof . These clones differ from BG505 . T332N in several positions , including at position-241 ( S1 Fig ) . The MG505 cl . A2 virus was partially or completely resistant to 28 of the 30 sera but , in 26 of those 28 cases , it became more sensitive when the K241S change was made to restore the residue found at this position in the BG505 . T332N virus ( Fig 7A , S3A Table ) . The latter effect is just as expected for the sera that fail to neutralize the BG505-S241K mutant . However , we noted four examples ( #5723–2 , #5746–7 , #5751–8 and #5738–6 ) where the presence of a lysine at position-241 was associated with neutralization resistance in the MG505 context ( i . e . , compare MG505 cl . A2 with MG505 cl . A2-K241S ) , but not in the BG505 context ( compare BG505 . T332N with BG505-S241K , and also the BG505 glycan knock-in mutants with the same phenotype ) . In those four cases , a lysine at position-241 apparently acts in concert with other sequence differences between the MG505 cl . A2 and BG505 . T332N viruses to impair crucial epitopes . Overall , the data suggest that residue-241 has a direct or indirect influence on the NAb epitope recognized by the above four sera , in addition to the more clear-cut role it plays for 17 more of the 30 sera ( Fig 7A , S3A Table , S4 Fig ) . The MG505 cl . H3 virus was resistant to all 30 of the rabbit sera ( Fig 7A , S3A Table , S4 Fig ) , and has a glycan at position-241 ( S1 Fig ) . Taken together with the data on the cl . A2 viruses , it is possible that the 241/289-region of the BG505 trimer may be targeted more frequently than the studies on the BG505 . T332N mutants suggest ( i . e . , by >17/30 sera ) . However , the wider sequence context of the MG505 clones presumably plays a role . For example , changes elsewhere may indirectly affect how the epitope ( s ) in the 241/289 region is presented on MG505 viruses . Alternatively , if other , unknown NAb epitopes on the BG505 . T332N virus are also targeted by 13 of the rabbit sera and they are absent on the MG505 viruses , this would increase the impact of changes at residue-241 in the MG505 context ( e . g . , the K241S change to MG505 cl . A2 ) . As noted above , heterologous NAbs against BG505 . T332N were induced in four group 1 rabbits immunized with B41 followed by DU422 trimers ( Figs 2A and 5 ) . These animals were never given BG505 trimers ( Fig 1A ) . When the peak sera from these animals were tested against the BG505 mutant viruses , three of them ( #5713 , #5716 and #5717 ) were found not to neutralize the 241/289 double glycan knock-in mutant ( Fig 7 ) . This finding suggests that the cross-reactive NAbs target the 241/289 region of the trimer . We note that the B41 trimer has a glycan at position-241 but not -289 , while the DU422 trimer has glycans at both positions . In two rabbits ( #5713 and #5715 ) , the extents to which the glycan knock-in mutants were neutralized increased over the course of the DU422 trimer immunizations ( S3A Table ) , possibly reflecting the shielding effect of the glycans in this region of the immunogen . Sera from B41 trimer-immunized rabbits were tested against two virus mutants containing knocked-in glycans to occlude potential holes at positions 130 and 289 ( Fig 7B , S3B Table ) . The B41-N132T variant retained the wild-type neutralization sensitivity to all sera . In marked contrast , the B41-A291T mutant ( which has an added glycan at position-289 ) was completely ( <10% neutralization relative to WT ) resistant to neutralization by 20 of the 22 sera and partially resistant to the other two ( Fig 7B ) . Thus , the autologous NAbs induced by the B41 trimer predominantly target a hole in the shield caused by the absence of a glycan at position-289 . The finer details of this epitope remain to be explored by the use of additional mutant viruses . We tested two clones of the CZA97 virus and found that they were either highly sensitive ( cl . 12; 12 of 12 rabbits ) or strongly resistant ( cl . 29; 10 of 12 rabbits ) to sera from rabbits immunized with CZA97 trimers ( Fig 7C ) . A notable difference between these two clones is the presence of a glycan at position-411 in the V4 region of cl . 29 that is absent from cl . 12 , together with 7 single-residue changes between residues 388 and 415 ( see alignment in S1 Fig ) . Thus , we hypothesized that the autologous NAb response to the CZA97 trimers targets the glycan hole at position-411 . Accordingly , we made the CZA97 cl . 12-D411N and cl . 29-N411D point-mutant viruses , respectively , to introduce and delete a glycan at this position in the sensitive and resistant clones . The addition of the 411-glycan made the sensitive cl . 12 virus partially or fully resistant to 7 of the 12 sera but had less effect on the other 5 . The neutralization titers of these five were nevertheless substantially reduced by the mutation ( S4 Fig ) . However , removing the 411-glycan from the resistant cl . 29 virus did not increase its neutralization sensitivity ( Fig 7C ) . We conclude that the autologous NAbs induced by the CZA97 trimer target the V4 region and are often influenced by the absence or presence of the 411-glycan , but that other nearby , variable residues also affect the presentation of the epitope ( s ) . Additional mutant viruses are required to delineate the epitope more fully . To assess the breadth of neutralization , we tested sera from all 40 rabbits at weeks 22 and 26 and , for the appropriate groups also at weeks 50 , 62 and 75 , against a panel of 9 heterologous Tier-2 viruses ( Table 2 ) . Sera from 25 of the 40 rabbits detectably neutralized at least one test virus at one or more time points when an IC50 cutoff of > 20 was used , and 12 sera did so when a more rigorous cut-off of > 40 was used . There were four examples of heterologous NAb titers > 100 ( of three different sera ) . Several of the heterologous responses were detected at the week-22 time point ( 2 weeks after the third immunization ) , but none at week 26 ( 2 weeks after the fourth ) . A few additional positive responses were seen at weeks 50 and 60 , but none persisted until week 62; several more , but still sporadic , responses were recorded at weeks 73 and 75 . The three most frequently neutralized viruses in the panel were: 25710–2 . 43 ( clade C ) , TRO . 11 ( clade B ) , and Ce1176_A3 ( clade C ) , while the least frequently hit was BJOX002000 . 03 . 2 ( CRF07_BC ) . The glycan sites in the Env proteins of these isolates are shown in S1D Fig . We note that virus 25710–2 . 43 lacks the 241-glycan but has the sequence KVS at that site . However , two of the three sera that neutralized 25710–2 . 43 at a titer >40 failed to neutralize the BG505-S241K mutant ( #5726–3 at week 22 and #5750–8 at week 22; Table 2 and S3A Table ) . The interpretation is that the heterologous neutralization is probably not specific for a glycan hole at the 241-site . Otherwise , there was no apparent pattern in the cross-neutralization data with respect to what trimers had been used to immunize the rabbits or the autologous NAb titers that were induced . We used ELISA to determine the half-maximal binding antibody titers ( EC50 ) to all four of the trimer immunogens ( Fig 9 ) . In contrast to the NAb responses , autologous trimer-binding antibodies were detectable in sera from all rabbits . Cross-reactive binding antibodies , including cross-clade , were commonly observed . The decay of the trimer-binding antibody titers in the resting period between weeks 24 and 36 was more variable than the autologous NAb responses , although titer drops of ~10-fold were again frequently seen . The rates of decay could not be quantified in detail because of the paucity of time points , but were generally high . For simplicity , correlation analyses for the binding antibody and autologous NAb responses were restricted to the monovalent immunogen groups 1 and 3 ( Fig 10 ) . These rabbits received first either the BG505 or the B41 trimer , and later either the DU422 or CZA97 clade C trimer ( Fig 1A ) . NAb responses during the period of only BG505 or B41 immunizations correlated better with the autologous than the heterologous trimer-binding antibody titers , although in group 3 , a good correlation was observed between the BG505 autologous NAb titers and the B41 heterologous binding antibody titers ( for correlation coefficients and significances , see Fig 10 ) . There was a wide spectrum of heterologous trimer-binding antibody titers in sera that lacked any heterologous NAbs , ranging up to 500 for BG505 trimers in group 1 and up to 3400 for B41 trimers in group 3 ( Fig 10 ) . For correlations between binding antibody titers to trimers and NAbs for the clade A , B and C trimers and viruses from weeks 38–75 , see S1 Text and S5 Fig . Overall , we conclude that the various immunizing trimers induce substantial amounts of antibodies that have at least moderate affinities for autologous and heterologous trimers , but that lack the ability to neutralize genetically diverge Tier-2 viruses consistently . The extent to which V3 non-neutralizing antibodies contribute to the reactivity in the ELISA we used to measure trimer-binding antibodies will also need to be considered ( 5 , 10 ) . The strongest correlations between autologous binding and neutralizing antibody titers were seen after the BG505 and B41 monovalent trimer immunizations , and thus support the conclusions we previously drew about how the soluble SOSIP . 664 immunogens are antigenic and structural mimics of virion-associated functional trimers [5 , 49] . Furthermore , the correlations for DU422 trimer binding and the heterologous or cross-boosted NAb responses in group 1 suggest that the NAb specificities induced by the initial B41 trimer remained dominant throughout the period of heterologous ( DU422 ) trimer boosting . At the same time , the NAb response broadened to cover a third virus ( i . e . , BG505 . T332N ) in animals that had not been immunized with that trimer . As there was no binding vs . neutralizing antibody correlation for DU422 itself , which was poorly neutralized , the data may indicate that the immune response to the boosting DU422 trimer was focused on previously seen epitopes . Additional studies would be needed to explore these various ideas .
Native-like soluble trimers , exemplified by SOSIP . 664 and next-generation derivatives , are now being used to explore how bNAbs may eventually be induced by immunization [24 , 26] . Studies of the immune responses to these trimers in various animals serve to identify which approaches are the more promising . In other words , how can the tool kit provided by the creation of native-like trimers be used most efficiently ? This rabbit study was designed to generate information on the immunogenicity of multiple trimers , delivered sequentially or simultaneously . As noted above , the original design of some study groups was compromised by contamination that we detected while the initial set of immunizations was in progress . Other groups were unaffected , and we believe that there remains considerable value in the information derived from this study . One key point relating to the robustness of any conclusions is the limitation imposed by using group sizes of only 5 rabbits . With autologous NAb titers that vary in magnitude from undetectable to >10 , 000 , random skews are inevitable . Moreover , the non-responsiveness of some rabbits , for reasons that are not understood , can also complicate data interpretations . Rabbits respond to immunization with either BG505 or B41 trimers by inducing autologous NAbs against the corresponding Tier-2 viruses , as well as cross-reactive NAbs to Tier-1 viruses [5] . The latter responses , which are predominantly directed against V3 , can be reduced by trimer-stabilization strategies that decrease the antigenicity and immunogenicity of V3 and other non-NAb epitopes [16] . However , to date the unmodified and stabilized versions of the BG505 , B41 or other SOSIP . 664 trimers have not induced substantial titers of bNAbs against heterologous Tier-2 viruses when delivered alone [5 , 16 , 51] . Can using more than one trimer improve matters in this regard ? When rabbits were immunized sequentially with the BG505 and B41 SOSIP . 664 trimers they eventually generated autologous Tier-2 NAbs against both viruses . The consistency and high titers of these NAb responses were little different from when the same trimers were administered alone . Thus , rabbits can respond efficiently to more than one native-like trimer , when each is given three times over a prolonged period ( ~20 weeks ) . Fewer immunizations do not generate consistent and strong responses; we have not yet explored the minimum rest period required for the third immunization to be optimally effective . After the later clade C trimer boosts , several rabbits given three different trimers raised autologous NAbs to each one at various times . The response to the first trimer did not interfere markedly with the ability to raise autologous NAbs to the second one . In some cases , there was evidence for cross boosting , in that a second trimer modestly increased previously primed NAb titers to the first; this effect was observed most consistently when clade C trimers were given as heterologous boosts , particularly in respect to boosting the BG505 . T332N NAb response . Both clade C trimers were immunogenic for autologous Tier-2 NAb responses , CZA97 being markedly more so than DU422 . We have not yet obtained data on the de novo responses to these two trimers in rabbits , but it seems unlikely they would be very different from what was seen in the present experiment . The strong responses to the CZA97 SOSIP . 664 trimers stand in marked contrast to the inability of uncleaved , non-native CZA97 gp140-Foldon proteins to induce autologous NAbs in immunized animals [1 , 4 , 52] . Taken together , the various observations on sequentially delivered trimers are relevant for devising how to use multiple trimer variants derived from a single genetic lineage; and perhaps also to immunization strategies involving priming with a germline bNAb-reactive trimer , followed by boosting with a more evolved trimer [24 , 26 , 53] . When BG505- and B41-based trimers were co-delivered at similar doses , autologous Tier-2 NAbs were raised against both viruses with no strong evidence for interference . Whether more substantial interference can arise in tri- and tetravalent immunizations will be determined from ongoing studies . The available evidence does not allow any strong conclusions about whether it is better to deliver two trimers each at the normal dose or at half of it ( e . g . , at 2 x 15 μg or 2 x 30 μg if the normal dose is 30 μg ) . The best way to co-deliver multiple ( 3 or more ) trimers as a cocktail is being assessed experimentally . We saw no sign , however , that sequential or simultaneous immunizations with two or three trimers was sufficient to generate consistent , high-level heterologous neutralization breadth at the Tier-2 level . Developing a bNAb response is unlikely to be as simple as mixing a cocktail , which is not to say that using more than one flavor of trimer has no value . When boosting previously primed responses , for example in germline antibody-targeting strategies , that approach may be exactly what is required . Nevertheless , we did detect NAbs against the BG505 . T332N virus at substantial titers in several rabbits immunized with clade B and then clade C trimers . Thus , the NAbs against this Tier-2 virus arose in animals that had never seen BG505 SOSIP . 664 trimers , and in some cases these NAbs appeared to target the 241/289 glycan hole of the BG505 . T332N virus . However , the overall breadth of the Tier-2 response was limited to sporadic neutralization of heterologous viruses at titers that were generally <100 . We mapped autologous NAb responses to the BG505 , B41 and CZA97 trimers to holes in their glycan shields , which supports an earlier suggestion about the nature of neutralization vulnerabilities for Tier-2 viruses [50] . Our new findings clearly supersede our earlier efforts to map the autologous NAb responses to BG505 trimers [5] . We can now show that the responses to these trimers in at least 17 of 30 rabbits target a glycan hole that is blocked by addition of a glycan at positions 241 and/or 289 . Modeling shows that these two sites are closely located on the BG505 SOSIP . 664 trimer structure and implies that each inserted glycan may partially occlude the same hole , from different directions ( Fig 8 ) . The data on the resistant phenotype of the BG505-S241K mutant , reinforced by the comparison of the MG505 cl . A2 and cl . A2-K241S viruses , indicate that serine-241 makes a major contribution to the autologous NAb epitope seen in these 17 rabbits . Whether additional NAb epitopes contribute to the overall response in the other 13 animals is not yet known , but any such site must be affected directly or indirectly by the additional sequence differences present in the MG505 clones ( resistant ) compared with BG505 . T332N ( sensitive ) . Here , variation in the C3 region between residues 354 and 363 may play a key role , perhaps by affecting how the 241/289 region of the trimer is presented [5] . Antibodies directed to gp120-gp41 interface epitopes and that directly or indirectly interfere with CD4 binding have been implicated in the autologous NAb response to BG505 trimers in the guinea pig [50] . How this antibody specificity relates to those described here remains to be determined . We note that the lack of the 241-glycan creates a site of vulnerability on the BG505 virus and a corresponding immunogenic site on the BG505 trimer , and that no such exposed site is present on the maternal MG505 virus . It seems plausible that changes at position-241 , perhaps in concert with others arising elsewhere ( e . g . , in C3 ) , are involved in neutralization escape in this particular virus lineage . Because most circulating HIV-1 strains ( ~97% ) resemble MG505 by bearing a glycan at the position-241 site they will be resistant to the induced antibodies , which is what we observed when the rabbit sera were tested against a Tier-2 virus panel . Moreover , even if a glycan hole were present at position-241 on a particular virus , the identity of the now-exposed residues at the base of the hole would also be important . Thus , the BG505 . T332N S241K mutant , which lacks the glycan but contains a point substitution at position-241 compared with the trimer immunogen , was resistant to many rabbit sera . Once more , the sequence diversity of HIV-1 will be a formidable obstacle to vaccine development . The rabbit response to the B41 trimers was predominantly directed against a glycan hole at residue-289 in C2 ( where 60% of HIV-1 isolates have a glycan ) , while that to the CZA97 trimers seems centered on the V4 region and is likely to involve a glycan hole at position-411 . Studies with additional mutant viruses would be required to define these epitopes in greater detail . We have not yet mapped the autologous NAb response induced in 2 of the 5 rabbits given the DU422 trimers . The finding that the BG505 . T332N and B41 viruses share a glycan hole that is impeded by the introduction of a glycan at position-289 may be relevant to understanding some observations of heterologous boosting and heterologous cross-neutralization . Thus , in the group-1 animals , we observed that immunization with B41-based trimers followed by boosting with DU422 trimers induced NAbs that neutralized the BG505 . T332N virus . These heterologous NAbs targeted the same 241-glycan hole as the autologous NAbs raised against the BG505 trimers . Perhaps some antibodies induced by the immunogenic 289-glycan hole on the B41 trimers can recognize the similar hole on the BG505 . T332N virus , and hence drive a degree of cross-neutralization . Cross boosting of the BG505 . T332N and B41 NAb responses by the clade C trimers , again in a manner sensitive to the 289-glycan in both cases , may be rooted in the same mechanism . Among the three most frequently , but still sporadically , neutralized heterologous Tier-2 viruses we note that 25710–2 . 43 lacks the 241-glycan . Various observations in individual rabbits or sub-groups are not easily understood , including some of the boosting effects of the DU422 trimers that have glycans at both positions 241 and 289 . Overall , these various ideas are tentative and would require additional experiments to confirm or refute . Although our overall emphasis is on suppressing rather than inducing Tier-1 NAb responses , we did quantify them [16] . The kinetics of the Tier-1 NAb responses induced by the BG505 and B41/B41-D7324 trimers clearly show that these antibodies are detectable earlier than , and not correlated with , the Tier-2 autologous NAb responses in the same animals ( Figs 2 and 8 and S2 Fig ) . This finding is generally consistent with our previous conclusions ( 5 ) . The lack of correlation between the Tier-1 and autologous Tier-2 NAb responses , both temporally within a rabbit and among a group of rabbits , should now be considered in light of what we are learning about the epitopes involved . Thus , Tier-1 NAb responses are dominated by antibodies to V3 , a region of Env that contains a cluster of continuous , peptide epitopes that are relatively devoid of any shielding glycans . In contrast , we now show that the autologous Tier-2 NAb epitopes involve holes in the glycan shield that are presumably influenced by the surrounding glycans . We suggest that antibody responses to these different categories of neutralization epitopes may not be elicited or boosted in the same way . If so , what has been learned from earlier generations of Env glycoproteins that induce predominantly V3-NAbs would not be informative about how NAb responses to Tier-2 epitopes are best elicited; different immunization regimens and/or adjuvants may be needed . The trimer genotype is now emerging as an additional important variable , in this and in other experiments that we have been conducting . For example , we observed here that , in the heterologous boost context , the CZA97 trimer was more immunogenic than its DU422 counterpart . Moreover , a few rabbits were non-responders to CZA97 , DU422 and B41 trimers but none to BG505 , through influences that we are yet to understand . We are now generating and testing hypotheses about what specific features of individual trimers , and the corresponding viruses , most influence the induction of autologous Tier-2 NAb responses . Whether the same characteristics will also affect how different trimers cross boost existing NAb responses , or induce heterologous responses , will also need to be understood . Host factors are also important . Within a group of rabbits , there can be a wide range ( 2–3 orders of magnitude ) of autologous NAb titers to the same trimer , and some animals do not respond at all . Whether an animal responds or not to a particular trimer immunogen by generating autologous Tier-2 NAbs , and the magnitude of any response , may be influenced by the genetics of both the antibody repertoire and of factors that affect the development of humoral immunity more generally . Whether different genetic factors affect the generation of Tier-1 and autologous Tier-2 NAb responses should also be considered in light of the different characteristics of the relevant epitopes . An influence of host genetic factors will not , of course , be unique to rabbits , although the identities of the factors involved may be species-dependent . This series of experiments was performed in rabbits , because the strong and consistent autologous NAb responses to the BG505 and B41 trimers in this species allow endpoints to be determined [5] . In contrast , BG505 trimers are only moderately immunogenic in macaques [5] and do not induce any autologous NAb response in mice that can be quantified [54] . Guinea pigs , however , do respond to BG505 trimers about as well as rabbits [51] . Which species provide the best models for predicting human responses to these trimers will be better understood as key immunological parameters become identified , and if and when data from human studies allow a retrospective comparison . We are now using the techniques reported here to map the targets of the autologous NAbs induced by BG505 SOSIP trimers in responding macaques and guinea pigs and determine whether they are different from those we find in the rabbit . Based on the current assumption that the macaque is the most likely of these species to mimic the human response , these ongoing studies may be of considerable value for identifying which , if any , small animal model is suitable for further testing of trimer immunogens . In conclusion , the observations and inferences from this complex set of experiments may be a valuable guide on how best to design and use trimer-based immunogens in the quest to induce bNAb responses in humans . The limitations of the study , the restricted statistical power of small groups , and perhaps other issues inherent to the choice of model species , must be recognized . Nevertheless , the present results suggest that the induction of sufficient neutralization breadth will probably not be achieved through current regimens simply by combining a few native-like trimers into a cocktail without further advances in immunogen delivery methods . The combination approach may , however , be useful for boosting responses primed by germline bNAb-targeting trimers , particularly if our understanding of genotype-dependent influences on immunogenicity advances . Knowledge of the nature of the epitopes responsible for neutralization of Tier-2 viruses is likely to aid the design of immunogens intended to convert narrow-specificity NAb responses that , perhaps , evade glycans into broader ones that , in some cases , accommodate their presence . Thus , glycan holes may be the initial targets for several bNAbs in their ontogeny in the infected human [55–57] and the continued removal and insertion of glycans at particular positions during natural infection may eventually lead to the emergence of broader responses . If so , better understanding the characteristics of epitopes and immunogenic sites , both natural and engineered variants , would be useful . Among other strategies to consider are the targeted opening or closing of glycan holes to try to alter the immunogenicity of trimers beneficially , as well as more sophisticated , structure-guided approaches towards accommodating glycans that abut key epitopes . | Native-like SOSIP trimers are a platform for development of immunogens aimed at inducing broadly neutralizing antibodies and , hence , a possible vaccine against HIV-1 infection . No previous study has reported on immune responses to more than one such trimer . Here , we assess how rabbits respond to immunization with two or three different trimers , based on virus sequences from HIV-1 clades A , B and C , to gain insights into whether each is immunogenic under various regimens . We find that autologous Tier-2 neutralizing antibody responses can be raised against each trimer immunogen , whether they are delivered simultaneously or sequentially . We also observed some boosting of the neutralization response to the first trimer when a second trimer was administered later . Cross-reactive neutralizing antibodies were seen but only sporadically . We also found that the key immunogenic epitopes on the Env trimers involved holes in the glycan shield , which normally protects the virus from antibody binding . These various findings will guide the design of future experiments in animals and eventually in humans . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"enzyme-linked",
"immunoassays",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"microbiology",
"vertebrates",
"rabbits",
"cloning",
"animals",
"mammals",
"retroviruses",
"animal",
... | 2016 | Sequential and Simultaneous Immunization of Rabbits with HIV-1 Envelope Glycoprotein SOSIP.664 Trimers from Clades A, B and C |
In mammals , mothers are the primary caregiver , programmed , in part , by hormones produced during pregnancy . High-quality maternal care is essential for the survival and lifelong health of offspring . We previously showed that the paternally silenced imprinted gene pleckstrin homology-like domain family A member 2 ( Phlda2 ) functions to negatively regulate a single lineage in the mouse placenta called the spongiotrophoblast , a major source of hormones in pregnancy . Consequently , the offspring’s Phlda2 gene dosage may influence the quality of care provided by the mother . Here , we show that wild-type ( WT ) female mice exposed to offspring with three different doses of the maternally expressed Phlda2 gene—two active alleles , one active allele ( the extant state ) , and loss of function—show changes in the maternal hypothalamus and hippocampus during pregnancy , regions important for maternal-care behaviour . After birth , WT dams exposed in utero to offspring with the highest Phlda2 dose exhibit decreased nursing and grooming of pups and increased focus on nest building . Conversely , ‘paternalised’ dams , exposed to the lowest Phlda2 dose , showed increased nurturing of their pups , increased self-directed behaviour , and a decreased focus on nest building , behaviour that was robustly maintained in the absence of genetically modified pups . This work raises the intriguing possibility that imprinting of Phlda2 contributed to increased maternal care during the evolution of mammals .
High-quality maternal care is vitally important for newborn survival and their behavioural and metabolic health later in life , evidenced by the catastrophic consequences when maternal care is poor or absent [1–3] ( S1 Table ) . Maternal care provision comes at the cost of the mother’s later reproductive fitness , whereas the paternal interest is best served by prolonged care of progeny exclusively by the dam . Imprinted genes , expressed from a single parental allele as a consequence of germline epigenetic events [4] , are thought to be a physical embodiment of this conflict between male and female mammals over maternal investment in offspring [5 , 6] . Essentially , ‘paternalisation’ ( silencing of paternal allele ) is proposed to secure higher maternal investment , while ‘maternalisation’ counteracts to protect maternal reproductive fitness [5] . While there are further refinements , alternative hypotheses , and much debate [6–8] , most theories are consistent with the existence of imprinted genes that influence maternal care provision . Disruption of the expression of three imprinted genes in the mother ( paternally expressed gene 1 [Peg1] , paternally expressed gene 3 [Peg3] , and type 3 deiodinase ) results in deficits in maternal care [9–11] , and we have recently shown that loss of Peg3 in the offspring increases maternal anxiety and decreases pup retrieval in wild-type ( WT ) mothers [12] . However , deficits in maternal care are relatively common and have been reported with a myriad of genetic modifications as well as exposures to adverse environments in pregnancy ( S1 Table ) . In contrast , examples of increased maternal care providing more compelling evidence for a purposeful phenomenon are exceptionally rare and almost invariably involve hormonal manipulations . The placenta is a foetally derived organ fundamental to pregnancy [13 , 14] . A number of hormones produced by , or dependent on , the placenta are implicated in the programming of maternal care during pregnancy [15] ( S1 Fig ) . Placental lactogens , produced by placental endocrine lineages in contact with the maternal circulation [16] , are known to accumulate in cerebrospinal fluid during pregnancy [17] and can stimulate maternal behaviour when infused directly into the medial preoptic area in nonpregnant rodents [18] . Placental lactogens are related to the pituitary hormone prolactin required for maternal behaviour [19 , 20] , and a subset bind and activate the prolactin receptor [21] also required for maternal care [22 , 23] . In addition , whereas not all the genes encoding steroidogenic enzymes are expressed in the mature mouse placenta ( S1 Fig ) , lactogenic hormones support the production of oestrogens and progesterone from the corpus luteum in rodents [24] , which act in concert in the rapid induction of maternal behaviour at term [25 , 26] . Together , these data strongly suggest that placental lactogens have a direct effect on maternal care behaviour . The neurotransmitter serotonin , important in maternal care [27] , is manufactured by the placenta , although thought to be directed at the foetal brain [28–31] . There is less evidence that dopamine [32–34] , oxytocin [35–37] , or vasopressin [38 , 39] could be influenced by placental genes ( S1 Fig ) . Using genetic mouse models , we identified several paternally silenced , maternally expressed imprinted genes that function to restrain the placental lineages that produce hormones thought to be important for maternal care behaviour in mice [40–42] . This led us to hypothesise that the silencing of one or more genes by the paternal genome may have acted to boost maternal care in mammals . The mouse pleckstrin homology-like domain family A member 2 ( Phlda2 ) gene negatively regulates the expansion of a single placental endocrine lineage called the spongiotrophoblast , a lineage that expresses placental lactogens , pregnancy-specific glycoproteins , and a number of other hormones that induce and maintain the physiological adaptations required for a successful pregnancy [16] , [41 , 43–46] . Phlda2 is expressed from the maternal allele primarily in the placenta [47 , 48] and encodes a pleckstrin homology ( PH ) domain–only protein [49] that functions to inhibit cell proliferation by repressing AKT activation [50] . Early in mouse placental development , Phlda2 supresses the proliferation of the spongiotrophoblast [51] . Doubling the Phlda2 gene dosage ( potentially resembling the ancient preimprinted state ) by means of a bacterial artificial chromosome ( BAC ) transgene reduces the contribution of the spongiotrophoblast lineage to the mature placenta by approximately 50% , whereas loss of expression by knockout ( KO ) of the maternal allele results in a 2-fold expansion of this lineage [41 , 52–54] . Phlda2 is consequently a rheostat for placental hormones expressed from , or dependent on , the spongiotrophoblast , with ‘paternalisation’ increasing their expression and ‘maternalisation’ decreasing expression . To ask whether expression of Phlda2 in the offspring influenced maternal behaviour , we exposed dams to three different doses of offspring Phlda2 , excluding the confounder of genetically modified dams by using recipient transfer of embryos exclusively into WT females . All experimental dams were generated by recipient transfer to control for any effect of this manipulation on gene expression , and litters of comparable size were used throughout . We observed changes in the maternal brain that preceded exposure to the pups and altered maternal behaviour after birth , consistent with our original hypothesis .
Before undertaking a behavioural assessment , we asked whether changes were present in the maternal brain during pregnancy . Preimplantation Phlda2+/+BACx1 ( BAC transgenic overexpression; 2x ) , Phlda2+/+ ( wild-type; 1x ) , and Phlda2−/+ ( loss of function; 0x ) embryos were transferred into WT recipient females to generate WT ( 2x ) ( WT dams carrying and caring for >60% Phlda2+/+BACx1 embryos ) , WT ( 1x ) ( WT dams carrying and caring for all WT embryos ) , and WT ( 0x ) ( WT dams carrying and caring for all Phlda2−/+ embryos ) dams , respectively ( Fig 1A ) . After normalising for the number of foetuses per dam across the three groups ( S2 Table ) , and consistent with our previous data from naturally mated litters [41 , 54] , foetal growth was reduced at E16 . 5 for both Phlda2+/+BACx1 ( 2x ) and Phlda2-/+ ( 0x ) foetuses carried by WT dams in comparison to fully WT foetuses carried by WT dams ( S3a Fig; S2 Table ) . Gene expression in the maternal hypothalamus ( onset , maintenance , and regulation of maternal behaviour ) and the hippocampus ( memory , learning , and responses to fear and stress ) [55] was assessed at embryonic day ( E ) 16 . 5 , 4 days before parturition at a time of maximal placental phenotype [41 , 54] . Heat maps were plotted for the genes that showed changes in expression levels at p < 0 . 05 . This resulted in a dendrogram grouping the WT ( 2x ) , WT ( 1x ) , and WT ( 0x ) gene profiles by foetal Phlda2 dosage ( Fig 1B ) . Two hundred ninety-six probes ( 0 . 79% ) were altered between WT ( 0x ) and WT ( 2x ) hypothalamus , and 1 , 333 probes ( 3 . 6% ) were altered between WT ( 0x ) and WT ( 2x ) hippocampus . Analysis of hypothalamic and hippocampal pathways identified disturbances in olfactory transduction pathways important for maternal care and G protein–coupled receptor ( GPCR ) pathways through which neuropeptides and hormones mediate their action ( Fig 1C and 1D; S3 Fig ) . Specific to the hippocampus were disturbances in the gonadotropin-releasing hormone signalling pathway , cell differentiation , and circadian entrainment ( Fig 1D ) . Together , these findings indicated a physiological response to the placental cues with potential for a behavioural response . Following these encouraging findings , a second cohort of dams was generated by the same protocol ( Fig 2A ) and allowed to deliver . As for the gene expression analysis , dams with small litters ( <6 ) were excluded to generate comparable litter sizes across the three groups ( F2 , 36 = 0 . 83 , p = 0 . 45; S2 Table; S2 Fig ) . Behaviour was assessed concurrently by the same team of researchers between postnatal day ( P ) 2 and P4 , when maternal behaviours are most intense in rodents [56 , 57] . To assess anxiety behaviour , we made use of the elevated plus maze ( EPM ) test on P2 ( Fig 2A ) . Duration spent within the open and closed zones ( F2 , 36 = 1 . 44 , p = 0 . 25 ) and frequency of entry into each arm ( F2 , 36 = 1 . 7 , p = 0 . 19 ) did not differ between cohorts . There was no difference in time spent in the most anxiogenic open zone ( F2 , 36 = 0 . 16 , p = 0 . 85 ) . The total distance moved within each zone was the same ( F2 , 36 = 1 . 58 , p = 0 . 22 ) , and there was no difference between the speed of travel of the dams across the different zones ( F2 , 36 = 2 . 87 , p = 0 . 069 ) . There were no differences in rearing ( duration or frequency ) , stretch-attend ( duration or frequency ) , or head-dips ( duration or frequency ) in any zone ( data in S2 Table ) . On P3 , dams were tested on the pup-retrieval task after 24-hour acclimatisation in Phenotyper cages . The time taken by dams to first sniff their pups did not vary with Phlda2 dosage ( F2 , 36 = 1 . 1 , p = 0 . 33; Fig 2C ) , indicating intact response to olfactory cues . However , the time taken to retrieve the first pup was different between the three groups ( F2 , 36 = 4 . 8 , p = 0 . 015; Fig 2C ) . WT ( 0x ) dams who carried and cared for offspring with the lowest expression of Phlda2 took 3 times longer to retrieve their first pup compared to either WT ( 1x ) dams ( p = 0 . 01 ) or WT ( 2x ) dams ( p = 0 . 01; additional data in S2 Table ) . Lactating dams are usually more active during the dark phase [58] , which was observed with the WT ( 1x ) dams exposed to the normal dose of Phlda2 ( p < 0 . 001; Fig 2D ) . In contrast , neither WT ( 0x ) ( p = 0 . 83 ) nor WT ( 2x ) ( p = 0 . 13 ) dams showed a difference in activity between light and dark phases . Although all three travelled similar distances over a 23-hour period ( 36 , 022 versus 39 , 191 versus 39 , 542; WT[2x] versus WT[1x] versus WT[0x]; F2 , 36 = 5 . 1 , p = 0 . 608 ) , WT ( 0x ) dams showed a tendency to be more active during the day , whereas WT ( 2x ) dams were less active at night ( Fig 2D ) . The time dams spent in the nest during the light phases was similar across the three cohorts ( F2 , 36 = 0 . 29 , p = 0 . 75 ) , but during the dark phase ( F1 , 25 = 5 . 8 , p = 0 . 024; Fig 2E ) , WT ( 0x ) dams spent more time in the nest compared to WT ( 1x ) dams ( p = 0 . 005; Fig 2E ) . Dams showed differences in nest building behaviour ( F2 , 36 = 4 . 3 , p = 0 . 022; Fig 2F ) . On P4 , WT ( 2x ) dams were more effective at making nests and placing at least 1 pup in the nest than either WT ( 1x ) ( p = 0 . 05 ) or WT ( 0x ) dams ( p = 0 . 03 ) . WT ( 0x ) dams were least effective in this task , with only 1 WT ( 0x ) dam building a nest and placing at least 1 pup within the nest during the 60-minute test . Consistent with success rate in this task , there were differences across the cohorts in time spent building nests ( F2 , 36 = 6 . 967 , p = 0 . 003; Fig 2G ) , with WT ( 0x ) dams spending less time building nests than either WT ( 1x ) ( p = 0 . 029 ) or WT ( 2x ) dams ( p = 0 . 002 ) . Isolation-induced ultrasonic vocalisations ( USVs ) are known to stimulate maternal behaviours such as nest building and pup retrieval [59 , 60] . We found no difference between Phlda2+/+ and Phlda2−/+ ( KO ) pup vocalisation at P2 or P4 that might explain delayed retrieval or the failure to build nests ( S4 Fig; data from naturally mated litters ) . Pup retrieval and nest building are classic tests used to assess maternal behaviour but are not necessarily indicators of enhanced maternal care . Dams normally divide their time between nursing and grooming their pups ( pup focused ) , maintaining themselves ( self-directed ) , and nondirected maternal behaviours ( protection of the young , nest building ) . During the nest building task , there is a conflict between these essential behaviours . Consistent with this conflict , during the task , there were differences in the duration and frequency of pup nursing ( F2 , 36 = 5 . 8 , p = 0 . 007 and F2 , 36 = 6 . 9 , p = 0 . 003; Fig 3A ) . Dams exposed to the lowest dose of Phlda2 ( 0x ) engaged in crouched nursing for longer than dams exposed to the highest dose ( WT[0x] versus WT[2x]; p = 0 . 005 and p = 0 . 003; Fig 3A ) . The number of crouched nursing events was greater for WT ( 0x ) females , as was the number of passive nursing events ( Fig 3A ) . Both the duration ( F2 , 36 = 4 . 39 , p = 0 . 02 ) and frequency ( F2 , 36 = 4 . 59 , p = 0 . 017 ) of pup grooming events was also different between the three groups ( Fig 3B ) . WT ( 0x ) dams spent 5 times longer grooming their pups ( p = 0 . 02 ) and engaged in twice the number of grooming events ( p = 0 . 012 ) than WT ( 2x ) dams . In addition to changes in pup-directed behaviour , dams differed in both the duration ( F2 , 36 = 6 . 61 , p = 0 . 004 ) and frequency ( F2 , 36 = 4 . 4 , p = 0 . 02 ) of self-grooming ( Fig 3C ) . WT ( 0x ) spent longer than either WT ( 1x ) ( p = 0 . 021 ) or WT ( 2x ) dams ( p = 0 . 003 ) in self-grooming and engaged in self-grooming more frequently ( p = 0 . 02 and p = 0 . 01 , respectively ) . WT ( 0x ) dams travelled further during the nest building task ( Fig 3D ) and visited the food zone more often during the nest building task than dams exposed to higher doses of Phlda2 ( Fig 3E ) . Additional data are in S2 Table . As the dosage of Phlda2 decreased in the offspring , dams spent more time caring for their pups and maintaining themselves and less time on the ‘housekeeping’ task . Pup growth can be regarded as a measure of maternal care correlating with time spent nursing and lactation . Phlda2+/+BACx1 and Phlda2−/+ foetuses were approximately 16% lighter than Phlda2+/+ foetuses at E16 . 5 ( S2 Fig ) . Pups were not weighed at birth to avoid disturbing the dams in this critical period , but by P7 , there was no difference in pup weight between the three cohorts ( F2 , 164 = 1 . 89 , p = 0 . 15; S2 Table ) . Adequate preweaning weight gain indicated that the altered maternal behaviour did not negatively impact pup welfare , at least in the short term . Our understanding of the function of Phlda2 in placental endocrine lineage development alongside alterations in the maternal brain that preceded exposure to the pups suggested the prenatal programming of the behavioural changes . To test this theory , WT dams exposed in utero to Phlda2−/+ ( 1x ) or Phlda2−/+ ( 0x ) embryos were provided with fully WT pups immediately after birth ( Fig 4A ) , followed by an assessment of behaviours previously shown to be altered in WT ( 0x ) dams . Delayed retrieval was not different between the two groups ( Fig 4B ) . WT ( 0x ) wt dams ( WT dams carrying all Phlda2−/+ embryos and caring for all WT pups ) were less effective at nest building than WT ( 1x ) wt dams ( WT dams carrying all WT embryos and caring for all WT pups from a different WT litter ) , despite a similar duration and number of events ( Fig 4C ) . Total number of nursing events was greater for WT ( 0x ) wt dams ( F1 , 30 = 4 . 33 , p = 0 . 046 ) , with arched nursing events showing the greatest difference between the two groups ( F1 , 30 = 8 . 58 , p = 0 . 006 ) ( Fig 4D ) . Both the duration and number of pup grooming events by WT ( 0x ) wt dams were increased relative to WT ( 1x ) wt dams ( F1 , 30 = 6 . 74 , p = 0 . 014 and F1 , 30 = 5 . 0 , p = 0 . 033; Fig 4E ) . Self-grooming behaviour was similar between the two groups ( Fig 4F ) . Additional data are in S3 Table . Despite the significant disruption imposed by removing their own newborns and replacing them with another dam’s newborns on the day of birth , WT ( 0x ) wt dams retained their enhanced nursing and grooming behaviours 4 days later . These data support prenatal programming of at least some aspects of the maternal care phenotype . Moreover , replicating our findings from the original experiment demonstrates that the phenotype is robustly reproducible .
Previous studies have shown that imprinted genes expressed in the dam are required for high-quality maternal care provision [9–11] . Loss of expression of imprinted genes in the offspring can also negatively influence maternal care postnatally through reduced demand for milk [61] or reduced communication by the pups [12] . Here , we show that the offspring can influence the maternal care they will receive even before they are born by modulating expression of the imprinted Phlda2 gene in their placenta . Critically , this is a rare example of an alteration that elicits enhanced care from the mother , reproducible over two independent studies . The direction of imprinting ( paternalisation ) supports our original hypothesis that silencing of Phlda2 in the male germline contributed to the evolution of enhanced maternal care in mammals . In mammals , male involvement in the care of young is rare , although not unknown [62] , because only females lactate , and internal fertilisation ensures maternity but not paternity . Genomic conflict over maternal care predicts that ‘paternalisation’ should result in increased maternal care , securing higher maternal investment [5] . However , this investment should not come at a substantial cost to maternal well-being , as pups are dependent on the mother until weaning . Consistent with this , we observed that ‘paternalised ( 0x ) ’ dams exposed to the lower Phlda2 dose spent more time nursing and grooming their pups and more time on self-maintenance behaviours than maternalised ( 2x ) dams . Indeed , it may be that paternal silencing of Phlda2 increases the fitness of the mothers , at least in the short term while caring for her litter . Importantly , despite removing newborn pups from the nest and replacing them with WT pups , paternalised ( 0x ) dams retained their enhanced nurturing behaviour in the absence of continued exposure to genetically modified offspring 4 days later . Not only does this finding support a prenatal mechanism , our ability to replicate findings at a different time and under considerably more disruptive circumstances demonstrates the robustness of the phenotype . In contrast , WT ( 2x ) dams exposed to the higher dose of Phlda2 in their offspring spent less time grooming and nursing their pups and less time on self-grooming . In most behaviours , WT ( 1x ) dams displayed either an intermediate phenotype or were similar to WT ( 2x ) dams . Changes in nurturing behaviour were subtle , and the decreased nurturing observed with WT ( 2x ) dams did not have an immediate impact on pup welfare . Following up the longer-term metabolic and behavioural outcomes for the offspring will be required to fully establish the consequences of decreased maternal care . Poor maternal care can be passed on [2 , 63 , 64] , and it will be equally interesting to ask whether the altered behaviour persists in subsequent pregnancies . Maternal nurturing was increased in response to the lower dose of Phlda2 , but WT ( 0x ) dams were slower to retrieve their first pup . Dams locate displaced pups through olfactory and auditory cues . Dams from the three groups were equally effective at locating and sniffing their pups , and Phlda2−/+ pups made normal USVs when separated from their mothers , arguing against a major deficit in either medium . Rat dams can selectively retrieve the best-developed pups of the litter [65] . Both Phlda2−/+ or Phlda2+/+BACx1 pups are growth restricted at birth [41 , 54] , and the delay in WT ( 0x ) wt dams retrieving the foster WT pups was not statistically significant . It is possible that the quality of the offspring is responsible for the observed delay in retrieval . WT ( 0x ) females were less focused on nest building , spending a shorter duration on this task , with fewer events and with only 1/13 females successfully building a nest and placing at least 1 pup inside . In contrast , WT ( 2x ) dams showed greater focus on nest building , spending longer on this task , with an increased number of events and 7/13 dams successfully completing the task , performing even better than WT ( 1x ) dams . When pups are displaced from the nest , dams normally rapidly return pups [66] , with both retrieval and nest building important for neonatal pup thermoregulation . Pup retrieval and nest building are classically used to assess maternal behaviour , but these behaviours per se are not necessarily an indicator of ‘good mothering’ . During nest building , there is a conflict between the time the dam spends on the nest and the time spent direct nurturing her pups and herself . Although more elaborate nest building and pup retrieval are necessary aspects of maternal behaviour , one interpretation is that paternalised ( 0x ) dams are better mothers , as they are more focused on direct nurturing . The most simple explanation for our findings is that alterations in priming of mammary development by placental lactogens drive increased or decreased milk availability , impacting nursing behaviour with altered nest building as a secondary effect . However , dams with lactation deficits can initiate normal maternal behaviour [67 , 68] . Moreover , changes in the dams’ hypothalamic and hippocampal transcriptomes were present 4 days before birth , indicating a prenatal component . The programming of maternal behaviour during pregnancy requires prolonged exposure to a number of hormones expressed from , or dependent on , the placenta . Phlda2 plays a key role in restricting the expansion of the spongiotrophoblast , which is the major endocrine cell type of the placenta [69] . Just 2-fold expression of Phlda2 prevents the expansion of this lineage by 50% , with a concurrent decrease in the expression of spongiotrophoblast-expressed hormones , while loss of expression results in a 200% increase in spongiotrophoblast and increased hormone expression [41] . Consistent with a direct action of hormones on the maternal brain , analysis of hypothalamic and hippocampal pathways identified alterations in GPCR pathways through which neuropeptides and hormones mediate their action . Olfactory transduction pathways important for maternal care [15 , 70 , 71] were altered in both the maternal hippocampus and hypothalamus . Olfaction is important for pup recognition connected to the emotional system regulating social motor behaviour , which is initiated by odour , and the way in which positive signals of rewards are valued , which are important for mother–pup bonding . The hippocampal olfactory circuit is linked to odour-guided learning and memory , potentially also influencing mother–pup bonding . The gonadotropin-releasing hormone signalling pathway altered in the hippocampus has been implicated in maternal care and is known to respond to prolactin via this receptor [72 , 73] . Placental lactogens have previously been implicated in the induction of maternal care [18] , potentially via an interaction with the prolactin receptor [22 , 23] , expressed in both the maternal hippocampus and hypothalamus [74] . Together , these data suggest changes in the expression of one or more placental lactogens contribute to the changes in behaviour we observe in our dams . In mice , there are 22 placenta-specific , Prl-related genes expressed from the trophoblast giant cells , spongiotrophoblast , and glycogen cell lineages [16 , 75] . Of these , Prl3d1 ( PL-I ) and Prl3b1 ( PL-II ) have been shown formally to bind to the Prl receptor [21] . Prl3b1 is the dominant form during the second half of pregnancy [76] , and expression of this hormone is significantly affected by loss and gain of the spongiotrophoblast mediated by Phlda2 [41 , 53] . However , while it seems likely that Phlda2 acts , at least in part , on maternal behaviour via regulating the production of Prl3b1 , given the complexity of maternal care , it is unlikely that Phlda2 functions via a single hormone or pathway . While we have not sought here to identify the specific hormone ( s ) that contributes to the prenatal programming of maternal care , we have shown that imprinted genes can overcome the rapid evolution of placental hormone gene families [77] by regulating the lineages that express these hormones rather than directly regulating individual hormones , at least in mice . The placenta is the most diverse organ in mammals [78 , 79] , and this phenomenon may not be conserved across species . We have previously reported that in human pregnancies , placental PHLDA2 expression is inversely correlated with maternal serum human placental lactogen [80] . It remains to be seen if this relationship applies to other eutherian mammals . In summary , our data provide further support for conflict related to pregnancy in mammals [81] and the prediction that placental hormones ( or genes that regulate placental hormones ) might be imprinted [82] . This is also a very rare , possibly unique , example of increased maternal care prenatally programmed by a single gene modification in the offspring . Phlda2 acquired an imprinted status during the transition from a marsupial to eutherian mode of reproduction [83] , raising the intriguing possibility that imprinting of Phlda2 contributed to increased maternal care during the evolution of mammals ( Fig 5 ) . As a final point , this study may also have important implications for human health . Elevated placental PHLDA2 is a common findings in foetal growth restriction [84] , which is associated with mood disorders in human pregnancies [85] . It will be important to know whether human mothers carrying infants with elevated placental PHLDA2 also show deficits in maternal behaviour .
Animal studies and breeding were approved by the University of Cardiff’s ethical committee , performed under a United Kingdom Home Office project license ( RMJ; PPL 3003134 ) , and abide by ARRIVE guidelines . At the end of the study , animals were euthanised as required by the UK Home Office . Mice were housed in a conventional unit on a 12-hour light–dark cycle with lights coming on at 07:00 hours , with a temperature range of 21 °C +/− 2 °C , with free access to tap water and standard chow . The Phlda2-targeted allele [52] and the Phlda2 BAC transgenic line [53] were maintained ( >12 generations ) and studied on the 129S2/SvHsd ( 129 ) strain background , with control embryos and recipient females generated from a concurrently maintained , fully WT 129 colony . Embryos and pups were genotyped as previously described [52 , 53] . To generate experimental dams , 5–7-week-old virgin female mice of the appropriate genotype ( Phlda2+/+ , Phlda2−/− , or Phlda2+/+BACx1 ) were superovulated by 5 units ( 100 μl ) intraperitoneal pregnant mare serum followed by 5 units ( 100 μl ) intraperitoneal human chorionic gonadotropin 44–48 hours later , and mated with either WT or Phlda2+/+BACx1 stud males . Embryos were flushed from oviducts in M2 media ( Sigma ) 2 days after a discernible plug on E1 . 5 and incubated at 37 °C under 5% CO2 for 1–2 hours in KSOM ( Specialty Media ) before 14–16 embryos were transferred via bilateral oviduct transfer into E0 . 5 pseudopregnant WT 129 females aged 5–7 weeks . Phlda2+/+BACx1 animals carry a single copy of a nonimprinted 85-kb transgene spanning the genomic Phlda2 locus [54] . Homozygosity was found to be unviable , and 2 heterozygous animals were mated to generate embryos with only litters with >60% transgenic pups used in the downstream analysis . Pregnant females were either killed at E16 . 5 or taken to term for the behavioural analysis . Day of delivery ( D0 ) and litter size were recorded . Dams with litter sizes between 6 and 12 live pups were used for analysis , and there were no significant differences in litter size within any experiment ( Data in S2 and S3 Tables ) . All experimental samples were generated by recipient transfer . Dams were weighed and euthanised at E16 . 5 by cervical dislocation; whole brains were rapidly removed , followed by fine dissection of the hypothalamus and hippocampus , which were rapidly frozen on dry ice . RNA was prepared by homogenisation in RNA-Bee ( Ams Biotechnology ) following manufacturer’s instructions , and RNA concentration was evaluated using a NanoDrop Spectrophotometer , with RNA concentration adjusted to approximately 1 μg/μl using 10 mM Tris ( pH 8 ) . Gene expression analysis was performed using the Affymetrix GeneChip Mouse Gene 2 . 0 ST Array . Microarray data analysis was performed using the R programming language . Linear Model for Microarray data Analysis ( LIMMA ) was used to identify differentially expressed ( DE ) genes between all pairwise comparisons across the sample groups to generate heat maps of changes in gene expression levels at a p value <0 . 05 . Pathway analysis was performed using Database for Annotation Visualization and Integrated Discovery ( DAVID ) and iPathwayGuide [86–88] . Impact analysis uses two types of evidence to define the most greatly affected pathways affected in a system: ( i ) the overrepresentation of DE genes in a given pathway and ( ii ) the perturbation of that pathway , computed by propagating the measured expression changes across the pathway topology . The first probability , pORA , expresses the probability of observing the number of DE genes in a given pathway that is greater than or equal to the one observed by random chance . The second probability , pAcc , was calculated based on the amount of total accumulation measured in each pathway . A perturbation factor was computed for each gene on the pathway . The two types of evidence , pORA and pAcc , were combined into one final pathway score by calculating a p value using Fisher’s method . This p value was then corrected for multiple comparisons using false discovery rate ( FDR ) or Bonferroni corrections . The EPM test was carried out on P2 when dams were in their home cage between 8:00 AM and 10:00 AM hr to evaluate anxiety levels . Each new dam was given 10–15 minutes to habituate to the test room before embarking on the test away from her pups . The EPM consisted of 2 opposite open arms and 2 opposite closed arms ( 19 cm × 8 cm × 15 cm; length × width × height ) elevated 1 metre above ground . The middle section that allows the animal to transit from arm to arm consisted of a square with dimensions of 12 × 12 cm . Testing was carried out under low light conditions ( 230 lux on open sections ) . Each mouse was placed in the centre of the maze , and the amount of time spent in each arm was recorded automatically by EthoVision XT 8 . 0 video tracking software ( Noldus Information Technology , Netherlands ) over 5 minutes . Frequency and duration of rearing , stretch-attend , and head-dips over the end or side of the open arms were recorded . Pup retrieval and nest building took place on P3 between 8:00 AM and 10:00 AM after a 24-hour habitation period in the Phenotyper cage ( Noldus , Netherlands ) . Cages were equipped with cameras in the roof for video recording , and activity was tracked using Ethovision XT software ( Noldus Information Technology , Netherlands ) . The dam was removed from her cage , and her pups were removed from the nest . Pups were placed at the opposite end of the cage . The female was returned to her cage in the opposite corner to the pups and nest , and the latency to sniff and then retrieve the first pup was recorded . Video recording was continued for the following 23 hours before the nest building task , and videos were analysed by 2 researchers blind to the experimental groups . The quality of the nest building was scored on P4 between 8:00 AM and 10:00 AM . The Phenotyper recording was paused 23 hours into the recording . Dams , pups , and nests were removed to the original home cage . A cardboard tube ( International Product Supplies , UK ) and 30 x 30 cm strips of tissue paper were placed in the Phenotyper cage , pups were placed next to the new bedding material , dams were returned to the cage , and recording resumed for 1 hour . Nest quality and presence of pups in the nest were scored from videos , using a simple scale of no nest , nest built but no pups , or nest built and pups inside the nest , by 2 researchers blind to the genotype of the pups . Distance moved during this trial and visits to virtual ‘zones’ were calculated automatically using Ethovision XT software . Behaviours scored were total , crouched , arched , and passive nursing , grooming of pups , contacts with pups , self-grooming , and visits to food hopper or water zone . WT ( 1x ) and WT ( 0x ) dams were generated by recipient transfer by the same protocol used to generate the original groups . Dams were checked every 6–8 hours from E19 . 5 for the presence of pups . When newly born pups were found at the same time in both models , WT pups were removed from their dams , and 6 WT pups were either fostered to a WT ( 0x ) dam or to a different WT ( 1x ) dam ( control ) . All litters were then left undisturbed in Phenotypers until testing from P3 . USV was performed on fully WT pups and fully Phlda2−/− pups on P2 in a separate cohort generated by natural mating . The dam’s litter was removed from the home cage and placed in a separate home cage . USVs made by the pups were recorded for 180 seconds using Avisoft-UltraSoundGate 116Hb ( Avisoft Bioacoustics e . K . , Germany ) . The experimental data for the all the behavioural tests were analysed using SPSS Version 23 ( SPSS , United States of America ) unless otherwise stated . All behavioural data were presented as mean values with the standard error of the mean ( ± SEM ) also displayed . Each behavioural test was analysed separately according to genotype , followed by any appropriate comparisons between genotypes or across genotypes . This was done using separate ANOVAs for various between-subject factors of genotype and within-subject factors . | Female mammals are primed during pregnancy for their new role as a mother caring for their newborn . Indirect evidence suggests that this behaviour is , in part , instructed by hormones produced by the foetally derived placenta . We previously reported that the Phlda2 gene controls the size of the placental endocrine compartment that produces hormones . Phlda2 is subject to the remarkable epigenetic process called genomic imprinting , in which one parental allele is switched off . In the case of Phlda2 , this is the paternal allele . This raises the intriguing possibility that the father’s genome influences the quality of care that offspring receive from their mothers . Here , we show via genetic manipulation of pup embryos that female mice pregnant with pups carrying the highest dose of Phlda2 favour nest building over caring for their young or themselves . In contrast , dams exposed to the lowest Phlda2 dose while pregnant prioritise nurturing and grooming of their young and personal grooming . These changes in maternal focus suggest the possibility that imprinting of Phlda2 contributed to enhanced maternal care in mammals . This study also presents a unique example of how the genetic makeup of offspring can influence maternal behaviour . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"medicine",
"and",
"health",
"sciences",
"reproductive",
"system",
"maternal",
"health",
"obstetrics",
"and",
"gynecology",
"brain",
"vertebrates",
"animals",
"mammals",
"hormones",
"endocrine",
"physiology",
"reproductive",
"physiology",
"developmental",
"biology",
"wome... | 2018 | Maternal care boosted by paternal imprinting in mammals |
Accumulating evidence indicates that the capacity to integrate information in the brain is a prerequisite for consciousness . Integrated Information Theory ( IIT ) of consciousness provides a mathematical approach to quantifying the information integrated in a system , called integrated information , Φ . Integrated information is defined theoretically as the amount of information a system generates as a whole , above and beyond the amount of information its parts independently generate . IIT predicts that the amount of integrated information in the brain should reflect levels of consciousness . Empirical evaluation of this theory requires computing integrated information from neural data acquired from experiments , although difficulties with using the original measure Φ precludes such computations . Although some practical measures have been previously proposed , we found that these measures fail to satisfy the theoretical requirements as a measure of integrated information . Measures of integrated information should satisfy the lower and upper bounds as follows: The lower bound of integrated information should be 0 and is equal to 0 when the system does not generate information ( no information ) or when the system comprises independent parts ( no integration ) . The upper bound of integrated information is the amount of information generated by the whole system . Here we derive the novel practical measure Φ* by introducing a concept of mismatched decoding developed from information theory . We show that Φ* is properly bounded from below and above , as required , as a measure of integrated information . We derive the analytical expression of Φ* under the Gaussian assumption , which makes it readily applicable to experimental data . Our novel measure Φ* can generally be used as a measure of integrated information in research on consciousness , and also as a tool for network analysis on diverse areas of biology .
Although its neurobiological basis remains unclear , consciousness may be related to certain aspects of information processing [1 , 2] . In particular , Integrated Information Theory of consciousness ( IIT ) developed by Tononi and colleagues [2–9] predicts that the amount of information integrated among the components of a system , called integrated information Φ , is related to the level of consciousness of the system . The level of consciousness in the brain varies from a very high level , as in full wakefulness , to a very low level , as in deeply anesthetized states or dreamless sleep . When consciousness changes from high to low , IIT predicts that the amount of integrated information changes from high to low , accordingly . This prediction is indirectly supported by recent neuroimaging experiments that combine noninvasive magnetic stimulation of the brain ( transcranial magnetic stimulation , TMS ) with electrophysiological recordings of stimulation-evoked activity ( electroencephalography ) [10–14] . Such evidence implies that if there is a practical method to estimate the amount of integrated information from neural activities , we may be able to measure levels of consciousness using integrated information . IIT provides several versions of mathematical formulations to calculate integrated information [2–8] . Although the detailed mathematical formulations are different , the central philosophy of integrated information does not vary among different versions of IIT . Integrated information is mathematically defined as the amount of information generated by a system as a whole above and beyond the amount of information generated independently by its parts . If the parts are independent , no integrated information should exist . Despite its potential importance , the empirical calculation of integrated information is difficult . For example , one difficulty involves making an assumption when integrated information is calculated according to the informational relationship between the past and present states of a system . The distribution of the past states is assumed to maximize entropy , which is called the maximum entropy distribution . The assumption of the maximum entropy distribution severely limits the applicability of the original integrated information measure Φ as indicated by [15] . First , the concept of the maximum entropy distribution cannot be applied to a system that comprises elements whose states are continuous , because there is no unique maximum entropy distribution for continuous variables [15 , 16] . Second , information under the assumption of the maximum entropy distribution can be computed only when there is complete knowledge about the transition probability matrix that describes how the system transits between states . However , the transition probability matrix for actual neuronal systems is practically impossible to estimate . To overcome these problems , Barrett and Seth [15] proposed using the empirical distribution estimated from experimental data , thereby removing the requirement to rely on the assumption of the maximum entropy distribution . Although we believe that their approach does lead to practical computation of integrated information , we found that their proposed measures based on the empirical distribution [15] do not satisfy key theoretical requirements as a measure of integrated information . Two theoretical requirements should be satisfied as a measure of integrated information . First , the amount of integrated information should not be negative . Second , the amount of integrated information should never exceed information generated by the whole system . These theoretical requirements , which are satisfied by the original measure Φ , are required so that a measure of integrated information is interpretable in accordance with the original philosophy of integrated information . Here , we propose a novel practical measure of integrated information , Φ* , by introducing the concept of mismatched decoding developed from information theory [17–20] . Φ* represents the difference between “actual” and “hypothetical” mutual information between the past and present states of the system . The actual mutual information corresponds to the amount of information that can be extracted about the past states by knowing the present states ( or vice versa ) when the actual probability distribution of a system is used for decoding . In contrast , hypothetical mutual information corresponds to the amount of information that can be extracted about the past states by knowing the present states when the “mismatched” probability distribution is used for decoding where a system is partitioned into hypothetical independent parts . Decoding with a mismatched probability distribution is called mismatched decoding . Φ* quantifies the amount of loss of information caused by the mismatched decoding where interactions between the parts are ignored . We show here that Φ* satisfies the theoretical requirements as a measure of integrated information . Further , we derive the analytical expression of Φ* under the Gaussian assumption and make this measure feasible for practical computation . We also compute Φ* and the previously proposed measures in electrocorticogram ( ECoG ) data recorded in monkeys to demonstrate that the previous measures violate the theoretical requirements even in real brain recordings .
In IIT , information refers to intrinsic information as opposed to extrinsic information ( See S1 Text for details ) . Intrinsic information is quantified from the intrinsic perspective of a system itself and only depends on internal variables of the system . On the other hand , extrinsic information is quantified from the extrinsic perspective of an external observer and depends on external variables . For example , in neuroscience , extrinsic information is quantified as mutual information between neural states X and external stimuli S , I ( X;S ) [21–24] . In contrast , intrinsic information can be quantified by the mutual information between the past states Xt−τ and the present states Xt of the system , I ( Xt−τ;Xt ) . The mutual information , I ( Xt−τ;Xt ) , is expressed by I ( X t - τ ; X t ) = H ( X t - τ ) - H ( X t - τ | X t ) , ( 1 ) where H ( Xt−τ ) is the entropy of the past states and H ( Xt−τ|Xt ) is the conditional entropy of the past states given the present states . In IIT , the distribution of the past states is assumed to be the maximum entropy distribution so that the entropy of the past states is maximized , i . e . , the past states are maximally uncertain . We can interpret that intrinsic information , I ( Xt−τ;Xt ) , quantifies to what extent uncertainty of the past states can be reduced by knowing the present states from the system’s intrinsic point of view . IIT considers such quantity as the amount of information intrinsically generated by the system . Consider partitioning a system into m parts such as M1 , M2 , ⋯ , and Mm and computing the quantity of information that is integrated across the m parts of a system . As detailed in S1 Text , the measure of integrated information proposed in IIT 2 . 0 can be expressed as follows: Φ = I ( max X t - τ ; X t ) - ∑ i = 1 m I ( max M i t - τ ; M i t ) , ( 2 ) where the superscriptmax indicates that the distribution of the past states is the maximum entropy distribution . The first term of Eq 2 , I ( max Xt−τ;Xt ) , represents the mutual information between the past and present states in the whole system , and the second term represents the sum of the mutual information between the past and present states in the i-th part of the system I ( max M i t - τ ; M i t ) . Thus , Φ , the difference between them , gives the information generated by the whole system above and beyond the information generated independently by its parts . If the parts are independent , no extra information is generated , and the integrated information is 0 . We can rewrite Eq 2 in terms of entropy H as follows: Φ = ∑ i = 1 m H ( max M i t - τ | M i t ) - H ( max X t - τ | X t ) . ( 3 ) To derive the above expression , we use the fact that the entropy of the whole system H ( max Xt−τ ) equals the sum of the entropy of the subsystems ∑ i = 1 m H ( max M i t - τ ) when the maximum entropy distribution is assumed . To interpret a measure of integrated information as the “extra” information generated by a system as a whole above and beyond its parts , it should satisfy theoretical requirements , as follows: First , integrated information should not be negative because information independently generated by the parts should never exceed information generated by the whole . Integrated information should equal 0 when the amount of information generated by the whole system equals 0 ( no information ) or when the amount of information generated by the whole is equal to that generated by its parts ( no integration ) . Second , integrated information should not exceed the amount of information generated by the whole system because the information generated by the parts should not be negative . In short , integrated information should be lower-bounded by 0 and upper-bounded by the information generated by the whole system . One can check the original measure Φ satisfies the lower and upper bounds . 0 ≤ Φ ≤ I ( max X t - τ ; X t ) . ( 4 ) As shown in S1 Text , Φ can be written as the Kullback-Leibler divergence . Thus , Φ is positive or equal to 0 . Further , as can be seen from Eq 2 , the upper bound of Φ is the mutual information in the entire system , because the sum of mutual information in the parts is larger than or equal to 0 . Although using an empirical distribution instead of the maximum entropy distribution makes integrated information feasible to calculate , it is still difficult to compute Φ* in a large system , because the summation over all possible states must be calculated . The number of all possible states grows exponentially with the size of the system and therefore , computational costs for computing Φ* also grow exponentially . Thus , for practical calculation of Φ* , we need to approximate Φ* in some way such as approximating the probability distribution of neural states using the Gaussian distribution [15] . Φ* can be analytically computed using the Gaussian approximation ( see Methods ) . The Gaussian approximation significantly reduces the computational costs and makes Φ* practically computable even in a large system . In this section , by considering two extreme cases , we demonstrate that the previously proposed measures ΦH and ΦI[15] do not satisfy either the lower or upper bound .
In this study , we consider the two theoretical requirements that a measure of integrated information should satisfy , as follows: The lower and upper bounds of integrated information should be 0 and the amount of information generated by the whole system , respectively . The theoretical requirements are naturally derived from the original philosophy of integrated information [3 , 6] , which states that integrated information is the information generated by a system as a whole above and beyond its parts . The original measure of integrated information Φ satisfies the theoretical requirements so that we can interpret a measure of integrated information according to the original philosophy . To derive a practical measure of integrated information that satisfies the required lower and upper bounds , we introduced a concept of mismatched decoding . We defined our measure of integrated information Φ* as the amount of information lost when a mismatched probability distribution , where a system is partitioned into “independent” parts , is used for decoding instead of the actual probability distribution . In this framework , Φ* quantifies the amount of information loss associated with mismatched decoding where interactions between the parts of a system are ignored and therefore quantifies the amount of information integrated by the interactions . We show that Φ* satisfies the lower and upper bounds , that ΦI does not satisfy the lower bound , and that ΦH does not satisfy the upper bound . We consider Φ* a proper measure of integrated information that can be generally used for practical applications . Here , we briefly note a potential reason why the previous study [15] failed to identify these problems of ΦI and ΦH . Although they calculated their measures in small networks by using the autoregressive model in Eq 12 , they did not extensively vary the connectivity matrix A and the Gaussian noise E . In particular , they fixed the covariance of the Gaussian noise E to 0 . As we can clearly see in Fig 3 and S1 Fig , both connectivity strength a and the covariance of the noise c strongly affect the amount of integrated information . In particular , when the covariance of E is large , ΦI and ΦH violate the theoretical requirements . For future investigations of calculating integrated information in networks described by autoregressive model , we should note that it is very important to take account of not only the effects of connectivity matrix A but also the effects of covariance of E on the amount of integrated information . The basic concept of Integrated Information Theory ( IIT ) was tested by conducting empirical experiments , and the evidence accumulated supports the conclusion that when consciousness is lost , integration of information is lost [10–14] . In particular , Casali and colleagues [14] found that a complexity measure , motivated by IIT , successfully separates conscious awake states from various unconscious states due to deep sleep , anesthesia , and traumatic brain injuries . Although their measure is inspired by the concept of integrated information , it measures the complexity of averaged neural responses to one particular type of external perturbation ( e . g . a TMS pulse to a target region ) and does not directly measure integrated information . There are few studies that directly estimate integrated information in the brain [27 , 28] using the measure introduced in IIT 1 . 0 [2] or ΦH . Our new measure of integrated information , Φ* , will contribute to experiments designed to test whether integrated information is a key to distinguishing conscious states from unconscious states [29–31] . We considered the measure of integrated information proposed in IIT 2 . 0 [3 , 6] , because its computations are feasible . There are several updates in the latest version , IIT 3 . 0 [8] . In IIT 2 . 0 , integrated information is quantified by measuring how the distribution of the past states differs when a present state is given ( see S1 Text for details ) whereas in IIT 3 . 0 , it is quantified by measuring how the distribution of the past and future states differs when a present state is given . In other words , IIT 2 . 0 considers only the information flow from the present to the past while IIT 3 . 0 additionally considers the information flow from the present to the future . Our measure Φ* does not asymmetrically quantify integrated information from the present to the past or from the present to the future , because the mutual information is a symmetric measure for the time points t − τ and t . An unanswered question is how integrated information should be practically calculated taking account of the both directions of information flow , using an empirical distribution . An unresolved difficulty that impedes practical calculation of integrated information is how to partition a system . In the present study , we considered only the quantification of integrated information when a partition of a system is given . IIT requires that integrated information should be quantified using the partition where information is least integrated , called the minimum information partition ( MIP ) [3 , 6] . To find the MIP , every possible partition must be examined , yet the number of possible partitions grows exponentially with the size of the system . One way to work around this difficulty would be to develop optimization algorithms to quickly find a partition that well approximates the MIP . Besides the practical problem of finding the MIP , there remains a theoretical problem of how to compare integrated information across different partitions . Integrated information increases as the number of parts gets larger , because more information is lost by partitioning the system . Further , integrated information is expected to be larger in a symmetric partition where a system is partitioned into two parts of equal size than in an asymmetric partition . IIT 2 . 0 [6] proposes a normalization factor , which considers these issues . However , there might be other possible ways to perform normalization . It is unclear whether there is a reasonable theoretical foundation that adjudicates the best normalization scheme . Moreover , it is unclear if the normalization factor , which is proposed for systems whose states are represented by discrete variables , is appropriate for systems whose states are represented by continuous variables . The normalization factor , which is based on the entropies of the parts of a system , can be negative because entropy can be negative for continuous variables . Thus , we need a different normalization factor when we deal with continuous variables . Further investigations are required to resolve the practical and theoretical issues related to the MIP . Although we derived Φ* , because we were motivated by IIT and its potential relevance to consciousness , Φ* has unique meaning from the perspective of information theory , which is independent of IIT . Thus , it can be applied to research fields other than research on consciousness [32] . Φ* quantifies the loss of information when interactions or connections between the units in a system are ignored . Thus , Φ* is expected to be related to connectivity measures such as Granger causality [33] or transfer entropy [34] . It will be interesting to clarify mathematical relationships between Φ* and the other connectivity measures . We expect that information geometry [25 , 26 , 35 , 36] plays an important role for studying the properties of these quantities . Here , we indicate only an apparent difference between them as follows: Φ* intends to measure global integrations in a system as a whole , while traditional bivariate measures such as Granger causality or transfer entropy intends to measure local interactions between elements of the system . Consider that we divide a system into parts A , B , and C . Using integrated information , our goal is to quantify the information integrated among A , B , and C as a whole . In contrast , what we quantify using Granger causality or transfer entropy is the influence of A on B , B on C , C on A and the reverse . It is not obvious how a measure of global interactions in the whole system should be defined and derived theoretically from measures of the local interactions . As an example , one possibility is simply summing up all local interactions and considering the sum as a global measure [37] . Yet , more research is required to determine whether such an approach is a valid method to define global interactions [36] . Φ* , in contrast , is not derived from the local interaction measures but is derived directly by comparing the total mutual information in the whole system with hypothetical mutual information when the system is assumed to be partitioned into independent parts . Thus , the interpretation of Φ* is straightforward from an information theoretical viewpoint . Our measure , which we consider a measure of the global interaction , may provide new insights into diverse research subjects as a novel tool for network analysis .
The amount of information for mismatched decoding can be evaluated using the following equation , I * ( X t - τ ; X t ) = - ∑ X t p ( X t ) log ∑ X t - τ p ( X t - τ ) q ( X t | X t - τ ) β + ∑ X t - τ , X t p ( X t - τ , X t ) log q ( X t | X t - τ ) β , ( 20 ) where β is the value that maximizes I* . The maximization of I* with respect to β is performed by differentiating I* and solving the equation , dI* ( β ) /dβ = 0 . In general , the solution of the equation can be found using the standard gradient ascent method , because I* is a convex function with respect to β[17 , 18] . For comparison , the mutual information is given by I ( X t - τ ; X t ) = - ∑ X t p ( X t ) log p ( X t ) + ∑ X t - τ , X t p ( X t - τ , X t ) log p ( X t | X t - τ ) . ( 21 ) If a mismatched probability distribution q ( Xt|Xt−τ ) is replaced by the actual distribution p ( Xt|Xt−τ ) in Eq 20 , the derivative of I* becomes 0 when β = 1 . By substituting q = p and β = 1 into Eq 20 , one can check that I* is equal to I in Eq 21 , as it should be . The amount of information for mismatched decoding , I* , was first derived in the field of information theory as an extension of the mutual information in the case of mismatched decoding [17] . I* was first introduced into neuroscience in [18] and was first applied to the analysis of neural data by [19] . However , I* in the prior neuroscience application [18 , 19] was quantified between stimuli and neural states , not between the past and present states of a system , as described in the present study . Assume that the probability distribution of neural states X is the Gaussian distribution , p ( X ) = 1 ( 2 π ) N | Σ ( X ) | 1 / 2 exp - 1 2 ( X - X ¯ ) T Σ ( X ) - 1 ( X - X ¯ ) . ( 22 ) where N is the number of variables in X , X ¯ is the mean value of X , and Σ ( X ) is the covariance matrix of X . The Gaussian assumption allows us to analytically compute Φ* , which substantially reduces the costs for computing Φ* . When Xt−τ and Xt are both multivariate Gaussian variables , the mutual information between Xt−τ and Xt , I ( Xt−τ;Xt ) , can be analytically computed as I ( X t - τ ; X t ) = 1 2 log | Σ ( X t - τ ) | | Σ ( X t - τ | X t ) | , ( 23 ) where Σ ( Xt−τ|Xt ) is the covariance matrix of the conditional distribution , p ( Xt−τ|Xt ) , which is expressed as Σ ( X t - τ | X t ) = Σ ( X t - τ ) - Σ ( X t - τ , X t ) Σ ( X t ) - 1 Σ ( X t - τ , X t ) T , ( 24 ) where Σ ( Xt−τ , Xt ) is the cross covariance matrix between Xt−τ and Xt , whose element Σ ( Xt−τ , Xt ) ij is given by cov ( X i t - τ , X j t ) . Similarly , we can obtain the analytical expression of I* as follows: I * ( β ) = 1 2 TrΣ ( X t ) R + 1 2 log | Q | | Σ ( X t - τ ) | - β N 2 , ( 25 ) where Tr stands for trace . Q and R are given by Q = Σ ( X t - τ ) - 1 + β Σ D ( X t - τ ) - 1 Σ D ( X t , X t - τ ) T Σ D ( X t | X t - τ ) - 1 Σ D ( X t , X t - τ ) Σ D ( X t - τ ) - 1 , ( 26 ) R = β Σ D ( X t | X t - τ ) - 1 - β 2 Σ D ( X t | X t - τ ) - 1 T Σ D ( X t , X t - τ ) Σ D ( X t - τ ) - 1 Q - 1 Σ D ( X t - τ ) - 1 Σ D ( X t , X t - τ ) T Σ D ( X t | X t - τ ) - 1 , ( 27 ) where ΣD ( Xt−τ ) , ΣD ( Xt , Xt−τ ) and ΣD ( Xt|Xt−τ ) are diagonal block matrices . Each block matrix is a covariance matrix of each part , Σ ( M i t - τ ) , Σ ( M i t , M i t - τ ) , and Σ ( M i t | M i t - τ ) where Mi is a subsystem . For example , ΣD ( Xt−τ ) is given by Σ D ( X t - τ ) = Σ ( M 1 t - τ ) Σ ( M 2 t - τ ) 0 0 ⋱ Σ ( M m t - τ ) . ( 28 ) The maximization of I* with respect to β is performed by solving the equation dI* ( β ) /dβ = 0 . The derivative of I* ( β ) with respect to β is given by d I * ( β ) d β = 1 2 TrΣ ( X t ) d R d β + 1 2 Tr Q - 1 d Q d β - N 2 , ( 29 ) where d R d β = Σ D ( X t | X t - τ ) - 1 - 2 β Σ D ( X t | X t - τ ) - 1 T Σ D ( X t , X t - τ ) Σ D ( X t - τ ) - 1 Q - 1 Σ D ( X t - τ ) - 1 Σ D ( X t , X t - τ ) T Σ D ( X t | X t - τ ) - 1 - β 2 Σ D ( X t | X t - τ ) - 1 T Σ D ( X t , X t - τ ) Σ D ( X t - τ ) - 1 d Q - 1 d β Σ D ( X t - τ ) - 1 Σ D ( X t , X t - τ ) T Σ D ( X t | X t - τ ) - 1 , ( 30 ) d Q d β = Σ D ( X t - τ ) - 1 Σ D ( X t , X t - τ ) T Σ D ( X t | X t - τ ) - 1 Σ D ( X t , X t - τ ) Σ D ( X t - τ ) - 1 , ( 31 ) and d Q - 1 d β = - Q - 1 d Q d β Q - 1 , ( 32 ) = - Q - 1 Σ D ( X t - τ ) - 1 Σ D ( X t , X t - τ ) T Σ D ( X t | X t - τ ) - 1 Σ D ( X t , X t - τ ) Σ D ( X t - τ ) - 1 Q - 1 . ( 33 ) Inspection of the above equations reveals that dI* ( β ) /dβ = 0 is a quadratic equation with respect to β . Thus , β can be analytically computed without resorting to numerical optimization such as gradient ascent . The detailed recording protocols were described in [38] . Here , we briefly describe the aspects of the protocols that are relevant for our analysis . We used customized multichannel ECoG electrode arrays . An array of ECoG electrodes was embedded in an insulating silicone sheet . The surface of the sheet was dimpled to expose the surface of ECoG electrodes with the diameter of 1 mm . The electrodes were made of platinum discs , and inter-electrode distance was 5 mm . We implanted 128 ECoG electrodes in the subdural space in four adult macaque monkeys . The ECoG electrodes covered the left hemisphere over the frontal , parietal , temporal , and occipital lobes . ECoG signal was recorded at a sampling rate of 1 kHz . All experimental and surgical procedures were performed in accordance with the protocols approved by the RIKEN ethics committee . During the experiments , the monkeys were seated in a primate chair with both arms and head restrained . We analyzed the data recorded when the monkeys were awake . To remove line noise and reduce artifacts in the ECoG data , we computed bipolar re-referenced signals between two neighboring electrodes . We calculated integrated information Φ* using all the bipolar re-referenced signals ( 64 in total ) . We considered the simplest partition scheme , “atomic partition” [39] , in which the system is partitioned into its individual elements . For this data set , it meant that we computed Φ* assuming that all the 64 channels are independent . The atomic partition gives the upper bound of Φ* among all the possible partitions because it quantifies the amount of information loss when all the interactions in the system are ignored for decoding . We approximated the probability distributions of the continuous ECoG signals with the Gaussian distribution . Under the Gaussian assumption , we analytically computed Φ* by using the equations derived in Methods . We estimated the covariance matrices of the data with a time window of 2s and a time step of 2s . Then , we averaged the covariance matrices over 600s and used the average of the covariance matrices for computation of Φ* . | Integrated Information Theory ( IIT ) of consciousness attracts scientists who investigate consciousness owing to its explanatory and predictive powers for understanding the neural properties of consciousness . IIT predicts that the levels of consciousness are related to the quantity of information integrated in the brain , which is called integrated information Φ . Integrated information measures excess information generated by a system as a whole above and beyond the amount of information independently generated by its parts . Although IIT predictions are indirectly supported by numerous experiments , validation is required through quantifying integrated information directly from experimental neural data . Practical difficulties account for the absence of direct , quantitative support . To resolve these difficulties , several practical measures of integrated information have been proposed . However , we found that these measures do not satisfy the theoretical requirements of integrated information: First , integrated information should not be below 0; and second , integrated information should not exceed the quantity of information generated by the whole system . Here , we propose a novel practical measure of integrated information , designated as Φ* that satisfies these theoretical requirements by introducing the concept of mismatched decoding developed from information theory . Φ* creates the possibility of empirical and quantitative validations of IIT to gain novel insights into the neural basis of consciousness . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"vertebrates",
"random",
"variables",
"neuroscience",
"covariance",
"animals",
"mammals",
"primates",
"regression",
"analysis",
"cognitive",
"neuroscience",
"probability",
"distribution",
"mathematics",
"statistics",
"(mathematics)",
"thermodynamics",
"research",
"and",
"anal... | 2016 | Measuring Integrated Information from the Decoding Perspective |
Despite the clinical ubiquity of anesthesia , the molecular basis of anesthetic action is poorly understood . Amongst the many molecular targets proposed to contribute to anesthetic effects , the voltage gated sodium channels ( VGSCs ) should also be considered relevant , as they have been shown to be sensitive to all general anesthetics tested thus far . However , binding sites for VGSCs have not been identified . Moreover , the mechanism of inhibition is still largely unknown . The recently reported atomic structures of several members of the bacterial VGSC family offer the opportunity to shed light on the mechanism of action of anesthetics on these important ion channels . To this end , we have performed a molecular dynamics “flooding” simulation on a membrane-bound structural model of the archetypal bacterial VGSC , NaChBac in a closed pore conformation . This computation allowed us to identify binding sites and access pathways for the commonly used volatile general anesthetic , isoflurane . Three sites have been characterized with binding affinities in a physiologically relevant range . Interestingly , one of the most favorable sites is in the pore of the channel , suggesting that the binding sites of local and general anesthetics may overlap . Surprisingly , even though the activation gate of the channel is closed , and therefore the pore and the aqueous compartment at the intracellular side are disconnected , we observe binding of isoflurane in the central cavity . Several sampled association and dissociation events in the central cavity provide consistent support to the hypothesis that the “fenestrations” present in the membrane-embedded region of the channel act as the long-hypothesized hydrophobic drug access pathway .
Voltage gated sodium channels ( VGSCs ) , which mediate the upstroke of the action potential in most excitable tissues , are key targets of anesthetics . The binding site and molecular mechanism of action for local anesthetics have been well characterized in the last few decades , while the role of VGSCs in general anesthetic action is less well understood and both mechanisms continue to be studied . Thus far , the VGSC binding sites for general anesthetics have not been identified . Identifying binding sites and access pathways for volatile general anesthetics is key to understanding their mechanism of action and to designing new drugs . Sodium channels can be inhibited by a number of compounds , including toxins , quaternary ammonium compounds and local anesthetics [1] . When the pore is open , local anesthetics can enter from the intracellular side blocking conduction of ions . More recent work shows that general anesthetics also inhibit sodium channels , possibly through pore-blocking mechanisms [2]–[7] . Anesthetic block by both local and general anesthetics is preserved in the archetypal bacterial voltage-gated channel NaChBac [4] , [8] . Early observations [9] pointed to a local anesthetic binding site inside the channel , and indeed , in Nav 1 . 2 , the binding site suggested by mutagenesis is in the central cavity [10] . The effect of local anesthetics on NaChBac is consistent with a cavity-binding site , although the precise binding pocket is probably not the same as in Nav 1 . 2 [8] . It is thus conceivable that local and inhaled general anesthetic sites in the channel cavity in NaChBac overlap . The classical studies of charged local anesthetics and their analogs showed that blocking and unblocking seemed to require open channels . This led to the description of the “hydrophilic pathway” for drug access . However , additional experiments showed that hydrophobic local anesthetics could bind and unbind even when channels are closed [11]–[13] . This finding suggested an additional “hydrophobic pathway” that circumvents the closed gate . However , no structural correlate for this pathway has yet been conclusively identified . Although large mammalian VGSCs have remained resistant to structural characterization , the discovery of the smaller bacterial VGSCs has provided a tool to characterize the structural features of these important channels [14] , [15] . In general , the major structural domains resemble those found in the crystal structures of voltage-gated K+ channels ( Figure 1A ) . Transmembrane domains S1–S4 form the voltage-sensing domain ( VSD ) , which is connected to the pore domain through the S4–S5 linker . The pore domain , formed by domains S5 and S6 includes the pore-loops ( P-loops ) , which form the ion selectivity filter and may also constitute an inactivation gate [16] , [17] . On the intracellular side , four S6-domains form the bundle crossing that constitutes the activation gate [18]–[20] . The S4–S5 linkers form a cuff around the bundle crossing , such that in the presence of a depolarizing stimulus , conformational changes in the VSD widen this cuff , allowing the bundle crossing to open to permit conduction . A striking feature of the bacterial VGSC crystallized thus far is the presence of the so-called fenestrations [21]–[23] . These fenestrations provide a hydrophobic tunnel through the lipid-embedded portion of the channel to the central cavity , where the consensus local anesthetic site is located . In X-ray crystal structures , lipid tails occupy the fenestrations . These fenestrations may provide the hydrophobic drug access pathway that has been postulated for sodium channels since the 1970s [12] , [13] , [24] . The available atomic structure of a bacterial VGSC offers the first opportunity to address some of the most fundamental questions about the mechanism of volatile anesthetic action on VGSCs , namely what are the likely structural determinants of general anesthetic effects on VGSCs and could the fenestrations really provide access to the central cavity for small hydrophobic drugs ? To investigate these questions , we present the results of a molecular dynamics ( MD ) simulation study of the bacterial VGSC NaChBac embedded in a lipid bilayer in presence of the inhaled general anesthetic isoflurane . Our unbiased “flooding” technique [25] is a way to flexibly dock isoflurane molecules to high affinity sites of the membrane-bound protein , as well as to thereby suggest energetically favorable drug binding pathways .
To identify putative isoflurane binding sites , we performed MD “flooding” simulations [25] on a structural model of NaChBac [26] inserted in a hydrated lipid bilayer ( Figure 1 , A and B , Movie S1 ) . We considered a system comprising the membrane-bound channel along with a large number of drug molecules , initially located in the aqueous phase ( Figure 1B ) . As expected from the Meyer-Overton rule , we observed almost complete partitioning into the membrane within the first 100 ns of MD simulation ( Figure S1 ) . In the subsequent MD trajectory , we observed binding to several hydrophobic pockets on the protein surface with at most a single isoflurane molecule in the aqueous phase . Importantly , the conformation of NaChBac was not significantly affected on the time-scale of our MD simulation by the binding of isoflurane molecules , as inferred from the RMSD of the channel along the MD trajectory ( Figure S2 ) . The potentially large number of binding sites poses a challenge in identifying which of these may be pharmacologically relevant . To overcome this challenge we applied a two-fold strategy in which we first explored all the possible binding modes of isoflurane , and then used an established data-mining approach [27] to extract distinct configurations from our large data-set of molecular configurations . Specifically , we perform a cluster analysis on the positions of the centers of mass of the whole ensemble of isoflurane molecules and ranked the clustering solutions according to their local density . The rationale for this choice is to prioritize those regions that show continuous occupancy , which would suggest tighter binding of the drug molecule . This analysis revealed three major binding regions ( Figure 1C ) : 1 ) a region near the selectivity filter , termed the extracellular site , 2 ) a region near the S4–S5 linker , termed the linker site , and 3 ) a region within the cavity , termed the cavity site . Due to the protein's four-fold symmetry , a functional channel has four equivalent copies of each of the three putative binding sites . To investigate whether the three putative sites are occupied symmetrically in the tetramer , and to determine which amino acids line each site , we analyzed the interactions between isoflurane and all the residues within the sites in each subunit . In particular , we monitored the contacts between any atom of the drug molecule and any atom of a given amino acid for each instantaneous configuration after isoflurane partitioning has reached equilibrium . Residues were classified as non-interacting , “possibly” interacting , and “likely” interacting according to the number of configurations in which the contact was detected . For the extracellular site , all amino acids in all four subunits are in contact with the drug . In the cavity site as well as through the fenestrations , most residues are possibly or likely interacting in all four subunits . Intriguingly , the linker site is asymmetrically occupied in two adjacent sites . We used the clustering data and proximity time analyses , in combination with analysis of hydrogen-bonding-like interactions and mobility of the drug molecule in the sites , to structurally characterize the sites and elucidate the key determinants of binding . Furthermore , to assess the pharmacological relevance of these sites , we estimated the binding free energy of isoflurane for each site using well-established FEP methods [28] , allowing us to rank the relative free energy at each site . We also estimated binding affinities ( Kd ) and found that they were within a reasonable physiological range ( see Methods ) [29] .
All three identified sites are in regions of the protein that are predicted by mutagenesis to be critical to gating and conduction , and we hypothesize that some or all of them will play a role in inhibition of NaChBac by isoflurane . Since the time-scales involved in the molecular events relevant for gating are beyond normal MD simulation time-scales , we cannot directly probe the effect of drug interactions on protein conformation . However , based on knowledge of the mechanisms of gating , conduction , and inhibition by local anesthetics , we postulate possible mechanisms of drug action involving the binding sites suggested by our simulations . Isoflurane binding to the extracellular site positioned in the P-loops could affect the conformation of the selectivity filter , leading to inactivation through filter collapse [16] , [17] and consequent reduction in peak current . Isoflurane binding to the linker site , which sits between the cuff formed by the S4–S5 linkers and the S6 bundle crossing , could affect the conformation or the coupling between the cuff and the bundle crossing to halt gating [30]–[32] . Finally , isoflurane in the pore site could simply occlude the pore through classical local anesthetic mechanisms [1] , which has also been recently proposed as an inhibition mechanism of the GLIC channel by the same anesthetic [33]; this open channel block could be either purely steric or enhanced by the desolvation of the pore . Though theories of anesthetic action focus primarily on anesthetic-protein interactions , the role of lipids cannot be discounted as partitioning into the bilayer changes lipid properties [34] , [35] . For the structurally unrelated channel gramicidin the effect of halothane on channel's function has been shown to be dependent on lipid composition [36] . Channel interactions with lipids are important in regulating voltage sensing , gating and voltage-sensor-pore coupling [37]–[39] in voltage-gated ion channels . NaChBac's lipid sensitivity [40] suggests that appropriate lipid interactions regulate NaChBac function , and thus anesthetic partitioning could indirectly regulate the channel via disruption of lipid interactions . However , this simulation study is not designed to give realistic insights into lipid-anesthetic regulation for two reasons: first , because the lipid composition is not representative of a realistic bacterial or mammalian membrane and second , because the time-scale of the simulation is too short to observe lipid effects on channel structure and dynamics . This simulation of isoflurane binding sites and access pathways offers a number of experimentally testable hypotheses . While simulation shows that it is thermodynamically favorable for isoflurane to occupy these sites , it is possible that isoflurane occupation may have little or no physiological function . The functional relevance of each site can be ascertained by mutating each site individually to remove key determinants of isoflurane binding and evaluating whether NaChBac retains isoflurane sensitivity through electrophysiological assays [4] . The putative hydrogen bonding partners in the extracellular and linker sites ( Gln 186 , Asn 225 respectively ) are particularly good candidates for mutagenesis , as point mutations changing the polar character of the residue are likely to cause a large change in binding affinity at these sites . Mutation of the cavity is likely to prove difficult , as isoflurane is highly mobile and interactions are nonspecific – the cavity site may be better probed through classical use-dependence and trapping assays [1] . Most intriguingly , the relevance of the fenestrations as a drug access pathway could also be probed through mutations that occlude the fenestration pore at the extracellular side . Though no one has previously performed molecular dynamics simulation on voltage-gated sodium channels to identify anesthetic sites , numerous simulations have been done on ligand-gated ion channels ( LGICs ) due to the availability of the structures of bacterial homologues [41]–[43] . Simulation of anesthetic sites in LGICs , primarily based on the ELIC structure , have identified predominantly intersubunit sites , as well as a few intrasubunit sites [33] , [44]–[50] . These sites are also the most common types identified experimentally [51] , [52] , though pore sites have also been found [45] , [52]–[55] . In fact , all isoflurane sites identified in NaChBac are intersubunit sites , with the pore site also being intersubunit . Importantly , the observation that binding occurs at the intersubunit regions in both voltage- and ligand-gated channels suggests that anesthetics may impair the cooperative conformational transitions that have been shown to be crucial for the function of ion channels . Experimental verification of the aforementioned predictions will prove crucial to deepen our understanding of sodium channels modulation by anesthetics and will likely prompt further computational investigations . Indeed , despite providing a relatively accurate and detailed description of drug-channel binding events , our computational model is characterized by several limitations: ( i ) simulations were performed on a homology-based model of NaChBac; ( ii ) only one of the metastable structural conformations of the channel was probed for drug binding; ( iii ) lipid dependence of drug action was not addressed . Though the high degree of sequence identity with NavAb gives us confidence on the overall architecture of NaChBac , inaccuracies in the structure of the binding sites or major reorganizations of these pockets along the activation pathway can both potentially affect our results . Increased availability of experimental structures [22] , [56] and a detailed characterization of the conformational transitions entailed by the activation/deactivation cycle [57] will allow us to address these issues in future computational studies , as well as being able to begin to work with eukaryotic channels [58] , [59] .
Simulations were initialized using the theoretical model of NaChBac in a closed conformation obtained previously [26] . This homology model was built on the basis of the X-ray crystal structure of NavAb ( PDB ID: 3RVY ) , which was crystallized in a closed-pore conformation with all four voltage-sensing domains partially activated [57] . The model for the closed NaChBac channel , embedded in a fully hydrated lipid bilayer , was equilibrated by MD simulation . Specifically , the membrane is comprised of 1-palmytoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine ( POPC ) lipid molecules . Isoflurane molecules were initially placed in the aqueous phase with random positions and orientation . The membrane protein complex contained a total of ∼120 , 000 atoms , including NaChBac , 434 lipid molecules , 25310 water molecules , 236 ions in solution and 145 isoflurane molecules . The resulting initial aqueous concentration of isoflurane is 300 mM; the equilibrium aqueous concentration 0 . 9 mM . Two Na+ ions were initially placed in the channel selectivity filter , in agreement with a previous computational study of NavAb showing double occupancy of the filter by Na+ ions [60] . All charged amino acids were protonated using their respective pKas and the assumption that the solution was at pH 7 . MD trajectories were collected for 0 . 5 µs . MD simulation used the CHARMM22-CMAP force field with torsional cross-terms for the protein and CHARMM27 for the phospholipids [61] , [62] . A united-atom representation was adopted for the acyl chains of the POPC lipid molecules [63] . The water molecules were described using the TIP3P model [64] , [65] . Periodic boundary conditions were employed for all of the MD simulations and the electrostatic potential was evaluated using the particle-mesh Ewald method [66] . The lengths of all bonds containing hydrogen were constrained with the SHAKE/RATTLE algorithm [67] . The system was maintained at a temperature of 300 K and pressure of 1 atm using the Langevin thermostat and barostat methods as implemented in the MD code NAMD2 . 8 [68] ( Figure S1 ) . The rRESPA multiple time step method was employed , with a high frequency timestep of 2 . 0 fs and a low frequency time step of 4 . 0 fs . This computational setup has several limitations . First , the time-scales achievable by our MD simulation are too short to observe channel gating . Second , the small size of the lipid bilayer in the simulation box results in an excessive drug concentration in the bilayer . The final limitation stems from the use of a oversimplified POPC lipid bilayer . Bacterial sodium channel function is strongly lipid dependent [40] . However , NaChBac is functional in liposomes containing POPC and in mammalian cells , making POPC a reasonable choice for NaChBac simulations . To identify binding sites , we analyzed equilibrated configurations of the system and seeking regions characterized by high density of isoflurane . We first computed the center of mass ( COM ) of each isoflurane molecule in the MD trajectory frame . We then performed a cluster analysis on the resulting set of COM positions using the geometric distance between each pair of positions to build a proximity matrix . Partitioning of the set ( clustering ) was obtained using the Jarvis-Patrick algorithm [27] with a nearest-neighbor list of 8 and shared-neighbor threshold of 3 . We then ranked the clusters according to their average density . Here , we treated each COM position as an isotropic Gaussian density of width 1 Å , and summed over all the COM positions belonging to a given cluster . Integration of the density for each cluster was performed on a 3-dimensional grid with bin dimensions of 0 . 5×0 . 5×0 . 5 Å3 . The top three ranked clusters , comprising 15% of the total number of configurations , are discussed as putative binding regions . Free energy calculations were performed using the free energy perturbation ( FEP ) method . The binding free energies are calculated using the following scheme: DGbind = DGgas–prot− ( DGsolv+DGrstr ) . Here , DGbind is the free energy of binding isoflurane to NaChBac , DGvac->prot is the free energy of transferring an isoflurane from the gas phase to the binding site , DGsolv is the isoflurane solvation free energy , and DGrstr is a measure of the entropy cost associated with the reduction in volume from a 1 M solution ( V1M ) to the volume available at the binding site ( Vrstr ) , i . e . , DGrstr = RT ln ( Vrstr/V1M ) . For all reported binding energies , V1M is given by the volume associated with the flat-bottom spherical restraint applied to keep the isoflurane in the binding site during the interaction decoupling . Calculations were performed in NAMD 2 . 8 by varying the coupling parameter in steps of 0 . 025 at the ends and 0 . 05 in the middle . This approach has been successfully applied to binding of anesthetics to proteins including the binding of R- and S- isoflurane enantiomers to apoferritin [69] as well as the binding of isoflurane and propofol to the GLIC bacterial ion channel [33] . The estimated binding free energy for isoflurane was assuming that the only two relevant thermodynamic states are protein-bound isoflurane aqueous isoflurane . Therefore the water-lipid partitioning is not considered . However , since the affinity of isoflurane for a hydrophobic phase is significantly lower than our estimated affinities ( ΔG of transfer from water to dodecane is 3 . 0 kcal/mol ) , we expect the lipid partitioning to have a marginal effect on the apparent Kd . | The molecular mechanisms mediating the pharmacologically induced state of general anesthesia are , in general , poorly understood . Modulation of voltage gated sodium channels is thought to play a major role in anesthesia , as several members of this class of channels show a significant response to general anesthetics . However , the detailed mechanism of inhibition or potentiation of these channels is completely unknown . Recently , the structures of several members of the bacterial family became available , thereby offering the opportunity to shed light on some of these issues . We have performed molecular dynamics simulations on one of these bacterial voltage gated sodium channels , NaChBac , to identify binding sites and access pathways for the volatile general anesthetic isoflurane . We found that isoflurane , at physiologically relevant concentrations , binds the channel at three distinct sites . One site is in the pore of the channel , suggesting that isoflurane may hinder the permeant sodium ions . Surprisingly , we found that this binding site is accessible to the drug even when the pore and the aqueous compartment at the intracellular side are disconnected . In our simulations , the “fenestrations” present in the membrane-embedded region of the channel act as the long-hypothesized hydrophobic drug access pathway . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"anesthetic",
"mechanisms",
"medicine",
"biophysic",
"al",
"simulations",
"general",
"anesthesia",
"anesthesiology",
"biology",
"computational",
"biology",
"biophysics",
"simulations",
"biophysics"
] | 2013 | Exploring Volatile General Anesthetic Binding to a Closed Membrane-Bound Bacterial Voltage-Gated Sodium Channel via Computation |
We systematically determined which spectrotemporal modulations in speech are necessary for comprehension by human listeners . Speech comprehension has been shown to be robust to spectral and temporal degradations , but the specific relevance of particular degradations is arguable due to the complexity of the joint spectral and temporal information in the speech signal . We applied a novel modulation filtering technique to recorded sentences to restrict acoustic information quantitatively and to obtain a joint spectrotemporal modulation transfer function for speech comprehension , the speech MTF . For American English , the speech MTF showed the criticality of low modulation frequencies in both time and frequency . Comprehension was significantly impaired when temporal modulations <12 Hz or spectral modulations <4 cycles/kHz were removed . More specifically , the MTF was bandpass in temporal modulations and low-pass in spectral modulations: temporal modulations from 1 to 7 Hz and spectral modulations <1 cycles/kHz were the most important . We evaluated the importance of spectrotemporal modulations for vocal gender identification and found a different region of interest: removing spectral modulations between 3 and 7 cycles/kHz significantly increases gender misidentifications of female speakers . The determination of the speech MTF furnishes an additional method for producing speech signals with reduced bandwidth but high intelligibility . Such compression could be used for audio applications such as file compression or noise removal and for clinical applications such as signal processing for cochlear implants .
Human speech , like most animal vocalizations , is a complex signal whose amplitude envelope fluctuates timbrally in frequency and rhythmically in time . Horizontal cross-sections of the speech spectrogram as in Figure 1A describe the time-varying envelope for a particular frequency while vertical cross-sections at various time points show spectral contrasts , or variation in the spectral envelope shape ( Audio S1 ) . Indeed , the structure in the spectrogram of speech is not characterized by isolated spectrotemporal events but instead by sinusoidal patterns that extend in time and frequency over larger time windows and many frequency bands . It is well known that it is these patterns that carry important phonological information , such as syllable boundaries in the time domain , formant and pitch information in the spectral domain , and formant transitions in the spectrotemporal domain as a whole [1] . In order to quantify the power in these temporal and spectral modulations , the two-dimensional ( 2D ) Fourier transform of the spectrogram can be analyzed to obtain the modulation power spectrum ( MPS ) of speech [2] , [3] . In this study , first we repeated this analysis using a time-frequency representation that emphasized differences in formant structure and pitch structure . Then we used a novel filtering method to investigate which spectral and temporal modulation frequencies were the most important for speech intelligibility . In this manner we obtained the speech modulation transfer function ( speech MTF ) . We were then able to compare the speech MTF with the speech MPS in order to interpret the effect of modulation filters on perception of linguistic features of speech . Our study both complements and unifies previous speech perception experiments that have shown speech intelligibility to depend on both spectral and temporal modulation cues , but to be surprisingly robust to significant spectral or temporal degradations . Speech can be understood with either very coarse spectral information [4]–[8] or very coarse temporal information [9]–[11] . Our goal was to unify spectral and temporal degradation experiments by performing both types of manipulations in the same space , namely , the space of joint spectrotemporal modulations given by the speech MPS . The approach makes advances in the rigor of signal processing , in the specificity of the manipulations allowed , and in the comparison with speech signal statistics . First , the approach depicts visually and quantifies the concomitant effects that temporal manipulations have on the spectral structure of the signal , and that spectral filtering has on temporal structure . Second , the technique offers the possibility of notch filtering in the spectral modulation domain , something which has not been done before . Whereas degradation by low-pass filtering can reveal the minimum spectral or temporal resolution required for comprehension , notch filtering can distinguish more limited regions of spectrotemporal modulations that differ in levels of importance for comprehension . Third , the modulation filtering technique can be used to target specific joint spectral and temporal modulations . In this study , this advantage was exploited in a two-step filtering procedure to measure the effects of precise temporal and spectral degradations in the range of modulations most important for intelligibility . In this procedure , we first removed potentially redundant information in higher spectral and temporal modulations , and then we applied notch spectral or temporal filters within the remaining modulation space . Finally , we were able to compare the results of the speech filtering experiments to the MPS of speech , in order to make an initial characterization of the speech MTF in humans . As far as we know , this is the first such comparison using a linear frequency axis and a modulation transfer function obtained directly from speech intelligibility experiments . The resultant speech MTF could be used to design more optimal speech compression such as that required by cochlear implants . Neurophysiological research on animal perception of modulations inspired our study . While the cochlea and peripheral auditory neurons represent acoustic signals in a time-frequency decomposition ( a cochleogram ) , higher auditory neurons acquire novel response properties that are best described by tuning sensitivity to temporal amplitude modulations and spectral amplitude modulations ( reviewed in [12] and [13] ) . By designing human psychological experiments using the same representations used in neurophysiological research , we can begin to link brain mechanisms and human perception . Speech signals carry information about a speaker's emotion and identity in addition to the message content . As a final thrust of investigation , we tested whether modulations corresponding to acoustic features embedded in the speech signal enabled listeners to detect the gender of the speaker . Vocal gender identity has been shown to depend on some spectral features in common with , and some distinct from , the spectral features conferring speech intelligibility [14] , [15] .
The MPS of American English ( Figure 1C ) was calculated from a corpus of 100 sentences ( see Materials and Methods ) . This speech modulation spectrum shares key features observed in other natural sounds . As in all natural sounds , most of the power is found for low modulation frequencies and decays along the modulation axes following a power law [3] . Moreover , as typical of animal vocalizations , the MPS is not separable; most of the energy in high spectral modulations occurs only at low temporal modulation , and most high temporal modulation power is found at low spectral modulation [3] , [16] . This characteristic non-separability of the MPS is due to the fact that animal vocalizations contain two kinds of sounds: short sounds with little spectral structure but fast temporal changes ( contributing power along the x-axis at intermediate to high temporal frequencies ) , and slow sounds with rich spectral structure ( found along the y-axis at intermediate to high spectral frequencies ) . In normal speech , this grouping of sounds corresponds roughly to the vocalic ( slow sounds with spectral structure , produced with phonation ) and non-vocalic acoustic contrasts ( fast sounds with less spectral structure , produced without phonation ) . Animal vocalizations and human speech do have sound elements at intermediate spectrotemporal modulations , but these have less power ( or in other words are less frequent ) than expected from the power ( or average occurrence ) of spectral or temporal modulations taken separately , reflecting the non-separability of the MPS . An additional aspect of human speech is that modulations separate into three independent areas of energy along the axis of spectral modulation , at low temporal modulation ( Figures 1C and 2 ) . First , the triangular energy area at the lower spectral modulation frequencies corresponds to the coarse spectral amplitude fluctuations imposed by the upper vocal tract , namely the formants and formant transitions ( labeled in Figure 2B ) . The other two areas of spectral modulation energy , found at higher levels , correspond to the harmonic structure of vocalic phones produced by the glottal pulse; this energy diverges into two areas because of the difference in pitch between the low male voice ( highest spectral modulations ) and the higher female voice ( more intermediate spectral modulations ) . The lower register of the male voice produces higher spectral modulations because of the finer spacing of harmonics over that low fundamental . Equivalent pitches corresponding to the spectral modulations are labeled parenthetically in white on the y-axis of Figure 2 . The MPS can also be estimated from time-frequency representations that have a logarithmic frequency axis ( see Materials and Methods , and Figure S1 ) . Although log-frequency representations are better models of the auditory periphery , the linear-frequency representation is more useful for describing the harmonic structure present in sounds . For example , the separation of the spectral structure of vocalic phones into three regions is a property that is observed only in the linear frequency representation ( Figure S1 ) . Thus , in the speech MPS with linear frequency , not only do vocalic and non-vocalic sounds occupy different regions within the modulation space , but the spectral modulations for vocalic sounds corresponding to formants and male and female pitch occupy distinct regions . Also , human speech is symmetric between positive and negative temporal modulation frequencies , showing that there is equal power for upward frequency modulations ( Figure 1C , left quadrant ) and downward frequency modulations ( right quadrant ) . Our modulation filtering methodology allowed us not only to rigorously degrade speech within its spectral and temporal structure but also to relate the results from the degradation to acoustic features of the signal that are important for different percepts , as described above . Our psychophysical experiments are organized in three sections . We first report results from the two sets of modulation filters applied to the whole spectrotemporal modulation spectrum of speech—low-pass filters and notch filters—which indicated a subset of modulations that are critical for speech understanding , thereafter designated the “core” modulations . Subsequently , we report results from notch filters applied to sentences containing only core modulations , further refining our identification of crucial spectrotemporal modulations . Subjects reported the gender of the speakers of the notch-filtered sentences . Even though sentences having modulations restricted to the “core” ( Figures 4D and 5C ) were well comprehended , gender identification of the speakers of these sentences fell to 77% , where chance would be 50% . Of the gender errors , 91% occurred when the speaker was female . When modulations outside the core were spared , the notch-filter of spectral modulations between 3 and 7 cycles/kHz ( Figures 2C and 4F ) significantly decreased gender identification ( to 79% ) . Of these misidentified speakers , 95% were female . Both the core condition and this spectral notch condition lacked modulations in the 3–7 cycles/kHz range , where female speech has more power ( core spectral modulations are below 3 . 75 cycles/kHz ) . Male speech has more power shifted to higher spectral modulations ( 6–11 cycles/kHz ) . Thus , spectral modulation filters in the uniquely male range produced no significant decrease in gender identification . These results can be explained by the fact that whenever the filtered sentences lacked spectral modulation information unique to the female vocal register , subjects guessed that the speaker was male .
This study attempts to use dynamic properties of sound , rather than the traditionally stereotyped cues of acoustic phonetics , to refashion a parsimonious account of speech perception . Specifically , we used a novel filtering technique to remove spectrotemporal modulations from spoken sentences in order to isolate the acoustic properties critical for identifying linguistic features and for recognizing gender as a personal attribute of the voice . We first systematically degraded sentences by filtering specific temporal and spectral modulation frequencies and then examined the effect on the number of words comprehended . As we will discuss in more detail below , our study replicates , but also has several advantages over , previous degradations performed in the temporal or spectral domain alone . First , it provides a rigorous mathematical framework to precisely quantify what is being removed from the speech signal . Second , the framework unifies manipulations across two lines of research that are otherwise described orthogonally , namely AM and FM filtering [20] . Finally , we can make a direct connection between the acoustical space we manipulated and the information-bearing features of speech distributed in each particular region [21] . In the MPS of American English ( our prototype for human speech ) , the distinctive non-separable distribution of energy—namely , close to the x and y axes—corresponds roughly to a division between vocalic and non-vocalic sounds [3] , [16] , [21] . Non-vocalic phones in speech tend to be rapid and to have little spectral structure whereas vocalic phones are longer and have more spectral structure . Our categorization of speech sounds along the spectral and temporal axes of the MPS remains , of course , rather coarse . For example , voicing is associated with multiple acoustic properties and is only one of the linguistic features ( e . g . , place of articulation , manner , rounding ) needed in order to categorize phonemes [22] . A more detailed analysis of the MPS of individual phonemes or simple combinations of phonemes would further illustrate the usefulness of this methodology for speech analysis [21] . Within the spectral structure especially associated with vocalic sounds , we also found a clear separation between pitch information and phonetic information ( formants and formant transitions ) . The separation of the formant spectral frequencies from the pitch spectral frequencies had been described before and is one reason that cepstral analysis works well for the determination of formant frequencies [23] . In the discussion that follows , we will relate performance on the comprehension task to the acoustic features of speech we filtered from the MPS . Our low-pass spectral-modulation filtering experiment shows that speech intelligibility begins to degrade significantly when modulations below 4 cycles/kHz are removed . Not surprisingly , this definitive point corresponds to the upper extent of the area in the speech MPS occupied by energy associated with formants ( Figures 1 and 2 ) . The separation between formant peaks in English vowels is greater than 500 Hz ( or 2 cycles/kHz ) [24] , but finer spectral resolution ( up to 4 cycles/kHz ) would be beneficial to capture further the overall spectral shape of the formant filters and to detect formant transitions . There is a large literature on how spectral degradation affects speech comprehension , the most similar studies being those of Shannon and Dorman and colleagues [5] , [6] who have investigated speech intelligibility with very limited spectral resolution as one would experience with a cochlear implant . Shannon et al . [6] reported that speech intelligibility in a noise-free setting was fully preserved with spectral structure present in only 4 frequency channels below 4 kHz . These spectral bands would correspond in our implementation to a low-pass filter cutoff of approximately 1 cycles/kHz , or 1 . 7 cycles/octave , which is below the level needed for fully resolving formant spectral peaks and considerably below our cutoff of 4 cycles/kHz ( or 2 cycles/octave as shown in Figure S1 ) . However , when noise is present , Friesen et al . have shown that intelligibility increased with additional spectral channels [25] . In that study , for 0 dB SNR , 16 channels spaced below 6 kHz ( or approximately 3 . 75 cycles/kHz ) yielded significant additional comprehension over that of more degraded speech with fewer frequency channels . Our results are consistent with that result , and our study brings several additional insights to this analysis . First , the notch filtering experiments unequivocally demonstrate that the spectral MTF is truly low-pass . Removing lower ( or intermediate ) spectral modulations while preserving higher spectral modulations results in significant decreases in speech intelligibility . In other words , there does not appear to be information that is redundant between the spectral modulations below 4 cycles/kHz and higher spectral modulations ( further details of the notch filtering results are discussed below ) . Second , our comparison between the speech MTF and the speech MPS offers an obvious explanation for the critical spectral frequency cutoff: it corresponds to the modulation power boundary of formants and formant transitions . Finally , by examining how much modulation power was removed in the filtering operations , we can also say that the crucial modulation areas are not simply the ones with the higher power in the speech MPS . For example , the region of the core notch filter between 0 . 25 and 0 . 75 cycles/kHz contributes less to intelligibility than the 0–0 . 25 cycles/kHz area , although the former contains higher power . Humans appear to be particularly adept at detecting very low modulations in the spectral envelope and at using that information for speech intelligibility . In the temporal dimension , we showed that filtering the amplitude envelope of the speech signal below 12 Hz results in significant intelligibility deficits . Our results are similar to experiments in which the temporal envelope of speech was low-pass filtered or degraded . For Dutch , English and Japanese , it was shown that the region below 8 Hz is critical for speech comprehension [9]–[11] . This critical modulation is somewhere between the temporal modulations corresponding to the rate of syllables , at around 2 to 5 Hz [22] , and those corresponding to phonemes , at around 15–30 Hz [26]–[28] . In the MPS , we observe that frequencies below 10 Hz account for approximately 85% of all the spectrotemporal modulation power . Examining the temporal modulation spectrum as a function of frequency bands ( rather than as a function of spectral modulations , as shown in the MPS ) , Greenberg and Arai showed that the peak in power lies between 4 and 6 Hz [29] . By preserving frequencies below 8 Hz , one therefore retains most of the power in the temporal modulation spectrum . Qualitatively , the speech sounds that were heavily temporally filtered ( below 5 Hz ) sounded like reverberated speech , consistent with the observation that it is the higher temporal modulation frequencies that are affected in reverberant environments [30] . Interpreting which modulations proved crucial in the low-pass spectral or temporal filtering results is problematic because each relative lowering of the cutoff frequency removed increasingly more modulations . Furthermore , comparisons between low-pass cutoffs do not exclude the possibility that higher intelligibility could be achieved with isolated regions of the MPS . To obtain something akin to a modulation transfer function ( MTF ) for speech intelligibility , low-pass filtering manipulation must be complemented with high-pass filtering . Alternatively , a transfer function can be obtained directly from notch filtering experiments . We chose the latter approach and based the design of our notch filters on the results from the low-pass experiments . The combination of notch filtering and low-pass filtering also allowed us to examine areas in the speech MPS that carry redundant information . Two conclusions can be made from the notch filtering experiments . First , the results show a low-pass spectral tuning with most of the gain between 0 and 1 cycles/kHz , and a band-pass temporal tuning with most of the gain between 1 and 7 Hz . Second , the results show the high level of redundancy in the speech signal . The intelligibility of most notch-filtered stimuli remained excellent . This is even more remarkable considering that tests were done with a SNR of 2 dB . Redundancy is evident also when one examines the difference in results obtained from the low-pass and notch filters . Notably , the low-pass cutoff spectral frequency of 2 cycles/kHz significantly reduced performance as compared to the 4 cycles/kHz condition , whereas the 1–3 and 3–7 cycles/kHz notch filters straddling that range of modulations produced no significant decrease in performance . This discrepancy suggests that some of the information about formant structure in the 1–4 cycles/kHz range can also be found at higher spectral modulation frequencies . For this reason , we conducted the second notch filter experiment that operated on the core modulations ( modulations below ∼4 cycles/kHz and ∼8 Hz ) . This second notch experiment allowed us to obtain a more detailed MTF . The final speech MTF was obtained by combining the results of the spectral and temporal notch filters applied to the whole MPS . For this purpose , we calculated the average percentage error in word comprehension , and divided by the average control comprehension . Then we multiplied the normalized comprehension errors from the spectral notch filters ( Figure 6A , vertical stripes ) , and the temporal notch filters ( Figure 6A , horizontal stripes ) . The resulting color plot indicates which MPS areas are more important ( red ) for speech comprehension , and which are less important ( blue ) . For comparison , we also generated a summary plot from the low-pass spectral and temporal modulation filters ( Figure 6B ) . In this case , the subsequent increases in error caused by each lowering of the cutoff modulation frequency were used . A similar analysis of the notch filters applied to sentences containing only core modulations ( Figure 6C , redness indicates importance for comprehension ) gave an overall impression in general agreement with the respective areas of Figure 6A . It should be noted that , to generate this initial speech MTF , we assumed that spectral and temporal degradations affect the speech signal independently , which allowed us to multiply the normalized comprehension errors . We know , however , from the discrepancy between the comprehension after notch-filtering of core modulations , and the comprehension after notch-filtering of all modulations ( namely , removal of intermediate spectral modulations is more detrimental to performance if higher spectral modulations have been removed as well ) , that there must exist some spectrotemporal interdependence . We also assumed that the MTF is symmetric along positive and negative modulation frequencies , in other words , that the gain in the MTF for joint spectrotemporal modulations corresponding to down-sweeps equals the gain for up-sweeps . Although we have not further explored the interdependence of the spectral and temporal modulations , our joint spectrotemporal modulation filtering technique opens the door to future studies directly assessing the degree of interdependence and potential asymmetry . The shape of our final speech MTF ( temporally band-pass and spectrally low-pass ) approximately matches the shape of a psychophysical MTF that was obtained from detection thresholds for broadband moving ripples ( corresponding to a single point in the MPS ) in white noise [2] , but with some important differences . Chi et al . found that the human MTF was low-pass for spectral and temporal modulations , with increases in threshold detection for modulations greater than 2 cycles/octave and 16 Hz ( Footnote: Chi et al . state that their MTF is low-pass in the temporal domain but their psychometric function does show that detection at the very low temporal modulations is somewhat more difficult than at the low intermediate temporal modulations ) . In comparison , if we examined only our low-pass filtering results , we would find modulation cutoff values around 4 cycles/kHz and 12 Hz . ( Note that 4 cycles/kHz corresponds to 2 cycles/octave for center frequencies of 500 Hz , and that we too obtained a cutoff value of 2 cycles/octave using log frequency as shown in Figure S1 ) . The estimation of these upper boundaries is therefore very similar between the two studies . However , our complete speech MTF based on the combination of notch and low-pass filters shows a much more restricted area of high gain . For example , while the MTF of Chi et al . is relatively flat all the way to 2 cycles/octave , our speech MTF shows that the lowest spectral modulations ( <0 . 25 cycles/kHz ) play a more important role than the higher ones ( >0 . 5 cycles/kHz ) . There are therefore important differences between the MTF obtained by measuring detections of ripple sounds in noise and the one obtained by performing notch filtering operations on speech . While humans might be equally good at detecting low and intermediate spectral modulations , the lower ones carry more information for speech intelligibility . The intermediate modulations should carry more information for other auditory tasks such as pitch perception . While animal models of speech perception remain a stretched analogy , models of animal sensitivity to relevant modulations hold more immediate potential . The shape of our speech MTF also resembles the MTFs that have been obtained for mammalian [31] and avian [32] high-level auditory neurons . This correlation between the power of the spectrotemporal modulations in speech ( the speech MPS ) , the MTF resulting from tests of speech intelligibility , the MTF derived from detection of synthetic sounds [2] , and the tuning properties of auditory neuron ensembles suggests a match between the speech signal and the receiver . The most informative modulations in speech , and in other animal communication signals , occur in regions of the modulation spectrum where humans show high sensitivity and where animals' high-level auditory neurons have the highest gain [13] , [33] , [34] . We also examined the role of modulations in the task of recognizing the gender of a speaker . The MPSs of male and female speech differ in the frequency rate at which power is concentrated in the higher spectral modulations ( Figure 2 ) . In our MPS representation , the pitch-associated spectral frequencies of male and female speakers showed a bimodal distribution: the two modes correspond to the glottal action of the vocal cords pulsing at ∼150 Hz in adult male speakers and at above 200 Hz in females [22] . The spectral notch filter that removed the high spectral modulation power unique to the female voice confused listeners' percept of gender , such that half of the female stimuli notch filtered between 3–7 cycles/kHz sounded male to subjects . Control stimuli containing only the core modulations , which likewise lack the female-specific modulation power , similarly confused listeners . We conclude that modulations between 3 and 7 cycles/kHz give rise to the percept of female vocal pitch . It is interesting that removal of the modulations underlying the male vocal register did not appear to detract from perception of speaker masculinity . Although fundamental frequencies provide the basis for gender recognition particularly in vowels [35] , it has also been shown that the fundamental and the second formant frequency are equally good predictors of speaker gender [36] . Therefore the lower spectral modulations could carry additional gender information , but the acoustic distinction fails to explain the bias for male identification . Alternatively , the perception of vocal masculinity could depend more on gender-specific articulatory behaviors which account for social “dialectal” gender cues distinguishing even pre-pubescent speakers [37] . Our results have implications for speech representation purposes including compression , cochlear design , and speech recognition by machines . In both speech compression applications and signal processing for cochlear design , the redundancy of the speech signal allows a reduction in the bandwidth of a channel through which the signal is represented . For this purpose , limiting spectral resolution has been a favorite solution both because of the robustness of the signal to such deteriorations [6] , [29] and because of engineering design constraints for cochlear implants . However , in noisy environments , additional spectral information results in significant speech hearing improvement [20] , [25] . Our approach provides a guided solution: after determining the speech MTF , one can selectively reduce the bandwidth of the signal by first representing key spectral modulations and then systematically including the most important adjacent spectrotemporal modulations to capture the greatest overall space within constraints , as illustrated in cartoon form in Figure 6 ( see also [2] ) . Our initial experiment with gender identification , together with research in music perception [38] , shows that the most advantageous solution will depend on the task and the desired percept . Finally , the speech MTF could also be used as a template for filtering out broadband noise: a modulation filtering procedure can be used to emphasize the modulations important for speech and to de-emphasize all others . Both the speech compression and the speech filtering operation require a decomposition of the sound in terms of spectrotemporal modulations , as well as a re-synthesis . These are not particularly simple operations ( see Materials and Methods ) , but with advances in signal processing they will become possible in real time . After all , a similar operation appears to happen in real time in the auditory system [12] , [21] , [39] .
Subjects gave written consent as approved by the Committee for the Protection of Human Subjects at University of California , Berkeley . Native American-English speakers of mixed gender ( 20 in the low-pass experiment , aged 18–34 yr; and 17 in the notch experiment , age range 18–36 yr ) volunteered to participate in the study . Audiograms showed that their hearing thresholds were normal from 30 to 15 , 000 Hz; one subject was excluded due to high-frequency hearing loss . Acoustically clean recordings of spoken sentences were obtained from the soundtrack of the Iowa Audiovisual Speech Perception videotape [40] . The soundtrack was digitized at 32 kHz sampling rate in our laboratory using TDT System II hardware . This corpus consists of 100 short complete sentences read without emotion by six adult male and female American-English speakers . See Figure 1 for the spectrogram of one example , “The radio was playing too loudly . ” The corpus has been used by previous studies of speech perception [5] , [6] . The original speech sentences were normalized for power . The synthetic degraded speech signals were generated from this original set by a novel filtering procedure performed on the log spectrogram , as described below . The modulation power spectrum ( MPS ) of a sound is the amplitude spectrum of the 2D Fourier Transform of a time-frequency representation of the sound's pressure waveform [3] . The MPS can be estimated for a single sound ( e . g . one sentence ) or for an ensemble of sounds ( e . g . 50 sentences from adult male speakers ) . In our analysis , the time-frequency representation is the log amplitude of a spectrogram obtained with Gaussian windows . Gaussian windows are used because of their symmetry in time-frequency and because they result in time-frequency representations that are more easily invertible [41] . As in cepstral analysis [23] , the logarithm of the amplitude of the spectrogram is used to disentangle multiplicative spectral or temporal modulations into separate terms . For example , in speech sounds , the spectral modulations that constitute the formants in vowels ( timbre ) separate from those that constitute the pitch of the voice ( Figure 2B ) . The MPS is then the amplitude squared as a function of the Fourier pairs of the time and frequency axis of the spectrogram of the log amplitude of this spectrographic representation . We call these two axes the temporal modulations ( in Hz ) and the spectral modulations ( in cycles/kHz ) . One of these two axes must have positive and negative frequency modulations to distinguish upward frequency modulations ( e . g . , cos ( ωsf-ωtt ) ) from downward modulations ( e . g . , cos ( ωff+ωtt ) ) . By convention , we use positive and negative temporal modulations . The time-frequency resolution scale of the spectrogram ( given by the width of the Gaussian window ) determines the upper bounds of the temporal and spectral modulation in an inverse relationship as a result of the uncertainty principle or time-frequency tradeoff . The time-frequency scale must therefore be chosen carefully so that modulation frequencies of interest are considered . The choice of time-frequency scale can be made in a somewhat systematic fashion by using a value for which the shape of the modulation spectrum does not change very much . At these values of time-frequency scale , most of the energy in the modulation spectrum would be far from the boundaries determined by the time-frequency tradeoff [3] . For analyzing our original and filtered signals , we used a time-frequency scale given by a Gaussian window of 10 ms in the time domain or 16 Hz in the frequency domain . For obtaining the MPS of sound ensembles , sounds in their spectrographic representation were divided into segments of 1 s and the MPS for each segment was estimated before averaging to obtain a power density function . The boundaries of the modulation spectrum at the time-frequency scale of 10 ms–16 Hz are 50 Hz and 31 cyc/kHz . At this time-frequency scale , approximately 90% of the power in the modulation spectrum was found for temporal modulations below 25 Hz and for spectral modulations below 16 cycles/kHz , justifying the choice ( Figure 1 ) . Moreover , the temporal and spectral modulation cutoffs correspond approximately to the critical modulation frequency at which amplitude modulated tones and noise start to promote a pitch percept [33] . Thus , when we use this particular time-frequency scale , the temporal modulation frequencies analyzed are perceived predominantly as temporal changes , while higher temporal modulations ( those above 50 Hz ) which would mediate a percept of pitch are found along the spectral modulation axis . Using wider frequency filters might cause spectral modulation power that is plotted high on the ordinate ( e . g . , 5 cycles/kHz corresponding to a 200 Hz pitch ) to appear instead at a correspondingly high temporal modulation ( 200 Hz ) on the abscissa . For the modulation filtering operation described below we used other time-frequency scales which were adapted to the filter's cutoff frequencies and thus improved the required spectrogram inversion step in that process . The MPS can be obtained from a time-frequency decomposition with a linear frequency axis ( resulting in spectral modulations in units of cycles/kHz ) , or from a decomposition with a log frequency axis ( resulting in spectral modulation in units of cycles/octave ) . The log frequency axis is a better model of the decomposition that occurs in the auditory periphery , but we found that the linear-frequency scale is a better decomposition for describing sounds that have harmonic structure . We suggest that higher level neurons may be equally well described as representing either linear or log scale frequency [42] . In any case , both representations are useful . To be able to compare our results to other published work , we additionally obtained the speech MPS and psychometric curves using the log-frequency representation . These results are shown in Figure S1 . The sentences were degraded by a novel modulation filtering procedure . In brief , the sound is first represented in its spectrographic representation using a log-spectrogram calculated with Gaussian windows as described above . Then a new log-spectrogram is obtained by a 2D filtering operation . This filtering operation is performed in the Fourier domain of the modulation amplitude and phase in the following way . First the 2D FFT of the log spectrogram is calculated . Then the amplitudes of specific temporal and spectral modulations that we want to filter out are set to zero . The inverse 2D FFT yields the desired filtered log-spectrogram . After exponentiation , the spectrogram is then inverted using an iterative spectrogram inversion algorithm [43] . We then verified the procedure by calculating the spectrogram and MPS of the filtered sound . For a measure of the errors introduced by spectrogram inversion , we squared the differences between the desired spectrogram and the spectrogram obtained , and divided by the desired spectrogram power , summing the resulting values over time and frequency . Across the 100 stimulus sentences in the control condition , the residual error at the end of 20 algorithm iterations averaged 2 . 5% . When the 100 sentences were low-pass filtered in one step to create stimuli with only the core modulations , the average residual error after the 20 algorithm iterations was 6 . 3% . The modulation filtering was written in Matlab using modified code from Malcolm Slaney's Auditory Toolbox for the spectrogram inversion routine [44] . The complete program is available upon demand . The iterative method improves upon earlier overlap-and-add methods that had to compensate for the retention of phase information that unintentionally preserves some spectral information targeted for removal [7] , [8] . For the low-pass modulation filtering procedure , the time-frequency scale of the spectrogram was adjusted depending on the desired modulation frequency cutoffs of the modulation filter . For example , if the amplitude of spectral modulation frequencies above 2 cycles/kHz was to be set to zero , then using a time-frequency scale where spectral modulations were represented only up to values approaching 2 cycles/kHz gave better results . In this example , one could use a time-frequency window in the spectrogram of 1 . 25 ms–128 Hz to obtain a MPS with boundaries at 402 Hz and 3 . 9 cycles/kHz . Such adjustments made the inverting process much more efficient . Moreover , for low-pass filtering only , one could take this procedure to the extreme and calculate the spectrogram at a time-frequency scale that corresponds exactly to the modulation frequency cut-off of the filter . In that case , the spectrogram would not require any additional filtering and the spectrogram inversion routine can be by-passed altogether . One can instead directly obtain the filtered sounds by using the amplitude envelopes in each frequency band of the spectrogram and using these to modulate a set of narrowband signals of the same bandwidth and center frequency but unitary amplitude . These unit-amplitude narrowband signals can be obtained from Gaussian white-noise that is decomposed through the same spectrographic filter bank [45] or , equivalently , by generating them directly using an analytic signal representation [46] . In the analytical representation the amplitude is set to 1 and the instantaneous phase is random but band limited so that the instantaneous frequency remains within the band . In this study , this direct method was used to generate the low-pass modulation-filtered sentences . The modulation filtering that involved notch or band-stop filtering was done with the complete spectrogram filtering and inverting procedure . In the direct methods , the frequency cutoff for temporal frequencies is inversely related to the frequency cutoff for spectral frequencies but the conjugate boundary was always far from the limits being considered here . For example , a 49 Hz low-pass temporal filter had a conjugate spectral frequency cutoff of 32 cycles/kHz and any temporal filtering with cutoff frequencies below 49 Hz has spectral modulations cutoffs higher than 32 cycles/kHz ( Figure 3A and 3B ) . Because of this relationship the panels C and D of Figure 3 could be merged into one plot that would show a unimodal ( inverted U ) psychometric curve as a function of a spectrotemporal cutoff ( as in Figure S1 ) . More details on these sound synthesis procedures and on time-frequency scale effects can be found in [46] and [3] . A control ( unfiltered ) speech sentence was obtained by inverting the unfiltered log-spectrogram obtained with the 10 ms–16 Hz time-frequency scale ( low-pass experiment ) or 5 ms–32 Hz scale ( notch experiment ) . The control sentences sounded very similar to the original sentences and yielded high levels of intelligibility . Errors calculated during resynthesis depend on the bandwidth of the time-frequency scale . Residual errors in the control case of spectrogram inversion without filtering would barely be affected by changing the time-frequency scale from 5 ms–32 Hz to 1 . 25 ms–128 Hz ( 2 . 92% vs . 2 . 52% after 20 iterations , averaged over all 100 sentences ) . Similarly , in the case of temporal and spectral low-pass filtering leaving only core modulations , this time-frequency change would make a minimal improvement in the residual errors ( 5 . 49% vs . 6 . 29% ) . However , in the case of low-pass spectral modulation filtering with a 2 cycles/kHz cutoff , the 128 Hz time-frequency scale would double residual errors ( 12 . 18% vs . 6 . 41% ) . Using the 128 Hz time-frequency scale for temporal low-pass filtering with a 6 Hz cutoff would similarly increase residual error ( 5 . 64% vs . 2 . 02% ) . All sounds were presented through headphones ( Sennheiser HD265 Linear ) to subjects who sat in a sound attenuated chamber . An audiogram from 30 Hz to 15 kHz was obtained initially for each subject , using an adaptive staircase procedure ( Tucker Davis Technologies software PsychoSig ) and subjects who had thresholds of 20 dB above normal were excluded . For the comprehension test , the sentences were embedded in Gaussian white noise ( 0–20 kHz ) . The white-noise lasted 6 seconds and the sentences ( filtered and control ) started at random times between 300 ms and 2 s after the onset of the noise . The white noise was played at a level of 65 dB SPL ( B&K Level Meter , A-weighting , measured with headphone coupler from B&K ) . The modulated speech sentences were played at 3 different levels: 72 dB , 67 dB , and 62 dB SPL ( B&K level meter , A-weighting , peak level with slow integration , headphone coupler ) . The 5 dB attenuation steps were obtained using a programmable attenuator ( Tucker Davis Technologies ) . The signal to noise ratios ( SNR ) calculated from the SPL measurements of the speech and noise signals were therefore +7 , +2 and −3 dB . We also calculated the SNR in terms of the RMS values of the sound pressure waveform of the noise and speech and found almost identical values ( 6 . 7 dB , 1 . 7 dB and −3 . 3 dB ) . These SNRs were chosen in pilot data to yield complete sigmoidal psychometric tuning curves in the low-pass filtered conditions , and almost perfect speech intelligibility for the control condition [47] . Furthermore , these SNRs cover the 3 dB SNR level that presents little difficulty for normal listeners but reduces comprehension in the hearing impaired [48] , [49] . Subjects listened to the sentences at their own pace , pressing a button to elicit the next stimulus . They were instructed to type whatever words they heard followed by whether they perceived the speaker's gender to be male or female . Subjects were asked to guess if necessary , but not to force sentences into making sense if any words did not make sense together . The typed response files were scored for the percentage of words reported correctly , with an algorithm to compensate for small spelling errors . Baseline performance under control conditions and with +2 dB SNR was around 90% . During an experiment each subject heard all 100 sentences in the corpus without repetitions , so that each sentence was pseudorandomly assigned only to one normal ( control ) or filtered condition at one level . The SNR levels and the filtering conditions were presented in pseudorandom order . The notch-filtered sentences were presented only at +2 dB SNR . | The sound signal of speech is rich in temporal and frequency patterns . These fluctuations of power in time and frequency are called modulations . Despite their acoustic complexity , spoken words remain intelligible after drastic degradations in either time or frequency . To fully understand the perception of speech and to be able to reduce speech to its most essential components , we need to completely characterize how modulations in amplitude and frequency contribute together to the comprehensibility of speech . Hallmark research distorted speech in either time or frequency but described the arbitrary manipulations in terms limited to one domain or the other , without quantifying the remaining and missing portions of the signal . Here , we use a novel sound filtering technique to systematically investigate the joint features in time and frequency that are crucial for understanding speech . Both the modulation-filtering approach and the resulting characterization of speech have the potential to change the way that speech is compressed in audio engineering and how it is processed in medical applications such as cochlear implants . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"neuroscience/behavioral",
"neuroscience",
"neuroscience/sensory",
"systems",
"computational",
"biology",
"neuroscience/psychology",
"neuroscience/experimental",
"psychology"
] | 2009 | The Modulation Transfer Function for Speech Intelligibility |
The large variability in mRNA and protein levels found from both static and dynamic measurements in single cells has been largely attributed to random periods of transcription , often occurring in bursts . The cell cycle has a pronounced global role in affecting transcriptional and translational output , but how this influences transcriptional statistics from noisy promoters is unknown and generally ignored by current stochastic models . Here we show that variable transcription from the synthetic tetO promoter in S . cerevisiae is dominated by its dependence on the cell cycle . Real-time measurements of fluorescent protein at high expression levels indicate tetO promoters increase transcription rate ∼2-fold in S/G2/M similar to constitutive genes . At low expression levels , where tetO promoters are thought to generate infrequent bursts of transcription , we observe random pulses of expression restricted to S/G2/M , which are correlated between homologous promoters present in the same cell . The analysis of static , single-cell mRNA measurements at different points along the cell cycle corroborates these findings . Our results demonstrate that highly variable mRNA distributions in yeast are not solely the result of randomly switching between periods of active and inactive gene expression , but instead largely driven by differences in transcriptional activity between G1 and S/G2/M .
At the single-cell level , mRNA and protein levels of regulable genes are often found to be highly variable [1]–[5] . The resulting long-tailed mRNA and protein distributions are well-described by stochastic models [1] , [5]–[7] of transcriptional bursting , where a promoter undergoes random and intermittent periods of highly active transcription . Real-time observations of transcription in multiple organisms appear consistent with this behavior [5] , [8]–[14] . Thus , both static and dynamic views attribute much of the observed mRNA variability to the stochastic nature of reactions intrinsic to transcription . Consequently , the standard stochastic model of gene expression has been widely used to infer steady-state dynamics [1] , [2] , [15]–[17] . However , earlier studies examining the origin of variability in protein expression found such variability is not solely due to stochasticity in reactions intrinsic to gene expression , but also extrinsic factors . This was done by looking for correlations in expression between identical copies of one promoter [18]–[20] and/or between that promoter and a global or pathway-specific gene [21] , [22] . Not only is the importance of extrinsic factors clear , without time-series measurements the intrinsic noise measured by these techniques may not completely be ascribed to stochastic reactions in gene expression [23] , [24] . While global extrinsic factors have been suggested to largely impact translation [1] , their influence on transcription and transcriptional bursting is unclear . The cell cycle has global effects on total protein and RNA synthesis that should play a role in transcription [12] , [20] , [25] . With few exceptions [20] , most deterministic and stochastic models of gene regulation do not account for cell cycle variability . Using both dynamic real-time protein and static single molecule mRNA measurements in single cells , we show that much of the variability in a synthetic tetO promoter typical of noisy genes in yeast is driven by differences in transcription rate between G1 and S/G2/M .
We examined cell-cycle dependent effects by microscopically monitoring fluorescent protein expression every 5 minutes in growing monolayers of yeast within a microfluidic chamber . We used a “3-color” diploid yeast strain with homologous 7xtetO promoters ( P7xtetO ) driving either Cerulean ( CFP ) or Venus ( YFP ) and a constitutive PGK1 promoter ( PPGK1 ) driving tdTomato ( RFP ) ( Fig . 1A ) . Correlations in transcriptional activity between different promoters allowed distinction between different sources of fluctuations [20] , [22] . We developed a method to infer the instantaneous transcription and protein production rates in single cells from the microscopy movies . We segment cells ( Fig . 1B ) , define budding times , track cell lineages and generate time series for volume and protein concentration ( measured by average fluorescence , Fig . 1C&D ) , similar to [26] , [27] . Two new features of our analysis enabled us to identify transitions in transcriptional state from these time series: precise assignment of cytokinesis ( Fig . S1 ) and using splines to stably estimate first and second time derivatives ( Text S1 ) . The product of volume and protein concentration time series gives total protein , P ( t ) ( Fig . 1E ) , from which we infer the protein production rate ( proportional to mRNA per cell , M ( t ) ) ( Fig . 1F ) , and transcription rate , A ( t ) ( Fig . 1G ) using a system of two differential equations describing transcription and translation ( Methods ) . The single-cell , instantaneous transcription rate estimated ( Fig . 1G ) represents a rate smoothed over a 15–20 minute window ( due to measurement and spline fitting errors ) and delayed 10–15 minutes due to fluorophore maturation ( Text S1 ) . To examine the effects of cell-cycle phase on expression , we aligned growth and expression data with respect to cell-cycle progression . We subdivided single-cell time series data between division events , synchronized the data by bud formation time , and rescaled time such that the pre- and post-bud phases mapped to the population average time in those phases . Thus , division occurs at 0 and 1 , and all traces bud at the same cell-cycle progression ( Fig . S2 ) . While reporter concentration is nearly flat across the cycle ( Fig . 2A ) , cells exhibit a slow growth phase up to bud formation corresponding to G1 and very early S , followed by faster growth in S/G2/M ( Fig . 2B ) , consistent with [27] , [28] . The instantaneous protein production rate similarly has two modes , but lags the instantaneous growth rate ( Fig . 2C ) . In contrast , the instantaneous transcription and growth rate correlate , approximately doubling in S/G2 relative to G1 ( Fig . 2D ) . This does not invalidate observations that total protein production rates correlate with average growth rate [29] ( Fig . S3 ) . The gradual rise in transcription rate over ∼30 min may mask a sharper change because of smoothing and reporter maturation ( Fig . S4 ) . These findings are robust to changes in cells' average growth rate ( Fig . 2E and Fig . S5 ) and remain present even without rescaling time ( Fig . S2 ) . We next added 50 ng/mL dox to reduce P7xtetO expression in the 3-color diploid to levels where transcription is thought to occur in infrequent , independent bursts at each locus that are presumably resolvable by the real-time analysis . Single-cell traces of transcription rate show occasional “on” periods ( Fig . S6 ) that are restricted to S/G2 , generally begin within 20 minutes of bud formation , and last until division ( Fig . 3A ) . If both P7xtetO copies turn on , >70% of the time they do so within 15 minutes of each other ( Fig . 3B ) . The “on” periods are not independent ( p<10−5 , χ2 test; ρ = 0 . 42 ) at each locus ( Fig . 3C ) . These results are in striking contrast to the view of transcriptional bursting as intrinsically driven with exponential interarrival times [1] , [4] , [8] , [30] . We sought further support for these real-time observations by using single-molecule mRNA fluorescence in situ hybridization ( FISH ) to probe how mRNA numbers in single cells varied with cell-cycle phase , classified based on the presence and size of a bud ( Methods ) . Fig . 4A&B describes mRNA distributions from cells with P1xtetO and P7xtetO , but no activator present ( basal expression ) . The G1 distributions are zero-peaked with a long-tail that disappears by early S , suggesting transcription does not occur in G1 , consistent with Fig . 3A . Progression through S/G2/M leads to a unimodal non-zero-peaked distribution in G2/M consistent with real-time observations ( Fig . 3B ) indicating the time when an inactive promoter turns ON in S/G2 is variable . With intermediate expression , there is also increased activity in S/G2/M , but the G1 distribution is qualitatively different: a non-zero peak indicates transcription now occurs in G1 ( Fig . 4C&D ) . However , low transcription activity does not imply G1 inactivity . The weak but constitutive DOA1 promoter ( PDOA1 ) [31] has a non-zero G1 peak even with low mRNA levels ( Fig . 4E ) . While the overall mRNA distributions exhibit excellent fits to a negative binomial predicted by the standard model [1] , [32] ( Fig . 4 ) , partitioning the data by cell-cycle phase reveals it is incorrect . We sought to understand to what extent the overall variability could be explained by a model with a constant transcription rate that increased by f-fold between G1 and S/G2/M ( Text S1 ) . When f = 2 as expected based on S-phase replication , the model qualitatively describes the progression of the observed distributions for PDOA1 ( Fig . 4E&F , Table S3 , Fig . S7 ) . However , f>2 better describes tetO promoter measurements , with f>100 for basal expression , consistent with no G1 transcription ( Fig . 4A–D , Fig . S7 ) . Still , noisy tetO promoters have more variable transcription than this model can generate . The real-time protein data ( Fig . S6 ) also indicate variability in the amount of mRNA produced in S/G2/M . Incorporating this variability by randomizing the timing of the transcription rate transition in the model gives predictions that agree well with P1/7xtetO but not PDOA1 ( Fig . S8 , Text S1 ) . The mRNA FISH images for tetO promoters tend to have bright spots thought to represent nascent mRNA transcription that are more likely in S/G2/M ( Fig . S9 ) and may indicate “bursty” expression as another source of variability . Finally , we asked if cell-cycle phase affected the kinetics of gene activation by measuring the time to activate P1xtetO and P7xtetO in response to a step change in transcription factor ( TF ) input . Because TF-regulating pathways in single cells may respond to chemical inducers like dox slowly and at variable times [33] , we developed a “kinetic” strain where TF activation was observable in its nuclear localization . The TF Pho4p localizes to the nucleus in response to low phosphate [34] . Reengineering the phosphate pathway ( Text S1 ) allows rapid , reversible control of a Pho4-tetR-YFP fusion capable of activating P1/7xtetO-CFP by toggling phosphate concentration ( Fig . 5A ) , with minimal effects on cell growth ( Fig . S10 ) . In response to a step change in phosphate concentration , both TF localization and subsequent gene activation from P1xtetO or P7xtetO driving CFP expression are identical ( Fig . 5B&C ) . The difference between localization and transcription activation timing in single cells ( Fig . 5D&E , Text S1 ) yields a response delay time distribution ( Fig . 5F ) with a median 17 min delay likely dominated by fluorophore maturation ( ∼10 min delay , Fig . S12 ) . Therefore , on average both promoters respond quickly to the TF . However , when we separate cells by the cell-cycle phase when TF localization occurs , the post-budding response delay distribution is significantly different from the pre-budding delay distribution ( Fig . 5G , 2-sample K-S test: p = 0 . 05 , P1xtetO; p<0 . 001 , P7xtetO ) , with median response times in post-budded cells 7 ( P1xtetO ) or 10 ( P7xtetO ) minutes earlier . We repeated the P7xtetO step test in 2% raffinose to extend G1 and better sample G1 cells . Both the median 20 min response delay and the 10 min gap between pre- and post-bud cells' delay ( K-S test: p = 0 . 005 ) are similar to results in glucose , with prolonged delays in activation restricted to G1 ( Fig . 5H , Fig . S13 ) .
Transcriptional bursting , whereby a promoter occasionally transitions from a long-lived inactive to a short-lived active state that produces a burst of mRNA , is commonly invoked to explain the observed single-cell variability in mRNA numbers [1] of noisy genes . In contrast , we demonstrate that large differences in transcriptional activity between G1 and S/G2 that go beyond gene dosage effects drive much of the observed single-cell variability in mRNA numbers . Genome-wide studies in yeast [3] , [35] have identified noisy promoters as those associated with strong TATA boxes and highly regulated by chromatin remodeling factors . The tetO promoters , whose core region is derived from the native CYC1 promoter [36] , have similar characteristics . Likewise , constitutive , housekeeping promoters and highly expressed tetO promoters previously associated with low expression variability [2] , [3] , [31] exhibit only ∼2-fold changes in transcription throughout the cell cycle consistent with gene dosage effects . None of these promoters are classified as cell-cycle regulated [37] . While the cell cycle has been appreciated to be an important source of extrinsic noise , our findings suggest there may be a specific role beyond gene dosage for noisy genes that have not been associated with cell-cycle regulation . Transcriptional bursting may still occur , but it is not needed to explain most of the variability in mRNA levels given the variability in timing of the G1 to S transition [38] , [39] . The heretofore unexplored connection between noisy gene expression and large differences in G1 and S/G2 transcriptional activity raises fundamental questions concerning its origin and prevalence amongst noisy genes in various organisms and its implications in stable gene network regulation . Studies monitoring transcripts in synchronized yeast cells with tiling microarrays identify ∼10% of genes that exhibit cell-cycle-dependent expression [37] . The mechanistic basis of their cycling , through regulation by cyclin-dependent transcription factors , is well-understood . The CYC1 gene , whose core promoter is present in the tetO promoters , is not part of this class . Because we observed cycle-dependent basal expression for tetO promoters in the absence of any tTA , the CYC1 gene might be expected to behave similarly . If we surmise that this cycle-dependent expression occurs at a majority of the ∼20% of genes whose expression has been identified in genome-wide studies as noisy [35] , why has not previous work , which monitors expression in populations of synchronized cells , identified these genes ? First , normalized microarray data represent a mole fraction of mRNA species in the population . As such , they will never identify , say , two-fold changes in mRNA abundance due to replication when normalized to the concurrent rise in total mRNA [40] . Second , we see large differences in transcription rate between G1 and S/G2 but because of the finite mRNA lifetime , differences in mRNA abundance are smaller . The discrepancies are much more pronounced with slower growth conditions ( as the mRNA reaches the new steady-state in each growth phase ) , but the microarray studies were done in fast growth conditions . Third , under conditions with the largest cell-cycle phase-dependent differences in transcription , S/G2 transcription is a probabilistic event . Microarray experiments measure the population average mRNA mole fraction in each phase , and hence the relative difference between S/G2 and G1 will be quite low on average . Therefore , monitoring expression in single living cells through multiple cell cycles is crucial to observing cell-cycle-dependent transcription . Single-cell studies do not suffer from the averaging effects of microarray analysis . Still , our data alters the interpretation of static single-cell studies where mRNA/protein distributions are fit to stochastic models of gene expression to infer steady-state dynamics [1] , [2] , [15] , [17] . This difficulty of using static data to pinpoint origins of variability has been anticipated [5] , [23] although here we have shown that even static mRNA FISH data can reveal additional dynamic information by disaggregating mRNA distributions by cell-cycle phase . Consistent with our observations , control measurements made in a study examining cycle-dependent mRNA degradation kinetics using mRNA FISH show an approximately 2-fold increase in mRNA levels as cells progress through S/G2/M for the ADH1 and DOA1 promoters [25] . Our results suggest that burst sizes and burst frequencies obtained from fitting static mRNA distributions to a negative binomial distribution may be of limited use beyond the fact that they describe the first two moments of the distribution ( and hence its noise ) . If we titrate tTA levels and infer burst statistics from the aggregate distribution , it appears that burst size increases at low expression levels , then burst frequency increases at high expression levels . This transition roughly corresponds to moving from little or no transcription in G1 ( with a burst frequency corresponding to one cell cycle ) to robust G1 expression ( Quinn , Maheshri , unpublished results ) . Because of the ease of grossly classifying cell-cycle phase in yeast , an exciting and straightforward exercise would be to reanalyze existing mRNA FISH datasets published from multiple labs to determine the prevalence of cycle-dependent expression . Morphological measurements might also be combined with two-color FISH measurements using a second transcript with known cell-cycle regulation [41] . Such an approach with high abundance transcripts has already revealed independent cell and metabolic cycles [42] . Recent real-time single-cell measurements of transcription may not have seen the cycle-dependent transcription we report for various reasons . In the one yeast study , mRNA expression of a housekeeping gene was measured in real-time with no accounting for cell-cycle phase [12] . In mammalian cells [14] , transcriptional bursts with a refractory period were inferred from real-time measurements of luciferase protein levels expressed by multiple promoters , but again there was no accounting for cell-cycle phase . The bursts occurred on time scales much shorter than cell-cycle transitions , but revisiting the data still may reveal a cell-cycle phase dependence . Interestingly , real-time measurements of mRNA expression from two housekeeping promoters in undifferentiated single Dictyostelium cells do show a weak cell-cycle dependence , with the frequency of transcriptional burst or pulses dropping 2–3 fold from the beginning to the end of the cell cycle [43] . However , Dictyostelium lack a G1 period and have an extended G2 [44] , [45]; the reduction in expression later in the cell cycle might be due to mitotic repression of transcription , which is a well-established phenomenon in mammalian cells [46] . It remains unclear how widespread cycle-dependent transcription is in other organisms and awaits further study . What might be the mechanistic basis for the large differences in G1 and S/G2 transcriptional activity ? At low expression tetO promoters exhibit transcriptional pulses that originate around the G1 to S transition when DNA replication occurs . S phase progression has been suggested to affect transcription [47] and has been established as important to ( de ) silence gene loci in a manner that either requires replication fork progression [48] , [49] or not [50] , [51] . Replication-dependent transcriptional activation of viral genes has been reported in transient transfection assays , and depends on trans activators that bind to either the proximal promoter or and enhancer region [52] , [53] . In these cases , there does not appear to be cycle-dependent changes in the activity of the trans activators . Furthermore , in an in vitro system replication potentiates Gal4-VP16 transcriptional activation during S phase [54] . In our case , cycle-dependent activity of tTA can be ruled out because we observe the phenomenon even in the absence of tTA ( Fig . 4A&B ) . In S phase , activity of the Cdc28 kinase leads to upregulation of the basal transcriptional machinery which has been shown to lead to similarly timed pulses of transcription in ∼200 yeast genes not classified as cycle-dependent , but these are enriched in housekeeping genes and CYC1 did not exhibit this behavior [55] . Therefore , we favor the hypothesis that temporary disruption of a repressed promoter's chromatin architecture during DNA replication could explain the pulse timing ( Fig . 3B ) and the increased tendency of tetO promoters to not activate until the G1 to S transition even when Pho4-tetR enters the nucleus in G1 ( Fig . 5H ) . ( We observe a similar cycle-dependence at the native PHO5 promoter where gene activation tends to occur in early S/G2 – Zopf , Maheshri , unpublished results . ) Because nucleosomes that are deposited following passage of the replication fork consist of an equivalent mixture of maternal and newly synthesized histones [56] , as this chromatin matures there may exist a period permissive for transcription . Even if this permissive period is short-lived , once activated , transcription-mediated deposition of histone marks could maintain an active transcription state [57] . The return to inactive state in G1 might be brought about during mitosis . Repressive modifications in general transcriptional machinery , changes in histone marks , and chromatin condensation all contribute to mitotic repression of RNA transcription in higher eukaryotes [58] . In budding yeast , transcription was originally proposed to occur throughout mitosis as total RNA synthesis appeared to increase at and after nuclear migration [59] , but recent work shows Cdc14 inhibiting RNA Pol I transcription during anaphase [60] . Still , mitotic repression of RNA Pol II transcription remains unclear , and we did observe an increased number of nascent spots of transcription in cells with large buds indicative of late G2 or M phase ( Fig . S9 ) . Exploring genetic and chemical perturbations in the activity of factors involved in chromatin maturation during DNA replication and mitosis for their effect on the cycle-dependent transcription pattern should aid in elucidating precise mechanisms . Our results should spur the development of new models that incorporate cell-cycle linked pulses of transcription and analyze its effects on the dynamics and function of genetic regulatory networks . Global extrinsic factors influence the behavior [22] and previous work identifies two important roles of cell division: in the stochastic partitioning of cellular contents [24] and in setting the timescale at which extrinsic fluctuations decay [61] . The latter especially has been incorporated in models of global extrinsic fluctuations [62] . Revised models in light of our results may be particularly exciting to develop in networks containing positive feedback loops , where switching from an OFF to ON state relies heavily on the statistics of transcription at low expression . Because here transcription appears prohibited in G1 , this would introduce a sizeable refractory period to switching . Finally , since poorer nutrient conditions increase cell division time by extending G1 in yeast , prohibited G1 transcription may alter the dynamics of regulatory networks in cells in different growth conditions in unexpected ways .
All S . cerevisiae strains were constructed in a W303 background [63] using standard methods of yeast molecular biology [64] . Details of strain and plasmid construction are listed in Tables S1 and S2 . The strain referred to as the “3-color” diploid yeast strain contains homologous 7xtetO promoters ( P7xtetO ) driving either Cerulean ( CFP ) [65] or Venus ( YFP ) [66] and a constitutive PGK1 promoter ( PPGK1 ) driving tdTomato ( RFP ) [67] ( Fig . 1A ) . Construction of the “kinetic” strain ( Fig . 5A ) used to measure the rate of gene activation is described in detail in the Text S1 . Yeast nitrogen base without phosphate ( MP Biomedicals , Santa Ana , CA; #4027-812 ) was mixed with monobasic potassium phosphate solution ( Sigma-Aldrich P8709 ) to set phosphate levels , and the pH was lowered to 4 . Synthetic complete ( SC ) media contained all amino acids , 2% glucose as the carbon source , and 5000 µM phosphate unless otherwise indicated . For each time series experiment , cells were picked from a single colony freshly grown on a minimal synthetic solid media ( no amino acids ) with 2% glucose agar , inoculated into SC media , grown overnight on a roller drum at 30°C to OD600 nm∼0 . 1 , diluted in fresh media , and grown again for 6–8 hrs to OD600 nm∼0 . 1 . These cells were loaded into a pre-washed and Y04C microfluidic plate ( CellAsic , Hayward , CA ) primed with SC with appropriate carbon source and/or phosphate level , according to the manufacturer's instructions . Cells were perfused with SC at 6 psi throughout the experiment . Flow was controlled using the programmable ONIX system ( CellAsic ) to rapidly switch between various media conditions discussed in the text . Cell growth and expression was observed using a Zeiss Axio Observer . Z1 inverted microscope at 63× magnification ( Zeiss Plan-Apochromat 63×/1 . 40 Oil DIC ) . Bright field ( BF ) , BF out of focus ( BFOOF , for segmentation ) and fluorescence images were acquired every 5 min with a Cascade II EMCCD camera ( Photometrics , Tuscon , AZ ) using MetaMorph software ( Molecular Devices , Sunnyvale , CA ) , a Lumen 200 metal-halide arc lamp ( PRIOR Scientific , Rockland , MA ) for fluorescence excitation , appropriate filters for CFP , YFP , and RFP ( Chroma Technology Corp , Bellows Falls , VT; set 89006 ) , and acquisition settings optimized for rapid time points . ( Detailed protocol available online at http://openwetware . org/wiki/Maheshri:Internal . ) We used custom-written , GUI-based software in MATLAB ( Mathworks , Natick , MA ) for semi-automated analysis of the fluorescence microscopy movies to extract single-cell volume and fluorescence data series [68] . Briefly , the cell regions were first segmented using a focused and an unfocused bright field image , tracked through time , and assigned mother-bud lineages , all of which were manually curated . Each bud's data were incorporated into its mother's time series until the automatically-assigned cytokinesis time . Next , in order to obtain stable time derivatives for each data series , we implemented a smoothing algorithm based on spline fitting . Finally , the total protein ( fluorescence ) spline P ( t ) for each cell was used to calculate the protein production rate ( proportional to mRNA per cell , M ( t ) ) , and the transcription rate , A ( t ) , using a simple continuous-time model of transcription and translation: ( 1 ) ( 2 ) where γM is the mRNA degradation rate and kt is the translation rate of mRNA to protein . We argue translation rate is nearly constant across the cell-cycle ( Text S1 ) . Further details of the analytical methods are available in the Text S1 . Single-stranded DNA probes to vYFP were coupled to tetramethylrhodamine ( TMR ) or indodicarbocyanine ( Cy5 ) fluorophores and probes to tdTomato were coupled to TMR fluorophores , as in [2] . Yeast were grown to mid log-phase ( OD600 nm = 0 . 5–1 ) then fixed , spheroplasted , hybridized and washed similarly to [69] with modifications as described in [2] . DNA probes at ∼5 µM were diluted 50-fold into hybridization solution containing 10% formamide . Cells were imaged on a Zeiss AxioObserver inverted microscope equipped with a PRIOR Lumen200 mercury arc lamp , a 100×/1 . 40 objective ( Zeiss ) and a rhodamine- and Cy5-specific filter set ( Chroma Technology Cat . No . 31000v2 and 41024 respectively ) . For each sample , eight Z-stack images 0 . 3 microns apart were obtained and analyzed using custom software written in MATLAB based on that used in [2] . The algorithm used to identify spots corresponding to single mRNA applies region-based thresholding and identifies local maxima as spots . Three parameters used by the algorithm can change due to day-to-day variation in staining: ( 1 ) the minimum intensity for a pixel to be considered as part of a spot , manually set by examining several z-stacks , picking a threshold that identifies spots and not background , and verified by insuring a false positive rate of <5% in negative control samples; ( 2 ) the average intensity of a single mRNA spot , chosen using the mode of spot intensities for lower expressing samples ( Fig . S14A ) , to allow counting of multiple overlapping mRNA; and ( 3 ) the threshold intensity at which a spot is classified as a site of nascent transcription , chosen as the transition between the peaked and flat sections of a histogram of spot intensities ( ∼5–10 fold higher than the intensity of a single spot – Fig . S14A ) . Mean protein levels in different samples were used as an internal control , and we always verified the ratio of mean protein level to mean mRNA count was consistent across samples and expression levels . Fig . S14B&C are examples of processed images showing mRNA spots . In order to investigate cell-cycle dependence of transcription , we manually classified cells in FISH images in either S/G2/M or G1 based on the presence of a bud . Imaging at low cell density assisted in distinguishing buds from adjacent cells . Cells in S/G2/M were further sub-divided into three equal-sized bins , based on ranked bud size , which approximates progression through the cell cycle ( as in Fig . 4 ) . | There is an astonishing amount of variation in the number of mRNA and protein molecules generated from particular genes between genetically identical single cells grown in the same environment . Particularly for mRNA , the large variation seen from these “noisy” genes is consistent with the idea of transcriptional bursting where transcription occurs in random , intermittent periods of high activity . There is considerable experimental support for transcriptional bursting , and it is a primary feature of stochastic models of gene expression that account for variation . Still , it has long been recognized that variation , especially in protein levels , can occur because of global differences between genetically identical cells . We show that in budding yeast , mRNA variation is driven to a large extent by differences in the transcriptional activity of a noisy gene between different phases of the cell cycle . These differences are not because of specific cell-cycle regulation , and in some cases transcription appears restricted to certain phases , leading to pulses of mRNA production . These results raise new questions about the origins of transcriptional bursting and how the statistics of gene expression are regulated in a global way by the cell cycle . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"systems",
"biology",
"cell",
"growth",
"gene",
"expression",
"genetics",
"biology",
"computational",
"biology",
"molecular",
"cell",
"biology"
] | 2013 | Cell-Cycle Dependence of Transcription Dominates Noise in Gene Expression |
The functional consequences of trait associated SNPs are often investigated using expression quantitative trait locus ( eQTL ) mapping . While trait-associated variants may operate in a cell-type specific manner , eQTL datasets for such cell-types may not always be available . We performed a genome-environment interaction ( GxE ) meta-analysis on data from 5 , 683 samples to infer the cell type specificity of whole blood cis-eQTLs . We demonstrate that this method is able to predict neutrophil and lymphocyte specific cis-eQTLs and replicate these predictions in independent cell-type specific datasets . Finally , we show that SNPs associated with Crohn’s disease preferentially affect gene expression within neutrophils , including the archetypal NOD2 locus .
In the past seven years , genome-wide association studies ( GWAS ) have identified thousands of genetic variants that are associated with human disease [1] . The realization that many of the disease-predisposing variants are non-coding and that single nucleotide polymorphisms ( SNPs ) often affect the expression of nearby genes ( i . e . cis-expression quantitative trait loci; cis-eQTLs ) [2] suggests these variants have a predominantly regulatory function . Recent studies have shown that disease-predisposing variants in humans often exert their regulatory effect on gene expression in a cell-type dependent manner [3–5] . However , most human eQTL studies have used sample data obtained from mixtures of cell types ( e . g . whole blood ) or a few specific cell types ( e . g . lymphoblastoid cell lines ) due to the prohibitive costs and labor required to purify subsets of cells from large samples ( by cell sorting or laser capture micro-dissection ) . In addition , the method of cell isolation can trigger uncontrolled processes in the cell , which can cause biases . In consequence , it has been difficult to identify in which cell types most disease-associated variants exert their effect . Here we describe a generic approach that uses eQTL data in mixtures of cell types to infer cell-type specific eQTLs ( Fig 1 ) . Our strategy includes: ( i ) collecting gene expression data from an entire tissue; ( ii ) predicting the abundance of its constituent cell types ( i . e . the cell counts ) , by using expression levels of genes that serve as proxies for these different cell types ( since not all datasets might have actual constituent cell count measurements ) . We used an approach similar to existing expression and methylation deconvolution methods [6–11]; ( iii ) run an association analysis with a term for interaction between the SNP and the proxy for cell count to detect cell-type-mediated or-specific associations , and ( iv ) test whether known disease associations are enriched for SNPs that show the cell-type-mediated or-specific effects on gene expression ( i . e . eQTLs ) .
We applied this strategy to 5 , 863 unrelated , whole blood samples from seven cohorts: EGCUT[12] , InCHIANTI [13] , Rotterdam Study [14] , Fehrmann [2] , SHIP-TREND [15] , KORA F4 [16] , and DILGOM [17] . Blood contains many different cell types that originate from either the myeloid ( e . g . neutrophils and monocytes ) or lymphoid lineage ( e . g . B-cells and T-cells ) . Even though neutrophils comprise ~60% [18 , 19] of all white blood cells , no neutrophil eQTL data have been published to date , because this cell type is particularly difficult to purify or culture in the lab [20] . For the purpose of illustrating our cell-type specific analysis strategy in the seven whole blood cohorts , we focused on neutrophils . Direct neutrophil cell counts and percentages were only available in the EGCUT and SHIP-TREND cohorts , requiring us to infer neutrophil percentages for the other five cohorts . We used the EGCUT cohort as a training dataset to identify a list of 58 Illumina HT12v3 probes that correlated positively with neutrophil percentage ( Spearman’s correlation coefficient R > 0 . 57; S1 Fig , S1 Table ) . We observed that 95% of these genes show much higher expression in purified neutrophils , as compared to 13 other purified blood cell-types , based on RNA-seq data from the BLUEPRINT epigenome project [21] ( S2 Fig ) . We then summarized the gene expression levels of these 58 individual probes into a single neutrophil percentage estimate , by applying principal component analysis ( PCA ) and determining the first principal component ( PC ) . We then used this first PC as a proxy for neutrophil percentage , an approach that is similar to existing expression and methylation deconvolution methods [6–11] . In the EGCUT dataset , the actual neutrophil percentage showed some correlation with age ( Pearson R = 0 . 08 , P = 0 . 02 ) , but no association with gender ( Student's T-test P = 0 . 31; S3 Fig ) in the EGCUT dataset . The proxy for neutrophil percentage showed the same behavior: some correlation with age ( Pearson R = 0 . 14 , P = 6 x 10–5 ) and no association with gender ( P = 0 . 11 ) . This predicted neutrophil percentage strongly correlated with the actual neutrophil percentage ( Spearman R = 0 . 75 , Pearson R = 0 . 76; Fig 2 ) . Including more or fewer top probes than the top 58 probes resulted in similar correlations ( S4 Fig ) . We then used this set of 58 probes in each of the other cohorts as well , and used PCA per cohort on the probe correlation matrix of these 58 probes . In the SHIP-TREND cohort , for which the actual neutrophil percentage was available as well , we observed that the inferred neutrophil proxy strongly correlated with the actual neutrophil percentage ( Spearman R = 0 . 81 , Pearson R = 0 . 82; Fig 2 ) . Here we limited our analysis to 13 , 124 cis-eQTLs that were previously discovered in a whole blood eQTL meta-analysis of a comparable sample size [22] ( we note that these 13 , 124 cis-eQTLs were detected while assuming a generic effect across cell-types , and as such , genome-wide application of cell-type specificity strategy might result in the detection of additional cell-type-specific cis-eQTLs ) . To infer the cell-type specificity of each of these eQTLs , we performed the eQTL association analysis with a term for interaction between the SNP marker and the proxy for cell count within each cohort , followed by a meta-analysis of the interaction term ( weighted for sample size ) across all the cohorts . We identified 1 , 117 cis-eQTLs with a significant interaction effect ( 8 . 5% of all cis-eQTLs tested; false discovery rate ( FDR ) < 0 . 05; 1 , 037 unique SNPs and 836 unique probes; S2 Table and S3 Table ) . The power to detect these effects depends on the original cis-eQTL effect size , as we observed that the main effect of cis-eQTLs that have a significant interaction term generally explained more variance in gene expression in the previously published cis-eQTL meta-analysis [22] , as compared to the cis-eQTLs that did not show a significant interaction: 71% of the cell-type specific eQTLs had a main effect that explained at least 3% of the total expression variation , whereas 25% of the cis-eQTLs that did not show a show significant interaction explained at least 3% of the total expression variation ( S5 Fig ) . Out of the total number of cis-eQTLs tested , 909 ( 6 . 9% ) had a positive direction of effect , which indicates that these cis-eQTLs show stronger effect sizes when more neutrophils are present ( i . e . ‘neutrophil-mediated cis-eQTLs’; 843 unique SNPs and 692 unique probes ) . Another 208 ( 1 . 6% ) had a negative direction of effect ( 196 unique SNPs and 145 unique probes ) , indicating a stronger cis-eQTL effect size when more lymphoid cells are present ( i . e . ‘lymphocyte-mediated cis-eQTLs’; since lymphocyte percentages are strongly negatively correlated with neutrophil percentages; Fig 1 ) . Overall , the directions of the significant interaction effects were consistent across the different cohorts , indicating that our findings are robust ( S6 Fig ) . We validated the neutrophil- and lymphoid-mediated cis-eQTLs in six small , but purified cell-type gene expression datasets that had not been used in our meta-analysis . We generated new eQTL data from two lymphoid cell types ( CD4+ and CD8+ T-cells ) and one myeloid cell type ( neutrophils , see online methods ) and used previously generated eQTL data on two lymphoid cell types ( lymphoblastoid cell lines and B-cells ) and another myeloid cell type ( monocytes , S4 Table ) . As expected , compared to cis-eQTLs without a significant interaction term ( ‘generic cis-eQTLs’ , n = 12 , 007 ) the 909 neutrophil-mediated cis-eQTLs did indeed show very strong cis-eQTL effects in the neutrophil dataset ( Wilcoxon P = 2 . 5 x 10–81 when compared to generic cis-eQTLs; Fig 3A ) . We observed that these cis-eQTLs also showed an increased effect-size in the monocyte dataset ( Wilcoxon P-value = 4 . 9 x 10–31 when compared to generic cis-eQTLs; Fig 3A ) . Because both neutrophils and monocytes are myeloid lineage cells , this suggests that some of the neutrophil-mediated cis-eQTL effects also show stronger effects in other cells of the myeloid lineage . Conversely , the 208 lymphoid-mediated cis-eQTLs had a pronounced effect in each of the lymphoid datasets ( Wilcoxon P-value ≤ 7 . 8 x 10–14 compared to generic cis-eQTLs; Fig 3A ) , while having small effect sizes in the myeloid datasets . These validation results indicate that our method is able to reliably predict whether a cis-eQTL is mediated by a specific cell type . Unfortunately , the cell type that mediates the cis-eQTL is not necessarily the one in which the cis-gene has the highest expression ( Fig 3B ) , making it impossible to identify cell-type-specific eQTLs directly on the basis of expression levels . Myeloid and lymphoid blood cell types provide crucial immunological functions . Therefore , we assessed five immune-related diseases for which genome-wide association studies previously identified at least 20 loci with a cis-eQTL in our meta-analysis . We observed a significant enrichment only for Crohn’s disease ( CD ) , ( binomial test , one-tailed P = 0 . 002 , S5 Table ) : out of 49 unique CD-associated SNPs showing a cis-eQTL effect , 11 ( 22% ) were neutrophil-mediated . These 11 SNPs affect the expression of 14 unique genes ( ordered by size of interaction effect: IL18RAP , CPEB4 , RP11-514O12 . 4 , RNASET2 , NOD2 , CISD1 , LGALS9 , AC034220 . 3 , SLC22A4 , HOTAIRM2 , ZGPAT , LIME1 , SLC2A4RG , and PLCL1 ) . CD is a chronic inflammatory disease of the intestinal tract . While impaired T-cell responses and defects in antigen presenting cells have been implicated in the pathogenesis of CD , so far little attention has been paid to the role of neutrophils , because its role in the development and maintenance of intestinal inflammation is controversial: homeostatic regulation of the intestine is complex and both a depletion and an increase in neutrophils in the intestinal submucosal space can lead to inflammation . On the one hand , neutrophils are essential in killing microbes that translocate through the mucosal layer . The mucosal layer is affected in CD , but also in monogenic diseases with neutropenia and defects in phagocyte bacterial killing , such as chronic granulomatous disease , glycogen storage disease type I , and congenital neutropenia , leading to various CD phenotypes [23] . On the other hand , an increase in activated neutrophils that secrete pro-inflammatory chemokines and cytokines ( including IL18RAP which has a neutrophil specific eQTL ) maintains inflammatory responses . Pharmacological interventions for the treatment of CD have been developed to specifically target neutrophils and IL18RAP , including Sagramostim [24] and Natalizumab [25] . These results show clear neutrophil-mediated eQTL effects for various known CD genes , including the archetypal NOD2 gene . Although CD has previously been shown to have a slightly higher incidence in females [26 , 27] , we did not find any relationship between NOD2 expression and gender or age ( Student's T-test P = 0 . 08 and Pearson's correlation P = 0 . 39 respectively; S7 Fig ) . As such , our results provide deeper insight into the role of neutrophils in CD pathogenesis . Large sample sizes are essential in order to find cell-type-mediated cis-eQTLs ( Fig 4 ) : when we repeat our study on fewer samples ( ascertained by systematically excluding more cohorts from our study ) , the number of significant cell-type-mediated eQTLs decreased rapidly . This was particularly important for the lymphoid-mediated cis-eQTLs , because myeloid cells are approximately twice as abundant as lymphoid cells in whole blood . Consequently , detecting lymphoid-mediated cis-eQTLs is more challenging than detecting neutrophil-specific cis-eQTLs . As whole blood eQTL data is easily collected , we were able to gather a sufficient sample size in order to detect cell-type-mediated or-specific associations without requiring the actual purification of cell types .
Here we have shown that it is possible to infer in which blood cell-types cis-eQTLs are operating from a whole blood dataset . Cell-type proportions were predicted and subsequently used in a G x E interaction model . Hundreds of cis-eQTLs showed stronger effects in myeloid than lymphoid cell-types and vice versa . These results were replicated in 6 individual purified cell-type eQTL datasets ( two reflecting the myeloid and four reflecting the lymphoid lineage ) . This indicates our G x E analysis provides important additional biological insights for many SNPs that have previously been found to be associated with complex ( molecular ) traits . Here , we concentrated on identifying cis-eQTLs that are preferentially operating in either myeloid or lymphoid cell-types . We did not attempt to assess this for specialized cell-types within the myeloid or lymphoid lineage . However , this is possible if cell-counts are available for these cell-types , or if these cell-counts can be predicted by using a proxy for those cell-counts . As such , identification of cell-type mediated eQTLs for previously unstudied cell-types is possible , without the need to generate new data . However , it should be noted that these individual cell-types typically have a rather low abundance within whole blood ( e . g . natural killer cells only comprise ~2% of all circulating white blood cells ) . As a consequence , in order to have sufficient statistical power to identify eQTLs that are mediated by these cell-types , very large whole blood eQTL sample-sizes are required and the cell type of interest should vary in abundance between individuals ( analogous to the difference in the number of identified lymphoid mediated cis-eQTLs , as compared to the number of neutrophil mediated cis-eQTLs , which is likely caused by their difference in abundance in whole blood ) . For instance we have recently investigated whole peripheral blood RNA-seq data in over 2 , 000 samples and identified only a handful monocyte specific eQTLs ( manuscript in preparation ) . As such this indicates that in order to identify monocyte specific eQTLs using whole blood , thousands of samples should be studied . We confined our analyses to a subset of cis-eQTLs for which we had previously identified a main effect in whole peripheral blood [22]: for each cis-eQTL gene , we only studied the most significantly associated SNP . Considering that for many cis-eQTLs multiple , unlinked SNPs exist that independently affect the gene expression levels , it is possible that we have missed myeloid or lymphoid mediation of these secondary cis-EQTLs . The method we have applied to predict the neutrophil percentage in the seven whole blood datasets involves correlation of gene expression probes to cell count abundances and subsequent combination of gene expression probes into a single predictor using PCA . This approach is comparable to other deconvolution methods for methylation and gene expression data [6–11] . However , methods have also been published [28 , 29] that take a more agnostic approach towards identifying different cell-types and their abundances across different individuals . The Preininger et al method [28] identified different axes of gene expression variation in peripheral blood , of which some reflect proxies of certain cell-types . We quantified these axes for each of the samples of the EGCUT and Fehrmann cohorts by creating proxy phenotypes , and subsequently conducted per axis an interaction meta-analysis and indeed identified eQTLs that were significantly mediated by these axes ( S6 Table ) . We first ascertained the Z-scores of the eQTL interaction effects for axis 5 of Preininger et al , an axis that is known to correlate strongly with neutrophil percentage . As expected , we observed a very strong correlation with the Z-scores of the eQTL interaction effects for the neutrophil proxy ( R = 0 . 72 ) . While some other axes also mediate eQTLs that might be attributable to differences in other cell-type proportions ( e . g . axis 2 that is likely due to differences in reticulocyte proportions ) , while other axes might even reflect differences in the state in which these cells are ( e . g . stimulated immune cells or senescent cells ) . We believe it is possible that some of these axes might reflect differences in gene activity ( rather than differences in cell-type proportions ) , and that it could be that such differences in gene activity might mediate certain eQTLs . This indicates that by applying novel methods that can summarize gene expression [28 , 29] and using these summaries as interaction terms when conducting eQTL mapping , various ‘context specific’ eQTLs might become detectable . Although we have shown that the proxy that is created by our method is able to predict neutrophil percentage accurately , this may not be the case for all cell types available in whole blood , which may be greatly dependent upon the ability of individual gene expression probes to differentiate between cell types However , we anticipate that the ( pending ) availability of large RNA-seq based eQTL datasets , statistical power to identify cell-type mediated eQTLs using our approach will improve: since RNA-seq enables very accurate gene expression level quantification and is not limited to a set of preselected probes that interrogate well known genes ( as is the case for microarrays ) , the detection of genes that can serve as reliable proxies for individual cell-types will improve . Using RNA-seq data , it is also possible to assess whether SNPs that affect the expression of non-coding transcripts , affect splicing [30] or result in alternative polyadenylation [31] are mediated by specific cell-types . The method we have used did not account for heteroscedasticity while estimating the standard errors of the interaction term , which may lead to inflated statistics . We therefore compared the standard errors that we used , with standard errors that have been estimated while accounting for heteroscedasticity ( using the R-package 'sandwich' ) . The heteroscedasticity-consistent standard errors and p-values were very similar ( Pearson correlation > 0 . 95; S8 Fig ) to the standard errors that did not account for heteroscedasticity . We also note that measurement error in the covariates of the model we have used may cause the inferred betas to be biased . Structural equation modeling may be used to determine unbiased estimates by taking measurement errors of the covariates into account ( particularly the neutrophil percentage proxy ) . However , typically these methods require replicate measurements of the covariates , which were not available for the cohorts in our study . As such , the observed interaction effects may be either underestimated or overestimated , depending on the character and the degree of the measurement error in our covariates . Although we applied our method to whole blood gene expression data , our method can be applied to any tissue , alleviating the need to sort cells or to perform laser capture micro dissection . The only prerequisite for our method is the availability of a relatively small training dataset with cell count measurements in order to develop a reliable proxy for cell count measurements . Since the number of such training datasets is rapidly increasing and meta-analyses have proven successful [2 , 22] , our approach provides a cost-effective way to identify cell-type-mediated or-specific associations that can supplement results obtained from purified cell type specific datasets , and it is likely to reveal major biological insights .
This eQTL meta-analysis is based on gene expression intensities measured in whole blood samples . RNA was isolated with either PAXgene Tubes ( Becton Dickinson and Co . , Franklin Lakes , NJ , USA ) or Tempus Tubes ( Life Technologies ) . To measure gene expression levels , Illumina Whole-Genome Expression Beadchips were used ( HT12-v3 and HT12-v4 arrays , Illumina Inc . , San Diego , USA ) . Although different identifiers are used across these different platforms , many probe sequences are identical . Meta-analysis could thus be performed if probe-sequences were equal across platforms . Integration of these probe sequences was performed as described before [22] . Genotypes were harmonized using HapMap2-based imputation using the Central European population [32] . In total , the eQTL genotype x environment interaction meta-analysis was performed on seven independent cohorts , comprising a total of 5 , 863 unrelated individuals ( full descriptions of these cohorts can be found in the Supplementary Note ) . Mix-ups between gene expression samples and genotype samples were corrected using MixupMapper [33] . Each cohort performed gene expression normalization individually: gene expression data was quantile normalized to the median distribution then log2 transformed . The probe and sample means were centered to zero . Gene expression data was then corrected for possible population structure by removing four multi-dimensional scaling components ( MDS components obtained from the genotype data using PLINK ) using linear regression . Additionally , we corrected for possible confounding factors due to arrays of poor RNA quality . We reasoned that arrays of poor RNA quality generally show expression for genes that are normally lowly expressed within the tissue ( e . g . expression for brain genes in whole blood data ) . As such , the expression profiles for such arrays will deviate overall from arrays with proper RNA quality . To capture such variable arrays , we calculated the first PC from the sample correlation matrix and correlated the first PC with the sample gene expression measurements . Samples with a correlation < 0 . 9 were removed from further analysis ( S9 Fig ) . In order to improve statistical power to detect cell-type mediated eQTLs , we corrected the gene expression for technical and batch effects ( here we applied principal component analysis and removed per cohort the 40 strongest principal components that affect gene expression ) . Such procedures are commonly used when conducting cis-eQTL mapping [2 , 5 , 22 , 30 , 31 , 34] . To minimize the amount of genetic variation removed by this procedure , we performed QTL mapping for each principal component , to ascertain whether genetic variants could be detected that affected the PC . If such an effect was detected , we did not correct the gene expression data for that particular PC [22] . As a result , this procedure also removed the majority of the variation that explained the correlation between neutrophil percentage and gene expression ( S10 Fig ) , minimizing issues with possible collinearity when testing the interaction effects . We chose to remove 40 PCs based on our previous study results , which suggested that this was the optimum for detecting eQTLs [22] . We would like to stress that while PC-corrected gene expression data was then used as the outcome variable in our gene x environment interaction model , we used gene expression data that was not corrected for PCs to initially create the neutrophil cell percentage proxy . To be able to determine whether a cis-eQTL is mediated by neutrophils , we reasoned that such a cis-eQTL would show a larger effect size in individuals with a higher percentage of neutrophils than in individuals with a lower percentage . However , this required the percentage of neutrophils in whole blood to be known , and cell-type percentage measurements were not available for all of the cohorts . We therefore created a proxy phenotype that reflected the actual neutrophil percentage that would also be applicable to datasets without neutrophil percentage measurements . In the EGCUT dataset , we first quantile normalized and log2 transformed the raw expression data . We then correlated the gene expression levels of individual probes with the neutrophil percentage , and selected 58 gene expression probes showing a high positive correlation ( Spearman R > 0 . 57 ) . Here , we chose to use the quantile normalized , log2 transformed gene expression data that was not corrected for principal components , since correction for principal components would remove the correlation structure between gene expression and neutrophil percentage ( S10 Fig ) . In each independent cohort , we corrected for possible confounding factors due to arrays with poor RNA quality , by correlating the quantile normalized and log2 transformed gene expression measurements against the first PC determined from the sample correlation matrix . Only samples with a high correlation ( r ≥ 0 . 9 ) were included in further analyses . Then , for each cohort , we calculated a correlation matrix for the neutrophil proxy probes ( the probes selected from the EGCUT cohort ) . The gene expression data used was quantile normalized , log2 transformed and corrected for MDS components . Applying PCA to the correlation matrix , we then obtained PCs that described the variation among the probes selected from the EGCUT cohort . As the first PC ( PC1 ) contributes the largest amount of variation , we considered PC1 as a proxy-phenotype for the cell type percentages . Considering the overlap between the cohorts in this study and our previous study , we limited our analysis to the 13 , 124 cis-eQTLs having a significant effect ( false discovery rate , FDR < 0 . 05 ) in our previous study [22] . This included 8 , 228 unique Illumina HT12v3 probes and 10 , 260 unique SNPs ( 7 , 674 SNPs that showed the strongest effect per probe , and 2 , 586 SNPs previously associated with complex traits and diseases , as reported in the Catalog of Published Genome-Wide Association Studies [1] , on 23rd September , 2013 ) . We defined the model for single marker cis-eQTL mapping as follows: Y≈I+β1*G+e where Y is the gene expression of the gene , β1 is the slope of the linear model , G is the genotype , I is the intercept with the y-axis , and e is the general error term for any residual variation not explained by the rest of the model . We then extended the typical linear model for single marker cis-eQTL mapping to include a covariate as an independent variable , and captured the interaction between the genotype and the covariate using an interaction term: Y≈I+β1*G+β2*P+β3*P:G+e where P ( cell-type proxy ) is the covariate , and P:G is the interaction term between the covariate and the genotype . We used gene expression data corrected for 40 PCs as the predicted variable ( Y ) . The interaction terms were then meta-analyzed over all cohorts using a Z-score method , weighted for the sample size [35] . Since the gene-expression data has a correlated structure ( i . e . co-expressed genes ) and the genotype data also has a correlated structure ( i . e . linkage disequilibrium between SNPs ) , a Bonferroni correction would be overly stringent . We therefore first estimated the effective number of uncorrelated tests by using permuted eQTL results from our previous cis-eQTL meta-analysis [22] . The most significant P-value in these permutations was 8 . 15 x 10–5 , when averaged over all permutations . As such , the number of effective tests = 0 . 5 / 8 . 15 x 10–5 ≈ 6134 , which is approximately half the number of correlated cis-eQTL tests that we conducted ( = 13 , 124 ) . Next , we controlled the FDR at 0 . 05 for the interaction analysis: for a given P-value threshold in our interaction analysis , we calculated the number of expected results ( given the number of effective tests and a uniform distribution ) and determined the observed number of eQTLs that were below the given P-value threshold ( FDR = number of expected p-values below threshold / number of observed p-values below threshold ) . At an FDR of 0 . 05 , our nominal p-value threshold was 0 . 009 ( corresponding to an absolute interaction effect Z-score of 2 . 61 ) . For each trait in the GWAS catalog , we pruned all SNPs with a GWAS association P-value below 5 x 10–8 , using an r2 threshold of 0 . 2 . We only considered traits that had more than 20 significant eQTL SNPs after pruning ( irrespective of cell-type mediation ) . Then , we determined the proportion of pruned neutrophil-mediated cis-eQTLs for the trait relative to all the neutrophil-mediated cis-eQTLs . The difference between both proportions was then tested using a binomial test . The source code and documentation for this type of analysis are available as part of the eQTL meta-analysis pipeline at https://github . com/molgenis/systemsgenetics Summary results are available from http://www . genenetwork . nl/celltype Discovery cohorts: Fehrmann ( GSE 20142 ) , SHIP-TREND ( GSE 36382 ) , Rotterdam Study ( GSE 33828 ) , EGCUT ( GSE 48348 ) , DILGOM ( E-TABM-1036 ) , InCHIANTI ( GSE 48152 ) , KORA F4 ( E-MTAB-1708 ) . Replication Cohorts: Stranger ( E-MTAB-264 ) , Oxford ( E-MTAB-945 ) . | Many variants in the genome , including variants associated with disease , affect the expression of genes . These so-called expression quantitative trait loci ( eQTL ) can be used to gain insight in the downstream consequences of disease . While it has been shown that many disease-associated variants alter gene expression in a cell-type dependent manner , eQTL datasets for specific cell types may not always be available and their sample size is often limited . We present a method that is able to detect cell type specific effects within eQTL datasets that have been generated from whole tissues ( which may be composed of many cell types ) , in our case whole blood . By combining numerous whole blood datasets through meta-analysis , we show that we are able to detect eQTL effects that are specific for neutrophils and lymphocytes ( two blood cell types ) . Additionally , we show that the variants associated with some diseases may preferentially alter the gene expression in one of these cell types . We conclude that our method is an alternative method to detect cell type specific eQTL effects , that may complement generating cell type specific eQTL datasets and that may be applied on other cell types and tissues as well . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Cell Specific eQTL Analysis without Sorting Cells |
The Nrf family of transcription factors mediates adaptive responses to stress and longevity , but the identities of the crucial Nrf targets , and the tissues in which they function in multicellular organisms to promote survival , are not known . Here , we use whole transcriptome RNA sequencing to identify 810 genes whose expression is controlled by the SKN-1/Nrf2 negative regulator WDR-23 in the nervous system of Caenorhabditis elegans . Among the genes identified is the synaptic cell adhesion molecule nlg-1/neuroligin . We find that the synaptic abundance of NLG-1 protein increases following pharmacological treatments that generate oxidative stress or by the genetic activation of skn-1 . Increasing nlg-1 dosage correlates with increased survival in response to oxidative stress , whereas genetic inactivation of nlg-1 reduces survival and impairs skn-1-mediated stress resistance . We identify a canonical SKN-1 binding site in the nlg-1 promoter that binds to SKN-1 in vitro and is necessary for SKN-1 and toxin-mediated increases in nlg-1 expression in vivo . Together , our results suggest that SKN-1 activation in the nervous system can confer protection to organisms in response to stress by directly regulating nlg-1/neuroligin expression .
Oxidative stress is generated in cells when an imbalance occurs between the production of electrophilic reactive species and the endogenous defenses against these harmful molecules [1] . Increased reactive oxygen species ( ROS ) can cause damage to cellular components such as lipids , DNA , and proteins , and can result in a myriad of detrimental effects , including protein aggregation , changes in cell signaling , or altered cell cycle progression . Given the nervous system's high metabolic demands , high lipid and iron content , and low regenerative ability , it is not surprising that oxidative stress is particularly detrimental to this tissue; indeed , elevated levels of reactive species have been implicated in a range of neurodegenerative diseases including Parkinson's disease , Alzheimer's disease and amyotrophic lateral sclerosis [2]–[4] . In multicellular organisms , investigating pathways that can mitigate the consequences of elevated oxidative stress on a cellular and organismal level is an ongoing area of investigation . The Nrf ( nuclear factor E2-related factor ) family of transcription factors controls the primary response to oxidative and xenobiotic stress in mammals [5]–[7] . Studies of Nrf2 have established a critical role for the transcription factor in defending tissues from oxidative damage [7] , [8] . During normal conditions , Nrf2 is sequestered in the cytoplasm by the kelch-domain containing protein Keap1; in response to stress , Keap1 releases Nrf2 , allowing the transcription factor to translocate into the nucleus and initiate transcription of downstream targets by binding to the antioxidant response element ( ARE ) in the promoter region of stress response genes [6] , [9] . Nrf2 is ubiquitously expressed , and in neurons , increased Nrf2 activity protects against toxicity by hydrogen peroxide , glutamate , and mitochondrial toxins [10] , [11] . Microarray studies using neuronal cultures lacking Nrf2 have identified detoxifying , antioxidant and defense genes to be regulated by the transcription factor [10]–[12] . Interestingly , these studies have also found changes in the expression of genes involved in a variety of processes , including cell signaling and calcium homeostasis , as well as neuron-specific genes , but the functional significance of these changes has not been determined . In C . elegans , the Nrf family homolog SKN-1 [13] confers cellular and organismal protection from a variety of environmental stressors . Although most studies have examined detoxification and stress response in the intestine , a role for SKN-1 in the nervous system is emerging . SKN-1 functions in a pair of neurons to promote longevity [14] , and systemic knockdown of skn-1 increases dopaminergic neuron degeneration during methylmercury , aluminum and manganese toxicity [15]–[17] . Furthermore , loss of skn-1 enhances manganese-induced organismal death [18] . In C . elegans , the abundance of SKN-1 is negatively regulated in part by the DDB-1/CUL-4 ubiquitin ligase substrate targeting protein WDR-23 , which is proposed to function in an analogous manner as Keap1 to promote degradation of SKN-1 during non-stressed conditions . [19] , [20] . Mutants lacking wdr-23 have increased SKN-1 protein levels , express high levels of genes involved in antioxidant and xenobiotic responses , and are resistant to stress [19]–[21] . In this study , we find that SKN-1 is negatively regulated by WDR-23 in cholinergic motor neurons . Using whole transcriptome RNA sequencing ( RNAseq ) of wdr-23 mutants expressing functional wdr-23a in the nervous system , we identify 810 genes whose expression is likely to be regulated by SKN-1; one of these is the cell adhesion molecule nlg-1/neuroligin . Both nlg-1 expression in neurons and NLG-1 protein abundance at synapses increase in mutants with increased SKN-1 activity as well as in animals exposed to toxins that generate mitochondrial stress . Furthermore , increasing NLG-1 protein abundance enhances survival following toxin treatment , and loss of nlg-1 diminishes SKN-1-mediated toxin resistance . Together , these results support a role for SKN-1 in promoting organismal survival by regulating synaptic neuroligin abundance .
The skn-1 locus encodes three isoforms , skn-1a , skn-1b and skn-1c ( wormbase . org ) , which differ in their N termini and utilize unique transcriptional start sites . skn-1c is primarily detected in the intestine; skn-1b , on the other hand , is expressed principally in a pair of sensory neurons and is involved in dietary restriction induced longevity [13] , [14] . The expression of the longest isoform , skn-1a , however , has not been examined . To examine the expression of skn-1a , we generated a fluorescent reporter by using a 7 . 3 kb fragment upstream of the skn-1a start site to drive gfp fused to a nuclear localization signal ( nls-gfp ) . skn-1a is a downstream gene within the bec-1/beclin operon [14] , [22] , [23] , and the 7 . 3 kb fragment includes the bec-1 coding region and 3 . 3 kb upstream of bec-1 ( Figure 1A ) . In animals expressing this construct , fluorescence was observed in many tissues , as previously reported [24] , [25] , as well as in cholinergic neurons of the ventral cord ( labeled with the unc-17/VAChT promoter expressing mCherry ) and GABAergic neurons ( unlabeled ventral cord neurons; Figure 1B ) . WDR-23 has two isoforms , WDR-23a and WDR-23b , which are expressed in neurons of the ventral cord [19]–[21] . In cholinergic neurons of the ventral cord , a WDR-23a::GFP fusion protein was found in cell bodies and in axons where it adopted a punctate pattern of fluorescence ( Figure 1C ) , whereas WDR-23b::GFP localized exclusively to nuclei ( Figure 1D ) . In axons , WDR-23a puncta co-localized with a synaptic vesicle marker ( Figure 1C ) , and WDR-23a fluorescence became diffuse in animals lacking unc-104/kinesin ( Figure 1E′ ) , which is required for trafficking of organelles along microtubules in neurons [26] . This suggests that WDR-23a associates with presynaptic organelles . WDR-23a also co-localized with a mitochondrial marker at synapses ( Figure 1E ) , and in unc-104 mutants , WDR-23a puncta remained co-localized with displaced mitochondria in axons ( Figure 1E′ ) . In mutants defective in drp-1/Drp1 , a protein required for normal mitochondrial fission [27] , both WDR-23 and mitochondrial markers became displaced in axons ( Figure 1E″ ) . Together , these results suggest that WDR-23a may associate with presynaptic organelles , including mitochondria , at synapses . Consistent with this , WDR-23a co-localizes with outer mitochondrial membrane markers when expressed in muscle cells [20] . To test whether WDR-23 regulates the abundance of SKN-1 in the nervous system , as it does in the intestine [19] , [21] , we examined the abundance of SKN-1a::GFP fusion proteins in motor neurons . We drove SKN-1a::GFP expression using the cholinergic unc-17/VAChT promoter since unc-17 is not transcriptionally regulated by wdr-23 [20] , and therefore changes in SKN-1a::GFP should reflect changes in protein abundance and not skn-1 expression . In wild type animals , SKN-1a::GFP fluorescence was detected in 49% ( n = 272 ) of ventral cord cholinergic neurons ( Figure 2A ) , consistent with low levels of SKN-1 reported in the intestine in non-stressed animals [13] . In wdr-23 mutants , SKN-1a::GFP fluorescence was detected in a larger fraction of neurons compared to wild type animals ( 77% , n = 349 ) . In addition , the average fluorescence intensity of SKN-1a::GFP in neuronal cell bodies of wdr-23 mutants increased compared to wild type controls ( Figures 2A and 2B; mean fluorescence±sem wt: 44 . 7±2 . 8; wdr-23: 62 . 6±3 . 1; p<0 . 001 , Student's t-test ) . Together , these results suggest that WDR-23 negatively regulates the abundance of SKN-1a in motor neurons . wdr-23 mutants have a variety of developmental and behavioral defects including reduced locomotion , resistance to stress , small size , and developmental delay . Each of these defects is suppressed by loss of skn-1 [19]–[21] , suggesting that the primary function of WDR-23 is to negatively regulate SKN-1 . We previously identified 2 , 285 transcripts that are significantly up-regulated in wdr-23 mutants compared to wild type controls using RNAseq [20] . To identify genes among these that are regulated by WDR-23 specifically in the nervous system , we performed RNAseq of wdr-23 mutants expressing an integrated array of full-length wdr-23a cDNA driven by the pan-neuronal snb-1 promoter ( referred to as WDR-23 rescue ) . Of the 2 , 285 genes whose expression increased in wdr-23 mutants , transcripts of 810 were significantly reduced in the WDR-23 rescue animals ( Figure 3A and Table S1 ) . The average expression levels of the rescued genes in wdr-23 mutants was 20 . 74 fold above wild type which is similar to the average of 23 . 99 for all 2 , 285 genes , indicating that the rescue was not caused by a bias generated by rescuing specifically low expressing genes in wdr-23 mutants . We used the Database for Annotation , Visualization and Integrated Discovery ( DAVID ) [28] , to assign GO terms for the 810 rescued genes; of the 492 genes with predicted functions , many genes involved in stress , detoxification and metabolism were identified , consistent with the known stress response roles of skn-1 ( Figure 3B ) . The expression of 177 genes has been experimentally examined ( wormbase . org ) , and 93 of these are expressed in the nervous system ( Figure 3C and Table S1 ) . A subset of these is expressed exclusively in neurons ( Table 1 ) , including eight genes encoding insulins or FMRFamide related peptides , as well as three genes encoding proteins involved in neuropeptide processing—egl-3/proprotein convertase , egl-21/carboxypeptidase , and sbt-2/7B2 . We also identified five genes encoding synaptic proteins , including unc-13/Munc13 , nlg-1/neuroligin , pde-4/cyclic nucleotide phosphodiesterase , dlk-1/MAPKKK , and cab-1/NPDC-1 [29]–[32] . Together , these results suggest that neuronal WDR-23 , possibly through SKN-1 , regulates the expression of genes involved in neuroendocrine signaling and synaptic function . To identify genes that may be direct SKN-1 targets , we examined the promoters of the neuronal genes for the presence of consensus SKN-1 binding sites . SKN-1 is predicted to bind the consensus sequence WWTDTCAT on either strand in the promoters of target genes . Using the web based program Regulatory Sequence Analysis Tools ( RSAT ) , we scanned 1000 bp promoter fragments for each gene for the SKN-1 consensus [33]; the promoter fragments for 70 of the 93 neuronal genes contained at least one potential binding site ( Table S1 ) . We then cross-referenced the list of rescued neuronal genes with SKN-1 ChIP-seq datasets taken from L1 , L3 and L4 stage animals ( ModEncode and [34] ) and found that 31 genes contained significant SKN-1 peaks within 2000 bp upstream or 500 bp downstream of the transcription start site ( Table S1 ) . These results indicate that a subset of the neuronal genes identified by RNAseq may be direct binding targets of SKN-1 . Among the neuronal genes containing SKN-1 consensus sites in their promoter , the cell adhesion molecule nlg-1/neuroligin emerged as an interesting candidate since it has been implicated in C . elegans stress response and in synaptic function [30] , [35] , [36] . nlg-1 is the sole neuroligin family ortholog in C . elegans , and NLG-1 is expressed in cholinergic motor neurons where it localizes to presynaptic terminals [35] , [37] . To examine the effects of skn-1 on nlg-1 expression , we constructed reporter strains consisting of a 3 . 6 kb nlg-1 rescuing promoter fragment [30] driving soluble gfp ( Pnlg-1::gfp , vjIs47 , and vjIs48; Figure 4A ) . In wild type animals , we detected Pnlg-1::gfp fluorescence in a few head neurons and in ventral nerve cord neurons , previously reported to be DA and VA class cholinergic motor neurons [30] , [35] . nlg-1 reporter fluorescence has been reported at low levels in muscles [30] , but we did not detect Pnlg-1::gfp in muscle cells in either transgenic line . In mutants lacking wdr-23 , Pnlg-1::gfp fluorescence in motor neurons increased approximately 5 fold ( Figure 4B and 4C ) , in agreement with the 5 . 6 fold increase in nlg-1 transcript levels detected by RNAseq [20] . Cell-specific expression of either wdr-23a or wdr-23b cDNA in nlg-1 expressing cells ( using the nlg-1 promoter ) fully rescued the increased Pnlg-1::gfp fluorescence to wild type levels ( Figures 4B and 4C ) , indicating that the regulation of nlg-1 expression by wdr-23 is cell autonomous . In skn-1 mutants lacking the skn-1a/c isoforms ( zu67 mutants ) , Pnlg-1::gfp expression was similar to wild type . However , skn-1 is required for the increased Pnlg-1::gfp fluorescence caused by loss of wdr-23 , since skn-1 mutations reduced the Pnlg-1::gfp reporter fluorescence of wdr-23 mutants to wild type levels ( Figures 4D and 4E ) . Conversely , Pnlg-1::gfp fluorescence increased by ∼35% in mutants in which skn-1 is hyperactive ( lax120gf or lax188gf; Figure 4F ) . lax120gf and lax188gf are thought to prevent SKN-1a/c interaction with mitochondrial docking proteins , resulting in an activated pool of SKN-1a/c [38] . Together , these results show that skn-1 is not necessary for baseline nlg-1 expression in motor neurons , but skn-1 activation positively regulates nlg-1 expression . The nlg-1 promoter has four SKN-1 binding consensus sites within 500 bp upstream of the transcriptional start site ( Figure S1 ) . Promoter alignments between the nematode Caenorhabditis species elegans , briggsae , japonica and remanei revealed that one of these sites located 396 bp upstream of the C . elegans start is completely conserved in all species ( Figure S1 ) . This site has the sequence AATGTCAT , which matches the consensus perfectly . The underlined region is predicted to be a largely invariant sequence that directly interacts with SKN-1 [13] , [39] . We mutated AATGTCAT at −396 to AACTGCAG in the Pnlg-1::gfp reporter to create a reporter with a deleted binding site ( the Pnlg-1 ( Δbs ) ::gfp reporter; Figure 4A ) . Basal motor neuron fluorescence of transgenic animals expressing Pnlg-1 ( Δbs ) ::gfp was similar to transgenic animals expressing the Pnlg-1::gfp reporter ( Figure 4B ) . However , Pnlg-1 ( Δbs ) ::gfp reporter fluorescence did not increase in either wdr-23 mutants or in skn-1 ( gf ) mutants compared to wild type controls ( Figure 4B , 4C and 4F ) , suggesting that this site is critical for skn-1-mediated increases in nlg-1 expression . To test whether SKN-1 binds to this site , we performed electrophoretic mobility shift assays . In vitro translated SKN-1a bound to labeled probes containing the putative SKN-1 binding site in the nlg-1 promoter , and binding was disrupted by the addition of excess unlabeled probe ( Figure 4G ) . These results indicate that the SKN-1 binding site at −396 in the nlg-1 promoter can be bound by SKN-1 in vitro and is critical for skn-1-induced expression of nlg-1 in vivo . To determine whether the transcriptional regulation of nlg-1 by SKN-1 impacts NLG-1 protein levels at synapses , we examined synaptic levels of NLG-1 in animals expressing a fusion protein in which GFP was inserted near the C-terminus of NLG-1 ( NLG-1-GFP , vjEx561 and vjIs105 ) . This fusion protein is functional [30] and localizes to presynaptic terminals in motor neurons [35] , [37] . NLG-1-GFP driven by the nlg-1 promoter adopted a punctate pattern of fluorescence in the dorsal and ventral cords , where presynaptic terminals of DA and VA class motor neurons are located , respectively ( Figure 5A ) . We examined changes in the synaptic abundance of NLG-1-GFP by measuring the average punctal fluorescence intensity ( peak fluorescence ) and synapse number ( interpunctal interval ) [40] , [41] . In skn-1 ( gf ) mutants , the punctal fluorescence of NLG-1-GFP significantly increased in both the dorsal and ventral cords , while the interpunctal interval did not change ( Figure 5A , 5B and Table S2 ) . These results indicate that SKN-1 positively regulates synaptic NLG-1 protein abundance but does not affect synapse number . We next tested whether toxins that induce oxidative stress could increase neuroligin expression and abundance in motor neurons . We found that exposure to the mitochondrial stressors juglone or sodium arsenite , both of which have been shown to activate skn-1 [42] , [43] , robustly induced Pnlg-1::gfp fluorescence in motor neurons compared to untreated animals ( Figure 6A and 6B ) . Treatment with an organic mercury ( thimerosal ) increased nuclear SKN-1::GFP in the intestine ( Figure S2 ) and also increased Pnlg-1::gfp fluorescence ( Figure 6B ) . However , these toxins had no effect on Pnlg-1 ( Δbs ) ::gfp fluorescence ( Figure 6A and 6B ) . We found that juglone treatment significantly increased punctal fluorescence of NLG-1-GFP , without changing the interpunctal interval ( Figure 6C , 6D and Table S2 ) . These results suggest that activation of skn-1 by oxidative stress increases NLG-1 synaptic abundance by increasing nlg-1 expression in neurons . To further explore how SKN-1 activity is regulated in the nervous system , we tested the impact of altering insulin signaling , mitochondrial respiration or synaptic activity on Pnlg-1::gfp expression . The insulin-like signaling ( IIS ) pathway regulates SKN-1 in the intestine [44]; activation of DAF-2/insulin-like receptor leads to SKN-1 phosphorylation by SGK-1/SGK , resulting in decreased SKN-1 activity . We examined putative null sgk-1 mutants and found that Pnlg-1::gfp reporter fluorescence increased by ∼25% ( Figure S3 ) . Conversely , during conditions of stress , PMK-1/p38 MAPK phosphorylates SKN-1 , resulting in nuclear SKN-1 translocation [45] . We found no change of baseline Pnlg-1::gfp fluorescence in mutants lacking either sek-1/MAPKK or pmk-1 ( Figure 6E , 6F and S3 ) . However , sek-1 and pmk-1 mutations suppressed the juglone-induced increase in Pnlg-1::gfp reporter fluorescence ( Figure 6E and 6F ) . C . elegans mutants with impaired mitochondrial respiration , for example the conserved clk-1/COQ7 , necessary for the biosynthesis of coenzyme Q , and isp-1/ISP , a subunit of mitochorial complex III , are resistant to toxins that increase oxidative damage [46] . pink-1/PINK1 , on the other hand , is predicted to act in conjunction with the E3 ligase Parkin to initiate mitophagy of damaged mitochondria [47] , [48] . Pnlg-1::gfp expression increased mildly in mutants lacking clk-1 , but not isp-1 . Conversely , loss of pink-1 resulted in a significant decrease of Pnlg-1::gfp fluorescence ( Figure S3 ) . Finally , in order to test whether neuronal activity regulates SKN-1 in the nervous system , we examined mutants with increased synaptic activity ( dgk-1/diacylglycerol kinase or goa-1/Gαo ) and mutants with decreased activity ( unc-2/VGCC or unc-18/nSec1 ) , as well as mutants lacking mef-2/MEF and mir-1/microRNA , which are involved in a retrograde synaptic signaling pathways that is dependent on nlg-1 [35] . We detected no change in reporter fluorescence in these mutants ( Figure S3 ) . Together , these results indicate that insulin signaling and mitochondrial metabolism contribute to the activation of SKN-1 in neurons , while changes in synaptic transmission do not seem to impact neuronal SKN-1 activity . We next sought to determine whether nlg-1 mediates the protective effects of skn-1 activation in response to environmental toxins . skn-1 mutants are hypersensitive to toxicity induced by arsenite and juglone treatment ( Figure 7A and S4; [42] , [43] ) . In addition , we found that skn-1 mutants were sensitive to thimerosal-induced toxicity ( Figure S5 ) , in agreement with studies showing SKN-1 protects from metal toxicity [15] , [18] . In contrast , animals lacking wdr-23 were resistant to toxicity of both thimerosal and juglone , and resistance was completely blocked by loss of skn-1 ( Figures 7A and S5; [43] ) . Similarly , hyperactive skn-1 ( gf ) mutants were resistant to juglone toxicity ( Figure 7B and S6 ) . Interestingly , loss of wdr-23 did not confer protection against sodium arsenite ( Figure S4 ) , suggesting specificity in drug resistance in mutants lacking wdr-23 . nlg-1 mutants are hypersensitive to heavy metal toxicity by thimerosal and oxidative stress induced by paraquat [30] . We found that mutants lacking nlg-1 were also more sensitive to juglone toxicity ( Figure 7A and S5 ) . In contrast , transgenic animals over-expressing NLG-1-GFP were significantly more resistant to juglone-induced toxicity compared to non-transgenic controls ( Figure 7C ) . To confirm that the juglone resistance caused by NLG-1-GFP transgenes was due to nlg-1 expression , we examined juglone responses of animals expressing Pnlg-1::gfp and found that they were not as resistant to juglone as NLG-1-GFP expressing animals ( Figure S7 ) . Finally , nlg-1 mutations dramatically reduced the ability of activated skn-1 to protect animals from the toxic effects of juglone: skn-1 ( gf ) ;nlg-1 double mutants were significantly less resistant to juglone than skn-1 ( gf ) mutants alone ( Figure 7B ) . Together , these data suggest that the dosage of nlg-1 is a critical determinant of survival in response to stress , and that nlg-1 contributes to skn-1-dependent survival .
Previous studies have demonstrated SKN-1/Nrf2 dependent transcriptional programs are initiated in response to oxidative and xenobiotic stress , and these programs are critical for organismal survival and longevity [42] , [43] , [49] . Among the genes regulated by SKN-1 , few have been shown to mediate the protective effects of SKN-1 , and fewer still have been shown to be direct binding targets . Furthermore , attempts to identify a comprehensive set of genes regulated by SKN-1/Nrf2 required to protect organisms from stress have been difficult due to the lack of tissue-level resolution . Here , we have used comparative whole transcriptome RNA sequencing to identify an inclusive set of genes that are likely to be regulated by SKN-1 in the nervous system . For this analysis , we examined wdr-23 mutants expressing a rescuing wdr-23a transgene driven by the pan-neuronal snb-1 promoter . Genetic studies indicate that several distinct phenotypes displayed by wdr-23 mutants are completely suppressed by skn-1 , indicating that SKN-1 is selectively activated in wdr-23 mutants , and a yeast-two hybrid screen identifies the only binding target of WDR-23 to be SKN-1 [19] , indicating that SKN-1 is selectively activated in wdr-23 mutants . Interestingly , wdr-23 mutants appear to activate SKN-1 to a greater extent than skn-1 ( gf ) mutations or toxin treatment . For example , loss of wdr-23 results in greater survival in response to juglone treatment than skn-1 ( gf ) mutations . In addition , wdr-23 mutations increase nlg-1 expression in the ventral cord neurons fivefold , whereas skn-1 ( gf ) or toxin treatment increased it by approximately 30% . Thus , wdr-23 mutants may provide increased sensitivity when used for transcriptional profiling , maximizing our ability to identify SKN-1 targets expressed in low abundance or in a small subset of cells . It is , however , possible that WDR-23 has functions beyond SKN-1 regulation , in which case some of the genes identified by this approach may not be targets of SKN-1 . While the snb-1 promoter fragment drives expression of GFP strongly in the nervous system , the possibility that this promoter may be leaky raises the prospect that some of the genes identified here may be regulated by WDR-23 in other tissues in addition to the nervous system . The 810 genes identified here most likely represent either direct SKN-1 targets or indirect targets of SKN-1 that are secondarily activated in neurons or other tissues . SKN-1 ChIP-seq of larval stage animals identified a list of approximately 3000 genomic peaks bound by SKN-1 in vivo [34]; we cross-referenced our gene list of rescued neuronal genes with SKN-1 ChIP-seq datasets taken from L1 , L3 and L4 stage animals [34] and found that a subset of the genes we identified contain a significant SKN-1 peak near to their transcriptional start sites ( Table S1 ) , but many genes do not . Differences between these datasets may be a byproduct of using non-stressed animals for the ChIP-seq experiments , reflecting basal , but not stress-induced , promoter occupancy by SKN-1 . Among the genes we identified were several neuropeptides and insulins , as well as peptide processing enzymes reported to be expressed in neurons . We speculate that stress-induced peptide processing and release may be part of a humoral response to promote organismal survival . Identification of the precise peptidergic signaling pathways will help to elucidate the mechanisms by which SKN-1 confers survival . We also identified a handful of known synaptic genes , including two cell adhesion molecules , nlg-1 and ncam-1 , indicating SKN-1 might play a role in maintaining the stability of neural networks . A theory has emerged suggesting axonal retraction might precede neuronal apoptosis in neurodegeneration , called dying back degeneration [50] , and synaptic breakdown may occur prior to axonal retraction . The identification of cell adhesion molecules in this study suggests that prior to synaptic breakdown , neurons might initiate transcriptional programs to protect the integrity of the synapse . In support of this , recent evidence suggests that increased expression of NCAM in SH-SY5Y cultures prevents oxidative stress-induced apoptosis , and over-expression of a truncated NCAM molecule protects neuronal tissue in lesioned rats [51] , [52] . Activation of SKN-1/Nrf2 can protect against cell toxicity induced by oxidative stress [53]–[55] . In primary neuronal cultures , for example , activation of Nrf2 prevents cell death in response to rotenone and MPP , potent inhibitors of mitochondrial respiration [11] . Loss of OPA1 , a key regulator of the morphology of mitochondria , results in increased Nrf2 activation [56] . These studies and others collectively suggest a potential role for SKN-1/Nrf2 as a sensor for increased mitochondrial dysfunction in the nervous system . We found that genes involved in drug detoxification , including the glutathione precursors gst-4 , gst-10 , gcs-1 , and gst-1 were reduced by neuronal expression of wdr-23a . Interestingly , gst-1 has also been shown to contribute to dopaminergic neuron survival after manganese treatment [16] . Given their confirmed expression in neurons , is possible that these genes have a role in directly enhancing neuronal protection . SKN-1 associates with purified mitochondrial fractions , and the skn-1 ( gf ) alleles are proposed to reduce mitochondrial association , suggesting that mitochondria may act as a sink for SKN-1 [38] . Our data is consistent with the idea that WDR-23a associates with presynaptic organelles , including mitochondria . First , WDR-23a localizes to presynaptic terminals , where mitochondria are abundant . Second , WDR-23a remains associated with mitochondria in mutants in which mitochondria are displaced . Third , WDR-23a localizes to the outer membrane of mitochondria in muscle cells [20] . We speculate that WDR-23 may be a mitochondrial stress sensor that regulates SKN-1 abundance . Our data supports the idea that SKN-1 activation leads to increased NLG-1 abundance at synapses . We found that nlg-1 confers some , but not all , of the protective effects of activation of SKN-1 , as nlg-1 mutations reduced , but did not eliminate , the resistance of skn-1 ( gf ) mutants to juglone . Additional SKN-1 targets either in neurons or in other tissues are likely to contribute to organismal survival in response to juglone treatment . Interestingly , loss of nlg-1 did not suppress wdr-23 mutants resistance to thimerosal or juglone ( Figure S5 and data not shown ) . This may be due to higher SKN-1 activity in wdr-23 mutants compared to skn-1 ( gf ) mutants , which may compensate for the lack of nlg-1 . Because nlg-1 mutants themselves are hypersensitive to stress , it is possible that nlg-1 and skn-1 function in parallel pathways to promote resistance; however , our data do not support this idea , but rather support the notion that nlg-1 is a direct transcriptional target of SKN-1 . Basal nlg-1 transcription is not likely to be under skn-1 regulation , since nlg-1 reporter expression remains unchanged in skn-1 mutants . Consistent with this , SKN-1 does not occupy the SKN-1 binding site we identified at position −396 in the nlg-1 promoter in unstressed larval animals by ChIP-seq analysis . How might increased expression of a synaptic cell adhesion molecule promote organismal survival in response to stress ? Recent work has established the presence of “mitokines” in neurons—a signal produced in the neurons in response to dysfunctional mitochondrial electron transport [57]; release of these mitokines results in increased organismal survival . It is possible that nlg-1 may be required in neurons for proper release of mitokines . Furthermore , the neuropeptides identified in this study are candidates for being signals released in response to stress to promote resistance . nlg-1 expression has been detected in head neurons , motor neurons , and muscle cells [30]; thus , it is possible that nlg-1 functions in any of these tissues to convey protection . In mammals , neuroligin is a post-synaptic cell adhesion molecule that binds to the presynaptic protein neurexin; this junction is necessary for maintaining mature synaptic connections and normal synaptic transmission [58] , [59] . In humans , rare mutations in neuroligin are associated with autism and other cognitive disorders . Some of these mutations reduce neuroligin delivery to the cell surface , interfering with synapse development and synaptic transmission [60]–[62] . Interestingly , it has been suggested that oxidative stress and mitochondrial dysfunction may play a role in the pathogenesis of autism , as certain biomarkers for oxidative stress are elevated in autistic patients [63]–[65] . It is possible that individuals with these mutations are unable to increase synaptic neuroligin levels in response to stress , and it will be interesting to identify the cellular and molecular mechanisms underlying neuroligin-dependent survival in response to stress .
Strains were cultured at 20° using standard methods . All experiments were performed on young adult hermaphrodites unless otherwise indicated . The following strains were provided by the Caenorhabditis Genetics Center , which is funded by the NIH National Center for Research Resources ( NCRR ) : sek-1 ( km4 ) , pmk-1 ( km25 ) , sgk-1 ( ok538 ) , skn-1 ( zu67 ) , nlg-1 ( ok259 ) , clk-1 ( e2519 ) , mef-2 ( gv1 ) , mir-1 ( gk276 ) , isp-1 ( gm150 ) , and pink-1 ( ok3538 ) . Strain wdr-23 ( tm1817 ) was provided by the National BioResource Project ( Japan ) . The wild type reference strain was N2 Bristol . The following strains were also used: drp-1 ( tm1108 ) , dgk-1 ( nu62 ) , goa-1 ( sa734 ) , unc-2 ( lj1 ) , unc-18 ( md299 ) , glo-1 ( zu391 ) , unc-104 ( e1265 ) , skn-1 ( lax120gf ) , skn-1 ( lax188gf ) , nuIs152[Pttx-3::RFP , Punc-129::GFP::SNB-1]II , nuIs321[Pmyo-2::GFP , Punc-17::mCherry] , nuIs225[Pmyo-2::GFP , Psnb-1::WDR-23a] , idIs7[rol-6 ( su1006 ) , Pskn-1::skn-1b/c::GFP] , yuIs25[Pmyo-2::GFP , Punc-129::mito-GFP]V , and vjIs26[Pmyo-2::GFP , Punc-129::WDR-23a::GFP]III . Mutant strains were outcrossed a minimum of 4 times; all integrants were outcrossed at least 8 times . C . elegans cDNA was used to clone all genes into pPD49 . 26 using standard molecular biology techniques , unless otherwise noted . Promoter elements were amplified from mixed stage genomic DNA . The following plasmids were generated: pTS147[Punc-129::invom::rfp] , pTS31[Pbec-1::nls-gfp] , pDS284[Pnlg-1::gfp] , pDS286[Pnlg-1 ( ΔBS ) ::gfp] , pKG8[Pnlg-1::nlg-1-gfp] , pTS79[Punc-17::skn-1a::gfp] , pDS139[Punc-129::snb-1::mCherry] , pDS237[Psnb-1::wdr-23a::gfp] , pTS85[Psnb-1::wdr-23b::gfp] , pDS334[Pnlg-1::wdr-23a] , pDS335[Pnlg-1::wdr-23b] , and pTS199[T7::skn-1a] . Oligo sequences: Pbec-1 oTS44: ccccccGGATCCcgacaattatacatgttcccc oTS45: ccccccGCTAGCcgactgactggattatgatagatcc Pnlg-1 oDS694: ccccccGCATGCtaagcccccgtacgctaacacc oXL13: ccccccGGATCCgcctgttcacttccaaattcgc Pnlg-1 ( Δbs ) oDS692: cctgttgccccccaaatgCTGCAGttacctcttttcctcccttctacc oDS693 ggtagaagggaggaaaagaggtaaCTGCAGcatttggggggcaacagg Transgenic strains were generated by injecting N2 with expression constructs ( 2 . 5–90 ng/µL ) and the co-injection marker KP#708 ( Pttx-3::rfp , 40 ng/µL ) or KP#1106 ( Pmyo-2::gfp , 10 ng/µL ) . Microinjection was performed using standard techniques as previously described [66] . At least three lines for each transgene were examined for expression , and representative lines were quantified . The following strains were made: vjEx7[Punc-129::wdr-23b::gfp] vjEx663[Punc-129::invom::rfp] , vjEx254[Pbec-1::nls-gfp] , vjEx391[Pnlg-1 ( Δbs ) ::gfp] , vjEx756[Pnlg-1 ( Δbs ) ::gfp] , vjEx561[Pnlg-1::nlg-1-gfp] , vjEx339[Punc-129::snb-1::mCherry] , vjEx423[Psnb-1::wdr-23a::gfp] , vjEx426[Psnb-1::wdr-23b::gfp] , vjEx447[Pnlg-1::wdr-23a] , vjEx436[Pnlg-1::wdr-23b] , vjIs45[Punc-17::skn-1a::gfp]II , vjIs47[Pnlg-1::gfp]IV , vjIs48[Pnlg-1::gfp]I , vjIs105[Pnlg-1::nlg-1-gfp]III . To image animals , adult worms were paralyzed using 2 , 3-butanedione monoxime ( BDM , 30 µg/µL; Sigma ) and mounted on 2% agarose pads for imaging . Images were captured with a Nikon eclipse 90i microscope equipped with a Nikon PlanApo 60× or 100× objective ( NA = 1 . 4 ) and a PhotometricsCoolsnap ES2 camera . For fluorescence imaging of synapses , images were captured either from the ventral or dorsal cord near the posterior gonadal bend of the worm , as indicated . Images of animals expressing nlg-1 reporters were captured at the ventral cord near the posterior gonadal bend of the worm . We found that skn-1 mutants expressing fluorescent integrants arrest at larval stages when the integrant is homozygous , making quantification of fluorescence markers challenging; however , skn-1 mutants expressing heterozygous integrants develop fully into adults . Metamorph 7 . 0 software ( Universal Imaging/Molecular Devices ) was used to capture serial image stacks , and the maximum intensity projection was used for analysis of the dorsal and ventral cords . Line scans of the maximum intensity projection image were also recorded using Metamorph . The fluorescence intensity values were then quantified using Puncta 6 . 0 software written with Igor Pro ( Wavemetrics ) , as previously described [67] . For all experiments , fluorescence values were normalized to the values of 0 . 5 µm FluoSphere beads ( Invitrogen ) captured during each imaging session . This was performed to provide a standard for comparing absolute fluorescence levels between animals from different sessions . To quantify changes in neuronal SKN-1a::GFP , an anterior and posterior image was taken for each animal in both RFP and GFP channels . Cell bodies were identified by the presence of soluble mCherry . Average cell body fluorescence was calculated by outlining the entire cell body and taking the average intensity . Background values were determined by finding the average fluorescence of the area immediately adjacent to the cell body and were subtracted from each cells' average fluorescence . Cells were categorized as low , medium , or high expressing cells using arbitrary cut off levels after background subtraction—low expressing cells were those cells below the level of detection ( less than 30 units different than background fluorescence ) , medium expressing cells were between 30–70 arbitrary units , and high expressing cells had a total fluorescence greater than 70 units . Stock solutions of 50 mM juglone ( Calbiochem ) and 20 mM thimerosal ( Enzo ) were freshly dissolved in DMSO or water , respectively , prior to addition to molten NGM . Sodium arsenite ( Ricca ) was maintained in aqueous solution at 0 . 5% w/v and stored at room temperature . Plates were freshly made approximately 24 hours before use and seeded with concentrated OP50 the night before being used . To assess longevity , age matched young adult animals were transferred to 100 mm NGM plates containing either drug or control and assayed over time . Animals which escaped the plates were excluded from the analysis . At least four replicates of n = 40 animals per genotype per stress were tested; final samples sizes reported in Table S3 . Animals were stored at 20° except during time points . Animals were scored as dead if they did not respond to repeated light prodding . Percentages alive for each genotype were determined by averaging the fraction alive per plate at each time point and plotting graphically . For fluorescence toxicity studies using the nlg-1 reporters , L4 stage animals were exposed to NGM plates supplemented with drug or control overnight for 14 hours; animals were allowed 2–4 hours recovery time before imaging . Concentrations were chosen that did not result in animal death after 14 hours . For juglone imaging , control plates were supplemented with an equal volume of DMSO , as the juglone was diluted to 50 mM in DMSO prior to addition to the test plates . A Student's t test was used to determine significance when comparing fluorescence of nlg-1 reporters in different conditions , unless otherwise specified . Log rank tests with a Bonferroni correction were calculated by JMP Pro version 10 . 0 software and were used to determine significant differences between genotypes for toxicity studies; differences between genotypes are reported in Table S3 . Total RNA was isolated from approximately 10 , 000 mixed stage animals for wild type , wdr-23 ( tm1817 ) mutants and wdr-23;nuIs225 using Stat60 ( Tel-test B , Texas ) . Transcriptome libraries were prepared using TruSeq RNA sample preparation kit ( Illumina ) according to manufacturer's TruSeq protocol as previously described [20] . Libraries were amplified by PCR and quality and quantity of libraries were evaluated on BioAnalyzer 2100 ( Agilent ) . Sequencing was performed on HiSeq2000 ( Illumina ) . Sequencing reads were aligned to the C . elegansgenome ( release WS210 ) using TopHat [68] . Gene models were downloaded from ENSEMBL and quantified using Cufflinks . Differentially expressed genes at false discovery rate ( FDR ) of 0 . 05 were identified using the Cuffdiff module of the Cufflinks package . Full-length SKN-1a cDNA was cloned into a pBS backbone driven by the T7 promoter and expressed using TnT T7 Quick Coupled Transcription/Translation System for DNA ( Promega ) . EMSA was performed using the LightShift Chemiluminescence EMSA Kit ( Pierce ) according to manufacturer's protocols . Complementary 5′ biotinylated oligonucleotides containing the pnlg-1 SKN-1 binding site were self-annealed and incubated with 1 µl of SKN-1 lysate , 1 µg Poly ( dIdC ) and 5 mM MgCl2 for 20 minutes at room temperature . Samples were separated on a 5% native polyacrylamide gel and blotted on Biodyne B nylon membranes ( Pierce ) . Pnlg-1 probes: oTS300: 5′ biotin-gttgccccccaaatgATGACATTacctcttttcctccc 3′ oTS310: 5′ biotin-gggaggaaaagaggtAATGTCATcatttggIgggcaac 3′ | Organisms have evolved mechanisms to protect themselves at the cellular level in response to a variety of environmental stresses . Oxidative stress , caused by an imbalance in the cellular production of free radicals and endogenous antioxidant defenses , can be particularly detrimental to the nervous system . Indeed , elevated levels of oxidative stress have been linked to nearly all neurodegenerative diseases . Therefore , understanding how living creatures protect themselves against oxidative stress , from a cellular to a systemic level , is vital . Here , we have found that increased stress activates a pathway that increases the amount of certain proteins found at neuronal synapses . This presents an interesting model in which , in response to stress , neurons might attempt to enhance the strength of the synapse to prevent degeneration . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"animal",
"models",
"caenorhabditis",
"elegans",
"cellular",
"stress",
"responses",
"model",
"organisms",
"biology",
"molecular",
"cell",
"biology"
] | 2014 | Regulation of Synaptic nlg-1/Neuroligin Abundance by the skn-1/Nrf Stress Response Pathway Protects against Oxidative Stress |
Switching of the Variant Surface Glycoprotein ( VSG ) in Trypanosoma brucei provides a crucial host immune evasion strategy that is catalysed both by transcription and recombination reactions , each operating within specialised telomeric VSG expression sites ( ES ) . VSG switching is likely triggered by events focused on the single actively transcribed ES , from a repertoire of around 15 , but the nature of such events is unclear . Here we show that RNA-DNA hybrids , called R-loops , form preferentially within sequences termed the 70 bp repeats in the actively transcribed ES , but spread throughout the active and inactive ES , in the absence of RNase H1 , which degrades R-loops . Loss of RNase H1 also leads to increased levels of VSG coat switching and replication-associated genome damage , some of which accumulates within the active ES . This work indicates VSG ES architecture elicits R-loop formation , and that these RNA-DNA hybrids connect T . brucei immune evasion by transcription and recombination .
The genome provides the blueprint for life and is normally protected from rapid content change by high fidelity DNA replication and a range of repair pathways . However , strategies for elevated rates of genome variation have evolved , some of which are genome-wide , such as in developmental chromosome fragmentation in ciliates [1] and chromosome and gene copy number variation during Leishmania growth [2 , 3] . More commonly , enhanced genome change is more localised and caused by deliberate lesion generation , such as during yeast mating type switching , which is induced by HO endonuclease-mediated cleavage in Saccharomyces cerevisiae [4] and locus-directed replication stalling in Schizosaccharomyces pombe [5] . Rearrangements to generate mature receptors and antibodies expressed by T and B cells [6] occur throughout mammalian immune genes and are generated by RAG endonuclease-catalysed DNA breaks [6] or transcription-linked base modification [7] . Reflecting the diversity in routes capable of initiating genome change , homologous recombination ( HR ) , non-homologous end-joining and microhomology-mediated end-joining repair reactions have been implicated in the catalysis of these different reactions . Antigenic variation is a very widespread pathogen survival strategy , involving stochastic switches in surface antigens to thwart host adaptive immunity [8] , and locus-directed gene rearrangement is a common route for the differing reactions used in bacteria , fungi and protists [9 , 10] . Since antigenic variation impedes vaccination , it is important to understand the potentially pathogen-specific events that initiate surface antigen gene switching . However , only in the bacteria Neisseria gonorrhoeae is initiation well understood; here , HR catalyses movement of silent , non-functional pilS genes into a pilE expression locus via transcription-induced guanine quadruplex formation [11] . In no other pathogen has the initiation event ( s ) during antigenic variation been resolved in this detail . Antigenic variation in T . brucei displays remarkable mechanistic complexity , since it involves both HR-directed rearrangement and transcriptional control of VSG genes . In any cell a single VSG is expressed by RNA Polymerase ( Pol ) I transcription of one of the multiple telomeric ES [12 , 13] . In each ES the VSG is proximal to the telomere and is co-transcribed with multiple ESAGs ( expression site associated genes ) [14] . Invariably , the VSG and ESAGs are separated by stretches of 70 bp repeats , sequences also found upstream of thousands of further VSGs in the T . brucei genome [15–17] . To execute a VSG coat switch , T . brucei can use HR to move one of around 1 , 000 silent subtelomeric VSGs into the active VSG ES , displacing the resident VSG , or silence transcription from the active VSG ES and activate transcription from one of the silent VSG ES . Though a wide range of factors have been described that influence singular ES expression [12 , 18–21] and execute VSG recombination [22] , initiation of VSG switching remains poorly understood . Targeting yeast I-SceI endonuclease activity to the active VSG ES elicits VSG recombination [23 , 24] , but no endogenous ES-focused endonuclease has been described . Impaired telomere protection results in VSG ES breaks [25 , 26] and critically short telomeres have been associated with increased VSG switching [27] , but how such processes act in unperturbed cells is unknown . Finally , DNA replication mapping suggests that the active VSG ES , uniquely amongst these telomeric loci , is subject to early replication in mammal-infective ( bloodstream form , BSF ) parasite cells [28] , but how this extrapolates to VSG switch initiation is unclear [29] . R-loops are stable RNA-DNA hybrids that form within the DNA helix , displacing single-stranded DNA . Though R-loops can arise from transcription , active roles are emerging in an range of genomic processes [30] , including replication initiation [31 , 32] and arrest [33] , transcription activation and termination [30] , telomere homeostasis [26 , 34] and chromatin formation [35 , 36] . In addition , R-loops can lead to genome instability and mutation [37–39] . To date , only one example of R-loop involvement in targeted genome rearrangement has been described: class switch recombination to alter the antibody type expressed by mature mammalian B lymphocytes [40–42] . Amongst a potentially wide range of activities described that prevent , reverse or interact with R-loops [30 , 43 , 44] , RNase H enzymes function to degrade the RNA within the hybrids . In virtually every cellular organism , two distinct ribonuclease H enzymes , which are termed RNase H1 and RNase H2 in eukaryotes , have been described [45] . Here , we describe R-loop distribution in the VSG ES of both wildtype BSF T . brucei and in null mutants that lack a homologue of RNase H1 . We show that R-loops accumulate throughout all VSG ES in the absence of the RNase H enzyme , indicating RNA-DNA hybrids form in these transcription sites and are normally resolved by removing the RNA . Loss of the RNase H results in elevated levels of replication-associated damage and leads to increased VSG coat switching , suggesting a model for the events that initiate antigenic variation in T . brucei and potential parallels with R-loop directed mammalian class switch recombination .
T . brucei encodes an RNase H1 homologue ( TbRH1 ) that has been predicted to be nuclear by fusing an N-terminal fragment to GFP [46] . By expressing full-length TbRH1 C-terminally fused to 12 copies of the myc epitope ( S1A Fig ) , from its own locus , we confirmed nuclear localisation ( Fig 1A ) , since anti-myc signal co-localised with the larger ( nuclear ) DAPI signal and not the smaller ( kinetoplast ) signal . Expression of TbRH1-myc was constitutive throughout all discernible cell cycle stages of BSF T . brucei cells and did not display any obvious sub-nuclear localisation ( Fig 1B ) , though increased signal appeared present in cells undergoing nuclear replication ( S1B Fig ) . By integration of TbRH1 targeting constructs ( S2A Fig ) we generated heterozygous ( +/- ) and then homozygous ( -/- ) TbRH1 mutants ( S2B Fig ) . No growth perturbation ( Fig 1C ) or alteration in cell cycle stage distribution ( Fig 1D ) was apparent in the mutants , indicating TbRH1 does not provide essential genome functions ( at least in culture ) , unlike mammalian RNase H1 [47] . However , it should be noted that the analyses described here cannot exclude the potential for secondary mutations during the generation of Tbrh1-/- cells , or uncharacterised adaptation following loss of the RNase H enzyme . To ask if TbRH1 targets R-loops within the VSG ES , we performed RNA-DNA immunoprecipitation coupled to next generation sequencing ( DRIP-seq ) [48] in both wild type ( WT ) and Tbrh1-/- cells , aligning DNA reads to the VSG ES using MapQ filtering [49] to ensure ES-specific mapping of Illumina short reads across regions of homology ( Fig 2 , S3 Fig ) . Though a range of strategies to isolate and map R-loops might have been considered [50] , DRIP-seq proved valuable in yeast [48 , 51 , 52] and we therefore deemed it appropriate as a means to provide the first genome-wide view of R-loop distribution in T . brucei [53] . Here , we have focused on R-loop distribution in the VSG ES . In WT cells there was limited read enrichment across the ES region spanning the promoter to the VSG , either in the actively transcribed ES ( BES1 , containing VSG221 ) or the 13 distinct silent ES ( Fig 2A , S3 Fig ) . Pronounced enrichment was only observed proximal to the ends of the ES , downstream of the VSG , which most likely represents TERRA RNA since levels of the signal appeared to increase in Tbrh1-/- mutants ( Fig 2A and 2E ) , the opposite of decreased TERRA RNA when TbRH1 is over-expressed [26] . Loss of TbRH1 resulted in DRIP-seq signal throughout all ES , both active and silent ( Fig 2A , S3 Fig ) . To check the mapping , we performed qPCR on DRIP samples ( Fig 2B ) . Enrichment of sequence in the IP relative to input ( non-IP ) was substantially higher ( ~10 fold ) from Tbrh1-/- cells relative to WT for two ESAGs ( 6 and 8 ) , confirming intra-ES R-loops and the increased DRIP-seq signal in the ES of Tbrh1-/- mutants compared with WT . The same differential between mutant and WT was also seen with qPCR using primers recognising VSG221 ( BES1 , active ) or VSG121 ( BES3 , inactive ) , confirming R-loops in both transcribed and untranscribed sites . Finally , on-bead treatment of samples with E . coli RNase HI prior to DNA recovery clearly reduced the IP enrichment in the Tbrh1-/- cells , confirming recovery of RNA-DNA hybrids . To examine the distribution and abundance of ES R-loops further , we performed K-means clustering of the read alignments from WT and Tbrh1-/- mutant DRIP-seq , separating the analysis into three ES components: the ESAG-containing region from the promoter to the 70 bp repeats ( Fig 2C ) , the 70 bp repeats ( Fig 2D ) , and the VSG plus 500 bp of flanking sequence ( Fig 2E ) . In all components of the ES , as expected , read abundance was greater in the Tbrh1-/- mutants than WT . However , the extent and pattern of enrichment was not equivalent in the three components , and nor was it always equivalent in active and silent ES . For the ESAG and 70 bp repeat components ( Fig 2C and 2D ) , clustering analysis separated the active ES from all silent ES both in WT and Tbrh1-/- cells , suggesting differences dictated by transcription . In contrast , no such separation was seen around the VSG , where the active and silent ES could not be distinguished ( Fig 2E ) . Indeed , the level of signal across the VSG ORFs was relatively low compared with upstream and , in particular , downstream ( presumably telomeric ) regions , despite low levels of R-loops being present ( as confirmed by VSG DRIP-qPCR; Fig 2B ) . Enrichment of DRIP-seq signal in Tbrh1-/- cells relative to WT extended across the ESAGs ( Fig 2C ) and did not appear to be due to localisation to any specific sequence elements , such as the ORFs or untranslated intergenic regions ( Fig 2A , S3 Fig ) , indicating R-loops became more abundant throughout the region of potential transcription upstream of the 70 bp in the absence of TbRH1 . In contrast , the 70 bp repeats were notable for three features ( Fig 2D ) . First , the level of enrichment across the repeats was notably higher in Tbrh1-/- mutants relative to WT than in both other components of the ES ( ~2 fold in the 70 bp repeats , compared with ~1 . 5 fold elsewhere; see also Fig 2A , S3 Fig ) . Second , in WT cells the small signal levels in the inactive ES were notably lower across the 70 bp repeats than surrounding sequence , whereas in Tbrh1-/- cells signal was greater across the repeats than the flanks . Third , a different pattern was seen in the active ES: in WT cells there was no DRIP-seq signal ‘dip’ within the 70 bp repeats , and in the Tbrh1-/- mutants the signal was more enriched at promoter-proximal repeats than telomere-proximal , following the direction of transcription . Taken together , the clustering analysis indicates TbRH1 plays a key role in removing R-loops in the VSG ES , with the 70 bp repeats being a focus for accumulation of the hybrids; furthermore , the distinct features of the mapping within the active ES relative to inactive ES suggest R-loop accumulation and removal by TbRH1 is co-transcriptional in WT cells . Given that R-loops accumulate within the VSG ES in the absence of TbRH1 , we next asked if the increased abundance of the RNA-DNA hybrids is associated with altered VSG expression in Tbrh1-/- mutants relative to WT . To test this association , we first performed RT-qPCR to measure RNA levels of a selection of ES VSGs ( Fig 3A ) . Five VSGs within silent ES in the Lister 427 T . brucei strain used here [14] ( Fig 3A ) displayed significantly higher RNA abundance in the Tbrh1-/- cells relative to WT . In addition , a small reduction ( p value 0 . 0284 ) in RNA levels was observed for VSG221 , which is present in the predominantly active VSG ES in WT cells ( BES1; Fig 2 ) . To ask if RNA changes are limited to ES VSGs , we performed RNA-seq on RNA from the WT and Tbrh1-/- cells and mapped the reads to all available annotated VSGs in the Lister 427 T . brucei strain [14 , 16] . In total , 63 VSGs displayed 1 . 5 fold or greater numbers of mapped RNA-seq reads in the Tbrh1-/- mutants compared with WT ( Fig 3B ) . Amongst these genes were nine bloodstream ES VSGs , of which four showed particularly pronounced read increases , though comparable changes in read depth were not obvious for the associated ESAGs within the ES ( see VSG121 in BES3 , S4 Fig ) . Though the RNA-seq and RT-qPCR data do not wholly match ( for instance , RT-qPCR predicts some increased VSG224 RNA in the mutants , whereas RNA-seq predicts a decrease; Fig 3 and S4 Fig ) , it should be noted that the two approaches used independently grown cells . Increased RNA-seq reads in the mutants were also detected for VSGs from all parts of the silent archive , with intact and pseudogenic array VSGs more frequently detected than metacyclic ES or minichromosomal VSGs ( Fig 3B and 3C ) . Taken together , the RT-qPCR and RNA-seq data indicate that loss of RNaseH1 leads to increased levels of transcription from normally silent VSGs , including some not predicted to be resident in VSG ES . To test if these RNA changes extend to the VSG surface coat , we used immunofluorescence on unpermeabilised cells to evaluate the stability of VSG221 expression , since this VSG is normally resident in the predominantly transcribed ES ( BES1 ) . To do this , we first examined VSG221 expression over time , comparing the frequency with which three Tbrh1-/- and WT clones no longer expressed the protein during prolonged passage ( Fig 4A ) . Despite the absence of immune selection against VSG221 expression , greater numbers of cells without surface VSG221 were seen in the Tbrh1-/- cells than in WT at each time point examined , indicating elevated levels of VSG switching throughout growth in culture . Notably , such elevated VSG switching is not associated with changes in population doubling time of the RNaseH1 mutants . To examine this effect in more detail , we next performed co-immunofluorescence ( IF ) on unpermeabilised WT and Tbrh1-/-cells ( grown for >45 generations in culture ) using antiserum recognising VSG221 ( active BES1 ) or VSG121 ( silent BES3 ) . In this WT population , all cells analysed expressed VSG221 ( Fig 3B ) , whereas ~3 . 5% of the Tbrh1-/- mutant cells no longer expressed VSG221 on their surface ( Fig 3B and 3C ) . Most of the cells ( ~3 . 1% ) that did not react with VSG221 antiserum also did not react with VSG121 antiserum , indicating they expressed a distinct VSG or VSGs on their surface . However , in a small proportion of cells ( ~0 . 35% ) VSG121 could be detectably expressed , indicating this gene had been activated . To determine if all VSG121-expressing Tbrh1-/- cells had switched off VSG221 , we looked amongst the VSG121-expressing cells for co-staining with both antisera ( Fig 3C and 3D ) . As a control , immunofluorescence was also performed in a distinct WT strain ( i . e . not lacking TbRH1 ) in which a transcription elongation blockade within BES1 has silenced this ES and predominantly activated BES3 [28 , 54 , 55] , containing VSG121 ( Fig 3C and 3D ) . Most ( ~68% ) Tbrh1-/- cells stained only with anti-VSG121 antiserum , though a minority ( ~32% ) were VSG221-VSG121 double expressers . Taken together , these data indicate that loss of TbRH1 results in an increased frequency at which expression of the active VSG is lost , which mainly reflects complete VSG switching events where the active VSG is no longer detected and expression of a distinct VSG occurs . Loss of mono-allelic control , which results in co-expression of VSGs from the active and at least one previously silent ES , is less common but was observed . Two models might be considered to explain elevated VSG switching in Tbrh1-/- mutants relative to WT cells ( Fig 5 ) . In one model , ES R-loop accumulation impedes complete transcription of the active ES , selecting for cells in which a previously silent ES has been transcriptionally activated . Once activated , these newly expressed ES then accumulate R-loops , propagating the DRIP-seq signal across all ES ( Fig 2 ) . However , R-loops have also been linked to DNA breaks and rearrangement [39] , through impeding DNA replication [56–60] , as a result of elevating the levels of transcription-associated breaks [61–63] , or because the RNA-DNA hybrids form in response to transcription-associated breaks [64–70] . A second model , therefore , is that increased R-loops in Tbrh1-/- cells reflect the accumulation of damage in the ES , leading to recombination-based VSG switching . To try and separate these models , we compared levels of nuclear genome damage in WT and Tbrh1-/- cells by assessing expression of Thr130-phosphorylated histone H2A ( γ-H2A ) , which increases in abundance after a range of genotoxic insults [71 , 72] and in repair mutants [28] . Western blotting suggested comparable levels of overall γ-H2A in WT and Tbrh1-/- cells ( S5 Fig ) . However , IF analysis revealed a ~2 . 3 fold increase in the number of Tbrh1-/- cells with detectable nuclear γ-H2A signal , rising from around 7% in WT ( Fig 6A ) . Super-resolution structure-illumination microscopy revealed that in both WT and Tbrh1-/- cells most γ-H2A signal appeared as a single subnuclear focus , though some cells with >1 foci were present ( Fig 6B; examples in Fig 6D and S6 Fig ) . In addition , some cells displayed diffuse staining throughout the nucleus ( Fig 6B ) , suggesting γ-H2A signal may represent various types of damage . DAPI staining of a T . brucei population provides a means to determine the cell cycle stage of individual cells , since replication and segregation of the nuclear ( N ) and kinetoplastid ( K ) genomes occur with different timings [73] . In keeping with previous work [71] more WT cells displayed γ-H2A signal ( Fig 6C ) when they were undergoing nuclear replication ( 1N1eK ) or were in G2-M phase ( 1N2K ) , with reduced numbers of signal-positive cells from the end of M phase ( 2N2K ) through G1 ( 1N1K ) . Cell cycle quantification of the Tbrh1-/- mutants showed that the increased proportion of cells with γ-H2A signal was nearly entirely accounted for by greater numbers of 1N1eK ( ~2 . 4-fold increase ) or 1N2K ( ~3 . 6-fold ) cells with foci relative to WT ( Fig 6C ) , indicating increased accumulation of nuclear damage occurs during replication of the genome in the absence of TbRH1 . To ask if some of the damage detected by microscopy of γ-H2A localises to the VSG ES , we performed ChIP-seq with anti-γ-H2A antiserum in WT and Tbrh1-/- cells , mapping the reads to the 14 ES using MapQ filtering ( Fig 6E , S7 Fig ) . In WT cells low levels of γ-H2A ChIP reads were seen in all ES , but the level of enrichment was notably greatest and most widespread in the active ES ( BES1 , containing VSG221 ) , in particular around the VSG and 70 bp repeat-proximal ESAGs . In the Tbrh1-/- cells , γ-H2A ChIP reads were detected at even greater levels in the active ES ( BES1 ) , both at the locations detected in WT cells and due to increased reads more proximal to the ES promoter . In contrast , though γ-H2A ChIP reads also increased in the silent ES of Tbrh1-/- cells ( Fig 6E , S7 Fig ) , the extent of this change was more modest than was seen in the active ES . These data indicate DNA damage is present in the active ES , in particular proximal to the 70 bp repeats and VSG , where distinct assays have suggested the presence of DNA breaks [23–25] . As the extent of γ-H2A signal increases after ablation of TbRH1 , this response correlates with the increased abundance of R-loops , which may form preferentially in the active ES .
In this work we reveal that interplay between RNA Pol I transcription and sequence composition of the VSG ES leads to R-loops acted upon by RNase H1 , and that loss of the ribonuclease results in increased replication-associated damage , including in the VSG ES , and increased VSG coat switching . These findings suggest VSG transcription and VSG-associated sequences may have evolved to generate R-loops during the expression of the trypanosome’s crucial surface antigen , indicating RNA-DNA hybrids may be harnessed to provide pathogen-specific , discrete functions , such as antigenic variation . These data provide only the second example we are aware of for R-loops providing locus-targeted rearrangement , adding to established roles for R-loops in the initiation of mammalian immunoglobulin class switching [42] . The data we present provide insight into the initiation of antigenic variation in T . brucei ( Fig 5 ) . We suggest that transcription through the VSG ES leads to the formation of R-loops that appear to be rapidly resolved , including by TbRH1 , to ensure continued high rates of VSG coat expression [74] . Within the ES the 70 bp repeats appear to be a pronounced site of VSG RNA-DNA hybrid accumulation , given the strongest R-loop ES enrichment is seen in this location in Tbrh1-/- mutants . R-loops may form more readily on the 70 bp repeats due to their sequence composition: individual 70 bp repeats show considerable size and sequence variation but are , in part , comprised of ( TRR ) repeats [15 , 75] that can become non-H bonded [76] and promote recombination [77] when transcribed . Increased DRIP-seq signal across the active VSG ES of Tbrh1-/- mutants might therefore be most readily explained by R-loops forming initially in the 70 bp repeats and extending back toward the RNA Pol I promoter due to retrograde spreading of the VSG ES transcription blockade . This scenario explains the less pronounced enrichment of R-loops within the telomere-proximal VSGs . In addition , it explains the localisation of R-loops both to ES coding and non-coding sequences , which contrasts with the predominant R-loop distribution in multigene transcription units elsewhere in the genome , where R-loops show a very pronounced intragenic localisation [53] . Thus , it is very unlikely that R-loops in the ES are associated with transcription processing , but instead emerge from impaired RNA Pol I movement . However , as we discuss below , R-loops may not only accumulate in RNase H1 mutants because transcription is impeded across the ES , but because of lesions forming within the transcription units . Our data suggest that R-loop accumulation leads to VSG switching and spreading of R-loops into silent VSG ES by two routes: transcriptional and recombinational . The transcriptional model is consistent with the observation that RNAi-mediated loss of VSG expression from the active ES is lethal [74] and can select for switching [78] . Co-transcriptional R-loop formation might itself lead to reduced transcription of the active VSG ES , or R-loops might form because of lesions generated during ES transcription ( see below ) . Irrespective , once R-loops form in the ES , in particular in the absence of TbRH1 , they could provide a blockade to full transcription , causing further R-loops to form in the ES and providing a selection for activation of a new ES . Once a new ES is activated , transcription in this locus then gradually suffers the same R-loop blockade , which is exaggerated by the absence of TbRH1 . The model is consistent with the presence of cells expressing both VSG221 and VSG121 , since co-expression of at least two VSGs indicates altered transcriptional control . However , such double expressers are relatively rare and the seemingly equivalent accumulation of R-loops in both the active ES and in all silent ES is not simple to reconcile with the continued predominant transcription of VSG221 ( BES1 ) in the Tbrh1-/- mutants . Therefore , a second possibility is that R-loops do not merely reflect or cause transcription blockade , but are associated with recombination of silent VSGs ( and in some circumstances ESAGs from silent ES ) into the active ES . An R-loop VSG recombination model is consistent with the greater levels of γ-H2A signal in the active ES and , in this scenario , the relatively similar level of R-loop enrichment in both the active and silent ES could be explained by gene conversion of silent ES sequences into the active ES . However , the lack of evidence for increased RNA spanning the silent ES ( such as BES3 ) in the Tbrh1-/- mutants contrasts with increased RNA-seq reads for multiple ES VSGs , and does not readily match the greater density of γ-H2A signal proximal to the 70 bp repeats in the mutants . Instead , the γ-H2A ChIP mapping appears consistent with damage initially arising in the telomere-proximal region of the ES , perhaps due the nature of the repeats ( above ) . If correct , this may explain the RNA-seq data , with telomere-proximal damage leading to the activation not simply ( or even predominantly ) of silent ES VSGs , but any VSG in the silent archive . In this model , accumulation of R-loops throughout the silent ES may indicate trans formation of the RNA-DNA hybrids after their generation in the active site , rather than transcription-associated formation . Irrespective of the precise details , the accumulated data presented here suggests that R-loop-associated recombinational switching predominates over R-loop-associated transcriptional switching . Nonetheless , since we could detect Tbrh1-/- cells in which at least two VSGs were expressed on the cell surface , a commonality of R-loops acting during transcription-mediated and recombination-mediated antigenic variation in T . brucei is possible . In other words , these reactions may not be entirely independent in mechanisms but , instead , share a common initiating event . Such commonality might explain why ablation of the homologous recombination factors RAD51 , BRCA2 and RAD51-3 impairs both VSG gene conversion and transcriptional switching [79–81] , as well as why induction of DNA damage can elevate levels of silent VSG expression [82] . Precisely how R-loops intersect with T . brucei VSG switching will need to be explored further , though similarities and differences can be drawn with R-loop involvement in mammalian Ig class switch recombination . In the mammalian IgH locus , R-loops form due to transcription of non-coding RNA upstream of Ig constant region exons [83–85] , with evidence for isotype transcription specificity [86] . In T . brucei each VSG ES is multigenic , with the VSG and ESAGs transcribed from a common upstream RNA Pol I promoter , and no stable intra-ES non-coding RNAs have been described . Thus , it seems likely that R-loops mediating T . brucei VSG switching are less gene- or exon-specific than those directing mammalian Ig class switching . If so , this difference may explain the mechanistic connection we propose between VSG switching through transcriptional and recombinational routes , whereas class switching is strictly a recombination-catalysed reaction . Nonetheless , our data suggest the 70 bp repeats within the VSG ES may be a focus for R-loop formation , perhaps suggesting some broad similarity with the role of G-rich sequences around Ig exons [87] , such as a need to use DNA features to target R-loops . However , in class switch recombination the key role of G-rich sequences appears to be in generating RNA quadruplexes , which recruit the cytidine deaminase enzyme AID [85]; such a role is in keeping with Ig exon expression specificity in B lymphocytes , since G quadruplexes are abundant throughout mammalian genomes but rarely form such structures when transcribed into RNA [88] . 70 bp repeats , in contrast , appear to be nearly exclusively associated with VSGs [89] and no evidence has been documented that they form stable RNAs . Despite the above potential similarities , execution of Ig class switching and VSG switching by recombination is very different . During Ig class switching RNA quadruplexes target AID to the Ig exons , at which time the enzyme catalyses insertion of uracil residues , leading to DNA double strand breaks ( as a result of mismatch and base excision repair ) that result in exon deletion through non-homologous end-joining [42] . T . brucei VSG switching involves homologous recombination [22] , and how such a difference in gene rearrangement strategy might emerge from a common R-loop intermediate is unclear . To date , no enzymatic activity , akin to AID , has been described to generate the DNA lesions expected to initiate VSG switching . It is possible that accumulation of R-loops due to ES transcription pausing might alone be enough to generate the increased damage we detect in the active ES . For instance , prolonged negative supercoiling downstream of RNA Pol I might increase the likelihood of R-loop formation and greater exposure of the single-stranded DNA in the RNA-DNA hybrid . Alternatively , R-loops may not be the cause of damage but may form in response to the formation of lesions by another route . Both possibilities are consistent with the detection of putative DSBs within VSG ES [23–25] . Determining whether R-loops cause or respond to DNA breaks remains challenging in any setting [38 , 64] , since it is clear that chromatin and repair pathways can modulate the damaging effects of R-loops [90–93] , while at the same time single- or double-stranded DNA breaks can induce R-loop formation [65 , 70] . Key to understanding how R-loops act in VSG switching might be examination of the factors that act on these structures , such as DNA or RNA helicases [94] . In the above scenarios it is assumed that R-loops form during transcription , or that transcription of the ES provides access to a lesion-forming machinery . It is equally possible that R-loops form not simply due to transcription , but impede DNA replication through the active VSG ES and lead to breaks that elicit a VSG switch . This final route would be compatible with the elevated levels of γ-H2A in replicating Tbrh1-/- cells , and with previous observations showing the active VSG ES replicates earlier than all the silent ES [28] . In other words , it is possible that ES structure and function has evolved to target replication-transcription clashes to the site of VSG expression to facilitate switching . If so , there may be a parallel with Ig class switching , since R-loops in the IgH locus recruit replisome components and lead to DNA replication [95] . Whether such a putative connection between R-loops , transcription and replication in the VSG ES could recruit repair machinery related to that described in yeast and mammals [90 , 96 , 97] is unknown . Currently none of the above scenarios can be ruled out , but the effects of TbRH1 loss on inducing VSG switching appear consistent with observations in other eukaryotes , and with the previously characterised roles of several T . brucei repair factors . Mutation of T . brucei RAD51 , the key catalytic enzyme of homologous recombination , impairs VSG switching [79] . Intriguingly , yeast Rad51 has recently been shown to promote formation of R-loops and rearrangement at certain loci [98] , an effect that is abrogated when factors that promote Rad51 activity are mutated , consistent with the impairment of VSG switching in T . brucei BRCA2 and RAD51-3 mutants [80 , 81] . In both yeast and mammals , loss of the RecQ helicases Sgs2 and BLM , respectively , causes elevated levels of R-loops and locus-specific instability [99] , a response that may explain increased VSG switching by gene conversion when RECQ2 ( the T . brucei orthologue ) is mutated [28] . Moreover , the R-loop role of Sgs2/BLM has been interpreted as being necessary to tackle replication-transcription clashes , a role that may explain the distinct phenotypes of T . brucei RECQ2 mutants when acting on DNA double strand breaks and during VSG switching [28] . Finally , mutation of both RNase H enzymes in yeast has been documented to cause elevated levels of DNA damage ( detected as RAD52 localisation ) at rRNA genes , an effect that is due to RNA Pol I transit and results in gene conversion by break-induced replication [100] , a process that has been suggested to mediate VSG switching in the RNA Pol I-transcribed ES [101 , 102] . To understand the relationship between R-loops and VSG switching , a number of questions need to be addressed , including: do R-loops generate or follow from ES lesions; what form of DNA lesion results from , or generates the R-loops; how are R-loop-associated lesions signalled to initiate repair; and do increased levels of R-loops in the silent ES indicate movement of the hybrids in trans from the active ES ? Irrespective of the detailed mechanism , positioning of the 70 bp repeats immediately upstream of the VSG appears advantageous , targeting breaks to allow recombination-mediated break repair to access any of the ~1000 VSGs outside the VSG ES . Notably , the increased expression of silent VSG RNA in the Tbrh1-/- mutants , encompassing the range of genes and loci that comprise the VSG archive , differs from the more limited range of VSG types activated after the controlled generation of a DSB in the active ES [15] . In addition , R-loops that form upstream of the 70 bp repeats in the active VSG ES could drive intra-ES recombination , which is observed frequently [103] . Beyond the proposed mechanistic involvement of R-loops in directing VSG switching , this work reveals wider overlap with emerging roles for RNA in many immune evasion strategies . Antigenic variation in Neisseria gonorrhoeae relies upon the expression of a small non-coding RNA , upstream of and antisense to the pilE expression site , across a guanine quartet-forming DNA sequence [104] , which results in DNA nicks that may elicit recombination [105] . Intriguingly , small RNAs may also be generated from silent pilS recombination substrates [106] . Non-coding RNA is widespread in Plasmodium , including sense and antisense non-coding RNA ( ncRNA ) that emanates from the promoter [107] and intron [108] of var genes , which mediate antigenic variation , as well as from a var-associated GC-rich ncRNA gene family [109] . Modulation of the expression of the ncRNAs , as well as mutating a novel exoribonuclease [107] , undermines the transcriptional controls that determine singular var gene expression during antigenic variation . Though it has not to date been reported that R-loops form in these settings , as we describe for T . brucei , and the generation and action of effector transcripts is very likely to be organism-specific , it is notable that both recombination and transcription events during antigenic variation are influenced by RNA , which in at least two cases interacts with DNA . Characterising the factors and reactions that act on the T . brucei ES R-loops to dictate the dynamics of antigenic variation will reveal how similar or distinct the processes are in the different pathogens .
All cell lines used were bloodstream form parasites , which were maintained in HMI-9 medium supplemented with 10% ( v/v ) FBS ( Sigma-Aldrich , Missouri , USA ) and 1% ( v/v ) of penicillin-streptomycin solution ( Gibco ) at 37°C and 5% CO2 . C-terminus endogenous tagging of TbRH1 was carried out as previous described [110] . Briefly , the C-terminal 626 bp sequence of the TbRH1 ORF was PCR-amplified using primers CGACGAAGCTTCTGCGGATGACGGTAATG and CGACGAGATCTTGTGAATCGCCCTTTGGC and cloned into the pNATx12M plasmid containing 12 copies of the c-myc epitope . The construct was then stably transfected into T . brucei brucei Lister 427 MITat1 . 2 cells after digestion with PstI . Heterozygous ( -/+ ) and homozygous ( -/- ) Tbrh1 knockout cell lines were generated using two constructs containing cassettes of either blasticidin or neomycin resistance genes between α-β tubulin and actin intergenic regions , flanked by sequences homologous to the 5' and 3' UTRs of TbRH1 , essentially as described in [28] . Homologous flanking regions were PCR-amplified using the following primers: 5' UTR CGACGGGATCCTTGCCTTACCCGTGTTTT and CGACGTCTAGACCTTTTCTTTCCCATGGAC , 3' UTR CGACGCCCGGGAGGTGTGTATGGGAATGA and CGACGCTCGAGGCACCACCCAGTATAGAAA . Total RNA was extracted using an RNeasy Mini Kit ( Qiagen ) and reverse transcribed with SuperScript II Reverse Transcriptase ( Invitrogen ) using random hexamer primers . Power SYBR Green Master Mix ( Invitrogen ) was used to perform qPCR and fold change was calculated using the 2-ΔΔCT method [111] . Both DRIP and γH2A ChIP sample preparation was performed using a ChIP-IT Enzymatic Express kit ( Active Motif ) . Briefly , ~ 2x108 cells were grown to log phase before fixing in 1% formaldehyde for 5 min whilst shaking at room temperature , before 1 mL of 10X Glycine Buffer was added directly to the cells to stop fixation . Cells were then pelleted , re-suspended in Glycine Stop-Fix Solution and shaken at room temperature for 5 min . Cells were next lysed , according to the manufacturer’s protocol , allowing chromatin to be extracted and digested for 5 min with Enzymatic Shearing Cocktail at 37 ˚C to produce ~200 bp fragments . IP was performed overnight at 4 ˚C with 4 . 5 ng of S9 . 6 ( Kerafast ) or 3 μg of anti-уH2A antibody . For DRIP , on-bead treatment of control DRIP samples was performed as previously described [48] . qPCR was performed directly from DNA recovered from DRIP samples , with or without EcRH1 treatment , using SYBR Select Master Mix ( Invitrogen ) . The amount of DNA in IP samples was expressed as a percentage of input DNA , using CT values first adjusted by the dilution factor of each sample . Library preparation was performed using a TruSeq ChIP Library Preparation Kit ( Illumina ) and fragments of 300 bp , including adaptors , were selected with Agencourt AMPure XP ( Beckman Coulter ) . Sequencing was performed with an Illumina NextSeq 500 platform . Reads were trimmed using TrimGalore ( https://github . com/FelixKrueger/TrimGalore ) under default settings before alignment to the Lister 427 bloodstream VSG expression sites using Bowtie2 [112] in "very-sensitive" mode . Reads with a MapQ value <1 were removed using SAMtools [113] , leaving at least 30 million aligned reads per sample . The fold change between input and IP read depth was determined for each sample using the DeepTools bamCompare tool ( library size was normalised by read count and fold change was expressed as a ratio ) and visualised as tracks with IGV [114] . Normalised ratio files were also used to generate plots and perform kmeans cluster analysis using deepTools computeMatrix , plotProfile and plotHeatmap [115] tools . For RNA-seq analysis , total RNA was extracted using the RNeasy Mini Kit ( Qiagen ) . Poly ( A ) selection and library preparation was then performed using the TruSeq Stranded Total RNA kit ( Illumina ) and sequencing of 75 bp paired-end reads was performed using the Illumina NextSeq 500 platform . RKPM was calculated for each available VSG coding region [16] ignoring duplicate reads . Fold-change in RPKM for each VSG was calculated for Tbrh1-/- relative to WT . To ask what transcripts displayed altered levels in the Tbrh1-/- mutants relative to WT cells fold in RPKM was determined in RStudio . RNA-seq reads were aligned to the Lister 427 VSG ES [14] and annotated VSGs [16] using HISAT2 [116] in ‘no splice alignment’ mode; reads with a MapQ value <1 were removed using SAMtools , which has been shown to remove >99% of short read alignment to the wrong ES [49] . Read mapping was visualised using Matplotlib and a custom python script . VSG immunofluorescence analysis was performed as previously described [18] . Briefly , cell were fixed in 1% formaldehyde ( FA ) at room temp for 15 min . Cells were then blocked in 50% foetal bovine serum ( FBS ) for 15 min before primary ( α-VSG221 , 1:10000; α-VSG121 , 1:10000: gift from D . Horn ) and secondary ( Alexa Fluor 594 goat α-rabbit ( Molecular Probes ) 1:1000; Alexa Fluor 488 goat anti-rat ( Molecular Probes ) , 1:1000 ) antibody staining was carried out at room temp for 45 mins in both cases . Cells were then mounted in Fluoromount G with DAPI ( Cambridge Bioscience , Southern Biotech ) . For 12myc-TbRH1 and yH2A staining , cells were first adhered to slides before fixing in 4% FA for 4 min and then quenched in 100 nM glycine . Cells were permeabilised in 0 . 2% triton-X 100 for 10 min . Blocking was performed for 1 hour with 3% FBS before antibody staining ( α-myc Alexa Fluor 488 conjugated ( Millipore ) , 1:500; α-γH2A , 1:1000 , and Alexa Fluor 488 goat α-rabbit ( Molecular Probes ) , 1:1000 ) and mounting in DAPI as described above . For counting purposes cells were imaged using an Axioscope 2 fluorescence microscope ( Zeiss ) with a 60x objective . Higher resolution of VSG staining was performed with a DeltaVision Core Microscope ( Applied Precision ) , using a 100x 1 . 4 oil objective ( Olympus ) . Super-resolution structured-illumination imaging of 12myc-RH1 and yH2A signal was performed using an Elyra PS . 1 microscope ( Carl Zeiss ) using a 63x 1 . 4 objective . Sequences used in all mapping analyses are available in the European Nucleotide Archive ( accession number PRJEB21868 ) . | All pathogens must survive eradication by the host immune response in order to continue infections and be passed on to a new host . Changes in the proteins expressed on the surface of the pathogen , or on the surface of the cells the pathogen infects , is a widely used strategy to escape immune elimination . Understanding how this survival strategy , termed antigenic variation , operates in any pathogen is critical , both to understand interaction between the pathogen and host , and disease progression . A key event in antigenic variation is the initiation of the change in expression of the surface protein gene , though how this occurs has been detailed in very few pathogens . Here we examine how changes in expression of the surface coat of the African trypanosome , which causes sleeping sickness disease , are initiated . We reveal that specialised nucleic acid structures , termed R-loops , form around the expressed trypanosome surface protein gene and increase in abundance after mutation of an enzyme that removes them , leading to increased changes in the surface coat in trypanosome cells that are dividing . We therefore shed light on the earliest acting events in trypanosome antigenic variation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"blood",
"serum",
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"body",
"fluids",
"nucleases",
"molecular",
"probe",
"techniques",
"enzymes",
"immunology",
"dna-binding",
"proteins",
"enzymology",
"antigenic",
"variation",
"parasitic",
"protozoans",
"... | 2018 | Ribonuclease H1-targeted R-loops in surface antigen gene expression sites can direct trypanosome immune evasion |
The clinical outcomes of human infections by Plasmodium falciparum remain highly unpredictable . A complete understanding of the complex interactions between host cells and the parasite will require in vitro experimental models that simultaneously capture diverse host–parasite interactions relevant to pathogenesis . Here we show that advanced microfluidic devices concurrently model ( a ) adhesion of infected red blood cells to host cell ligands , ( b ) rheological responses to changing dimensions of capillaries with shapes and sizes similar to small blood vessels , and ( c ) phagocytosis of infected erythrocytes by macrophages . All of this is accomplished under physiologically relevant flow conditions for up to 20 h . Using select examples , we demonstrate how this enabling technology can be applied in novel , integrated ways to dissect interactions between host cell ligands and parasitized erythrocytes in synthetic capillaries . The devices are cheap and portable and require small sample volumes; thus , they have the potential to be widely used in research laboratories and at field sites with access to fresh patient samples .
Cellular events that contribute to severe malaria are multi-faceted [1] . Binding of malaria-infected red blood cells ( iRBCs ) to host endothelium may alter blood flow , affect blood vessel integrity , and contribute to physical blockages of narrow capillaries [2–4] . Loss of red blood cell ( RBC ) deformability may also contribute to capillary occlusions [5–8] . Activation of endothelial cells can trigger localized inflammatory responses [9–11] , and phagocytic activity can lead to anemia through destruction of both infected and uninfected RBCs , particularly in the spleen [12 , 13] . Since such events can occur in all infected individuals , it is a mystery why some infected patients progress to severe forms of the disease while others harbor high parasitemia and remain symptom free . Dissection of the molecular basis for variations in malaria pathogenesis relies on many approaches , each with unique advantages but also significant limitations . For instance , human genetics plays an important role in determining disease severity; human RBC disorders , which cause abnormal expression of surface ligands on iRBCs , can protect patients from disease [14] . Animal studies have also contributed significantly to our understanding of severe malaria , but disease pathology in animals can be very different from that in humans [15 , 16] . Finally , postmortem autopsies provide a direct link to human pathology , but offer limited flexibility for systematic testing of hypotheses for early events leading to pathogenesis [17–20] . In vitro models for malaria pathogenesis can complement studies of human genetics , autopsies , and animal models . In the past , traditional cell binding assays helped identify relevant host cells [21] , host proteins [22–25] , and parasite proteins [26–28] involved in cytoadherence . Experiments in flow chambers were particularly important in addressing shear stress–dependent adhesion , as seen in capillaries in vivo [29–33] . However , improved in vitro models are needed that more accurately depict iRBC interactions with host cells in small capillaries . Flow chamber dimensions tend to be much larger than typical capillaries where adhered parasitized cells are found; therefore , the peculiarities of blood flow in narrow blood vessels are not accurately mimicked . Flow chambers also offer no design flexibility needed to mimic complex capillary networks found in the microvasculature , where varying shear stresses could be critical in precipitating cytoadhesion or capillary occlusions . Traditional flow chambers are made of gas-impermeable , rigid materials , which do not readily model the elastic , multicellular properties of blood vessels . Finally , the chambers are bulky , cumbersome to transport , and not well suited for work with small biological sample volumes over many hours . Microfluidic devices made from elastomeric materials can overcome many of the limitations posed by bulk flow chambers . The fabrication techniques have been specifically developed to engineer devices of diverse shapes with micron-sized dimensions . Thus microfluidics allows for flow experiments in channels that have the same dimensions as capillaries in the microvasculature , using very small sample volumes . We recently showed that a 20-μm microfluidic channel with a 2-μm constriction was able to mimic certain aspects of the physical blockage of a narrow capillary by an infected erythrocyte , even in the absence of any host cells [7] . However , the onset of disease in infected individuals is more complicated than capillary obstruction by rigid erythrocytes . Here , we show how to integrate the adhesive interactions between host ligands and parasitized erythrocytes with the unique effects of blood flow in capillary-like environments . We also show that host–parasite interactions over long time periods can be studied in microfluidic devices by demonstrating phagocytic responses to infected erythrocytes under flow conditions . The addition of such complexity to microfluidic systems opens up the potential to use the devices in the field to explore the highly individual responses to malaria infections .
To mimic blood flow and cytoadherance of infected erythrocytes in capillaries , microfluidic channels of a variety of shapes and sizes were fabricated , including straight , 50-μm-wide channels , channels with narrow , 4-μm constrictions in them , and bifurcating channels that resembled a network of capillaries . The use of polydimethylsiloxane ( PDMS ) to create the molds that form the channel walls and ceilings is described in the Methods section and is supported by well established chemistry [34] . In physical appearance , the PDMS devices are approximately 3 cm long , 1 or 2 cm wide , and 0 . 5–1 cm tall . The ends of the channel are perforated with plastic tubing to allow flow to and from the channels . A syringe is connected to the outlet tubing to generate negative pressure across the channel , and a digital manometer is attached through a T-junction to measure this pressure ( see Methods for more details ) . This configuration requires no external pumps , is easy to transport , and can be mounted on practically any inverted microscope for flow measurements . Because the PDMS is irreversibly sealed to the glass substrate , the channels can withstand very high pressures—greater than 8 kPa across the channels—without any leakage . Typical pressure drops across the channels used in subsequent experiments are comparable with what has been measured in vivo for small capillaries . ( For example , a 1-kPa pressure drop was measured across a capillary in the cat mesentery [35] . ) Lower pressures ( below 0 . 2 kPa ) are obtained by adjusting the height of the column of fluid in the inlet reservoir of the channel . The elasticity of PDMS ( Young's modulus ∼ 750 kPa ) resembles that of many blood vessels in vivo ( 40 or 1 , 200 kPa , depending on the type of vessel and age ) [36 , 37] . The Methods section describes the approaches used to coat the floor of the channels with either purified protein ligands known to be important in cytoadherance or mammalian cells expressing such ligands . Fluid flow in microfluidic channels with at least one dimension less than 100 μm is well understood . The flow is laminar , has a low Reynolds number , and has a typical parabolic velocity profile , with the maximum velocity at the center of the channel [38] . The velocity at different spatial positions in microfluidic channels has been measured in previous experiments using submicron-sized fluorescent beads and was found to be in excellent agreement with predicted velocities [38] . Flow at low Reynolds numbers is entirely reversible and is governed only by the pressure drop across the channel and the viscosity of the fluid . The viscosity of blood can be calculated from the hematocrit . Thus , in our system the direct measurement of the pressure and hematocrit allowed us to calculate other parameters related to the fluid flow , such as average fluid velocity or wall shear stress . Although fluid velocity decreases with increasing hematocrit , the wall shear stress remains unchanged since it depends on the product of the viscosity and the fluid velocity . To establish our methodology , we used the parasite strain ItG-ICAM-1 , which is known to bind intracellular adhesion molecule 1 ( ICAM-1 ) , an important ligand mediating cytoadhesion in vivo . Both rolling and stationary adhesion of ItG-ICAM-1 to ICAM-1 under flow conditions have been previously measured at low shear stresses ( 0 . 05 Pa and 0 . 1 Pa ) [39] . No previous experiments have measured binding of this strain to CD36 , another important receptor that binds iRBCs in vivo; however , other ICAM-binding parasite strains are known to bind both ICAM-1 and CD36 [31] . Variations in binding to receptors between different strains have been previously reported , although the qualitative behavior is expected to be similar [39] . Our stocks of ItG-ICAM-1 were regularly selected for binding to purified ICAM-1 prior to introduction in channels [40] . Adhesion to ICAM-1 is important for malaria pathogenesis in vivo . ICAM-1 may be particularly important for mediating cytoadhesion in the brain , since immunohistochemical studies have shown that it is upregulated in the cerebral vasculature in fatal malaria cases [41] . Evidence also suggests that without ICAM-1 , binding to the endothelium under flow conditions is impaired [30] . Although previous work has shown that ICAM-1 works synergistically with other receptors to mediate stable adhesion to the endothelium [30 , 33] , we provide the first evidence to our knowledge that ICAM-1 alone may be able to mediate stable adhesion in a microfluidic environment . Adhesion of iRBCs to purified ICAM-1 was confirmed in our 50-μm wide × 29-μm tall microfluidic channels even under physiologically relevant shear stresses ( applied pressures: 0 . 5–5 kPa; corresponding shear stress: 0 . 2–2 . 5 Pa; [42–44] ) . These shear stresses are about an order of magnitude higher than shear stresses reported in previous adhesion experiments [31 , 33 , 39] . At all measured shear stresses , infected erythrocytes displayed rolling behavior on purified ICAM-1 adsorbed to the channels ( Figure 1A−1C and Video S1 ) . Fluorescence labeling confirmed that all rolling or attached RBCs were infected . Large numbers of uninfected erythrocytes flowed past the attached cells , usually without knocking them off the protein-coated surface . At all measured pressures , about 86% of iRBCs that interacted with the surface-adsorbed ICAM-1 rolled rather than remained stationary . Finally , at all pressures , ∼99% of cells that rolled continued rolling for as long as they were followed along the length of the channel ( typically 180 μm ) , rather than arresting on the surface or detaching . Indeed , several cells were observed to roll for several millimeters ( over 4 min ) without stopping or detaching from the surface , in agreement with previous published results at lower wall shear-stress values [31] . Trajectories of individual rolling iRBCs on ICAM-1 showed that the rolling occurred in a jerky , stepwise manner at all pressures , with periodic changes in velocity . In control experiments , uninfected cells , or erythrocytes infected with non-adherent strains of parasites such as unselected 3D7 or HB3 , showed no adhesion to the ICAM-1 surfaces . On purified CD36 , only stationary adhesion was observed at pressures below 1 kPa , while both rolling and stationary adhesion were observed at higher pressures [31] . We compared the adhesion of iRBCs under flow conditions to adsorbed ICAM-1 in the presence and absence of soluble ICAM-1 . At a pressure of 2 kPa , we found that soluble ICAM-1 inhibited adhesion of ItG-ICAM-1 by up to 85% . Using the microfluidic system , we performed this adhesion inhibition experiment using less than 50 μl of fluid . The use of small volumes of fluid for such experiments will greatly facilitate testing of potential drug or vaccine candidates that block adhesion . Adhesion was also studied in synthetic microcapillaries seeded with mammalian CHO cells expressing ICAM-1 ( CHO-ICAM ) and grown to confluence over 2 d . In contrast to behavior on cell-free ICAM-1 ligand , the majority of iRBCs exhibited stationary adhesion on CHO-ICAM ( at 0 . 1 kPa , as well as 3 kPa ) . Those iRBCs that did roll on CHO-ICAM displayed sporadic behavior , showing large variations in their instantaneous rolling velocities , sometimes coming to a complete halt for several seconds and sometimes detaching from the surface ( Figure 1D−1F and Video S2 ) . In control experiments , uninfected cells or erythrocytes infected with unselected 3D7 or HB3 parasites showed no binding to the CHO-ICAM cells . Stationary adhesion was observed on CHO cells expressing CD36 at pressures below 1 kPa , while an increasing fraction of iRBCs rolled at higher pressures , similar to purified CD36 receptor adsorbed to the channels ( unpublished data ) . The difference in binding to pure ligand versus ligand expressed on mammalian cells was not previously seen in bulk flow chambers that compared rolling of iRBCs on purified ICAM-1 and HUVECs—cells that primarily express ICAM-1 [31] . Since ICAM-1–transfected CHO cells do not express any other known cytoadherence receptors in abundance , it is unlikely that the stable binding we observed on CHO cells is mediated by additional known proteins such as CD36 or thrombospondin [45 , 46] . Previous computational and experimental work has shown that the presence of cells in a micron-sized channel can greatly alter the flow microenvironment [47] . Therefore , the stable adhesion of iRBCs to CHO-ICAM may be a direct consequence of such flow alterations . The ability of microfluidic devices to support both pure ligands and ligand-expressing human cells in culture , under high shear stress , should facilitate a detailed understanding of the role of the ligand environment in precipitation of capillary blockage in the microvasculature . Even at high pressures in microchannels , iRBCs carrying the ITG strain of Plasmodium falciparum displayed remarkable “meta-stable” attachment to protein surfaces . iRBCs continued to roll on host ligands at pressures as high as 5 kPa without detaching from the surface . This suggests that an interesting and important molecular mechanism promotes such behavior , because in the absence of a compensating mechanism , most cells that roll on substrates are expected to increase their rolling velocities in response to increasing shear stresses and eventually detach from the surface [48] . Furthermore , the ligand ICAM-1 , when compared with another important cytoadhesion ligand , CD36 , imposed different rolling properties on different iRBCs within a population . To illustrate how rolling velocities responded to increasing pressure , we tracked individual iRBCs in a population rolling on either purified CD36 or ICAM-1 . On purified CD36 , significant rolling required pressures higher than 1 kPa . As pressure increased beyond 1 kPa , iRBCs showed no significant increases in the mean rolling velocities ( Figure 2A ) . A similar analysis of individual iRBCs rolling on ICAM-1 revealed different and even more complex behavior . First , the average rolling velocities on ICAM-1 were higher than those on CD36 . In addition , on ICAM-1 , populations of iRBCs showed significant differences in the variances of rolling velocities at different pressures , indicating that all cells within a population did not roll in an identical fashion . A large proportion of iRBCs appeared to stabilize adhesion and did not show proportional changes in velocity with increasing pressure , particularly at lower pressures . This agrees with previous work that showed rolling velocities for ICAM-1 did not change when shear stress was increased [39] . However , at high pressures , a fraction of iRBCs did indeed increase rolling velocities with increasing pressures ( Figure 2B ) . The increase in rolling velocities of some cells but not others was not a result of the parabolic fluid velocity profile in a microchannel . First , the measured velocities showed no correlation with the spatial position of iRBCs in the channel; many iRBCs in the same part of the channel had different velocities . Second , all velocity measurements were taken at least 10 μm from the channel walls to exclude any RBCs that may be affected by interactions with the wall . For the aspect ratio used in our devices , the maximum variation in velocity attributable to the parabolic flow profile is approximately 25% [38] . In contrast , the differences in velocity between rolling iRBCs at a particular pressure was typically much greater and increased with increasing pressures . Finally , iRBCs with large variations in rolling velocities were not seen on CD36 , indicating that the velocity profile in the channel has a negligible effect on the rolling velocity . The plateau in rolling velocities of iRBCs at increasing pressure is qualitatively similar to the stability of leukocyte-rolling velocities on selectins at a wide range of shear stresses , both in vivo and in vitro [49 , 50] . For leukocytes , this has been attributed to a shear-dependent increase in the number of receptor–ligand bonds per rolling step , to compensate for the predicted increase in receptor-ligand dissociation [51] . Cellular characteristics like deformability also contribute to the stabilization of rolling velocities displayed by leukocytes [52] . Similar mechanisms could explain stabilization of rolling velocities for the present iRBC–protein interactions . Stabilization of rolling velocities of iRBCs on host ligands could have clinical significance . Regulated rolling on capillaries in vivo may allow iRBCs to evenly sample the endothelium , independent of changing dimensions of the blood vessels and the accompanying changes in wall shear stress . Slightly enhanced stabilization of rolling velocities , even in a subpopulation of infected cells , could thus play an important role in promoting accumulation of iRBCs in capillaries . Branching capillaries are natural sites in the circulatory system where changes in blood flow patterns can lead to alterations in wall shear stress [43] . Microfluidic technology enabled us to fabricate a device that mimicked branching capillaries , with a main channel connected to a network of secondary channels ( Figures 3 and 4 ) . Given the complex responses of rolling iRBCs to changes in flow pressures , we hypothesized that individual iRBCs in such branched channels functionalized with ICAM-1 would show different rolling behavior at the sites of shear-stress changes . iRBCs displayed continued rolling behavior upon encountering a fork and followed the path dictated by the bulk fluid flow ( Video S3 ) . Velocities of rolling erythrocytes upon reaching the branches , however , varied from cell to cell and displayed one of two patterns . Some cells did not change rolling velocities as they moved from the bifurcating branch into the main artery of the channel , despite the increase in wall shear stress ( Figure 3B ) . Yet , other cells displayed significant increases in rolling velocity ( Figure 3D ) . Flow in microfluidic channels is entirely reversible and depends only on the pressure difference between the entrance and exit of the channel [38] . Thus , iRBCs flowing in the reverse direction—from the main artery into the branch of the channel—displayed the same behavior . Branched channels were also used to determine whether the accumulation of stably adhering iRBCs was dependent on the shear stress in a simulated capillary network . In a channel functionalized with CD36 , at pressures where primarily static adhesion is observed , we found increased accumulation of iRBCs in the branches of a model capillary network relative to the main artery ( Figure 4 ) . These studies demonstrate that microfluidic devices can be fabricated to identify and possibly select cell types that will most likely stabilize rolling upon encountering lower shear stresses . They also show how changing shear stresses due to the shape of a capillary in vivo may be critical in determining where cytoadhesion will likely occur . Clearly , sequestration of infected erythrocytes may depend on the location of host cells with adhesive ligands in the microvasculature , as well as the type and quantity of expressed ligands and the nature of the individual iRBCs . Future microfluidic studies can be designed to explore the influence of ligand concentrations , or even mixtures of ligands on cytoadherance by RBCs harboring different parasite clones . Erythrocytes in the microvasculature can encounter capillaries with dimensions smaller than the RBC diameter . Historically , such constrictions have been thought to interfere with circulation of rigidified iRBCs [3 , 6 , 10] . Intuitively , it would appear that coating of constricting capillaries with adhesive proteins would further promote blockage . To test this simple hypothesis , we designed an ICAM-1–coated channel that was 5-μm tall and began with a 20-μm width that constricted to 5 μm before returning to 20 μm ( Figure 5A ) . The tight constriction was just wide enough to permit a normal RBC , as well as infected erythrocytes , to squeeze through in the absence of ligand [7] . The behavior of rolling iRBCs as they approached and passed through 5-μm-wide ligand-coated constrictions dramatically illustrated how microfluidic technology permits experiments that would be impossible in conventional flow chambers . As the rolling iRBCs entered the constriction , they briefly ceased rolling and actually accelerated through the pore . This was recorded as a jump in the distance traveled over the length of the constriction and a corresponding spike in the iRBC velocity ( Figure 5B and Video S4 ) . Upon exiting the 5-μm constriction , the iRBCs efficiently reattached on the other end and continued rolling at a velocity similar to that before entering the constriction . The rapid traverse of iRBCs in the narrow part of the channel was not due to uneven coating of ICAM-1 on the channel walls; a fluorescently labeled antibody to ICAM confirmed the presence of the ICAM-1 protein throughout the channel , including in the 5-μm constriction . The decreased interaction of iRBCs with adhesive proteins in confined spaces could be due to one of two other reasons . The large pressure drop across the narrow constriction could create wall shear stresses that readily override the adhesion capabilities of iRBCs . Alternatively , the inability of iRBCs to roll in the confined environment could reduce their affinity for adsorbed ligands . Regardless , the presence of the adhesive protein on the surface of the narrow channels did not augment the formation of obstructions within RBC-sized channels . These results suggest that , unless additional events are in play , the narrowest capillaries in vivo may not necessarily be the first to become obstructed with iRBCs . Clearance of parasites from a naive infected individual is largely dependent on the phagocytosis of iRBCs by macrophages in the spleen . To build on experiments on phagocytosis of iRBCs by macrophages in static cultures [53 , 54] , we studied the interactions between iRBCs and macrophages under shear flow in a 50-μm-wide channel for over 20 h . iRBCs rolled slowly on RAW macrophages and finally halted , similar to the behavior exhibited on CHO-ICAM cells . Fluorescently labeled parasite DNA confirmed that over 90% of bound RBCs were parasitized . After 30 min of RBC flow , the attached cells were subjected to continuous media flow , without additional RBCs . After 2 h , 65% of iRBCs remained attached to the macrophages . Of these , approximately half were internalized after 20 h under continuous media flow ( Figure 6 ) . Phagocytosis of infected erythrocytes under shear flow occurred in one of several ways ( Figure 6 ) . Some macrophages ingested parasites together with the intact RBC membrane , as judged by simultaneous fluorescence labeling of the RBC membrane and parasite DNA . However , in many cases , the parasites were internalized without any accompanying RBC membrane , reminiscent of in vivo observations of parasite “pitting” by macrophages [55–57] . We also saw evidence of phagocytosis of uninfected cells , which could represent the phagocytosis of a previously “pitted” erythrocyte , an aging uninfected RBC , an uninfected RBC tagged with parasite proteins [58] , or a residual erythrocyte ghost from a ruptured schizont . This observation agrees with autopsy data that show phagocytosis of non-parasitized erythrocytes in large numbers in the spleen [59] . Finally , our microfluidic system revealed hemozoin internalization in the macrophages , not only in the same cell as a fluorescent parasite but also in some cells that contained no parasites . Again , this is consistent with human autopsy data [59] . Microfluidic devices offer a powerful new opportunity to study malaria pathogenesis and other human diseases that involve the microvasculature . The present laboratory-based applications of this advancing technology illustrate the types of questions in malaria pathogenesis that may be addressed with microfluidics . Since the devices are portable and require mere microliter volumes of samples , future applications should be possible at field sites , using matched patient samples . Such studies could include parasitized blood , serum , platelets , antibodies , phagocytic cells , and possibly biopsied host samples . We expect that the most valuable insights into the causes of severe malaria will arise from detailed studies of variations in human and parasite samples at field sites . Technically , even though the fabrication of the silicon master for a specific experimental application requires an experienced materials science engineer and a specialized clean room facility , subsequent production of dozens of PDMS devices from a common master is inexpensive and easy to learn . As illustrated , the soft lithography methodology allows for the design of channels of a wide variety of shapes and very small sizes , and the gas-permeable PDMS polymer readily accommodates long-term cell growth of multiple cell types in channels . The microfluidic devices can be mounted on a microscope , and data on single cells can be collected as still photos or as movies on a personal computer for further detailed analysis . In addition to their use in field sites , we expect the devices to be popular in standard research laboratories where access to traditional flow adhesion apparatus is either unavailable or impractical due to the large volumes of sample needed .
Microfluidic silicon masters were fabricated using standard photolithographic techniques [7 , 34] . Briefly , channel patterns were created on a quartz-chrome mask ( Photo Sciences , http://www . photo-sciences . com/ ) and imprinted on silicon wafers ( Montco Silicon Technologies , http://www . silicon-wafers . com/ ) using photoresist . Relief features were etched using the Bosch deep reactive ion etch process ( Oxford Instruments , http://www . oxford-instruments . com/ ) . To make the elastomeric microfluidic devices from the silicon masters , PDMS ( Dow Corning , http://www . dowcorning . com/ ) was poured over the silicon master , cured , and then cut from the master and irreversibly sealed to a glass coverslip after oxygen plasma treatment ( Harrick Scientific Products , http://www . harricksci . com/ ) . Access to the channels was possible through a 5-mm hole punched on one end of the channel to form a reservoir for the sample , with a smaller hole punched by a 21-gauge needle at the other end , into which polyethylene tubing ( PE20 ) attached to a syringe could be inserted . Pressure was controlled by manually adjusting the plunger on a syringe and was measured with a digital manometer inserted between the tubing and the syringe . Flow rates were calculated from applied pressures using Poiseuille's equation for flow in a channel with a rectangular cross section [60] , from which wall shear stresses could be calculated [31] . The ICAM-1–adherent laboratory line of P . falciparum ( ItG-ICAM-1 ) was used for all assays and was a kind gift from Joseph Smith ( Seattle Biomedical Research Institute , http://www . sbri . org/ ) . Parasites were periodically selected by passage over recombinant ICAM-1 protein ( R&D Systems , http://www . rndsystems . com/ ) spotted on Falcon 1007 petri dishes . Parasites were cultured in human RBCs according to standard protocols . Unsynchronized cultures were used for all experiments , with parasitemia ( assessed by Giemsa-stained blood smears ) ranging between 5% and 17% . For all adhesion assays , parasite cultures were washed twice in pre-warmed binding medium , consisting of RPMI 1640 with 0 . 5% BSA ( pH 7 . 2 ) and resuspended in the binding medium at a hematocrit of 2%–10% . CHO cells transfected with CD36 were a gift from Joseph Smith , and CHO cells transfected with ICAM-1 were obtained from ATCC ( http://www . atcc . org/ ) . Both cells lines were cultured in F-12K nutrient mixture supplemented with 10% fetal bovine serum , 5% penicillin-streptomycin , and 0 . 25 mg/ml Geneticin ( Invitrogen , http://www . invitrogen . com/ ) . RAW macrophages were cultured in DMEM supplemented with 10% fetal bovine serum and 5% penicillin-streptomycin ( Invitrogen ) . The channels were first rinsed continuously with a flow of ethanol for about 10 min , followed by rinsing with a 4% solution of aminopropylethoxysilane ( APES; Sigma-Aldrich , http://www . sigmaaldrich . com/ ) in ethanol for about 15 min , to prepare the glass surface for protein adsorption . Solutions of either ICAM-1 or CD36 in MilliQ water ( both at concentrations of 50 μg/ml , R&D Systems ) were introduced into the channels at low flow rates for approximately 2 h at 37 °C . Channels were then blocked for 2 h with a 2% BSA solution . A similar protocol was previously used to functionalize glass microslides with both CD36 and ICAM-1 [61] . To ensure reproducibility of the protein surface , we used concentrations of CD36 and ICAM-1 that are well above those that saturate the surface . For adhesion blocking with soluble ICAM-1 , 3 μl of packed RBCs enriched to 30% parasitemia using Plasmion plasmagel were incubated in 50 μl of ICAM-1 at a concentration of 50 μg/ml for 15 min at 37 °C . The RBC solution was then flowed through the microfluidic chamber at a pressure of 2 kPa for 12 min , after which the number of attached cells were counted over at least eight different fields of view . The number of attached cells was compared with the number obtained by flowing into the channel an equivalent concentration of iRBCs that were not exposed to soluble ICAM-1 at the same pressure for the same time . Channels were first incubated with the appropriate cell culture media for approximately 1 h at 37 °C prior to introducing cells . About 200 μl of cells in media were pipetted into the channel reservoir at a concentration of about 5 million cells/ml . The cells were pulled into the channel and allowed to settle . Unattached cells were rinsed away and the process was repeated to achieve an attached cell density that would support the growth of a confluent monolayer . Cells in the channels were grown under continuous fluid flow for up to 3 d and shown to be alive using a fluorescent Live/Dead Cell Vitality Assay ( Molecular Probes , http://probes . invitrogen . com/ ) . All imaging of cells and channels was carried out on an inverted fluorescence microscope ( Nikon TE200 or TE2000; http://www . nikon . com/ ) , with either a 40× ( Plan Fluor , 0 . 75 NA , Nikon ) or an oil immersion 100× ( Plan Fluor , 1 . 3 NA , Nikon ) objective . Movies and images of infected erythrocytes in channels were captured on a high-sensitivity CCD camera ( a Hamamatsu Digital Camera C4742–98 , Hamamatsu CCD Camera [video] C2400 , or a Photometrics CoolSnap ES [Roper Scientific , http://www . roperscientific . com/] ) . Image and movie acquisition was with Metamorph Imaging System ( Molecular Devices , http://www . moleculardevices . com/ ) . A home-built temperature-controlled stage maintained a 37 °C environment for the experiments with live mammalian cells in the channels . Movies of rolling iRBCs were analyzed using the tracking software on the Metamorph Imaging System . x and y coordinates of cells at each acquisition frame ( every 0 . 1 s ) were recorded , from which instantaneous velocities , average velocities , and distances from the origin were calculated . Data were further analyzed using Igor Pro software ( Wavemetrics , http://www . wavemetrics . com/ ) . Statistical analysis was performed with the Igor Pro ANOVA package . RAW macrophages were seeded and grown in 50 μm × 29 μm channels and iRBC cultures introduced at a pressure of 0 . 1 kPa . Channels were kept overnight in an incubator at 37 °C and 5% CO2 , with the flow rate maintained by gravity . Infected erythrocytes were counted by taking an average of approximately 20 random fields of view of the attached macrophages in the channel . Phagocytosis was measured after lysis of attached erythrocytes with cold water , as previously described [53 , 54] . | With over 500 million clinical cases and 1 million deaths per year , malaria presents a devastating global health problem . Samples from patients with severe disease suggest that binding of malaria-infected red blood cells ( iRBCs ) to host mammalian cells plays an important role in precipitating blood vessel blockages that can cause organ failure . Yet , some individuals in endemic countries harbor parasites without significant clinical symptoms . To help explore variations in disease outcomes , we developed microfluidic channels that mimic many potential features of severe disease . Synthetic microfluidic channels , with sizes and shapes resembling small capillary networks , were coated with pure host proteins or cultured mammalian cells expressing host ligands . We could therefore simulate binding of iRBCs under high-pressure fluid flow in a realistic capillary environment . By tracking the fate of individual iRBCs , we observed parasite-to-parasite variation in adhesion and an unexpected drop in adhesion when iRBCs passed through the thinnest capillaries . We also showed engulfment of iRBCs by phagocytic cells under fluid flow . The microfluidic devices should serve as powerful field tools for understanding severe malaria because the system is easy to use , requires very small sample volumes , and is portable for on-site analysis of patient samples in the field . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases",
"public",
"health",
"and",
"epidemiology",
"cell",
"biology",
"pathology",
"plasmodium",
"in",
"vitro",
"immunology",
"microbiology",
"eukaryotes",
"chemical",
"biology",
"hematology"
] | 2007 | Microfluidic Modeling of Cell−Cell Interactions in Malaria Pathogenesis |
In the dark , etiolated seedlings display a long hypocotyl , the growth of which is rapidly inhibited when the seedlings are exposed to light . In contrast , the phytohormone ethylene prevents hypocotyl elongation in the dark but enhances its growth in the light . However , the mechanism by which light and ethylene signalling oppositely affect this process at the protein level is unclear . Here , we report that ethylene enhances the movement of CONSTITUTIVE PHOTOMORPHOGENESIS 1 ( COP1 ) to the nucleus where it mediates the degradation of LONG HYPOCOTYL 5 ( HY5 ) , contributing to hypocotyl growth in the light . Our results indicate that HY5 is required for ethylene-promoted hypocotyl growth in the light , but not in the dark . Using genetic and biochemical analyses , we found that HY5 functions downstream of ETHYLENE INSENSITIVE 3 ( EIN3 ) for ethylene-promoted hypocotyl growth . Furthermore , the upstream regulation of HY5 stability by ethylene is COP1-dependent , and COP1 is genetically located downstream of EIN3 , indicating that the COP1-HY5 complex integrates light and ethylene signalling downstream of EIN3 . Importantly , the ethylene precursor 1-aminocyclopropane-1-carboxylate ( ACC ) enriched the nuclear localisation of COP1; however , this effect was dependent on EIN3 only in the presence of light , strongly suggesting that ethylene promotes the effects of light on the movement of COP1 from the cytoplasm to the nucleus . Thus , our investigation demonstrates that the COP1-HY5 complex is a novel integrator that plays an essential role in ethylene-promoted hypocotyl growth in the light .
The phytohormone ethylene plays significant roles in many developmental processes and stress responses in plants . Molecular and genetic analyses have revealed a linear signalling pathway , which is initiated by ethylene perception at the endoplasmic reticulum membrane , resulting in transcriptional regulation in the nucleus [1]–[3] . The ethylene receptors ETHYLENE RESPONSE 1 ( ETR1 ) , ETHYLENE RESPONSE SENSOR 1 ( ERS1 ) , ETR2 , ERS2 , and ETHYLENE INSENSITIVE 4 ( EIN4 ) are members of a family of two-component His protein kinase receptors that negatively affect ethylene signaling [4] , [5] . In the absence of ethylene , these receptors directly suppress the ethylene response by interacting with a Raf-like mitogen-activated protein kinase kinase kinase family protein , CONSTITUTIVE TRIPLE RESPONSE 1 ( CTR1 ) [6]–[8] . This negative regulator interacts with and directly phosphorylates the cytosolic C-terminal domain of EIN2 in Arabidopsis [9] . EIN2 is a central positive regulator of ethylene signalling that is localised to the endoplasmic reticulum membrane through its N-terminal domain [10] . The phosphorylation of EIN2 by CTR1 prevents EIN2 from signalling in the absence of ethylene , whereas the inhibition of CTR1 upon ethylene perception is a signal for cleavage and translocation of EIN2 from the cytoplasm to the nucleus [9] , [11]–[13] . The transcription factors EIN3 and EIN3-LIKE 1 ( EIL1 ) act downstream of EIN2 , which is required for the ethylene-induced stabilisation of EIN3/EIL1 [14] , [15] . EIN3 further activates the expression of ethylene-responsive genes in different physiological processes [15]–[21] . When emerging from the soil , newly etiolated Arabidopsis thaliana seedlings display long hypocotyls , apical hooks , and closed cotyledons . Exposure to light inhibits hypocotyl growth and promotes the greening and expansion of the cotyledons and leaves [22] . During this processes , light modulates multiple hormonal pathways , including those involving gibberellins , abscisic acid , auxin , brassinosteroids , cytokinins , and ethylene , to regulate these developmental changes [23]–[32] . Increasing evidence suggests that gibberellins and cytokinins regulate the accumulation of the light signalling component LONG HYPOCOTYL 5 ( HY5 ) , a basic leucine zipper ( bZIP ) transcription factor that acts downstream of the light signal and positively regulates the transcription of light-induced genes [33] , [34] . Light promotes the accumulation of HY5 protein by inhibiting the accumulation of CONSTITUTIVELY PHOTOMORPHOGENIC 1 ( COP1 ) in the nucleus [35] , [36] . Furthermore , HY5 reportedly promotes photomorphogenesis , in part by modulating auxin , cytokinin , and gibberellins signalling [32] , , revealing the integration of light and phytohormone signalling . In addition , a recent study reported that EIN3/EIL1 cooperate with phytochrome interacting factor 1 ( PIF1 ) and COP1 to optimise the de-etiolation of Arabidopsis seedlings and demonstrated that ethylene plays a key role in the establishment of green seedlings upon exposure to light [19] . Importantly , ethylene also inhibits Arabidopsis hypocotyl elongation in the dark , whereas ethylene and its precursor 1-aminocyclopropane-1-carboxylate ( ACC ) increase hypocotyl elongation in light-grown Arabidopsis seedlings [24] , [39] , [40] , and EIN3 transcriptionally activates two contrasting pathways: the PIF3-dependent growth-promoting pathway and an ethylene response factor 1 ( ERF1 ) -mediated growth inhibiting pathway , to fine-tune ethylene-promoted hypocotyl growth in the light [39] . Moreover , ethylene regulates the biosynthesis , transport , and distribution of IAA during light-mediated hypocotyl growth , dependent on the effect of COP1 on gene transcription downstream of EIN3 [29] , [40] . These findings indicate the regulation of seedling growth by ethylene-light interactions; however , it remains unclear how the ethylene and light signalling pathways are integrated at the protein level to achieve the conserved regulation of plant hypocotyl growth . In this study , we found that ethylene enhances the movement of COP1 to the nucleus to degrade HY5 in the light , revealing that the COP1-HY5 complex is a novel integrator of light-ethylene interactions during hypocotyl growth . Our data show that ethylene regulates hypocotyl growth by reducing HY5 protein levels . More importantly , ethylene-promoted hypocotyl elongation was inhibited in the absence of COP1 function , and COP1 was found to be required for HY5 stability . Further , ACC affected the EIN3-dependent shuttling of COP1 between the nucleus and cytoplasm . These results reveal that the light-regulated COP1-HY5 cascade is involved in ethylene-promoted hypocotyl growth .
Ethylene confers opposing effects on hypocotyl growth that are dependent on the conditions under which plants are grown ( light or dark ) . Ethylene and its precursor ACC can stimulate hypocotyl growth in the light , but inhibit hypocotyl elongation in etiolated seedlings . HY5 is an important positive regulator of photomorphogenesis; the hy5 mutant displays a long hypocotyl when grown in the light , but a normal hypocotyl when grown in the dark [41] . A recent report revealed that HY5 binds to the promoters of cell elongation-related genes and recruits PKL/EPP1 through their physical interaction to regulate hypocotyl growth [42] . Interestingly , although treatment with 10 µM ACC diminished hypocotyl growth in dark-grown Col-0 and hy5 plants , ACC did not significantly promote hypocotyl growth in light-grown hy5 seedlings ( Figure 1A–D ) . This result was confirmed in hy5-215 ( Col-0 background ) and hy5-ks50 ( Ws background , Figure S1 ) . In addition , the root of hy5 mutants responded normally to ethylene because ACC inhibited the growth of root as the same of Col-0 in the light or in the dark ( Figure 1A and 1C , Figure S1A ) . These results indicate that HY5 plays an essential role in ethylene-promoted hypocotyl growth . As a key factor in photomorphogenic development , HY5 represents a convergence point for light and multiple hormone signalling pathways [29]; combined with our results , these observations suggest that HY5 participates in ethylene-promoted hypocotyl elongation in the light but not in the dark . To confirm the function of HY5 in regulating hypocotyl growth , we generated a hybrid of the ethylene overproducing mutant eto2 [43] and transgenic plants constitutively expressing HY5 ( 35S::HY5 ) [44] . As shown in Figure S2A and S2B , eto2 displayed longer hypocotyls than Col-0 plants , while the hypocotyl length of eto2×35S::HY5 was intermediate between that of eto2 and Col-0 . The long hypocotyl caused by endogenous ethylene overproduction was diminished by the constitutive expression of HY5 ( Figure S2 ) , indicating that HY5 negatively regulates ethylene-promoted hypocotyl growth in the light . To determine how ethylene modulates HY5 accumulation , we first examined whether ethylene transcriptionally activates HY5 expression . The expression levels of HY5 and its homologue HYH were similar in Col-0 , eto , ein2 , and ein3-1 under normal growth conditions ( Figure S3 ) , suggesting that ethylene does not transcriptionally regulate HY5 expression . We next detected HY5 protein accumulation in Col-0 and hy5 mutant plants . There was no band detected in hy5 mutant while ACC greatly reduced HY5 protein accumulation in Col-0 ( Figure 1E ) , suggesting that ethylene mediates an increase in HY5 instability at the post-transcriptional level to regulate hypocotyl growth . To further understand how ethylene suppresses HY5 accumulation , we analysed the stability of HY5 in the transgenic line 35S::HY5 . Because ACC induces hypocotyl growth in a dose-dependent manner [24] , we first examined the induction of HY5 degradation by ethylene using different concentrations of ACC . Treatment with 5 , 10 , or 25 µM ACC for 16 h slightly , greatly , and significantly suppressed HY5 accumulation in 35S::HY5 , respectively . When the applied ACC concentration increased , HY5 protein accumulation gradually decreased ( Figure 1F ) . Furthermore , we observed that treatment with ACC for 8 h resulted in obvious degradation of HY5; when the treatment was increased to 16 h , the HY5 protein level was further reduced ( Figure 1G ) , suggesting that ethylene significantly suppresses HY5 stability in a dose- and time-dependent manner . We also found that treatment with the 26S protein degradation inhibitor MG132 prevented the decrease in HY5 protein levels caused by ACC treatment ( Figure 1H ) , suggesting that ethylene regulates HY5 stability via 26S proteasome-mediated degradation . To further study the mechanism of ethylene-promoted hypocotyl elongation in the light , we examined the hypocotyl lengths of ethylene signalling mutants . Similar to eto2 , the constitutive ethylene response mutant ctr1-1 displayed longer hypocotyls than Col-0 or the ethylene-insensitive mutants etr1-1 , ein2 , ein3-1 , and ein3-1 eil1 . Unlike Col-0 , the hypocotyl lengths of etr1-1 , ein2 , ein3-1 , and ein3-1 eil1 did not increase following treatment with 10 µM ACC ( Figure 2A and 2B ) . An analysis of HY5 expression in these mutants by Western blotting produced results that were consistent with those of our hypocotyl elongation assays ( Figure 2C ) . Namely , HY5 accumulation was reduced in eto2 and ctr1-1 compared to Col-0 in the absence of ACC , and ACC treatment enhanced the decrease in HY5 protein , especially in ctr1-1 . However , in etr1-1 , ein2 , ein3-1 , and ein3-1 eil1 , only slight differences from Col-0 were observed , and ACC did not promote HY5 degradation ( Figure 2C ) . Consistent with the hypocotyl lengths shown in Figure 2A and 2B , these results indicate that HY5 acts downstream of ethylene signalling to regulate hypocotyl growth . We next analysed the genetic relationship between HY5 and ethylene signalling . The hypocotyl lengths in the etr1-1 hy5 , ein2 hy5 , and ein3-1 hy5 double mutants were obviously longer than that in Col-0 and similar to that in hy5 ( Figure 3A and 3B ) . Taken together with the finding that ACC did not promote the degradation of HY5 in ein3-1 ( Figure 2C ) , our data indicate that HY5 acts downstream of EIN3 to mediate ethylene-promoted hypocotyl growth . COP1 is an E3 ligase that negatively modulates HY5 activity via protein-protein interactions [33] , [36] . COP1 has three structural domains that are critical for its molecular association with HY5: an N-terminal RING-finger domain followed by a predicted coiled-coil domain and C-terminal WD-40 repeats [33] . For the regulation of HY5 stability by ethylene through 26S proteasome-mediated degradation , we hypothesised that COP1 may also function in this process . Therefore , we measured the ethylene response in Arabidopsis cop1 mutants ( cop1-4 and cop1-6 , both in Col-0 background ) and transgenic lines . It has previously evidenced that the cop1-4 mutation alters the Gln-283 CAA codon to a UAA stop codon , resulting in a truncated COP1 protein containing only the N-terminal 282 amino acids , and the cop1-6 mutation changes the splicing junction “AG” at the 3′ end of intron 4 to “GG” , generating a five-amino acid insertion in the COP1 protein [45] . Interestingly , those cop1 mutants that reduced activity of COP1 displayed shorter hypocotyls than Col-0 , and ACC failed to promote hypocotyl elongation in cop1-4 or cop1-6 , whereas the full-length COP1 overexpressor ( GUS-COP1 ) displayed longer hypocotyls than Col-0 ( Figure 4A and 4B ) . These results indicate that COP1 positively regulates ethylene-promoted hypocotyl growth . The RING-finger domain , coiled-coil domain , and WD-40 repeats of COP1 are essential for its interaction with HY5 [33] , [46] . Accordingly , a transgenic line overexpressing COP1 without the RING-finger domain ( ΔRING , No-0 background ) showed significantly longer hypocotyls than the wild type , whereas a transgenic line overexpressing COP1 without the coiled-coil domain ( ΔCoil , Col-0 background ) displayed a similar hypocotyl length to the wild type in the absence of ACC under white light ( Figure 4A and 4B ) . Distinctive to the cop1 mutant , ACC significantly enhanced hypocotyl growth in the ΔRING and ΔCoil transgenic lines , similar to the effect of ACC on their corresponding controls ( Figure 4A and 4B ) , suggesting the redundant function of the RING-finger and coiled-coil domains in hypocotyl growth . Furthermore , transgenic line overexpressing COP1 without the RING-finger and coiled-coil domains ( ΔRΔC , Col-0 background ) did not show obvious hypocotyl elongation compared to Col-0 in the absence of ACC under white light , and ACC did not significantly improve hypocotyl growth in the ΔRΔC line ( Figure 4A and 4B ) . Thus the above observations indicate that mutation of protein-protein interaction domains of COP1 alters ethylene responsive hypocotyl elongation . To elucidate the mechanism through which ethylene suppresses HY5 accumulation , we analysed the HY5 protein levels in the cop1 mutants and transgenic lines . As expected , HY5 accumulation was detected in cop1-4 and cop1-6 , while ACC-induced HY5 degradation was inhibited in cop1-4 and cop1-6 ( Figure 4C ) , revealing the essential role of COP1 in ethylene-promoted hypocotyl growth and HY5 stability . Moreover , the abundance of HY5 in these transgenic lines was correlated with the observed extent of photomorphogenic development . When grown under the same white light conditions , GUS-COP1 and ΔRING had reduced while ΔCoil and ΔRΔC showed similar levels of HY5 protein relative to wild type , and ACC treatment enhanced the degradation of HY5 in these transgenic lines ( Figure 4D ) , indicating that the ACC-induced suppression of HY5 is mediated by the E3 ligase COP1 , dependent on the interaction of COP1 with HY5 . Because HY5 is not the unique targets of COP1 [47]–[49] , then we further questioned whether the role of HY5 in ethylene-promoted growth is specifically targeted by the interaction with COP1 . To address this query , we performed assays of ethylene responsiveness using transgenic lines of the deletion of N-terminal 77 amino acids of HY5 ( HY5-ΔN77 ) in Col-0 or hy5 background , which has been evidenced that this fragment is essential for the interaction of HY5 with COP1 , and HY5-ΔN77 transgene is being driven by the CaMV35S promoter [33] . Transcriptional detections with quantitative real-time PCR ( qPCR ) showed that the expression of truncated HY5 ( deletion of the N-terminal 77 amino acids ) or full-length HY5 was overexpressed in individual transgenic lines ( Figure S4 ) . Further observations showed that the transgenic lines of HY5-ΔN77 displayed shortened-hypocotyl in Col-0 background , but the HY5-ΔN77 partially complemented the hy5 long hypocotyl phenotype . More importantly , ACC did not significantly promote the elongation of hypocotyl in HY5-ΔN77/Col-0 and HY5-ΔN77/hy5 ( Figure 5 ) . These results indicate that the COP1-HY5 interaction is required for the degradation of HY5 by ethylene , and subsequent hypocotyl growth . We then generated eto2 cop1-4 and ctr1 cop1-4 double mutants to determine whether ethylene-mediated hypocotyl elongation is COP1-dependent . Both of the mutants displayed hypocotyl lengths that were similar to those of cop1-4 and were not affected by ACC treatment ( Figure S5 ) . To investigate the relationship between COP1 and ethylene signalling , we evaluated the hybrids EIN3ox×cop1-4 and ein3-1×GUS-COP1 . Both EIN3ox and GUS-COP1 displayed longer hypocotyls than the controls , indicating that EIN3 and COP1 positively regulate ethylene-promoted hypocotyl growth . However , EIN3ox×cop1-4 , which lacked COP1 , displayed a short hypocotyl phenotype that was similar to that of cop1-4 , and ein3-1×GUS-COP1 , which lacked EIN3 , showed the same long hypocotyl phenotype as GUS-COP1 ( Figure 6A and 6B ) , indicating that COP1 might be genetically downstream of EIN3 . This conclusion was supported by Western blotting analyses: as shown in Figure 6C , twice as much HY5 protein accumulated in EIN3ox×cop1-4 as in cop1-4 , and this accumulation was not affected by ACC treatment . Correspondingly , the HY5 level in ein3-1×GUS-COP1 was dramatically reduced compared to that in the controls ( Figure 6C ) , consistent with the function of HY5 , COP1 likely acts downstream of EIN3 in ethylene-promoted hypocotyl elongation . Thus , our results demonstrate that the ethylene-enhanced degradation of HY5 is mediated by COP1 and that the COP1-HY5 complex acts downstream of EIN3 to mediate ethylene-promoted hypocotyl growth . To further confirm the ethylene responsiveness of the physical interaction between COP1 and HY5 , we examined the expression of the ethylene-regulated genes ERF1 [16] , ESE1 [21] , and CHIB [50] in hy5 . Our data show that the expression of ERF1 , ESE1 , and CHIB was inhibited in cop1-4 , but increased in hy5 ( Figure S6 ) , indicating the participation of COP1-HY5 not only in ethylene-promoted hypocotyl growth , but also in ethylene-regulated gene expression . Interestingly , we did not observe an obvious difference in root length between Col-0 and the ethylene signalling mutants in the absence of ACC . In particular , root growth in the mutants eto2 and ctr1-1 was constitutively inhibited . However , ACC significantly diminished root growth in Col-0 , but not in ein2 or the ein3 eil1 double mutant , and it partially inhibited root growth in etr1-1 and ein3-1 ( Figure S7A ) , demonstrating the essential role of ethylene in the regulation of root growth . However , the root length in the hy5 or cop1 mutants and the transgenic lines expressing truncated versions of COP1 was reduced by treatment with ACC ( Figure S7B ) , indicating that the COP-HY5 complex mainly takes part in ethylene-regulated hypocotyl growth rather than root growth . COP1 activity has been correlated with its movement between the cytoplasm and nucleus [35] . COP1 is excluded from the nucleus in the light , whereas in the dark or shade it accumulates in the nucleus and directs its targets , including HY5 , to the proteasomal degradation machinery [36] , [47]–[49] , [51] , [52] . This nucleocytoplasmic partitioning of COP1 is required to promote HY5 protein accumulation [53] , [54] . To determine whether ethylene influences COP1 localisation to the nucleus , we quantified the nuclear and cytoplasmic localisation of COP1 in a transgenic line that constitutively expresses a GUS-COP1 fusion ( GUS-COP1 ) . GUS was mainly localised to the nucleus in the dark , but it translocated to the cytosol upon light irradiation in the absence of ACC ( Figure 7A and 7B ) , consistent with previous data [35] . Importantly , ACC treatment did not induce obvious changes in GUS localisation in the dark , whereas under continuous white light for 24 h , ACC treatment enhanced the nuclear localisation of GUS-COP1 ( Figure 7A ) , suggesting that ACC affects the distribution of COP1 and hypocotyl growth in the light rather than in the dark . Further , we used 4-day-old seedlings grown in long days ( 16/8 h ) , and then the plants were shifted to continuous white light for 36–48 h to conduct experiments . The level of GUS staining in the nucleus did not change following treatment with ACC under continuous white light ( Figure S8 ) . Moreover , under an 8-h/16-h dark/light cycle approximately 18% of the hypocotyl cells exhibited nucleus-enriched GUS staining in the absence of ACC; however , after addition of ACC under the same growth conditions , the proportion of cells with nucleus-enriched GUS staining increased to 58% ( Figure S9A–D ) . These results indicate that ACC reversed the localisation of COP1 in the light and further enhanced HY5 degradation . Because COP1 is a downstream component of EIN3 , we examined whether ethylene modulates COP1 localisation via the action of EIN3 . To address this issue , we constructed a hybrid of GUS-COP transgenic line using the ein3-1 mutant to detect GUS-COP1 shuttling between the cytoplasm and nucleus and grew the hybrid under continuous white light for 24 h . Our results indicate that the loss of function of EIN3 in the ein3×GUS-COP1 hybrid disabled the shuttling of COP1 to the nucleus promoted by ACC . In comparison , the shuttling of COP1 in ein3×GUS-COP1 was controlled by darkness ( Figure 7B ) . These results demonstrate that ethylene modulates COP1 localisation via the action of EIN3 . To directly demonstrate that ACC promotes the shuttling of COP1 between the nucleus and cytoplasm , we performed cell fractionation experiments using 5-day-old Col-0 plants grown under an 8-h/16-h dark/light cycle . To exclude the influence of de novo protein production and degradation during the process , we supplemented with cycloheximide ( CHX ) and MG132 with or without ACC in the light for 15 h . Our results showed that COP1 was mainly localized in the cytoplasm in the light without ACC treatment , while it shuttled to the nucleus in the presence of ACC ( Figure 8A ) , evidencing that ACC enriches the nuclear distribution of COP1 in the light . Therefore our data suggest that ACC promotes COP1 translocation to the nucleus , resulting in the subsequent degradation of HY5 via 26S proteolysis . This experiment was also performed using ein2 and ein3-1 plants . In accordance with the localisation of GUS-COP1 in the ein3-1 background , COP1 was only weakly detected in the nucleus with or without ACC ( Figure 8B ) , demonstrating that ethylene-regulated COP1 shuttling occurs downstream of EIN3 .
To understand the interactions between environmental cues and phytohormones , it is critical to elucidate the coordination of ethylene and light signalling during seedling hypocotyl growth . A recent study showed that EIN3 transcriptionally activates the expression of PIF3 and ERF1 to integrate these pathways [39] , revealing the transcriptional regulation of a light regulator by ethylene . The bZIP transcriptional regulator HY5 has been shown to be a key factor in hypocotyl growth through its interaction with COP1 as a photomorphogenic repressor [33] , [44] . In the present report , we demonstrated that the light regulators COP1 and HY5 are essential for ethylene-promoted hypocotyl growth in the light but not in the dark , and that these proteins act downstream of the ethylene signalling pathway component EIN3 . Importantly , the ethylene-dependent activity of EIN3 enhanced the shuttling of COP1 between the nucleus and cytoplasm to promote HY5 degradation . Therefore , to our knowledge , this study presents the first evidence that the COP1-HY5 complex integrates ethylene-promoted hypocotyl growth in the light , increasing our understanding of the mechanism by which light suppresses ( while ethylene promotes ) hypocotyl elongation in the light . Ethylene suppresses hypocotyl growth through the ethylene-induced gene ERF1 whereas it promotes hypocotyl elongation downstream of EIN3 via PIF3 in the light [39] . This is consistent with the finding that PIF3 and HY5 independently regulate PHYB-mediated hypocotyl elongation inhibition [55] . Moreover , regulation of the biosynthesis , transport , and distribution of IAA by ethylene was observed during light-mediated hypocotyl growth , further confirming the transcriptional regulation of hypocotyl growth by ethylene [29] , [40] . In addition , increasing evidence indicates that the components of the ethylene signalling pathway play critical roles in plant growth and development [56]–[58] . In the present investigation , we demonstrated that ethylene-promoted hypocotyl growth in the light , but not in the dark , is mediated by the photomorphogenic factor HY5 . The hypothesis that HY5 functions downstream of the ethylene signalling pathway is further supported by biochemical assays demonstrating that , under normal growth conditions , HY5 protein levels were not altered in loss-of-function ethylene signalling mutants but were significantly depressed in the ctr1-1 background . Moreover , the addition of ACC greatly reduced HY5 accumulation in eto2 , ctr1-1 , and transgenic EIN3ox plants but not in etr1-1 , ein2 , ein3 , and ein3-1 eil1 plants . Therefore , the results of our genetic and phenotypic analyses and biochemical assays suggest that HY5 represents a novel negative regulator that functions downstream of the ethylene signalling component EIN3 and participates in ethylene-promoted hypocotyl growth . The direct interaction between COP1 and CSN1 reportedly stimulates the nuclear localisation of COP1 [59] , and the plant CSN-interacting F-box protein COP9 INTERACTING F-BOX KELCH 1 contributes to the control of hypocotyl elongation [60] . Indeed , several components of the photomorphogenic response , including CSN3 , CSN6A , and CSN6B , interact with EIN2 . It has been reported that EIN2 probably functions to direct the activity of the CSN through ENHANCED ETHYLENE RESPONSE 5 , which is part of the ethylene response pathway [61] , suggesting that photomorphogenic factors are involved in ethylene signalling . Importantly , we found that the EIN3-dependent nuclear localisation of COP1 could be enhanced by ethylene in the light but not in the dark , suggesting that ethylene affects the shuttling of COP1 to the nucleus . The cop1 mutant has been shown to display a photomorphogenic response in the dark but a short hypocotyl in the light [35] . In the present study , exogenous ACC and endogenous ethylene stimulated hypocotyl growth , but this stimulation was not observed in cop mutant plants . These results suggest that ethylene-mediated hypocotyl growth is COP1-dependent , consistent with the finding that COP1 is involved in the ethylene-promoted photomorphogenic response [19] , [40] . We found that cop1 mutants have shorter hypocotyls , while eto2 and ctr1 display longer hypocotyls , in the light . The short hypocotyl lengths of the eto2 cop1-4 and ctr1 cop1-4 double mutants , however , were similar to that of the cop1-4 mutant , and this phenotype was unaffected by the addition of ACC . Furthermore , genetic analyses and biochemical assays showed that the hypocotyl length and HY5 stability in the hybrids EIN3ox×cop1-4 and ein3-1×GUS-COP1 were consistent with the function of COP1 , indicating that COP1 acts downstream of the ethylene signalling component EIN3 , similar to the manner in which HY5 is regulated by EIN3 during hypocotyl growth . Interestingly , we also found that ACC enhanced the movement of COP1 to the nucleus in root cells upon light illumination ( Figure S9E and S9F ) , suggesting the general regulation of the movement of COP1 in the light by ethylene . However , root elongation in hy5 , cop1 , and Col-0 seedlings was not significantly affected by ACC treatment , suggesting that the effect of ACC/ethylene on root growth is independent of COP1-HY5 . Moreover , the length of the root in the ethylene signaling mutants was similar to that in Col-0 in the absence of ACC treatment . After ACC treatment , root growth was completely abolished in Col-0 , but it was similar in terms of length in ein2 and ein3-1 eil1 with the control , implying that ethylene is essential for root growth . Reflecting the redundancy of EIN3 and EIL1 , root growth in ein3-1 and the gain-of-function mutant etr1-1 was partially inhibited , compared to the corresponding controls . This observation suggests that the regulatory role of ethylene in hypocotyl growth is different from that in root elongation , further supporting the regulation of HY5 in hypocotyl growth rather than in root growth . Thus , the involvement of HY5 in the ethylene response might be an organ-specific effect . Importantly , the data from our phenotypic and biochemical analyses indicate that a loss of function of CTR1 constitutively improves hypocotyl growth by decreasing the level of HY5 , while the hypocotyl length in ctr1 is similar to that in ACC-treated Col-0 , but shorter than that in hy5 , suggesting that other factors coordinate with ethylene to affect HY5 protein levels . This hypothesis is supported by the observation that ACC moderately induces but does not significantly improve growth of the hypocotyl in the hy5 mutant , consistent with previous observations [40] , indicating that ethylene coordinates with other phytohormones , including auxin , to affect hypocotyl growth [62] , [63] . The finding that ethylene promotes hypocotyl elongation in the light by increasing nuclear-localised COP1 , resulting in a decreased level of HY5 , combined with the observation that the effect of ethylene on HY5 protein levels was abolished in etr1-1 , ein2 , ein3-1 , and ein3/eil1 , indicates that the effect of ethylene on hypocotyl growth requires transcriptional regulation by EIN3 . This hypothesis is supported by the following observations . EIN3 overexpression decreased the HY5 protein level in a Col-0 background but not in a cop1-4 background , whereas COP1 overexpression reduced the HY5 level irrespective of EIN3 . Moreover , the loss of function of EIN3 abolished the movement of COP1 in the light , suggesting that EIN3 is required for the COP1-mediated decrease in HY5 . Therefore , based on the present data and previous reports showing that ethylene promotes EIN3/EIL1 accumulation whereas high EIN3/EIL1 levels are detrimental to plant growth and development [14] , [15] , we propose that COP1-HY5 functions as a regulatory module in ethylene-promoted hypocotyl growth . In the absence of ethylene , COP1 accumulates mainly in the cytoplasm in the light , releasing the suppression of COP1 on HY5 . In the presence of ethylene , this effect is reversed , even though in the light the nucleus is enriched with COP1 , resulting in the degradation of HY5 . Ethylene production is negatively regulated by HY5 [64] , [65]; decreases in HY5 accumulation enhance the synthesis of ethylene . Thus , in the present study , we found that the COP1-HY5 complex acts as an integrator downstream of EIN3 to fine-tune the regulation of hypocotyl growth by light and ethylene . Although previous data [19] , [39] , [40] and our findings suggest that cotyledon greening and hypocotyl growth are COP1-related , multiple factors downstream of COP1 may participate in these growth processes . Additional genetic and biochemical tests should be performed to better understand this integration .
All Arabidopsis thaliana mutants and transgenic lines were generated in the Columbia ( Col-0 ) background with the exception of overexpression line of GUS-COP1 , overexpressing COP1 without the RING-finger domain ( ΔRING , Nossen , No-0 background ) and hy5-ks50 ( Wassilewskija , Ws background ) . Homozygous eto1-1 ( CS3072 ) , eto2 ( CS8059 ) , eto3 ( CS8060 ) , etr1-1 ( CS237 ) , ctr1-1 ( CS8057 ) , ein2 ( CS8844 ) , ein3-1 ( CS8052 ) , and hy5 ( SALK_096651 ) lines were obtained from the Arabidopsis Biological Resource Center ( Columbus , OH ) . The sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: COP1 ( At2g32950 ) , HY5 ( At5g11260 ) , HYH ( At3g17609 ) , ETR1 ( Ag1g66340 ) , CTR1 ( At5g03730 ) , EIN2 ( At5g03280 ) , EIN3 ( At3g20770 ) , EIL1 ( At2g27050 ) , ERF1 ( AT3G23240 ) , ESE1 ( AT3G23220 ) , and CHIB ( AT3G12500 ) . The mutants or transgenic plants hy5-215 and hy5-ks50 [41]; cop1-4 and cop1-6 [45]; ein3-1 eil1 , EIN3ox , and EIN3ox×cop1-4 [19]; 35S::HY5 [44]; ΔRING , ΔCoil , and ΔRΔC [46]; HY5-ΔN77/Col-0 and HY5-ΔN77/hy5 driven by the CaMV35S promoter [33] were generated previously . Seeds were surface-sterilised and sown on Murashige and Skoog ( MS ) medium containing 3% sucrose and 0 . 5% Phytagel . The plates were chilled at 4°C in the dark for 3 days and then moved to 22°C under a 16-h white light ( 50 µmol/m2s ) /8-h dark cycle . All of the chemicals used were obtained from Sigma-Aldrich ( St . Louis , MO ) . Total RNA was extracted from 5-day-old seedlings using TRIzol reagent ( Invitrogen , Carlsbad , CA ) and treated with RNase free-DNase I ( Promega , Madison , WI ) before the latter procedures were performed . Five micrograms of total RNA were reverse-transcribed to cDNA with M-MLV reverse transcriptase ( Promega ) according to the manufacturer's instructions . Gene expression was measured by qPCR analysis ( SYBR Premix; Takara Bio Inc . , Shiga , Japan ) . All amplification reactions were performed in 96-well optical reaction plates with 45 cycles of denaturation for 15 s at 95°C , annealing for 20 s at 56°C , and extension for 45 s at 72°C . The expression levels were normalised to that of TUB4 . The primers used for qPCR are listed in Table S1 . Each qPCR was repeated independently three times with the same expression pattern . For the measurement of hypocotyl length , surface-sterilised seeds were deposited on plates containing MS medium with or without ACC . The plant materials and growth conditions were described in the previous section . At least 50 seedlings were measured from digital images of 5-day-old seedlings using ImageJ software . Proteins were extracted from approximately 50 seedlings treated with or without ACC in 100 µL of extraction buffer ( 50 mM Tris , pH 7 . 5 , 150 mM NaCl , 1 mM EDTA , 10 mM NaF , 25 mM β-glycerophosphate , 2 mM sodium orthovanadate , 10% glycerol , 0 . 1% Tween 20 , and 1 mM PMSF ) . After centrifugation , 10 µg of the supernatant was separated via SDS-PAGE and Western blotting was performed as described previously [36] using anti-HY5 antibodies ( 1∶500 ) . TUB4 ( 1∶5 , 000 ) was used as a loading control . The relative protein levels of HY5 were calculated using Image Gauge V3 . 12 ( Fujifilm , Tokyo , Japan ) . The values were normalised to 100 in Col-0 without ACC treatment based on the TUB4 signal as an internal loading control . 35S::HY5 and eto2 plants were crossed , and putative homozygous F2 progeny were screened for kanamycin resistance during seed germination . Kanamycin-resistant homozygous eto2 seedlings were identified as described previously [66] . Non-segregating lines of the progeny of the eto2 homozygous kanamycin-resistant plants were screened to identify homozygous hybrids of eto2×35S::HY5 . The double mutants ein2 hy5 , ein3-1 hy5 , and etr1-1 hy5 were generated by crossing the recessive ethylene mutant ein2 , ein3-1 , or etr1-1 with hy5 . The F2 progeny of the crosses were grown in the dark for 4 days with 10 µM ACC to isolate ethylene-insensitive individuals that were homozygous for ein2 , ein3-1 , or etr1-1 . The isolated individuals were subsequently screened for the presence of long hypocotyls , and the homozygous hy5 mutations were confirmed by PCR-based genotyping . The double mutants eto2 cop1-4 and ctr1-1 cop1-4 were generated by crossing the ethylene mutant eto2 , or ctr1-1 with cop1-4 . The F2 progeny from the crosses were grown in the dark for 4 days to isolate homozygous cop1-4 lines by the phenotype . Individuals were subsequently screened by PCR-based genotyping , and the PCR products were digested with AvaI ( New England Biolabs , Ipswich , MA ) for eto2 or Tsp509I ( New England Biolabs ) for ctr1-1 to identify homozygous double mutant lines . The primer information is summarised in Table S1 . ein3-1 and GUS-COP1 plants were crossed , and the resulting putative homozygous plants ( F2 progeny ) were screened for kanamycin resistance during seed germination . After the identification of non-segregating progeny of the ein3-1 homozygous plants , seedlings were screened to identify homozygous hybrids of ein3-1×GUS-COP1 . Transgenic GUS-COP1 seeds were germinated and grown on MS medium for 5 days . After treatment with or without 25 µM ACC for 24 h in the dark , continuous light , or light-dark ( 16 h/8 h ) conditions , the seedlings were fixed with cold acetone and then soaked in GUS-DAPI buffer ( 0 . 5 µg/mL DAPI , 0 . 5 mg/mL 5-bromo-4-chloro-3-indolyl-β-D-glucuronic acid , 5 mM ferricyanide and ferrocyanide , 10 mM EDTA , and 0 . 1% [v/v] Tween 20 in 100 mM sodium phosphate , pH 7 . 0 ) and left in the dark at 37°C overnight . The tissues were cleared in 70% ( v/v ) ethanol and mounted on glass slides . To quantify the nuclear-cytoplasmic localisation of GUS-COP1 , we calculated the percentages of cells that exhibited nucleus-enriched GUS staining . The cells in which GUS-COP1 staining was stronger in the nucleus than in the cytoplasm were scored as ‘nuclear , ’ and the ratio of these cells to the total number of cells that displayed GUS staining , including both nucleus-enriched cells and cells displaying evenly distributed GUS staining between the nucleus and cytoplasm , was calculated . At least 10 seedlings with 20 cells per seedling were analysed for each GUS fusion under both the MS and ACC conditions . Nuclear proteins were extracted at 4°C using a CelLytic PN Extraction Kit ( Sigma-Aldrich ) as described previously [67] with minor modifications . After the samples grown on MS medium for 5-days were treated with 50 µM CHX and 5 µM MG132 ( to exclude de novo protein production and degradation during the process ) , supplemented with or without 25 µM ACC , in the light for 15 h , the seedlings ( 3 g ) were ground to a fine powder in liquid nitrogen using a precooled mortar and pestle . Next , 15 mL of 1× NIB were mixed with the samples . The suspension was passed through a 100-µm filter mesh into 50-mL tubes . The organelles in the tubes , including the nuclei , were pelleted by centrifugation at 1 , 200 g for 10 min , and the supernatant , including the cytoplasmic fraction , was collected and analysed as the soluble fraction . The pellet was completely resuspended in 1 mL of 1× NIBA ( NIB buffer containing a protease inhibitor cocktail ) , and the organelle membranes were lysed by adding 10% Triton X-100 to a final concentration of 0 . 3% . To produce a semi-pure nuclear preparation , the lysates were applied to a 0 . 8-mL cushion of 1 . 5 M sucrose in 1× NIBA in 1 . 5-mL tubes . After centrifugation at 12 , 000 g for 10 min , the upper phase and sucrose cushion were removed , and the pellet was washed twice with NIBA buffer . The nuclear pellet was resuspended in 50 µL of nuclear extraction buffer and vortexed for 30 min . The insoluble material was removed by centrifugation at 12 , 000 g for 10 min . The final nuclear protein fraction was transferred to a new tube and stored at −80°C . | It is well known that light suppresses hypocotyl growth in seedlings , while the phytohormone ethylene and its precursor 1-aminocyclopropane-1-carboxylate ( ACC ) enhance hypocotyl growth in the light . However , the mechanism by which light and ethylene oppositely affect this process at the protein level is unclear . Here , we demonstrate that ethylene enhances the movement of CONSTITUTIVE PHOTOMORPHOGENESIS 1 ( COP1 ) to the nucleus where it promotes the degradation of LONG HYPOCOTYL 5 ( HY5 ) in the light , contributing to hypocotyl growth . Our data indicate that HY5 is required for ethylene-promoted hypocotyl growth in the light , but not in the dark . Using genetic and biochemical analyses , we found that HY5 functions downstream of ETHYLENE INSENSITIVE 3 ( EIN3 ) during ethylene-promoted hypocotyl growth . Further , the regulation of HY5 stability by ethylene is COP1-dependent , and COP1 is genetically located downstream of EIN3 , indicating that the COP1-HY5 complex integrates light and ethylene signalling downstream of EIN3 . Importantly , ACC enriched the nuclear localisation of COP1 in an EIN3-dependent manner in the presence of light , suggesting that ethylene rescued the effects of light on the movement of COP1 from the cytoplasm to the nucleus . Thus , our investigation shows that the COP1-HY5 complex is a novel integrator that plays an essential role in ethylene-promoted hypocotyl growth in the light . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Ethylene Promotes Hypocotyl Growth and HY5 Degradation by Enhancing the Movement of COP1 to the Nucleus in the Light |
Infectious diseases impose considerable burden on society , despite significant advances in technology and medicine over the past century . Advanced warning can be helpful in mitigating and preparing for an impending or ongoing epidemic . Historically , such a capability has lagged for many reasons , including in particular the uncertainty in the current state of the system and in the understanding of the processes that drive epidemic trajectories . Presently we have access to data , models , and computational resources that enable the development of epidemiological forecasting systems . Indeed , several recent challenges hosted by the U . S . government have fostered an open and collaborative environment for the development of these technologies . The primary focus of these challenges has been to develop statistical and computational methods for epidemiological forecasting , but here we consider a serious alternative based on collective human judgment . We created the web-based “Epicast” forecasting system which collects and aggregates epidemic predictions made in real-time by human participants , and with these forecasts we ask two questions: how accurate is human judgment , and how do these forecasts compare to their more computational , data-driven alternatives ? To address the former , we assess by a variety of metrics how accurately humans are able to predict influenza and chikungunya trajectories . As for the latter , we show that real-time , combined human predictions of the 2014–2015 and 2015–2016 U . S . flu seasons are often more accurate than the same predictions made by several statistical systems , especially for short-term targets . We conclude that there is valuable predictive power in collective human judgment , and we discuss the benefits and drawbacks of this approach .
It is perhaps unsurprising that societal advances in technology , education , and medicine are concomitant with reduced morbidity and mortality associated with infectious diseases [1] . This is exemplified by the staggering drop in infectious disease mortality in the United States during the 20th century [2] . Yet despite continued technological advances , the U . S . death rate due to infectious disease has not improved significantly since around the 1960s , and in fact has been rising since the 1980s [3] . It is evident that incremental advances in medical treatment are failing to reduce overall infectious disease mortality . One way in which the situation can be improved is through expanding our capacity for preparedness and prevention—we need forewarning [4] . This is the defining problem out of which the nascent field of epidemiological forecasting has risen . There is widespread interest in predicting disease outbreaks to minimize losses which would otherwise have been preventable given prior warning . In recent years , the U . S . Centers for Disease Control and Prevention ( CDC ) has sponsored three challenges to predict Influenza epidemics [5–7] , Defense Advanced Research Projects Agency ( DARPA ) has sponsored a challenge to predict the invasion of chikungunya [8 , 9] , several agencies working together under the Pandemic Prediction and Forecasting Science and Technology ( PPFST ) Working Group have sponsored a challenge to predict Dengue outbreaks [10] , and the Research and Policy for Infectious Disease Dynamics ( RAPIDD ) group of the National Institutes of Health hosted a workshop for forecasting Ebola outbreaks [11] . Given the magnitude of time , energy , and resources collectively invested in these challenges by both participants and organizers , it is critical that qualitative and quantitative assessments be made to help understand where epidemiological forecasting excels and where it lags . As exemplified by the fields of meteorology and econometrics , statistical and computational models are frequently used to understand , describe , and forecast the evolution of complex dynamical systems [12 , 13] . The situation in epidemiological forecasting is no different; data-driven forecasting frameworks have been developed for a variety of tasks [14–16] . To assess accuracy , forecasts are typically compared to pre-defined baselines and to other , often competing , forecasts . The focus has traditionally been on comparisons between data-driven methods , but there has been less work toward understanding the utility of alternative approaches , including those based on human judgment . In addition to developing and applying one such approach , we also provide an intuitive point of reference by contrasting the performance of data-driven and human judgment methods for epidemiological forecasting . Methods based on collective judgment take advantage of the interesting observation that group judgment is generally superior to individual judgment—a phenomena commonly known as “The Wisdom of Crowds” . This was illustrated over a century ago when Francis Galton showed that a group of common people was collectively able to estimate the weight of an ox to within one percent of its actual weight [17] . Since then , collective judgment has been used to predict outcomes in a number of diverse settings , including for example finance , economics , politics , sports , and meteorology [18–20] . A more specific type of collective judgment arises when the participants ( whether human or otherwise ) are experts—a committee of experts . This approach is common in a variety of settings , for example in artificial intelligence and machine learning in the form of committee machines [21] and ensemble classifiers [22] . More relevant examples of incorporating human judgment in influenza research include prediction markets [23 , 24] and other crowd-sourcing methods like Flu Near You [25 , 26] . Here we assess the performance of a forecasting framework based on collective human judgment—“Epicast” . In particular , we assess its performance as a competitor in the aforementioned influenza and chikungunya forecasting challenges . Each of these challenges dealt with a different set of data and objectives , and we analyze them separately . For both influenza challenges , the data of interest was population-weighted percent influenza-like illness ( wILI ) in 10 regions of the U . S . and the U . S . as a whole . wILI is syndromic surveillance defined as the percent of patients having flu-like symptoms ( fever over 100°F and either cough or sore throat ) without a known cause other than influenza . wILI is reported voluntarily by health care providers through ILINet and distributed by CDC [27 , 28] . The DARPA chikungunya challenge focused instead on predicting case counts within 55 countries and territories in the Americas . The data of interest was the number of weekly cases ( including suspected , confirmed , and imported ) within each location , cumulatively since the beginning of 2014 . These case counts are published by the Pan American Health Organization ( PAHO ) [29] . In regards to terminology in subsequent discussion , there is an important distinction to be made between a prediction and a forecast . These words are often used interchangeably elsewhere , but here we use them to refer to subtly different concepts . A prediction makes an absolute statement about the future and says nothing about other potential outcomes . In contrast , a forecast is a generalization of a prediction in which a probability is assigned to all possible outcomes . In the case of Epicast , we collect a prediction from each participant—a single possibility . Epicast aggregates many such predictions to produce a forecast—a probability distribution over all possibilities . Because predictions and forecasts make different claims , they are evaluated by different metrics; we use mean absolute error ( MAE ) to assess predictions and mean negative log likelihood ( based on “logarithmic score” [30] , also known as “surprisal” in other contexts [31] ) to produce a figure of merit for forecasts . As part of our evaluation of the Epicast system in forecasting flu , we compare with several competing statistical and/or data-driven systems . For the 2014–2015 flu contest , we compare Epicast with “Empirical Bayes” and “Pinned Spline”; for the 2015–2016 flu contest , we compare Epicast with “Stat” and “ArcheFilter” . All of these systems were serious and successful competitors in their respective contest years . Consequently , the performance of these systems provides a measure of the state of the art in flu forecasting . None of these competing systems were used in both flu contests , hence we compare against a different set of systems in different years . The Empirical Bayes system [32] is based on the notion that future epidemics will generally resemble past epidemics , up to basic transformations and noise . This system defines a prior distribution over wILI trajectories , draws samples from that distribution , assigns a likelihood-based weight to each sample , and finally produces a posterior distribution over trajectories . The reported forecast for each target is derived from the posterior distribution of wILI trajectories . The Pinned Spline system [33] attempts to smoothly interpolate current and past wILI . To do this , two partial wILI trajectories are defined . The first partial trajectory spans the start of the season through the current week and is defined to be wILI as published by CDC for the current season . The second partial trajectory spans the next week through the end of the season and is defined to be the week-wise mean of the wILI trajectories of past epidemics over the same span . Finally , the two partial trajectories are connected with smoothing splines . In subsequent analysis we only evaluate point predictions—not forecasts—made by the Pinned Spline system . The Stat system is a weighted ensemble of statistical methods , including both Empirical Bayes and Pinned Spline . It additionally contains baseline components ( including a uniform distribution and an empirical distribution ) and other non-mechanistic methods ( including delta density and kernel density methods ) . The cross-validation weight assigned to each constituent method is recomputed for each forecast . Stat forecasts are a weighted combination of the forecasts of each method . The ArcheFilter system [33] assumes that there is a latent archetype wILI trajectory , and that the observed wILI trajectory of each flu epidemic is a transformed , noisy version of the archetype . The ArcheFilter defines this archetype roughly as the peak-aligned , week-wise mean wILI trajectory of all past epidemic seasons . As an epidemic progresses , a Kalman filter is used to estimate the time-shift and wILI-scale parameters that , when applied to the archetype , most parsimoniously explain the observed wILI trajectory of the current epidemic . Uncertainty in the state of the filter—the shift and scale parameter values—gives rise to a distribution over wILI trajectories from which the forecast for each target is derived .
This study was granted Carnegie Mellon University IRB exemption with ID STUDY2015_00000142 . We developed a website [34] for collecting predictions of epidemiological time series ( Fig 1 ) . Participants were shown a partial trajectory and were asked to hand-draw a continuation of the trajectory as a prediction . At regular intervals , user-submitted trajectories were collected and an aggregate forecast was generated . Participants were not shown the predictions made by other participants . We produced , in real-time , forecasts for the 2014–2015 and 2015–2016 U . S . flu seasons and predictions for the 2014–2015 chikungunya invasion of the Americas . For influenza , we collected predictions from the general public; for chikungunya , we only collected predictions from a set of selected experts in related fields . All three forecasting challenges were carried out as the event ( epidemic or invasion ) progressed; we did not produce retrospective forecasts . Each week during the 2014–2015 and 2015–2016 flu seasons , we asked participants to predict wILI for each remaining week of the season . Each individually submitted prediction was a trajectory of varying length ( depending on the week of submission ) of wILI values , and we asked users to provide such predictions for each of the HHS and U . S . regions . Similarly , once each month , from August , 2014 through January , 2015 , we asked participants to predict the cumulative weekly chikungunya case count in each of the 55 Pan American Health Organization ( PAHO ) locations through the end of February 2015 . Each predicted trajectory was extended to cover the entire time series of interest by concatenating the observed time series at the time of submission with the predicted time series . By doing this , all time series were made to have the same length ( for example , 32 weekly wILI values spanning the 2014–2015 flu season ) . The objectives of the influenza challenges were to forecast a number of features , or “targets” , of the weekly wILI time series . These included: “Peak Height” , the maximum value of wILI reported throughout the flu season; “Peak Week” , the Morbidity and Mortality Weekly Report ( MMWR ) week number [35] on which wILI reaches its maximum value; and the next four values of wILI , called “1–4 Week Lookaheads” . Each of these values was forecasted separately for each of the ten Health and Human Services ( HHS ) regions of the U . S . and also for the U . S . as a whole—a super-region which we include in our analysis as one of eleven total regions . The forecast for each target in each region consisted of a set of probability bins over a pre-defined range of outcomes and a prediction of the single most likely outcome . The chikungunya challenge objective was simply to predict the trajectory of cumulative case counts in each country or territory . Of the six flu targets , two are season-wide: Peak Week and Peak Height . The rest—the 1–4 Week Lookaheads—are short-term targets . Peak Week is an integer representing the number of weeks elapsed between the start of the season and the week during which wILI peaks . The start of the season is defined as MMWR week 40 of the first year of the season , written as “2014w40” or “2015w40” , depending on the season . The remaining five targets are measured in wILI on a continuous scale from 0% to 100% . An additional season-wide target , the MMWR week of epidemic onset , is discussed in S1 Text . The Epicast point prediction for any target was defined as the median of the target values measured on user predictions . The Epicast forecast for any target was a Student’s t distribution with location equal to the median value ( the point prediction ) , scale equal to the sample standard deviation of values , and degrees of freedom equal to the number of participants . The same general methodology was used for both forecasting challenges , with the exception that we produced both a prediction and a forecast for influenza and only a prediction for chikungunya . We primarily assess the quality of predictions in terms of mean absolute error ( MAE ) . Given a set of N true outcomes , y , and corresponding predictions , y ^ , MAE can be written as: MAE = 1 N ∑ i = 1 N ∣ y i - y ^ i ∣ . We further assess wILI and case-count predictions ( for example , Peak Height ) by measuring how often predictions fall within some range ( ± 10% , 20% , 30% , 40% , or 50% ) of ground truth . We similarly assess predictions of weeks ( for example , Peak Week ) by measuring how often predictions fall within some range ( ± 1 , 2 , 3 , 4 , or 5 weeks ) of ground truth . We report the fraction of the time that predicted values fall within each range , aggregated over regions and potentially also weeks . We refer to these analyses as “fraction of predictions accurate within a target range” . We assess the quality of flu forecasts in terms of a likelihood-based score . We define the “log score” as the negative average of the logarithm of the probability assigned to a range of values surrounding the true outcome . In the case of Peak Week , we consider the log score of the range of the actual Peak Week plus or minus one week ( for example , if the Peak Week was 5 , we compute the log likelihood of the probability assigned to a peak being on week 4 , 5 , or 6 ) . Suppose that PkWk r obs denotes the observed value of Peak Week in region r and that P ( ⋯ ) represents the probability assigned by the forecaster to a given outcome . Then the score across all regions can be written as: score = - 1 11 ∑ r = 1 11 log P ( PkWk r ∈ [ PkWk r obs - 1 , PkWk r obs + 1 ] ) . For the five wILI targets , we only have available a set of probability bins each of width 1 wILI , as this is what was required by CDC’s 2014–2015 flu contest . To determine which bins to include in the likelihood calculation , we select ( 1 ) the wILI bin containing the actual value and ( 2 ) the adjacent wILI bin nearest to the actual value . For example , the actual Peak Height in the U . S . National region was 6 . 002 , and we select the two bins which together give the probability assigned to the event that actual Peak Height falls between 5 and 7 . Suppose a forecast was made that P ( 5 ≤ wILI < 6 ) = 0 . 215 and P ( 6 ≤ wILI < 7 ) = 0 . 412; the log score assigned to this forecast is − log ( 0 . 215 + 0 . 412 ) = 0 . 467 . For Peak Height ( and similarly for the Lookahead targets ) across all regions: score = - 1 11 ∑ r = 1 11 log P ( PkHt r ∈ [ round ( PkHt r obs ) - 1 , round ( PkHt r obs ) + 1 ] ) . To compensate for the varying difficulty over time of predicting and forecasting , we often treat accuracy as a function of “lead time” , the number of weeks preceding the region-specific Peak Week . Positive , zero , and negative lead times indicate predictions and forecasts made before , on , and after the epidemic peak , respectively . We consider lead times for the 2014–2015 flu season that range from +10 to −10 weeks . However , due to the unusually late Peak Week within most regions in the 2015–2016 flu season , we constrain lead times in that season to the range of +10 to −5 weeks . To contextualize the accuracy of both predictions and forecasts , we compare Epicast with individual participants and/or other forecasting methods . To further contextualize log scores , we show also the log score of a hypothetical “Uniform” system in which uniform probability is assigned to all plausible outcomes . For Peak Week , we define this as a uniform distribution over weeks 2014w46 through 2015w12 and 2015w45 through 2016w12 ( p = 1 20 per week , per season ) , and for the wILI targets we define this as a uniform distribution over 0% to 12% wILI ( p = 1 12 per bin ) . The Uniform system is intended to provide a lower bound on the performance of a reasonable forecaster . Our main challenge in presenting results is that the space in which comparisons can be made consists of several orthogonal dimensions: regions ( U . S . and 10 HHS regions ) , targets ( Peak Week , Peak Height , and 1–4 Week Lookaheads ) , season weeks ( depending on season and target , up to 32 ) , and error metrics ( MAE and log score ) . To concisely compare system performance , we are given the non-trivial task of reducing this dimensionality , otherwise we would come to thousands of separate figures of merit . Several confounding issues impede aggregation along any one axis; forecasting difficulty varies over time as the season progresses , the various regions may peak at different times in the season , long-term targets are often more difficult to predict than short-term targets , and targets are measured in different units . To work around these complications in the case of point predictions , we rank systems and participants in terms of absolute error and perform our analysis on the relative ranking assigned to each predictor . More specifically , we consider the pairwise ranking in absolute error of Epicast versus individual participants and statistical frameworks . For each lead time , region , and target , we ask whether Epicast or the competitor had a smaller absolute error , and we measure the fraction of instances where Epicast had the smaller error—a “Win Rate” . To assess the statistical significance of each result , we use a Sign test with the null hypothesis that the pair of forecasters is equally accurate . It should be noted that this test assumes that all observations are independent , but results across adjacent weeks , for example , are likely to be correlated to some extent . We define ground truth to be the version of wILI published by CDC 15 weeks after the end of each flu season—MMWR week 35 . Specifically , we use values published on 2015w35 and 2016w35 for evaluating the results of the 2014–2015 and 2015–2016 flu contests , respectively .
For the 2014–2015 flu season we gathered a total of 5 , 487 trajectories from a set of 48 volunteer participants during the 32 week period spanning 2014w41 through 2015w19 . For the 2015–2016 flu season we gathered a total of 3 , 833 trajectories from a set of 23 volunteer participants during the 30 week period spanning 2015w42 through 2016w19 . Participants varied in self-identified skill , from experts in public health , epidemiology , and/or statistics , to laypersons . Participation varied over time with an average of 16 participants per week during the 2014–2015 season and 12 participants per week during the 2015–2016 season ( Fig 2 ) . In the following analysis we did not handle expert and non-expert predictions differently , but we compare the performance of the two groups in S1 Text—the experts on average made slightly more accurate predictions . In what follows , we group errors across regions for brevity , but a breakdown of performance within each region is also given in S1 Text . We first consider the fraction of predictions accurate within a target range , aggregated over weeks of the season ( Fig 3 ) . For the four short-term Lookahead targets , the Epicast prediction is within 10% of the actual value just under half the time when predicting one week into the future; this falls to roughly one quarter of the time when predicting 4 weeks into the future . The trend is similar , though perhaps less abrupt , at other accuracy thresholds . Accuracy within 50% is achieved near or above 85% of the time , even predicting up to 4 weeks ahead . We next consider the number of regional predictions accurate within a target range , as a function of lead time ( Fig 4 ) . For 2 , 3 , and 4 weeks ahead , the lead time with lowest accuracy is roughly 2 , 3 , and 4 weeks ahead of the Peak Week , respectively , which suggests that there is a distinct challenge in forecasting the Peak Height . This is to be expected because there is significantly more volatility around the peak of the epidemic . In general , accuracy in season-wide targets rises sharply 2–5 weeks before the epidemic peak and remains high through the remainder of the season . For the 2015–2016 season , short-term accuracy ( relative , not absolute ) is surprisingly low in several regions around 10 weeks before the peak . This is likely due to the fact that the Peak Week in most regions was exceptionally late during this season . As a result , wILI was still near to baseline values for several weeks after predictions were made . Predicting a premature rise in wILI when ground truth was small in magnitude resulted in errors exceeding 50% of the actual value in several regions . Once it became clear that this would likely turn out to be a mild and/or late-peaking season , accuracy rose to nominal pre-peak levels . The remainder of our analysis is focused on comparing the accuracy of Epicast with individual participants and competing methods; we begin with Win Rate ( Fig 5 ) . Overall , considering all targets , Epicast has Win Rates above 0 . 5 ( lower absolute error than a competitor on a majority of predictions ) when compared with all but one individual participant and all four statistical frameworks . In season-wide targets , Epicast performs reasonably well; however , six participants and the ArcheFilter method bring Epicast’s Win Rate below 0 . 5 . In short-term targets , Epicast has Win Rates uniformly above 0 . 5 . Epicast has a Win Rate significantly higher than the Spline method in all categories and significantly higher than the Empirical Bayes and ArcheFilter methods both overall and in short-term targets . Epicast never has a significantly lower Win Rate than any of the competing statistical systems . Next , we compare predictions in terms of MAE . We calculate , separately for each target , MAE across regions as a function of lead time ( Fig 6 ) . In agreement with previous results , Epicast MAE in season-wide targets generally decreases with lead time and is highest in short-term targets when predicting the peak value of wILI . MAE is occasionally elevated in short-term targets on the Peak Week ( lead time = 0 ) , suggesting a relative increase in uncertainty immediately after the true peak ( which is not known at the time to be so ) . Compared with the statistical methods , Epicast particularly excels when predicting short-term targets . Finally , we compare forecasts in terms of log score . Our analysis in this context does not include the 2014–2015 Spline method , but the hypothetical Uniform method is included for both seasons . We compute the average log score for Epicast and competing methods separately for each target and lead time ( Fig 7 ) . In the 2014–2015 season , Empirical Bayes scored within the bounds of the Uniform system more consistently than Epicast . However , in the 2015–2016 season , all systems consistently scored within the Uniform bounds for wILI targets , likely due at least partly to a relatively low peak wILI in this season . Across both seasons , Epicast has average log score in short-term targets as good as , or better than , that of the statistical systems . However , the statistical systems almost uniformly outperform Epicast in season-wide targets . In total , we gathered 2 , 530 trajectories from a set of 12 volunteers with expertise in vector-borne diseases , public health , and/or epidemiology ( Fig 8 ) . Predicting chikungunya fundamentally differed in two ways from predicting flu . First , the chikungunya invasion of the Americas was a rare event for which little historical precedent was available , whereas flu epidemics are a regular occurrence for which we have significant historical data . Second , errors in ( cumulative ) chikungunya predictions accumulated over weeks , whereas errors in ( non-cumulative ) flu predictions were separated out across weeks . While it would have been trivial to convert a cumulative trajectory into a non-cumulative trajectory , the published counts which were defined to be ground truth are only available sporadically over time , preventing us from converting the true cumulative trajectory into a non-cumulative trajectory . The increased difficulty of the task is reflected by a reduction in accuracy . At best ( 1 week ahead ) , less than one in three predictions were within 10% of the actual value; and at worst ( 10 weeks ahead ) , over half of the predictions were off target by more than 50% . Even in such conditions , when comparing pair-wise absolute error between Epicast and each user , Epicast more frequently predicts closer to the true value than any individual user .
Epicast was one of two winning methods in the 2014–2015 flu contest [7] . Epicast was one of three winning methods in the 2015–2016 flu contest [36] ( the other winners were Stat and ArcheFilter ) . We expect a future CDC publication to provide additional details and analysis . Epicast was not selected as one of the six chikungunya challenge winners [37] , but we are told that it ranked in the top quartile of submissions . There are two caveats to point out regarding our incarnation of the crowd prediction method for forecasting flu . First , the ILINet data we showed participants , and also asked them to predict , was subject to weekly revision—in some cases significantly so ( for example , wILI in HHS region 2 on 2015w03 was first reported as 6 . 2% , and then the next week as 5 . 6%; the final , stable value was reported as 5 . 0% ) . The changing values are due to a backfill process whereby data from late-reporting providers is used to retrospectively update prior values of wILI . This is one reason , for example , that MAE after the Peak Week is non-zero; even once the peak has been observed with high confidence , there is still some non-negligible chance that a subsequent update due to backfill will result in a revision of the peak timing . A further discussion of the effects of backfill can be found in S1 Text . Second , the data used in our present analysis is the same data collected for the various contests , but our methodology in the 2014–2015 flu contest differed slightly from what we present here . Namely , our 2014–2015 contest submissions assumed a normal distribution over user inputs , whereas here , and in our 2015–2016 contest submissions , we assumed a Student’s t distribution . There are several important limitations of the human judgment approach relative to purely data-driven methods that should be made clear . First , these results are only representative of two flu seasons and a single chikungunya outbreak . This highlights one of the biggest shortcomings of this approach—collecting predictions is a tedious and time-consuming process . Unlike statistical methods which can be applied retrospectively to any outbreak , the approach here requires a significant amount of work from a large number of participants . For example , because of this we are unable to perform cross validation across seasons . Second , these results do not necessarily provide us with an improved understanding of epidemiological dynamics . In contrast , statistical methods aim to learn from past data in order to better describe and model the epidemic process . On the other hand , the human judgment approach does have unique advantages over purely data-driven systems . Humans have the innate and powerful ability to assimilate , with little to no effort , diverse data sources and considerations . An example of this is using news headlines , which we display within the Epicast interface , to inform predictions . Another advantage of human judgment is the ability to make reasonable predictions for events with little historical precedent , like the outbreak of a new disease or a disease invasion in a new location . The task of predicting trajectories is not trivial , and we asked each of the participants to provide us with many such trajectories over quite a long period of time . This resulted in some tedium , which we suspect is the reason for the relatively high attrition rate in the flu Epicast . There are many guidelines describing ways to make crowd work streamlined and sustainable , and we made every effort to implement these ideas . To minimize the overall amount of effort required and to streamline the process as much as possible we: allowed participants to use their previously entered forecasts as a starting point; accepted any number of regional predictions ( not requiring all eleven to be completed ) ; reduced the entire process to one drag and one click per region; and sent URLs tailored with a unique identifier via email each week to bypass having to manually login . Additionally , we tried to increase interest and participation by including a leader board of both weekly and overall high scores . We also had the competing objective of collecting the most informed forecasts from the participants . To this end we included a section of links to educational resources , and for the flu Epicast we embedded within each participant’s home page a Google news feed on the topic of “flu” . Epicast is not well suited for all forecasting situations . “Wisdom of crowds” methods are robust to high variance among individual predictions , but require that the overall distribution of predictions is unbiased . By showing wILI trajectories of past flu epidemics on Epicast’s forecasting interface , we undoubtedly bias predictions toward typical flu seasons . While this may be beneficial when forecasting a typical flu season , it is almost certainly harmful when predicting highly atypical flu epidemics , and especially pandemics . Epicast is not robust to “long-tail” events such as these . In general , Epicast is best suited for situations where: ( 1 ) the event is regularly occurring , ( 2 ) historical surveillance data is available for many examples of the event , and ( 3 ) ongoing surveillance data is available with relatively short lag in comparison to the length of the event . This may explain the increased difficulty of Epicast in predicting the chikungunya invasion: it was a one-time event with no historical data and relatively lagged and intermittently available ongoing data . Alternative prediction methods which solicit and aggregate human judgment exist; the Delphi method [38] is one such example . Both the Delphi method and the Epicast method herein collect predictions from human participants and produce a single output prediction . However , the Delphi method is iterative , requiring more time , effort , and coordination . The Delphi method requires from each participant not only a prediction , but also reasoning or justification for that prediction . Then , all participants are shown the predictions , and reasons for them , of all other participants . Participants are then given the opportunity to revise their predictions , and the process continues iteratively either until convergence or some other stopping criteria are met . One of the design goals of Epicast was to minimize human time and effort required , and so we did not pursue the Delphi method . However , it would be of value to compare the Epicast and Delphi methods to learn the relative advantages and disadvantages of each method in terms of both human effort and prediction accuracy . It was our hope that the number of participants would grow organically , for example through word of mouth and social media . Instead , we found it difficult to recruit new participants and to maintain participation throughout the flu season . The failure to achieve a true “crowd” is most likely due to the tedium of the task , and we are working on ways to both reduce this tedium and to make the task more gratifying for participants . While we strove to design the user interface in a way that minimizes the level of effort required to input predictions , there is always room for further improvement . One option we considered , but did not implement because of the small number of participants , is to reduce workload by asking participants to provide a prediction only for a randomized subset of regions . Another option we considered , but did not completely implement due to time constraints , was gamification . This was partially implemented in the form of leader boards , but it would be difficult to provide a more immediate reward because of the inherent delay between prediction and revelation of true outcomes . There are several additional ways in which the Epicast method could be improved . First , there is an important relationship between a prediction , and the level of confidence in that prediction , that we were unable to capture . We asked participants to give us their best point predictions , but there was no way for them to communicate with us their level of confidence in those and other predictions—a forecast . We made the implicit assumption that disagreement among user predictions implies lack of confidence , which is probably true to some extent . The inverse however—that uniformity in predictions implies high confidence—is clearly untrue . Consider as an example the case where everyone believes that next week’s wILI has a 60% chance of staying the same as this week’s wILI , resulting in all point predictions strongly concentrated on the same wILI , and the distributional spread being very narrow , in contrast with the participants’ beliefs . It would be ideal to collect from each user a more informative measure of their confidence , but this would undoubtedly complicate the user interface and degrade the overall experience ( which we were averse to ) . Another improvement to consider is a weighted combination of predictions whereby participants who have historically had more accurate predictions are given more weight in the aggregation process . This is similar in spirit to weighting user recommendations and rankings , which has been shown to increase accuracy in those settings [39 , 40] . In the case of Epicast , there is limited evidence suggesting that some participants are overall more ( or less ) accurate than other participants . One example of this is in Fig 5B where one participant has significantly higher Win Rate than Epicast and several other users . On the other hand , it is not clear whether the variance of prediction error is sufficiently small to learn which users are the most accurate in a reasonable amount of time—before the epidemic peak , for example . In other words , differences in accuracy may be exploitable , but only if precision is sufficiently high . If this is the case , then an adaptive weighting scheme could benefit the overall forecast . However , there is a critical obstacle that hinders the practical implementation of such a scheme: because of backfill , the final measure of accuracy is not known for many months . Despite this , we propose , implement , and analyze one such scheme in S1 Text—the result is a small and statistically insignificant increase in accuracy . A natural evolution of systems such as those for epidemiological forecasting is the combination of human and statistical ( machine ) methods [41 , 42] . The first question in such a project is whether human predictions should be given as input to statistical methods or whether the output of the statistical methods should be shown to humans for more informed predictions . In theory both directions are viable , and there are intuitive reasons for each . In support of the latter , people are naturally inclined to trust forecasts made by humans ( or to distrust forecasts made by machines ) , a phenomenon known as algorithm aversion [43] . Supporting the former , on the other hand , is the observation that in many settings and in a variety of tasks , objective machine prediction is often superior to subjective human prediction [44 , 45] . We have begun to explore both directions; currently , we show a subset of Epicast participants a confidence band derived from a separate statistical method , and we are developing a retrospective analysis wherein we compare performance of various statistical methods with and without a supplemental input of human prediction as an independent data source . In the meantime , we plan to continue to host Epicast and collect predictions for the current flu season , and we end this section with an open invitation to participate [34] . For years , both humans and machines have been employed to tackle difficult prediction problems , and the biases involved and the relative advantage of data-driven approaches are at least well documented [43 , 44] , if not well understood . We do not make the claim that human judgment is intrinsically more valuable or more capable than machines when making epidemiological forecasts , but we do posit that there is value in understanding the strengths in each approach and suspect that both can be combined to create a forecasting framework superior to either approach alone . | Despite advanced and widely accessible health care , a large number of annual deaths in the United States are attributable to infectious diseases like influenza . Many of these cases could be easily prevented if sufficiently advanced warning was available . This is the main goal of epidemiological forecasting , a relatively new field that attempts to predict when and where disease outbreaks will occur . In response to growing interest in this endeavor , many forecasting frameworks have been developed for a variety of diseases . We ask whether an approach based on collective human judgment can be used to produce reasonable forecasts and how such forecasts compare with forecasts produced by purely data-driven systems . To answer this , we collected simple predictions in real-time from a set of expert and non-expert volunteers during the 2014–2015 and 2015–2016 U . S . flu seasons and during the 2014–2015 chikungunya invasion of Central America , and we report several measures of accuracy based on these predictions . By comparing these predictions with published forecasts of data-driven methods , we build an intuition for the difficulty of the task and learn that there is real value in collective human judgment . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"epidemiological",
"statistics",
"infectious",
"diseases",
"medicine",
"and",
"health",
"sciences",
"mathematics",
"forecasting",
"statistics",
"(mathematics)",
"chikungunya",
"infection",
"epidemiology",
"mathematical",
"and",
"statistical",
"techniques",
"influenza",
"negle... | 2017 | A human judgment approach to epidemiological forecasting |
A major goal in evolutionary biology is to understand how adaptive evolution has influenced natural variation , but identifying loci subject to positive selection has been a challenge . Here we present the adaptive loss of a pair of paralogous genes in specific Saccharomyces cerevisiae subpopulations . We mapped natural variation in freeze-thaw tolerance to two water transporters , AQY1 and AQY2 , previously implicated in freeze-thaw survival . However , whereas freeze-thaw–tolerant strains harbor functional aquaporin genes , the set of sensitive strains lost aquaporin function at least 6 independent times . Several genomic signatures at AQY1 and/or AQY2 reveal low variation surrounding these loci within strains of the same haplotype , but high variation between strain groups . This is consistent with recent adaptive loss of aquaporins in subgroups of strains , leading to incipient balancing selection . We show that , although aquaporins are critical for surviving freeze-thaw stress , loss of both genes provides a major fitness advantage on high-sugar substrates common to many strains' natural niche . Strikingly , strains with non-functional alleles have also lost the ancestral requirement for aquaporins during spore formation . Thus , the antagonistic effect of aquaporin function—providing an advantage in freeze-thaw tolerance but a fitness defect for growth in high-sugar environments—contributes to the maintenance of both functional and nonfunctional alleles in S . cerevisiae . This work also shows that gene loss through multiple missense and nonsense mutations , hallmarks of pseudogenization presumed to emerge after loss of constraint , can arise through positive selection .
Biologists have long sought to understand the process of natural selection and the signatures left behind in extant species . Finding evidence of adaptive evolution has been a holy grail for evolutionary biologists , because it can provide insights into how and why organisms evolve . However , examples of adaptive selection from which to glean insights remain relatively scarce [1] . The recent explosion in the number of genomes available for different organisms provides an exciting opportunity to identify loci with unusual patterns of variation indicative of selection ( for example [2]–[7] ) . However , even for loci with strong signatures of selection , the affected phenotypes are often a complete mystery . In contrast , mapping studies link quantitative trait variation to genomic loci that can then be interrogated for evidence of selection . The challenge in most organisms is identifying responsible SNPs within candidate regions , which are often megabases long and contain hundreds of functional elements , hindering further study [8] . Here , we used the power of yeast genetics and genomics to uncover a unique example of adaptive gene loss , involving multiple paralogous genes and several sequential evolutionary events . We previously surveyed phenotypic variation in Saccharomycetes collected from diverse environments and found that relatively few of those strains ( 12% ) could survive freeze-thaw ( FT ) stress [9] . Many tolerant strains were isolated from oak soil in the Northeastern United States , whereas sensitive strains were typically isolated from warm environments , often from fruit or fermentations . This suggested that FT tolerance has been selected for in strains from cold climates but lost in other isolates . Several genes have been linked to freeze-thaw tolerance in yeast and other organisms , including water transporters . The paralogous yeast aquaporins ( AQYs ) AQY1 and AQY2 were implicated in FT stress by the baking industry , which found that AQY over-expression increases yeast viability in frozen bread dough [10] . Rapid export of water through AQYs is thought to increase FT survival by preventing intracellular shearing due to water crystallization [10] , [11] . The paralogs may have arisen in the whole-genome duplication ( WGD ) event in the Saccharomyces lineage [12] , since all post-WGD species all have two aquaporins whereas most pre-WGD species have a single ortholog ( Dana Wohlbach and A . P . G . , unpublished ) . It has been observed that laboratory and industrial strains as well as several vineyard isolates harbor non-functional alleles of AQY2 , while several strains harbor a non-functional version of AQY1 [13]–[16] . However , without population-level analysis or knowledge of potential ecological driving forces , it is difficult to distinguish selection at these loci from neutral gene loss in the progenitor of the related strains . Here , we provide the first evidence of adaptive loss of AQY paralogs in natural populations of S . cerevisiae , leading to incipient balancing selection via spatial variation in selective pressures .
We applied several tests to assess if loss of AQY function may have been selected for in some strains . Under the neutral model , the rate of polymorphism within strains should be similar to the rate of divergence across species . Instead , both AQY1 and AQY2 show an excess of replacement polymorphism , assessed by the McDonald-Kreitman test [22] that compares non-synonymous ( A ) to synonymous ( S ) codon changes ( Table S3 ) . AQY1 showed an A/S ratio of polymorphism ( 5/4 = 1 . 25 ) that was significantly higher than that of divergence ( 11/49 = 0 . 22 , p = 0 . 026 , Fisher's exact test ) . AQY2 also showed an excess of polymorphic sites ( A/S of 8/20 = 0 . 4 ) compared to divergent sites ( 3/40 = 0 . 075 , p = 0 . 019 ) , as well as an excess of deletions ( 3/20 versus 0/40 , p = 0 . 045 ) . AQY2 ( but not AQY1 ) also deviated from neutral evolution at synonymous sites , showing an excess of SNPs compared to 8 intergenic sequences ( p = 0 . 028 , multi-locus HKA test [23] , Table S4 and Table S5 ) . For the most part , the tests were not significant if subgroups of strains , defined by AQY haplotypes in Figure 2 ( see Table S1 for details ) , were considered separately ( Table S3 and Table S5 ) . This result indicates that much of the variation is between strain groups . Excess polymorphism can result from relaxed constraint in the species , or if local adaptation is driving divergence between populations [24] , [25] . To distinguish between these models , we used non-imputed genome sequence data of Liti et al . [21] to characterize sequence variation flanking the AQY genes . We applied several empirical tests , which can handle the missing data in the low-coverage genomic sequences and are less subject to the unusual features of S . cerevisiae populations ( including extensive population structure , unknown population dynamics , ambiguous balance between clonal vs . sexual reproduction , and human-associated migration [21] , [26]–[29] ) that can confound standard tests [2] , [30] . To monitor the variation surrounding AQY2 , we subdivided 21 strains with data at AQY2 into strains harboring the Asian G25 deletion , the 11 bp deletion , or the full-length AQY2 ( clonal Malaysian strains were not considered , see Methods ) . We calculated the average pairwise nucleotide differences surrounding AQY loci within and between groups , and then compared this variation to other regions across the genome . We tested for several signatures: a recent selective sweep is predicted to reduce variation flanking the selected allele in the affected population , while balancing selection can increase variation between strain groups [25] . Since much of the genome may be evolving neutrally , loci with extreme values display the strongest evidence for non-neutral evolution . For strains harboring the Asian G25 allele of AQY2 , we saw a high correlation in between-group variation and within-group variation across much of the genome , including the right arm of chromosome 12 ( Figure 3A , right side ) . However , a 50 kb stretch on the left arm of chromosome 12 showed below-average variation within the strains ( 0 . 76th percentile compared to other similarly sized regions genome-wide , see Methods ) but high variation between groups . There was a sharp break in this pattern at ∼72 kb , which may represent the breakpoint of a selective sweep . To further explore this , we calculated the difference in between-group variation minus within-group variation , then calculated the area under contiguous peaks in the difference curve for comparison ( see Figure S2 and Methods ) . This procedure identified a 5 . 6 kb region spanning the 870 bp AQY2 ORF that ranked in the top 1 . 2 percentile of 4 , 600 regions genome-wide with skewed between-group versus within-group variation ( Figure 3A ) . Strains harboring the 11-bp deletion displayed a 4 , 300 bp region encompassing AQY2 with a significant skew in the between- versus within-group variation ( 1 . 8th percentile of 3 , 238 regions genome-wide , Figure 3B ) and below-average within-group variation ( 6th percentile genome-wide ) . Strains harboring the full-length AQY2 showed a smaller peak of 1 , 800 bp with high between-group variation ( 6 . 3rd percentile , Figure 3C ) , but average within-group variation ( >50th percentile ) . These results show that strains harboring either deletion have low variation within those strain groups , and that the high variation at AQY2 distinguishes the three groups from one another . Indeed , a genome-wide plot of FST [2] , which measures the population differentiation based on these three groupings , identified a clear peak of 6 . 4 kb over AQY2 with above-average FST , ranking among the top 1 . 5th percentile genome-wide ( Figure 3D ) . A confounding feature is the extensive population structure within S . cerevisiae [21] , [29] , which can mimic some signatures of selection . Several controls indicate that the observed patterns are unlikely due to demographics . First , these regions were among the most extreme across the genome , which is not expected if population structure is the underlying cause . However , many S . cerevisiae strains have mosaic genomes , for which large regions have distinct lineages [21] , [29] . To control for this , we performed a partitioning sampling: strains were partitioned at each of 1 , 370 randomly chosen SNPs across the genome . The difference in between-group minus within-group variation was scored surrounding the partitioning SNP and compared to the difference profile when strains were partitioned based on AQY2 allele ( see Methods ) . The regions observed for the Asian G25 or 11-bp deletion classes remained among the most extreme ( 4 . 7th and 5 . 4th percentile , respectively ) . Thus , the profiles we observe in Figure 3A and 3B are uncommonly found at random SNPs , most of which likely reflect neutral variation . In contrast , the skew in variation found in strains with full-length AQY2 was not significant by this assessment ( 26th percentile ) . We conclude that the observed skew in polymorphism observed in strains with the Asian G25 deletion and the 11-bp deletion in AQY2 resulted from two separate partial selective sweeps that reduced variation within each group . The high variation distinguishing strain groups is a signature of balancing selection , which may be maintaining both functional and non-functional AQY2 alleles in the population . Indeed , we observed a positive Tajima's D at AQY2 , assessed on a smaller set of high-quality sequences ( D = 0 . 851 , p<0 . 05 , Figure S3 ) , indicating an excess of intermediate-frequency polymorphism that is consistent with balancing selection [24] . The patterns at AQY1 were less clear . Strains harboring the aqy1 V121M allele or the A881 deletion showed reduced variation within each group and high variation between groups at the AQY1 region ( Figure S4 ) . Although these were highly significant compared to other loci across the genome ( 0 . 77th and 1 . 23rd percentile , respectively ) , they were not significant compared to random-SNP partitioning described above ( 16th and 48th percentile , respectively ) . Thus , the slight skew in between-group versus within-group variation at AQY1 could be due to demographic factors , incorrect strain groupings , or older or weaker selective sweep ( s ) that have since recovered variation through recombination or mutation . The above results strongly suggest selective pressure to lose AQY function in some strains , perhaps driven by environmental factors . We previously reported an anti-correlation between FT survival and osmo tolerance across a wide range of S . cerevisiae strains ( R = −0 . 35 , p = 0 . 006 ) [9] . Furthermore , a lab strain with functional AQYs was shown to be sensitive to hypo- and hyper-osmotic cycling , but not to consistently high osmolarity [13] , [14] . Instead , we found that loss of both AQY genes provides a major growth advantage in high osmolar conditions found in nature ( Figure 4 ) . A YPS163 mutant lacking both AQYs displayed ∼1 . 7X greater survival in 1 . 5 M sorbitol , whereas introducing a functional AQY into the S288c-derived lab strain decreased survival 2–3X . Furthermore , sorbitol tolerance was anti-correlated to both freeze-thaw tolerance ( R = −0 . 38 ) and the number of functional aquaporins ( R = −0 . 31 ) in these strains ( Table S1 ) . The sugar concentration used here is comparable to that found in the fruit substrates of many wild strains [31] . Thus , AQY function presents a substantial fitness defect in conditions relevant in nature , likely due to passive water loss triggered by the high osmolarity of sugary substrates . In the course of these experiments , we also discovered that YPS163 lacking either aquaporin had a major defect in spore formation during meiosis ( Figure 5 ) . Although AQY1 had been previously implicated in a late step in spore maturation [32] , our phenotype is distinct in that it affects spore production . Whereas >70% of the parental YPS163 formed full tetrads within 2 days , only 18–24% of the double or single mutants produced full tetrads . After 9 days , the mutant produced more spores but was still defective compared to the parental strain ( <60% full tetrads compared to ∼85% , Figure S5 ) . The AQY requirement is ancestral , since an S . paradoxus aqy1Δ mutant displayed an identical defect ( Figure 5A ) . In contrast , strains without functional AQY genes produce full tetrads ( albeit with lower efficiency than YPS163 [33] ) , consistent with a previous report showing AQY1 is not required for sporulation in vineyard strains [34] . More importantly , introducing the functional YPS163 allele of AQY1 into strains with different combinations of non-functional AQY alleles ( including strains M22 , K1 , SK1 , and S288c ) did not significantly improve spore production ( Figure 5B ) . Thus , strains lacking AQY function have also lost the ancestral need for AQY during spore production .
This work provides the first clear evidence for adaptive loss of AQY function in subgroups of wild S . cerevisiae isolates . The excess polymorphism at AQY genes ( McDonald-Kreitman and HKA tests ) , high between-group variation surrounding AQY2 that distinguishes strain groups ( group variation and FST plots , Figure 3 ) , and skew in the frequency spectrum toward intermediate-frequency AQY2 alleles ( Tajima's D ) are all consistent with non-neutral evolution . Furthermore , AQY paralogs have been lost at least 6 independent times , through 2 partial selective sweeps at AQY2 and possibly others at AQY1 . The high variation between strain groups , and the non-random retention or loss of both paralogs in diverse strains , is consistent with the establishment of balanced polymorphism . We propose that the antagonistic pleiotropy of aquaporin function , coupled with spatial differences in selective pressures , provide pressure to maintain both functional or both non-functional alleles in distinct subpopulations of S . cerevisiae . FT tolerance may be crucial for survival in cold climates , and along with sporulation efficiency may impart strong pressure to retain AQY genes in strains from wintry niches . Indeed , the ratio of non-synonymous to synonymous differences in YPS163 compared to S . paradoxus is 2 - 6X lower for AQYs compared to the genomic average ( Ka/Ks of 0 . 018 and 0 . 059 for AQY2 and AQY1 , respectively , versus 0 . 1 across all genes [35] ) . This is consistent with purifying selection acting to remove deleterious codon changes . The oak strains likely represent the ancestral state , since close relatives S . paradoxus and S . mikatae are also recovered from tree exudates and soil [17] , [36] , display high FT tolerance [9] , and require aquaporins for sporulation ( Figure 5 and data not shown ) . Interestingly , Northeastern-US oak strains display unique phenotypes suggestive of other evolutionary forces as well . AQY2 is expressed on average 14-fold higher in YPS163 compared to 17 other surveyed strains [9] , [37]; those levels are doubled in YPS1009 , which underwent a duplication of the entire chromosome 12 [9] . Although further studies will be needed , that over-expression of AQY2 is known to enhance FT tolerance in industrial strains [10] hints that the elevated expression may have been selected for , further underscoring the importance of AQY function in these strains . In contrast , many other strains exist in warm environments that never experience freezing . Most of these were sampled from fruit substrates and distillations , which typically consist of ∼25% sugars [31] , in contrast to oak soil [38] , [39] from which many cold-climate strains have been isolated . Thus , the significant advantage in osmo-tolerance due to AQY loss likely played a major role in selection at this locus . It is unclear which came first–loss of aquaporin requirement during sporulation , or loss of aquaporin function that drove subsequent loss of the sporulation role . Loss of sporulation dependency on aquaporins , coupled with migration to warmer climates , would have relaxed constraint on the genes and facilitated their adaptive loss when cells moved to high-sugar substrates . This model could have involved a single loss of sporulation requirement followed by multiple independent losses of aquaporin function . Alternatively , strong selective pressure to lose aquaporins could have forced multiple independent losses of the sporulation requirement , just as it lead to multiple independent losses of aquaporin function . S . cerevisiae strains are thought to have migrated globally through human association , after two domestication events produced sake/distillation strains and vineyard/wine-making lines ∼10 , 000 years ago [20] , [21] , [26] , [27] , [29] , [40] . Human-facilitated migration may have significantly increased exposure of S . cerevisiae to diverse climates , which may have imposed new selective pressures when strains encountered new niches . Increased migration may also have facilitated outcrossing of domesticated strains with natural strains , allowing several of these alleles to spread through natural populations . It is important to note that Malaysian strains , not previously associated with domestication events , show unique non-functional AQY alleles , revealing that loss of aquaporins is not strictly driven by domestication . The selective sweeps of nonfunctional aquaporin alleles appear to have been recent events , given the strength of the signal at AQY2 , and may reflect an ongoing process . A remaining question is the fate of the emerged balance in polymorphism . Given sufficient migration of strains between the two niches and unequal fitness costs of the opposing haplotypes ( i . e . two functional or two nonfunctional AQY alleles ) , one haplotype may eventually win out to fixation , eliminating the balanced alleles . On the other hand , long-term balancing selection could result if equivalent selective constraints are maintained in each respective niche . In the extreme case , strongly opposing selective forces could restrict yeast migration between environments to promote ecological speciation [41] . Little is known about S . cerevisiae migration between tree soil and fruits , although oak-soil strains are genetically well separated from vineyard/fermentation isolates [21] , [29] , [40] , [42] . The antagonistic forces driving aquaporin loss at the cost of freeze-thaw sensitivity may be one factor that has limited gene flow between these niches .
Strains and plasmid constructs are described in Table S6 . Two S . cerevisiae strains ( DY8 and DY9 ) were isolated from oak-tree soil from Maribel , Wisconsin using the method of [17] , and typed by a mating/sporulation assay with a tester S . cerevisiae strain ( Dan Kvitek and APG , unpublished ) . Gene deletions were created by homologous recombination , replacing AQY1 and/or AQY2 with KanMX3 or NatMX3 drug-resistance cassettes , respectively . Homothalic wild strains capable of mating-type switching ( including YPS163 , M22 , and S . paradoxus ) were sporulated and dissected , and drug-marked colonies were selected as homozygous diploids . In all cases , homozygous gene deletions were confirmed by diagnostic PCR . The region corresponding to the 870 bp full-length AQY2 ORF plus 971 bp upstream and 393 bp downstream sequence was cloned from YPS163 or BY4741 , by homologous recombination replacing a GFP-ADH1-terminator cassette in plasmid BA1924 ( provided by P . Kainth and B . Andrews ) , which is derived from pRS315-based CEN plasmid BA1926 [43] but with the NatMX3 cassette replacing the LEU2 marker . The region corresponding to the 918 bp full-length AQY1 ORF with the flanking 947 bp upstream and 747 bp downstream was similarly cloned . All clones were verified by sequencing . To assess functionality of the different alleles , AQY1 ORFs representing M22 , BY4741 , or Y55 alleles ( identical to the YPS163 allele but harboring the A881 deletion ) or the Malaysian AQY2 allele ( identical to YPS163 except for the G528 deletion ) were cloned between the native upstream and downstream AQY1 sequence from YPS163 . This was done to prevent confounding influences on expression through variation in the flanking regulatory regions . Plasmids were introduced into YPS163 aqy1Δ , BY4741 , or other naturally AQY-minus strains , and complementation of spore production in the YPS163 aqy1Δ mutant or of FT tolerance in BY4741 was scored ( Table S2 ) . Yeast strains were grown at 30°C in YPD medium to an optical density at 600 nm ( OD600 ) of 0 . 3–0 . 4 in 24-well plates . To measure freeze-thaw tolerance , 200 µl of cells was transferred to 1 . 5 ml tubes and frozen in a dry ice/ethanol bath ( <−50C ) for two hours or on ice as control . Viability was measured by scoring serial dilutions spotted onto agar plates as previously described [9] , or using Live/Dead stain ( Invitrogen , Carlsebad , CA ) read on a Guava EasyCyte Plus flow cytometer ( Millipore , Billerica , MA ) . Scores in Figure 2 correspond to high ( >80% of YPS163 viability , three pluses ) , medium ( 50–80% viability , two pluses ) , low ( <50% viability , one plus ) , or no detectible ( minus sign ) FT tolerance . Osmotic tolerance was measured by plating cells onto agar plates containing 1 . 5 M sorbitol . Percent viability was scored as the number of colony-forming units compared to the no-stress control plate . Cells were grown in YPD rich medium to OD600 nm of 1 . 0 , harvested by centrifugation , resuspended in 1% potassium acetate , and incubated at 25°C for 2 or 9 days . Cells were harvested , diluted and the number of spores per tetrad was counted on a hemocytometer . QTL mapping strains and analysis were as previously described [18] , using the Haley-Knott algorithm implemented in R-QTL [44] . Two additional peaks in Figure 1B ( left arm of Chromosome 2 and right arm of Chromosome 8 ) were not significant when Chromosome 12 and 16 QTL were held as fixed terms , suggesting the additional peaks may be false positives . Sequencing using Big-Dye ( Applied Biosystems , Carlsbad , CA ) scored at least 3 reads ( including forward and reverse ) per basepair from 2 independent genomic preparations ( GenBank accessions GQ848552-74 and GQ870433-54 ) . The vast majority of sequence data represented homozygous sites . The few base pairs with evidence of heterozygocity were represented by one of the alleles , randomly chosen . MK-tests , Tajima's D , and Ka/Ks were calculated in DNASP 5 . 0 [45] and ML-HKA was done as in [23] using sequence data from [9] , [20] and here . Genome-wide sequence analysis was performed using unimputed , aligned data from [21] with quality scores > = 40 ( generously provided by Alan Moses ) , treating all gaps as missing data to avoid alignment errors . Strains were grouped according to AQY2 or AQY1 alleles ( see Table S1 for details ) , and the average number of pairwise SNPs was calculated every 1000 bp with a 100 bp step size , for all pairs of strains within each group and for all strains in a given group compared to each strain outside that group . Within-group variation was scored for all 50 kb regions across the genome with a step size of 20 kb , and for all 5 kb regions with step size of 2 kb . These regions were compared to the 50 kb region highlighted in the text ( position 22 , 000–72 , 000 in Figure 3 ) for strains with the G25 AQY2 allele and to a 5 kb region centered on AQY2 for other strain groups . All regions were ranked based on the average pairwise within-group variation to calculate the percentile rank of regions in question . To monitor the skew variation within and between groups , a difference profile of between-group variation minus within-group variation ( calculated as described above ) was taken across the genome , and all contiguous regions ( “peaks” ) where the difference value was >1 . 5X the chromosome-wide average were identified ( see Figure S2 and Table S7 ) . The area under each peak was estimated by the trapezoidal method , and compared to the area under the peaks in Figure 3A and 3B spanning AQY2 . For the partitioning sampling , we scanned for SNPs with at least 3 strains harboring the minor allele , every 10 , 000 bp across each of the 16 yeast chromosomes . Strains were partitioned based on that SNP , then the between-group and within-group variation was measured for 20 , 000 bp centered on the partitioning SNP , based on the average-pairwise differences every 1 , 000 bp with a 100 bp step size as above . A profile of the between-group variation minus the within-group variation was taken in every window . For each partitioning SNP , a peak in the difference profile was identify by walking outward until the difference value was <3 . 54 , the cutoff used the genomic scan shown in Figure 3B . The area under the curve was calculated as above and compared to that measured at AQY2 by an identical procedure except that strains were partitioned by Asian G25 allele vs . all others strains or by 11-bp deletion vs . all other strains . Very similar percentile rankings were obtained if we scored 5 kb windows centered on each SNP ( data not shown ) . | Local adaptation is thought to be a driving force in population differentiation and the formation of new species . Yet , there are few examples of ecologically relevant phenotypes that have been mapped to individual genes , making it difficult to know what drives the evolution of such genes and contributes to the molecular mechanisms underlying divergence . Here , we provide a unique case of local adaptation through multi-gene loss . We mapped the genetic basis for natural variation in yeast freeze-thaw tolerance to two water transporters , AQY1 and AQY2 . Although tolerant strains harbor functional alleles of both genes , the set of sensitive strains lost aquaporins at least 6 independent times , through missense mutations and frame-shifting deletions . Genome-wide scans reveal several signatures of recent , partial selective sweeps at the aquaporin loci , indicating positive selection for gene loss . This was likely driven by a major fitness advantage of aquaporin loss when cells grow in high sugar concentrations common to many strains' niche . Surprisingly , strains that lost aquaporins also lost the ancestral requirement for these genes during sexual reproduction . This work provides a compelling example of how gene loss through nonsense mutations , a hallmark of pseudogenization , is caused not by loss of constraint but by positive selection . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"evolutionary",
"biology/microbial",
"evolution",
"and",
"genomics",
"evolutionary",
"biology/genomics",
"evolutionary",
"biology/evolutionary",
"ecology",
"evolutionary",
"biology/evolutionary",
"and",
"comparative",
"genetics"
] | 2010 | Incipient Balancing Selection through Adaptive Loss of Aquaporins in Natural Saccharomyces cerevisiae Populations |
The Calsequestrin ( Csq ) transgenic mouse model of cardiomyopathy exhibits wide variation in phenotypic progression dependent on genetic background . Seven heart failure modifier ( Hrtfm ) loci modify disease progression and outcome . Here we report Tnni3k ( cardiac Troponin I-interacting kinase ) as the gene underlying Hrtfm2 . Strains with the more susceptible phenotype exhibit high transcript levels while less susceptible strains show dramatically reduced transcript levels . This decrease is caused by an intronic SNP in low-transcript strains that activates a cryptic splice site leading to a frameshifted transcript , followed by nonsense-mediated decay of message and an absence of detectable protein . A transgenic animal overexpressing human TNNI3K alone exhibits no cardiac phenotype . However , TNNI3K/Csq double transgenics display severely impaired systolic function and reduced survival , indicating that TNNI3K expression modifies disease progression . TNNI3K expression also accelerates disease progression in a pressure-overload model of heart failure . These combined data demonstrate that Tnni3k plays a critical role in the modulation of different forms of heart disease , and this protein may provide a novel target for therapeutic intervention .
Heart failure is the common final outcome for many forms of acute and chronic heart disease . The prognosis of heart failure is highly variable between patients , and the differences in phenotypic expression ( symptoms , disease progression and course , and final outcome ) create difficulties in the construction of predictive models [1] . Previous research has suggested that genetic factors can considerably modify the progression and outcome of heart failure [2] . However , these factors are difficult to identify directly in the human population because of wide genetic variability , uncontrollable environmental factors , and the intervention of medical therapy . We have employed a disease-sensitized mouse model to map genetic factors that modify the progression and outcome of heart disease . In the Calsequestrin transgenic mouse , cardiac-specific overexpression of the calcium binding protein Calsequestrin ( CSQ ) leads to dilated cardiomyopathy [3] . This murine model recapitulates many of the hallmarks of human dilated cardiomyopathy including cardiac enlargement , depressed contractile function , abnormal beta-adrenergic receptor signaling and premature death [4] . Although all mice that overexpress CSQ develop dilated cardiomyopathy , disease progression and outcome varies greatly depending on the genetic background ( inbred mouse strain ) harboring the Csq transgene ( Csqtg ) [5] , [6] . These differences are due to modifier genes whose multiple alleles differentially modulate the phenotype . We have exploited these strain-specific phenotypic differences to map seven different loci ( Heart failure modifier or Hrtfm ) that modify the progression of cardiac dysfunction and the outcome of heart failure [5]–[7] . Hrtfm2 , mapping to mouse chromosome 3 , was identified in a cross between inbred strain DBA/2J ( DBA ) , which harbors the original Csqtg , and C57BL/6 ( B6 ) . In this cross , the B6 allele at Hrtfm2 imparted a dominant , disease-accelerating effect on both cardiac dysfunction ( as measured by echocardiography ) and reduced survival [5] . Subsequently , Hrtfm2 was re-identified in a second cross between DBA/Csqtg animals and inbred strain AKR/J ( AKR ) [7] . In this cross , the AKR allele of Hrtfm2 also imparted a dominant , disease-accelerating effect on cardiac dysfunction and survival . The phenotypic effects of Hrtfm2 were robust , accounting for 30% of the genetic variability for survival and 22% for cardiac dysfunction . Capitalizing on the ancestral nature of the Hrtfm2 allele ( ie , mapping within a murine haplotype block that has been retained throughout evolution and now found in multiple inbred strains ) , we employed haplotype-sharing analysis to effectively narrow the candidate interval from 16 . 5 Mb to 2 Mb , a region containing only 7 known genes [7] . We had previously suggested the Tnni3k gene as an attractive candidate based on its location within the shared haplotype interval and its biological significance as a cardiac-specific kinase that reportedly interacts with cardiac Troponin I ( cTnI ) [8] . Here we report the molecular characterization of allelic variation at the murine Tnni3k gene , and present in vivo functional evidence showing that Tnni3k underlies the heart failure modifier locus , Hrtfm2 .
As part of an effort to identify candidate genes for the Hrtfm loci , we performed microarray expression analysis of normal heart tissue from the inbred strains used in our mapping crosses to identify genes showing innate differences in transcript levels . Of the genes mapping within the Hrtfm2 linkage peak , only one gene exhibited significantly different transcript levels between the less susceptible strain DBA and the more susceptible strains B6 and AKR . Transcript levels of Tnni3k were elevated 12-fold in B6 and AKR compared with DBA , whereas levels of other transcripts mapping within the interval were not significantly different ( Figure 1A ) . These expression differences were validated by more-sensitive qRT-PCR analysis , where Tnni3k message levels were found to be 25-fold higher in B6 and AKR strains than those in DBA ( Figure 1B ) . In parallel , we genetically isolated the Hrtfm2 locus by creating a congenic line that carried AKR alleles across Hrtfm2 ( an approximately 20 Mb region between rs13477425 and rs13477504 ) and DBA alleles throughout the rest of the genome . Quantitative RT-PCR showed that Tnni3k transcript levels in hearts from DBA . AKR-Hrtfm2 congenic mice were comparable to levels observed in B6 and AKR ( AKR being the source of the Hrtfm2 locus ) , and not that seen in DBA ( the genomic background ) , suggesting that the Tnni3k expression differences were driven by cis-acting sequence elements within the Hrtfm2 locus , rather than trans-acting factors mapping elsewhere in the genome . We analyzed heart tissue prepared from six inbred mouse strains to determine if these differences in levels of Tnni3k transcript would be observed at the protein level . We chose three additional strains that shared either the DBA or B6 haplotype at Tnni3k ( Table S1 ) . As predicted by the transcript levels , robust levels of TNNI3K protein were detected in B6 , AKR , 129×1/SvJ ( 129 ) and the DBA . AKR-Hrtfm2 congenic , which share the B6 haplotype . Surprisingly , no apparent protein was detected for DBA , C3H/HeJ and BALB/cByJ ( BALB/c ) strains , which share the DBA haplotype ( Figure 1C ) . Therefore , within the limits of detection of the antiserum , TNNI3K protein was apparently absent from hearts of strains sharing the DBA haplotype at the Tnni3k locus . The latter strains effectively represent Tnni3k null or extreme hypomorphic genotypes with no apparent effect on development or survival , and with no obvious pathological consequence . The Tnni3k coding region of the mapping strains differed by a single , relatively conservative , non-synonymous coding SNP ( rs30712233 , T659I ) . By sequencing Tnni3k cDNA from the strains , we noted another , more consequential , strain-specific sequence difference . All strains sharing the B6 haplotype showed a single major transcript identical to the published cDNA . By contrast , all strains sharing the DBA haplotype exhibited a mixture of two transcripts consisting of the published transcript and a second transcript containing a 4-nucleotide insertion between exons 19 and 20 ( Figure 2A ) . This insertion was not present in the genomic DNA , but instead represented the addition of the first 4 nucleotides from intron 19 into the Tnni3k transcript . The insertion created a frameshift resulting in a premature termination codon immediately downstream ( Figure 2B ) . We determined that the frameshifted transcript accounted for approximately 70% of the message in DBA heart mRNA , but was not present in B6 or AKR ( Figure 2C ) . This transcript was not found in any of the EST databases for mouse or any other species , suggesting that it represented an aberrant message created by defective splicing , possibly caused by the use of a second ‘gt’ splice donor site 4 nucleotides downstream of the normal donor site . The genomic region surrounding exons 19 and 20 harbors over 50 SNPs . Although in principle any of these could have caused the aberrant splicing , we focused on the SNP nearest to the splice donor junction . B6 and related strains ( AKR , 129 , MRL ) possess an ‘a’ nucleotide at rs49812611 , whereas DBA and related strains ( A/J , C3H/HeJ ( C3H ) , BALB/c ) possess a ‘g’ . This SNP lies at the +9 position for the normal splice site , but importantly , this SNP lies at the +5 position with reference to the aberrant splice site . Thus , DBA and related strains harbor the consensus ‘g’ nucleotide at the +5 position for the aberrant site . During mRNA processing , the ‘g’ at the +5 splice donor position pairs with a ‘c’ in the U1 or U6 snRNA , resulting in a preference for ‘g’ at this position . Weight matrix scores for splice donor strength [9] , [10] for each possible splice donor site confirmed that the second ( aberrant ) splice site was the strongest splice site in the region only when the ‘g’ nucleotide is present at rs49812611 ( Figure 2D ) . We tested the hypothesis that rs49812611 is the cause of aberrant splicing using an in vitro splicing system . Genomic DNA spanning exons 18 through 20 from both B6 and DBA were sub-cloned and transfected into 293T cells ( Figure 3A ) . These in vitro constructs recapitulated the splicing pattern observed in vivo , confirming that the splicing defect was caused by cis-acting sequences residing within the cloned 4 kb genomic fragment ( Figure 3B ) . Site-directed mutagenesis was used to investigate the role of rs49812611 in aberrant splicing . A single change at this SNP completely reversed the splicing pattern . DBA genomic DNA altered to carry the ‘a’ allele at rs49812611 generated no aberrant splice product , whereas the B6 DNA carrying the ‘g’ allele exhibited the aberrant product ( Figure 3B ) . These results showed that rs49812611 was responsible for the presence or absence of the aberrantly spliced message , although the extent of aberrant splicing may be modulated by other flanking sequence variation . Since Tnni3k was originally identified as a positional candidate for Hrtfm2 due to differences in transcript levels between the mapping strains , we hypothesized that nonsense-mediated decay ( NMD ) was responsible for the drastically reduced levels of the frameshifted message seen in DBA . We investigated this in the mouse cardiomyocyte cell line , HL-1 [11] , which shares the DBA haplotype at Tnni3k . We first confirmed that HL-1 cells expressed both aberrant and normal Tnni3k at levels comparable to wild-type DBA hearts , with the majority of the message being the aberrant variant that includes the 4-nucleotide insertion . HL-1 cardiomyocyte cells were then treated with either cycloheximide and emetine , two drugs commonly used to block NMD [12] . Treatment with either drug increased the level of aberrantly spliced transcript relative to the normally spliced message ( Figure 4A ) . As predicted , these treatments increased levels of total Tnni3k mRNA 16-fold ( Figure 4B ) , supporting a major role for NMD in the observed differences in transcript levels between strains . Although these experiments determined the molecular mechanism underlying the observed differences in Tnni3k transcript levels , they did not address the in vivo role of Tnni3k in the progression of cardiomyopathy . We next investigated whether Tnni3k was the gene underlying the Hrtfm2 locus . We created transgenic mouse lines that expressed human TNNI3K protein in the heart . TNNI3K protein is highly conserved between human and mouse ( 91% identity ) , and transgenic expression of the human transcript enabled discrimination between the endogenous murine transcript and that derived from the transgene . Three independent founder lines were created , and qRT-PCR indicated that the human transgene was expressed at levels ranging from 5 to 20-fold above the endogenous B6 mouse transcript , depending on the founder . F1 generation mice from all three lines survived over a year , and cardiac function in 12 and 21-week transgenic animals were indistinguishable from wild-type animals . Consequently , TNNI3K expression alone did not result in overt cardiomyopathy or decreased survival due to heart failure . This was not unexpected , since in the absence of the Csq transgenic disease-sensitizer , there were no measurable differences in heart function between B6 and DBA animals , even though B6 express robust levels of TNNI3K whereas DBA shows no detectable protein . By repeated backcrosses to DBA , the TNNI3K transgenes were introgressed into the DBA background that shows no detectable endogenous murine TNNI3K protein to test the hypothesis that in the presence of the Csq transgenic sensitizer , increased expression of TNNI3K would accelerate disease progression . The backcrossed transgenic lines continued to express robust levels of the human TNNI3K protein ( Figure S1 ) . Two lines were chosen for all subsequent experimental crosses , and phenotypic data from N5 ( or N6 ) animals from both lines were combined , as there was no discernable difference in the data derived from either transgenic line . In addition to an apparently normal lifespan , TNNI3K transgenic animals did not show any signs of cardiac pathology by echocardiography . By contrast , expression of TNNI3K in the context of the Csq transgenic sensitizer resulted in profoundly premature death ( Figure 5 ) . Of the four possible genotypes from a cross between Csq ( sensitizer ) and TNNI3K ( modifier ) transgenic lines , only the double transgenic mice showed a decrease in survival ( p<0 . 00001 ) . The observed survival differences were profound . All other genotypes survived on average at least 150 days , but all animals expressing both Csq and TNNI3K died within 21 days . This extreme premature death phenotype resembled that which we had previously observed when attempting to introgress the Csq transgene into the B6 background , which exhibits robust levels of endogenous mouse TNNI3K protein [5] . Starting with the sensitizer in the DBA background [4] , we were unable to move the Csq transgene beyond the second generation , as N2 animals died within 40 days , precluding further backcrosses with B6 mice [5] . We next determined whether natural levels of the murine TNNI3K protein would also exhibit disease-accelerating effects . This was investigated by crossing the congenic mice harboring the AKR allele at Hrtfm2 with mice containing the Csq transgene , both held for many generations in the DBA background . The progeny from this cross would harbor a only single AKR allele of Tnni3k , the appropriate genotype for Hrtfm2 which exhibited dominant effects in the original mapping crosses [5] , [7] . When the congenic mice DBA . AKR-Hrtfm2 , were crossed with the Csq sensitized mice , the Csqtg offspring with even only a single AKR Hrtfm2 allele showed a decrease in survival ( on average 107 days ) compared to those with two DBA Hrtfm2 alleles ( more than 150 days , Figure 5 ) . The Csq-sensitized Hrtfm2 congenic line also survived longer than Csq-sensitized F1 ( DBA/B6 ) animals , which survived on average to only 50 days [5] . In the original mapping cross , Hrtfm2 contributed approximately 30% of the genetic variance towards the survival trait [5] . Thus , as expected , the isolated B6/AKR Hrtfm2 allele contributed a robust but only partial effect on reduced survival when compared to the F1 animals , as other modifier loci had been crossed out of the congenic line . To determine whether the premature death was related to cardiac dysfunction , we performed echocardiography on animals with all four possible genotypes resulting from cross between the Csq and TNNI3K transgenic lines . Echocardiography was performed at 14 days , the earliest possible age for reproducible echocardiographic data . Due to the extremely accelerated disease course and profound reduction in survival , only six of fourteen double transgenic mice survived to their scheduled 14-day echocardiogram . Fractional shortening in these TNNI3Ktg/Csqtg mice was significantly decreased compared to the other three genotypes of animals ( P<0 . 0232 ) , demonstrating severely abnormal heart function of the double transgenic animal ( Figure 6A and 6B , Table S2 ) . Even by 14 days , hearts from the TNNI3Ktg/Csqtg mice were larger than those of the other genotypes , and by histological staining showed obvious chamber dilation ( Figure 6C ) . Thus , the double transgenic animals developed dilated cardiomyopathy by 14 days ( or earlier ) and all mice of this genotype died before this or shortly thereafter due to heart failure . Many of the double transgenic animals displayed bradycardia ( a severe slowing of the heart rate ) , clearly evident in the echocardiograms . This phenotype , while a feature of the natural disease progression in the Csq transgenic model , is normally observed only in adult animals just prior to heart failure [5] . Thus , this hallmark of the natural progression of the Csq transgenic model is also greatly accelerated with overexpression of TNNI3K . We also investigated the echocardiographic parameters of the DBA . AKR-Hrtfm2/Csqtg ( see Figure 5 ) . Echocardiography performed on these mice at 4 and 8 weeks of age showed decreased fractional shortening in DBA . AKR-Hrtfm2/Csqtg mice compared to the DBA/Csqtg littermates , indicating more severe level of cardiomyopathy ( Figure 7 , Table S3 ) . Furthermore , from age 4 to 8 weeks , percent fractional shortening decreased more rapidly in DBA . AKR-Hrtfm2/Csqtg mice ( 36% , p = 0 . 0004 ) than the littermates ( 18% , p = 0 . 186 ) , suggesting that the presence of even a single AKR-Hrtfm2 locus ( essentially half the normal AKR/B6 level of TNNI3K expression ) can accelerate the progression of the Csq-induced cardiomyopathy . We next investigated whether TNNI3K expression would exhibit a disease accelerating effect in a model of cardiomyopathy that was unrelated to Calsequestrin over-expression . Transverse aortic constriction ( TAC ) induces left ventricular hypertrophy in response to pressure overload [13] . We performed TAC on TNNI3K transgenic animals and wild-type littermate controls . Cardiac function was analyzed by echocardiography at 4 and 8 weeks following TAC surgery . At 4 and 8 weeks post-surgery , the transgene-positive mice showed greater diastolic and systolic dysfunction ( increased left-ventricular end diastolic diameter ( LVEDD ) and left-ventricular end systolic diameter ( LVESD ) ) , and significantly reduced fractional shortening compare to the control mice ( Figure 8A–8C , Table S4 ) . This confirmed that TNNI3K expression has a detrimental effect on heart function outside the context of the transgenic Csq sensitizer .
We had previously mapped 7 loci ( Hrtfm1-7 ) that modify heart disease progression using the Calsequestrin ( Csq ) transgenic mouse model [5]–[7] . Here we report Tnni3k ( cardiac Troponin I-interacting kinase ) as the gene underlying Hrtfm2 . We have shown that the murine Tnni3k locus harbors an ancestral SNP in intron 19 that activates a cryptic splice site , generating an aberrant transcript that undergoes NMD , leading to drastically reduced message levels and an apparent absence of TNNI3K protein . In DBA and other inbred mouse strains sharing the same haplotype at Tnni3k , drastically reduced levels of TNNI3K protein have no obvious effect on normal development or physiology , suggesting that any trace amounts of protein that remain are sufficient for its normal function , or that the lack of protein is compensated by functional redundancy of another gene . In vivo transgenic and congenic mouse lines confirm that TNNI3K levels are a significant determinant of the rate of disease progression and outcome , since expression of this protein accelerates disease progression in two independent and unrelated models of cardiomyopathy . However , we did not observe a simple , linear relationship between the level of TNNI3K transgenic overexpression and the strength of the modifying effect . In these models , the levels of overexpression in the transgenic lines may have crossed the threshold required for maximal phenotypic effects . The modifying effects of TNNI3K expression were not dependent on the allele at the nonsynonymous coding SNP ( rs30712233 , T659I ) . DBA and most inbred mouse strains encode Threonine at this position . Most other species also encode Threonine at the homologous position . By contrast , B6 and AKR inbred strains encode the Isoleucine variant . The human TNNI3K transgene employed in the validation experiments coincidentally encodes the highly conserved Threonine variant . The modifying effects of the human transgene carrying the conserved variant ( 695T ) directly parallel , and are even stronger , than those observed using the congenic line containing the B6/AKR variant ( 695I ) . Thus , robust phenotype modifying effects were observed independent of the murine Tnni3k coding variant . Nonetheless , these experiments did not address whether the 695I polymorphism also alters TNNI3K protein function . TNNI3K was identified as a cardiac-specific protein kinase that interacted with cardiac Troponin I ( cTnI ) in the yeast-two hybrid interaction assay [8] , however , cTnI has not been established as a phosphorylation target . TNNI3K protein contains seven ankyrin repeats in the N-terminus followed by a dual-specificity protein kinase domain and a short C-terminal serine-rich domain . The overall domain structure of TNNI3K resembles that of Integrin-linked Kinase ( ILK ) . ILK mediates communication from the cellular matrix to intracellular signaling molecules such as PKB and GSK3β , and plays important roles in cardiac growth , contractility and repair [14] , [15] . Sequence and structural homology might imply similar functions for TNNI3K . A yeast two-hybrid interaction screen with a C-terminal fragment of TNNI3K identified several additional sarcomeric proteins as putative binding partners such as cardiac α-actin and myosin binding protein C [8] . These studies suggest that TNNI3K might modulate sarcomere function through interactions with key components of the sarcomeric complex . However , to date , none of these proteins has been validated as a phosphorylation target of TNNI3K in cardiomyocytes , and the in vivo function of TNNI3K remains unknown . Recently , expression of TNNI3K was shown to be protective in a different cardiomyopathic disease context [16] , [17] . In a murine model of cardiac ischemia , intramyocardial transplantation of Tnni3k-overexpressing P19CL6 cells promoted cardiomyogenesis and improved cardiac function . We note that P19CL6 cells were originally derived from the C3H lineage of mice [18] , which share with DBA , the “null” haplotype for Tnni3k ( see Figure 1 ) . Thus , the resulting phenotype may have been due to the restoration of Tnni3k expression in otherwise null cells , rather than to overt overexpression of the gene . A locus for susceptibility to coxsackievirus B3-induced myocarditis maps to a locus on distal mouse chromosome 3 ( Vms1 , viral myocarditis susceptibility locus ) that virtually overlaps Hrtfm2 and which includes Tnni3k in its confidence interval [17] . In this disease model , inbred strain C57BL/10 provides the protective allele at the locus , and strain A/J provides the susceptibility allele . Assuming that Tnni3k also underlies the Vms1 locus , viral-induced myocarditis might represent another disease context where expression of TNNI3K is protective . These combined data suggest that expression of TNNI3K may be detrimental in certain pathological conditions such as pressure overload or aberrant sarcomeric calcium regulation , but protective in other disease contexts . In either scenario , TNNI3K appears to play a critical role in modulating disease progression and outcome in heart disease . Since protein kinases are critical cell cycle regulators , kinase inhibitors have become a major avenue for the development of novel cancer therapeutics . TNNI3K may be an ideal candidate for the development of small molecule kinase inhibitors for categories of heart disease where TNNI3K expression is detrimental . In these cases , selective inhibition of TNNI3K would be particularly useful as it might slow disease progression , which may prove beneficial in treating individuals with rapidly progressing disease . In other scenarios of disease , augmentation of TNNI3K activity or protein levels may instead prove beneficial . Further investigation of TNNI3K function in these and other cardiomyopathic mouse models will lead to increased understanding of its role in both normal and pathological contexts , and may provide a novel target for therapy for heart disease .
All mice were handled according to approved protocol and animal welfare regulations of the Institutional Review Board at Duke University Medical Center . All inbred mouse strains used in the course of this study were obtained from Jackson Laboratory ( Bar Harbor , ME ) . Transgenic mice overexpressing Csq [3] were maintained on a DBA/2J genetic background . Through repeated backcrossing to DBA/2J , a congenic mouse was created which retains AKR genomic DNA at the Hrtfm2 locus in the DBA genetic background . At generation N2 , breeders were selected which were heterozygous at Hrtfm2 and homozygous DBA at the other mapped modifier loci [7] . Genome-wide SNP genotyping was carried out using the Mouse MD linkage panel with 1449 SNPs ( Illumina , San Diego , CA ) . By generation N6 , the animals were homozygous for DBA alleles throughout the genome and only showed heterozygosity for an approximately 20 Mb interval on chromosome 3 , the region containing Hrtfm2 . Once we had reached the generation N10 backcross , the DBA . AKR-Hrtfm2 mouse was maintained by intercross . Whole hearts removed from age- and sex-matched wild type animals from each of the three primary strains ( B6 , DBA , AKR ) were used to examine RNA transcript levels . Total RNA was isolated using the RNeasy Kit ( Qiagen , Valencia , CA ) . Microarray analysis was done on an Affymetrix Mouse probe set ( Mouse 430 2 . 0 Array , Affymetrix , Santa Clara , CA ) . Analysis was done using GeneSpring GX 7 . 3 Expression Analysis ( Agilent Technologies , Santa Clara , CA ) . For the TaqMan expression analysis , total RNA was extracted from whole mouse hearts using TRIzol reagent ( Invitrogen , Carlsbad , CA ) . cDNA was synthesized from 1 µg total RNA using the High Capacity cDNA Archive Kit ( Applied Biosystems , Foster City , CA ) and used as the template for qRT-PCR . Tnni3k cDNA was amplified using the predesigned gene expression assay ( TaqMan , ABI , assay ID: Mm01318633_m1 ) . Beta-actin ( Actb ) was used as the endogenous control ( TaqMan , ABI , catalogue number 4352341E ) . All amplifications were carried out in triplicate on an ABI Prism 7000 Real Time PCR system and analyzed with ABI software . All statistical analyses were done using an unpaired , two-tailed T-test . Whole heart protein lysates were prepared using flash-frozen heart tissue resuspended in lysis buffer with protease inhibitors . Lysates were analyzed by SDS-PAGE and Western blot performed with standard methods . A polyclonal peptide antiserum was developed to the C-terminal 14 amino acids of mouse TNNI3K protein ( LHSRRNSGSFEDGN ) . Antiserum from 2 rabbits was purified on a Protein A column ( GenScript , Piscataway , NJ ) . TNNI3K antibody was used at a 1∶1000 dilution in TBST with 5% dry milk . Secondary anti-rabbit antibody conjugated to HRP followed by incubation with Pierce SuperSignal West Pico Chemiluminescant Substrate ( Thermo Fisher Scientific , Rockford , IL ) and exposure to X-OMAT film ( Kodak ) to visualize protein bands . Western blot analysis was used to confirm specificity of the antibody . As predicted , the mTNNI3K antibody detects a 90 kDa protein from lysates prepared from 293T cells transiently transfected with a full length Tnni3k expression vector and in protein lysates from wild-type mouse hearts . cDNAs were subjected to qRT-PCR using primers designed to detect either a 116 bp or a 120 bp cDNA PCR product . The forward primer was targeted 25 bp upstream of the predicted 4 base insertion and was fluorescently labeled: 5′-6FAM-AGATTTCTGCAGTCCCTGGAT-3′ while the unlabeled reverse primer was targeted 48 bp downstream of the predicted 4 base insertion with the sequence: 5′-AAGACATCAGCCTTGATGGTG-3′ . Accumulation of both fragments was quantified using the GeneMapper analysis program on the ABI Prism 3730 DNA Sequencer ( Applied Biosystems ) . Ratios of properly spliced and mis-spliced products were calculated based on relative amplification of both cDNA products . To create the Tnni3k genomic splicing constructs , DBA genomic DNA and B6 BAC clone RP23-180023 were used as templates to generate genomic 4 kb fragments that included part of intron 17 , exon 18 , intron 18 , exon 19 , intron 19 , exon 20 and part of intron 20 . The sequence of the forward PCR primer was 5′-ACTTACTTATGTGCTTCTCTTAGTTATGTGC-3′; the reverse primer was 5′-GGATTTAAACATAGGTGTGTACCTAATTGT-3′ . PCR products were sub-cloned into pSPL3 ( Invitrogen ) . Clones were verified by direct sequencing . Human embryonic kidney HEK293T ( 293T ) cells ( ATCC , Manassas , VA ) were maintained in Dulbecco's Modified Eagle's Medium ( DMEM , Gibco ) containing 10% fetal bovine serum at 37°C in 5% CO2 . Cells were grown on 35 mm2 plates and transfected with 1 µg plasmid DNA using FuGene reagent ( Roche , Indianapolis , IN ) according to the manufacturer's protocol . RNA was extracted with TRIzol ( Invitrogen ) 24 hr post-transfection and RT-PCR was carried out using standard methods . HEK293T cells were grown to approximately 80% confluence in 6-well plates , then transfected using with 1 µg of DBA- or B6-pSPL3 plasmid mixed with FuGene reagent . All transfections were performed in triplicate . Total RNA was extracted with TRIzol 20 hr post-transfection . RT-PCR was carried out using standard methods . Ratios of properly spliced and aberrantly spliced products for the Tnni3k construct were determined by the fluorescent RT-PCR assay described above . A single base was changed at rs49812611 ( IVS19+9 ) , in the DBA-pSPL3 construct ( G→A ) and the B6-pSPL3 construct ( A→G ) using the QuikChange Site-Directed Mutagenesis Kit ( Stratagene , LaJolla , CA ) with PfuTurbo proofreading DNA polymerase . All clones were sequenced to verify proper incorporation of the SNP . HL-1 cardiomyocytes [11] were cultured in Claycomb Medium ( SAFC Laboratories , Lenexa , KS ) supplemented with Fetal Bovine Serum at 10% , 2 mM L-Glutamine , 100 mg/ml Penicillin/Streptomycin , and 100 mM fungizone . Cells were cultured at 37°C with 5% CO2 . Although the HL-1 cardiomyocytes were derived from a heart isolated from a mixed B6-DBA mouse [11] direct sequencing of genomic DNA from the cell line showed that it is homozygous for DBA alleles at the Tnni3k locus . HL-1 cells were treated with 5 . 7×10−2 mM cycloheximide or 3 . 3×10−2 mM emetine . Each treatment was performed in triplicate and RNA was isolated from cells 24 hours post treatment . RT-PCR was performed on RNA isolated from cells treated with NMD blocking drugs and untreated controls . Ratios of properly spliced and aberrantly spliced products were measured using the fluorescent RT-PCR splicing assay as described above . Total transcript levels were determined using the Tnni3k TaqMan assay described above . A full-length 2 . 5 kb human TNNI3K cDNA was amplified from normal human heart RNA following RT-PCR and cloned into a vector downstream of the murine α-myosin heavy chain ( αMHC ) promoter . An artificial minx intron was inserted upstream of the TNNI3K start codon . The construct was linearized and an 8 kb fragment containing the αMHC promoter , cDNA and SV40 polyadenylation sequence was purified and used for microinjection . B6SJLF1/J blastocysts were injected with the linearized transgene and subsequently implanted into surrogate mice . The resulting founder animals were genotyped for presence of the TNNI3K transgene using a 5′ primer in the αMHC promoter and a 3′ primer in the TNNI3K transgene . Three transgenic lines were chosen for backcrossing to the DBA strain . Western blot analysis of heart lysates with a polyclonal antibody ( Bethyl Laboratories , Montgomery , TX ) raised against a human C-terminal TNNI3K peptide ( FHSCRNSSSFEDSS ) confirmed similar levels of expression of the TNNI3K transgene in each line ( Figure S1 ) . This was repeated for several generations of backcrossing to DBA . Southern blot analysis of DNA from founder animals and subsequent generations ( N2–N3 ) indicated that two founder lines carried 10–20 copies of the transgene while the third line appeared to have >100 copies . qRT-PCR with SYBRgreen ( Invitrogen ) was performed on heart cDNA from several transgenic mice to determine the relative expression difference between endogenous mouse Tnni3k and transgenic human TNNI3K expression . Transthoracic two-dimensional M-mode echocardiography was performed between 12 and 18 weeks of age in conscious mice using either a Vevo 770 echocardiograph ( Visual Sonics , Toronto , Canada ) or an HDI 5000 echocardiograph with a 15-MHz frequency probe ( Phillips Electronics , Bothell , WA ) . Measurements of cardiac function include heart rate , posterior and septal wall thickness , left-ventricular end diastolic diameter ( LVEDD ) and left-ventricular end systolic diameter ( LVESD ) . Fractional shortening ( FS ) was calculated with the formula: FS = ( LVEDD−LVESD ) /LVEDD , as previously described [4] . Hearts were fixed in 10% neutral buffered formalin , dehydrated in 75% , 90% and 100% ethanol , and embedded in paraffin; sections 5 mm in thickness were cut and then stained with Masson's trichrome stain . Mice were anesthetized with a mixture of ketamine ( 100 mg/kg ) and xylazine ( 2 . 5 mg/kg ) , and transverse aortic constriction ( TAC ) was performed as previously described [13] . TAC was performed on 14 TNNI3K transgene-positive animals and 14 transgene-negative ( wild-type ) littermates at 10 weeks of age . One of the transgene-negative controls and three transgene-positive animals died following surgery , which is a normal complication of this procedure . The remaining 24 mice were then analyzed by echocardiography ( as described above ) , at 4 and 8 weeks following the surgery . | Heart failure is the common final outcome of many forms of acute and chronic heart disease . The prognosis of heart disease is highly variable between patients , and these differences in the phenotypic expression ( symptoms , course , and final outcome ) are in part due to genetic factors that have proven difficult to directly identify in the human population . To overcome this limitation , we employed a disease-sensitized mouse model of dilated cardiomyopathy to identify genes that modify the progression and outcome of the phenotype . Here we report the identification of a novel heart disease modifier gene , Tnni3k , that accelerates disease progression in two distinct mouse models of cardiomyopathy . This gene appears to play a critical role in modulating heart disease phenotypes and may provide a novel target for therapeutic intervention . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"genetics",
"and",
"genomics/disease",
"models",
"genetics",
"and",
"genomics/genetics",
"of",
"disease",
"genetics",
"and",
"genomics/gene",
"discovery"
] | 2009 | Tnni3k Modifies Disease Progression in Murine Models of Cardiomyopathy |
When selective pressures differ between males and females , the genes experiencing these conflicting evolutionary forces are said to be sexually antagonistic . Although the phenotypic effect of these genes has been documented in both wild and laboratory populations , their identity , number , and location remains unknown . Here , by combining data on sex-specific fitness and genome-wide transcript abundance in a quantitative genetic framework , we identified a group of candidate genes experiencing sexually antagonistic selection in the adult , which correspond to 8% of Drosophila melanogaster genes . As predicted , the X chromosome is enriched for these genes , but surprisingly they represent only a small proportion of the total number of sex-biased transcripts , indicating that the latter is a poor predictor of sexual antagonism . Furthermore , the majority of genes whose expression profiles showed a significant relationship with either male or female adult fitness are also sexually antagonistic . These results provide a first insight into the genetic basis of intralocus sexual conflict and indicate that genetic variation for fitness is dominated and maintained by sexual antagonism , potentially neutralizing any indirect genetic benefits of sexual selection .
Males and females differ in the optimal value for most behavioural , morphological , and physiological traits [1] , as a consequence of the different strategies they adopt to maximize their fitness [2] , [3] . At the genetic level , these differences trigger an evolutionary conflict between the sexes . For any given genetic locus , an allele may be favoured by selection in males , while a different allele is favoured in females . Hence , intralocus sexual conflict occurs when selection acts differentially on the same locus in the two sexes [4] . If many loci experience this sexually antagonistic selection , sets of alleles that are positively selected in males will produce a “good” male phenotype but a “bad” female phenotype , while the opposite will be true for other sets of alleles positively selected in females . Over the past decade , the phenotypic effects of intralocus sexual conflict have been demonstrated using two major lines of evidence: first , from studies showing a negative genetic correlation for fitness between the sexes , both in wild and laboratory populations [5] , [6] , and second , from experimental evolution studies , where gender-limited selection resulted in relatively higher fitness of the selected sex [7] , [8] . Furthermore , sexually antagonistic selection appears to be a taxonomically widespread phenomenon [9] . Although the effects of intralocus sexual conflict on the whole organism are receiving increasing attention [10] , very little is known about the genetics underlying the patterns observed , namely the identity , number , or location of the genes involved . So far , two predictions have been made about the features of sexually antagonistic genes . First , sexually antagonistic loci should accumulate on the sex chromosomes [1] due to their patterns of inheritance in the two sexes [11] . Second , since the genetic information available to males and females is largely coincidental , sexual dimorphism is expected to arise through differences in where , when , and to what extent genes are expressed [12] , as a way to resolve the conflict and to mitigate the “gender load” [1] . Numerous studies have employed sex bias in gene expression as a proxy for sexual antagonism [13]–[17] with the assumption that sexual dimorphism in expression levels reflects the current extent to which sexual conflict is present at each locus . However , as some authors explicitly note [9] , [12] , [13] , sex-biased expression is more likely to represent a partial or total resolution to the conflict , and the assumption that sex-biased expression equals sexual antagonism remains to be demonstrated . An explicit test of these predictions at the gene level is only possible when a set of candidate genes has been identified . The aim of this study was therefore to provide an empirical test of current sexual conflict theory with respect to the genome-wide number , location , and function of sexually antagonistic genes in an outbred population of D . melanogaster .
Our results provide the first direct test , to our knowledge , of the identity , quantity , and location of sexually antagonistic genes in any organism . These data show that sexually antagonistic selection has a non-negligible effect on fitness-related genes , and as such its neutralizing effect on “good genes” processes in sexual selection should no longer be overlooked [19] . They also give an indication of the extent to which this process may maintain genetic variation in the face of sexual ( i . e . , the lek paradox [29] ) or natural selection [5] , [30] . The presence of sexual antagonism in sex-limited tissues other than the gonads also provides evidence of a link between intralocus and interlocus sexual conflict , since the accessory gland in males and sperm-storage organs in females are known to play an important role in male-female coevolution [31] , [32] . We expect our results will be a starting point from which a more detailed functional genomic analysis of sexual conflict can proceed . In particular , a better understanding of the function , genomic location , and the degree of linkage in a gene network ( epistasis and pleiotropy ) of each locus under conflict might provide insights into the processes that allow or prevent conflict resolution [10] .
The base population of Drosophila melanogaster ( LHM ) has been maintained as a large , outbred population for over 400 non-overlapping generations . One hundred haplotypes were sampled from LHM and maintained as heterozygous stock hemiclonal lines using double-X clone-generator females [C ( 1 ) DX , y , f; T ( 2;3 ) rdgC st in ri pP bwD][6] , [18] . Hemiclonal haplotypes were expressed as males by mating stock hemiclone males with virgin double-X LHM females [C ( 1 ) DX , y , f] and expressed as females by mating with virgin LHM females . Each hemiclonal fly therefore shares one nearly complete genomic haplotype ( with the exception of the fourth dot chromosome ) , the other being a random sample from the base population . Given the patterns of inheritance of a hemiclonal genotype , the variation across lines does not include any non-additive dominance variation or maternal effects , although some epistatic interactions remain [18] . Adult fitness of hemiclones was assayed in competition with individuals from a replica population of LHM marked with the bw− eye-colour allele . All flies were reared in 25 mm vials on cornmeal-molasses-agar food . The total adult lifetime fitness of 100 hemiclonal haplotypes when expressed as either males or females was assayed under competitive conditions that closely match those experienced by adults in the base population [18] . Competitor flies homozygous for the brown eye-colour allele bw- were generated following nine rounds of backcrossing into LHM . For the male assays , hemiclonal males were first generated by mating stock hemiclonal males to 30 virgin double-X LHM [C ( 1 ) DX , y , f] females . These females were allowed to oviposit in vials for 18 h , after which the density of eggs was reduced so that approximately 150 viable zygotes remain ( 3/4 of the zygotes are lethal aneuploids ) . Five hemiclonal wild-type males arising from this cross were then placed together with 10 competitor bw− males and 15 virgin bw− females ( reared at the same larval density and matched for age ) in yeasted vials for 2 d . The females were then isolated in test tubes and allowed to oviposit for 18 h . On Day 12 , the progeny from each female was scored for eye colour . This assay was replicated 6 times , representing a total of 30 hemiclonal males per line . The relative adult male fitness for each line was calculated by averaging the relative fitness across replicates , obtained by dividing the proportion of offspring sired by hemiclonal males ( bw+/bw− ) by the maximum proportion across all hemiclonal lines and replicates . For the female assays , the protocol was identical except that hemiclonal females were obtained by mating hemiclonal stock males to groups of 16 virgin LHM females ( producing half aneuploids ) . Groups of 5 hemiclonal females were housed with 10 competitor females and 15 bw− males in yeasted vials for 2 d . The hemiclonal females were then placed in individual test tubes and allowed to oviposit for 18 h . This assay was replicated 4 times , representing a total of 20 hemiclonal females per line . Relative adult female fitness for each line was calculated by averaging across replicates the mean number of progeny emerging by Day 12 divided by the maximum fecundity across all lines and replicates . All statistical analyses were performed using R [33] 2 . 9 ( http://www . R-project . org ) . Fitness assay data were analysed by fitting a linear mixed model using Bayesian methods and Markov chain Monte Carlo sampling techniques ( MCMCglmm package ) to data on relative male and female fitness: Y = S + L + ε , where S ( sex ) is a fixed effect , L ( line ) is a 2×2 matrix that specifies the variance structure of the random effects , allowing for estimates of sex-specific variances among lines and their covariance , and ε is a matrix of sex-specific , within-line residual variances . Flat priors for the correlation were used . Fifteen lines showing hyper-dispersed variation in relative male and female fitness based on ranks were selected for expression analysis with DNA microarrays . We chose five lines each showing low-male/high-female fitness ranks , high-male/low-female fitness ranks , and average-male/average-female fitness ranks ( see Figure 1 ) as well as low variance . Four independent replicates of hemiclonal males and females from each of the 15 selected lines were generated following the same crosses described above ( but with 12 hemiclonal stock males:30 females ) . Adult hemiclonal and LHM tester flies of both sexes ( reared following the base population protocol ) were then collected in groups of 16 on Day 10 . On Day 12 , each group of hemiclones was placed together with a group of tester flies of the opposite sex in yeasted vials . After 24 h , the tester flies were removed and after a further 20 h a group of six hemiclonal flies were randomly chosen from each vial under brief CO2 anaesthesia . Four hours after sorting , the flies were frozen using liquid nitrogen and stored at −80°C for no more than 6 d until RNA extraction . Total RNA was extracted using Trizol ( Invitrogen ) and purified with an RNeasy Mini Kit ( Qiagen ) , from four independent groups of six flies for each sex/line ( 2 sexes , 15 lines , 4 replicates , giving a total of 120 arrays and 720 flies ) . RNA quantity and quality was assessed with an Agilent Bioanalyzer ( Agilent Technologies ) prior to sample preparation and hybridisation following the manufacturer's instructions to GeneChip Drosophila Genome 2 . 0 Affymetrix microarrays at the Uppsala Array Platform . The 120 microarrays were processed in 8 batches of 15 . Several packages within BioConductor [34] 2 . 4 ( http://www . bioconductor . org ) were used for gene expression data analysis . Microarray data were pre-processed using Robust Multichip Average ( RMA ) as implemented by the affy package [35] . The phenotypic variation in gene expression was partitioned using the following linear restricted maximum-likelihood mixed model ( lme4 package ) : Y = B + S + L + S × L + ε , where S ( sex ) is a fixed effect , L ( line ) is a random effect , and B is a random effect introduced to block for the effect of batch . A similar model ( without the S and interaction terms ) was fitted to sex-specific subsets of the data . The p values for random effects were calculated using a 0 . 5χ02+0 . 5χ12 mixture distribution from a Likelihood Ratio Test on the full and reduced ( without the random effect to be evaluated ) models . All the reported p values were corrected for FDR [36] . We used the following regression model: Y = B + S + F + S × F + ε ( S = sex as fixed effect; F = sex-specific line fitness , covariate; B = batch as random blocking factor ) to identify transcript associated with fitness ( limma package ) . A similar model ( without the S and interaction term ) was fitted to sex-specific subsets of the data . A Bayesian approach to pool information across genes has been used to moderate the variance [37] . All the reported p values were corrected for FDR [36] . We identified tissue-specific transcripts using the Flyatlas database [38] . Raw data were downloaded by GEO ( Gene Expression Omnibus , accession number GSE7763 ) and pre-processed with RMA ( as default in affy package [35] ) separately for each tissue . Expression values were then averaged across replicates and rescaled to whole-fly baseline expression ( also obtained from FlyAtlas , to ensure homogeneity of the experimental procedures ) using the average expression of unexpressed genes ( n = 599 , expression value in the whole fly smaller than 3 . 4 ) . Rescaling was necessary only to ensure an equal signal baseline for all the tissues . Transcripts were considered tissue specific if the expression level in the target tissue was 2-fold higher than in the whole fly . To test for overabundance of genes of interest in a target tissue , we performed a one-tailed Fisher's exact test on the observed and expected tissue-specific genes of interest compared to the overall number of tissues-specific genes in each tissue . All the reported p values were Bonferroni-corrected for testing on multiple tissues ( n = 17 ) . To identify GO categories and chromosomes ( or chromosomal bands ) enriched for particular subsets of transcripts , we used a hypergeometric test for overrepresentation ( p <0 . 05 , GOstats and Category packages , modified ) . Microarray data are deposited on the GEO database , accession number GSE17013 . | Males and females of many species are different: many of these differences are thought to have evolved because the sexes often have needs and strategies that do not coincide . For example , in fruit-flies , females may do best by concentrating their efforts in acquiring resources to be able to lay more eggs , while males would benefit most from increasing their mating and fertilization success . Such differences generate a sexual “conflict of interests” , and since as a general rule each behavioural , morphological or physiological characteristic is regulated by the same set of genes in the two sexes , this conflict takes place ultimately at the genetic level . In our study , we combined data on the reproductive success of different lines of fruit-flies with their gene expression profiles . We show that a large proportion of genes that contribute to male fertilization success are detrimental for female fecundity , and vice-versa . These results indicate that an optimal genotype for both sexes does not exist: many genes maintain different variants because they have opposite effects in males and females , perhaps helping to explain how genetic diversity is maintained in the face of selection . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/gene",
"expression",
"evolutionary",
"biology/evolutionary",
"ecology",
"evolutionary",
"biology/sexual",
"behavior",
"genetics",
"and",
"genomics/complex",
"traits",
"evolutionary",
"biology/genomics"
] | 2010 | The Sexually Antagonistic Genes of Drosophila melanogaster |
The apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3 ( APOBEC3 ) proteins are cell-encoded cytidine deaminases , some of which , such as APOBEC3G ( A3G ) and APOBEC3F ( A3F ) , act as potent human immunodeficiency virus type-1 ( HIV-1 ) restriction factors . These proteins require packaging into HIV-1 particles to exert their antiviral activities , but the molecular mechanism by which this occurs is incompletely understood . The nucleocapsid ( NC ) region of HIV-1 Gag is required for efficient incorporation of A3G and A3F , and the interaction between A3G and NC has previously been shown to be RNA-dependent . Here , we address this issue in detail by first determining which RNAs are able to bind to A3G and A3F in HV-1 infected cells , as well as in cell-free virions , using the unbiased individual-nucleotide resolution UV cross-linking and immunoprecipitation ( iCLIP ) method . We show that A3G and A3F bind many different types of RNA , including HIV-1 RNA , cellular mRNAs and small non-coding RNAs such as the Y or 7SL RNAs . Interestingly , A3G/F incorporation is unaffected when the levels of packaged HIV-1 genomic RNA ( gRNA ) and 7SL RNA are reduced , implying that these RNAs are not essential for efficient A3G/F packaging . Confirming earlier work , HIV-1 particles formed with Gag lacking the NC domain ( Gag ΔNC ) fail to encapsidate A3G/F . Here , we exploit this system by demonstrating that the addition of an assortment of heterologous RNA-binding proteins and domains to Gag ΔNC efficiently restored A3G/F packaging , indicating that A3G and A3F have the ability to engage multiple RNAs to ensure viral encapsidation . We propose that the rather indiscriminate RNA binding characteristics of A3G and A3F promote functionality by enabling recruitment into a wide range of retroviral particles whose packaged RNA genomes comprise divergent sequences .
The apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3 ( APOBEC3 , or A3 ) proteins have been identified as potent antiviral effector proteins [1 , 2] . There are seven family members in humans , each of which contains one ( A3A , A3C and A3H ) or two ( A3B , A3D , A3F and A3G ) characteristic zinc-coordination domains , one of which is catalytically active [3] . These proteins have been identified as inhibitors of retroviruses such as human immunodeficiency virus type-1 ( HIV-1 ) [4] , simian immunodeficiency viruses , murine leukaemia virus [5–7] and mouse mammary tumour virus [8] , as well as viruses from other families such as hepatitis B virus [9] , adeno-associated virus [10] and also endogenous retroelements [11] . Viruses have developed assorted strategies to evade A3-mediated inhibition , the most prominent of which is the expression of the dedicated regulatory protein , Vif , by most lentiviruses . Specifically , HIV-1 Vif counteracts APOBEC3 proteins by inducing their proteasomal degradation through the direct recruitment of CBF-β and a cellular E3 ubiquitin ligase comprising CUL5 , ELOB/C , and RBX2 [12–15] . When Vif is absent or defective , APOBEC3 proteins are packaged into progeny virions and transferred to target cells during new infections , where they inhibit reverse transcription and hypermutate nascent cDNAs through excessive cytidine-to-uridine editing [5 , 7 , 16–19] . Thus , the encapsidation of APOBEC3 proteins into viral particles is essential for their antiviral activity , and a complete description of APOBEC3 protein function will require a full understanding of the packaging mechanism . APOBEC3 proteins are RNA binding proteins [20–22] . A3G associates in an RNA-dependent mechanism with multiple ribonucleoprotein ( RNP ) complexes and accumulates in cytoplasmic RNA-rich microdomains such as P-bodies , stress granules and Staufen-containing granules [23–26] . Localisation to these regions does not appear to be important for antiviral function [27 , 28] , and it has been suggested that sequestration in RNPs may be important for downregulation of APOBEC3 protein activity within cells . These findings further raise the question of how APOBEC3 proteins are packaged into assembling virions . One elegant study has demonstrated that it is newly synthesised protein that is encapsidated , presumably by averting entrapment into cytoplasmic RNPs [29] . The packaging of A3G into HIV-1 particles requires the nucleocapsid ( NC ) region of the viral Gag protein [30–33] . It has been established that the A3G interaction with NC is RNA-dependent , leading to the consensus view that its packaging is reliant upon its capacity to bind RNA [30–33] . However , although several groups have sought to define specific RNAs that are responsible for A3G packaging , a clear consensus has not yet emerged . In particular , Kahn et al . suggested that viral genomic RNA ( gRNA ) is required for A3G packaging [34] , Wang et al . have concluded that 7SL RNA is the responsible RNA [35] , and Svarovskaia et al . proposed that both viral and cellular RNAs play a role [32] . A3G interacts with diverse RNAs such as the 7SL RNA ( the RNA component of the signal recognition particle , SRP ) , Alu RNAs , human Y RNAs and several mRNAs [24 , 36] , many of which are also packaged into retroviruses ( reviewed in [37] ) . In our study , we used an unbiased strategy to identify the RNAs that interact with A3G or A3F in HIV-1 infected cells and in budded virions , and then undertook specific experiments to investigate the involvement of such RNAs in A3G and A3F packaging . To define the interacting RNAs , we employed a cross-linking and immunoprecipitation technique ( iCLIP ) followed by next generation sequencing . This method has successfully defined RNAs that interact with proteins such as neuro-oncological ventral antigen ( NOVA ) , hnRNP C and Argonautes [38–40] . Here , we demonstrate that A3G and A3F interact with a broad range of RNA molecules , including HIV-1 gRNA , cellular mRNAs and a number of small non-coding RNAs . A series of cell-based assays revealed that no single/unique RNA mediates the encapsidation of A3G or A3F , suggesting that multiple , diverse RNAs can recruit APOBEC3 proteins into viral particles , provided that they are themselves packaged . We therefore propose that A3G and A3F exploit their relatively non-specific RNA binding capabilities to patrol the cytoplasm for nascent retroviral RNA , thereby ensuring effective capture by assembling viruses and resultant antiviral function .
Packaging of A3G into HIV-1 particles requires the nucleocapsid ( NC ) region of p55Gag , and it has been established that this interaction is RNA-dependent [30–33] . This led to the consensus view that A3G packaging depends on RNA binding . In order to identify the specific RNAs to which A3G binds for efficient packaging , we first applied an unbiased , deep sequencing method to catalogue the RNAs that are bound to A3G and A3F in living cells productively infected with HIV-1 . We also extended this study to determine A3G and A3F target RNAs in cell-free HIV-1 virions . To generate libraries of RNAs bound to A3G or A3F , we first generated CEM-SS human T-cell lines that stably expressed GFP ( negative control ) , GFP-A3G , GFP-A3F , T7-GFP ( negative control ) , T7-A3G or T7-A3F ( S1 Fig . ) . By using two different immunoprecipitation tags for A3G and A3F , we could identify if either the GFP or T7-epitope tag biased the resulting library . Importantly , our study is distinguished from all others ( to the best of our knowledge ) by inclusion of GFP-only controls: this is an important addition as it allows the determination of RNA binding enrichment relative to background . These cultures were challenged with vif-deficient HIV-1 such that more than 90% of the cells were infected , as judged by intracellular p24Gag staining ( S2 Fig . ) . 48 h after infection , the supernatants were collected and used to assess A3G and A3F antiviral efficacy . Regardless of the tag , both A3G and A3F were antiviral and inhibited single-cycle virus infectivity by 100- and 30-fold , respectively ( S3 Fig . ) . Virus-producing cells and viruses produced from these cells were used to generate iCLIP libraries . Fig . 1A shows RNA cross-linked to the proteins of interest after T7 or GFP directed immunoprecipitation , RNAse digestion ( used to shorten the length of bound RNAs for later deep sequencing ) and ligation of a linker . Of note , A3G and A3F each cross-link vastly more RNA than GFP , consistent with the fact that these proteins are established RNA binding proteins . The RNAs that migrate at higher molecular weights than the protein of interest were extracted . A primer that anneals to the linker was then used to generate cDNA . The cDNAs were circularised , a further primer annealed over the linker to create a region of double stranded DNA and digested with BamHI . This procedure generated DNAs where the unknown sequence was flanked by linkers , allowing PCR amplification of the library ( Fig . 1B ) and next generation sequencing . The sequencing provided a total of 20 million reads of 50 nucleotides each . The libraries obtained for each protein contained between 0 . 1 and 5 million sequences . We aligned the reads to the human genome and to the HIV-1 genome using bowtie [41] . Only reads that aligned uniquely in the genomes with a maximum of two mismatched nucleotides were considered for analysis . Furthermore , the reads containing the same random barcode and truncated at the same nucleotide were considered PCR artefacts , and only one of such reads was considered . These unique reads represented between 80% and 95% of the total reads for each replicate experiment , demonstrating the high level of sequencing library complexity obtained in the experiments . Human mRNAs were split into regions , specifically the 5’- and 3’-untranslated regions ( UTRs ) , introns and open reading frames ( ORFs ) , and each region type was analysed independently . The number of reads aligning to a specific sequence or class of sequences was divided by the total number of aligned reads obtained from the library and then compared with the GFP control to provide measurements of enrichment . We performed two independent experiments for each cell line . By comparing these replicates , we observed that the data sets were highly correlated ( r>0 . 98 ) . The data obtained from the differently tagged ( GFP versus T7 ) proteins also exhibited high correlation ( r>0 . 95 ) . Therefore , we present here the averaged data obtained for the four libraries with standard deviations . As summarised in Fig . 1C , A3G and A3F binding to mRNAs was enriched compared to the GFP control . While the binding of both A3G and A3F to the 3’-UTRs was higher ( 2-fold ) , only A3G was enriched in the ORF or 5’-UTR regions ( 2- and 3-fold , respectively ) ; Fig . 1C and S1 Table ) . We next looked in detail at virion-associated RNA: both the gRNA and host RNAs such as tRNAs , U RNAs , 7SL , ribosomal RNAs and Y RNAs [37] . Because these cellular RNAs are transcribed from repeat elements in the human genome , the reads were aligned using the repeat masker software [42] to a library containing consensus sequences of the elements , allowing misalignment of up to 3 nucleotides per read . As before , hits in each repeat consensus were divided by the total number of reads and compared to the GFP negative control . Fig . 1D and S2 Table show that A3G and A3F bound 2-fold more to the viral gRNA compared with GFP . Also , both A3G and A3F bound to Y RNAs ( 17- and 7-fold , respectively ) and to U RNAs ( 2-fold ) , as compared to GFP . Interestingly , A3G bound to 7SL ( 14-fold ) while A3F did not . As depicted in Fig . 1E , we observed that in infected cells , 20% of the RNA that is bound to A3G or A3F is of viral origin . However , in viral particles this RNA constitutes the majority of the library ( 80% of reads aligned to HIV-1 gRNA ) . This implies that , in vif-deficient HIV-1 , A3G/F may be packaged mainly through interactions with HIV-1 gRNA , or that APOBEC3 proteins bind to encapsidated cellular RNAs and then transfer to gRNA once inside the viral particles . In light of the diversity of RNA substrates bound by A3G or A3F , we undertook a series of studies designed to determine which RNAs can mediate the encapsidation of A3G or A3F into budding HIV-1 particles . One obvious RNA species that could potentially mediate the packaging of A3G into particles is the viral gRNA . Indeed , at least one previous report has considered this RNA to be essential for A3G packaging [34] . Since A3G and A3F are clearly able to bind to this RNA in infected cells , we performed packaging assays to test this hypothesis . First , we used lentiviral vectors with or without gRNA ( Fig . 2A ) . Succinctly , 293T cells stably expressing HA-tagged A3G or A3F were co-transfected with the HIV-1-based packaging plasmid ( p8 . 91 ) [43] , the VSV glycoprotein envelope expression vector , and either the pHR’SIN-cPPT-SEW lentiviral vector plasmid ( denoted lt vector ) [44] that expresses gRNA with an intact packaging signal ( Ψ ) or a mock plasmid . Immunoblot analysis of particles harvested 48 h post transfection and isolated through a sucrose cushion shows that A3G and A3F were packaged into both viral vectors with almost identical efficiency , indicating that viral gRNA is not required for effective A3G or A3F packaging . These data were next confirmed using an alternative experimental system where the gRNA also serves as the mRNA for Gag ( Fig . 2B ) . 293T cells stably expressing HA-tagged A3G or A3F were transfected with expression vectors encoding: ( i ) Gag-Pol , ( ii ) Gag , ( iii ) Gag in which NC had been replaced by a leucine zipper domain ( Zwt ) to allow for Gag multimerisation and VLP production ( Gag ΔNC ) [45] , or ( iv ) a Gag-Pol vector in which the previously defined gRNA packaging determinant between SL2 and SL3 [46] had been deleted ( ΔΨ ) ; ( S4 Fig . ) . The protease region of Pol was rendered inactive in the Gag-Pol constructs to facilitate the detection and comparison of Gag across the different samples . Analysis of viral-like particles ( VLPs ) confirmed that NC is necessary for efficient A3G and A3F packaging , but showed that the Ψ element is dispensable ( Fig . 2B , lanes 3 and 7 ) [30–33] . Since Gag alone was able to package A3G or A3F efficiently , we conclude that Pol is also not required for the packaging of A3G and A3F ( Fig . 2B , compare lanes 4 with 1 and 8 with 5 ) . We also determined the amount of gRNA packaged into wild type Gag , Gag ΔNC and Gag-Pol ΔΨ VLPs by quantitative real time PCR ( qRT-PCR ) analysis ( Fig . 2C ) . As anticipated , the Gag ΔNC , and ΔΨ VLPs each contained <5% of the level of gRNA compared to wild type Gag VLPs . We next examined a second RNA , the non-coding 7SL RNA of the SRP , for its importance in A3G/F packaging . This RNA is incorporated into retroviral particles [47–49] . In particular , our iCLIP analyses ( Fig . 1D ) confirmed earlier work showing that A3G , unlike A3F , binds to 7SL RNA [50] . Of note , one group has previously reported that 7SL is the RNA required for A3G packaging [35] , whereas a second concluded the opposite [51] . Over-expression of SRP19 , a protein component of the SRP , reduces the amount of 7SL RNA packaging into HIV-1 particles [35] , presumably by binding to free cellular 7SL RNA and precluding its interaction with Gag and resultant incorporation into assembling virions . Accordingly , we transfected 293T cells stably expressing A3G or A3F with a plasmid encoding the vif-deficient NL4-3 provirus together with an expression vector for SRP19 or a control vector . Immunoblot analysis 48 h post transfection showed that both A3G and A3F were still packaged into virions , irrespective of the reduction in virion-associated 7SL RNA ( Fig . 3A ) . Interestingly , reducing the amount of packaged 7SL did not influence the infectivity of the viruses or the antiviral activity of the packaged A3G or A3F ( S5 Fig . ) . The data were then confirmed using Gag VLPs , with Gag ΔNC serving as a negative control ( Fig . 3B ) . Quantitative RT-PCR analysis of RNA extracted from these VLPs confirmed that SRP19 overexpression reduced virion-associated 7SL RNA levels by more than 90% ( Fig . 3C ) . Our data therefore rule out a unique requirement for 7SL RNA for efficient A3G or A3F incorporation into virions , suggesting that their packaging may be mediated by other RNAs . To investigate a potential redundancy between HIV-1 gRNA and host 7SL RNA , we next transfected 293T cells expressing A3G or A3F with a vector encoding SRP19 to inhibit 7SL packaging , as well as the aforementioned ΔΨ construct to generate VLPs depleted of gRNA . We observed that A3G and A3F were both packaged with normal efficiencies irrespective of SRP19 over-expression and/or the prevention of gRNA packaging ( Fig . 4 ) . Our observations imply that APOBEC3 proteins are packaged into these particles through the action of other RNAs . Our data suggested that a variety of host cell RNAs might be involved in APOBEC3 protein packaging into virions . To investigate this further , we next devised an alternative experimental approach whereby the packaging of RNA was dictated by heterologous RNA binding proteins ( or domains thereof ) rather than by the product of binding to NC and intracellular abundance . To do so , RNA binding domains were genetically fused to the carboxy-terminus of Gag ΔNC , and these proteins were used to generate VLPs in the presence or absence of A3G or A3F . In case of poor virion production , Gag ΔNC was co-expressed to ensure efficient VLP production [52 , 53] . To verify the experimental system , we initially expressed SRP19 fused to Gag ΔNC , anticipating that SRP19 would bind 7SL RNA and recruit it into VLPs together with A3G or A3F . Fig . 5A demonstrates that the addition of 6 amino acids to the C-terminus of Gag ΔNC , creating convenient restriction sites to allow fusions to Gag ΔNC ( lanes 3 and 7 ) , does not mediate APOBEC3 packaging . Also , A3G/F packaging was not mediated by the fusion of Gag to GFP , a protein that does not specifically bind to RNA ( lanes 4 and 8 ) . Although SRP19 was cleaved from Gag ( lanes 5 and 10 ) , the produced particles still contained intact Gag ΔNC-SRP19 . Quantification of the amount of 7SL RNA in these VLPs demonstrated that the presence of SRP19 enables the efficient packaging of 7SL RNA ( Fig . 5B ) . Importantly , these Gag ΔNC-SRP19 particles were able to package A3G efficiently , but not A3F . This observation is in agreement with the iCLIP data , which show that A3F does not preferentially bind to 7SL . Thus , while 7SL RNA is not required for the packaging of A3G into HIV-1 particles with an intact NC domain ( Fig . 3 ) , it is evidently able to promote packaging of A3G when selectively captured by VLPs . Our iCLIP data suggested that both A3G as well as A3F can bind to Y RNAs , small non-coding RNAs of ∼100 nucleotides that are components of RoRNPs [54 , 55] ( Fig . 1D ) . These RNAs are also incorporated into HIV-1 particles [35 , 49] . To investigate their role in A3G/F packaging , we first tried to “knock down” their levels using siRNA mediated silencing . Although we could reduce cellular Y RNA concentrations , the levels that were encapsidated remained unaltered , in line with previous findings where cellular Y RNAs were reduced following RNAi-mediated Ro60 depletion but packaging into MLV particles was unchanged [56] . We then used our established Gag ΔNC fusion system to ask if Y RNAs would be able to recruit A3G or A3F into viral particles . Ro60 was fused to Gag and 293T cells were co-transfected to express HA tagged A3G or A3F with wild type Gag , Gag ΔNC , Gag ΔNC-Ro60 , Gag ΔNC-Ro60 together with wild type Gag or with Gag ΔNC . VLPs were produced and immunoblotting showed that , although Gag fused to Ro60 alone did not produce detectable quantities of particles , mixed particles containing the Gag-Ro60 fusion protein and either Wt Gag or Gag ΔNC were formed ( Fig . 5C ) . Quantitative RT-PCR specific for Y3 RNA was performed on RNA extracted from these particles; both mixed Gag ΔNC + Gag ΔNC-Ro60 and wild type Gag particles each contained ∼7-fold more Y3 RNA compared to Gag ΔNC VLPs ( Fig . 5D ) , thus validating our approach . Analysis of the A3G/F contents demonstrated a strong restoration of packaging for the Gag ΔNC + Gag ΔNC-Ro60 mixed particles , though the levels did not match those noted with wild type Gag VLPs ( Fig . 5C ) . These results indicate that controlled packaging of specific RNA ligands of A3G or A3F can determine their encapsidation , presumably by bridging between assembling Gag fusion proteins and APOBEC3 proteins , further demonstrating that specific RNAs can promote the encapsidation of APOBEC3 proteins . Having shown that specific RNAs can recruit APOBEC3 proteins into assembling HIV-1 particles , we next asked whether this was a general property of packaged RNAs . To address this , a series of RNA binding domains ( RBDs ) of cell-encoded RNA binding proteins known to possess broad RNA binding capabilities were fused to Gag ΔNC as above . Specifically , we used the RBDs from two heterogeneous ribonucleoprotein ( hnRNP ) proteins: hnRNP C1 , that binds to uridine tracts of RNAs [39] , and hnRNP K , which is the prototypic protein for the KH RNA binding motif and has high affinity to poly ( C ) RNA [57 , 58] . We also fused the splicing factor SRSF2 that has a degenerate RNA binding sequence motif [59] and the double stranded RNA binding protein Staufen-1 [60 , 61] . These expression constructs were co-transfected with A3G/F into 293T cells together with vectors for wild type Gag or Gag ΔNC , and VLPs analysed by immunoblot ( Fig . 6 ) . Remarkably , all four RBDs readily rescued packaging of A3G and A3F in mixed virions with Gag ΔNC with ( generally ) similar efficiency as the wild type Gag or NC itself when reconnected to Gag ΔNC ( compare lanes 6 , 8 , 10 and 12 with 1 , 2 and 4; and 18 , 20 , 22 and 24 with 13 , 14 and 16 ) . Given that A3G and A3F exhibit very broad RNA binding characteristics ( Fig . 1 ) , we conclude that a multitude of such RNA substrates , if packaged , can serve to draw A3G/F into VLPs . Fig . 6 shows that A3G and A3F can be incorporated into VLPs when diverse RBDs are fused to Gag . This is consistent with our iCLIP data , demonstrating that A3G and A3F are able to bind to multiple diverse RNAs . One obvious question that is raised by these observations is whether A3G/F are preferentially encapsidated by assembling HIV-1 VLPs , or whether they are inevitably packaged as a natural consequence of their promiscuous RNA binding capabilities . To address this directly , we carefully quantified the ratios of A3G and A3F to RNA in the lysates of virus producing cells and the matching particle preparations ( Fig . 7 ) . Accordingly , 293T cells were transiently transfected with vectors expressing T7-tagged versions of A3G or A3F , as well as the wild type Gag expression vector . Cells were also transfected with an irrelevant plasmid to serve as a negative control . VLPs and cell lysates were collected at 48 h post transfection . VLPs were isolated through a continuous sucrose gradient and the fractions containing VLPs were identified by immunoblot ( S6 Fig . ) . Protein quantities were then determined against a standard curve of recombinant T7-A3G ( Fig . 7A and 7B ) , and RNA in cell lysates and VLPs were extracted and quantified by Qubit ( Fig . 7C ) . Culture supernatant from cells transfected with an irrelevant plasmid was used to assess background , and RNA was not detected in these samples ( threshold of detection , 20 pg/ml ) . Interestingly , the calculated ratios of A3G/F to RNA were similar in cells and in virus particles ( Fig . 7D , mean of 4 independent experiments ) . In other words , there is no evident enrichment of A3G or A3F in virions relative to virus-producing cells . Taking all our findings together leads us to conclude that the packaging of A3G and A3F into HIV-1 particles is driven by RNA binding , and that multiple/diverse RNAs can fulfil this role provided they are themselves packaged .
A3G and A3F are antiviral proteins that have to be packaged into newly synthesised retrovirus particles to exert their activity . However , the details of the packaging mechanism remain incompletely understood , though a preference for encapsidating newly synthesised A3G has been established [29–35 , 47 , 48 , 51] . Here , we used for the first time a high throughput method to identify the RNAs that these two proteins bind to in living cells productively infected with vif-deficient HIV-1 and in cell-free viral particles . We then systematically addressed which RNAs are either necessary or sufficient for A3G and A3F incorporation into budding virions . An important innovation with our study was the employment of a reference ( “non-RNA binding” ) protein during the iCLIP procedure , in this case GFP . This enabled us to identify A3G/F RNA ligands that are enriched over background RNA associations that are a presumed property of any protein ( Fig . 1; exemplified here by the generation of iCLIP libraries from GFP-containing cells ) . While some preferential binding to certain classes of RNAs was apparent , it was nonetheless evident that the patterns of A3G/F binding were mostly non-discriminatory . We speculate that such promiscuity in RNA binding could be explained by the capacity of A3G/F to interact with myriad RNA sequences that are available for binding simply because they are not already occupied by other proteins . We investigated in detail some specific RNAs that are found in retroviral particles [37] . We found that A3G and A3F do not require HIV-1 gRNA or 7SL for packaging into HIV-1 particles ( Fig . 2 , 3 and 4 ) . However , using our Gag ΔNC fusion assay , we could show that A3G ( but not A3F ) can utilise 7SL to be incorporated into viral particles ( Fig . 5 ) . These observations correlated well with our iCLIP data , where it was shown that A3F does not bind to 7SL . In accordance with previous studies [30–33] , we observed that A3G and A3F are not packaged into Gag ΔNC VLPs . Importantly , these particles can clearly incorporate APOBEC3 proteins when a variety of unrelated RNAs are recruited into assembling virions via fusions of Gag ΔNC to a series of unrelated RBDs ( Fig . 5 and 6 ) . In other words , under experimental conditions , A3G and A3F can be packaged into HIV-1 particles via interactions with diverse and unrelated RNAs . Our data also show that 80% of the RNA sequences that A3G/F bind to inside vif-deficient ( but otherwise normal ) viral particles are of viral origin . If we assume that this distribution correlates with the binding of A3G/F to RNAs that are being packaged during particle production , our data imply that viral genomic RNA ordinarily mediates A3G/F packaging . An alternative possibility is that non-viral RNAs recruit A3G/F into particles , but that A3G/F then release these RNAs and bind to gRNA following particle formation and release . Lastly , fastidious quantification of A3G/F and total RNA levels in cells and budded virions revealed that viral particles are not enriched for APOBEC3 protein content ( Fig . 7 ) . While RNA binding is clearly required for APOBEC3 protein packaging , these observations indicate that there is no selectivity for engaging RNAs that are destined for incorporation into assembling viruses . Accordingly , we speculate that cytoplasmic APOBEC3 proteins exploit their relatively non-specific RNA binding capabilities to patrol the cytosol and bind to unoccupied sites on RNAs . In the context of cellular RNAs , this may account for the pronounced accumulation of APOBEC3 proteins in RNA-rich microdomains such as P-bodies and stress granules [23–26] . For viruses with RNA genomes , such as retroviruses , this can result in encapsidation into newly formed viral particles . Given the DNA editing function of APOBEC3 proteins , viruses that have RNA genomes and replicate via DNA intermediates—namely retroviruses and hepadnaviruses—will be susceptible to hypermutation and inhibition [4 , 5 , 9] . Indeed , we propose that the relatively non-specific RNA binding characteristics of A3G/F render these proteins well suited to the inhibition of a wide variety of viral and transposon targets . Moreover , this feature may further ensure that viral sequence variation , a noted hallmark of HIV-1 , will not afford a means of escape from APOBEC3-mediated restriction: perhaps this underlies the evolution of an entirely different ( protein-based ) evasion mechanism , namely Vif-induced protein degradation ?
cDNAs encoding SRP19 , GFP , A3G or A3F were cloned between the XbaI and BamHI sites of pCGTHCFFLT7 [62] . DNA fragments encoding T7-tagged derivatives of GFP , A3G or A3F were cloned between the XhoI and EcoRI sites of the retrovirus vector , pCMS28 [26] . A GFP-containing fragment with a GST linker sequence at its 3’-end was cloned between the EcoRI and XhoI restriction sites of pCMS28 , and A3G or A3F cDNAs were then inserted using the NotI and XhoI sites . Plasmids expressing HA-tagged A3G and A3F were previously described [50] . vif-deficient HIV-1NL4-3 [27] and HIV-1IIIB [63] strains were used where indicated . The wild type Gag-Pol vector , pCMS446 , was generated by inserting the HIV-1 5’ UTR-Gag-Pol , Protease activity inactivated ( nucleotides 455–5096 from the HV-1HXB2 isolate [GenBank: K03455 . 1] [64] ) fragment into pcDNA3 . 1 ( Invitrogen ) that contains the HIV-1 RRE [65] . pGag was generated by deleting the Pol sequence 3’ to the Gag stop codon . pGag ΔNC was generated by cloning the analogous 5’ UTR-Gag-Pol fragment from the Zwt-p6 proviral construct [45] into the same RRE-containing vector . SL2 and 3 were deleted from pGag by overlapping PCR using the primers 5′-AGGGGCGGCGACTGGTGAGAGATGGGTGCGAGAGCGTCAGTATTAAGC-3′ and 5′-TGACGCTCTCGCACCCATCTCTCACCAGTCGCCGCCCCTCGCCTCTTGC-3′ to generate pGag ΔΨ [46] . pGag ΔNC NB was created by inserting NheI and BamHI restriction sites in frame 5’ to the stop codon of Gag in pGag ΔNC . NC ( HIV-1HXB2 strain ) , GFP , SRP19 , Ro60 , hnRNP C1 ( amino acids 1–104 ) , hnRNP K ( amino acids 38 to 464 with amino acids 323–338 deleted ) , SRSF2 ( amino acids 2–93 ) and Staufen 1 cDNAs were inserted into pGag ΔNC NB using the NheI and BamHI sites . 293T and HeLa cells were obtained from the American Tissue Culture Collection ( ATCC ) . TZM-bl cells were obtained through the NIH AIDS Reagents Repository Program ( ARRP ) . These cell lines were cultured in Dulbecco’s modified Eagle’s medium ( Invitrogen , UK ) supplemented with 10% foetal bovine serum and 1% penicillin/streptomycin . CEM-SS T cells , from ARRP , were cultured in Roswell Park Memorial Institute 1640 medium ( Invitrogen , UK ) supplemented with 10% foetal bovine serum and 1% penicillin/streptomycin . Stable CEM-SS T cell lines were generated by standard retroviral transduction using MLV-based vectors expressing GFP , GFP-A3G GFP-A3F , T7-GFP , T7-A3G or T7-A3F and selected with 1 μg/ml puromycin . 293T cells stably expressing HA-tagged A3G or A3F were generated by transduction with MLV based vectors expressing the proteins of interest and selection with 1 μg/ml puromycin . Expression levels of the A3 proteins were assessed by immunoblot using rabbit polyclonal sera specific for A3G [66] or A3F [67] for primary detection . iCLIP has been described in detail previously [39] . Briefly , CEM-SS T cells stably expressing the proteins of interest were infected with vif-deficient HIVIIIB . 48 h later , a sample was collected to assess infection by intracellular p24Gag staining and flow cytometry , confirming that at least 80% of cells were productively infected . The supernatant was collected , filtered through a 0 . 45 μm pore filter , and viruses isolated through a 20% sucrose cushion ( wt/vol ) at 21000 × g for 2 h at 4°C and resuspended in PBS . Cells were collected , washed 6 times with PBS and resuspended in PBS . Cells and viruses were then radiated with 400 mJ/cm2 using a Stratlinker 2400 . Cells were pelleted by centrifugation and the supernatant discarded . Cells and viruses were then resuspended in 1 ml of lysis buffer ( 50 mM Tris-HCL , pH 7 . 4; 100 mM NaCl; 1% NP-40; 0 . 1% SDS; 0 . 5% sodium deoxycholate and protease inhibitor ) and sonicated . 0 . 16 μg of RNase A were added to High RNase samples and 0 . 04 ng to the other samples . Tubes were incubated at 37°C for 3 min and added to protein G dynabeads previously incubated with anti-T7 antibody ( Novagen ) or anti-GFP antibody ( Roche ) . The RNAs were dephosphorylated using Shrimp alkaline phosphatase ( Promega ) and a pre-adenylated linker was ligated to the 3’-end of RNAs on beads . RNAs were radiolabeled with P32-ϒ-ATP and separated using SDS-polyacrylamide gel electrophoresis , electrophoretically transferred to nitrocellulose and visualised on film . Pieces of the membrane containing the RNAs of interest were excised and resuspended in PK buffer ( 100 mM Tris-HCl pH 7 . 5 , 50 mM NaCl , 10 mM EDTA ) containing 2 mg/ml proteinase K . RNAs were isolated with phenol/chloroform ( Ambion ) and precipitated with 2 . 5 volumes of 100% ethanol , 0 . 1 volumes of sodium acetate ( 3 M , pH 5 . 5 ) and 0 . 5 μl of glycoblue . RNAs were then pelleted and reverse transcribed using barcoded primers . cDNAs were run on a TBE-urea polyacrylamide gel and products ranging from 70–85 , 85–110 and >110 base pairs were excised . Nucleic acids were extracted and circularised . A primer that anneals to the linker previously ligated to the RNAs was used to create a double stranded region and this DNA was digested with BamHI . cDNA was then amplified by PCR and sequenced on one lane of an Illumina GA2 flow cell with 50 nucleotides run length . Data is available at ArrayExpress with the accession number E-MTAB-2700 . Before mapping reads , adapter sequences were removed , and the barcodes for each sample within each library were used to identify which sequence was immunoprecipitated from each protein . Mapping of the reads was performed against the human genome ( version Hg19/ GRCh37 ) and HIV-1IIIB ( GenBank ID EU541617 ) genome using bowtie [41] . Reads that aligned to a single position on the human genome or HIV-1 genome with at most two mismatches were considered for analysis . Genomic annotations were then assigned based on gene annotations provided by Ensembl ( v59 ) . Reads were also aligned to human repeat sequences using repeat masker [42] and a database of consensus sequences provided by the software . A maximum of 3 mismatches were allowed . HIV-1 virions were produced by transfecting 293T cells with vif-deficient pNL4-3 or pIIIB using polyethyleneimine ( PEI ) . Virus were then harvested 48 h later and filtered through a 0 . 45 μm pore filter . Lentiviral vectors were produced in 293T cells by transfecting the p8 . 91 packaging plasmid , a lentiviral vector and the VSVg envelope plasmid pMDG2 . 1 [43] at a ratio of 2:2:1 using PEI . Vectors were harvested 48 h after transfection and filtered . Viral particles were quantified by p24Gag enzyme linked immunosorbent assay ( Perkin Elmer ) . VLPs were produced by transfecting the packaging plasmid of interest and the Rev expression vector , at a ratio of 2:1 . Wherever stated , HA- or T7-tagged GFP , A3G or A3F vectors were co-transfected at a ratio of 1:5 to the packaging plasmid . The supernatant was collected 48 h later , filtered through a 0 . 45μm pore filter and isolated through a 20% sucrose cushion ( wt/vol ) at 21000 × g for 2 h at 4°C . Cells and viral pellets were lysed in radioimmunoprecipitation assay ( RIPA ) buffer ( 50 mM Tris-HCl pH 7 . 4 , 100 mM NaCl , 1 mM MgCl2 , 1% NP-40 , 0 . 1% SDS , 0 . 5% sodium deoxycholate ) . T7-tagged A3G was purified as described before [68] . Expression , VLP production and packaging of APOBEC3 proteins were assessed by standard immunoblot using anti-HA ( mouse monoclonal; 12CA5 ) or anti-T7 ( Novagen ) and anti-p24Gag ( mouse monoclonal; p24-2 [69] ) antibodies and detected and quantified by Li-cor Odyssey infrared imaging using IRDye800CW or IRDye680LT-labeled secondary antibodies . VLPs produced for RNA and protein quantification were filtered , layered over a continuous sucrose gradient ( 60–20% ) and centrifugated at 150 , 000 × g for 1h15min at 4°C . 1mL fractions were collected and centrifugated at 21000 × g for 2 h at 4°C . The supernatant was removed and the pellet resuspended in RIPA buffer . The fractions containing p24Gag were identified by immunoblot . The fraction with highest level of p24Gag was then used to quantify the packaged A3G/F by immunoblot using a standard curve of purified T7-A3G . The remainder of the fraction was used to extract RNA using a microRNA extraction kit ( Promega ) . Total RNA was then quantified using the Qubit RNA HS assay kit ( Life Technologies ) , following the manufacturer’s instructions . RNA was extracted from viral particles using Tri Reagent LS ( Sigma ) according to the manufacturer’s instructions . 0 . 5 μg of total RNA was used to synthesise cDNA with the high-capacity cDNA reverse transcription kit ( Applied Biosystems ) and random primers using the manufacturer’s protocol . qPCR was then performed using the primers 5′-TAACTAGGGAACCCACTGC-3′ , 5′-GCTAGAGATTTTCCACACTG-3′ and the probe 5′-6-carboxyfluorescein [FAM]-ACACAACAGACGGGCACACACTA-6-carboxytetramethylrhodamine [TAMRA]-3′ to detect HIV-1 gRNA; SYBR Green ( Applied Biosystems ) was used to detect 7SL with the primers 5′-GGGCTGTAGTGCGCTATGC-3′ and 5′-CCCGGGAGGTCACCATATT-3′ , and to quantify hY3 RNA with the primers 5′-GGCTGGTCCGAGTGCAGTG-3′ and 5′-AAAGGCTAGTCAAGTGAAGCAGTGG-3′ . | APOBEC3 proteins are cell-encoded restriction factors that counteract infections , particularly by retroviruses such as HIV-1 , and retrotransposons . When packaged into HIV-1 particles , APOBEC3G and APOBEC3F both inhibit reverse transcription and induce destructive hypermutation in viral DNA . The mechanism of APOBEC3 virion packaging awaits elucidation , though a dependency on RNA binding has been established . Here , we employed a cross-linking and next generation sequencing approach to determine which RNAs are bound to A3G and A3F in HIV-1 infected cells . We show that both proteins bind to multiple different RNAs , including viral RNA as well as cellular coding and non-coding RNAs , with relatively little evidence of selectivity . We then developed a complementation assay to address the diversity of RNAs that can act as substrates for A3G/F virion packaging . Consistent with the RNA binding profiles , many RNAs can promote packaging provided that those RNAs are , themselves , packaged . These observations suggest that APOBEC3 packaging lacks selectivity and is driven simply by the non-specific RNA binding capabilities of these proteins . We speculate that this model accounts for the broad range of retro-elements that are susceptible to repression by individual APOBEC3 proteins , and also that such substrates cannot escape APOBEC3-mediated inhibition through sequence variation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Promiscuous RNA Binding Ensures Effective Encapsidation of APOBEC3 Proteins by HIV-1 |
In sub-Saharan Africa , over 200 million individuals are estimated to be infected with urinary and intestinal schistosomiasis . In a bid to lay a foundation for effective future control programme , this study was carried out with the aim of assessing the diagnostic efficacy of some questionnaire-based rapid assessment indices of urinary schistosomiasis . A total number of 1 , 363 subjects were enrolled for the study . Questionnaires were administered basically in English and Hausa languages by trained personnel . Following informed consent , terminal urine samples were collected between 09:40 AM and 2:00 PM using clean 20 ml capacity universal bottles . 10μl of each urine residue was examined for the eggs of S . haematobium using x10 objective nose of Motic Binocular Light Microscope ( China ) . The average age ± Standard Deviation ( SD ) of school children examined was 15 . 30 ± 2 . 30 years and 40 . 87% were females . The overall prevalence and geometric mean intensity of S . haematobium infection were 26 . 41% ( 24 . 10─28 . 85 ) and 6 . 59 ( 5 . 59─7 . 75 ) eggs / 10 ml of urine respectively . Interestingly , a questionnaire equivalence of the prevalence obtained in this survey was 26 . 41% ( 24 . 10─28 . 85 ) for Rapid Assessment Procedure based on self-reported blood in urine . The results of correlation analyses demonstrated significant associations between the prevalence of S . haematobium infection and contact with potentially infested open water sources ( r = 0 . 741; P = 0 . 006 ) . By regression model , cases of respondents with self-reported blood in urine are expected to rise to 24 . 75% if prevalence of the infection shoots up to 26 . 5% . The best RAP performance was obtained with self-reported blood in urine . Based on the overall prevalence value , the study area was at a “moderate-risk” of endemicity for urinary schistosomiasis . Chemotherapeutic intervention with Praziquantel , the rationale behind rapid assessment procedure for schistosomiasis , has been recommended to be carried out once in every 2 years for such communities .
Schistosomiasis , a water-borne neglected tropical disease ( NTD ) , has been reported as the second most prevalent parasitic disease after malaria [1] . The causative agent of human schistosomiasis is a digenetic trematode blood fluke of the genus Schistosoma with a complex , indirect life cycle involving different species of freshwater snails [2 , 3] . These snails serve as intermediate hosts to S . haematobium , S . intercalatum , S . japonicum , S . mansoni and S . mekongi which parasitize humans [4 , 5] . Humans become infected when the infective larvae mechanically penetrate their skin after contact in fresh water bodies located in environment characterized by poor hygiene and sanitation [6] . The distribution of schistosomiasisis is more abundant in the African region with 42 countries endemic for the infection . In sub-Saharan Africa , over 200 million individuals are estimated to be infected with urinary and intestinal schistosomiasis [7] , with approximately 393 million people at risk of infection from Schistosoma mansoni , of which 54 million are infected while 436 million people are at risk of S . haematobium infection and 112 million are infected [8] . The most widely used approach for the diagnosis in endemic settings is the detection of schistosome eggs in either stool or urine specimens by light microscopy . The first step in targeting health interventions is to map the disease geographically and rank it according to the risk of infection and morbidity [4] . The use of geographical information systems highlighted the scarcity of data in endemic region such as Africa , and emphasizes the need for a rapid , non-invasive and inexpensive epidemiological assessment tool that can be fully integrated within existing administrative systems [9] . Simple school questionnaires were developed for S . haematobium and has since been validated in many ecological , epidemiological , and sociocultural settings across sub-Saharan Africa . It is well accepted and operationally feasible . It is faster and less expensive than the standard parasitological diagnosis [10] . The basis of this method was that , being a chronic disease , the only clear symptom school-age children could observe and easily remember was the presence of blood in urine [11] . They build directly on a community’s perception of disease , involve the active participation of teachers and schoolchildren , and represent a first step towards involving the community in control activities . Macrohaematuria , microhaematuria , and proteinuria are assessed by reagent strips [12] . In a bid to lay a foundation for effective future control programme for urinary schistosomiasis in the study area , we embarked on this cross-sectional survey with the aim of assessing the diagnostic efficacy of some questionnaire-based rapid assessment indices of urinary schistosomiasis .
Written ethical clearance to conduct the survey was issued by the Ethical Committee of the Katsina State Ministry of Education , Dutsin-Ma Zonal Office . School heads and students gave oral informed consent to participate after appropriate briefing on the background and objectives of the study . Oral assent , aided by an interpreter , was provided by students after appropriate briefing on the background and objectives of the study . They demonstrated this by willingly providing their names for a written documentation during the interview . Information obtained from the subjects was kept confidential . Noteworthy is the fact that formal consent could not be obtained from the parents and guardians of the subjects partly because the cultural and religious situation of the study area was volatile . To buttress this point , there is history of attempted physical attack on healthcare officials in the study area . The study was undertaken in twelve ( 12 ) high schools from six ( 6 ) communities of Dutsin-Ma and Safana ( 809 km2 ) Local Government Areas ( LGAs ) of Katsina State , Northwestern Nigeria ( see Fig 1 ) . As at 2006 National Census , both neighboring LGAs were inhabited by 353 , 450 people [13] . Noteworthy is the fact that the study area , characteristically sandy with a rocky terrain ( typical of western upland plateau ) is drained by different water bodies , the largest being Zobe Dam . The study covered a bio-geographical Sudan Savannah area characterized by low to moderate endemicity for urinary schistosomiasis . By Agro-ecological classification , it belongs to the Sudan savanna vegetation zone of Nigeria [14] . The main economic activity there is farming , with millet as the subsistence crop . The predominant ethnic groups , Hausa and Fulani , complement crop production with trading and nomadism . Both LGAs have a mean annual rainfall and temperature less than 800mm and 30°C respectively [15] . A cross-sectional study design was adopted in this present survey . By estimating the prevalence of Schistosoma haematobium at 30% with power and sampling error of 90% and 5% respectively , a sample size of 912 was obtained . This calculation was made based on the standard of World Health Organization for sample size estimation [16] . Simple random sampling technique was employed to select the total number of 1 , 363 secondary school students who participated in the study between May and August , 2015 . This sample size accounted for effect size and any anticipated non-response . School based questionnaire with questions relating to the knowledge of urinary schistosomiasis , sources of water , and local name associated with the disease was used in the survey . For information on urinary schistosomiasis , question asked was: “Do you know any student in this school who reportedly pass blood-stained urine ? ” For a positive response , the next question was: “What is the local language for this condition ? ” The individual questionnaire was designed , among other things , to obtain responses from subjects on water contact activities ( fetching , swimming and play in shallow water ) , experiences of itching , haematuria , and pain while urinating . To elicit responses for some of these experiences , interviewees were asked the following questions: “Have you ever experienced: ( i ) pains while urinating ? ( ii ) blood in your urine ? ” Whenever a response was positive for the latter , each subject was further questioned: “How have you been treating it ? ” Because the study population was divided along the language lines of English and Hausa , questionnaires were administered accordingly by trained personnel which included the investigators and selected teachers from the participating schools . In the study area , S . mansoni is co-endemic with S . haematobium . Consequently , some interviewees suffered a mixed infection . Consequently , there was unusual discovery of the eggs of the former ( S . mansoni ) in a few urine samples that were as well positive for the eggs of the latter ( S . haematobium ) . They were distinguished using their unique identification keys , that is , the possession of a lateral and terminal spine by the eggs of S . mansoni and S . haematobium respectively [18] . However , it has been reported that under special epidemiological settings with a very high prevalence of urinary schistosomiasis but a very low prevalence of intestinal schistosomiasis , eggs of S . mansoni do occur in urine [19] . 10μl of each urine residue was examined for the eggs of S . haematobium using x10 objective nose of Motic Binocular Light Microscope ( China ) . Each average egg count was recorded as number of eggs per 10 ml of urine sample using a multiplier factor of two . While prevalence was grouped into low ( ˂ 10% ) , moderate ( ≥ 10%–49% ) and high ( ≥ 50% or more ) [20] , intensity of infection was categorized into light ( ˂ 50 eggs / 10 ml of urine ) and heavy ( ≥ 50 eggs / 10 ml of urine ) infections according to standard method [4] . All data obtained from the survey were entered into Microsoft Excel 2010 ( USA ) and analysed using SPSS 15 . 0 ( Chicago , USA ) . The relationships between Rapid Assessment and Parasitological Indices were assessed using Spearman’s rank correlation and linear regression analyses . The prevalence for S . haematobium infection , as well as for micro-haematuria ( shown in Figs 2–7 ) was calculated on a school-level basis . These school-level estimates were correlated with the following rapid assessment indicators [i . e . water contact ( in % ) , self-reported blood in urine ( in % ) , itching , urethral pain , and combined RAPs based on water contact/ self-reported blood in urine/ itching/ urethral pain ( in % ) , and water contact/ itching ( in % ) ] using the Spearman's rank correlation coefficient test . Both school-level prevalences of S . haematobium infection and micro-haematuria were used in a linear regression model as continuous dependent outcomes ( in % ) and were controlled for , in a univariable manner , each rapid assessment indicator . Prior to reporting our findings , necessary diagnostic tests were performed . Normality was achieved and no key model assumptions were violated . Statistical significance was considered at 95% confidence level ( CL ) with a P value of 0 . 05 . The diagnostic performances of indices for identifying “low risk” , “moderate risk” or “high risk” schools were assessed by calculating sensitivities , specificities , and positive and negative predictive values .
In the six ( 6 ) communities surveyed , 1 , 363 students with ages ranging from 10–25 years were interviewed and examined . The average age ± Standard Deviation ( SD ) of school children examined was 15 . 30 ± 2 . 30 years and 40 . 87% were females . Of the total number interviewed , 360 respondents were infected . It is worthy of note that the prevalence and geometric mean of S . haematobium egg counts in the study communities ranged from 15 . 59% to 37 . 28% and 3 . 27 to 16 . 42 eggs / 10 ml of urine respectively ( see Table 1 and Fig 2 ) . Hence , the overall prevalence and geometric mean intensity of S . haematobium infection were 26 . 41% ( 24 . 10–28 . 85 ) and 6 . 59 ( 5 . 59–7 . 75 ) eggs / 10 ml of urine respectively . The arithmetic mean intensity of infection was 27 . 90 ( 19 . 55–36 . 25 ) eggs / 10 ml of urine . Males recorded a higher prevalence [40 . 07% ( 36 . 69–43 . 56 ) ] and geometric mean intensity of S . haematobium infection [7 . 52 ( 6 . 33–8 . 94 ) eggs / 10 ml of urine] . Furthermore , males were 9 times [COR ( 95% CI ) : 9 . 39 ( 6 . 54–13 . 49 ) ] more likely to be infected with the cercariae of S . haematobium ( see Table 1 ) . Spearman’s rank correlation ( see Table 2 and Figs 3–7 ) demonstrated significant associations between: S . haematobium prevalence and contact with potentially infested open water sources ( r = 0 . 741; P = 0 . 006 ) ; prevalence of micro-haematuria and contact with potentially infested open water sources ( r = 0 . 643; P = 0 . 024 ) ; self-reported blood in urine and prevalence of S . haematobium infection ( r = 0 . 629; P = 0 . 028 ) ; prevalence of micro-haematuria and the combined RAP of water contact , self-reported blood in urine and painful urination ( r = 0 . 741; P = 0 . 006 ) and prevalence of micro-haematuria and the combined RAP of self-reported blood in urine and painful urination ( r = 0 . 727; P = 0 . 007 ) . Significant associations ( see Table 2 ) were also obtained with: prevalence of S . haematobium infection and the combined RAP of water contact , and itching ( r = 0 . 636; P = 0 . 026 ) ; and prevalence of S . haematobium infection and the combined RAP of self-reported blood in urine , and painful urination ( r = 0 . 587; P = 0 . 045 ) . The questionnaire equivalence of the prevalence revealed a similar prevalence of 26 . 41% ( 24 . 10–28 . 85 ) for RAP based on self-reported blood in urine ( see Fig 5 ) . However , no significant association was recorded between the prevalence of light infection intensity and water contact ( r = 0 . 573; P = 0 . 051 ) ; school community egg load and water contact ( r = 0 . 364; P = 0 . 245 ) ; prevalence of micro-haematuria and the combined RAP of water contact , and itching ( r = 0 . 357; P = 0 . 255 ) . Noteworthy was the fact that the single RAP index of pain while urinating recorded the poorest performance since it had no statistically significant association ( P ˃ 0 . 05 ) with any Parasitological indices ( see Table 2 ) . We discovered that , for a 26 . 5% rise in contact with infested water sources , the: prevalence of S . haematobium infection will increase by 1% ( using the equation: y = 0 . 6667 * x– 16 . 6667 shown in Fig 3 ) ; prevalence of microhaematuria will only increase by 3 . 6% ( see Fig 4 ) . However , cases of respondents with self-reported blood in urine are expected to rise to 24 . 75% if prevalence of the infection shoots up to 26 . 5% ( see Fig 5 ) . Meanwhile , using equation 0 . 6 * x + 9 , the prevalence of microhaematuria is anticipated to increase by 24 . 9% with a similar 26 . 5% rise in the number of respondents with combined experiences of contact with infested water sources , self-reported blood in urine and painful urination ( see Fig 6 ) . Similarly , for 26 . 5% increase in the number of respondents with combined experiences of contact with infested water sources and self-reported blood in urine , the prevalence of microhaematuria is expected to remain 24 . 9% ( see Fig 7 ) . All RAPs showed sensitivities which ranged from 33 . 33–98 . 08% . However , RAPs based on water contact , pain while urinating and self-reported blood in urine respectively recorded high sensitivities in descending order of magnitude . Meanwhile , specificity ranged from 45 . 46–94 . 32% . The least value was recorded for RAP based on water contact while the highest was obtained with the combined RAP of water contact , self-reported blood in urine and pain while urinating . In summary , the best RAP performance was obtained with self-reported blood in urine which had a sensitivity of 60 . 28% and specificity of 91 . 43% . Co-incidentally , this RAP also recorded the best combined values for Positive Predictive Value ( PPV ) and Negative Predictive Value ( NPV ) ( Table 3 ) .
Currently , the most widely used clinical approach to determining the prevalence and intensity of infection due to S . haematobium is manual egg count by means of urine microscopy . Our data showed that by this gold standard [17 , 21] , the overall prevalence and geometric mean intensity of urinary schistosomiasis were 26 . 41% ( 24 . 10–28 . 85 ) and 6 . 59 ( 5 . 59–7 . 75 ) eggs / 10 ml of urine respectively , with males being 9 times [COR ( 95% CI ) : 9 . 39 ( 6 . 54–13 . 49 ) ] more likely to be infected compared to females . It is pertinent to state that this gold standard employed vis a vis the rapid assessment procedures in this cross-sectional survey was rather cumbersome and frustrating . However , it is of interest that a questionnaire equivalence of the prevalence obtained in this survey revealed a similar prevalence of 26 . 41% ( 24 . 10–28 . 85 ) for RAP based on self-reported blood in urine . Again , findings in this survey showed that the best RAP performance was obtained with self-reported blood in urine with a sensitivity and specificity of 60 . 28% and 91 . 43% respectively . To corroborate the reliability of this RAP index , a report from Yemen shows that 72 . 2% of respondents who suffered heavy intensity of infection with Schistosoma haematobium visibly experienced blood in their urine [17] . Furthermore , high sensitivity and specificity have previously been reported in other urinary schistosomiasis endemic settings . For example , in a similar survey conducted in southwestern Nigeria , a specificity of almost 100% was obtained [22] . In northern Ghana , self-reported haematuria showed a sensitivity of 53% and a specificity of 85% [23] . Co-incidentally , in this survey , self-reported blood in urine also recorded the best results for Positive Predictive Value ( 71 . 62% ) and Negative Predictive Value ( 86 . 51% ) . The latter simply means that , of all the subjects who tested negative for urinary schistosomiasis by microscopic examination , 86 . 51% were actually negative while 13 . 49% were positive , going by questionnaire-based rapid means of assessment using self-reported blood in urine ( macro-haematuria ) . In addition , when this RAP was combined with water contact and pain while urinating ( dysuria ) , a higher Positive Predictive Value ( 75 . 64% ) was obtained . That is , this combination unmasked 4 . 02% of more subjects that were infected compared to the single RAP . This is indeed a cost-effective means of improving on the quality of data obtained in urinary schistosomiasis research . Moreover , the result of correlation analysis demonstrated a statistically significant association between self-reported blood in urine and prevalence of S . haematobium infection ( r = 0 . 629; P = 0 . 028 ) . Better still , when self-reported blood in urine was combined with RAP based on painful urination , a stronger association ( r = 0 . 727; P = 0 . 007 ) was obtained between them and micro-haematuria . Meanwhile self-reported blood in urine , micro-haematuria and painful urination ( dysuria ) have been previously identified as morbidity markers of urinary schistosomiasis [17 , 24 , 25] . The implication of these is that we can use a RAP based on self-reported blood in urine to predict the parasitological prevalence of urinary schistosomiasis in either moderate or high endemic settings . It could as well produce a reliable result in areas where biomedical reagent strips are not available [26] . Previous studies carried out in some African countries ( Cameroon , Congo , Democratic Republic of the Congo , Ethiopia , Malawi , Zambia and Zimbabwe ) also showed that macro-haematuria had a very good diagnostic ability to detect “high-risk” schools while ruling out “low-risk” ones [10] . In a survey conducted in the Tanga region of the United Republic of Tanzania , average of 75% school-age children were reportedly accurate in their self—diagnosis of urinary schistosomiasis using the presence of blood in urine ( haematuria ) as a rapid diagnostic procedure [27] . To the best of our knowledge , contact with potentially infested , open , and unwholesome water sources is not in use as a rapid assessment indicator for urinary schistosomiasis . However , in this survey , the result of correlation analysis demonstrated significant association between prevalence of S . haematobium infection and contact with potentially infested open water sources ( r = 0 . 741; P = 0 . 006 ) . This did not come as a surprise because urinary schistosomiasis has been constantly reported as a water-borne disease [10 , 22 , 28] . When employed as a single RAP index in this present survey , it recorded a very high sensitivity ( 98 . 06% ) and Negative Predictive Value ( 98 . 49% ) but low values for the duo of specificity ( 45 . 46% ) and Positive Predictive Value ( 39 . 22% ) . Meanwhile , when combined with self-reported blood in urine and dysuria , its sensitivity markedly reduced to almost half ( 49 . 17% ) while its Negative Predictive Value dropped to 83 . 79% . Its specificity ( 94 . 32% ) and Positive Predictive Value ( 75 . 64% ) , however , approximately doubled . It also demonstrated a statistically significant association ( r = 0 . 643; P = 0 . 024 ) with the prevalence of micro-haematuria . More interestingly , when combined with other RAPs based on self-reported blood in urine and painful urination , a stronger relationship was obtained with the prevalence of micro-haematuria ( r = 0 . 741; P = 0 . 006 ) . The implication of these findings is that when subjects are carefully interviewed as regards their water contact activities , to a large extent , a good rapid diagnostic result for urinary schistosomiasis could be obtained . This is a good news to all high risk endemic settings where diagnostic kits and microscopes are very short in supply . Bearing in mind that indiscriminate mass chemotherapeutic intervention with Praziquantel is indeed not harmful [27] , on the basis of this finding , it could be achieved successfully without anticipating any severe adverse reactions . In the context of Schistosomiasis Elimination Strategy and Potential Role of a Vaccine in Achieving Global Health Goals co-sponsored by Bill and Melinda Gates Foundation and the National Institute of Allergy and Infectious Diseases [29] , self-reported blood in urine , as a single or combined RAP index , could play a major diagnostic role in unraveling new endemic foci for mass drug administration . As it stands , self-reported blood in urine will continue to be a relevant rapid diagnostic RAP index until schistosomiasis is eradicated . This survey is , however , subject to some limitations . To start with , adults were not included . Therefore , the result reported here may not be applicable to the whole population of the study area because previous findings have shown that the diagnostic efficacy of haematuria as a RAP index is inversely proportional to the age of subjects but stable in teenage children [30] . An extrapolation is only applicable after a painstaking assessment of the school enrollment , and the overall socio-cultural and epidemiological condition of the study area [11] . Moreover , previous report has shown that the accuracy of macro-haematuria as a yardstick for rapid assessment of urinary schistosomiasis may be better when a day-to-day variation in eggs excretion is considered [31] . However , this survey did not capture a serial assessment of each subject for macro-haematuria . Since report has shown that the identification of schools and communities endemic for schistosomiasis is a key issue in any control programme [32] , the high performing RAPs in this study could be employed to discover more endemic foci in the existence of ongoing regular Mass Drug Administration ( MDA ) in Nigeria .
Based on this overall prevalence value of 26 . 41% ( 24 . 10–28 . 85 ) obtained in this survey , it is obvious that the study area was at a “moderate-risk” of endemicity for urinary schistosomiasis [4 , 20] . Meanwhile , chemotherapeutic intervention with Praziquantel , the rationale behind rapid assessment procedure for schistosomiasis , has been recommended to be carried out once in every 2 years for such communities [20] . Although , water contact was found to have a good diagnostic efficacy , The best RAP in this survey was self-reported blood in urine . Both RAPs performed better when combined with other RAP indices . | Schistosomiasis is a water-borne neglected infectious disease of poverty that has consistently plagued over 200 million helpless inhabitants of the tropics , particularly , sub-Sahara Africa . Under the auspices of different nomenclatures and affiliations , many control programmes based on Praziquantel have been inaugurated over the past decades . Bearing in mind that globally , schistosomiasis exist in focal pockets within peri-urban and rural settings , and the lowest cost of a generic 600-mg tablet is approximately US$ 0 . 08 , it is imperative to focus control resources more on high risk settings . In order to identify such settings , rapid means of mapping schistosomiasis prevalence are carried out either with questionnaires or biomedical testing with reagent strips . Rapid assessment procedure for urinary schistosomiasis , the focus of this study , builds substantially on the perception of respondents about the disease through visible blood in their urine ( where applicable ) . We conducted this present survey in 6 communities of Katsina State , northwestern Nigeria by interviewing and examining the urine of 1 , 363 high schools students for the eggs of Schistosoma haematobium . A unique discovery in this survey was that contact with unwholesome water bodies , where properly defined , was significantly associated with urinary schistosomiasis , both as a single index and when combined with itching experience . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion",
"Conclusion"
] | [
"schistosoma",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"education",
"helminths",
"sociology",
"tropical",
"diseases",
"social",
"sciences",
"parasitic",
"diseases",
"animals",
"parasitology",
"urine",
"physiological",
"processes",
"negl... | 2017 | Rapid mapping of urinary schistosomiasis: An appraisal of the diagnostic efficacy of some questionnaire-based indices among high school students in Katsina State, northwestern Nigeria |
Obesity is a worldwide health problem that is closely linked to many metabolic disorders . Regular physical exercise has been found to attenuate the genetic predisposition to obesity . However , it remains unknown what kinds of exercise can modify the genetic risk of obesity . This study included 18 , 424 unrelated Han Chinese adults aged 30–70 years who participated in the Taiwan Biobank ( TWB ) . A total of 5 obesity measures were investigated here , including body mass index ( BMI ) , body fat percentage ( BFP ) , waist circumference ( WC ) , hip circumference ( HC ) , and waist-to-hip ratio ( WHR ) . Because there have been no large genome-wide association studies on obesity for Han Chinese , we used the TWB internal weights to construct genetic risk scores ( GRSs ) for each obesity measure , and then test the significance of GRS-by-exercise interactions . The significance level throughout this work was set at 0 . 05/550 = 9 . 1x10-5 because a total of 550 tests were performed . Performing regular exercise was found to attenuate the genetic effects on 4 obesity measures , including BMI , BFP , WC , and HC . Among the 18 kinds of self-reported regular exercise , 6 mitigated the genetic effects on at least one obesity measure . Regular jogging blunted the genetic effects on BMI , BFP , and HC . Mountain climbing , walking , exercise walking , international standard dancing , and a longer practice of yoga also attenuated the genetic effects on BMI . Exercises such as cycling , stretching exercise , swimming , dance dance revolution , and qigong were not found to modify the genetic effects on any obesity measure . Across all 5 obesity measures , regular jogging consistently presented the most significant interactions with GRSs . Our findings show that the genetic effects on obesity measures can be decreased to various extents by performing different kinds of exercise . The benefits of regular physical exercise are more impactful in subjects who are more predisposed to obesity .
Obesity is one of the most challenging public health issues worldwide [1–6] . According to the World Health Organization , a person with a body mass index ( BMI ) of 30 kg/m2 or above is generally considered obese . Although BMI is easy to calculate and is commonly used to identify obesity , it does not take into account lean body mass or identify central obesity . Four important metrics , body fat percentage ( BFP ) , waist circumference ( WC ) , hip circumference ( HC ) , and waist-to-hip ratio ( WHR ) , are complementary to BMI . The BFP of an individual is the total fat mass divided by the total body mass , multiplied by 100 . HC is a useful predictor of metabolic syndromes such as diabetes [7] . WC and WHR are indicators of central obesity [8] . Obesity is complicated as it is caused by genetics , lifestyle , and the interplay between them [9 , 10] . The heritability of BMI was reported to range from 24% to 81% [11] , and many genes have been shown to be related to obesity [12] . Although hereditary factors are critical , some lifestyle factors can modify the genetic influences on BMI [13–24] . For example , regular physical exercise has been found to blunt the genetic effects on obesity [13–16 , 18 , 20 , 24] . However , most of these studies focused on only BMI , without discussing central obesity . Moreover , investigations specific to particular kinds of exercise remain limited . It is unknown what kinds of exercise ( jogging , mountain climbing , cycling , etc . ) can attenuate the genetic effects on obesity measures . To fill the research gap , we here comprehensively investigated gene-exercise interactions on the 5 obesity measures: BMI , BFP , WC , HC , and WHR . Moreover , we investigated whether 18 kinds of exercise could modify the associations between genetic risk scores ( GRSs ) and these 5 obesity measures .
TWB received ethical approval from the Institutional Review Board on Biomedical Science Research/IRB-BM , Academia Sinica , Taiwan , and from the Ethics and Governance Council of Taiwan Biobank , Taiwan . Written informed consent was obtained from each participant in accordance with institutional requirements and the principles of the Declaration of Helsinki . Moreover , the current study was approved by the Research Ethics Committee of National Taiwan University Hospital ( NTUH-REC no . 201805050RINB ) . Taiwan Biobank ( TWB ) is the largest government-supported biobank in Taiwan . The aim of TWB is to collect lifestyle and genomic data from Taiwan residents [25 , 26] . TWB keeps recruiting community-based volunteers who are 30 to 70 years of age and have no history of cancers . Participants signed informed consent , provided blood samples and a range of information via a face-to-face interview and physical examination . Our study comprised 20 , 287 TWB individuals who have been whole-genome genotyped until October , 2018 . To remove cryptic relatedness , we estimated the genome-wide identity by descent ( IBD ) sharing coefficients between any two subjects . The IBD scores for all pairs of subjects , i . e . , PI-HAT = Probability ( IBD = 2 ) + 0 . 5×Probability ( IBD = 1 ) , were obtained from PLINK 1 . 9 [27] . Similar to many genetic studies [28–30] , we excluded third-degree relatives by removing one individual from a pair with PI-HAT ≥ 0 . 125 . After this step , 18 , 424 unrelated subjects ( 9 , 093 males and 9 , 331 females ) remained in our analysis . The majority of TWB subjects were of Han Chinese ancestry [25] . The TWB chip is based on Axiom Genome-Wide Array Plate System ( Affymetrix , Santa Clara , CA , USA ) . It genotyped a total of 646 , 783 autosomal single-nucleotide polymorphisms ( SNPs ) . We excluded 51 , 293 SNPs with genotyping rates < 95% , 6 , 095 SNPs with Hardy-Weinberg test P-values < 5 . 7×10−7 [31] , and 1 , 869 variants with minor allele frequencies ( MAFs ) < 1% . The remaining 587 , 526 SNPs were used to construct ancestry principal components ( PCs ) for the adjustment of population stratification . The TWB measured body height and weight for each participant . BMI was calculated by weight ( kg ) /[height ( m ) ]2 . In addition to BMI , 4 measures including BFP , WC , HC , and WHR were also investigated . BFP is the percentage of an individual’s weight that is made up of fat . WHR is the ratio of WC to HC and is a commonly used index for central obesity [8] . In addition to a physical examination , each participant completed a questionnaire through a face-to-face interview with one of the TWB researchers . Questions addressed personal information and lifestyle factors . Regular exercise was defined as engaging in 30 minutes of “exercise” three times a week . “Exercise” included only leisure-time activities such as jogging , yoga , mountain climbing , cycling , swimming , dance dance revolution ( DDR , a computer game based on dancing with music videos ) , playing basketball , etc . Occupational activities such as physical work or heavy manual work were not counted as “exercise” . Sex and age ( in years ) have been considered as important covariates in most obesity studies [13–16 , 18 , 20 , 24 , 32–34] . Moreover , some studies also adjusted for drinking status , smoking status , and educational attainment [16] . A previous large-scale study has found an inverse association between BMI as well as WC and education level [35] . Therefore , we also considered educational attainment as one of the covariates for obesity measures . Educational attainment was recorded as a value ranging from 1 to 7 , where 1 indicated “illiterate” , 2 meant “no formal education but literate” , 3 represented “primary school graduate” , 4 indicated “junior high school graduate” , 5 meant “senior high school graduate” , 6 represented “college graduate” , and 7 indicated “Master’s or higher degree” . Drinking was defined as a subject having a weekly intake of more than 150 cc of alcohol for at least 6 months and having not stopped drinking at the time his/her obesity measures were being assessed . Smoking was defined as a subject who had smoked for at least 6 months and had not quit smoking at the time his/her obesity measures were being assessed . In most gene-environment interaction ( G×E ) studies , investigators typically constructed a GRS and tested the significance of the GRS×E interaction term ( E represents the environmental factor ) [13–24] . A GRS was a weighted sum of risk-allele counts , where the weights were usually retrieved from large published genome-wide association studies ( GWASs ) or meta-analyses [13–24] . Recent G×E studies related to obesity measures [14 , 16–19 , 21 , 23] usually constructed a GRS according to the results of a large meta-analysis [34] , in which 97 BMI-associated SNPs reaching the genome-wide significance level ( p < 5×10−8 ) were reported [34] . A total of 20 out of the 97 SNPs were genotyped in the TWB chip . We imputed the genotypes of other SNPs using the Michigan Imputation Server ( https://imputationserver . sph . umich . edu/index . html ) , with the reference panel based on the East Asian ( EAS ) population from the 1000 Genomes Phase 3 v5 . After removing SNPs with MAFs < 1% and SNPs with Hardy-Weinberg test P-values < 5 . 7×10−7 [31] , 86 SNPs remained in S1 Table . The European-based GRS was calculated as EuGRS=∑j=186wjSNPj , where the weights ( wj , j = 1 , ⋯ , 86 ) were the effect sizes reported by Locke et al . [34] , and SNPj was the number of effect alleles at the jth SNP . Each EuGRS was then transformed into a z-score that indicated how many standard deviations an EuGRS was from the mean . Although EuGRS is positively associated with the 5 obesity measures ( S2 Table ) ( the results of EuGRS×exercise interactions can be found from S3–S5 Tables ) , it may not be an efficient GRS to detect TWB G×E for the following three reasons . First , the 97 SNPs account for 2 . 70% of BMI variation in Europeans [34] . However , in TWB subjects , these SNPs can only explain 1 . 92% , 1 . 05% , 1 . 43% , 1 . 60% , and 0 . 79% of variation of BMI , BFP , WC , HC , and WHR , respectively ( S6 Table ) . Second , all the 97 BMI-associated SNPs reached the genome-wide significance level ( p < 5×10−8 ) in Europeans . However , in TWB , only rs1558902 located in the fat mass and obesity-associated ( FTO ) gene was associated with BMI at the genome-wide significance level , and only 29 were associated with BMI at the significance level of 0 . 05 ( S1 Table ) . Third , none of the 97 BMI-associated SNPs were associated with the other 4 obesity measures at the genome-wide significance level ( S1 Table ) . BMI is the most commonly investigated obesity measure . SNPs robustly associated with other obesity measures have not been reported . Based on the above three reasons , using EuGRS may be inefficient for Han Chinese and for obesity measures other than BMI . However , large obesity-related GWASs in Han Chinese are unavailable . To overcome this problem , we used internal weights to construct a GRS , and then tested the GRS×E interaction term in a regression model . This approach has been proposed in genome-wide [36] , pathway-based [37 , 38] , and gene-based G×E studies [39 , 40] . Initially , SNPs in high linkage disequilibrium ( LD ) were first pruned to avoid multicollinearity [41 , 42] . We used PLINK 1 . 9 command “plink--bfile TWBGWAS--chr 1–22--indep 50 5 2” to prune SNPs in high LD [27] . In this way , we removed SNPs with a variance inflation factor > 2 within a sliding window of size 50 , where the sliding window was shifted at each step of 5 SNPs . After this pruning stage , 142 , 040 SNPs remained . We then regressed BMI on each of the 142 , 040 SNPs while adjusting for covariates including sex , age , educational attainment , drinking status , smoking status , and the first 10 PCs . The 142 , 040 regression models were built as follows: BMI=β0+βSNP , iSNPi+βCCovariates+ε , i=1 , ⋯ , 142040 , ( 1 ) where SNPi is the number of minor alleles at the ith SNP ( 0 , 1 , or 2 ) and ε is the error term . By testing H0: βSNP , i = 0 vs . H1: βSNP , i ≠ 0 , we obtained a P-value regarding the marginal association of the ith SNP with BMI . Considering the model incorporating SNP-by-environment interactions , as follows: BMI=γ0+γSNP , iSNPi+γEE+γInt , iSNPi×E+γCCovariates+ε , i=1 , ⋯ , 142040 , ( 2 ) β^SNP , i ( estimated from model 1 ) and γ^Int , i ( estimated from model 2 ) are asymptotically independent under the null hypothesis of no SNP-by-environment interaction ( proved in corollary 1 of [43] ) . A two-stage approach that first filters SNPs by a criterion independent of the test statistic ( γ^Int , i estimated from model 2 ) under the null hypothesis , and then only uses SNPs that pass the filter , can maintain type I error rates and boost power [44 , 45] . Given a P-value threshold ( a filter ) , the 142 , 040 SNPs were allocated into a BMI-associated set and a BMI-unassociated set according to their marginal-association P-values . Suppose there were m SNPs associated with BMI , the BMI genetic risk score ( BMIGRS ) was calculated as ∑j=1mβ^SNP , jSNPj , where the weights ( β^SNP , j , j=1 , ⋯ , m ) had been estimated from model ( 1 ) , and SNPj was the number of minor alleles at the jth SNP in the BMI-associated set . Because BMI-unassociated SNPs were filtered out from the construction of BMIGRS , this approach is the so-called “marginal-association filtering” in G×E analyses [40 , 43 , 45] . Following the suggestion from our previous methodological study [36] , 10 P-value thresholds were considered: 0 . 0001 , 0 . 00025 , 0 . 0005 , 0 . 001 , 0 . 0025 , 0 . 005 , 0 . 01 , 0 . 025 , 0 . 05 , and 0 . 1 . S7 Table shows the numbers of SNPs in the BMI-associated sets under the 10 P-value thresholds . For each TWB subject , 10 BMIGRSs were calculated based on the 10 sets of SNPs . For example , the 9th BMIGRS accumulated the information of 7 , 753 SNPs ( S7 Table ) . Similar with model ( 1 ) , BFP , WC , HC and WHR were regressed on each of the 142 , 040 SNPs while adjusting for the same covariates , respectively . A total of 10 BFPGRSs , 10 WCGRSs , 10 HCGRSs , and 10 WHRGRSs were obtained under the 10 P-value thresholds . Each GRS was then transformed into a z-score that indicated how many standard deviations a GRS was from the mean . The number of SNPs to form each GRS was listed in S7 Table . We investigated whether the association of BMIGRS with BMI could be modified by regular physical exercise ( yes or no ) . BMI was regressed on a BMIGRS , regular exercise or not ( E: 1 vs . 0 ) , and the interaction between them ( BMIGRS×E ) , while adjusting for sex , age , educational attainment , drinking status , smoking status , and the first 10 PCs . The regression model was built as follows: BMI=β0+βGRSBMIGRS+βEE+βIntBMIGRS×E+βCCovariates+ε . ( 3 ) With 10 BMIGRSs , 10 regression models like ( 3 ) were fitted and 10 P-values regarding testing H0: βInt = 0 vs . H1: βInt ≠ 0 were obtained . To adjust for multiple testing , the Bonferroni-corrected P-value was calculated as 10 times the minimum P-value of the 10 BMIGRS×E interaction tests . This approach is called “the GRS approach based on marginal effects of SNPs” , abbreviated as the “GRS-M” method [36] . The comprehensive simulations performed by Hüls et al . [37 , 38] and Lin et al . [36] have confirmed the validity of building GRS with marginal effects of SNPs in detecting G×E . Extracting weights from other cohorts or splitting data in two subsets is not required for the GRS-M approach [36] . The GRS-M approach is valid in the sense that the empirical type I error rate is satisfactorily controlled . Furthermore , it is generally the most powerful test if some phenotype-associated SNPs also exhibit interactions with E [36] . Similarly , we also investigated GRS-exercise interactions on the other 4 obesity measures . The significance level throughout this work was set at 0 . 05/550 = 9 . 1x10-5 because 275 tests for GRS-exercise interactions and 275 tests for main effects of exercises were performed .
Table 1 presents the basic characteristics of the TWB subjects , stratified by the quartiles of the 9th BMIGRS . The aim of this study was to test whether the genetic effects on obesity measures can be modified by any of 18 kinds of exercise . A previous large-scale study has found an inverse association between BMI as well as WC and education level [35] . Our TWB analysis results also show improvements when including educational attainment as a covariate for all 5 obesity measures . By including educational attainment as a covariate , the adjusted R-square increased from 5 . 9% to 7 . 3% for BMI , from 34 . 8% to 35 . 9% for BFP , from 14 . 3% to 15 . 6% for WC , from 4 . 5% to 4 . 8% for HC , and from 23 . 2% to 24 . 6% for WHR , respectively . To explore the associations of covariates with the 5 obesity measures , Table 2 shows the results of regressing each obesity measure on sex , age , educational attainment , drinking status , smoking status , regular exercise , and the first 10 PCs . Sex was the most significant predictor for all 5 obesity measures . Except for BFP , males had larger mean values than females in the other 4 obesity measures . Educational attainment and regular exercise were also significant predictors for all 5 metrics . These results were consistent with previous findings: attaining a higher education degree [35] and performing regular physical exercise [46] were associated with a decrease in obesity measures . Among the 18 , 424 subjects , 7 , 652 ( 41 . 5% ) reported performing regular exercise , while 10 , 764 reported no regular exercise . A total of 8 subjects did not respond to this question . For a subject who reported performing regular exercise , he/she would then be asked questions regarding the kinds of exercise , the frequency of engaging in a particular exercise per month , and the duration in each practice . An individual could enumerate up to 3 kinds of regular exercise . Table 3 shows that each 1 s . d . increase in BMIGRS was associated with a 0 . 43 kg/m2 lower BMI in exercisers than in nonexercisers ( p = 1 . 3×10−32 ) . Each 1 s . d . increase in BFPGRS was associated with a 0 . 62% lower BFP in exercisers than in nonexercisers ( p = 1 . 2×10−15 , Table 3 ) . Regular physical exercise also significantly attenuated the genetic effects on WC and HC . However , the WHRGRS-exercise interaction was not significant ( p = 1 ) . Fig 1 shows the average BMI , BFP , WC and HC stratified by GRS quartiles and regular exercise . The effects of GRSs on these 4 obesity measures were smaller in physically active subjects than in physically inactive subjects . Regular exercise attenuated the genetic predisposition to obesity measures . We then performed a specific analysis for the 18 kinds of exercise . Some TWB individuals reported multiple kinds of regular exercise , and a limit of 3 kinds could be recorded by TWB interviewers . Therefore , when we assessed the interaction between a GRS and a kind of exercise , whether a person also engaged in other kinds of exercise should be considered . The regression models were similar with model ( 3 ) , but more covariates were adjusted in the models . For example , to investigate the BMIGRS-jogging interaction on BMI , we regressed BMI on a BMIGRS , jogging or not ( 1: yes vs . 0: no ) , the interaction between them , while adjusting for sex , age , educational attainment , drinking status , smoking status , the first 10 PCs , 17 covariates regarding engaging in the other 17 kinds of exercise or not , and the 17 BMIGRS-exercise interaction terms . As shown in Table 3 , all types of exercise generally attenuate the genetic contributions of BMI , BFP , WC and HC , as indicated by the direction of the interaction terms ( β^Int<0 ) . Among the 18 kinds of exercise , jogging , mountain climbing , walking , exercise walking , and international standard dancing significantly attenuated the genetic effects on BMI ( p < 9 . 1x10-5 ) . Moreover , jogging additionally attenuated the genetic effects on BFP and HC . As shown in Table 3 , across all 5 obesity measures , jogging consistently presented the most significant interactions with GRS ( i . e . , the smallest P-value ) . Fig 2 shows the average BMI , BFP , WC and HC stratified by GRS quartiles and jogging . The effects of GRSs on these 4 obesity measures were smaller in joggers than in nonjoggers . The results of exercise frequency ( Table 4 ) and duration ( Table 5 ) were similar to those of engaging in the kind of exercise ( Table 3 ) . Additionally , a longer practice of yoga could blunt the genetic effects on BMI ( Table 5 ) . Fig 3 shows the effect of BMIGRS on BMI , stratified by exercise types . All types of exercise generally attenuate the genetic effects of BMI , as indicated by β^GRS of each exercise type < β^GRS of no exercise . The GRS effects on other 4 obesity measures can be found from S1–S4 Figs . S11–S13 Tables present the results of GRS×exercise interactions , stratified by sex . The directions of β^Ints were in line with the results in Tables 3–5 where sex was treated as a covariate adjusted in model ( 3 ) . All types of exercise generally attenuate the genetic contributions of BMI , BFP , WC and HC , as indicated by the direction of the interaction terms ( β^Int<0 ) .
Because 95% of the subjects in Locke et al . ’s study [34] were of European descent , building GRS according to these 97 SNPs may not be appropriate for other ethnic populations . Although the same data set is used to estimate βSNP , i ( i = 1 , ⋯ , 142040 ) and to test the significance of GRS×E , this GRS-M approach is valid in the sense that the type I error rates are satisfactorily controlled [36] . Corollary 1 of Dai et al . [43] has justified the validity of using marginal associations ( between SNP and an obesity measure ) as the filtering test statistics , and the data-splitting strategy is not required . Building GRS with internal weights has been used in some G×E analyses [36–40 , 48] . Previous G×E analyses have typically constructed a GRS using SNPs that reached the genome-wide significance level ( i . e . , p < 5×10−8 ) [13–24] . However , some studies have suggested that a GRS comprising more SNPs can improve the prediction for a phenotype [41 , 49–51] . SNPs that interact with an environmental factor may not necessarily present a strong marginal association with the phenotype . To explore G×E , it is worthwhile to consider a more liberal threshold than the genome-wide significance level ( 5×10−8 ) . For example , the “Set-Based gene-EnviRonment InterAction test” ( SBERIA ) constructs a GRS by using all SNPs with a marginal-association P-value < 0 . 1 [39 , 40] . In fact , the optimal filtering P-value threshold varies with environmental factors and phenotypes [52] . Therefore , the GRS-M method considers 10 P-value thresholds for marginal-association filtering [36] . For each obesity measure , 10 GRSs were calculated , and then 10 regression models were fitted . To adjust for multiple testing , the GRS-M P-value was reported as 10 times the minimum P-value of the 10 GRS-exercise interaction tests . The GRS-M test is a valid statistical method by controlling type I error rates well [36] . As summarized in S7 Table , significant GRS-exercise interactions were detected at a marginal-association P-value threshold between 0 . 0025 and 0 . 05 , and the number of SNPs used to construct each of the GRSs ranged from 481 to 7 , 753 . With the development of relatively inexpensive SNP arrays , using more SNPs than those achieving the genome-wide significance level is currently feasible [53] . Previous studies have found that performing regular physical exercise could blunt the genetic effects on BMI [13–16 , 18 , 20 , 24] . However , few studies have investigated BFP or measures of central obesity . These obesity measures are even more relevant to health than BMI . For example , central obesity is considered to be a predominant risk factor for metabolic syndrome [54 , 55] . We here show that performing regular exercise attenuates the genetic effects on 4 obesity measures , including BMI , BFP , WC , and HC ( Table 3 ) . Regarding exercise types , regular jogging mitigated the genetic effects on BMI , BFP , and HC . Mountain climbing , walking , exercise walking , and international standard dancing also attenuated the genetic effects on BMI ( Table 3 ) . Moreover , a longer practice of yoga blunted the genetic effects on BMI ( Table 5 ) . These results indicated that although hereditary factors are critical to obesity , performing different kinds of exercise can modify this relationship to various extents . A BMI that is too high or too low is associated with an increased mortality rate . According to studies from western Europe and North America [56] , a BMI ranging from 22 . 5 to 25 kg/m2 corresponded to the lowest overall mortality . Fig 1 ( A ) shows that regular physical exercise was associated with an increase in BMI at a low BMIGRS ( the bottom quarter: Q1 ) but a decrease in BMI at a high BMIGRS ( the top quarter: Q4 ) . Performing regular exercise was associated with a reduced risk of having a too-high or a too-low BMI . Summarizing Tables 3–5 , a total of 12 kinds of exercise did not achieve significance for the attenuation of the genetic risk of obesity measures . Plausible reasons included ( 1 ) less popularity or ( 2 ) a smaller GRS-exercise interaction effect . Exercises such as cycling ( 989 subjects ) , stretching exercise ( 602 subjects ) , swimming ( 486 subjects ) , DDR ( 420 subjects ) , and qigong ( 377 subjects ) were more popular or as popular as yoga ( 379 subjects ) , but their evidence of interacting with GRS was relatively weak ( Table 3 ) . These 5 kinds of exercise may have limited effects on mitigating the genetic risk of obesity measures . In contrast , although the evidence of GRS-Tai Chi interactions did not achieve the Bonferroni-corrected significance level ( 9 . 1x10-5 ) , the small P-values implied that engaging in Tai Chi ( 449 subjects ) might potentially blunt the genetic effects on obesity measures . Few studies have investigated the interplay between particular kinds of exercise and genetic risk of obesity measures . Therefore , we can hardly compare our results with previous findings . We here provide possible explanations for these results . Cycling ( 989 subjects ) , stretching exercise ( 602 subjects ) , and qigong ( 377 subjects ) usually require less energy expenditure than the 6 exercises that demonstrate interactions with GRS [57] . Exercises in cold water such as swimming ( 486 subjects ) can especially stimulate appetite and food intake [58 , 59] . DDR ( 420 subjects ) , a computer game based on dancing with music videos , is not as formal as international standard dancing . These reasons may possibly explain why these 5 popular exercises ( cycling , stretching exercise , qigong , swimming , and DDR ) cannot mitigate genetic susceptibility to obesity measures . Because relatively few subjects engaged in weight training ( 218 subjects ) , badminton ( 204 subjects ) , table tennis ( 169 subjects ) , basketball ( 119 subjects ) , or tennis ( 110 subjects ) , the statistical power to detect the interplay between GRS and these exercises was limited . Further research on these 5 kinds of exercise will be interesting . A G×E study for BMI using 362 , 496 UK Biobank subjects has reported that a quicker walking pace attenuated the genetic effects on BMI ( the top row in Tables 2–3 of [14] ) . This is consistent with our findings in Tables 3–5 , i . e . , |β^Int| of BMIGRS×jogging > |β^Int| of BMIGRS×exercise walking > |β^Int| of BMIGRS×walking . Because pace of jogging > pace of exercise walking > pace of walking , our results also show that a quicker walking pace could more effectively attenuate the genetic effects on BMI . Moreover , the frequency of stair climbing in last 4 weeks has been found to blunt the effect of BMIGRS ( Tables 2–3 of [14] ) . Similarly , we here detected significant interactions between BMIGRS and both the frequency ( Table 4 ) and duration ( Table 5 ) of mountain climbing . Some previous studies investigated the efficacy of performing several kinds of exercise in preventing obesity [60 , 61] . For example , a randomized controlled trial with 64 subjects assigned to the Tai Chi group and 78 assigned to the control group demonstrated that performing Tai Chi led to a marked but non-significant reduction in WC [60] . For comparison , in S8–S10 Tables , we listed the associations of 18 kinds of exercise with obesity measures , i . e . , β^E estimated from model ( 3 ) . Our results showed that performing Tai Chi was significantly associated with a reduction in WC and BFP ( p < 9 . 1x10-5 ) . Regular jogging , performing yoga and Tai Chi were associated with a decrease in multiple obesity measures . Moreover , playing table tennis was associated with a reduction in WHR . WC and WHR are indicators of central obesity [8] . Our results show that performing Tai Chi or playing table tennis was related to a reduced risk of central obesity , presumably because waist turning is frequently required when engaging in these two kinds of exercise . The results for associations of 18 kinds of exercise with obesity measures were robust to the exclusion of GRS and GRS-exercise interaction terms . In addition to obtaining βE^ from model ( 3 ) , we additionally fitted the following model without GRS and the relevant interaction terms: BMI ( oranotherobesitymeasure ) =β0+βEE+βCCovariates+ε , ( 4 ) where E was some kind of exercise , and covariates included sex , age , educational attainment , drinking status , smoking status , the first 10 PCs , and 17 covariates regarding engaging in the other 17 kinds of exercise or not . The results were similar to those obtained from model ( 3 ) , i . e . , regular jogging , performing yoga , Tai Chi and playing table tennis were associated with a decrease in obesity measures . To sum up , regular jogging and performing yoga were not only associated with a decrease in obesity measures , but they also attenuated the genetic predisposition to obesity measures . Exercises such as walking , exercise walking , mountain climbing , and international standard dancing , were not significantly associated with a change in obesity measures , but these 4 kinds of exercise could blunt the genetic effects on BMI . By comparing rows of “walking” and “yoga” in S8–S10 Tables , our result is consistent with a previous finding that engaging in yoga shows a larger reduction in BMI than walking [61] . It is interesting that , across all 5 obesity measures , regular jogging consistently presented the most significant interactions with GRSs ( Table 3 ) . The genetic effects on obesity measures can be decreased to various extents by performing different kinds of exercise . The benefits of regular physical exercise , especially jogging , are more impactful in subjects who are more predisposed to obesity . | The complex interplay of genetics and lifestyle makes obesity a challenging issue . Previous studies have found performing regular physical exercise could blunt the genetic effects on body mass index ( BMI ) . However , BMI does not take into account lean body mass or identify central obesity . Moreover , it remains unclear what kinds of exercise could more effectively attenuate the genetic effects on obesity measures . With a sample of 18 , 424 unrelated Han Chinese adults , we comprehensively investigated gene-exercise interactions on 5 obesity measures: BMI , body fat percentage , waist circumference , hip circumference , and waist-to-hip ratio . Moreover , we tested whether the genetic effects on obesity measures could be modified by any of 18 kinds of self-reported regular exercise . Because no large genome-wide association studies on obesity have been done for Han Chinese , we constructed genetic risk scores with internal weights for analyses . Among these exercises , regular jogging consistently presented the strongest evidence to mitigate the genetic effects on all 5 obesity measures . Moreover , mountain climbing , walking , exercise walking , international standard dancing , and a longer practice of yoga attenuated the genetic effects on BMI . The benefits of regularly performing these 6 kinds of exercise are more impactful in subjects who are more predisposed to obesity . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"body",
"weight",
"medicine",
"and",
"health",
"sciences",
"swimming",
"education",
"sociology",
"sports",
"and",
"exercise",
"medicine",
"social",
"sciences",
"physical",
"activity",
"biological",
"locomotion",
"physiological",
"parameters",
"obesity",
"public",
"and",... | 2019 | Performing different kinds of physical exercise differentially attenuates the genetic effects on obesity measures: Evidence from 18,424 Taiwan Biobank participants |
The Warburg effect - a classical hallmark of cancer metabolism - is a counter-intuitive phenomenon in which rapidly proliferating cancer cells resort to inefficient ATP production via glycolysis leading to lactate secretion , instead of relying primarily on more efficient energy production through mitochondrial oxidative phosphorylation , as most normal cells do . The causes for the Warburg effect have remained a subject of considerable controversy since its discovery over 80 years ago , with several competing hypotheses . Here , utilizing a genome-scale human metabolic network model accounting for stoichiometric and enzyme solvent capacity considerations , we show that the Warburg effect is a direct consequence of the metabolic adaptation of cancer cells to increase biomass production rate . The analysis is shown to accurately capture a three phase metabolic behavior that is observed experimentally during oncogenic progression , as well as a prominent characteristic of cancer cells involving their preference for glutamine uptake over other amino acids .
The Warburg effect , a phenomenon discovered by Otto Warburg in 1924 , reflects a shift to an inefficient metabolism in cancer cells , in which an increase in the inefficient production of adenosine 5′-triphosphate ( ATP ) via glycolysis leads to the secretion of non-oxidized carbons in the form of lactate , even in the presence of oxygen ( termed aerobic glycolysis ) [1] , [2] . Specifically , aerobic glycolysis allows the production of only 2 ATP molecules per one glucose molecule , whereas oxidative phosphorylation permits the generation of 32 ATP molecules per one molecule of glucose [3] . Nevertheless , the importance of aerobic glycolysis to cancer cells has been experimentally demonstrated [4] , [5] . Over the years , several hypotheses were raised regarding the potential cause of the Warburg effect: ( i ) Defective mitochondrion hypothesis – suggesting that cancer cells have defective mitochondria and hence rely on glycolysis [6] , however subsequent research revealed that mitochondrial function is not impaired in most cancer cells [7] , [8] . ( ii ) Hypoxia– suggesting that tumor hypoxia selects for cells dependent on anaerobic metabolism [9] , but previous studies have shown that cancer cells already resort to aerobic glycolysis before exposure to hypoxic conditions [10] , [11] . ( iii ) Avoiding ROS-mediated DNA damage – it was suggested that reducing oxidative phosphorylation in proliferating cells due to the Warburg shift reduces ROS and hence protects cells from DNA damage and subsequent apoptosis [12] . ( iv ) A game theoretical approach suggesting that the Warburg effect occurs as glycolysis provides higher ATP production rate than oxidative phosphorylation [13] , [14] , [15] . ( v ) An approach suggesting that a trade-off between the enzyme-synthesis costs and the ATP production yields of the different pathways that catabolize carbon sources may cause the Warburg effect: the high-yield oxidative phosphorylation pathway also has high enzyme costs , leading to a sub-optimal ATP production strategy , as it has lower production rates than glycolysis [16] . ( vi ) Metabolic adaptation to fast proliferation - it was argued that as opposed to metabolism in differentiated cells that is geared towards efficient ATP production , the aerobic glycolysis observed in cancer cells is adapted to facilitate biomass accumulation and high proliferation . Accordingly , in order to satisfy the requirements of anabolic metabolism in addition to the production of ATP , nutrients must be used to generate both the carbon building blocks of macromolecules and the reducing power needed for biosynthesis [17] . Previous computational investigations of the Warburg effect studied the role of either energy or biomass production in causing the Warburg effect , focusing solely on central carbon metabolism . For example , the study of Vander Heiden et al . manually computed the metabolic requirements for producing one essential biomass precursor , palmitate ( a major constituent of cellular membranes ) considering the stoichiometry of a few central metabolic pathways . They found that aerobic glycolysis enables maximal palmitate production yield due to specific reducing power requirements . In another recent study , Vazquez et al . employed a schematic model of ATP production in human cells ( considering two lumped reactions representing aerobic glycolysis and oxidative phosphorylation ) , elegantly showing that a switch to aerobic glycolysis should result from cellular maximization of ATP production [18] . Their schematic model accounts not only for the stoichiometry of glycolysis and oxidative phosphyrylation but also for the enzyme-volumetric costs of activating these pathways ( the latter bounded by the total cellular solvent capacity , also known as a macromolecular crowding constraint [19] ) . A similar approach was previously employed in the study of over-flow metabolism in E . coli [20] , [21] . Another interesting theory explaining overflow metabolism was suggested by Molenaar et al . , where the production costs of the metabolic enzymes involved were accounted for in a self-replicating model [16] . In this paper , we study the causes of the Warburg effect by accounting for both energy production and anabolism of essential biomass constituents , in a genome-scale stoichiometric network model [22] employing enzyme solvent capacity constraints . The usage of a large-scale metabolic network is essential if one aims to correctly account for the inter-connectivity of pathways that produce the various energy and biomass precursors required for proliferation , rather than examining just single factors in isolation , as has been previously performed in [17] , [18] . Towards this goal , we rely on a constraint-based modeling ( CBM ) framework that serves to analyze the function of metabolic networks by solely relying on simple physical-chemical constraints [23] . CBM has already been successfully used in the past to predict the metabolic state of various microorganisms [24] , [25] , and recently for studying human cellular metabolism [22] . The potential clinical utility of the human CBM model was previously demonstrated by its ability to identify functionally related sets of reactions that are causally related to hemolytic anemia , and potential drug targets for treating hypercholesterolemia [22] , to predict metabolic biomarkers in inborn errors of metabolism [26] and to predict a variety of metabolic behaviors of different human tissues , including the brain , liver , kidney and more [27] , [28] . Our analysis shows that while strictly stoichiometric considerations are insufficient for explaining the Warburg effect , the incorporation of enzyme solvent capacity constraints successfully predicts the emergence of the Warburg effect . The analysis is shown to accurately predict an experimentally observed metabolic trajectory occurring during oncogenic progression , as well as the preference of cancer cells for a high rate of glutamine uptake .
We utilized a genome-scale human metabolic network that includes 3 , 742 reactions [22] , adding a pseudo biomass reaction that represents the production of a pre-defined set of essential biomass precursors required for cellular proliferation , as conventionally done in Flux Balance Analysis ( FBA , [29] , see Methods ) . The biomass precursors include amino-acids , nucleotides , deoxy-nucleotides , ATP , lipids , etc ( based on prior knowledge of their relative concentrations; Methods ) . In our simulations , we assume a minimal growth medium with glucose as a carbon source , as glucose is known to serve as a major fuel in cancer cells ( below and in Text S1 we show that qualitatively similar results were obtained when considering also the presence of an additional major nutrient taken by cancer cells , glutamine ) . To predict plausible metabolic fluxes in cancer , we first employed a standard FBA method to identify a feasible flux distribution that satisfies stoichiometric mass-balance , while maximizing biomass production yield ( see Methods ) . We found that the predicted flux distribution does not display the prime characteristic of the Warburg effect , i . e . lactate secretion ( see also Text S1 ) . Interestingly , this finding is in accordance with a previous study showing a conceptually similar failure of FBA to predict the Crabtree effect in yeast , in which glucose is fermented into ethanol under aerobic conditions [30] . Thus , stoichiometric considerations alone are insufficient for explaining the Warburg effect and its relation to the metabolic requirements of highly proliferating cells . Notably , these results stand in difference from those presented by Vander Heiden et al . [17] , claiming that strictly stoichiometric considerations directly lead to the Warburg effect due to metabolic demands for cellular proliferation . A strictly stoichiometric analysis , such as the one presented above , implicitly assumes that metabolic flux rates can be tuned to achieve high biomass production yields , without considering constraints imposed by enzyme concentrations and catalytic rates , which are prime determinants of metabolic flux . Specifically , while cells might be free to regulate enzyme concentrations according to metabolic demands [17] , the total enzymes' concentration in the proliferating cells is bounded by the cell's solvent capacity , quantifying the maximum amount of macromolecules that can occupy the intracellular space [18] . To account for the functional effects of this additional fundamental constraint , we follow [18] , [21] and extend our stoichiometric genome-scale CBM analysis to compute for each enzyme the concentration required to facilitate the predicted flux , utilizing data on known human enzyme catalytic rates ( taken from the literature; see Methods ) . This modeling approach enables the prediction of metabolic flux distributions that maximize the biomass production rate and concomitantly obey the solvent capacity constraints – rather than predicting flux distributions that only maximize the biomass production yield as done in standard FBA . We applied the approach described above ( FBA with solvent capacity constraint ) to predict human cellular flux distributions that maximize the biomass production rate . To simulate varying growth rates we performed the optimization across a wide range of different glucose uptake rates . Indeed , under these combined sets of constraints we find that biomass yield does decline at high growth rates – in accordance with the Warburg effect [17]; Figure 1a ) . Specifically , the predicted metabolic behavior manifests three distinct growth phases ( Figure 1b ) : ( i ) optimal yield metabolism at a growth rate that is below 43% of the maximal possible rate , characterized by low glycolytic vs . high oxidative phosphorylation ( OXPHOS ) flux ( Figure 2a , phase I ) , with low oxygen uptake rates ( Figure 1b , phase I ) . ( ii ) Intermediate yield metabolism at growth rate between 43-92% , characterized by increased glycolytic and oxidative phosphorylation flux ( Figure 1a , phase II ) , the latter involving a significantly increased oxygen consumption ( Figure 1b , phase II ) . Notably , our prediction for an intermediate phase , involving increased oxygen consumption , presents a remarkable resemblance to two recent experimental studies examining the metabolic activity at different oncogenic progression stages ( [31] , Figure 1c and [32] , Figure 2b ) . Neither the stoichiometric model [17] nor an analysis using the schematic model of [18] give rise to similar predictions . ( iii ) Low yield metabolism at a growth rate above 92% of the maximal possible growth rate , characterized by a sharp increase in glycolytic flux and a decrease in oxidative phosphorylation ( and hence of O2 uptake ) . The increase in aerobic glycolysis flux ( Figure 2a , phase III ) leads to a rise in lactate secretion rates - a prime characteristic of the Warburg effect ( Figure 1b , phase III ) . To further validate the plausibility of the model , we examined the correlation between its enzyme concentration predictions ( based on predicted flux distributions; see Methods ) and mRNA expression values measured for 1 , 269 metabolic genes across 60 cancer cell lines of the NCI-collection [33] . The enzyme concentrations predicted with FBA accounting for the solvent capacity constraint show significant rank correlations with the gene expression data across the different cancer cell-lines ( mean Spearman correlation of 0 . 28 , mean p-value = 6 . 5e−21 ) . Notably , the strictly stoichiometric analysis provides significantly lower correlations with the expression measurements ( with a mean correlation of 0 . 1; Wilcoxon p-value = 3 . 5e−21 ) , further demonstrating the advantage of the genome-scale approach that accounts for enzyme solvent capacity . The shift towards low yield metabolism at high growth rates can be intuitively explained considering , on one hand , a flux distribution A with high growth yield ( YA ) and high ‘cost’ in terms of the required enzyme concentrations per unit of glucose uptake ( CA ) , and , on the other hand , a flux distribution B with a lower growth yield ( YB ) and lower cost ( CB ) . Considering a bound on the total enzyme concentration cost , one can observe that when the glucose uptake is unlimited flux distribution B will provide a higher growth rate if its growth yield normalized by its cost is higher than that of flux distribution A ( i . e . ; Figure 3 ) . When the glucose uptake rate is limited , maximal growth rate is achieved solely via flux distribution A or by a combination of A and B . Concretely , analyzing the results of our model , flux distribution A stands for a typical metabolic state in phase I , which is characterized by high mitochondrial oxidative phosphorylation , with a high growth yield of 0 . 094 and a high cost of 0 . 302 ( with a yield to cost ratio of 0 . 31 ) . Flux distribution B stands for a typical metabolic state at phase III , which involves a high rate of aerobic glycolysis , with a low growth yield of 0 . 035 and a low cost of 0 . 050 ( yield to cost ratio = 0 . 7 ) . These values indeed transcribe to a higher growth yield per unit of concentration cost of the enzymes participating in B , as alluded above . Figure 3 shows that , indeed , at low growth rates the glucose uptake rate is the sole limiting factor and hence the high yield oxidative phosphorylation route is taken; in contrast , at higher growth rates , the enzyme concentration constraint takes effect , and mixed solutions involving lactate secretion are necessarily formed . Notably , the predicted flux distributions across the range of growth rates described in this paper cannot be obtained from linear combinations of just two states ( as in the above simplified example ) , but are rather composed of multiple flux distributions with different growth yields per concentration cost ( as evident for example by the non-linear curve showing the predicted oxygen uptake rates across growth rates; Figure 1b ) . Thus , the flux distributions actually obtained in genome-scale models markedly differ from those that can be captured by a simplified analysis that describes the transition between just two metabolic states with different growth yields as above , or as in a previous study of Vazquez et al . [18] . The role of glutamine in cancer has been a topic of major interest as cancer cells are known to have a significant high glutamine uptake rate [34] . Repeating the previous analyses in the presence of both glucose and glutamine in the growth media shows qualitatively similar results to those described above . However , as expected , the addition of glutamine yields a higher maximal biomass production rate than the one obtained when only glucose was available in the medium ( Text S1 ) . To investigate the preference of cancer cells specifically to glutamine over other amino-acids , we applied FBA analysis to predict the contribution of each amino-acid separately to biomass production in the human model that accounts for enzyme solvent capacity constraints ( Methods ) . We find that , indeed , the contribution of glutamine to the proliferation rate is markedly higher than that of all other amino-acids ( Figure 4B ) . We further show that this result is robust to changing the bound on maximal amino-acid uptake rate , and that it remains valid across a large number of random samplings of enzyme turnover rates ( Text S1 ) . Repeating this analysis without accounting for enzyme solvent capacity constraints ( i . e . by considering only the network stoichiometry in the vanilla FBA model ) fails to predict the preference for glutamine ( Figure 4A ) . We carefully examined the flux distribution obtained with glutamine in the growth medium ( achieving a growth rate of 0 . 062 1/h ) vs . the one obtained with glutamate ( growth rate = 0 . 056 1/h ) . Interestingly , when glutamate is present in the medium , a large quantity of it is transformed into glutamine in an ATP consuming reaction catalyzed by the enzyme glutamine synthetase ( EC 6 . 3 . 1 . 2 ) . This satisfies the glutamine biomass requirement as well as the production of nucleotide precursors , among others . When removing the ATP requirement from this reaction , the growth rate achieved with glutamate in the medium increases to 0 . 059 1/h , which explains 50% of the growth rate difference . Notably , while this provides some intuitive explanation for the predicted preference for glutamine , we cannot identify a simple explanation for the entire effect due to the high complexity of the network model employed .
Metabolic adaptation to elevated growth requirements during cancer development has been recently suggested as the possible cause of the Warburg effect , a long-standing enigma of cancer metabolism . In this work we rigorously study this hypothesis using a genome-scale human metabolic model and demonstrate that stoichiometric considerations solely are insufficient to explain the shift to inefficient metabolism , in difference from recent claims [17] . However , integrating these constraints in a genome-scale model of human metabolism together with a constraint on enzyme solvent capacity does lead to the emergence of the Warburg effect at high proliferation rates . Furthermore , it accurately predicts a three phase metabolic behavior experimentally observed during oncogenic progression , as well as a marked preference to a high uptake rate of glutamine . The importance of enzyme solvent capacity in metabolic modeling has already been recognized in the earlier work of Beg et al . [20] , where applying such a constraint to the E . coli model improved phenotypic predictions . In their work , however , Beg et al . assumed an upper bound on the total cell-volume occupied by metabolic enzymes , as opposed to the method introduced here where we assume a bound on the enzyme mass per cell mass ( i . e . a bound on enzyme fractional concentration ) . In order to account for enzyme volumes , Beg et al . estimated enzyme volumes by assuming a uniform specific volume parameter ( representing the ratio between enzyme mass and volume ) for all enzymes . Here , we employed a simpler approach that does not depend on specific volume estimations , and explicitly constrains the total sum of enzyme mass . Notably , we further tested the effect of accounting for volumes instead of masses , and obtained results which are very similar to those obtained with masses only ( Text S1 ) . In a recent study by Molenaar et al . [16] , a metabolic shift at high growth rates was predicted based on a general self-replicating model . Another recent work ( by Vazquez et al . ) already pointed to the significance of the solvent-capacity constraint in explaining the Warburg effect [18] . Notably though , the work presented here provides a marked contribution over both studies: First , both employ abstract small-scale models . Specifically , the Vazquez et al . work introduces a schematic model of ATP production in central metabolism including just a handful of variables . Furthermore , similarly to the work of Pfeiffer et al . , their work does not explicitly account for the entire biomass composition and the associated energy requirements . In contrast , here we study a genome-scale biomass producing human model that , despite the scores of alternative biomass and energy production pathways existing in the human network , successfully shows that highly proliferating cells such as cancer cells are forced to display Warburg related phenotypes at high growth rates ( phase III ) . Additionally , and in contrast to the small-scale models , our genome-scale model correctly predicts an experimentally observed transitional phase ( II ) . Furthermore , on a mechanistic level , the genome-scale metabolic description provided by our analysis is significantly correlated with the gene expression patterns across the wide array of NCI-60 cancer cell-lines ( much stronger than the association displayed by the stoichiometric model alone ) , a result which could not have been predicted by the Vazquez et al . model . Lastly , the model was able to predict the marked contribution of glutamine to rapid cellular growth . As a further demonstration of the robustness of our results , we repeated the analyses using a model accounting for maintenance ATP production , obtaining qualitatively similar results ( Text S1 ) . While the data on reactions' stoichiometry is considered accurate and comprehensive , enzyme kinetic constant data are noisy and are currently available for only about 15% of the reactions in the model . In the analysis presented here , we addressed this problem by assigning enzymes with missing turnover rates with the median rate computed over the set of known turnover rates . Notably , the model's main findings are robust to random sampling of turnover rates from a distribution of known rates , as shown in Text S1 . However , repeating the analysis when assigning all reactions in the model with the median turn-over rate shows no Warburg characteristics - testifying to the importance of utilizing known turnover rates even if this data is sparse . Future measurements of additional enzyme turnover rates and improved methods for accurately predicting these parameters ( e . g . [35] ) are expected to further refine the predictions of cancer metabolic phenotypes using stoichiometric metabolic models with an enzyme solvent capacity constraint . In our work we accounted for a solvent capacity constraint assuming a limited protein mass per cell , without considering the effect of enzymes' sub-cellular compartmentalization . To investigate how the latter would affect our predictions , we repeated the analysis while considering separate solvent capacity constraints for cytoplasm and mitochondria ( Text S1 ) , yielding quantitatively similar results to those described above . The incorporation of solvent capacity constraints for different cellular compartments may lead to further improved prediction accuracy in the future , when additional data on enzyme turnover rates becomes available . Specifically , the addition of membrane-specific constraints may be a promising direction , as many metabolically important proteins are confined to membranes ( e . g . those of respiratory chain and membrane biosynthesis ) . The presented modeling approach is likely to contribute to more accurate metabolic modeling of highly proliferating human cells in general ( as was already shown regarding genome-scale models of microorganisms [21] ) and of cancer cells . The latter may be in turn utilized for anti-cancer drug target prediction and specifically , for predicting drugs that work to reverse the Warburg effect . While the current analysis has relied on the available human generic model , future studies may utilize a similar methodology in modeling the metabolism of specific cancers . These may be generated by integrating cancer-signature expression data with the generic human model to carve out different cancer types models ( using methods such as those outlined in [27] or [28] ) , and thus further advance the development of anti-cancer drugs specific to different cancers .
The Duarte et al . [22] human genome-scale metabolic model , accounting for 1 , 496 ORFs , 3 , 742 reactions and 2 , 766 metabolites , was used . The metabolic network is represented in a stoichiometric matrix S , where m is the number of metabolites , n is the number of reactions , and represents the stoichiometric coefficient of metabolite i in reaction j . Biomass production was modeled by adding a new growth reaction to the human model: this reaction was compiled using the steady state concentrations of 30 biomass compounds including amino acids ( 0 . 78 g/gDW; [36] , [37] ) , nucleotides ( 0 . 06 g/gDW; [38] ) , lipids ( 0 . 16 g/gDW; [39] ) as well as the growth-associated energy requirement ( 24 mmol/gDW of ATP; [40] ) . Essential amino acids were not accounted for since they were assumed not to take active part in the metabolic model besides flowing directly into the biomass reaction . The full list of biomass metabolites and their relative concentrations is available in Dataset S1 . The biomass reaction was defined as the objective function of the CBM method Flux Balance Analysis ( FBA; [29] ) . FBA looks for a flux distribution v that maximizes the objective function ( Equation 1 ) subject to steady-state , thermodynamic and growth medium constraints: ( 1 ) ( 2 ) ( 3 ) Equation 2 imposes the steady state constraints on the system , assuming that the metabolite concentrations remain constant in time . Thermodynamic constraints determining the reaction directionalities are accounted for via the flux limits and in Equation 3 . The uptake and secretion of a pre-defined set of metabolites from and to the environment is facilitated via the definition of exchange reactions in the stoichiometric matrix . The growth medium is defined via an upper bound on the glucose uptake exchange reaction ( as the carbon source ) and by allowing an unlimited uptake flux of oxygen , sodium , potassium , calcium , iron , chlorine , phosphate , sulfate and ammonia ( based on the RPMI- 1640 medium definition; as none of these substances can be used as a carbon source ) . Growth yield ( growth rate divided by the glucose uptake rate ) , oxygen uptake and lactate secretion rates were computed under a wide range of glucose uptake rates ( varying from 0 to 1 . 55 umol/mgDW/h , the uptake achieving maximal growth rate ) using Flux Variability Analysis ( FVA ) [41] , allowing us to determine minimal and maximal flux bounds on the reactions of interest . A constraint on the total enzyme concentration was added to the biomass production FBA model: The enzyme mass ( per mg dry weight ( DW ) of cells ) required to maintain the flux in the i-th reaction ( vi [mmol/ ( mgDW*h ) ] ) is given by the product of vi and the enzyme's molecular weight ( MWi [mg/mmol] ) divided by its turnover number ( [1/h] ) [21] . The limit on the total metabolic enzyme mass ( C = 0 . 078 [mg/mgDW] ) was estimated based on dry cell weight protein biomass measurements ( 0 . 779 [mg/mgDW]; [42] ) multiplied by the fraction of metabolic genes out of the total cellular protein mass , which was evaluated as the sum of metabolic gene expression readouts divided by the total sum of gene expression readouts ( [33]; equal to 0 . 1 ) . Notably , the reliability of this value was validated based on a recently published protein abundance dataset ( [43] , Text S1 ) . In order to account for positive fluxes only , each bidirectional reaction was split into two unidirectional reactions , resulting in a total of 4 , 894 reactions . Enzyme molecular weights were obtained from the BRENDA database ( [44]; Dataset S2 ) while turnover number data was taken from BRENDA and from the SABIO-RK databases ( [45]; Dataset S3 ) , and assigned as following: each reaction with a known Enzyme Commission ( EC ) number was queried against BRENDA for the maximal human wild-type kcat value . In case a human kcat value was not available , the maximal non-human wild-type turnover number was assigned . In case BRENDA data was not available , the SABIO-RK database was used in a similar manner . As a result , 729 reactions were assigned with kcat values while the other 4 , 165 reactions were assigned with the median kcat value across the set of known kcat values ( 25 1/s ) . Flux distributions were computed under maximal growth rates in the three growth phases ( phase I – 0 . 0243 1/h; phase II – 0 . 0515 1/h; phase III – 0 . 0557 1/h ) . For each phase , the median flux distribution across 1000 different uniform samples was calculated using ACHR sampling [46] . Mean pathway flux was calculated as the mean flux across the reactions belonging to the pathway of interest . Data on relative metabolomic measurements for lactate , and on relative transcriptomic measurements for genes which are important for mitochondrial biogenesis ( PGC-1-α , NRF-1 , TFAM and ATP5E ) was taken from [32] . Gene expression readouts for 1 , 269 metabolic genes across 60 cell lines from the NCI-60 collection [33] were correlated with enzyme concentrations predicted by ( i ) a stoichiometric only model and by ( ii ) a model accounting also for enzyme solvent capacity . Given a flux distribution vector v , for each reaction i , the enzyme concentration needed to maintain its flux ( the i-th entry in v ) was calculated as the product ofand the molecular weight of the enzyme catalyzing this reaction ( denoted ) , divided by its turnover number ( denoted ) , that is , . Total enzyme concentrations ( per gene ) were given by summing the enzyme concentrations across all of the reactions associated with the gene of interest ( i . e . reactions catalyzed by enzymes encoded by this gene ) , based on a gene-to-reaction mapping given in the human metabolic model . The Spearman correlation between the gene expression vector and inferred enzyme concentration vector was calculated for the two models in each of the 60 cell lines . The robustness of the results was validated against 1 , 000 uniformly sampled flux distributions from the solution spaces of the two models using ACHR sampling [46] . Each of the 20 amino acids was added , in turn , to the growth media , resulting in 20 different maximal biomass production rates calculated based on ( i ) the stoichiometric model , and on ( ii ) a model additionally accounting for the solvent capacity constraint . The maximal amino-acid uptake rate was set to the same uptake rate as glucose; the results are shown to be robust to the choice of this value ( Text S1 ) . | Cancer cells , as opposed to normal cells , produce a substantial amount of energy inefficiently via aerobic glycolysis , even in the presence of sufficient oxygen to support mitochondrial respiration . Despite the fact that this phenomenon , called the Warburg effect , has already been discovered back in 1924 , its causes remain poorly understood . Here we utilize a genome-scale human metabolic network model and show that by accounting for cellular capacity for metabolic enzymes , the Warburg effect is a direct consequence of cancer cells' adaptation to fast proliferation . We demonstrate that our model accurately captures several metabolic phenotypes observed experimentally during cancer development , as well as the preference of cancer cells to glutamine uptake over other amino acids . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"systems",
"biology",
"molecular",
"cell",
"biology",
"cell",
"biology",
"cell",
"growth",
"biology",
"computational",
"biology",
"metabolic",
"networks",
"genetics",
"and",
"genomics"
] | 2011 | Genome-Scale Metabolic Modeling Elucidates the Role of Proliferative Adaptation in Causing the Warburg Effect |
The mechanistic target of rapamycin ( mTOR ) is an established therapeutic target in renal cell carcinoma ( RCC ) . Mechanisms of secondary resistance to rapalog therapy in RCC have not been studied previously . We identified six patients with metastatic RCC who initially responded to mTOR inhibitor therapy and then progressed , and had pre-treatment and post-treatment tumor samples available for analysis . We performed deep whole exome sequencing on the paired tumor samples and a blood sample . Sequence data was analyzed using Mutect , CapSeg , Absolute , and Phylogic to identify mutations , copy number changes , and their changes over time . We also performed in vitro functional assays on PBRM1 in RCC cell lines . Five patients had clear cell and one had chromophobe RCC . 434 somatic mutations in 416 genes were identified in the 12 tumor samples . 201 ( 46% ) of mutations were clonal in both samples while 129 ( 30% ) were acquired in the post-treatment samples . Tumor heterogeneity or sampling issues are likely to account for some mutations that were acquired in the post-treatment samples . Three samples had mutations in TSC1; one in PTEN; and none in MTOR . PBRM1 was the only gene in which mutations were acquired in more than one post-treatment sample . We examined the effect of PBRM1 loss in multiple RCC cell lines , and could not identify any effect on rapalog sensitivity in in vitro culture assays . We conclude that mTOR pathway gene mutations did not contribute to rapalog resistance development in these six patients with advanced RCC . Furthermore , mechanisms of resistance to rapalogs in RCC remain unclear and our results suggest that PBRM1 loss may contribute to sensitivity through complex transcriptional effects .
Both everolimus and temsirolimus , analogs of rapamycin termed rapalogs , are FDA-approved and in common used for treatment of metastatic RCC based on seminal randomized clinical trials [1–3] . However , these drugs are known to cause disease stabilization in most cases , with a 5% objective response rate by standard RECIST criteria . The Phosphatidylinositol 3-kinase ( PI3K ) /AKT/mechanistic target of rapamycin ( mTOR] pathway plays a critical role in cell growth , differentiation , survival and metabolism . It is frequently activated in a variety of cancer types [4] , and new uses of rapalogs in combination with other therapies continue to be discovered [5 , 6] . mTOR is a serine threonine kinase which occurs in cells in two large multi-component complexes termed mTORC1 and mTORC2 [7 , 8] . mTORC1 is negatively regulated by the TSC protein complex which consists of TSC1 , TSC2 , and TBC1D7 , which converts the small GTPase RHEB into its inactive GDP-bound form . When both alleles of either TSC1 or TSC2 are mutated or lost , as is the rule in tumors occurring in individuals with the genetic disorder tuberous sclerosis complex , RHEB-GTP levels are high , leading to activation of mTORC1 . mTORC1 activity is also regulated by PI3K , AKT , MAPK , AMPK , growth factors , nutrient availability , stress levels and oxygen levels . Activation of mTORC1 leads to protein synthesis , lipid synthesis , nucleotide synthesis , autophagy inhibition , leading to cell enlargement and preparation for cell division [9] . Somatic mutations in MTOR which deregulate and activate its kinase [10 , 11] are known to occur in several cancer types , predominantly RCC in which mutation is seen in about 5% [12] . Activating RHEB mutations which activate mTORC1 are quite rare but also known to occur in cancer [13] . Rapalogs are allosteric inhibitors of mTORC1 through binding to FKBP12 , which binds to a specific domain of mTORC1 to inhibit its kinase activity . Previously we have reported that response to rapalog therapy in RCC is associated with mutation in the mTOR pathway genes: TSC1 , TSC2 , and MTOR [14] . Another recent study reported that mutations in PBRM1 were associated with response to rapalog therapy in the RECORD-3 trial [15] . To our knowledge no previous study has examined molecular mechanisms of acquired or secondary resistance of rapalog therapy in responding patients with RCC .
We identified six mRCC patients who developed resistance to rapalog therapy after initial clinical benefit , and who had available pre-treatment and post-treatment tumor samples . Five of these 6 patients had been studied in our earlier analysis of the association between mTOR pathway mutations and response to rapalogs in RCC [14] . ( However , note that the earlier study did not include analysis of post-treatment as well as pre-treatment samples , and was only gene panel sequencing , not whole exome sequencing . ) Five patients had clear cell RCC ( ccRCC ) and one had chromophobe RCC ( Table 1 ) . Most patients had received prior treatment with vascular endothelial growth factor targeted therapies ( n = 5 ) and received a rapalog in the second ( n = 3 ) or third ( n = 2 ) line setting . Five patients received treatment with everolimus and one with temsirolimus . The patients received rapalogs for a median of 9 . 5 months ( range: 5 . 5–46 months ) , after which they progressed . One ( the chromophobe RCC ) had a complete response , four had a partial response , and one had 10% tumor shrinkage ( Stable Disease ) . 434 somatic variants were identified in these six patients’ biopsies , including both pre-treatment and post-treatment samples ( S1 Table ) . The mutation profiles of the six individual tumors matched well with the expected genes and mutations from past studies in RCC . Three patients’ tumors showed mutations in TSC1 , while one had a mutation in PTEN , and some of these patients had been included in our previous study showing that there is enrichment for mutations in mTOR pathway genes in RCC patients who respond to rapalog therapy [14] . Recurrent mutations ( seen in more than one patients’ samples ) were seen in 10 genes ( Table 2 ) . VHL mutations were seen in all 5 ccRCCs , as expected . PBRM1 mutations were also seen in all 5 ccRCC samples . Three samples had inactivating mutations in TSC1 , as noted; three including the chromophobe RCC had TP53 mutations; while two had BAP1 mutations . Several genes with recurrent mutations were likely chance events , enhanced by their large size: DNAH11 ( 4516aa ) , TTN ( 34350aa ) , PIEZO1 ( 2521aa ) , TRPM6 ( 2022aa ) ; none are thought to be involved in the pathogenesis of any form of cancer . Furthermore , several of these mutations were silent , also suggesting that they were random events . Recurrent mutations in PTPRN2 may have also been due to random chance , and one of those was also silent . ABSOLUTE was used to calculate the tumor purity of the samples , and relative allelic frequency of each mutation in the cancer cells , termed clonal cell fraction ( CCF ) [16–18] . Phylogic was used to construct phylogenetic trees for the pre- and post-treatment pairs of samples , and to generate diagrams showing the major mutational events and clonal evolution ( S1 Fig ) . 46% of the somatic variants identified were clonal in both the pre-treatment and post-treatment samples , while 25% of variants were not seen in the pre-treatment sample but were clonal in the post-treatment sample ( S2 Table ) . VHL mutations were clonal or high subclonal in all 5 ccRCC samples , as expected , and showed no significant change in clonal representation in the paired samples . A TSC1 mutation went from subclonal to zero in one sample ( MT_002 ) after treatment , consistent with a model in which TSC1-mutation bearing cells might have been sensitive to rapalog therapy , and were selectively killed or inhibited by such therapy . However , in another sample ( MT_006 ) , an inactivating TSC1 mutation went from zero cancer cell fraction in the pre-treatment sample to clonal in the post-treatment sample , in direct contrast to this model . Mutations in each of TP53 and PTPRN2 also were enriched in one patient’s post-treatment sample but lost in another patient’s post-treatment sample . Multiple copy number ( CN ) events were also identified in these tumor samples ( S3 Table ) . All ccRCC samples showed loss of one copy of 3p , as expected , and the chromophobe sample showed loss of multiple chromosomes ( 1 , 2 , 6 , 8 , 10 , 11p , 13 , 17 , 21 ) . There were no focal amplifications identified in the post-treatment samples , but rather a variety of chromosome and arm level gains and losses of uncertain significance . PBRM1 mutations were clonal in both pre-treatment and post-treatment samples from two patients , but went from zero to clonal in 3 patients samples’ following treatment . Two of these acquired mutations were nonsense mutations in PBRM1 while the third was a synonymous change . PBRM1 mutations are seen in about 30% of ccRCC samples [19] , and it is possible that the appearance of PBRM1 mutations in the post-treatment samples was due solely to tumor heterogeneity and/or sampling issues . However , as noted above , VHL mutations were present at or near clonal frequency in all 5 ccRCC samples both pre- and post-treatment . Hence , the finding that PBRM1 mutations were present in all 5 ccRCC samples at the time of resistance , including acquisition of mutations in 3 ccRCC samples led us to explore the hypothesis that PBRM1 loss is a mechanism of resistance to rapalog therapy in ccRCC . We studied several ccRCC cell lines with native wild type ( 786-O , SNU-349 ) or native mutant PBRM1 ( A704 , RCC4 ) [20–22] . In 786-O cells , 4 different shRNA clones ( sh889 , sh890 , sh994 , sh326 ) were used to generate stable lines with reduced PBRM1 expression ( Fig 1A ) . The 3 lines with greatest knock-down showed somewhat variable growth rates in comparison to a similarly derived line expressing a control shRNA , and all lines showed moderate growth inhibition in response to rapamycin treatment at 20 nM for up to 3 days with no difference between the PBRM1 knock-down cells and those expressing control shRNAs ( Fig 1B ) . Similar results were obtained from SNU349 cells with stable downregulation of PBRM1 ( S2 Fig ) . We also studied RCC4 cells that are known to have biallelic mutation in PBRM1 that leads to a complete loss of expression of the protein [21] . We used derivative RCC4 cell lines expressing either control vector or wild type VHL [20] . We observed significant growth inhibition of both RCC4-vector and RCC4-VHL cells in response to each of rapamycin and Torin1 in 96well plate assays ( S3 Fig ) . This also suggested that PBRM1 loss did not lead to rapamycin resistance . We also performed the converse experiment , in that we examined rapalog sensitivity of derivatives of a native PBRM1 null cell line ( A704 ) expressing either empty vector ( A704_EV ) , wild type PBRM1 ( A704_WT ) , or a mutant Q1298* PBRM1 ( A704_Q1298* ) under regulation of doxycycline [23] ( Fig 2A ) . There were minor differences in the growth rate of these various A704 derivative lines , but there was no appreciable difference among them in response to rapamycin treatment for up to 6 days with and without doxycycline induction ( Fig 2B and 2C ) . We also examined the growth of these various A704 sublines in a clonogenic assay under rapamycin treatment for 30 days ( Fig 2D and 2E ) . Different numbers of colonies were seen for the different A704 derivative lines that varied to a small extent with and without doxycycline . All three lines showed a significant reduction in colony growth in response to rapamycin , and for the A704_WT line , a similar reduction in colony number was seen with and without doxycycline induction of wild type PBRM1 . One hundred twenty-nine genes showed a significant increase in the cancer cell fraction for a single mutation in a single post-treatment tumor sample in comparison to the pre-treatment sample ( S4 Table ) . To assess whether these genes were enriched in a pathway that might be consistent with resistance to rapalogs , we performed gene set enrichment analysis using hallmark gene sets [24] . The 129 genes showed modest overlap with two hallmark gene sets ( E2F targets , and mitotic spindle ) , 4 genes each with FDR q = 0 . 044 , and no enrichment for any other hallmark gene set . None of these genes were obvious members of the PI3K-AKT-mTOR signaling cascade . Some were known cancer genes , including CDKN2A , KEAP1 , MYCN , PLK4 , SETD2 , TP53 . It is possible that any of these singleton genetic changes contributed to resistance to rapalog therapy in an individual patient .
In this study , we sought to identify molecular mechanisms of secondary resistance to rapalog therapy in patients with RCC who had demonstrated initial clinical benefit ( 5 of 6 had PR/CR ) . Our analyses were limited by both the relatively small number of samples available to us ( n = 6 ) , and that only one of the six patients had experienced a durable CR , whereas four had PRs only , and one had prolonged stable disease . Furthermore , tumor heterogeneity is well-known in RCC [12 , 25–27] , and complicates interpretation of genetic differences seen in the two paired tumor samples . 63 to 69% of all somatic mutations were not detectable across every tumor region of RCC when multiple samples were analyzed [25] , similar to our findings that 46% of somatic mutations were not seen in both of our paired samples for these 6 patients ( S2 Table ) . Consequently the 129 genes with somatic mutations that were enriched ( CCF increased by > 0 . 5 ) in a single post-treatment sample likely reflect that heterogeneity , and are each unlikely to contribute to rapalog resistance . However , it is possible that some of those ‘acquired’ mutations seen only in the post-treatment samples may have contributed to resistance . PBRM1 was the only gene to show gain of mutation in more than one patient in the post-treatment sample . Two ccRCC patients showed gain of a nonsense mutation in PBRM1 , while a third showed gain of a synonymous mutation , and all three were clonal in the post-treatment sample . However , through analysis of RCC cell lines with both native PBRM1 expression , and those with native loss of PBRM1 , we could find no evidence of differential sensitivity to rapamycin therapy in standard and clonal growth assays at the standard dose of rapamycin 20nM , which is similar to serum trough levels of this compound achieved in patients with standard dosing , 10-15nM . It remains possible that the effects of PBRM1 loss with respect to rapalog sensitivity are not modeled well in tissue culture systems , and that this genetic event still contributed to rapalog resistance in patients . Alternatively , tumor heterogeneity or sampling issues may account for these PBRM1 mutations seen only in the post-treatment samples . Interestingly , Hsieh et al . reported that PBRM1 mutations were associated with longer progression free survival ( PFS ) in metastatic RCC patients treated with first-line everolimus in the RECORD-3 trial [15] . Hence it is possible that the finding of PBRM1 mutations in our 5 ccRCC patients correlates with response to rapalog therapy , though not seen in the pre-treatment tumor specimen , rather than representing a mechanism of resistance . On the other hand , if PBRM1 mutations cause response , then loss of PBRM1 mutation might be expected in the post-treatment resistance sample , given tumor heterogeneity in RCC , and we did not observe this . We were somewhat surprised that we did not identify any secondary mutations in MTOR in these samples . Previous studies have identified MTOR mutations capable of preventing each of rapalog and ATP-competitive kinase inhibition of mTOR kinase activity in vivo and in vitro [28 , 29] . Furthermore , a variety of activating mutations in MTOR are well-known in both RCC and other cancer types [10–12 , 27 , 30] , and in some cases are associated with exceptional response to rapalog therapy . Nonetheless , no MTOR mutations were seen in any of these six patients , nor were MTOR mutations associated with resistance development in even a single case . Hence , we conclude that mechanisms of resistance to rapalog therapy in RCC are not easily explained by mutations in most cases , and likely depend on more subtle transcriptional and/or epigenetic changes . Transcriptional effects of PBRM1 mutation have recently been identified in analysis of the association of response of ccRCC to immune checkpoint therapies [31] , and may have a similar effect in enhancing response to rapalogs .
This research study was approved by the Dana Farber/Harvard Cancer Center Office for Human Research Studies , protocol 07–336 , and all subjects provided written informed consent . We identified patients with metastatic RCC who initially responded to treatment with rapalog for at least 5 months , and then showed progressive disease , with available pre- and post-treatment ( at time of progression ) biopsies through a search of our own medical facilities and national and international collaborators . The six patients were treated with temsirolimus or everolimus at one of three medical centers: Dana-Farber Cancer Institute , Beth Israel Deaconess Medical Center , and the University of Utah Hospital . From each patient , we collected 1 ) a pre-treatment nephrectomy specimen , 2 ) post-treatment metastatic tumor specimen , and 3 ) a venous blood specimen . An expert genitourinary pathologist ( SS ) reviewed hematoxylin and eosin stained slides . For each case regions containing at least an estimated 50% tumor cells were selected for DNA extraction . Selected tumor areas were scraped off unstained slides and DNA was extracted using the QIAamp DNA FFPE Tissue Kit ( QIAGEN , Valencia , CA ) , according to the manufacturer guidelines . Whole exome sequencing ( WES ) was performed at the Broad Institute following standard protocols . Sequencing data was analyzed using standard analytic pipelines deployed in the Firehose environment . Mutect and Indelocator were used to identify somatic mutations in tumor-normal pairs . Every single mutation that was called by these pipelines was scrutinized using IGV to assess the reliability of the variant call , and to confirm allele frequencies seen in the various samples . Many variants were discarded due to misaligned reads , repetitive sequence tracts , low quality base or read scores , or reads seen in only a single direction . Recapseg and AllelicCapseg were used to determine copy number profiles . ABSOLUTE was used to estimate sample purity and ploidy , absolute copy number for each chromosome and segment , and clonal cell fraction ( CCF ) values for each mutation [16 , 17] . Phylogic was used to perform Bayesian clustering of mutation CCFs , and to construct phylogenetic trees for the pre- and post-treatment samples , as described previously [17 , 18] . We studied RCC cell lines SNU349 , 786-O , RCC4 , and several versions of the A704 cell line ( A704+BAF180_WT , A704+BAF180_Q1298* , A704+BAF180_EV ) previously generated by one of the co-authors ( WG ) [19] . Stable PBRM1 knock down was performed using four different shRNAs ( Sigma ) in lentiviruses following standard methods; reduced PBRM1 expression was confirmed by SDS-PAGE and immunoblotting of cell lysates . Cell growth assays were performed for the indicated time points in clear 96-well plates for Crystal Violet staining or white opaque 96-well plates ( Corning ) for cellular ATP measurement using Cell Titer Glo ( Promega ) . Rapamycin was used at 20nM and compared with vehicle ( DMSO ) treatment . Clonogenic cell proliferation assays were performed by plating 200 cells in 10 cm dishes ( n = 3 for each cell line/condition ) , treating them every 3 days with Rapamycin ( 20nM ) or corresponding DMSO control , with and without doxycycline ( 1ug/ml ) for 30 days , and then counting visible colonies without magnification following Crystal Violet staining . | Mammalian target of rapamycin ( mTOR ) inhibitors , everolimus and temsirolimus , are FDA-approved for treatment of metastatic renal cell carcinoma ( mRCC ) , but molecular mechanisms of acquired or secondary resistance to these agents are unknown . We evaluated six mRCC patients with available pre-treatment specimens who were treated with mTOR inhibitors and had a good clinical outcome , and then had a second biopsy at the time of resistance . We found that mutations in PBRM1 appeared to be enriched in post-treatment samples . However , modulation of PBRM1 levels in vitro in cell lines had no apparent effect on rapalog sensitivity . We conclude that mechanisms of resistance to rapalog therapy in RCC are not easily explained by gene mutations in most cases , and may depend on more subtle transcriptional and/or epigenetic changes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"antimicrobials",
"renal",
"cell",
"carcinoma",
"medicine",
"and",
"health",
"sciences",
"cancer",
"treatment",
"carcinomas",
"drugs",
"cancers",
"and",
"neoplasms",
"genitourinary",
"tract",
"tumors",
"microbiology",
"alleles",
"antimalarials",
"oncology",
"mutation",
... | 2018 | Mechanisms of acquired resistance to rapalogs in metastatic renal cell carcinoma |
Identification and characterization of CD8+ and CD4+ T-cell epitopes elicited by HIV therapeutic vaccination is key for elucidating the nature of protective cellular responses and mechanism of the immune evasion of HIV . Here , we report the characterization of HIV-specific T-cell responses in cART ( combination antiretroviral therapy ) treated HIV-1 infected patients after vaccination with ex vivo-generated IFNα Dendritic Cells ( DCs ) loaded with LIPO-5 ( HIV-1 Nef 66–97 , Nef 116–145 , Gag 17–35 , Gag 253–284 and Pol 325–355 lipopeptides ) . Vaccination induced and/or expanded HIV-specific CD8+ T cells producing IFNγ , perforin , granzyme A and granzyme B , and also CD4+ T cells secreting IFNγ , IL-2 and IL-13 . These responses were directed against dominant and subdominant epitopes representing all vaccine regions; Gag , Pol and Nef . Interestingly , IL-2 and IL-13 produced by CD4+ T cells were negatively correlated with the peak of viral replication following analytic treatment interruption ( ATI ) . Epitope mapping confirmed that vaccination elicited responses against predicted T-cell epitopes , but also allowed to identify a set of 8 new HIV-1 HLA-DR-restricted CD4+ T-cell epitopes . These results may help to better design future DC therapeutic vaccines and underscore the role of vaccine-elicited CD4+ T-cell responses to achieve control of HIV replication .
Globally , an estimated 36 . 9 million people were living with Human Immunodeficiency Virus ( HIV ) in 2017 , among them 59% were accessing antiretroviral therapy [1] . Introduction of combination antiretroviral therapy ( cART ) since the late 1990’s decreased HIV-related morbidity and mortality [2] and reduced the risk of HIV transmission [3] . However , cART requires a life-long adherence to treatment , has potential long-term side-effects [4] , and poor adherence can result in the development of resistance [5] . HIV infection has also been shown to result in more prevalent age-associated non-communicable comorbidities like peripheral arterial , cardiovascular disease , and impaired renal function as compared to HIV-uninfected controls [6] . Furthermore , among well-treated HIV-infected individuals ≥ 50 years without comorbidity or AIDS-defining events , the estimated median survival time remains lower than in the general population [7] . Importantly , cART is not able to eradicate HIV infection and the virus persists in long-lived reservoir cells [8] leading to the inevitable rebound of HIV within few days following cART interruption [9] . Thus , strengthened by recent scientific advances , the field is focused on development of strategies able to improve the immune control of HIV replication and to prolong the period of remission ( i . e . HIV control without cART ) leading to improvement of quality of life , decrease in morbidity and sparing costs . Vaccines are essential elements of these functional cure strategies . The rationale of therapeutic immunization relies on the induction of HIV-specific cellular responses through the improvement of the magnitude of preexisting T-cell immune responses elicited by the natural infection which are not able to control HIV replication [10] . A common endpoint in vaccine trials is measurement of the breadth and magnitude of vaccine-induced CD4+ and CD8+ T-cell responses . Identification and characterization of CD8+ and CD4+ T-cell epitopes , as well as the restricting HLA alleles can play a major role in elucidating the nature of protective cellular responses and mechanism of the immune evasion of HIV . CD8+ T-cell epitopes are typically 8-11-mer peptides restricted by HLA class I [11] and CD4+ T-cell epitopes are 12-20-mer peptides restricted by HLA class II [12] . A large number of HIV epitopes have been identified and deposited into various databases . The growing amount of class-I and class-II HLA / peptide-binding data generated by experimental methods has supported the generation of more sophisticated , computational tools which can predict CD8+ and CD4+ T-cell epitopes within viral proteomes by using different data-driven bioinformatic approaches including Artificial Neural Network ( ANN ) , Support Vector Machines , hidden Markov models , as well as other motif search algorithms . Using these HLA binding predictors can improve the efficiency of epitope mapping protocols in vaccine trials [13] . To achieve an optimal effect of therapeutic vaccination , the choice of antigens and vector is of critical importance . Several HIV-1 therapeutic candidate vaccines are currently under evaluation and one particularly considered approach is the use of dendritic cell ( DC ) -based vaccines [14] . Indeed , since DCs are the most powerful antigen-presenting cells ( APCs ) to initiate generation of antigen-specific CD4+ and CD8+ T-cell responses [15] , it has been suggested that ex vivo-generated DCs might be the most potent activator of T-cell responses avoiding the use of potentially toxic adjuvants . The DALIA phase I trial evaluated a new DC-based therapeutic vaccine that consisted of ex vivo-generated IFNα DCs loaded with 5 HIV-1 lipopeptides in 19 cART-treated HIV-1 infected patients [16] . These lipopeptides ( Gag 17–35 , Gag 253–284 , Nef 66–97 , Nef 116–145 and Pol 325–355 ) were composed of HIV-1 clade B regions rich in CD8+ and CD4+ T-cell epitopes [17–18] that have shown induction of cellular immune responses in a phase II clinical trial in healthy volunteers [19] . We have shown that DALIA vaccination was well tolerated and increased the breadth and functionality of HIV-1 specific immune responses . Moreover , the DALIA study design included a 6-month analytic treatment interruption ( ATI ) which allowed to show an inverse correlation between the peak of viral load after ATI and the frequency of polyfunctional CD4+ T-cell responses after vaccination [20] . In the present study , we performed more in-depth epitope mapping to dissect anti-HIV-1 specific immune responses in order to better characterize vaccine responses associated with the control of viral load post-ATI . We showed that viral control was associated with the breadth and magnitude of IL-2 and IL-13 CD4+ T-cell responses targeting both immunodominant and subdominant HIV-1 peptides included in Gag , Pol and Nef vaccine regions . These results highlight the potential use of IL-13 as a new biomarker of HIV-1 vaccine efficacy and may help to better design the future DC therapeutic vaccine trials .
We previously showed that DALIA DC-based therapeutic vaccination ( see Materials and Methods section and S1 Fig for study design ) induced and increased HIV-1 specific CD4+ and CD8+ T-cell immune responses and a broad repertoire of cytokine producing cells [20] . To better characterize these responses , we analyzed the production of cytokines before ( W-4 ) and after ( W16 ) vaccination in PBMCs stimulated with individual 15-mer peptides ( overlapping by 11 amino acids ) spanning the LIPO-5 vaccine regions using Luminex technology . Positive responses were defined as production of at least one detectable cytokine following individual peptide stimulation ( see Materials and Methods section and S1 Table ) . As shown in Fig 1A , at W16 a production of cytokines was observed in a higher proportion of patients as compared to baseline; IFNγ ( 94% at W16 / 56% at baseline ) , IL-2 ( 94% / 19% , p<0 . 05 ) , IL-13 ( 75% / 6% , p<0 . 05 ) , IL-21 ( 63% / 0% , p<0 . 05 ) and IP-10 ( 81% / 31% ) . As shown in Fig 1B , vaccination also significantly increased the breadth of responses . At baseline , the median number of epitopes inducing the production of at least one cytokine ( IFNγ , IL-2 , IL-10 , IL-13 , IL-17 , IL-21 or IP-10 ) were between 0–1 ( range 0–7 ) . After vaccination , these numbers were significantly increased ( p<0 . 05 ) for IFNγ , IL-2 , IL-13 , IL-21 and IP-10 ( median between 1 . 5–7 . 5 , range 0–28 ) . In contrast , neither IL-10 nor IL-17 responses were significantly increased after vaccination ( Fig 1A and 1B ) . Thus , DALIA vaccine regimen broadened breadth and improved functionality of HIV-specific immune responses targeting vaccine regions . We next correlated the breadth and the magnitude ( fluorescence intensity , FI ) of PBMC responses to individual HIV 15-mer peptides after vaccination ( W16 ) and peak of plasma HIV RNA in patients following cART interruption . We found an inverse correlation ( Spearman ) between the maximum of plasma HIV RNA values and the number of peptides eliciting IL-2 ( r = -0 . 58 , P = 0 . 020 ) , and IL-13 ( r = -0 . 66 , P = 0 . 007 ) production , and the magnitude of the response for IFNγ ( r = -0 . 51 , P = 0 . 047 ) , IL-2 ( r = -0 . 66 , P = 0 . 007 ) and IL-13 ( r = -0 . 71 , P = 0 . 003 ) . No correlation was observed for IL-21 or IP-10 ( Fig 1C ) . These results showed an association between the breadth and magnitude of vaccine-elicited functional responses at W16 and a lower viral rebound following ATI post vaccination . Next , we sought to characterize peptide-specific IFNγ , IL-2 and IL-13 responses further , since they were significantly associated with lower viral load after ATI . Before vaccination ( W-4 ) , only a weak cytokine secretion was observed ( S2 Fig ) . After vaccination ( W16 ) , we observed that each peptide induced a similar polyfunctional cytokine profile ( Fig 2A ) and that 6 peptides ( Gag17-2 , Gag253-3 , Gag253-4 , Gag253-5 , Gag253-6 , Pol325-5 ) induced a stronger response in at least half of the patients , either in FI ( Fig 2A ) or in concentration ( Fig 2B ) . These immunodominant peptides were derived from the HIV-1 Gag and Pol vaccine sequences and not from Nef . Individual concentration data for all peptide stimulations are provided in S2 Table . No secretion of IFN-γ , IL-2 and IL-13 after vaccination was observed with non-stimulated PBMC or PBMC stimulated with non-vaccine-specific HIV-1 peptides ( pools of 15-mer overlapping peptides from p2p6 Gag , outside the vaccine sequence ) and high secretion of these cytokines was observed using SEB stimulation ( S3 Fig ) . We then investigated whether peptide-specific secretion of IFNγ , IL-2 and IL-13 cytokines observed after vaccination ( W16 ) could be associated with viral dynamics after ATI . Thus , we correlated the magnitude of cytokine responses to the 36 individually tested peptides with the maximum viral load detected following cART interruption . For eighteen peptides , an inverse correlation ( Spearman ) between the maximum of HIV RNA values and magnitude of the specific response for at least one cytokine was found ( Fig 2C , peptides in bold ) . All of them , except N116-8 , passed positivity criteria described in Methods section ( labeled with * ) . Thus , responses elicited by the vaccination against 17 peptides of interest from Gag , Nef and Pol proteins were identified as potentially associated with control of viral load post-ATI . Interestingly , there was an overlap with 3 ( Gag253-3 , Gag253-4 , Pol325-5 ) of the 6 immunodominant peptides previously identified in Fig 2A and 2B ( highlighted within red rectangles ) . Intracellular Cytokine Staining ( ICS ) were assessed to analyze post-vaccination HIV-specific CD4+ and CD8+ T-cell responses from 14 patients . Cytometry dot plots from one representative individual ( D7 patient ) showing an HIV-specific CD4+ T-cell response with G253-4 peptide stimulation ( IFNγ and IL-13 ) , an HIV-specific CD8+ T-cell response with N116-6 peptide stimulation ( IFNγ and IL-2 ) , and a dual HIV-specific CD4+ and CD8+ T-cell response with P325-5 peptide stimulation ( IFNγ and IL-2 ) after vaccination are represented in Fig 3A . A total of 117 CD4+ and 46 CD8+ cytokine-producing T-cell responses directed to 29 and 22 peptides , respectively , were observed ( Fig 3B ) . Although all vaccine regions induced CD4+ and CD8+ T-cell responses , Gag 253–284 was the main targeted region for CD4+ T-cell responses and Nef 66–97 was the main targeted region for CD8+ T-cell responses . Immunodominant HIV 15-mer peptides previously identified in Fig 2A and 2B induced both CD4+ and CD8+ T-cell responses , but CD4+ T-cell responses were more frequent . Moreover , targeted peptides associated with lower viral rebound after ATI induced either CD4+ T-cell responses alone ( 4/16 peptides ) or both CD4+ and CD8+ T-cell responses ( 12/16 peptides ) . Interestingly , despite the fact that some peptides were not predicted to bind HLA-DR molecules of the patients according to NetMHCIIpan 3 . 2 server [21] , a significant frequency of CD4+ T-cell responses was observed ( S4A Fig ) . For 61 of these non-predicted HLA-DR CD4+ T-cell responses , we used single HLA-DRB1 transfected cell lines as APC to try to identify the restricting DRB1 alleles involved in the IFNγ-positive CD4+ T-cell responses . We were able to identify a total of 8 new HLA-DR restricted CD4+ T-cell epitopes covering 14 ( 23% ) of the 61 T-cell responses ( Table 1 and S3 Table ) . These epitopes were localized in HIV-1 Gag , Nef and Pol vaccine sequences . Interestingly , 3 of them were immunodominant ( Gag17-2 , Gag253-3 and Pol325-5 ) and 5 of them were associated with a lower viral rebound after ATI ( Gag253-3 , Nef66-5 , Nef66-6 , Pol325-5 , and Pol325-8 ) . For CD8+ T cells , only few responses were observed in comparison to the high frequency of patients whose HLA class I molecules were predicted to bind to the different 15-mer peptides according to NetMHCpan 4 . 0 server [22] ( S4B Fig ) . For 38 IFNγ-positive CD8+ T-cell responses observed and predicted , we used 8- to 10-mer peptides in ICS assays to identify the optimal CD8+ T-cell epitopes . A total of 30 CD8+ T-cell responses were identified ( 79% ) of which 6 ( 20% ) were directed towards an epitope shared by two different 15-mer peptides . So 24 unique IFNγ-positive CD8+ T-cell responses were observed , leading to the identification of a total of 18 different well-described CD8+ T-cell epitopes ( Table 2 ) . Thus , DALIA vaccine regimen induced more CD4+ than CD8+ T-cell responses and 8 new HIV-1 HLA-DR-restricted CD4+ T-cell epitopes with their restricting HLA-DRB1 allele were identified . Looking at the polyfunctionality of CD4+ and CD8+ T-cell responses induced by vaccination ( W16 ) using ICS , we observed that vaccine-specific T cells were mainly monofunctional IFNγ+ ( IFNγ+IL-2-IL-13- ) CD4+ and CD8+ T cells , and to a lesser extent IL-13+ ( IFNγ-IL-2- ) CD4+ T cells . Some CD4+ T cells were able to produce 2 cytokines ( IFNγ+IL-2-IL-13+ and IFNγ+IL-2+IL-13- CD4+ T cells ) , and only few IFNγ+IL-2+IL-13+ and IFNγ-IL-2+IL-13- CD4+ T cells were induced by vaccination ( Fig 4A ) . Since cytotoxic CD4+ T cells have already been described in HIV-1 seropositive patients [23–24] , we decided to characterize cytokine-positive CD4+ and CD8+ T cells using cytotoxic and degranulation markers such as granzyme A ( GRZA ) , granzyme B ( GRZB ) , perforin ( Perf ) and CD107a . We observed that CD4+ T-cells induced by vaccination were mainly non-cytotoxic ( CD107a-GRZA-GRZB-Perf- ) , and to a lesser extent able to express GRZA and GRZB or CD107a , but not perforin . Conversely , CD8+ T cells exhibited mostly a cytotoxic phenotype ( CD107a+GRZA+GRZB+Perf+ and CD107a-GRZA+GRZB+Perf+ ) ( Fig 4B ) . Cytometry dot plots from one representative individual ( D18 patient ) showing cytotoxic profile of HIV-1-specific CD4+ and CD8+ T cells stimulated with G253-4 peptide after vaccination are represented in Fig 4C .
We report that vaccination with ex vivo-generated DCs pulsed with HIV-1 long peptides ( HIV-1 lipopeptides ) induced and/or expanded HIV-1-specific CD4+ T cells secreting IFNγ , IL-2 and IL-13 , and also CD8+ T cells producing IFNγ , perforin and granzymes . All five HIV-1 peptide regions included in the vaccine induced CD8+ T-cell responses . Previous data have shown that these HIV-derived long peptides were enriched in CD8+ T-cell epitopes and that lipopeptide vaccination either alone or combined with other agents induced CD8+ T-cell responses in HIV-negative or HIV-positive individuals [17 , 19 , 25] . Interestingly , we previously showed that the DALIA vaccine elicited predominantly IFNγ+ and TNFα+ CD8+ T cell responses [20] , and here we showed that these CD8+ T cell responses also exhibit cytotoxicity markers , thus indicating a broader functional profile in contrast to those elicited by the administration of HIV-1 lipopeptides alone or in combination with a canarypox vector ALVAC [26] . Interestingly , a comparison of epitopes recognized by vaccine-elicited CD8+ T cells showed that among the 18 optimal CD8+ T-cell epitopes identified in our study , four belongs to the nine most promiscuously presented HIV epitopes described by Frahm and colleagues [27] . To provide global coverage of HIV-1 population diversity , mosaic vaccines containing conserved regions from different HIV-1 clades were developed [28] . It is noteworthy that RMYSPTSI , one of the five Gag conserved epitopes in the tHIVconsvX therapeutic vaccine , which had ability to suppress replication of circulating HIV-1 in HIV-1-infected individuals [29] , was present in the DALIA vaccine sequence ( Gag 253–284 region ) . Furthermore , three immunodominant HIV-1 clade B epitopes identified in our study have the ability to be recognized by CD8+ T cells from CRF02_AG-infected Ivorian patients [30] and two others were shown to be targeted by CTL from HIV-resistant Kenyan sex workers who had frequent exposure to HIV-1 clades A , C and D [31] . However , it is noteworthy that despite this broad repertoire and a highly cytotoxic phenotype of CD8+ T-cell responses , we did not observe any association between these responses and the control of HIV replication in patients following ATI . One limitation of our study is the lack of demonstration of the real killing capacity of these cells , which was precluded for technical reasons . Nevertheless , assuming that it is likely that a therapeutic epitope-based vaccine certainly needs to contain different CD8+ T-cell epitopes able to bind many different HLA haplotypes , in order to circumvent HIV-1 mutations , our data demonstrate that our DALIA platform is an efficient immunogenic vaccine strategy . Moreover , these data reinforce the need to combine immunotherapeutic strategies along with a vaccine to achieve a functional control of HIV replication . Our in-depth analysis extended previous results characterizing the HIV-specific CD4+ T-cell responses elicited by the DALIA vaccine . These results are supported by several studies showing that robust HIV-1-specific CD4+ T-cell responses are associated with natural control of primary HIV-1 infection [32] , predictive of persistent AIDS-free infection [33] and control of viremia in long-term non-progressors [34] . Furthermore , loss of non-progressor status was strongly associated with undetectable or declining Gag p24 CD4+ T-cell responses [35] . We described an inverse correlation between the breadth and magnitude of vaccine-induced IL-2 responses against individual 15-mer peptides and the maximum of viral load after ATI . The importance of IL-2 production by HIV-specific CD4+ T cells was already demonstrated , since IL-2 production by HIV-specific CD4+ T cells was shown to be associated with the persistence of a stable T-cell memory compartment [36] , and HIV-specific CD4+ lymphoproliferative responses were shown to enhance ex vivo proliferative activity of HIV-1-specific CD8+ T cells [37] . Moreover , IFNγ+IL-2+ CD4+ T cells have been associated with control of viremia in HIV- seropositive patients [38–41] , and Lu and colleagues found an inverse correlation between HIV-1 viral load and HIV-1-specific IFNγ and IL-2 producing CD4+ cells after vaccination of cART naïve HIV-1 individuals with a DC-based therapeutic vaccine pulsed with aldrithiol-2-inactivated HIV-1 [42] . Besides IL-2 responses , we also showed an inverse correlation between the breadth and magnitude of 15-mer peptides-mediated IL-13 responses and the maximum of viral load detected post-ATI . Similarly to the IL-2 , we showed that IL-13 was mostly produced by non-cytotoxic CD4+ T cells . IL-13 is considered a Th2 cytokine and is poorly studied in the HIV field . However , it has recently been shown that HIV-specific Th2 responses could predict HIV vaccine efficacy [43] and that Th2 responses induced after SIV vaccination were correlated with a decrease risk of SIV acquisition [44] . We have already observed IL-13 secretion after vaccination of healthy volunteers with LIPO-5 [45] but to our knowledge , the only other publication studying IL-13 secretion in a therapeutic HIV vaccine context showed an association between higher IL-13 secretion after vaccination and higher viral load after ATI [46] . These discrepancies could be explained by a difference in vaccine composition ( Gag/Pol/Nef lipopeptides-loaded activated DCs versus Gag p24 peptides + GM-CSF ) and a difference in cytokine measurement protocol ( 48h stimulation with 15-mer peptides versus 6 days stimulation with recombinant Gag p24 ) . In line with our results , it has been demonstrated that IL-13 inhibited HIV production in primary blood-derived human macrophages in vitro [47] and that IL-13 and IFNγ secretion by activated T cells in HIV-1 infection was associated with viral suppression and a lack of disease progression [48] . Moreover , IL-13 was shown to increase HIV-specific and recall responses from HIV-1-infected subjects in vitro by modulating monocytes [49] . Last but not least , IL-13 was also shown to be produced by seronegative subjects highly exposed to HIV [50] . All five HIV-1 lipopeptide sequences used to load the DALIA vaccine , in particular Gag 253–284 and Pol 325–355 regions , induced strong CD4+ T-cell responses . Accordingly , these five HIV regions were shown to induce CD4+ T-cell responses in non-HIV ( 19 ) or HIV-infected patients [25] vaccinated with HIV-1 lipopeptides , and some epitopes contained in these vaccine regions were previously described as present in a set of 11 immunogenic HLA-DR supertype CD4+ epitopes [51] or 31 immunodominant CD4+ T-cell epitopes [52] . We extended this repertoire by identification of 6 immunodominant 15-mer peptides elicited after DALIA vaccination: Gag17-2 ( EKIRLRPGGKKKYKL ) , Gag253-3 ( PVGEIYKRWIILGLN ) , Gag253-4 ( IYKRWIILGLNKIVR ) , Gag253-5 ( WIILGLKNIVRMYSP ) , Gag253-6 ( GLNKIVRMYSPTSIL ) and Pol325-5 ( EPFRKQNPDIVIYQY ) . The first five were already described as CD4+ T-cell epitopes [53] and listed in the HIV molecular immunology database ( http://www . hiv . lanl . gov/content/immunology/tables/helper_summary . html ) . Gag253-3 , Gag253-4 , Gag253-5 and Gag253-6 have been described as DRB1*04:01 restricted epitopes in a study using HLA DR4 transgenic mice [54] . Gag253-4 , Gag253-5 and Gag253-6 were also found to be promiscuous peptides , i . e . able to bind multiple HLA-DR alleles , using NetMHCIIpan 3 . 2 , the server we used to predict binding of 15-mer peptides to HLA Class II DRB1 alleles . Pol325-5 was not listed in the HIV molecular immunology database , nonetheless it is nearly totally contained ( 93% overlap ) in a Pol CD4+ T-cell promiscuous epitope identified by van der Burg and colleagues [55] . Of note , in some studies where the breadth of Gag-specific CD4+ T-cell responses was shown to be associated with lower viral load in HIV-infected individuals [56–58] , Gag17-4 , Gag253-4 , Gag253-5 , Gag253-6 and Gag253-7 were the main targeted epitopes . HIV-1 Nef-specific CD4+ T-cell responses are also important to control HIV-1 viral load and the Nef 66–97 peptide included in the DALIA vaccine was found among the 4 Nef epitopes previously associated with non-progression in HIV-1 infection [59] . In our study , we found that responses to seventeen 15-mer peptides of interest were potentially associated with lower viral load after ATI . Among them , Gag253-4 and Nef66-6 CD4+ T-cell epitopes were found in the studies mentioned above . All these data suggest that Gag253-4 is probably the most important HIV-1 sequence to be included in a therapeutic vaccine , and this is strengthened by its large capacity to bind to the 20 most frequent HLA-DRB1 alleles in the world population [60] . To study the reliability of the epitope prediction servers used in our study , we first looked whether the observed responses were predicted . Among the 117 cytokine-positive CD4+ T-cell responses detected in our study , only 35% were predicted to bind DRB1 molecules by NetMHCIIpan 3 . 2 server , but it is difficult to comment on the reliability of NetMHCIIpan 3 . 2 server because we only used HLA-DRB1 ( and not HLA-DP or -DQ ) molecules of the patients for epitope prediction with this server . By contrast , all 46 observed CD8+ T-cell responses were predicted to bind HLA class I A , B or C molecules of the patients by NetMHCpan 4 . 0 server . Then we looked whether the predicted responses were actually observed . Interestingly , only 10% of all predicted CD8+ T-cell responses were observed . A potential explanation could be that even the servers used in this study predict binding affinity of peptides to MHC molecules , affinity is only one of the determinant factors for a peptide to induce an immune response . Indeed , to be immunogenic , a peptide needs to be efficiently processed by the cell machinery and then presented by MHC molecules . Epitope flanking residues can influence the proteasome or peptidase processing of the antigen . TCR affinity for the peptide/MHC complex also has an important role on the induction of immune responses . For CD4+ T-cell responses , 51% of predicted responses were observed and we found 3 different patterns: 1 ) some peptides were predicted to bind HLA-DR molecules frequently expressed by the patients and for those we observed the majority of CD4+ T-cell responses ( this was the case for Gag253-4/5/6 ) ; 2 ) some peptides were predicted to bind HLA-DR frequently expressed by the patients and for those we observed only few CD4+ T-cell responses ( this was the case for Gag17-4 , Gag253-7 and Pol325-1/2 ) ; 3 ) some peptides were not predicted to bind HLA-DR molecules expressed by the patients , but for those we nevertheless observed a CD4+ T-cell responses ( this was the case for Gag253-3 , Nef66-4 and Pol325-4/5 ) . This latter result could easily be explained by the fact that these peptides could bind HLA-DP or -DQ alleles . However , another explanation could be that HIV-1 regions that bind some HLA class II alleles very weakly can induce CD4+ T-cell responses , and this is actually the case for the HIV-1 lipopeptides used in the DALIA vaccine , for which the hierarchy of CD4+ T-cell epitopes relies on the frequency of pre-existing peptide-specific T cells in healthy donors [61] . Recently , an association of HLA-DRB1 allele expression with HIV viral load profiles at a population level was reported [62] . DRB1*15:02 was shown to be associated with HIV control whereas DRB1*03:01 was shown to be associated with HIV progression . Here , we identified using a modified ICS protocol with single HLA DRB1 transfected cell lines as APC , in vaccinated patients , 8 new HLA-DR CD4+ T-cell epitopes restricted by different HLA-DRB1 alleles . Among them , Gag17-2 was restricted by DRB1*03:01 . Although , in our study , due to the small number of patients , we could not find any effect of these two HLA-DRB1 alleles on viral load dynamics after ATI , prediction analysis revealed that only two epitopes Gag253-4 and Gag253-5 could bind the DRB1*03:01 allele while Gag253-4 , Gag253-5 , Gag253-6 , Gag253-7 , Pol325-1 , Pol325-2 and Pol325-7 could bind the “protective” DRB1*15:02 allele . Our study has characterized epitope responses to a therapeutic HIV-1 vaccine beyond the “classical” analyses of responses to global HIV peptide pools . Our analysis benefited from the well described definition of HIV-1 peptides used for loading of the DC vaccine . This originality represents an advantage compared to similar studies using chemically or temperature inactivated whole autologous HIV or RNA-based pulsed DCs [42 , 63 , 64] . Regarding the need to significantly improve the breadth and the functionality of vaccine responses with the aim to better control HIV in the perspective of functional HIV cure , it is key to better identify the epitopes of interest . In contrast to several studies of these responses in natural infection , fine characterization of vaccine response at the level of individual peptides contained in therapeutic vaccine is lacking . We confirmed at the epitope level that the DALIA vaccine regimen induced CD8+ and CD4+ T-cell responses , and that IL-2 and IL-13 responses induced by the vaccine were mediated by CD4+ T cells . Beyond the dissection of precise epitopes , we described an inverse correlation between the breadth and magnitude of peculiar CD4+ T-cell responses producing IL-2 and/or IL-13 and a partial control of viral load after ATI . From a clinical stand point , these results may help to better design the future planned DC therapeutic vaccine trial . First , it is likely that a combined strategy to complement CD8+ T-cell responses ( immune checkpoint blockers , cytokines such as IL-7 or IL-15 ) to CD4+ elicited T-cell responses is warranted . Second , a particular attention will be paid to IL-2 and IL-13 CD4+ T-cell responses in relation to viral load dynamics after ATI . Since DC-based vaccines are difficult to make and will be challenging to use on a large scale , a new generation of vaccines targeting DC in vivo was developed using monoclonal antibodies directed against cell-surface receptors [65] . Thus , the sequences of the 5 lipopeptides used in the DALIA vaccine were recently fused with an anti-human CD40 antibody to target DCs in vivo . Targeting DC of HIV-1 infected patients in vitro with this new candidate vaccine ( αCD40 . HIV5pep ) showed induction of CD4+ T-cell responses towards Gag 253–284 and Pol 325–355 regions in a majority of patients [66] . Safety and immunogenicity of this vaccine construct administrated intradermally with poly-ICLC was recently demonstrated in rhesus macaques [67] . Moreover , in humanized mice infected with HIV-1 and treated with antiviral drugs , αCD40 . HIV5pep plus poly-ICLC vaccination rescued HIV-1-specific CD4+ and CD8+ T-cell responses and reduced the size of the HIV-1 reservoir , leading to significant control of HIV-1 rebound after cART interruption [68] . Thus , either ex vivo ( DALIA ) and in vivo ( anti-CD40 . HIV5pep ) DC-targeting platforms are promising immune interventions to combine with cytokines , immune checkpoint inhibitors , as well as broad neutralizing antibodies for an HIV cure [69] .
The ANRS/VRI DALIA was a phase I single-center single-arm clinical trial ( North Texas Infectious Diseases Consultants , Dallas , TX ) sponsored by Baylor Institute for Immunology Research and the French Agency INSERM-Agence Nationale de Recherches sur le SIDA et les hepatites ( INSERM-ANRS ) . Eligible participants were asymptomatic HIV-1-infected adults with CD4+ T cell counts >500 cells/μL and ≥25% of total lymphocytes , CD4 nadir ≥300 cells/μL , and no history of AIDS-defining events . Participants had to be on combination antiretroviral therapy ( cART ) with plasma HIV RNA <50 copies/mL at screening and within the previous 3 months . Participants received four vaccinations at weeks ( W ) 0 , 4 , 8 and 12 while on cART . Participants who had HIV-1 RNA <400 copies/mL at W22 were proposed to interrupt cART at W24 until W48 . Flowchart describing the ANRS/VRI DALIA clinical trial , as well as baseline characteristics of the patients , have been reported previously [20] and design of the DALIA trial is shown in S1 Fig . Twenty participants were screened and 19 enrolled in the trial between 2009 and 2010 . Among these 19 participants , three were not included in the present analyses due to lack of cryopreserved PBMCs to perform T-cell assays . Blood monocytes from each patient were obtained from an apheresis at W-4 while on cART . DC-based vaccines were generated by culturing blood monocytes with GM-CSF and IFNα for three days . Differentiating DCs were then pulsed for 24 hours with lipopeptides ( ANRS HIV-LIPO-5 ) : HIV-1 LAI ( clade B ) Gag 17–35 , Gag 253–284 , Pol 325–355 , Nef 66–97 and Nef 116–145 , activated during the last 6 hours with lipopolysaccharide ( LPS ) , purified under Good Manufacturing Practice guidelines , harvested and then frozen . Preclinical validation of the preparation and functional characterization of this vaccine has been previously reported [16] . Approximately 15x106 viable frozen-thawed HIV lipopeptide-loaded DC were injected subcutaneously in 3 separate injection sites ( 3 . 3 ml per site ) in the upper and lower extremities at each vaccination time point . The trial was approved by the IRB of Baylor Research Institute ( BRI ) and registered on ClinicalTrials . gov ( NCT00796770 ) . All participants provided written informed consent . Samples were sent to LABS Inc ( Centennial CO , USA ) for HLA Typing ( A , B , C , DRB1 and DQB1 ) using High Resolution Sequence Based Testing ( SBT ) . HLA characteristics of participants ( A , B , and C for HLA class I and DRB1 for HLA class II ) are reported in S4 Table . Fifteen-mer peptides ( n = 36 ) overlapping by 11 amino acids ( aa ) and covering the vaccine sequences were synthesized by Biosynthesis ( Lewisville , TX , USA ) and used at a final concentration of 2 μM ( sequences are listed in S1 Table ) . Eight- to ten-mer peptides synthesized by NeoMPS ( now PolyPeptide Laboratories France , Strasbourg ) were used at 2 μg/ml to identify optimal CD8+ T-cell epitopes . NetMHCIIpan 3 . 2 [21] server was used to predict binding of peptides to MHC class-II DRB1 molecules of the participants and their putative CD4+ T-cell responses . NetMHCpan 4 . 0 [22] was used to predict binding of peptides to MHC class-I ( A , B and C ) molecules of the participants and potential CD8+ T-cell epitopes . These prediction methods are based on ANN . For NetMHCIIpan 3 . 2 , the prediction values are given in IC50 values ( in nM ) and as %Rank . The percentile rank for a peptide is generated by comparing its score against the scores of 200 , 000 random natural peptides of the same length of the query peptide . Strong and weak binding peptides are identified based on %Rank , with customizable thresholds . We applied the default thresholds of 2% and 10% for strong and weak binders , respectively . For NetMHCpan 4 . 0 , the method is trained on a combination of more than 180 , 000 quantitative binding data and mass spectrometry derived MHC eluted ligands . Rank of the predicted affinity was compared to a set of random natural peptides . The peptide will be identified as a strong binder if the % Rank is below the specified threshold for the strong binders , by default 0 . 5% . The peptide will be identified as a weak binder if the % Rank is above the threshold of the strong binders but below the specified threshold for the weak binders , by default 2% . Frozen PBMCs from aphereses before ( W-4 ) and after ( W16 ) vaccination were thawed and rested at 37°C for 1h . Samples were filtered and resuspended in RPMI 1640 media supplemented with 10% Human serum ( HS ) . Plates containing individual overlapping peptides , positive ( Staphylococcal enterotoxin B ( SEB ) ) and negative ( no peptide ) controls , were thawed at room temperature and 1x106 cells were added in each well to a final volume of 300 μl . After 48h of culture , the plates were centrifuged , and 200 μl of each supernatant was collected and frozen at -80°C for cytokine detection . Measurements and analyses of secreted IL-2 , IL-10 , IL-13 , IL-17 , IL-21 , IFNγ and IP10 ( expressed in fluorescence intensity ( FI ) and concentration ) were performed using an “in house” multiplex bead-based technology ( Luminex ) assay with a Bio-Plex 200 instrument ( Bio-Rad ) after a two-fold dilution of supernatants . Human IFNγ , IL-2 , IL-10 , IL-13 , IL-17 , IL-21 and IP-10 monoclonal antibody reagent pairs were developed and validated by the Baylor Institute for Immunology Research ( BIIR ) . These pairs were conjugated to Luminex X-MAP SeroMAP microspheres according to manufacturer’s guidelines ( Luminex ) . Biotinylation was performed with EZ link Sulfo-NHS LC Biotin from Thermo Fisher . PhycoLink Streptavidin-RPE was purchased from Prozyme . This multiplex was assessed using a commercial 68-Plex cytokine standard cocktail produced expressly for the BIIR by BioLegend with protocols comparable to that for Millipore multiplex processing . The BIIR Luminex Core facility maintains compliance with the External Quality Assurance Program Oversight Laboratory , a National Institutes of Health and National Institute of Allergy and Infectious Diseases Division of AIDS program for quality assessment review and ratings of laboratories involved in HIV/AIDS research and vaccine trials around the world . Frozen PBMC from the apheresis at W16 were thawed and rested at 37°C for 3h in RPMI 1640 media with L-Glutamax supplemented with Penicillin / Streptomycin and 10% HS ( R-10HS ) . On day 0 ( D0 ) , 2x106 cells were cultured in 24 flat bottom wells culture plate with or without 2 μM of 15-mer peptides that induced a positive response by Luminex assay . IL-2 ( 10 U/ml ) was added on day 2 and half of the volume of each culture well was refreshed with fresh media containing IL-2 ( 10 U/ml ) at day 4 . On day 7 the cells were divided in two 5 ml polypropylene tubes and washed two times . D0 non-stimulated cells were then stimulated or not with SEB , and D0 stimulated cells were re-stimulated or not with the same peptide as D0 , in a 500 μl R-10HS final volume . For the cytotoxic panel , CD107a was also added during cell stimulation . One hour later , BD GolgiPlug and GolgiStop ( Becton Dickinson France ) were added and the culture was continued for additional 5 hours . Cells were then washed and stained for surface antibodies and live/dead marker for 20 minutes . Following fixation and permeabilization , cells were stained with intracellular antibodies for 30 minutes . Cells were then washed and resuspended in 250 μl PBS 1% PFA . To determine optimal CD8 responses , predicted CD8+ T-cell epitopes ( 8- to 10-mers ) were used in ICS assay . To determine the DRB1 HLA molecules involved in the observed CD4+ T-cell responses , 14 different DRB1 ( DRB1*01:01 , 03:01 , 04:01 , 04:02 , 04:03 , 04:04 , 04:05 , 07:01 , 11:01 , 11:04 , 13:01 , 13:02 , 14:01 , 15:01 ) single transfected DAP . 3 cell lines ( LJI , USA ) were pulsed for 1h with 15-mer peptides , washed four times and then added at D7 to D0-stimulated cells . Non-transfected DAP . 3 cell line was used as negative control . Two different ICS panels were tested , a first one for cytokine responses using anti-CD3 A700 ( BD ref 557943 ) , -CD4 BV605 ( BD ref 562658 ) , -CD8 efluor780 ( eBioscience ref 9047-0087-120 ) , -CD56 PECF594 ( BD ref 564849 ) , -IL-13 PE ( BD ref 340508 ) , -IFNγ PerCPCy5 . 5 ( BD ref 560704 ) , -IL-2 FITC ( BD ref 559361 ) , -IL-21 AF647 ( BD ref 560493 ) , Live dead fixable Aqua Dead ( Life Technologies ref L34957 ) , and a second for cytotoxic characterization of cytokine responses using anti-CD3 BV605 ( Biolegend ref 300460 ) , -CD4 BV785 ( Biolegend ref 300554 ) , -CD8 APC-H7 ( BD ref 560179 ) , -CD56 PC7 ( BD ref 557747 ) , -IL-13 PE ( BD ref 340508 ) , -IFNγ PECF594 ( BD ref 562392 ) , -IL-2 APC ( eBioscience ref 17-7029-41 ) , -Granzyme A Pacific Blue ( Biolegend ref 507207 ) , -Granzyme B AF700 ( BD ref 560213 ) , -Perforin PerCPCy5 . 5 ( Biolegend ref 353314 ) , -CD107a FITC ( BD ref 555800 ) , Live dead fixable Aqua Dead ( Life Technologies ref L34957 ) . Samples were analyzed with a BD LSRII and data were acquired with DIVA software . FlowJo 9 . 9 . 6 ( TreeStar , Inc . , Ashland , OR ) , Pestle 1 . 7 and Spice 5 . 35 ( M . Roederer , National Institute of Health ) softwares were used for data analysis . For a given cytokine , a Luminex quantification was considered as a positive response if the fluorescence intensity ( FI ) was > 90th percentile of all negative controls ( non-stimulated cells before and after vaccination for all participants ) and > 3 times the background ( median of three negative control wells for the considered participant and time point ) . An ICS response was considered positive for a given cytokine if the frequency of stimulated CD3+CD56-CD4+ or CD3+CD56-CD8+ cells was > 3 times the negative control ( unstimulated cells ) and >0 . 05% . Correlations between immunogenicity outcome at W16 and the maximum of HIV-1 RNA observed during ATI were analyzed by Spearman rank correlations . Comparison between frequency of responders before and after vaccination was performed using McNemar’s test and comparison between the breadth of the response observed before and after vaccination was performed using Wilcoxon matched-pairs signed rank test . Statistical analyses were carried out using GraphPad Prism ( version 8 . 0 . 1 , GraphPad Software , San Diego , USA ) . | Improvement of therapeutic vaccine strategies in the perspective of HIV cure is warranted and one of the determinant factors for elucidating the nature of protective cellular responses is the identification and characterization of CD8+ and CD4+ T-cell epitopes elicited by the vaccine . However , fine characterization of vaccine response at the level of individual peptides contained in therapeutic vaccines is lacking . Here , we report in-depth characterization of HIV-specific T-cell responses observed in antiretroviral therapy treated HIV-1 infected patients after vaccination with autologous ex vivo-generated IFNα DCs loaded with 5 long HIV peptides coupled to a lipid tail . Our study shows that dominant and subdominant epitopes from Gag , Pol , and Nef vaccine regions are targeted by HIV-specific CD8+ T and CD4+ T cells and that vaccine-elicited HIV-specific CD4+ T cells producing IL-2 and IL-13 are significantly correlated with a better viral control following treatment interruption . These results underscore the role of vaccine-elicited CD4+ T-cell responses to achieve control of HIV replication and thus may inform the design of better therapeutic vaccines . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"blood",
"cells",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"cytokines",
"pathogens",
"immunology",
"microbiology",
"retroviruses",
"viruses",
"immunode... | 2019 | Anti-HIV potency of T-cell responses elicited by dendritic cell therapeutic vaccination |
Lymphedema related to lymphatic filariasis ( LF ) is a disabling condition that commonly manifests in adolescence . Fifty-three adolescents , 25 LF infected and 28 LF non-infected , in age and sex-matched groups , using the Binax ICT rapid card test for filarial antigen were recruited to the study . None of the participants had overt signs of lymphedema . Lymphedema assessment measures were used to assess lower limb tissue compressibility ( tonometry ) , limb circumference ( tape measure ) , intra- and extra-cellular fluid distribution ( bioimpedance ) and joint range of motion ( goniometry ) . The mean tonometric measurements from the left , right , and dominant posterior thighs were significantly larger in participants with LF compared to participants who had tested negative for LF ( p = 0 . 005 , p = 0 . 004 , and p = 0 . 003 , respectively ) indicating increased tissue compressibility in those adolescents with LF . ROC curve analysis to define optimal cut-off of the tonometry measurements indicated that at 3 . 5 , sensitivity of this potential screening test is 100% ( 95%-CI = 86 . 3% , 100% ) and specificity is 21 . 4% ( 95%-CI = 8 . 3% , 41 . 0% ) . It is proposed that this cut-off can be used to indicate tissue change characteristic of LF in an at-risk population of PNG adolescents . Further longitudinal research is required to establish if all those with tissue change subsequently develop lymphedema . However , thigh tonometry to identify early tissue change in LF positive adolescents may enable early intervention to minimize progression of lymphedema and prioritization of limited resources to those at greatest risk of developing lifetime morbidity .
The mosquito-borne parasitic disease lymphatic filariasis ( LF ) is endemic in around 81 tropical countries , has a global burden of around 120 million cases , and is classified by the World Health organization as the second most common cause of long term disability after mental illness [1] . Three species of filarial parasites cause LF . Wuchereria bancrofti , the cause of Bancroftian filariasis , accounts for 90% of the cases worldwide . Brugian filariasis is caused by Brugia malayi , which is found in eastern Asia , and Brugia timori , which is confined to Timor and adjacent islands . All three species cause similar lymphatic disease but only Bancroftian filariasis causes hydrocele and all are controlled and treated by the same methods [2] . LF has a wide clinical spectrum ranging from debilitating acute bacterial dermatolymphgangioadenitis ( ADLA ) attacks , covert lymphatic and renal disease , and various degrees of lymphedema , to the terrible disfiguring , and often socially ostracizing , chronic manifestations of hydrocele and elephantiasis [3] . A global program to eliminate LF as a public health problem was introduced in 2002 [1] , [3] . It is based upon two “pillars . ” Pillar one is the interruption of transmission by the use of community-wide preventative chemotherapy ( PCT ) . This used to be called “mass drug administration” , MDA . The second pillar is the alleviation of suffering in those who already have chronic manifestations of the disease . Over time filarial infection can lead to the development of chronic lymphedema . Infection often occurs in childhood but the obvious clinical effects of the disease such as filarial lymphedema or hydrocele may not occur until they reach adolescence [4] . Shenoy et al [5] , [6] have shown covert abnormalities in the lymphatics with lymphoscintigraphy and worm nests by Doppler ultrasonography in B . malayi LF-infected adolescents as young as three years . Importantly , it has also been shown that these early lymphatic changes can be reversed by rug administration [7] . The mechanisms involved in the development of filarial lymphedema are not fully understood but they are known to be a complex interaction between the parasite and the host's immune system [8] , [9] . Filarial lymphedema is painful and debilitating , accompanied by skin changes , decreased joint range of motion and recurrent infections . The presence of childhood filarial lymphedema results in social problems including embarrassment , frequent absence from school and even discontinuation of studies [10] . It is generally accepted that acute dermatolymphangioadenitisis the main risk factor for progression of lymphedema and incidence of these attacks can be greatly reduced by the introduction of a basic “self care” program consisting of daily limb washing , treatment of skin entry lesions , antibiotic treatment of established infections , gentle exercise , appropriate footwear and limb elevation [11]–[13] . Although lymphoscintigraphy is an effective method for detecting covert lymphatic changes , it is invasive in that it requires the injection of a radioactive substance and is not suitable for use in field-based research and the monitoring and evaluation , or “fine tuning” of treatment programs for morbidity control . This study was undertaken to establish a non-invasive method to identify early , sub-clinical changes of lymphedema secondary to LF in the lower limbs of adolescents in Papua New Guinea ( PNG ) . PNG was chosen because of the high Bancroftian filarial antigen prevalence . There is a national plan for the elimination of LF but PCT has only been started in two out of 19 provinces , and there is currently no morbidity control component in place [2] .
Ethics approval was obtained from the PNG Medical Research Advisory Council ( MRAC number 08 . 25 ) and James Cook University Human Ethics Committee , Townsville , Australia . The proposed study was carefully explained to community leaders , school teachers , and health care workers in English , Hiri Motu and Tok Pisin . Parents provided verbal consent for their adolescents to participate . In this community of low literacy the use of verbal consent is appropriate and was approved by both the PNG Medical Research Advisory Committee and James Cook University Human Ethics Committee . The research was undertaken at Opau village , in central Province whilst a baseline survey for the PNG national LF program was conducted . Members of the village where inducted into the study based on their willingness to participate after the introduction ( tok save ) . As adolescents ( aged 10–21 years ) were inducted , their LF status was determined using the Binax ICT rapid card test for filarial antigen [2] and demographic data was collected . These individuals were divided into LF reactive and LF non-reactive groups by the investigators responsible for sampling and testing . Binax ICT testing was conducted according to the manufacturers' instructions . An age and sex matched sub-sample of each group were selected for participation . Verbal consent was sought and obtained from the parents of those inducted . The researcher responsible for all measurements of the adolescents was blinded to the results of the ICT test . There are many methods available to assess the lymphatic system . It was important that the methods included were inexpensive , portable and easy to apply in the field and therefore usable in the context of PNG . Hence the following methods were included in the study: Circumferential measures of each lower limb were undertaken using a tape measure following the protocol described by the Australasian Lymphology Association [14] . This provided a gross measure of limb size . Measurements were performed at the metatarsophalangeal joints ( MTP ) ( tape applied around the foot from MTP 1 to MTP5 ) , the foot adjacent to the leg ( tape applied around the dorsum and plantar aspect of the foot as close as possible to the leg ) , and 10 and 15 centimetres above and below the joint line of the knee . The distance from the knee was measured with a rigid ruler from the medial and lateral joint lines of the knee . Tonometry was used to measure alterations in tissue resistance . The tonometer consists of a plunger device which , when applied to the skin , provides a measure of the ability of the skin and underlying tissue to resist compression . A higher tonometry value indicates greater indentation and decreased tissue resistance . It has been used to investigate post-surgical lymphedema [15] and LF in adults with established stage II and III lymphedema [16] . Tonometry was undertaken using a Flinders Tissue Tonometer ( Flinders Medical Centre Biomedical Engineering , Australia ) consisting of a central plunger operating through a 6 cm diameter footplate that rests on the skin and applies a load of 200 grams . The degree of penetration of the plunger is measured by a micrometer on a linear scale [15] . The tonometer was calibrated according to the manufacturers' instructions prior to each measurement session . The length of the posterior thigh was measured from the gluteal fold to the posterior knee crease using a tape measure . The value was halved and this mid-point in the centre of the posterior thigh was marked . The same distance from the superior aspect of the patella was marked on the anterior thigh . The length of the calf was measured from the posterior knee crease to the base of the heel . This value was halved and a point marked on the centre of the posterior calf . The tonometer was placed on the skin at each marked point and readings to the nearest millimeter were recorded . Goniometry was used to measure joint range of motion ( ROM ) . When lymphedema is present and limb size increases the range of movement at joints can become restricted . Goniometry was undertaken for ankle dorsiflexion , knee flexion and hip flexion of each lower limb [17] . Bioimpedance spectroscopy ( BIS ) measures the opposition or impedance ( Z ) to the flow of an alternating electrical current passed through the body . The SFB7 used in this study performs measurements at 256 frequencies over the range 3–1000 KHz . . At low frequencies , <20 kHz , the current passes predominantly through the extra-cellular fluid ( ECF ) while at a high frequencies , >50–100 kHz , it passes through both the intra- cellular fluid ( ICF ) and the ECF . Thus the ratio of impedances , measured at high and low frequencies , provides an indication of change in fluid distribution between the tissue compartments . In secondary lymphedema , increased volumes of ECF are expected and hence the ratio of ICF to ECF should decrease . Since impedance is inversely related to fluid volume this would be observed as an increase in the impedance ratio Ri∶Re ( intracellular impedance: extracellular impedance1 ) . Evidence exists that bioimpedance spectroscopy can identify a change in fluid distribution in the arms of women at risk of lymphedema following management for breast cancer up to six months earlier than the usual clinical method of circumferential measurement [18] . Bioimpedance instruments are light and portable and may provide a more accessible and reasonable method to identify early accumulation of fluid in the limbs of adolescents . Tissue bioimpedance was measured for each lower limb using an Impedimed SFB7 bioimpedance spectrometer device ( ImpediMed Limited , Unit 1–50 Parker Court , Pinkenba , Qld , 4008 Australia ) . Adhesive electrodes were attached to the feet and hands of participants and the impedance recorded according to the manufacturer's instructions . All data was collected between the hours of 11am and 3pm over three consecutive days . All measurements were made with the subjects in the same position , supine . There is no reason to anticipate that BIS will perform any differently in a rural/village environment to elsewhere assuming a consistent procedure is used . However , as no reliability studies regarding the use of this equipment in a field environment were available two bioimpedance measures were undertaken for each limb . Bioimpedance values , goniometry and circumferential measures are altered by limb composition hence participants were asked which leg they would kick a ball with to determine limb dominance . Numerical variables were described using mean values and standard deviations ( SD ) when found to be approximately normally distributed . Median values and inter-quartile ranges ( IQR ) were used when the variable was skewed . Participants with and without lymphatic filariasis ( LF ) were compared using t-tests , Fisher's exact tests , and non-parametric Wilcoxon Mann-Whitney test . Binary logistic regression analyses were conducted to identify associations between lower limb measurements and LF . Demographic and body characteristics were considered as potential confounders . The model was adjusted for a confounder when the estimate changed by about 10% or more . Paired analyses comparing left and right limb were conducted using paired t-tests and paired non-parametric Wilcoxon signed rank tests . Throughout the analysis a significance level of 0 . 05 was assumed . Statistical analysis was conducted using PASW ( version 18 of SPSS; SPSS Inc . IBM; Chicago; Illinois ) .
In order to assess the reliability of the BIS measures we calculated the concordance correlation coefficient by Lawrence I-Kuei Lin [19] . The concordance correlation coefficient was high with 0 . 88 ( 95%-confidence interval 0 . 82 , 0 . 92 ) suggesting good reliability of the measurement ( Figure 1 ) . Mean age of the 53 participants was 16 . 5 years ( SD 2 . 5; range 10 to 21 years ) and 54 . 7% were female . Overall 47 . 2% tested positive for LF . The mean weight of the participants was 50 . 6 kg ( SD 7 . 1 ) and the mean body mass index ( BMI ) was 19 . 7 kg/m2 ( SD 1 . 9; range 14 . 5 to 23 . 6 ) . Four participants ( 7 . 5% ) were left leg dominant and one participant ( 1 . 9% ) would use both legs equally to kick a ball ( Table 1 ) . None of the demographic and body characteristics were significantly different for participants with and without LF ( Table 1 ) . The mean tonometric measurements from the left , right , and dominant posterior thighs were significantly higher in participants with LF compared to participants who had tested negative for LF ( p = 0 . 005 , p = 0 . 004 , and p = 0 . 003 , respectively; Table 2 ) . Logistic regression showed increasing the tonometric measurement from the dominant posterior thigh by 1 unit increased the odds of having tissue change secondary to LF by 2 . 2 ( 95%-CI = ( 1 . 2 , 4 . 1 ) ; p = 0 . 014 ) . This result was adjusted for the confounding effects of BMI . The mean circumferential measurements of the right and dominant thighs , 10 cm proximal to the knee , and of the left thigh , 15 cm proximal to the knee were significantly greater in participants with LF compared to participants who had tested negative for LF ( p = 0 . 038 , p = 0 . 042 and p = 0 . 043 , respectively; Table 2 ) . However when adjusted for the confounding effects of BMI those measurements were no longer significantly associated with LF ( p = 0 . 164 , p = 0 . 189 , p = 0 . 226 , respectively ) . There were no significant differences found in the impedance ratio of the legs between LF positive and LF negative subjects , irrespective of which leg was compared or of limb dominance ( Table 2 ) . However there was a significant difference in the bioimpedance ratios of the dominant and non-dominant legs of the LF negative group . This relationship was not identified in the LF positive group ( Table 3 ) possibly suggesting that BIS is detecting an LF-related change in ECW∶ICW ratios but that this is confounded by limb dominance effects upon the impedance measurements . These results support the use of tonometry of the dominant posterior thigh to indentify alteration in tissue compressibility in adolescents with sub-clinical lymphedema secondary to LF . Mean and standard deviation values for dominant posterior thigh tonometry in the positive LF group were 5 . 26 and 0 . 99 and in the negative LF group were 4 . 38 and 1 . 07 . Receiver operating characteristics ( ROC ) curves were plotted to identify the optimal cut-off for these measurements to differentiate between those with and those without altered tissue compressibility related to LF . Power when comparing tonometry of the dominant posterior thigh in adolescents who are LF positive and LF negative was 86 . 3% . ROC curve analysis to define optimal cut-off of the tonometry measurements of the dominant posterior thigh to identify early tissue changes is reported in Figure 2 and Table 4 . If we chose as the cut-off the 3 . 5 , then sensitivity of this potential screening test is 100% ( 95%-CI = 86 . 3% , 100% ) and specificity is 21 . 4% ( 95%-CI = 8 . 3% , 41 . 0% ) . All true LF cases are detected as such , but together with 22 false positive cases . The positive predictive value is 53 . 2% . However the false positive cases could be picked up in a second test .
Digital tonometry identified increased tissue compressibility in the posterior thigh in LF-infected adolescents with no overt signs of secondary lymphedema . As there was no significant concomitant change in BIS values these findings indicate a softening of tissue rather than altered intra- and extra-cellular fluid distribution as the first measurable LF tissue changes . This is in contrast to women who develop secondary lymphedema after breast cancer ( BC ) where alteration of extra- and intra-cellular fluid distribution is the earliest identifier of lymphatic change [18] . The difference in sensitivity of measurement tools in the LF and BC groups is likely to be related to the cause of lymphatic dysfunction . During intervention for cancer the removal of lymph nodes and direct trauma to the lymphatic tissue ( surgical , radiation ) are the main contributors to secondary lymphedema . In LF , dilation of the lymphatic vessels in response to the presence of the worm and the host inflammatory response to the living worm and their secreted antigens cause lymphatic damage [20] . Dilation of the lymphatics in this early stage is unlikely to lead to the accumulation of lymph seen in BC-related lymphedema and detected by BIS . Further , when the worm dies symbiotic Wolbachia organisms are released and introduce Wolbachia bacteria to the host which causes an inflammatory response in the lymphatic system [8] . It has recently been reported that living worms may also release bacteria and/or the products of the symbiotic Wolbachia into their host [21] . A study of Rhesus monkeys identified that they mount an antibody response to Wolbachia surface proteins that are temporarily associated with worm death and lymphedema development [8] . Specifically , increased Th1/Th17 responses and decreased regulatory T cells as well as regulation of Toll- and Nod-like receptors have been identified in the pathogenesis of filarial lymphedema [20] . Intra-subject between limb variation in BIS is expected due to limb dominance and consequent variation in limb mass and composition . This was not identified in the LF positive group . Research is required to determine if increased tissue compressibility is due to fatty infiltration , collagen breakdown , loosening of inter-cellular junctions or simply dilation of lymph vessels . Two sets of genetic risk factors have been suggested for the development of lymphedema , immune response genes associated with heightened inflammatory responses and lymphatic damage following filarial infection and genes associated with impaired lymphangiogenesis [8] . Thigh measures are likely to show the earliest change due to the heavy worm burden in the groin associated with LF . In adult males the worms have been identified by ultrasound to have a preference for the vessels associated with the spermatic cord [22] while in young boys they are found in the inguinal or other peripheral lymph nodes [8] and in the scrotal lymphatic vessels [23] . It has been suggested therefore that puberty and hormonal change may alter worm infestation . This may have contributed to why BIS did not detect differences in this study . The electrode placement in this study measured the impedance of the whole lower limb . As impedance is inversely proportional to the cross sectional area of the limb this results in the impedance of the whole leg being dominated by that of the smaller cross-sectional area of the lower leg , calf and below . Therefore even if impedance of the thigh is different the sensitivity to detect this difference is decreased when measuring the whole leg . BIS sensitivity should be improved by altering electrode placement and measuring the thigh region only . Tonometry is simple to learn , non-invasive , portable , does not require electricity or batteries and is relatively inexpensive ( <$500AUD ) . After a short training session ( 30 minutes ) , subsequent practice and establishment of intra-measurer reliability it would be suitable for use by community health workers . The training session should include the manufacturers requirements for calibration of the tonometer on a flat surface prior to measurement , placement of the tonometer in contact with the skin surface to be measured in a vertical position and reading of the measurement dial . As each measurement takes less than a minute one tonometer would allow a large number of people to be assessed . Tonometry could be used to screen for onset of tissue change , to monitor tissue change and the effect of interventions in those with established lymphedema . A digital tonometer is being developed that may allow even easier field use . Tonometry is most reliable when assessed on an even surface . This may account for the difference in findings for the posterior and anterior thigh tonometry measures as when lying in the prone position the posterior thigh provides a more even surface than the anterior thigh in the supine position . There are many factors which contribute to the development of lymphatic change in the LF population , many of which are not well understood . It is not known if all the positive LF adolescents identified in this study will undergo progressive lymphatic change . . Further research is required to track changes in tonometry , circumferential measures and BIS in these participants and better understand the progression of lymphatic and tissue change related to LF . This study suggests possible cut-off values for tonometry which may indicate tissue change characteristic of LF in an at-risk population of PNG adolescents . However , this should not be generalized to other populations and requires further research to establish its generalisability even within PNG . The cut-off chosen in this analysis results in low specificity and a high number of false positives . However the consequences of being wrongly treated are little in comparison to unidentified and unmanaged lymphedema . | The effects of lymphatic filariasis ( LF ) on the lymphatic system often become apparent during adolescence when the lower limb swells due to lymphedema and males develop hydrocele . Currently there is no simple or mobile field method to identify those at greatest risk of developing lymphedema or those with early subclinical lower limb change . Fifty-three adolescents , 25 LF infected and 28 LF non-infected were recruited to the study . The groups were compared with respect to lower limb tissue compressibility ( tonometry ) , limb circumference ( tape measure ) , intra- and extra-cellular fluid distribution ( bioimpedance ) and hip , knee and ankle joint range of motion ( goniometry ) . Tonometry , is a simple , inexpensive tool , which measures the distance a plunger will indent the soft tissues . Those adolescents who were LF positive had significantly increased soft tissue compressibility when assessed with tonometry than adolescents who were LF negative . Tonometry has high levels of sensitivity to identify adolescents who test positive to LF . If we are able to identify adolescents before they have overt symptoms , management practices to decrease disease progression can be implemented . This could prevent lifetime morbidity and allow allocation of scarce resources to those identified to be most at risk of developing lymphedema . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"diagnostic",
"medicine",
"public",
"health"
] | 2011 | Lymphatic Filariasis: A Method to Identify Subclinical Lower Limb Change in PNG Adolescents |
Cutaneous beta human papillomavirus ( HPV ) types are suspected to be involved , together with ultraviolet ( UV ) radiation , in the development of non-melanoma skin cancer ( NMSC ) . Studies in in vitro and in vivo experimental models have highlighted the transforming properties of beta HPV E6 and E7 oncoproteins . However , epidemiological findings indicate that beta HPV types may be required only at an initial stage of carcinogenesis , and may become dispensable after full establishment of NMSC . Here , we further investigate the potential role of beta HPVs in NMSC using a Cre-loxP-based transgenic ( Tg ) mouse model that expresses beta HPV38 E6 and E7 oncogenes in the basal layer of the skin epidermis and is highly susceptible to UV-induced carcinogenesis . Using whole-exome sequencing , we show that , in contrast to WT animals , when exposed to chronic UV irradiation K14 HPV38 E6/E7 Tg mice accumulate a large number of UV-induced DNA mutations , which increase proportionally with the severity of the skin lesions . The mutation pattern detected in the Tg skin lesions closely resembles that detected in human NMSC , with the highest mutation rate in p53 and Notch genes . Using the Cre-lox recombination system , we observed that deletion of the viral oncogenes after development of UV-induced skin lesions did not affect the tumour growth . Together , these findings support the concept that beta HPV types act only at an initial stage of carcinogenesis , by potentiating the deleterious effects of UV radiation .
Non-melanoma skin cancer ( NMSC ) is the most common cancer in adult Caucasian populations [1] . The cutaneous human papillomavirus ( HPV ) types belonging to genus beta are suspected , together with ultraviolet ( UV ) radiation , to be involved in NMSC [2 , 3] . The first two beta HPV types , 5 and 8 , were isolated from skin lesions of patients with a disorder called epidermodysplasia verruciformis ( EV ) . EV patients are highly susceptible to beta HPV infection in the skin and develop cutaneous squamous cell carcinoma ( cSCC ) at anatomical sites exposed to sunlight [4] . The fact that organ transplant recipients , due to their immunosuppressed status , have an elevated risk of beta HPV infection and development of cSSC provided evidence for the role of beta HPV types in skin carcinogenesis also in non-EV individuals [5 , 6] . Finally , many epidemiological studies support the link between these viruses and cSCC in the general population [2 , 3 , 7] . These studies showed that , compared with the general population , patients with a history of cSCC are more frequently positive for viral DNA in the skin and/or for antibodies against the major capsid protein L1 . Molecular analysis showed that not all cancer cells contain a copy of the beta HPV genome and that the copy number of the beta HPV genome is higher in pre-malignant actinic keratosis ( AK ) , a precursor lesion of SCC , than in SCC [8] . Thus , these data suggest that beta HPV types may act at an initial stage of skin carcinogenesis and that after full transformation of the infected cells , viral DNA can be lost . This model is consistent with the fact that additional carcinogens are involved in skin carcinogenesis . Considering that UV radiation is the key risk factor for cSSC development [9–11] , the most plausible hypothesis is that beta HPV types exacerbate the accumulation of a large number of UV-induced somatic mutations , facilitating cellular transformation . Subsequently , the expression of the viral oncogenes may become irrelevant for the maintenance of the malignant phenotype . Several studies in human keratinocytes , the natural host of beta HPV types , showed that E6 and E7 from some beta HPV types target key pathways linked to DNA repair , apoptosis , and cellular transformation [3] . Several transgenic ( Tg ) models for beta HPV have been generated [12–16] , some of which have highlighted the synergism between viral oncogene expression in the skin epithelium and UV radiation in promoting cSCC [3] . Tg mice expressing beta HPV38 E6 and E7 in the basal layer of the epidermis under the control of the cytokeratin K14 promoter ( K14 ) did not spontaneously develop any lesions during their life span . Upon long-term exposure to UV radiation ( 30 weeks ) , they developed first skin lesions closely resembling human AK and subsequently cSCCs . In contrast , wild-type ( WT ) mice developed neither pre-malignant lesions nor cSCCs when exposed to the same dose of UV radiation [15] . However , it is still unknown whether the high susceptibility of the K14 HPV38 E6/E7 Tg animals to UV-induced skin carcinogenesis is linked to the accumulation of mutations facilitated by the viral oncoproteins , which may become dispensable after cSCC development . In this study , we addressed this open question on the synergism between UV radiation and beta HPV38 E6 and E7 oncoproteins using the Tg mouse model . We showed that viral oncoproteins act at an initial stage of UV-induced skin carcinogenesis , facilitating the accumulation of a large number of somatic mutations in crucial genes that are associated with cSCC development in humans . In addition , silencing of the expression of the viral genes in established skin lesions does not affect further tumour growth .
We have previously shown that HPV38 E6/E7 expression in mouse skin strongly increases susceptibility to UV-induced carcinogenesis [15] . To evaluate whether the development of skin lesions present in K14 HPV38 E6/E7 Tg mice of chronic UV irradiation correlated with the number of accumulated DNA mutations , we used whole-exome sequencing of WT and Tg samples . For this analysis , we selected normal skin from WT mice not exposed or exposed to UV radiation for 30 weeks ( n = 2 ) and histologically confirmed skin specimens from three independent K14 HPV38 E6/E7 Tg mice UV-irradiated for 30 weeks , i . e . , ( i ) normal skin , ( ii ) pre-malignant skin lesions and ( iii ) cSCC . For the pre-malignant lesions , the histological analyses revealed that they have the classic features observed in humans of the precancerous condition of AK , including slight atypia , parakeratosis , and acanthosis ( S1 Fig ) [15] . Exome sequencing ( Illumina Hi-Seq ) of collected samples generated an average coverage of 141 . 71× ± 11 . 9 ( mean ± standard deviation ) . The genomic sequence of the WT mouse not exposed to UV radiation was used as a control sample in paired analysis . Only 10 mutations were detected in the skin of the UV-irradiated WT mouse . Similarly , less than 10 mutations were detected in the Tg mouse not exposed to UV irradiation . In both cases , all the mutations were in genes not directly linked to carcinogenesis ( S1 Table ) . In UV-irradiated Tg animals , the mutational load varied across our cohort of well-differentiated cSCC exomes , averaging 3541 somatic variants ( range , 3261–4027 ) or 68 . 58 ± 7 . 64 variants per Mb . The exome of the pre-malignant samples had substantially fewer variants , with an average of 1337 somatic variants ( range , 937–2026 ) or 23 . 14 ± 14 . 70 variants per Mb . The exome of the chronically UV-exposed normal skin of Tg mice harboured an average of 15 somatic variants ( range , 11–20 ) or 0 . 29 ± 0 . 08 variants per Mb ( S2 Table ) . Thus , the number of somatic mutations was proportional to the severity of the skin lesion; the average number in SCCs was approximately double that in the pre-malignant lesions ( Fig 1A ) . The vast majority of the somatic mutations detected in SCCs were C:G > T:A mutations , mutations that are also prevalent in the UV-induced mutational signature ( Fig 1B and 1C ) . We applied the non-negative matrix factorization ( NMF ) method to extract the mutational signatures composed of 96 single base substitution ( SBS ) types considering the sequence context ( one base upstream and one base downstream ) ( S2 Fig ) . The extracted signature was compared with known mutational signatures by the cosine similarity method [17 , 18] . The value of the similarity obtained for the new B signature is 0 . 86 for COSMIC signature 27 ( UV signature ) ( S2 Fig ) , indicating the clear prevalence of the impact of UV radiation on the etiology of these cSCCs . To assess the biological significance of the somatic mutations detected in the skin lesions of the K14 HPV38 E6/E7 Tg mice , we determined whether they were detected in the previously compiled lists of epi-driver and epi-modifier genes [19–23] , as well as genes identified in the Cancer Gene Census [24] . As shown in Fig 2 , three classes of genes were found to be recurrently mutated in pre-malignant and malignant skin lesions of K14 HPV38 E6/E7 Tg animals , suggesting a selective process for the enrichment of mutations in these groups of genes . Pathway analyses confirmed that the mutations detected in mouse cSCC affect key pathways intimately linked to cellular transformation ( S3 Table ) . A comparison of somatic mutations detected in our experimental Tg mouse model and in human cSCC [25] revealed that a large number of epi-driver , epi-modifier , and Cancer Gene Census genes were recurrently mutated in murine and human cSCC ( Fig 3A ) . A recent study identified the top human genes mutated in cSCC [26] . Interestingly , most of these genes are also found to be mutated in the UV-induced skin lesions of the K14 HPV38 E6/E7 Tg animals ( Fig 3B ) . In agreement with previous findings on human cSCC [25] , Trp53 showed up as the most mutated gene in the murine Tg-derived cSCC ( Figs 2A and 3B ) . Here , p53 mutations appear to be an early event in skin carcinogenesis , because they were detected in one sample of normal skin as well as in all pre-malignant lesions and cSCCs . In agreement with our data , it was reported that p53 mutations can be detected in keratinocytes of UV-exposed normal skin [27 , 28] . However , all mutations were identified in the p53 DNA-binding domain ( S4 Table ) , supporting their key role in the process of carcinogenesis . Consistent with the fact that in keratinocytes the Notch signalling pathway promotes cell-cycle exit and differentiation [29 , 30] , NOTCH1 and NOTCH2 have been found to be mutated in human cSCC [25] . In our Tg mouse model , mutated NOTCH1 and/or NOTCH2 were also detected in all three cSCCs , but never in pre-malignant lesions ( Fig 3B ) . Our previous data showed that HPV 38 E6 and E7 expression in human keratinocytes resulted in accumulation of TAp53 , which is recruited to the internal promoter located in intron 3 of p53 gene , with resulting transcriptional activation of ΔNp73α [31 , 32] . Fig 4 shows that also in the mouse skin , expression of the viral genes leads to increased ΔNp73α transcription . In contrast , in histologically confirmed pre-malignant and SCC lesions , p53 mutation correlates with a strong decrease in ΔNp73α mRNA levels ( Fig 4 ) . In conclusion , our findings show that the expression of HPV38 E6 and E7 oncogenes in mouse skin increases susceptibility to UV-induced cSCC by facilitating the accumulation of somatic mutations that have been clearly associated with skin cancer development in humans . Many studies support the role of beta HPV types , together with UV radiation , in the development of skin SCC [2 , 3] . However , in contrast to the mucosal high-risk HPV types such HPV16 that are required in all steps of cervical carcinogenesis , beta HPV types appear to have a role in the initial steps of carcinogenesis . To test this hypothesis , we constructed our K14 HPV38 E6/E7 Tg mice as a conditional expression model with two loxP elements , located immediately upstream and downstream of the viral genes [15] . Originally , we crossed the K14 HPV38 E6/E7 Tg mice with K14 Cre-ERT2 Tg animals overexpressing the Cre recombinase gene fused to a triple-mutant form of the human estrogen receptor that gains access to the nuclear compartment only after exposure to 4-hydroxytamoxifen ( TMX ) but not to the natural ligand 17β-estradiol , in order to silence E6/E7 expression by Cre-mediated deletion of the floxed viral genes at different times of the chronic UV irradiation , i . e . , different stages of SCC development . Although the expression of the viral genes could be efficiently silenced upon administration of TMX to 5-week-old K14 Cre-ERT2 HPV38 E6/E7 compound mice , in the compound mice a strong decrease in viral gene expression was observed during the 30 weeks of UV irradiation in the absence of TXM treatment ( S3 Fig ) . The loss of HPV38 E6 and E7 genes in long-term experiments was most likely due to a basal , non-specific Cre recombinase activity in the nucleus of mouse skin keratinocytes . None of the K14 Cre-ERT2 HPV38 E6/E7 Tg compound lines developed cSCC after 30 weeks of UV irradiation , further highlighting the importance of the viral proteins in UV-induced carcinogenesis . Therefore , we developed a different strategy to evaluate the requirement of HPV38 E6 and E7 genes for cancer maintenance ( Fig 5A ) . K14 HPV38 E6/E7 Tg mice were exposed to long-term UV irradiation , and after the appearance of well-defined skin lesions , after about 22–25 weeks of irradiation , two different DNA vectors were delivered by electroporation into the abnormal tissues . Because of the small size of the electroporated skin lesions , we could not perform any biopsy; therefore , we did not have any histological information about whether they correspond to pre-malignant or malignant lesions . Results obtained in several independent experiments showed that the lesions that occurred after 22–25 weeks of UV irradiation correspond to pre-malignant lesions or an early stage of cSCC [15 , 16] . Both vectors contain a scaffold/matrix attachment region ( S/MAR ) that keeps the plasmid in an episomal state , avoiding any integration-mediated toxicity , and ensures robust and persistent gene expression [33] . The vector codes for luciferase and Cre recombinase genes ( Cre-Luc ) separated by the P2A cleavage site , whereas the control vector expresses only a luciferase gene ( Luc ) . Luciferase was used to monitor the efficiency of transfection by non-invasive in vivo imaging , and Cre was used to induce the excision of the viral genes . A total of 23 lesions on 14 mice were transfected either with the Luc vector ( n = 9 ) or with the Cre-Luc vector ( n = 14 ) . When possible , the same mouse was injected with both vectors , each on a different lesion . Three representative mice are shown in Fig 5B . Luciferase activity was detected in the animals’ skin in each of the electroporated areas independently of the vector type . After electroporation , the animals were irradiated until the end of the 30-week UV irradiation protocol and closely monitored for several weeks to evaluate the progression of the skin lesions . No significant difference in tumour growth was observed in animals transfected with the Luc or Cre-Luc vectors ( Fig 6A ) . Histological analyses confirmed that 100% percent of the Luc-injected lesions and 93% of the Cre-Luc injected lesions ( 13 out of 14 ) evolved into invasive cSCC; a morphological examination revealed no major differences between the two groups of tumours ( Fig 6B ) . Detection of the viral RNA transcripts by RNA-RNA in situ hybridization confirmed that electroporation of skin lesions with the Cre-Luc vector , but not with the Luc vector , resulted in the loss of E6/E7 expression in large islands of cancer tissue ( Fig 6B ) . In conclusion , our findings show that after the accumulation of UV-induced DNA mutations and the development of skin lesions , the expression of the HPV38 E6/E7 genes is dispensable for the maintenance of the malignant phenotype of skin cancer cells .
Although the HPV family includes more than 200 types , to date only the mucosal high-risk ( HR ) HPV types have been clearly associated with human carcinogenesis . These viruses are the etiological agents of cervical cancers as well as a subset of other genital and oropharyngeal cancers [34] . Beta HPV types have been proposed to be associated with cSCC . They were initially linked to cSCC in EV patients , but now many epidemiological and biological studies support the role of beta HPV types in skin carcinogenesis also in non-EV individuals [3] . We have previously shown in a Tg mouse model that expression of beta HPV38 E6 and E7 in the skin strongly increases the risk of cSCC development upon UV irradiation [15] . Here , we showed that the higher susceptibility of K14 HPV38 E6/E7 Tg mice to UV-induced skin carcinogenesis tightly correlates with the accumulation of a high number of mutations in the keratinocyte genome . Remarkably , exposure of WT animals to the same doses of UV radiation did not lead to accumulation of DNA mutations and development of cSCC . These data suggest that the HPV38 oncoproteins can negatively affect the DNA repair machinery and/or immune pathways that lead to the elimination of damaged cells . We have recently shown that K14 HPV38 E6/E7 Tg mice are hampered in the production of interleukin 18 ( IL-18 ) during their exposure to UV radiation [16] . Upon UV irradiation and activation of the inflammasome , keratinocytes secrete high levels of cytokines from the IL-1 family , including IL-18 , thus inducing a broad spectrum of processes , such as infiltration and activation of inflammatory leukocytes , immunosuppression , DNA repair , and apoptosis [35–38] . Thus , it is likely that the high susceptibility to UV-induced DNA mutations and skin carcinogenesis of K14 HPV38 E6/E7 Tg mice may be linked to the negative impact of HPV38 on IL-18 production . Analysis of the mutational profile revealed that a large number of genes encoding for epi-drivers or epi-modifiers and proteins known to be associated with carcinogenesis ( Cancer Gene Census ) harbour missense or nonsense mutations . Most importantly , the gene mutation profile found in murine cSCC shows remarkable similarities to the mutational profile found in human cSCC . In particular , mutations in p53 appear to be an early event in murine and human skin carcinogenesis . We have previously shown that beta HPV38 E7 alters the p53/73 network by inducing accumulation of p53/p73 antagonist ΔNp73α [31 , 32] . In human keratinocytes expressing beta HPV38 E6 and E7 , ΔNp73α forms a transcriptional inhibitory complex , which binds a subset of p53-regulated promoters , preventing their activation in the presence of cellular stress [39] . Because the major role of p53 is to safeguard genome integrity , the high cancer susceptibility of K14 HPV38 E6/E7 Tg mice along with the high numbers of accumulated UV-induced DNA mutations can be explained , at least in part , by the properties of the beta HPV oncoproteins . However , once p53 , and likely other cellular genes , are irreversibly inactivated by DNA mutations induced by UV radiation , the progression and maintenance of the skin carcinogenic process could become independent of the expression of viral genes . In agreement with this view , ΔNp73α mRNA levels decrease strongly in UV-induced skin lesions of K14 HPV38 E6/E7 Tg animals after accumulation of p53 mutations . In addition , we observed that the deletion of the HPV38 E6 and E7 genes does not affect further growth of the tumour . In contrast , in K14 Cre-ERT2 HPV38 E6/E7 Tg the loss of the viral genes at early stages of the irradiation protocol prevents the development of UV-induced skin lesions , underlining the key function of HPV38 E6 and E7 in UV-mediated carcinogenesis . These findings in the K14 HPV38 E6/E7 Tg mouse model are in agreement with the studies on human skin lesions , supporting an early role of beta HPV types in skin carcinogenesis . Indeed , the copy numbers of the beta HPV genome appear to be higher in the pre-malignant lesion , AK , than in cSCC [8] . In addition , not all cancer cells contain a copy of a beta HPV genome [8] . Thus , the mechanisms of carcinogenesis induced by beta HPV types appear to be substantially different from those of the mucosal HR HPV types . In the case of the mucosal HR HPV types , the viral oncoproteins are the major drivers of cancer development ( e . g . in the cervix ) that , in addition , are required throughout the entire carcinogenic process ( Fig 7 ) . In contrast , UV-induced damage is the main carcinogen of cSCC . Here , however , beta HPV oncoproteins can facilitate the accumulation of UV-induced DNA damage but they are dispensable after full development of a malignant lesion ( Fig 7 ) . Why do different HPV types display different biological properties ? Cutaneous and mucosal HPV types infect cells at distinct anatomical sites exposed to different environmental stresses . Thus , it is not surprising that they have evolved with divergent biological properties . All HPV types rely on the DNA replication machinery of the host cell . Therefore , they must have developed several mechanisms to maintain the infected cell in a proliferative state to guarantee efficient viral genome replication . Exposure of skin keratinocytes to UV radiation leads to accumulation of DNA damage , which in turn induces cell-cycle arrest or apoptosis to allow repair or elimination , respectively , of the damaged cell . The cutaneous HPV types appear to be able to circumvent this adverse effect of UV radiation on keratinocyte proliferation , promoting the accumulation of damaged cells in the skin and , consequently , carcinogenesis . Our previous findings showed that different HPV38 E6/E7 expression levels in independent Tg lines influence the rate of SCC development [15] . Thus , it plausible to hypothesize that also in humans , the viral gene expression levels may have an impact on UV-induced skin carcinogenesis . Limited data are available on beta E6 and E7 gene expression in normal skin and pre-malignant and malignant skin lesions ( reviewed in [2 , 3] ) . There is no information on the different spliced forms of beta HPV genes and how they could determine a different efficiency in protein synthesis . Thus , additional studies are required in humans to corroborate the findings obtained in the Tg mouse model on the hit-and-run mechanism of HPV38 in UV-induced carcinogenesis . In conclusion , our findings in a Tg mouse model highlight a novel mechanism of infection-associated carcinogenesis , in which the virus is not the driving force but synergizes with UV radiation in promoting cSCC .
The transgenic animal model FVB/NTgN ( 38E6E7 ) 187DKFZ ( https://mito . dkfz . de/mito/Animal%20line/10954 ) has been previously described [15] . UVB irradiation was performed under sevoflurane anaesthesia , and every effort was made to minimize suffering . The animal facility of the German Cancer Research Center has been officially approved by responsible authority ( Regional Council of Karlsruhe , Schlossplatz 4–6 , 76131 Karlsruhe , Germany ) , official approval file number 35–9185 . 64 . Housing conditions are thus in accordance with the German Animal Welfare Act ( TierSchG ) and EU Directive 425 2010/63/EU . Regular inspections of the facility are conducted by the Veterinary Authority of Heidelberg ( Bergheimer Str . 69 , 69115 Heidelberg , Germany ) . All experiments were in accordance with the institutional guidelines ( designated veterinarian according to article 25 of Directive 2010/63/EU and Animal-Welfare Body according to article 27 of Directive 2010/63/EU ) and were officially approved by Regional Council of Karlsruhe ( File No 35–9185 . 81/G-64/13 and 35–9185 . 81/G-200/15 ) . To generate the Luc and the Luc-Cre vectors , the pS/MARt-GFP DNA vector was first digested with the restriction enzymes NheI and BglII to linearize the vector and eliminate the transgene GFP . The InFusion system provided by Clonetech was used to introduce the luciferase gene alone or in combination with the Cre recombinase gene to generate the vector pS/MARt-Luc or the vector pS/MARt-Luc-P2A-Cre , respectively . UVB irradiation was performed with a Bio-Spectra system ( Vilber Lourmat , Marne La Vallee , France ) at a wavelength of 312 nm as previously described [15] . Briefly , animals were anesthetized with 3% Sevorane ( Abbott , Wiesbaden , Germany ) in an inhalation anesthetizer ( Provet , Lyssach , Switzerland ) and placed in a covered compartment with an upper square opening ( 3×2 cm ) at a distance of 40 cm from the UVB lamp . To study UV-induced carcinogenesis , 7-week-old female FVB/N WT or K14 HPV38 E6/E7 Tg animals were shaved on the dorsal skin with electric clippers and irradiated 3 times a week for 10 weeks with increasing doses of UVB , starting from 120 mJ/cm2 to a final dose of 450 mJ/cm2 , with a constant weekly increase to allow skin thickening . For the following 20 weeks , mice were irradiated 3 times a week with 450 mJ/cm2 . The UV irradiation protocol was based on the data described in [40] and to mimic the situation in humans . For instance , the maximum dose of the UV irradiation protocol , 450 mJ/cm2 , corresponds to 50 minutes of solar exposure in July in Paris . The tumour incidence ( tumour bearers/group ) was recorded weekly . Tumours were identified first macroscopically and by histological diagnosis . After 30 weeks , or earlier if the tumour reached the ethically allowed maximal size , the animals were sacrificed and H&E-stained sections of dorsal skin were used for histological diagnosis . To study the effect of the loss of the viral genes on skin cancer development , 7-week-old K14 HPV38 E6/E7 Tg mice ( n = 14 ) were shaved on the dorsal skin and treated for 30 weeks with increasing doses of UVB as previously described [15] . As soon as skin lesions ( maximum diameter 2 . 6 mm ) became evident , 46 μg of pS/MARt-Luc or 50 μg of pS/MARt-Luc-P2A-Cre dissolved in isotonic saline solution was injected directly into the lesions . To facilitate the uptake of the injected DNA , an electric field was applied to the area of the injection site using a Tweezertrodes connected to a BTX ECM 630 generator ( Harvard Apparatus , Holliston , MA , USA ) . A first high-voltage electric pulse ( 1400 V/cm , 100 μs , 2 times ) , to induce temporary gaps in the keratinocytes cell membrane , was followed by a low-voltage electric field ( 140 V/cm , 400 ms , 2 times ) , to facilitate the migration of the DNA into the cells . At 72 h after the DNA injection , the mice were injected intraperitoneally with 150 mg/kg of luciferin in sterile water , and the luciferase activity was then assessed using an IVIS Lumina III imaging system ( Perkin Elmer , Rodgau , Germany ) . When possible , a single mouse received both plasmids at the same time , each on a different lesion . The UV irradiation continued until week 30 , according to the protocol [15] . The lesions were then closely monitored and the animals were sacrificed in accordance with an ethical protocol to avoid animal suffering . Skin lesions were collected for histological examination and detection for HPV38 E6/E7 RNA by in situ hybridization . Total RNA was isolated from dorsal skin of WT ( n = 4 ) or K14 HPV38 E6/E7 Tg animals ( n = 5 ) as well as histologically confirmed pre-malignant ( pre-m ) and SCC from three independent mice . cDNA was synthesized from 1 μg of total RNA using M-MLV reverse transcriptase ( Invitrogen , Darmstadt , Germany ) , and a mix of random hexamers were used as primers . Quantitative reverse transcription PCR ( RT-qPCR ) was performed in a 20 μl mixture containing 1 μl of 1:10 diluted cDNA and Mesa green quantitative PCR ( qPCR ) Master Mix ( Eurogentec , Angers , France ) with specific mouse ΔNp73α primers ( 5′-GCCAAAAGGGTCATCATC-3′ and 5′-TGCCAGTGAGCTTCCCGTTC-3′ ) or mouse GAPDH primers to amplify a housekeeping gene as internal control ( 5′-GTGACCCCATGAGACACCTC-3′ and 5′-GTATGTCCAGGTGGCCGAC–3′ ) , using an Applied Biosystems 7300 machine ( Applied Biosystems , Darmstadt , Germany ) . The fluorescence threshold value was calculated using the SDS analysis software from Applied Biosystems . Once the tumours reached the maximum ethically allowed size , the mice were killed and the lesions isolated . Half of the lesion was embedded in OCT medium and slowly cooled down to −80°C . Sense and antisense riboprobes were generated from linearized plasmid DNA containing full-length HPV38E6E7 cDNA using the Digoxigenin RNA labelling Mix from Roche . RNA-RNA in situ hybridization was performed as previously described[41] . In brief , serial 5 μm cryo-sections were mounted on Superfrost Plus slides ( Thermo Scientific ) , fixed in 4% paraformaldehyde in 2× SSPE , digested with proteinase K ( 0 . 5 μg/ml ) , and pre-hybridized at 42°C for 2–4 h . Hybridization was performed overnight at 42°C in 50% formamide , 2× SSPE , 10% dextran sulfate , 10 mM Tris-HCl pH 7 . 5 , 1× Denhardt’s solution , 500 μg/ml tRNA , 100 μg/ml herring sperm DNA , 0 . 1% SDS , and 10 μg/ml DIG-labelled riboprobe . After hybridization , slides were washed once in 50% formamide , 2× SSPE; 0 . 1% SDS for 30 min at 50°C , treated with RNaseA ( 50 μg/ml in 2× SSC , 0 . 1% SDS ) , and washed again in 50% formamide , 0 . 5× SSPE , 0 . 1% SDS for 30 min at 37°C . Hybridization signals were visualized using Biotin Tyramide ( TSA Biotin System , PerkinElmer ) according to the manufacturer’s protocol . Tumour growth values of lesions injected with the pS/MARt-Luc or pS/MARt-Luc-P2A-Cre vector were compared with the two-sample t-test . The statistical analysis was performed with GraphPad Prism ( version 6 , GraphPad Software Inc . , La Jolla , CA , USA ) . The quality of the raw reads was estimated with FastQC software ( version 0 . 11 . 5 , http://www . bioinformatics . babraham . ac . uk/projects/fastqc/ ) . Reads were mapped to the GRCm38 Mouse reference genome ( ftp://hgdownload . cse . ucsc . edu/goldenPath/mm10/ ) using Burrows-Wheeler Aligner ( BWA , http://bio-bwa . sourceforge . net/ ) version 0 . 7 . 15 and producing a BAM file . The following GATK Best Practice Recommendations were applied to the BAM files to improve variant detection quality . Picard ( version 2 . 4 . 1 , https://broadinstitute . github . io/picard/ ) SortSAM was used to sort and index BAM files , and the AddOrReplaceReadGroups tool was used to replace all read groups with a single new read group . The duplicate reads were marked with the MarkDuplicates tool from Picard , and the newly produced BAM file was indexed with the BuildBamIndex tool . GATK ( version 3 . 6 . 0 , https://software . broadinstitute . org/gatk/download/ ) RealignerTargetCreator was used to determine the position concerned by local realignment , and IndelRealigner was used to perform local realignment around these sites . The GATK BaseRecalibrator tool was used to detect systematic errors in base quality scores . Dbsnp and dbindel ( version 142 ) for the mm10 reference genome was downloaded from the Sanger website ( ftp://ftp-mouse . sanger . ac . uk/REL-1505-SNPs_Indels/ ) and considered as input . Lastly , the index of the output BAM file was created with Picard BuildBamIndex , and GATK PrintReads was used to write out sequence read data . The quality of the alignment was estimated with Qualimap ( version 2 . 0 . 2 , http://qualimap . bioinfo . cipf . es/ ) . Then the variant calling was done with Mutect ( version 1 . 1 . 7 , http://archive . broadinstitute . org/cancer/cga/mutect ) , by using a skin sample from a WT mouse not exposed to UV as the “normal sample” for paired analysis . Only somatic mutations passing Mutect internal filters were considered for the analysis . The VCF files are annotated with Annovar by using the MutSpec Annot Tool in Galaxy [42] . Variants were then filtered based on SegDup databases from UCSC ( version from 4 May 2014 , http://hgdownload . cse . ucsc . edu/goldenPath/mm10/database/genomicSuperDups . txt . gz ) , as well as Tandem Repeat and Repeat Masker ( version from 9 February 2012 , http://hgdownload . soe . ucsc . edu/goldenPath/mm10/bigZips/ ) . House-made scripts were then used to keep only SNPs that have a functional impact and fall in exonic or splicing regions . Non-negative matrix factorization mutational signatures were inferred with MutSpec-NMF tools , as previously reported . The pathway analysis was performed using the EnrichR web application ( http://amp . pharm . mssm . edu/Enrichr/; citations*2 ) . The input gene list was made by merging the mutations detected in the pre-malignant lesions ( n = 3 ) or cSCCs ( n = 3 ) of the K14 HPV38 E6/E7 Tg animals . The analysis included only genes harbouring mutations that are likely to alter the biological properties of the encoded products , i . e . , 3111 genes in the pre-malignant lesions and 6372 genes in the cSCCs . The gene lists were then loaded into the EnrichR software , and the result from the KEGG database ( version 2016 ) was considered . Only pathways with a significant adjusted p-value are shown in S1 Table . The list of pathways is ranked by combined score ( combined score is computed by taking the log of the p-value from the Fisher exact test and multiplying it by the z-score of the deviation from the expected rank ) . The list of epigenetic driver and modifier genes was constructed on the basis of genes reported in different publications [19–23] . The Cancer Gene Census list was downloaded from the COSMIC website ( 12 November 2016 , http://cancer . sanger . ac . uk/census ) and is based on a previous publication [24] . The comparison of the mouse data with the human data [25 , 26] was done with Bioconductor ( release 3 . 4 , https://www . bioconductor . org/ ) in R ( version 3 . 3 . 2 , “Sincere Pumpkin Patch” ) . The module BioMart[43 , 44] , version 2 . 3 enables the conversion of nearly 87 . 86% of human gene names from the Chitsazzadeh et al . publication [26] to their corresponding mouse gene names . | Many epidemiological and biological findings support the hypothesis that beta HPV types cooperate with UV radiation in the induction of NMSC , the most common form of human cancer . We have previously shown that K14 HPV38 E6/E7 Tg mice , when exposed to long-term UV radiation , developed NMSC , whereas WT animals subjected to identical treatments did not develop any type of skin lesions . Here , we show that the high skin cancer susceptibility of these Tg animals tightly correlates with their tendency to accumulate UV-induced mutations in genes that are frequently mutated in human NMSC . Importantly , deletion of the HPV38 E6 and E7 genes in existing skin lesions did not affect the further growth of the cancer cells . Together , these findings support the model that beta HPV infection is a co-factor in skin carcinogenesis , facilitating the accumulation of the UV-induced DNA mutations . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"keratinocytes",
"pathology",
"and",
"laboratory",
"medicine",
"ultraviolet",
"radiation",
"light",
"electromagnetic",
"radiation",
"epithelial",
"cells",
"radiation",
"animal",
"models",
"oncology",
"mutation",
"model",
"organisms... | 2018 | Beta HPV38 oncoproteins act with a hit-and-run mechanism in ultraviolet radiation-induced skin carcinogenesis in mice |
Adenoviruses infect epithelial cells lining mucous membranes to cause acute diseases in people . They are also utilized as vectors for vaccination and for gene and cancer therapy , as well as tools to discover mechanisms of cancer due to their tumorigenic potential in experimental animals . The adenovirus E4-ORF1 gene encodes an oncoprotein that promotes viral replication , cell survival , and transformation by activating phosphatidylinositol 3-kinase ( PI3K ) . While the mechanism of activation is not understood , this function depends on a complex formed between E4-ORF1 and the membrane-associated cellular PDZ protein Discs Large 1 ( Dlg1 ) , a common viral target having both tumor suppressor and oncogenic functions . Here , we report that in human epithelial cells , E4-ORF1 interacts with the regulatory and catalytic subunits of PI3K and elevates their levels . Like PI3K activation , PI3K protein elevation by E4-ORF1 requires Dlg1 . We further show that Dlg1 , E4-ORF1 , and PI3K form a ternary complex at the plasma membrane . At this site , Dlg1 also co-localizes with the activated PI3K effector protein Akt , indicating that the ternary complex mediates PI3K signaling . Signifying the functional importance of the ternary complex , the capacity of E4-ORF1 to induce soft agar growth and focus formation in cells is ablated either by a mutation that prevents E4-ORF1 binding to Dlg1 or by a PI3K inhibitor drug . These results demonstrate that E4-ORF1 interacts with Dlg1 and PI3K to assemble a ternary complex where E4-ORF1 hijacks the Dlg1 oncogenic function to relocate cytoplasmic PI3K to the membrane for constitutive activation . This novel mechanism of Dlg1 subversion by adenovirus to dysregulate PI3K could be used by other pathogenic viruses , such as human papillomavirus , human T-cell leukemia virus type 1 , and influenza A virus , which also target Dlg1 and activate PI3K in cells .
Human adenovirus type 9 ( Ad9 ) is a member of the subgroup D adenoviruses that cause eye infections in people [1] . In addition , infection of experimental animals with Ad9 generates estrogen-dependent mammary tumors , and the E4-ORF1 gene is the primary viral oncogenic determinant [2]–[4] . This viral gene likely evolved from a cellular dUTPase gene , which codes for an enzyme of nucleotide metabolism , and E4-ORF1 and dUTPase share a similar protein fold [5] , [6] . However , the E4-ORF1 protein lacks dUTPase catalytic activity , indicating functional divergence from dUTPase . Instead , E4-ORF1 functions to activate cellular class IA phosphatidylinositol 3-kinase ( PI3K ) at the plasma membrane of Ad9-infected human epithelial cells and Ad9-induced experimental tumor cells [7] . This function is conserved in other human adenovirus E4-ORF1 proteins and is essential for Ad9-induced oncogenesis [7] . E4-ORF1 activation of PI3K also enhances productive replication of human adenovirus type 5 ( Ad5 ) by overriding protein translation checkpoints [8] , [9] , prolongs survival of Ad5 vector-infected primary human endothelial cells [10] , and modulates lipid and glucose metabolism in human adenovirus type 36-infected cells [11] . Class IA PI3K is a lipid kinase that under normal physiological conditions functions as a key downstream effector of membrane receptors and ras [12] . PI3K exists as a heterodimer composed of p85 regulatory and p110 catalytic subunits . In the cytoplasm , the regulatory subunit stabilizes the catalytic subunit and inhibits its lipid kinase activity . Activated membrane receptors and ras can bind and recruit cytoplasmic PI3K to the plasma membrane , bringing it into contact with the lipid substrate phosphatidylinositol-4 , 5-bisphosphate ( PIP2 ) and also relieving enzymatic inhibition by the p85 regulatory subunit . PI3K converts PIP2 to the second messenger phosphatidylinositol 3 , 4 , 5-trisphosphate ( PIP3 ) , which in turn recruits PI3K effector proteins Akt and PDK1 to the plasma membrane . At this site , Akt is activated by phosphorylation on threonine 308 ( T308 ) by PDK1 and on serine 473 ( S473 ) by mTORC2 . Numerous downstream effectors of Akt act to regulate a broad range of cellular processes that include metabolism , protein synthesis , growth , survival , migration , and proliferation . Notably , the PI3K signaling pathway is one of the most frequently dysregulated pathways in human cancers [13] , and PI3K and its downstream effectors are subverted by many pathogenic human viruses to enhance virus-cell and virus-host interactions , such as viral entry , replication , reactivation from latency , and pathogenesis [14] . These observations underscore a prominent role for the PI3K pathway in human disease . Mutational analyses of the 125-residue Ad9 E4-ORF1 polypeptide identified multiple amino acid residues required for activation of PI3K , promotion of oncogenic transformation in cultured cells , and tumorigenesis in experimental animals [7] , [15]–[17] . Some of these crucial residues cluster at the E4-ORF1 carboxyl-terminus [18] , which defines an element that mediates direct binding to cellular PSD95/Dlg/ZO-1 ( PDZ ) domain proteins [19] , thereby revealing the first known virus-encoded PDZ domain-binding motif ( PBM ) . The identification of the E4-ORF1 PBM led to the discovery that the E6 and Tax proteins encoded by high-risk HPV or HTLV-1 , respectively , also possess a carboxyl-terminal PBM that is crucial for their oncogenic potential [19]–[24] . It is now recognized that many viruses code for PBM-containing proteins that target a wide variety of cellular PDZ proteins [23] . Dlg1 was the first cellular PDZ protein reported to interact with the PBM of Ad9 E4-ORF1 , as well as HPV E6 and HTLV-1 Tax [19] . Subsequent studies showed that the E4-ORF1 PBM also mediates binding to the cellular PDZ proteins MUPP1 , PATJ , MAGI-1 , and ZO-2 [25]–[28] . In polarized epithelial cells , Dlg1 localizes to the adherens junction ( AJ ) , whereas MUPP1 , PATJ , MAGI-1 , and ZO-2 localize to the tight junction ( TJ ) . These PDZ proteins normally function as scaffolds to assemble plasma membrane-associated protein complexes that control signal transduction , AJ and TJ formation , and polarity establishment [22] , [23] . In cells , E4-ORF1 exists as both a monomer and homo-trimer , each of which targets a different subset of PDZ proteins with distinct functional outcomes [5] . The E4-ORF1 monomer specifically binds and sequesters MUPP1 , PATJ , MAGI-1 , and ZO-2 within cytoplasmic punctae [25] , [26] , [28] and , as a result , disrupts the TJ and causes a loss of cell polarity [27] . Such defects are hallmarks of cancer cells and may directly contribute to carcinogenesis by deregulating normal cellular proliferation and differentiation programs [29]–[31] . More relevant to the current study , the E4-ORF1 trimer specifically binds to Dlg1 [5] , which consists of three PDZ domains along with Lin-2 and Lin-7 ( L27 ) , Src homology 3 ( SH3 ) , and guanylate kinase-homology ( GuK ) domains , as well as different insertion elements ( I1-I5 ) generated by alternative splicing [32] . While evidence indicates that Dlg1 has a tumor suppressor function [33] , Dlg1+/+ but not mutant Dlg1−/− mouse embryo fibroblasts are able to support oncogenic PI3K activation by Ad9 E4-ORF1 [34] . This finding not only revealed an absolute dependence on Dlg1 for this activity but also exposed a previously unrecognized Dlg1 oncogenic function , which may be widely important given that high-risk HPV E6 proteins require Dlg1 to promote invasive properties in cervical carcinoma cells [33] , [35] . The Dlg1 oncogenic function hijacked by E4-ORF1 derives from a specific Dlg1 splice isoform , Dlg1-I3 , which has an I3 insertion element that localizes Dlg1 to the plasma membrane by binding to membrane-associated protein 4 . 1 [36] , [37] . Moreover , Dlg1-I3 contributes to E4-ORF1-mediated PI3K activation , at least in part , by recruiting E4-ORF1 to the plasma membrane [34] , implying that the Dlg1:E4-ORF1 complex activates PI3K at this site . However , the molecular mechanism of activation has not been determined and remains an important gap in knowledge . The present study was undertaken to test the hypothesis that PI3K activation depends on an additional undetermined cellular factor recruited by E4-ORF1 into the Dlg1:E4-ORF1 complex . This idea prompted a search for new E4-ORF1-interacting proteins and led to identification of PI3K itself . Our findings revealed that E4-ORF1 binds directly to PI3K and recruits it into the Dlg1:E4-ORF1 complex , thereby forming a Dlg1:E4-ORF1:PI3K ternary complex that localizes to the plasma membrane and stimulates PI3K signaling . This novel mechanism of PI3K activation may serve as a paradigm to understand how other pathogenic human viruses dysregulate PI3K and how the common viral target Dlg1 , and possibly other PDZ proteins , contributes to viral infections and diseases .
To identify the postulated cellular factor that binds to E4-ORF1 in the Dlg1:E4-ORF1 complex , we subjected HeLa cell lysates to an in vitro pulldown assay with glutathione S-transferase ( GST ) fused to Ad9 E4-ORF1 , followed by a mass spectrometry analysis of associated proteins . Table S1 lists 10 selected identified proteins , eight of which were either identical ( Abl1 , ELAV1 , IGF2BP1 , IGF2BP3 ) or related to ( UPF2 , LARP1 , LARP2 , DDX17 ) Ad5 E4-ORF1-binding proteins reported in a recent proteomic study of DNA tumor virus oncoproteins [38] , suggesting that these eight proteins may be genuine E4-ORF1 cellular targets . More importantly , the other two proteins in the list were novel , and their identities as the PI3K regulatory subunits p85α and p85β prompted additional experiments to confirm and extend the finding . We first used immunoblots to examine proteins recovered from a similar pulldown conducted with lysates of human MCF10A cells , an immortalized but non-transformed human mammary epithelial cell line retaining properties of normal breast epithelial cells [39] . The results showed that GST-E4-ORF1 but not GST interacts with p85β and the PI3K catalytic subunit p110α and that neither GST protein associates with the cellular PDZ protein scribble ( Figure 1A ) . Moreover , GST-E4-ORF1 but not GST also interacted with the purified , catalytically-active recombinant human p85α:p110α heterodimer ( Figures 1B and 1C ) . The low binding efficiency of GST-E4-ORF1 seen in these assays is a general property attributed to aggregation of this bacterially expressed fusion protein . These collective data indicated specific and direct binding of Ad9 E4-ORF1 to functional class IA human PI3K . To investigate whether E4-ORF1 associates with endogenous PI3K in human epithelial cells , we generated and characterized MCF10A lines transduced either by the empty pBABE retroviral expression vector ( vector cells ) or by pBABE encoding the wild-type ( wt ) Ad9 E4-ORF1 protein ( wtORF1 cells ) , the Ad9 E4-ORF1 PBM mutant V125A protein unable to bind PDZ proteins ( V125A cells ) , or the mutant rasV12 protein that lacks a PBM but activates PI3K and the Raf/MAP kinase pathway ( rasV12 cells ) [7] , [19] . Consistent with previous results in rodent fibroblast lines [7] , immunoblots of extracts from the cell lines revealed higher levels of activated PI3K effector Akt dually phosphorylated at T308 and S473 ( P-Akt ) in wtORF1 and rasV12 cells than in vector and V125A cells ( Figure 2 , compare lanes 1–2 to lanes 3 and 5; Figures 3A and 3B , compare lane 1 and lane 3 ) , as well as higher levels of activated , phosphorylated MAP kinases ERK-1 and -2 ( P-ERK1/2 ) in rasV12 cells than in other cells ( Figure 2 , compare lane 5 to lanes 1–3 ) [34] . Compared to vector and V125A cells , wtORF1 and rasV12 cells also displayed higher levels of p85α , p85β , and p110α ( Figure 2 , compare lanes 1–2 to lanes 3 and 5 ) . Moreover , treatment of wtORF1 and rasV12 cells with the PI3K inhibitor drug LY294002 ( LY ) returned the high P-Akt levels to those of vector cells , yet had little or no effect on the high p85 and p110 levels ( Figure 2 , compare lane 1 to lanes 3–4 and 5–6 ) . In contrast , stable expression of a short hairpin RNA ( shRNA ) that depleted Dlg1 decreased the high levels of P-Akt , p85 , and p110 in wtORF1 cells but not in rasV12 cells ( compare Figure 3A , lanes 3–4 to Figure 3B , lanes 3–4 ) . From cumulative immunoblots , we quantified pertinent protein level differences for wtORF1 or rasV12 cells versus vector cells , as well as for wtORF1 cells transduced with the Dlg1 shRNA vector versus the scrambled shRNA vector ( Tables S2 , S3 , and S4 ) . Taken together , the data showed that while E4-ORF1 and rasV12 similarly elevate p85α , p85β , and p110α levels and activate Akt in a PI3K-dependent manner in human epithelial cells , E4-ORF1 differs from rasV12 in its dependence on both a PBM and Dlg1 for these activities and in not activating the Raf/MAP kinase pathway . The Dlg1-I2 isoform differs from the Dlg1-I3 isoform by having an I2 rather than I3 splice insertion element and by failing to support E4-ORF1-induced PI3K activation in rodent fibroblasts [34] . Given that Dlg1 depletion diminished both PI3K activation and PI3K protein elevation by E4-ORF1 in MCF10A cells ( Figure 3A ) , we sought to clarify roles for each Dlg1 isoform in these two E4-ORF1 activities . Our chosen approach was to determine whether reconstitution of Dlg1-depleted MCF10A cells with each Dlg1 isoform restores E4-ORF1-induced PI3K activation and/or PI3K protein elevation . We therefore transfected the Dlg1 shRNA-expressing MCF10A line with a wt E4-ORF1 expression plasmid alone or in combination with an expression plasmid encoding HA epitope-tagged ( HA- ) ΔNT-Dlg1-I3 or ΔNT-Dlg1-I2 ( Figure 4 , left panel ) , both of which have the amino-terminal ( NT ) region deleted . Deletion of the NT region , which contains the Dlg1 shRNA targeting sequence , renders HA-ΔNT-Dlg1 expression refractory to depletion by the Dlg1 shRNA , but does not affect the ability of Dlg1-I3 to support E4-ORF1-induced PI3K activation [34] . To detect the low levels of E4-ORF1 protein produced in these transient transfection assays , we immunoblotted for E4-ORF1 in the RIPA buffer-insoluble cell pellet fraction , which contains 91%±2 . 5% ( n = 3 independent experiments ) of E4-ORF1 protein expressed in MCF10A cells ( Fig . S1 ) . E4-ORF1 displays a similar cell fractionation profile in rodent fibroblasts [5] , [15] , [34] . A control immunoblot also verified Dlg1 depletion in the Dlg1 shRNA-expressing MCF10A line ( Figure 4 , right panel ) . The data showed that E4-ORF1-induced PI3K activation is increased by HA-ΔNT-Dlg1-I3 or decreased by HA-ΔNT-Dlg1-I2 and that E4-ORF1-induced p110α and p85α/β protein elevation is enhanced similarly by either HA-ΔNT-Dlg1-I3 or -I2 ( Figure 4 , left panel , compare lanes 2–4 ) . From cumulative independent experiments , we quantified these effects induced by Dlg1-I3 ( Table S5 ) or Dlg1-I2 ( Table S6 ) . Based on the findings , we concluded that , in Dlg1 shRNA-expressing MCF10A cells , Dlg1-I3 expression restores E4-ORF1-induced PI3K activation whereas either Dlg1-I3 or Dlg1-I2 expression restores PI3K protein elevation . The latter observation with Dlg1-I2 also demonstrated that E4-ORF1-induced PI3K activation and PI3K protein elevation are distinct Dlg1 activities , and revealed the first known function for the E4-ORF1:Dlg1-I2 complex . We previously reported that ras is required for E4-ORF1-induced PI3K activation in mouse fibroblasts [34] . To test whether this finding would extend to the current system , we transfected MCF10A cells with a wt E4-ORF1 expression plasmid alone or in combination with an expression plasmid encoding dominant-negative mutant rasN17 . The data showed that rasN17 over-expression blocked PI3K activation induced by mutant rasv12 ( Figure S2A , compare lanes 3–4 ) but not E4-ORF1 ( Figure S2B , compare lanes 3–4 ) . These findings suggested that ras does not mediate E4-ORF1-induced PI3K activation in MCF10A cells . We next immunoprecipitated ( IPed ) p110α from lysates of vector , wtORF1 , and V125A cells , as well as T123D cells expressing the Ad9 E4-ORF1 PBM mutant T123D protein unable to bind PDZ proteins [7] , [19] , and then tested for co-immunoprecipitation ( CoIP ) of E4-ORF1 , Dlg1 , p85α , and p85β in immunoblots ( Figure 5 ) . The data showed comparable coIP of wt and mutant E4-ORF1 proteins with p110α from each respective cell line ( lanes 6–8 ) , and also coIP of Dlg1 with p110α only from wtORF1 cells ( compare lane 6 to lanes 7–8 ) . IPed p85α or p85β yielded similar results ( data not shown ) . In a reciprocal experiment , we IPed Dlg1 from lysates of the same cell lines and tested for coIP of E4-ORF1 , p110α , p85α , and p85β ( Figure 5 ) . Here we observed coIP of the wt E4-ORF1 protein but not PBM mutant V125A or T123D protein with Dlg1 as expected , and also co-IP of p110α , p85α , and p85β with Dlg1 , but again only from wtORF1 cells ( compare lane 10 to lanes 11–12 ) . Significantly , the latter results with vector , wtORF1 , and V125A cells were mirrored when Dlg1 was IPed from lysates of mock-infected MCF10A cells or MCF10A cells infected with wt Ad9 virus or mutant Ad9 virus having the PBM mutation V125A introduced into the E4-ORF1 gene ( Ad9-V125A ) ( Figure 6A , compare lane 5 to lanes 4 and 6 ) . As controls , we verified that the infection with each virus was comparable by examining E4-ORF1 protein levels ( Figure 6A ) , viral major capsid protein hexon accumulation ( Figure 6B ) , and cytopathic effects ( Figure 6C ) . The comparable hexon protein accumulation was consistent with our failure to detect a replication defect for the Ad9-V125A mutant virus in MCF10A cells ( unpublished results ) . In addition , wt Ad9 virus-infected cells showed higher levels of activated P-Akt than either Ad9-V125A virus- or mock-infected cells ( Figure 6A; Tables S7 and S8 ) , similar to previous results examining P-Akt activation in human A549 cells mock-infected or infected with wt Ad9 or mutant Ad9-IIIA virus , which like the Ad9-V125A virus encodes an E4-ORF1 gene with a disrupted PBM [7] . We also observed decreased Dlg1 levels in cells infected with either wt Ad9 virus or Ad9-V125A virus compared to mock-infected cells ( Figure 6A , compare lane 1 and lanes 2–3 ) . A similar effect was detected in rasV12-expressing cells ( Figure 3B , compare lanes 1 and 3 ) , whereas the modest reduction in Dlg1 levels seen in wtORF1 cells compared to vector cells in Figure 3A was not reproducible . We do not yet understand how Dlg1 is downregulated in adenovirus-infected cells and RasV12-expressing cells . The biochemical data presented above importantly showed that in human epithelial cells , either stably expressing wt E4-ORF1 protein or lytically infected with wt Ad9 virus , the E4-ORF1 protein binds the PI3K p85:p110 heterodimer in a PBM-independent manner and , in conjunction with separate PBM-dependent binding to Dlg1 , tethers these two cellular factors together to form a Dlg1:E4-ORF1:PI3K ternary complex . Moreover , formation of this ternary complex strictly correlated with E4-ORF1-mediated PI3K/Akt activation as we observed the complex in wtORF1 cells and wt Ad9 virus-infected cells but not in V125A cells , T123D cells , or mutant Ad9-V125A virus-infected cells . Nonetheless , immunoblots of lysates from wt Ad9 virus- and mutant Ad9-V125A virus-infected cells revealed some differences from corresponding wtORF1 and V125A cells . Whereas the levels of p85α , p85β , and p110α were higher in wtORF1 cells than in V125A and vector cells ( Figure 2 ) , these protein levels were higher in both wt Ad9 virus- and Ad9-V125 virus-infected cells than in mock-infected cells , with comparable p85α and p110α elevation in the virus-infected cells and somewhat higher p85β elevation in wt Ad9 virus-infected cells than in Ad9-V125 virus-infected cells ( Figure 6A; Tables S7 and S8 ) . These results suggested that , in adenovirus-infected cells , p85 and p110 are upregulated by two distinct viral functions , E4-ORF1 and an undetermined viral factor , which differs from E4-ORF1 by inducing weaker elevation of p85β protein levels and not activating PI3K . To identify residues of E4-ORF1 required for binding to PI3K , we generated MCF10A lines expressing 16 different E4-ORF1 mutants . We then IPed p110α from lysates of these cell lines , and tested for co-IP of E4-ORF1 ( data not shown ) . One E4-ORF1 mutant , mutant KI , was defective for binding to PI3K compared to the wt E4-ORF1 protein ( Figure 7A , compare lanes 6 and 7 ) . Mutant KI has two point mutations that change the adjacent carboxyl-terminal residues K120 and I121 to alanine residues ( Figure 7B ) [5] . We also IPed Dlg1 from the lysates and tested for co-IP of E4-ORF1 . The data showed that mutant KI is similarly defective for binding to Dlg1 ( Figure 7A , compare lanes 9 and 10 ) , consistent with a previous report [5] . These binding defects of mutant KI are specific and do not reflect general protein misfolding or complete loss of function because it retains the capacity to form homo-trimers and to bind the cellular PDZ proteins MUPP1 , MAGI-1 , and ZO-2 [5] . Thus , we identified two E4-ORF1 residues that specifically determine both interactions required to form the ternary complex . Interestingly , the KI residues are located at the extreme carboxyl-terminus near the PBM required for binding to Dlg1 , as well as to other PDZ proteins , and additionally overlap the TRI element required for E4-ORF1 homo-trimerization ( Figure 7B ) . The requirement for Dlg1 to support E4-ORF1-induced PI3K activation correlates with its ability to recruit E4-ORF1 protein to the plasma membrane [34] . This observation , together with discovery of the Dlg1:E4-ORF1:PI3K ternary complex , led us to hypothesize that Dlg1 recruits not only E4-ORF1 but also PI3K to the plasma membrane . This hypothesis predicts that E4-ORF1 and PI3K would co-localize with Dlg1 at the plasma membrane in wtORF1 cells , but not in V125A cells where the PBM mutant V125A protein can neither bind Dlg1 nor activate PI3K , yet can bind PI3K ( Figure 5 ) . We investigated this idea in indirect immunofluorescence ( IF ) assays where vector , wtORF1 , and V125A cells were double labeled with antibodies to Dlg1 and either E4-ORF1 or p85 and then analyzed by confocal microscopy . We first examined each cell line for co-localization of E4-ORF1 and Dlg1 ( Figure 8 ) . In wtORF1 cells , E4-ORF1 protein localized in the cytoplasm , exhibiting both diffuse and punctate distributions , as well as at the plasma membrane . Previous findings indicated that the cytoplasmic punctae reflect a combination of E4-ORF1 monomer sequestration of TJ-associated PDZ proteins and E4-ORF1 trimer association with membrane vesicles whereas the plasma membrane fraction represents E4-ORF1 trimer binding to Dlg1 [5] , [34] . Consistent with these findings , the PBM mutant V125A protein failed to localize at the plasma membrane , but retained the cytoplasmic punctate distribution , likely through the PBM-independent association of E4-ORF1 with membrane vesicles [15] . Unlike the wt E4-ORF1 protein , the mutant V125A protein also accumulated in the nucleus , a defect of PBM mutants attributed to passive nuclear diffusion resulting from loss of PBM-dependent anchoring to PDZ proteins at extranuclear sites [15] . We also note that in IF assays described here and below , V125A and T123D cells yielded identical results ( data not shown ) . More importantly , while we found that Dlg1 similarly localized at cell-cell contact regions of the plasma membrane in vector , wtORF1 , and V125A cells , only the plasma membrane-associated E4-ORF1 protein fraction in wtORF1 cells co-localized with Dlg1 . Quantified IF data for this experiment , and other IF experiments detailed below , are presented in Table S9 . We next tested the cell lines for co-localization of p85 and Dlg1 ( Figure 9 ) . In vector and V125A cells , p85 exhibited a cytoplasmic distribution but , due to an absence at the plasma membrane , it failed to co-localize with Dlg1 . In wtORF1 cells , p85 was similarly distributed in the cytoplasm and , more importantly , was additionally detected at the plasma membrane where it co-localized with Dlg1 . Furthermore , the latter effect was decreased by shRNA-mediated Dlg1 depletion in wtORF1 cells ( Figure 10 ) . These data , together with those presented in Figure 8 , indicated that PBM-mediated binding of the wt E4-ORF1 protein to Dlg1 functions not only to assemble the Dlg1:E4-ORF1:PI3K ternary complex but also to localize PI3K to the plasma membrane . To obtain evidence that the membrane-associated Dlg1:E4-ORF1:PI3K ternary complex mediates PI3K signaling , we investigated whether P-Akt , which is recruited to membrane sites of PI3K activation by binding to PI3K product PIP3 , also co-localizes with Dlg1 at the plasma membrane ( Figure 11 ) . Consistent with results of immunoblot assays shown in Figures 2 and 5 , P-Akt staining was detected in wtORF1 cells but not in vector or V125A cells . In wtORF1 cells , P-Akt distributed in a speckled pattern in the cytoplasm , and also accumulated at the plasma membrane where it co-localized with Dlg1 . We observed a similar accumulation of total Akt protein at the plasma membrane of wtORF1 cells but not vector or V125A cells ( Figure S3 ) . In addition , this accumulation of P-Akt at the plasma membrane was diminished by shRNA-mediated Dlg1 depletion in wtORF1 cells ( Figure 12 ) . These data supported the conclusion that the ternary complex mediates PI3K signaling at the plasma membrane . Results presented in Figures 8 and 9 demonstrated that both E4-ORF1 and PI3K are recruited to Dlg1 located at the plasma membrane of MCF10A cells . We previously reported that E4-ORF1 also induces cytoplasmic Dlg1 to translocate to the plasma membrane [34] . This finding led to the expectation that wtORF1 cells would show higher Dlg1 membrane staining and lower Dlg1 cytoplasmic staining than vector or V125A cells . While this effect was evident in Figures 10 and 12 , it was either weak or absent in Figures 8 , 9 and 11 . These variable results can be explained by the fact that prior to formation of adherens junctions , Dlg1 localizes primarily in the cytoplasm whereas during formation of adherens junctions , Dlg1 increasingly concentrates at this site of the plasma membrane , even in the absence of E4-ORF1 . In experiments where the E4-ORF1 effect was weak or absent , the cells were post-confluent and had concentrated Dlg1 at mature adherens junctions , thereby partially or completely masking E4-ORF1-induced Dlg1 membrane recruitment . Thus , to optimize detection of this E4-ORF1 activity , we compared the localization of Dlg1 in confluent vector , wtORF1 , and V125A cells prior to formation of adherens junctions ( Figure 13 ) . Under these conditions , wtORF1 cells displayed higher Dlg1 membrane staining and lower cytoplasmic staining than vector or V125A cells . This finding demonstrated that E4-ORF1 also promotes cytoplasmic Dlg1 to translocate to the plasma membrane of MCF10A cells . We next investigated the functional importance of the ternary complex in cells . The capacity of cells to form colonies when suspended in soft agar measures anchorage-independent growth , a property that correlates best with tumorigenic potential [40] . Previous studies showed that Ad9 E4-ORF1-expressing rodent fibroblasts and Ad9-induced mammary tumor cells form colonies in soft agar and that this oncogenic property depends on E4-ORF1-induced PI3K activation [7] , [34] . In soft agar assays conducted with the MCF10A lines , we found that wtORF1 cells formed colonies with a high cloning efficiency ( 95%±1 . 5% SD; n = 3 independent experiments ) , whereas V125A and T123D cells behaved like vector cells by not forming colonies ( Figure 14A ) . Moreover , the PI3K inhibitor LY abolished colony formation by wtORF1 cells ( Figure 14B ) . Transformation assays measuring focus formation by these MCF10A lines yielded similar results ( Figures 14C and 14D ) . To implicate Dlg1 directly in E4-ORF1-induced cellular transformation , we compared colony formation by wtORF1 cells expressing the Dlg1 shRNA and control wtORF1 cells expressing the scrambled Dlg1 shRNA ( see Figure 3A , lanes 3 and 4 ) . Initial experiments , however , failed to detect lower colony formation by the Dlg1 shRNA-expressing wtORF1 cells . Given that , on the average , P-Akt levels in wtORF1 cells are elevated 39- to 63-fold ( Table S2 ) and the Dlg1 shRNA reduces these levels by only 3 . 2- to 3 . 6-fold ( Table S4 ) , we considered the possibility that the Dlg1 shRNA-mediated reduction in P-Akt levels may be insufficient to decrease colony formation . To test this idea , we performed the soft agar assays in the presence of the PI3K inhibitor drug LY at the sub-inhibitory dose of 25 µM , which is one-fourth the concentration required to abolish PI3K activation , soft agar growth , and focus formation in MCF10A cells ( Figures 2 , 14B , and 14D ) . Under these conditions , colony formation by wtORF1 cells expressing the Dlg1 shRNA was substantially reduced compared to control cells ( Figure 14E ) , consistent with previous findings [34] . Immunoblot assays confirmed that , when cultured with 25 µM LY , wtORF1 cells expressing the Dlg1 shRNA have lower levels of Dlg1 and P-Akt and but not E4-ORF1 than the control cells ( Figure 14F ) . Collectively , our findings supported the conclusion that both the Dlg1:E4-ORF1:PI3K ternary complex and its stimulation of PI3K signaling are essential for the function of E4-ORF1 in cells . The results additionally showed that the wt E4-ORF1 protein is capable of promoting transformation of a non-transformed human epithelial cell line .
The human adenovirus E4-ORF1 protein mediates constitutive activation of cellular PI3K , a function shown to promote tumorigenesis in experimental animals and to augment viral replication , prolong survival , and modulate lipid and glucose metabolism in cells [7]–[11] . While it is known that PI3K activation by E4-ORF1 depends on its interaction with the cellular PDZ protein Dlg1 and on localization of the resulting Dlg1:E4-ORF1 complex to the plasma membrane , the underlying molecular mechanism for activation had not been previously determined . We report here the first mechanistic insight into this activity by identifying the PI3K p85:p110 heterodimer as a new cellular target of the E4-ORF1 protein . Based on results reported here along with previously published findings , the mechanism for E4-ORF1-mediated PI3K activation shown in Figure 15 has emerged . We propose that in step 1 , E4-ORF1 via the KI residues binds directly to PI3K to form a cytoplasmic E4-ORF1:PI3K heterocomplex ( Figures 1 , 5 , and 7 ) . In step 2 , E4-ORF1 within this heterocomplex also interacts with Dlg1-I3 ( Figure 4 ) [34] , via cooperative binding between two PBM+KI elements in the E4-ORF1 trimer and two PDZ domains in Dlg1 ( Figure 7 ) [5] , [34] , thereby assembling a Dlg1:E4-ORF1:PI3K ternary complex ( Figures 5 and 6A ) . Our previous finding that only homo-trimeric E4-ORF1 can interact with Dlg1-I3 [5] , which is crucial for E4-ORF1-induced PI3K activation ( Figures 3A and 4 ) [34] , implies that PI3K likewise binds to the E4-ORF1 trimer in both the ternary complex and the E4-ORF1:PI3K heterocomplex , though it remains possible that the E4-ORF1 monomer also binds PI3K . More importantly , a key consequence of assembling the ternary complex is recruitment of the cytoplasmic E4-ORF1:PI3K heterocomplex to the plasma membrane in a Dlg1-dependent fashion ( Figures 8–10 and 13 ) [34] . Similar to growth factor receptor- and ras-mediated PI3K activation [12] , the ternary complex additionally stimulates PI3K catalytic activity ( Figure 4 ) [7] . An interesting possibility is that direct binding of PI3K to both E4-ORF1 and Dlg1 in the ternary complex contributes to the latter effect , as well as to PI3K protein elevation discussed later . In step 3 , the increased catalytic activity of PI3K in the ternary complex together with its location at the plasma membrane brings the activated PI3K enzyme into contact with lipid substrate PIP2 to produce PIP3 , which in turn recruits Akt to the plasma membrane ( Figure S3 ) , resulting in its activation by phosphorylation ( Figures 2 , 3A , and 11–12 ) . The ensuing dysregulated activation of Akt and its downstream effectors enhance viral replication and cellular metabolism and survival , and can also promote cellular transformation ( Figure 14 ) [7]–[9] , [11] , [34] . Also worth mention is that unpublished results suggest that the proposed mechanism of PI3K activation by Ad9 E4-ORF1 protein is shared by E4-ORF1 proteins encoded by other adenovirus serotypes and subgroups ( MK and KK , manuscript in preparation ) . Numerous viruses encode proteins that dysregulate PI3K in cells [14] , [41]–[51] , though molecular mechanisms are lacking in many cases . When mechanistic details are available , they often involve direct or indirect viral protein:PI3K interactions mediated by phosphorylated tyrosine residues on the viral protein itself or on an activated cellular receptor bound to the viral protein . Examples include mouse polyomavirus middle T antigen and EBV LMP1 and LMP2A that directly bind PI3K by mimicking tyrosine phosphorylated , activated membrane receptors; HBV X , HIV-1 Nef , and HCV NS5 that directly couple PI3K to non-receptor tyrosine kinases; and HPV E5 that indirectly binds to PI3K through an interaction with activated , cellular receptor protein tyrosine kinases [14] . These examples underscore novel differences in PI3K activation by Ad9 E4-ORF1 , including the requirement for Dlg1 ( Figures 3 and 4 ) and the lack of association of PI3K with tyrosine-phosphorylated proteins [34] . Our data indicated that the ternary complex not only activates PI3K but also elevates p85/p110 protein levels ( Figures 2 , 3A , 4 , and 7 ) . The latter effect may be important as PI3K over-expression is reported to dysregulate PI3K signaling . For example , an amplified and over-expressed PIK3CA gene coding for the PI3K catalytic subunit hyper-activates PI3K signaling in several different types of human cancers [52] , and an over-expressed wt PIK3CA cDNA in MCF10A cells increases PI3K signaling and cellular proliferation , and provokes cellular transformation [53] . Because treatment of E4-ORF1-expressing cells with a PI3K inhibitor for 30 min , or 2 h ( data not shown ) , failed to diminish p85/p110 levels yet lowered Akt and ERK levels ( Figure 2 ) , it seems unlikely that p85/p110 elevation results from the capacity of the downstream PI3K effector complex mTORC1 to increase protein synthesis [52] . We instead postulate that the ternary complex directly mediates Dlg1-dependent p85/p110 protein stability . Results showed that both p85/p110 protein elevation and PI3K activation in wtORF1 cells are diminished by shRNA-mediated Dlg1 depletion or are either severely impaired or absent in V125A and T123D cells ( Figures 2 and 3A ) , revealing a common dependence on Dlg1 . Other data showed that the Dlg1-I2 isoform does not support E4-ORF1-induced PI3K activation but does support E4-ORF1-induced PI3K protein elevation ( Figure 4 ) , indicating that these two Dlg1-mediated activities are separable . This observation hints to the possible existence of separate pools of ternary complexes containing either Dlg1-I3 or Dlg1-I2 . In this scenario , the Dlg1-I2 ternary complex may localize in the cytoplasm and function solely to elevate PI3K protein levels , whereas the Dlg1-I3 ternary complex localizes both in the cytoplasm and at the plasma membrane , with the cytoplasmic complex also functioning to elevate PI3K protein levels and the membrane complex functioning specifically to activate PI3K . Future studies will test these ideas as well as determine whether E4-ORF1-mediated p85/p110 elevation results from increased mRNA levels , protein synthesis , or protein half-life . Results of soft agar and focus formation assays implicated the ternary complex and its activation of PI3K in E4-ORF1-induced transformation of a human epithelial line that retains features of normal epithelial cells ( Figure 14 ) . This finding reinforces concerns that the use of adenovirus vectors retaining the E4-ORF1 gene for vaccination or various therapies , as well as the proposed use of the Ad36 E4-ORF1 gene to treat fatty liver disease and liver dysfunction or to improve glycemic control [11] , may increase patient risk for developing neoplasms . Evidence also indicates that Dlg1 functions to regulate normal PI3K signaling in cells . For example , Laprise et al . showed that Dlg1 is phosphorylated on tyrosine residues at the AJ of human intestinal epithelial cells and that these residues mediate direct binding to and activation of PI3K to promote cellular differentiation [54] . Furthermore , the lipid phosphatase and important human tumor suppressor protein PTEN [13] , which antagonizes PI3K signaling by dephosphorylating PIP3 , is known to bind Dlg1 . This interaction , mediated by the carboxyl-terminal PBM of PTEN and PDZ domain 2 of Dlg1 [55] , [56] , enhances PTEN activity to block proliferation and viability of MCF-7 breast carcinoma cells and to suppress Schwann cell myelination of peripheral nerves [57] , [58] . In addition to human adenovirus , several other pathogenic human viruses code for PBM-containing proteins that target Dlg1 and also bind to and/or activate PI3K , including the E6 oncoprotein of high-risk HPVs [42] , [59] , [60] , the Tax oncoprotein of HTLV-1 [61]–[65] , and the NS1 protein of influenza A viruses [66] . These observations suggest that , under normal physiological conditions , Dlg1 functions as a key regulator of PI3K signaling and that pathogenic human viruses commonly hijack this cellular PDZ protein , at least in part , to dysregulate the PI3K pathway and , in doing so , enhance viral infections associated with acute and chronic human diseases and cancer . Hence , studies of human adenovirus and its subversion of cellular PI3K , Dlg1 , and other PDZ proteins may yield mechanistic insights that aid development of new therapeutic strategies for treating viral diseases in people .
Plasmid pBABE-puro or -blasti containing a wt or mutant E4-ORF1 or rasV12 cDNA , plasmid pGEX-2TK containing a wt E4-ORF1 cDNA , and plasmid GW1 containing wt E4-ORF1 , rasN17 , HA-ΔNT-Dlg1-I3 , or HA-ΔNT-Dlg1-I2 cDNA were described [7] , [18] , [19] , [34] . Oligonucleotides encoding Dlg1 shRNA 5′GCAAGATACCCAGAGAGCA3′ or matched scrambled shRNA 5′GGACCACAACGACTAGAGA3′ were cloned into plasmid pSUPER-retro ( Oligoengine , Seattle ) . Human MCF10A mammary epithelial cells ( American Type Culture Collection ) were maintained , as described [39] , in complete medium consisting of DMEM/F-12 supplemented with 5% horse serum ( Invitrogen , Carlsbad , CA ) , 20 ng/ml epidermal growth factor ( EGF ) ( Peprotech , Rocky Hill , NJ ) , 100 µg/ml hydrocortisone , 10 µg/ml insulin , 1 ng/ml cholera toxin , and 20 µg/ml gentamicin ( Sigma-Aldrich , St . Louis , MO ) . MCF10A lines were generated by transduction with retroviral vector pBABE and/or pSUPER-retro followed by selection in complete medium containing 2 µg/ml puromycin and/or 25 µg/ml blasticidin . Experiments utilized pools of selected cells passaged five times or less , except for the experiment presented in Figure 12B , which used cells at passage number 8 . The same numbers of cells were plated for all experiments comparing different cell lines or treatments . For some experiments , cells were passaged into complete medium containing a lower concentration of EGF ( 5 ng/ml ) [39] . PI3K inhibitor LY294002 was purchased ( Cell Signaling Technology , Inc . , Beverly , Massachusetts ) . Transfections were performed with TransIT-LT1 Transfection Reagent ( Mirus Bio , Madison , WI ) . Wt and mutant Ad9 viruses and their propagation in human A549 cells were described [7] , [16] . Extracts were prepared , as described [19] , by lysis of cells in ice-cold RIPA buffer ( 150 mM NaCl , 50 mM Tris-HCl pH 8 . 0 , 1% Nonidet P-40 , 0 . 5% deoxycholate , 0 . 1% SDS ) containing protease inhibitors ( 2 mM PMSF , 20 µg/ml each of leupeptin and aprotinin ) and phosphatase inhibitors ( 50 mM NaF , 10 mM sodium pyrophosphate , 1 mM sodium orthovanadate ) . Protein concentrations were determined by the Bradford method . For cell fractionation assays [15] , cells were lysed in RIPA buffer and then centrifuged ( 10 , 000× g for 15 min at 4°C ) to separate the detergent-soluble supernatant and detergent-insoluble pellet fractions . The pellet fraction was subsequently solubilized in a volume of 2× sample buffer ( 125 mM Tris-HCl , pH 6 . 8 , 4% SDS , 0 . 2 M DTT , 20% glycerol , 0 . 001% bromophenol blue ) equal to that of the detergent-soluble supernatant fraction . Experiments compared equal volumes of the detergent-soluble supernatant and detergent-insoluble pellet fractions . Ad9 E4-ORF1 antiserum was described [3] . Antibodies to p110α , Akt , phospho-Akt ( Ser473 ) , phospho-Akt ( Thr308 ) , p42/44 MAPK and phospho-p42/44 MAPK ( Thr202/Tyr204 ) ( Cell Signaling Technologies ) , or p85β , SAP97 ( Dlg1 ) , and scribble ( Santa Cruz Biotechnologies ) , or p85α , p85α/β , and actin ( Millipore ) , or ras ( BD Biosciences ) , or HA ( Sigma-Aldrich ) were purchased . Greater than 90% pure , catalytically active recombinant PI3K generated by co-expression of histidine-tagged full-length human p85α and p110α in Sf9 cells was purchased ( SignalChem ) . GST pulldowns and immunoprecipitations with glutathione-sepharose beads or protein G-sepharose beads ( GE Healthcare Life Sciences ) , respectively , were carried out as described [4] , [18] . Recovered proteins and cell extract ( 30 µg of protein ) were resolved by SDS-PAGE , transferred to a PVDF membrane , and immunoblotted as described [4] , [18] . Immunoblotted membranes were imaged with a UVP Biospectrum 810 Imaging System ( Upland , CA ) and analyzed with VisionWorksLS software . Differences in the levels of specified proteins between two cell samples were quantified by comparing protein band intensities normalized to actin , and the Student's t-test was performed to determine statistical significance . E4-ORF1-binding proteins were identified by conducting a pulldown assay with extracts of suspension-cultured HeLa cells , resolving recovered proteins by SDS-PAGE , digesting separate gel sections with trypsin , and subjecting the released peptides to MALDI-TOF mass spectrometry . Glass slides ( Millicell EZ SLIDE , Millipore ) were coated with poly-L-lysine ( Sigma-Aldrich ) . Cells plated on the slides were fixed in 2% formaldehyde ( Polysciences , Inc . ) , permeabilized with 0 . 5% Triton X-100 , quenched with 100 mM glycine , blocked in 10% goat serum , and incubated with primary antibody and then with Alexa Fluor 488-conjugated goat anti-rabbit IgG and/or Alexa Fluor 594-conjugated goat anti-mouse IgG secondary antibodies ( Life Technologies Corp . ) . Prior or following incubation with primary antibody , cells were washed between each step with either phosphate-buffered saline ( PBS ) or immunofluorescence buffer ( IFB ) [7 . 7 mM sodium azide , 0 . 1% ( w/v ) BSA , 0 . 2% ( v/v ) Triton X-100 , 0 . 05% ( v/v ) Tween-20 in PBS] , respectively . Coverslips were mounted on slides with SlowFade Gold antifade reagent ( Life Technologies Corp . ) . Cells were analyzed by confocal microscopy with a Nikon A1-Rs inverted laser-scanning microscope and NIS Elements software . The percentage of cells in which specified proteins localized at the plasma membrane was quantified using Image J software ( NIH ) . Soft agar assays were carried out as described [3] . Briefly , in complete medium , 3×105 cells were suspended in 1 ml of 0 . 4% noble agar ( Affymetrix ) and placed atop a 2 ml 0 . 8% noble agar underlay in a 6-well plate . Cells were fed complete medium every other day . Colonies were documented with a Nikon D70S camera mounted on a Nikon TMS inverted microscope . ImageJ software ( NIH ) was used to score the numbers of cells that ( a ) did and ( b ) did not form a colony , and cloning efficiency ( a/a+b ) was calculated from >300 scored cells per experiment . For focus formation assays , 300 wtORF1 , V125A , or vector cells mixed with 3×105 vector cells in complete medium were seeded into a 6-well plate . Cells were fed every 3 days and , when visible , foci were stained with crystal violet ( 5 mg/ml in 25% methanol ) and photographed with a Canon PowerShot A1200 digital camera . ANOVA with Tukey post-hoc analysis was performed to determine statistical significance . Microsoft Excel was used to calculate the mean , standard deviation ( SD ) , and standard error of the mean ( SEM ) for data . R statistical software was used to determine statistical significance . Standard denotation of asterisks for p values was used ( * , p<0 . 05; ** , p<0 . 01; *** , p<0 . 001 ) . UniProtKB/Swiss-Prot accession numbers ( parentheses ) are indicated for proteins mentioned in text: Abl1 ( P00519 ) , actin ( P68032 ) , Ad9 E4-ORF1 ( P89079 ) , Ad9 hexon ( Q9QPU1 ) , Akt ( P31749 ) , DDX17 ( Q92841 ) , Dlg1 ( Q12959 ) , dUTPase ( P33316 ) , EBV LMP1 ( P03230 ) , EBV LMP2A ( P13285 ) , ELAV1 ( Q15717 ) , ERK1 ( P27361 ) , ERK2 ( P28482 ) , Ras ( P01112 ) , HBV X ( Q69027 ) , HCV NS5 ( C1IEN6 ) , HIV-1 Nef ( P04601 ) , HPV-16 E5 ( P06927 ) , HPV-16 E6 ( P03126 ) , HTLV-1 Tax ( P03409 ) , IGF2BP1 ( Q9NZI8 ) , IGF2BP3 ( O00425 ) Influenza virus A NS1 ( P03496 ) , LARP1 ( Q6PKG0 ) , LARP2 ( Q659C4 ) , MAGI-1 ( Q96QZ7 ) , MUPP1 ( O75970 ) , p110α ( P42336 ) , p85α ( P27986 ) , p85β ( O00459 ) , PATJ ( Q8NI35 ) , PDK1 ( Q15118 ) , PTEN ( P60484 ) , scribble ( Q14160 ) , UPF2 ( Q9HAU5 ) , ZO-2 ( Q9UDY2 ) | Adenoviruses cause acute illnesses in people , and are additionally utilized both as vehicles to cure genetic diseases , fight cancer , and deliver vaccines , and as tools to discover how cancers develop due to a capacity to generate tumors in experimental animals . The adenovirus E4-ORF1 protein reprograms cell metabolism to enhance virus production in infected cells and promotes cell survival and tumors by activating the important cellular protein phosphatidylinositol 3-kinase ( PI3K ) . How E4-ORF1 activates PI3K is not known , though this function depends on E4-ORF1 binding to the membrane-associated cellular protein Discs Large 1 ( Dlg1 ) , which many different viruses evolved to target . In this study , we identify PI3K as a new direct target of E4-ORF1 . Results further show that E4-ORF1 binds to PI3K in the cytoplasm and delivers it to Dlg1 at the membrane where the three proteins form a complex that activates PI3K and induces oncogenic growth in cells . This novel molecular mechanism in which adenovirus subverts Dlg1 to dysregulate PI3K may serve as a paradigm to understand PI3K activation mediated by other important pathogenic viruses , such as human papillomavirus , human T-cell leukemia virus type 1 , and influenza A virus , which also target Dlg1 in infected cells . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"protein",
"interactions",
"enzymes",
"regulatory",
"proteins",
"enzymology",
"microbiology",
"basic",
"cancer",
"research",
"oncology",
"model",
"organisms",
"population",
"modeling",
"enzyme",
"chemistry",
"research",
"and",
"a... | 2014 | The Human Adenovirus E4-ORF1 Protein Subverts Discs Large 1 to Mediate Membrane Recruitment and Dysregulation of Phosphatidylinositol 3-Kinase |
Trichomonas vaginalis is a causative agent of Trichomoniasis , a leading non-viral sexually transmitted disease worldwide . In the current study , we show Heat shock protein 90 is essential for its growth . Upon genomic analysis of the parasite , it was found to possess seven ORFs which could potentially encode Hsp90 isoforms . We identified a cytosolic Hsp90 homolog , four homologs which can align to truncated cytosolic Hsp90 gene products along with two Grp94 homologs ( ER isoform of Hsp90 ) . However , both Grp94 orthologs lacked an ER retention motif . In cancer cells , it is very well established that Hsp90 is secreted and regulates key clients involved in metastases , migration , and invasion . Since Trichomonas Grp94 lacks ER retention motif , we examined the possibility of its secretion . By using cell biology and biochemical approaches we show that the Grp94 isoform of Hsp90 is secreted by the parasite by the classical ER-Golgi pathway . This is the first report of a genome encoded secreted Hsp90 in a clinically important parasitic protozoan .
Trichomonas vaginalis is the causative agent of Trichomoniasis , a leading non-viral sexually transmitted disease worldwide [1] . The pathogen induces local inflammation of the lower genitourinary tract , can be involved in premature labor and low birth weight . Trichomoniasis affects both males and females with higher prevalence in females in both developed and developing countries [1] . Epidemiological studies show a two to three-fold increase in the risk of HIV infection in people with Trichomoniasis due to HIV target cells , CD4+ T-lymphocytes , recruitment at the site of infection [2 , 3] . This parasite infects more than 270 million people worldwide annually , despite which there is no clear consensus on its pathogenicity mechanisms . Due to the synergy with HIV infection and increase in reports of drug resistance , it is pivotal to understand the basic survival principles of the parasite [2 , 4] . This extracellular parasite survives in a dynamic and physiologically diverse niche of urogenital tract and therefore , is likely to possess a robust stress-response machinery to persist and establish infection . Our studies on related parasitic protozoa show the important role of stress-regulated heat shock protein 90 in their life cycle and virulence [5–14] . When we analyzed Hsp90 sequences in Trichomonas genome , we found two Hsp90 isoforms ( TVAG_030810 and TVAG_378910 , henceforth referred to as TV810 and TV910 respectively ) , which lacked classical motif of either cytosolic Hsp90 or Grp94 [15] . We show that these two isoforms are Grp94 homologs , however , both lacked an ER retention signal . This observation was very peculiar since most of the studied Grp94s have the canonical ER retention signal . Therefore , we were intrigued about what could be the possible fate of Trichomonas Grp94 homologs . In absence of ER retention signal , one possibility is that these two Hsp90 orthologs are secreted . In cancer cells , it is very well established that Hsp90 is secreted and regulates key clients like matrix metalloproteinases ( MMPs ) which are involved in migration and invasion [16] . It is known to play an important role in epithelial to mesenchymal transition , thereby regulating tumor metastasis [16–20] . Majority of these studies have focused on studying the role of extracellular Hsp90 in context of cancer and there are not many well-documented studies on extracellular Hsp90 in other organisms including parasitic protozoa . In the current study , we show that Hsp90 inhibition is lethal to Trichomonas growth and hence Hsp90 function is critical for the survival of the parasite . Using cell biology , biochemical and proteomics approaches , we investigated the localization of Grp94 ortholog , TV910 , and found it to be secreted by Trichomonas in the culture medium by the classical ER-Golgi secretory pathway . This is the first report of a secreted Hsp90 in a clinically important parasitic protozoan .
A polyclonal antibody was raised in New Zealand White strain rabbit against 6x-His-tagged full-length TVAG_378910 ( TV910 ) recombinant protein expressed in E . coli . Dilutions of the antibodies used for different experiments: For Western blot: anti-TV910 at 1:500; For IP: anti-TV910 at 1:50 . Anti- α-tubulin antibody , 12G10 produced in mouse ( a kind gift from Dr . Carsten Janke , Institute Curie , Orsay ) was used at the dilution of 1:500 for blot . HRP conjugated Goat anti-rabbit IgG and anti-mouse IgG antibodies ( Bangalore Genie ) . 1:10000 dilution was used for both the antibodies . Animals were maintained , and experiments were performed as per the principles , guidelines , and methods approved by the Institutional Animal Ethics Committee ( IAEC ) of the Central Animal Facility ( CAF ) , Indian Institute of Science , Bangalore . The methods were approved by Committee for the Purpose of Control and Supervision of Experiments on Animals ( CPCSEA ) , Ministry of Environment , Forest and Climate Change Welfare Division , Government of India . The committee approved the experiments performed ( Project Number: CAF/Ethics/269/2012 ) . TV560 Fwd: 5’-GGGGGGATCCATGTCTGCTGAAGTCGAAACACTTG-3’ , TV560 Rev: 5’-GGGGGAATTCTTAATCGACATCATCAAACTTATTAAGG-3’ , TV910 Fwd: 5’-GGGGGGATCCATGGACCTTCGTGAGAAGCTC-3’ , TV910 Rev: 5’-GGGGCCATGGCTAAAGTATACCTGCATAACATTCTTC-3’ , TV810 Fwd: 5’-GGGGGGATCCATGTTTCAAGTTATCTTTTTTGCAAAGG-3’ , TV810 Rev: 5’-GGGGCCATGGTTAGGGTTCATTGACGTTTTCG-3’ . Hsp90 isoforms were identified in Trichomonas genome by BLASTp in trichdb . org . Multiple sequence alignment was carried out for all three full-length Hsp90 isoforms–TVAG_153560 ( TV560 ) , TVAG_378910 , TVAG_030810 using MUSCLE algorithm at ebi . uk . For promoter analysis , motif search for Inr elements and M5 like elements was done for a region 400 bp upstream of transcription start site . Comparison of Hsp90 and Grp94 ( XP_641313 . 1 ) canonical motifs was done with human Hsp90 ( NP_005339 . 3 ) and Grp94 respectively . For phylogenetic clustering of Hsp90 isoforms , maximum likelihood algorithm was used as described above . Sequences used were retrieved from NCBI and Eupathdb . Gene IDs used for cytosolic Hsp90s were Saccharomyces cerevisiae NP_013911 . 1 , Homo sapiens NP_005339 . 3 , Entamoeba histolytica XP_653132 . 1/ EHI_196940 , Cryptosporidium parva XP_626924 . 1 , Babesia bovis XP_001611554 . 1 , Plasmodium falciparum XP_001348998 . 1 , Toxoplasma gondii XP_002368278 . 1 , Trypanosoma brucei A44983 , Giardia lamblia BAJ33526 . 1 , Trichomonas vaginalis TVAG_153560 , Neospora caninum XP_003881046 . 1 , Mus musculus NP_032328 . 2 , Gallus gallus NP_996842 . 1 , Danio rerio NP_571403 . 1 , Brugia malayi EDP29326 . 1 , E . coli HtpG EDV65681 . 1 and Klebsiella Htpg CCI78437 . 1 . Gene IDs for Grp94s used for analysis were: Entamoeba histolytica ( EHI_163480 ) , Cryptosporidium parva ( cgd7_3670 ) , Giardia lamblia ( DHA2_15247 ) , Plasmodium falciparum ( PF3D7_1222300 ) , Babesia bovis ( BBOV_IV008400 ) , Theileria annulata ( TA06470 ) , Toxoplasma gondii ( TGGT1_244560 ) , Neospora caninum ( NCLIV_019110 ) , Bos taurus ( NP_777125 . 1 ) , Mus musculus ( NP_035761 . 1 ) , Danio rerio ( NP_937853 . 1 ) , C . elegans ( NP_001255536 . 1 ) . Hsp90 protein sequences were retrieved by performing a simple BLASTp search with default parameters against major phyla of Protista clade including various taxonomic clades of Protista including Alveolata , Amoebozoa , Apicomplexa , Ciliophora , Diplomonadida , Euglenozoa , Microsporidia , Myxospora , and Parabasalia . Protein BLASTp search was done with default parameters using Human Grp94 protein sequence ( Accession number: AAH66656 ) and sequences were filtered on the basis of their Query coverage ( >60% ) , Identity ( >30% ) and e-value ( less than 0 . 01 ) . Protein sequences fulfilling the above criteria were considered for further analysis . The sequences not having ER retention signal ( KDEL motif or KDEL-like motifs ) were then filtered using an in-house Perl script . Multiple Sequence alignment of sequences lacking canonical retention motifs and some known Hsp90 orthologs , was performed using MUSCLE algorithm from MEGA 7 suites , followed by the construction of a Maximum Likelihood tree with 1000 bootstraps . The sequences clustering along with Cytoplasmic or Mitochondrial Hsp90 sequences were excluded from further analysis . The Hsp90 sequences clustering with Human Grp94 sequence were then confirmed to be indeed Grp94 homolog by reverse BLAST analysis . Sequences yielding known Grp94 sequence as the topmost hit were identified as true Grp94 homologs which do not have a canonical KDEL or KDEL-like retention motif . Trichomonas vaginalis isolate ( Strain is a kind gift from Prof . Daman Saluja and Dr . Manisha Yadav , ACBR , New Delhi ) was cultured in glass tubes in TYI-S-33 medium at 37°C containing 10% heat-inactivated adult bovine serum ( HiMedia ) , and 2 . 5% Diamond vitamin mix . G . lamblia Portland P1 parasites were cultured in TYI-S-33 [6] supplemented with 12% fetal bovine serum . RNA extraction was carried out using TriZol reagent ( Thermo Fisher Scientific ) according to manufacturer’s instructions . The concentration and purity of the RNA extracted were evaluated using the Nanodrop spectrophotometer ( Thermo Scientific; 1000 ) . 2 μg RNA of all samples was used to synthesize cDNA using Verso cDNA Synthesis kit ( Thermo Fisher Scientific ) according to manufacturer’s instructions . Amplification for genes was performed with respective primers as listed in previous section in Mastercycler ( Eppendorf ) . PCR products were analyzed on a 2% agarose gel with ethidium bromide . TV910 and TV560 were amplified from cDNAs of respective parasites using the primers . Amplified products were cloned in pRSET-A vector as a 6x-His tag fusion protein . Both proteins were expressed in E . coli BL21 pLysS expression strain and proteins were purified to homogeneity using Ni-NTA affinity chromatography . In brief , E . coli strains of clones were grown in LB broth and induced with 0 . 2 mM IPTG at 16˚C for 8 hours . Cells were pelleted down and lysed in buffer containing Tris-Cl pH 7 . 5 , 1 M NaCl and 10 mM imidazole and the protease inhibitors . Lysate supernatants were allowed to bind to Ni-NTA beads . Beads bound to proteins were then washed with buffer containing a gradient of imidazole concentration ranging from 10 mM to 50 mM . Proteins were finally eluted using a buffer containing 200 mM imidazole . Proteins were then dialyzed in suitable buffers described in following sections for different experiments . Protein concentration was estimated using Bradford’s reagent using BSA as standard . Trichomonas trophozoites grown to log phase were harvested by chilling on ice for 5 min and centrifuging at 600 g and were lysed by repeated freeze-thaw in liquid nitrogen in PBS containing 0 . 1% Triton X-100 and protease inhibitors . Trichomonas trophozoites grown to log scale were harvested and 15 , 000 cells were seeded per well in TYI-S-33 medium in a 96-well plate . Cells were allowed to adhere . Medium was replaced with the medium containing varying concentrations of 17-AAG from 10 nM to 100 µM . 0 . 2% DMSO was used as the vehicle control . Cells were treated with the drug for 24 hours . Post-treatment , viable Trichomonas cells were counted under a microscope using trypan blue dye exclusion method . Percent survival above control was plotted against the log concentration of 17-AAG and data was analyzed using GraphPad Prism 5 . 0 and GI50 ( 50% Growth inhibitory concentration ) was determined . Inhibition assay for extracellular Hsp90 was performed in a similar manner using Fluorescein isothiocyanate-Geldanamycin ( FITC-GA; Enzo Lifesciences ) , which is a cell impermeable inhibitor of Hsp90 [21] . ATP binding was determined using the method of fluorescence quenching upon ligand binding as described previously [8 , 12] . Briefly , 50 µg protein in binding buffer ( 40 mM Tris-Cl buffer pH 7 . 4 , 5 mM MgCl2 and 100 mM KCl ) was incubated with different concentrations of ATP ( 0 . 05–10 mM ) . Intrinsic tryptophan fluorescence was measured by scanning the emission spectrum in the wavelength range of 300–400 nm and excitation at 280 nm . The fluorescence intensity at λmax 340 nm was selected for calculations . The difference in intrinsic fluorescence of protein alone and in the presence of the ligand was plotted against the ligand concentration . Data was analyzed using GraphPad Prism 5 . 0 using non-linear regression analysis with single site-specific binding . A similar procedure was carried out for 17-AAG binding with 50 µg protein in binding buffer with the concentrations of 17-AAG ranging from 500 nM—50 μM . The final concentration of DMSO in the assay was 1% . 1 . 5 μM of Hsp90 protein in 40 mM Tris-Cl buffer , pH 7 . 4 containing 100 mM KCl and 5 mM MgCl2 was incubated with varying concentrations of ATP ( 50 to 4000 μM ) . γ32P-ATP with specific activity of 0 . 55 Ci/mmole was used as a tracer . 300 µM of 17-AAG was used in the control reaction to negate out non-specific or background activity . Control activity was subtracted from the total activity . ATPase activity was plotted against the ATP concentrations . Data was analyzed using GraphPad Prism 5 . 0 using Michaelis-Menten kinetics . ATPase inhibition assay was carried out in a similar manner as described above , except that the purified Hsp90 protein was incubated with a saturating concentration of ATP ( 2 mM ) and 17- AAG concentration was varied from 2 . 5 µM to 150 µM . 300 µM 17-AAG was used in the control reaction . Percentage residual ATPase activity was plotted against the log concentration of inhibitor and the result was analyzed using GraphPad Prism 5 . 0 . For secretion assay , log phase grown Trichomonas or Giardia trophozoites were pelleted at 600×g and washed thrice with PBS . Washed cells were resuspended at 105 cells/mL density in PBS sucrose ( 5% w/v ) . Cells were incubated at 37°C for desired time . Following incubation , cells were harvested by centrifuging at 600×g for 10 min , spent media was filtered through 0 . 22 μm syringe filter and concentrated using Amicon Ultra filters . The sample was further processed for either SDS PAGE , 2DGE or IP . Peptides were extracted either by in-gel trypsin digestion from each gel piece , or the protein solution was processed for digestion in solution . Proteins were first reduced using 10 mM DTT and alkylated by 55 mM iodoacetamide followed by 16 hours in-gel trypsin digestion ( 20 ng/µL , Promega ) . Peptides were then extracted in 5% formic acid and 60% acetonitrile . Extracted peptides were dried under vacuum . Vacuum dried digested peptide mixtures were dissolved in the solvent ( 2% acetonitrile and 98% water containing 0 . 5% formic acid ) and further fractionated by Reverse Phase chromatography on C-18 material on a nano- HPLC system connected online to a nanospray ESI hybrid Q-TOF mass spectrometer from Applied Biosystems . A 100 µm ID , 5 µm particle size , 100 Å porosity Michrom column was used for chromatographic separation . The RP chromatography runtime was set for 144 minutes with a flow rate of 300 nL/minute . The mobile phase was adjusted to an increasing concentration of acetonitrile from 5% to 90% , to elute peptides from the column based on their hydrophobicity . Analyst QS software was used to systematically acquire the TOF MS and MS/MS data for each precursor ion entering the instrument from the nanoLC . Eluted peptides were analyzed by one full MS scan and four consecutive product ion scans of the four most intense peaks , using rolling Collision Energy and Dynamic Background Subtract function . An Information Dependent Acquisition ( IDA ) experiment was used to specify the criteria for selecting each parent ion for fragmentation which included selection of ions in m/z range: > 400 and <1600 , of charge state of +2 to +5 , exclusion of former target ions for 30 seconds , accumulation time of 1 second for a full scan and 2 seconds for MS/MS , ion source voltage of 2200 V . The original data files were analyzed using the Protein Pilot software from a combined database ( Swiss Prot 2005 , TrEMBL , NCBI , and PDB ) . Peptide scores above 50 and a protein score of minimum 1 . 3 corresponding to a confidence level greater than 95% were used . We further developed an MRM ( Multiple reaction monitoring ) based detection method for TV910 . The Skyline was used to generate peptides by in silico trypsin digestion and to predict their MRM . The length of the peptide was kept between 8–16 amino acids . The peptide with consecutive K and R ( KR , KK , RR , and RK ) were not considered to avoid miscleavage sites by trypsin . Peptides with cysteine and methionine were avoided due to their instability . Two peptides were selected , based on the previously mentioned criteria [22] . Their uniqueness was confirmed to T . vaginalis TV910 by BLASTp analysis . The selected peptides were DELINNLGGIAK and IENVILSK . Using pure TV910 , MS response and retention time were determined . Optimized MRM transitions and chromatographic conditions of the selected peptides are described in S2 and S3 Tables . Secretion assay was performed as described in the previous section and secreted proteins were then resolved on an SDS-PAGE . The band near ~ 90 kDa for blank and the lysate was excised and processed as mentioned above . The digested peptide mixtures were dissolved in the solvent ( 2% acetonitrile and 98% water containing 0 . 5% formic acid ) and analysed on LC-MS/MS ( Agilent 1260 Infinity LC coupled with 6460 Triple Quad MS ) . Gradient chromatographic separation was achieved on Agilent Eclipse C18 ( 4 . 6 ×250 mm , 5μm ) column by using mobile phase A ( 0 . 1% formic acid in MilliQ water ) and mobile phase B ( 0 . 1% formic acid in Acetonitrile . The run time was 32 min , at the flow rate of 0 . 4 ml/min . Data analysis was done by Agilent Mass Hunter Qualitative Analysis B . 07 . 00 . Cells were treated with BFA at a concentration of 10 μg/mL for 2 hours at 37°C at a cell density of 105 cells/mL . Trichomonas cells were starved by incubating in Cys and serum-free medium for two hours . S35 met and cys radiolabel was added to the culture at 100 µCi/mL . Cells were incubated with the label for 2 hours and label was washed off by PBS washes . Chase for different time points or in presence or absence of BFA was carried out . Lysate and spent medium samples were prepared as described in previous sections . Pre-clearing of samples was done by incubating 1/10th volume of Protein A-agarose beads with samples at 4°C on an end-to-end rotor for 1 . 5 hours . After pre-clearing , beads were spun down and the supernatant was taken in a fresh tube , desired primary antibody was added to test samples and in Protein A control , the antibody was not added . Samples were incubated for 12 hours at 4°C on end to end rotor . After 12 hours 1/10th volume of Protein A- agarose beads were added to samples and incubated for 3 hours at 4°C . After this step , Protein-A beads were spun down and were washed with 1 ml IP buffer ( PBS with 0 . 1% Triton X-100 ) on vortex for 15 minutes each and four such washes were done . After washes , the beads alone were taken and suspended in 50 µL of Laemmli buffer and were boiled for 10 minutes and centrifuged at 20 , 000 rpm for 15 minutes . The supernatant was then resolved on an SDS- 10% Polyacrylamide gel . The gel was coomassie stained and dried . The dried gel was exposed on phosphor screen for 5 days and scanned by PhosphorImager .
There are seven genes annotated as Hsp90 in TrichDB . org ( Fig 1A ) . Amongst the 7 annotated genes , 5 genes are homologous to cytosolic Hsp90 . Among these 5 cytosolic isoforms , there is only one gene , TVAG_153560 ( TV560 ) , which codes for a full-length Hsp90 with 721 amino acids and is an intron less ORF of 2 . 1 kb . The other 4 genes are the products of non-sense mutation leading to premature termination of Hsp90 , which if expressed can code for truncated Hsp90 with an N-terminal domain and a part of the middle domain . These nonsense mutations in genes TVAG_155010 and TVAG_034730 have resulted in the C-terminal halves to being annotated as separate genes TVAG_155020 and TVAG_034730 . Two of the remaining genes , TVAG_030810 and TVAG_378910 showed significant identity to Grp94 , an ER resident paralog of Hsp90 . Both the genes were present on two different scaffolds , DS113541 ( TVAG_030810 ) and DS113184 ( TVAG_378910 ) . To ascertain their identity , phylogenetic analysis was carried out . A phylogenetic tree was constructed using the sequences of well-annotated cytosolic Hsp90s and Grp94 using the maximum likelihood method . These sequences were retrieved by BLASTp analysis using human cytosolic Hsp90 and Grp94 as queries in NCBI and EupathDB databases . The tree shows that full-length cytosolic Hsp90 of Trichomonas vaginalis ( TVAG_153560 ) clustered with cytosolic Hsp90s . However , both TVAG_030810 and TVAG_378910 clustered with Grp94 clade ( Fig 1B ) . Any potential homology with TRAP1 ( a mitochondrial Hsp90 paralog ) was also ruled out by sequence and phylogenetic analysis ( S2 Fig ) . For our further analysis , we focused only on the three full-length Hsp90 isoforms , cytosolic Hsp90 TV560 , and Grp94 orthologs , TV910 and TV810 . We performed sequence analysis and aligned the three isoforms using MUSCLE algorithm . Cytosolic Hsp90 had a canonical DDVD motif at the C-terminal end , which is conserved and is responsible for binding to co-chaperones and clients containing TPR domain . However , upon a closer look , we noticed that both the Grp94 orthologs TV910 and TV810 showed neither a canonical ER signal peptide at the N–terminus , nor a retention sequence at the C–terminus of the protein ( Figs 1C and S1 ) . It has been noticed in Trichomonas that many of the ER and exported proteins lack a canonical signal sequence [23 , 24] . In the alignment ( S1 Fig ) , we noticed TV810 and TV910 have additional 46 and 18 residues at the start of their N-termini , respectively . These extra nucleotides at the start of the N-terminal domain encoded by these genes could possibly indicate signal peptide ( Fig 1C ) . ER resident proteins are characterized by the presence of an ER retention signal KDEL ( or sequence variants of this signal ) . However , both TV810 and TV910 lack this ER retention sequence ( Figs 1C and S1 ) . When we analyzed sequences of other conserved ER resident proteins of Trichomonas like Bip ( TVAG_092490 ) , calreticulin ( TVAG_122020 ) etc . we found them to have an ER retention signal . Overall , we could count a total of 171 proteins in Trichomonas vaginalis to have an ER retention signal KDEL or variants ( sequences ending with DEL , EEL , EDL , and DDL ) ( S1 Table ) . We also searched for the homolog of KDELR1 ( KDEL endoplasmic reticulum protein retention receptor 1 ) in the Trichomonas genome . This receptor protein recognizes the ER retention signal and retrieves resident soluble proteins from the Golgi [25] . We could identify a homolog of KDELR1 , TVAG_242900 , in the Trichomonas genome . These observations suggested that the sorting mechanism of ER resident proteins is putatively conserved in Trichomonas . Overall , our observations suggest that Grp94 orthologs in Trichomonas lack an ER retention signal and do not share this peculiarity with other bonafide ER resident proteins of Trichomonas . To the best of our knowledge no other study describes a Grp94 ortholog lacking an ER retention signal . To check if the absence of an ER retention signal is a rare occurrence limited to Trichomonas Grp94s , or is a more common phenomenon we analyzed Grp94 homologs from various taxonomic clades of Protista including Alveolata , Amoebozoa , Apicomplexa , Ciliophora , Diplomonadida , Euglenozoa , Myxospora , Parabasalia and related microsporidia . Human Grp94 and TV910 were used as the query for BLASTp search in NCBI database for above-mentioned taxa . A total of 913 sequences were retrieved . From these putative Grp94 sequences , all proteins containing ER retention signal and its variants at the C-terminus ( DEL , EEL , EDL & DDL ) were eliminated . A second round of filtering was carried out for cytosolic Hsp90s , mitochondrial Trap1 , or apicoplast Hsp90 ( the organellar Hsp90 paralog of apicoplast ) , showing high similarity to Grp94 , by reverse BLAST analysis . Final validation of an ER retention signal lacking Grp94s was done by constructing a phylogenetic tree using sequences of annotated Hsp90 paralogs of different cellular organelles and cytosol using the maximum likelihood method . The analysis revealed that only 5 sequences of Grp94 lacked an ER retention signal among those analyzed ( S2 Fig ) . These sequences belong to three genera . Two of them were TV910 and TV810 from Trichomonas vaginalis , two from Mitosporidium daphniae ( KGG51005 . 1 and KGG52886 . 1 ) and one from Ichthyophthirius multifiliis ( MG5_078890 ) . This shows that the absence of an ER retention signal in Grp94 is indeed a rare phenomenon and warranted further investigation . The absence of an ER retention signal further suggests that these particular Grp94 orthologs in Trichomonas may not be retained in ER . Pseudogenes are common in the vastly expanded genome of Trichomonas , most of these pseudogenes are due to absence of promoter elements . In comparison with the promoter elements identified in Trichomonas , both the Grp94 orthologs , TV810 and TV910 , along with cytosolic Hsp90 TV560 , were found to be under the control of similar promoter-like elements ( Fig 2A ) . Sequences similar to M5-like elements and Inr elements were found to be present upstream of all the three Hsp90 isoforms [26] . Further , expression of all three isoforms was confirmed by RT PCR . Briefly , total RNA of Trichomonas trophozoites was isolated and cDNA was prepared . The presence of TV560 , TV810 and TV910 mRNA was confirmed by gene-specific primers by RT PCR . Fig 2B shows PCR amplicons of full-length genes for all three isoforms . This confirms that both the Grp94 orthologs which lack an ER retention signal , in addition to cytosolic Hsp90 , are expressed by T . vaginalis and are not pseudogenes . We further investigated if a functional Hsp90 is required for parasite growth . Therefore , to examine the effect of Hsp90 inhibition on growth , Trichomonas trophozoites were grown to log phase then treated with varying concentrations of 17-AAG in the range of 10 nM to 100 µM for 24 hours . Cell survival was measured by counting viable cells using trypan blue dye exclusion methodology . Percent survival was plotted against Log10 [17-AAG] concentration . Complete cell death was observed at higher drug concentrations . Growth inhibitory concentration GI50 for 17-AAG treatment was 708 nM ( Fig 2C ) . This shows a functionally important role for Hsp90 in Trichomonas . Grp94 orthologs TV810 and TV910 show high conservation and both lack an ER retention signal and possess a putative signal sequence ( Fig 1C ) , therefore , for all further experiments , only one isoform was investigated , and we chose TV910 for this purpose and compared its biochemical parameters with cytosolic Hsp90 TV560 . We showed that both TV560 and TV910 can bind ATP with a strong affinity with a kd of 538 . 3 µM and 722 μM , respectively ( Fig 3A ) . We then measured the catalytic activity of the two Hsp90 isoforms . Briefly , ATPase activity of Hsp90s was analyzed by incubating purified Hsp90s with increasing concentrations of ATP . γ32 P-ATP was used as a tracer , and ATP hydrolysis rate was analyzed on a TLC plate . Fractional cleavage of ATP by Hsp90 was used to calculate enzyme velocity , which was plotted against corresponding ATP concentrations . In the control reactions , 300 μM of the Hsp90 inhibitor , 17-AAG , was added to inhibit Hsp90 activity and reveal only background activity , which is subsequently subtracted from the activity of Hsp90 to negate nonspecific background . The data was analyzed by GraphPad Prism using non-linear regression analysis for Michaelis-Menten kinetics ( Fig 3B and 3C ) . KM for TV560 and TV910 were found to be 486 . 6 µM and 1206 µM , respectively . TV910 showed a higher KM value compared to TV560 . The kcat was found to be similar at 0 . 263 min-1 and 0 . 176 min-1 respectively for TV560 and TV910 . The catalytic efficiency was found to be 5 . 049×10−4 min-1µM-1 and 1 . 463 ×10−4 min-1µM-1 respectively for TV560 and TV910 suggesting cytosolic Hsp90 is a more active ATPase . 17-AAG is a known inhibitor of Hsp90 ATPase . The dissociation constant ( kd ) for 17-AAG was observed to be 11 . 45 µM and 15 . 73 µM respectively for TV560 and TV910 ( Fig 3D ) . 17-AAG could inhibit the ATPase activity of both the Hsp90 isoforms , though weakly for TV910 ( IC50: 87 . 05 µM ) when compared to cytosolic Hsp90 ( IC50: 22 . 98 µM ) . ( Fig 3E ) . Overall , our biochemical analysis suggests that both cytosolic Hsp90 and the Grp94 ortholog TV910 are active in vitro . Compared to human Hsp90 , Trichomonas Hsp90s are more active ATPases ( Fig 3F ) . Our observations suggest that TV910 encodes an Hsp90 isoform with functional ATPase activity . To study the localization of TV910 in Trichomonas , an antibody against full-length recombinant TV910 was raised in rabbit . The anti-TV910 antibody was purified from rabbit serum using Protein-A affinity column . Antibody α-TV910 was found to be specific to TV910 and did not cross-react with purified TV560 or TV810 proteins ( S3 Fig ) . A hallmark of secreted proteins is the presence of a signal sequence which is recognized by SRP ( signal recognition particle ) . The newly synthesized polypeptide is translocated to ER with the aid of SRP and SRP receptor on the ER membrane . The protein is folded in the ER and is transported to Golgi apparatus by COPII coated vesicles and those secretory proteins lacking an ER retention signal are further secreted through Golgi [29] . As described before , TV910 is a Grp94 homolog with an extended N-terminal sequence that could potentially be a signal peptide and lacks an ER retention signal ( Figs 1C and S1 ) , therefore , we hypothesized that probably TV910 is secreted by Trichomonas . To check if TV910 is indeed secreted , T . vaginalis trophozoites were harvested at log phase of growth . They were washed thrice with PBS to remove any medium and resuspended in PBS-sucrose ( 5% w/v ) and incubated for 3 hours at 37°C at a cell density of 105 cells/ml . Cells were pelleted and lysed and spent media was filtered using 0 . 22 μM filter and concentrated using Amicon ultra-15 filters . Both lysate and spent media were resolved on an SDS-PAGE followed by western transfer . The blot was probed with the α-TV910 antibody . A clear signal for TV910 was observed at its molecular weight of ~89 . 5 kDa in both the lysate and spent medium ( Fig 4A ) . As a control to rule out any lysis and leakage of cellular content into the spent media , the blot was probed with α-alpha tubulin for which the signal was observed only in the lysate and not in the spent media ( Fig 4A ) . Alpha tubulin has been previously shown not to be present in spent media of T . vaginalis and is used as a control for cellular integrity [23 , 30] . As an additional control , a secretion assay was set up for Giardia lamblia trophozoites . Giardia is an early branching close relative of Trichomonas . The experiment was set up in the same way as described for Trichomonas . A very prominent signal for GlHsp90 was observed in the cell lysate , but not in spent media ( Fig 4B ) . To further validate secretion of TV910 , proteins present in spent media were resolved on an SDS-PAGE and stained with Coomassie brilliant blue ( Fig 4C ) . The gel region between 75–100 kDa was excised and trypsin digestion was performed . The tryptic digest was analyzed on an LC-MS/MS ESI- QTOF system . Data was analyzed using Protein Pilot 5 against Trichomonas database downloaded from NCBI . TV910 was identified in spent media with high confidence . Fig 4D shows MS/MS spectrum of peptide “VTEDPRGNTLGR” that was identified with a confidence of more than 95% . Fig 4E shows the TV910 sequence and peptides identified highlighted in color . Thus , we confirmed by MS/MS that TV910 is indeed secreted by Trichomonas into the spent medium . We also developed an MRM-based method to detect TV910 in the secreted medium . The tryptic digest from blank ( only PBSS ) and the spent medium were analyzed using Agilent 1260 Infinity LC coupled with 6460 Triple Quad MS . Based on the detection of the two unique peptides , our analysis showed the presence of TV910 in the spent medium; further strengthening our biochemical data ( S4 and S5 Figs ) . Analysis using MRM can be further used to quantitate levels of protein ( s ) in the spent medium and can be a robust alternative method to antibody-based detection . Steady-state levels show the presence of TV910 in the secreted medium . To further study the kinetics of secretion , a pulse-chase experiment was performed . Briefly , cells were starved in cysteine- and serum-free medium for 2 hours following which they were pulsed for two hours by metabolic labeling of proteins using S35 containing Met-Cys mix . Following the pulse , cells were washed and labeled medium was removed . Labeled cells were divided into five equal aliquots and a chase in PBS-sucrose ( 5% w/v ) was carried out at 37°C for different time points ( 0 H , 0 . 5 H , 1 H , 2 H and 4 H ) . Labeled TV910 was pulled down from spent media using α-TV910 antibody . Signal was detected using autoradiography . Fig 4F ( a ) shows the immunoprecipitation profile of TV910 at different time intervals . It can be clearly seen that signal for TV910 starts to appear 2 H and increases by 4 H . At 0 H no signal was observed for TV910 . At 0 . 5 H and 1 H also there is no significant signal for TV910 . Fig 4F ( b ) shows the total profile of labeled proteins in the cell . A decrease in the labeled proteins can be seen by increasing time . Total protein content was equal in all the lanes . Fig 4F ( c ) shows the total profile of labeled proteins in spent media at different time points of the chase in the unlabeled medium . A clear increase in total secreted labeled proteins can be seen by increasing time . As an additional control , pulse-chase ( pulse of 2 H and chase of 4 H ) was performed for Giardia trophozoites , as described above followed by IP for GlHsp90 . No signal for GlHsp90 was seen in spent media and GlHsp90 was pulled down only in case of lysate ( Fig 4F d ) . Overall , we show that TV910 is actively secreted by Trichomonas in a time-dependent manner . To study if TV910 is secreted via a classical ER–Golgi secretory pathway , secretion assays were performed in the presence of Brefeldin-A ( BFA ) . BFA is a fungal metabolite that blocks the formation of COP-I coated vesicles by inhibiting the small GTPase Arf1p . This causes redistribution of Golgi proteins and thus , inhibits classical ER-Golgi pathway [31 , 32] . We looked at the effect of BFA on TV910 secretion using two different approaches . Firstly , the effect of BFA was tested on steady state levels of TV910 immunoblot . Cells were treated with 10 μg/mL of BFA for 2 hours , followed by secretion assay in PBS-sucrose for 3 hours as described before . An equal number of cells were treated with vehicle control ( ethanol ) . Spent media was filtered and concentrated . Total secreted proteins for control and BFA-treated cells in spent media were resolved on SDS PAGE followed by western blot . Fig 5A and 5B show blot for TV910 in control treated spent media and BFA-treated spent media and quantitation of TV910 signal , respectively . A clear decrease can be seen in TV910 secretion upon BFA treatment . Ponceau profile shows equal loading . Further , a labeled pulse-chase experiment was performed for TV910 , as described in the previous section with a pulse of two hours and chase of 0 and 3 hours . Cells were treated with vehicle ( ethanol ) or BFA for two hours during the pulse with S35-Met-cys label . This was followed by an IP for TV910 . As can be clearly seen in Fig 5C , no signal for TV910 was seen at 0 H in both control and BFA-treated spent media , however , at 3 H there is a clear signal for TV910 in control and no signal in BFA-treated spent media . Fig 5D shows total labeled proteins in the lysate and there is a decrease in the amount of labeled proteins from 0 H to 3 H as expected . There is a comparatively higher amount of labeled proteins in BFA-treated cells , which is due to inhibition of secretion and retention of more proteins in the cell compared to control . Fig 5E shows that there is an increase in the amount of total labeled proteins from 0 to 3 H in spent media of control cells , and at 3 H , spent media of BFA-treated cells have a lesser amount of labeled protein as expected . These results altogether suggest that BFA blocks TV910 secretion . Thus , we can conclude that TV910 follows classical ER-Golgi pathway for secretion .
Trichomonas vaginalis is a clinically important parasite and a common cause of STD , with millions of new infections worldwide every year . The problem of STD is compounded by poor hygiene in lower socio-economic classes , lack of good healthcare facility and social taboo associated with it . These limitations prevent the timely treatment of disease in many patients . Trichomonas ranks highest in terms of prevalence and incidence among the four major non-viral treatable sexually transmitted infections that include Chlamydia trachomatis , Neisseria gonorrhoeae , and Treponema pallidum . Trichomonas vaginalis has to deal with constant stress in its physiological niche of the urogenital tract that includes changes in pH , fluctuation in iron balance and other nutrients and desquamation of vaginal epithelial cells associated with menstrual cycle . In this study , we show that the parasite critically depends on Hsp90 for its growth and Hsp90 inhibition is lethal . We identified Grp94 orthologs lacking an ER retention signal . We show that TV910 is actively secreted by the parasite into the extracellular milieu . We provide biochemical and proteomics evidence for the secretion of this Hsp90 isoform . We observed that lack of an ER retention signal in a Grp94 ortholog is a rare phenomenon . Other canonical ER resident proteins of Trichomonas possess the conserved ER retention signal at their C-termini and the ER sorting machinery appears to be conserved at least at the genomic level . We show that TV910 follows the classical ER-Golgi secretory pathway and inhibition of this pathway by BFA blocks its secretion . To our knowledge , this is the first genome-annotated Hsp90 to be secreted by a parasitic protozoan . We analyzed mass-spectrometry data from our studies and other reports [23 , 33] for other secreted proteins of Trichomonas and found potential Hsp90 interactors in the secretome , including Hsp70 , peptidyl- prolyl-isomerase that are the known co-chaperones of Hsp90 . It will be of interest to see if the extracellular TV910 isoform can interact with them . Many potential Hsp90 client proteins , including metabolic and signaling molecules , are also present in the secreted proteome [23 , 33] . For establishing infection , Trichomonas requires to break through the mucus layer and adhere to epithelial cells of urogenital tract [34] . It also penetrates into basement membrane and binds to extracellular matrix proteins [35] . The role of extracellular Hsp90 has been previously shown to be important in migration and invasion of tumor cells [16 , 17 , 20] . Grp94 has also been shown to have a tumor-specific cell surface expression [15 , 36] . Inhibition of extracellular Hsp90 using cell-impermeable Hsp90 inhibitor FITC- Geldanamycin [21] has no apparent effect on the growth of the parasite in the axenic culture conditions ( S6 Fig ) . This is not surprising as one would expect the intracellular Hsp90 to be essential for growth which we have shown in this study . However , based on the functions described for Hsp90 isoforms in the extracellular space , we can speculate that extracellular Hsp90 may play an important role in virulence and pathogenesis of Trichomonas while infecting the host and this needs to be investigated . One of the major reasons for the high prevalence of Trichomonas infection is the lack of proper diagnosis of the infection due to various reasons including absence of reliable diagnostic methods , especially in developing countries . Based on communication with many clinicians , we found out that there is an absence of a defined diagnostic strategy , and often diagnosis happens based on symptoms . Very few labs carry out microscopic observations for parasites in infected samples . One of the major implications for secreted Hsp90 of Trichomonas is the possibility of its use as a biomarker to detect infection and our MRM based detection method can be further tested in clinical settings . Overall , we show that Trichomonas Grp94 orthologs lack the ER retention signal and the parasite secretes Hsp90 . To date , major focus on extracellular Hsp90 has been in the context of cancer . We show that a protozoan parasite is actively secreting Hsp90 which may have an important role in its virulence and survival . | Hsp90 is an essential chaperone in eukaryotes and it is often described as a master regulator of cellular homeostasis . In addition to its well-known functions inside the cell , extracellular Hsp90 has also been implicated in migration and invasion of tumor cells . We have , for the first time , identified the presence of an extracellular Hsp90 in a parasitic protozoan , Trichomonas vaginalis . The extracellular Hsp90 is a Grp94 homolog that lacks a canonical ER retention signal . Our analysis of Grp94 sequences from protozoa shows that it is uncommon for a Grp94 to lack ER retention signal . In the current study , we characterized the biochemical parameters and established the extracellular localization of this Hsp90 paralog . This secreted Hsp90 in Trichomonas can potentially modulate host-pathogen interaction . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"enzymes",
"endoplasmic",
"reticulum",
"cell",
"processes",
"enzymology",
"phosphatases",
"physiological",
"processes",
"sequence",
"motif",
"analysis",
"trichomonas",
"cellular",
"structures",
"and",
"organelles",
"research",
"and... | 2018 | A secreted Heat shock protein 90 of Trichomonas vaginalis |
Ebola virus ( EBOV ) causes severe and often fatal hemorrhagic fever in humans and nonhuman primates ( NHPs ) . Currently , there are no licensed vaccines or therapeutics for human use . Recombinant vesicular stomatitis virus ( rVSV ) -based vaccine vectors , which encode an EBOV glycoprotein in place of the VSV glycoprotein , have shown 100% efficacy against homologous Sudan ebolavirus ( SEBOV ) or Zaire ebolavirus ( ZEBOV ) challenge in NHPs . In addition , a single injection of a blend of three rVSV vectors completely protected NHPs against challenge with SEBOV , ZEBOV , the former Côte d'Ivoire ebolavirus , and Marburg virus . However , recent studies suggest that complete protection against the newly discovered Bundibugyo ebolavirus ( BEBOV ) using several different heterologous filovirus vaccines is more difficult and presents a new challenge . As BEBOV caused nearly 50% mortality in a recent outbreak any filovirus vaccine advanced for human use must be able to protect against this new species . Here , we evaluated several different strategies against BEBOV using rVSV-based vaccines . Groups of cynomolgus macaques were vaccinated with a single injection of a homologous BEBOV vaccine , a single injection of a blended heterologous vaccine ( SEBOV/ZEBOV ) , or a prime-boost using heterologous SEBOV and ZEBOV vectors . Animals were challenged with BEBOV 29–36 days after initial vaccination . Macaques vaccinated with the homologous BEBOV vaccine or the prime-boost showed no overt signs of illness and survived challenge . In contrast , animals vaccinated with the heterologous blended vaccine and unvaccinated control animals developed severe clinical symptoms consistent with BEBOV infection with 2 of 3 animals in each group succumbing . These data show that complete protection against BEBOV will likely require incorporation of BEBOV glycoprotein into the vaccine or employment of a prime-boost regimen . Fortunately , our results demonstrate that heterologous rVSV-based filovirus vaccine vectors employed in the prime-boost approach can provide protection against BEBOV using an abbreviated regimen , which may have utility in outbreak settings .
The viruses in the family Filoviridae and within the genera Ebolavirus ( EBOV ) and Marburgvirus ( MARV ) cause severe and often fatal hemorrhagic fever ( HF ) in humans and nonhuman primates ( NHPs ) [1] , [2] . Case fatality rates with these viruses range from 23–90% depending on the strain and/or species . The EBOV genus is diverse and , as of 2007 , consisted of four species: Sudan ebolavirus ( SEBOV ) , Zaire ebolavirus ( ZEBOV ) , Côte d'Ivoire ebolavirus ( CIEBOV ) , and Reston ebolavirus ( REBOV ) . A fifth species , Bundibugyo ebolavirus ( BEBOV ) was discovered during an outbreak in Uganda during 2007/08 [3] . Before 2012 , the EBOV genus had accounted for at least 22 outbreaks dating back to 1976 with 18 of these occurring within the last 20 years [4] . In 2012 there were two separate outbreaks of EBOV; SEBOV in Uganda [5] and BEBOV in the Democratic Republic of Congo ( DRC ) [6] . The increased frequency of EBOV outbreaks together with the potential for deliberate misuse has increased public health concerns regarding filoviruses . Case fatality rates frequently range between 70% and 90% in ZEBOV outbreaks , 50–55% for SEBOV episodes , and 40–48% for BEBOV outbreaks . CIEBOV caused deaths in chimpanzees and a severe nonlethal human infection in a single case in the Republic of Côte d'Ivoire in 1994 [7] . REBOV is highly lethal for macaques but is not thought to cause disease in humans [8] . Presently , there are no licensed vaccines or post-exposure treatments available for human use; however , there are at least seven different vaccine candidates that have shown the potential to protect NHPs from lethal EBOV and/or MARV infection using platforms based on DNA vectors , recombinant Adenovirus ( rAd ) vectors , combined DNA/rAd vectors , virus-like particles ( VLPs ) , alphavirus replicons , recombinant human parainfluenza virus 3 ( rHPIV3 ) , and recombinant vesicular stomatitis virus ( rVSV ) [9] , [10] , [11] , [12] , [13] , [14] , [15] , [16] , [17] , [18] , [19] , [20] , [21] , [22] , [23] , [24] , [25] , [26] , [27] , [28] , [29] . The EBOV vaccine systems rely on antigens specific for each species of virus to provide protection against lethal challenge in NHP models; however , there is no description of a vaccine approach yet that can provide 100% single immunization cross species protection against challenge with an emerging filovirus such as BEBOV . The rVSV filovirus vaccine platform , reported on herein , relies on the filovirus glycoprotein ( GP ) as the immunizing antigen [12] . Current data suggest that the GP from each filovirus species can only protect against homologous challenge when using the rVSV vaccine platform as a single injection [16] , [25] . Cross-protection with the rVSV vaccines has been achieved using a blended vaccination strategy where a mixture of three separate vaccine vectors , rVSV-MARV-GP , rVSV-ZEBOV-GP , and rVSV-SEBOV-GP were able to protect against separate challenge with either MARV , ZEBOV , CIEBOV , or SEBOV in NHPs [16] . Although cross-protection was achieved using this blended vaccination strategy against challenge of known species of EBOVs , the BEBOV outbreak in 2007 offered a new challenge to develop a strategy to protect against an emerging species of EBOV using existing vaccines that were available at the time of the outbreak . This strategy was tested in cynomolgus macaques using two different vaccine platforms against heterologous challenge with BEBOV; the DNA/rAd platform [23] and the rVSV-filovirus-GP platform [25] where the mortality rate for BEBOV in cynomolgus macaques was found to be 66 to 75% . The study using the DNA/rAd platform consisted of four ZEBOV-GP/SEBOV-GP DNA vaccinations given over the course of 14 weeks and a boost vaccination consisting of the ZEBOV rAd5 GP ZEBOV vector 12 months after the final DNA vaccination . This strategy , although long and complicated , was able to confer 100% protection to the NHPs used in the study [23] . In contrast , the rVSV vaccine strategy employed to protect against heterologous challenge with BEBOV was a single vector strategy . The NHPs in this study were vaccinated with rVSV-ZEBOV-GP or rVSV-CIEBOV-GP separately and challenged with BEBOV 28 days after vaccination . While the rVSV-CIEBOV-GP vector did not provide any additional protection when compared to mock-vaccinated control NHPs in the study ( 33% survival ) , the rVSV-ZEBOV-GP vaccine conferred 75% survival [25] . This result was surprising when one considers CIEBOV is more genetically related to BEBOV when compared to ZEBOV [3] . The use of SEBOV and ZEBOV GP as antigens to confer 100% protection against cross species challenge with BEBOV using the DNA/rAd strategy [23] suggested that if the rVSV-SEBOV-GP vaccine was used in combination with rVSV-ZEBOV-GP the cross species protection using the rVSV system would increase from 75% with just rVSV-ZEBOV-GP [25] to 100% protection . Here , we evaluated the utility of combining rVSV-SEBOV-GP and rVSV-ZEBOV-GP vectors using either a single injection blended vaccination approach or in a prime-boost regimen against heterologous BEBOV challenge in cynomolgus macaques . Furthermore , we assessed the ability of a single injection of a newly developed homologous rVSV-BEBOV-GP vaccine vector to provide protection against homologous BEBOV challenge .
Healthy , adult cynomolgus macaques ( Macaca fascicularis ) were handled in Animal BSL-2 and BSL-4 containment space in the Galveston National Laboratory ( GNL ) at the University of Texas Medical Branch ( UTMB ) , Galveston , Texas . Research was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals , and adhered to principles stated in the Eighth edition of the Guide for the Care and Use of Laboratory Animals , National Research Council , 2013 . The facility where this research was conducted ( UTMB ) is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International and has an approved OLAW Assurance #A3314-01 . Research was conducted under animal protocol number 1011057 approved by the UTMB Institutional Animal Care and Use Committee ( IACUC ) . All steps were taken to ameliorate the welfare and to avoid the suffering of the animals in accordance with the “Weatherall report for the use of nonhuman primates” recommendations . Animals were housed in adjoining individual primate cages allowing social interactions , under controlled conditions of humidity , temperature , and light ( 12-hour light/12-hour dark cycles ) . Food and water were available ad libitum . Animals were monitored ( pre- and post-infection ) and fed commercial monkey chow , treats and fruit twice daily by trained personnel . Environmental enrichment consisted of commercial toys . All procedures were conducted by trained personnel under the oversight of an attending veterinarian and all invasive clinical procedures were performed while animals were anesthetized . Endpoint criteria was specified and approved by the UTMB IACUC . The rVSV-filovirus GP-vectors , rVSV-ZEBOV-GP ( strain Mayinga ) , rVSV-SEBOV-GP ( strain Boniface ) , and rVSV-BEBOV-GP ( Fig . 1A ) were recovered from cDNA as previously described [30] , [31] . BEBOV , strain 200706291 , was isolated from a fatal human case in western Uganda during the outbreak in 2007 [3] . The challenge stock of BEBOV used in this study was propagated on Vero E6 cells twice making this a passage 2 virus . The viral genomes from this stock were Sanger-sequenced across the GP editing site and it was confirmed that the sequence from bases 6900 to 6907 was wild-type BEBOV ( Accession: NC 014373 . 1 ) . The BEBOV challenge stock was kindly provided by Dr . Thomas G . Ksiazek . The rVSV-filovirus-GP vector preparations and BEBOV challenge virus stocks were assessed for the presence of endotoxin using The Endosafe®-Portable Test System ( PTS ) ( Charles River , Wilmington , MA ) . Virus preparations were diluted 1∶10 in Limulus Amebocyte Lysate ( LAL ) Reagent Water ( LRW ) per manufacturer's directions and endotoxin levels were tested in LAL Endosafe®-PTS cartridges as directed by the manufacturer . Each preparation was found to be below detectable limits while positive controls showed that the tests were valid . Twelve , healthy , filovirus-naïve , adult ( 5 to 12 kg ) , male cynomolgus macaques ( Macaca fascicularis ) were randomized into four different groups of three animals each ( Groups 1 , 2 , 3 , and 4; Figure 1B ) . Animals were vaccinated by intramuscular ( i . m . ) injection of an identical volume of PBS ( Group 1 ) , ∼2×107 plaque-forming units ( PFU ) of rVSV-BEBOV-GP ( Group 2 ) , or ∼1×107 PFU of rVSV-SEBOV-GP and ∼1×107 PFU of rVSV-ZEBOV-GP ( Groups 3 and 4 ) . While Groups 3 and 4 received the same vaccine vectors the dosing regimens were different , as shown in Figure 1B and C . Group 3 received a single inoculation that was an equal blend of the two vaccines . Group 4 received two inoculations , the rVSV-SEBOV-GP vaccine first and the rVSV-ZEBOV-GP vaccine 14 days later . Four ( Groups 1 , 2 , and 3 ) or 5 ( Group 4 ) weeks after the initial vaccination , all animals were challenged i . m . with 1 , 000 PFU of BEBOV . Animals were monitored for clinical signs of illness ( temperature , weight loss , changes in blood count , and blood chemistries ) during the vaccination and BEBOV challenge portions of the study . Viremia was analyzed after vaccination and challenge . Physical exams were given when blood was collected on days of vaccination and 8 days before challenge and on days 0 , 3 , 6 , 10 , 14 , 21 , and 28 post-challenge ( Fig . 1C ) . Total white blood cell counts , white blood cell differentials , red blood cell counts , platelet counts , hematocrit values , total hemoglobin concentrations , mean cell volumes , mean corpuscular volumes , and mean corpuscular hemoglobin concentrations were analyzed from blood collected in tubes containing EDTA using a laser based hematologic analyzer ( Beckman Coulter , Brea , CA ) . Serum samples were tested for concentrations of albumin , amylase , alanine aminotransferase ( ALT ) aspartate aminotransferase ( AST ) , alkaline phosphatase ( ALP ) , gamma-glutamyltransferase ( GGT ) , glucose , cholesterol , total protein , total bilirubin ( TBIL ) , blood urea nitrogen ( BUN ) , creatine ( CRE ) , and C-reactive protein ( CRP ) by using a Piccolo point-of-care analyzer and Biochemistry Panel Plus analyzer discs ( Abaxis , Sunnyvale , CA ) . RNA was isolated from whole blood utilizing the Viral RNA mini-kit ( Qiagen ) using 100 µl of blood into 600 µl of buffer AVL . Primers/probe targeting the GP gene of BEBOV were used for quantitative real-time PCR ( qRT-PCR ) as used previously [25] with the probe used here being 6-carboxyfluorescein ( 6FAM ) -5′ AGGCTTCCCTCGCTGCCGTTATG 3′-6 carboxytetramethylrhodamine ( TAMRA ) ( Life Technologies ) . BEBOV RNA was detected using the CFX96 detection system ( BioRad Laboratories , Hercules , CA ) in One-step probe qRT-PCR kits ( Qiagen ) with the following cycle conditions: 50°C for 10 minutes ( min ) , 95°C for 10 seconds ( s ) , and 40 cycles of 95°C for 10 s and 59°C for 30 s . Threshold cycle ( CT ) values representing BEBOV genomes were analyzed with CFX Manager Software , and data are shown as + or − for genome equivalents ( GEq ) above or below 3 . 0 log10 respectively . To create the GEq standard , RNA from BEBOV stocks was extracted and the number of BEBOV genomes was calculated using Avogadro's number and the molecular weight of the BEBOV genome . Virus titration was performed by plaque assay with Vero E6 cells from all serum samples . Briefly , increasing 10-fold dilutions of the samples were adsorbed to Vero E6 monolayers in duplicate wells ( 200 µl ) ; the limit of detection was 25 PFU/ml . Serum collected at indicated time points ( Fig . 1C , vertical arrows ) was tested for cross-reactive immunoglobulin G ( IgG ) antibodies against SEBOV , ZEBOV , and BEBOV . Enzyme-linked immunosorbent assay ( ELISA ) using purified virus-like particles ( VLPs ) containing VP40 and GP antigen for the appropriate filovirus , was used to detect cross-reactive IgG . VLPs were produced as previously described [32] , with the exception of using baby hamster kidney ( BHK ) cells to produce the particles . Species specific VLPs were detergent lysed in 0 . 01% Triton-X 100-PBS and 1 µg of protein was used to coat the 96 well ELISA plates ( Nunc ) . The serum samples were assayed at 4-fold dilutions starting at a 1∶100 dilution in ELISA diluents ( 1% heat inactivated fetal bovine serum ( HI-FBS ) , 1×PBS , and 0 . 2% Tween-20 ) . Samples were incubated for 1 hour at room temperature , removed , and plates were washed . Wells were then incubated at room temperature for 1 hour with anti-monkey IgG conjugated to horseradish peroxidase ( Fitzgerald Industries International ) at a 1∶2500 dilution . These wells were washed and then incubated with 2 , 2′-azine-di ( 3ethylbenzthiazoline-6-sulfonate ) peroxidase substrate system ( KPL ) and read for dilution endpoints at 405 nm on a microplate reader ( Molecular Devices Emax system ) . Statistics were calculated for ELISA IgG titers utilizing GraphPad Prism 5 software by using a 2way ANOVA analysis comparing treatments and times between all groups . Neutralizing antibody titers were determined by performing plaque reduction neutralization titration assays ( PRNT ) . Briefly , Vero cells were seeded into 6 well plates to generate a confluent monolayer on the day of infection . Serum dilutions were prepared in DMEM and 100 µL were incubated with ∼100 pfu of rVSV-BEBOV-GP in a total volume of 200 µL . Media was removed from cells , the serum–virus mixture was added and samples were incubated for 60 min at 37°C . The mixture was removed from the cells and 2 ml of 0 . 9% agaraose EMEM ( 5% FBS v/v ) was overlayed on wells . Cells were observed 72 hours post-incubation and plaques were counted . The neutralizing antibody titer of a serum sample was considered positive at a dilution showing a ≥50% reduction ( PRNT50 ) compared with the virus control without serum .
To evaluate whether a homologous monovalent vaccine could protect against BEBOV and whether or not we could achieve cross-protection against BEBOV with heterologous vaccines available at the time of the original BEBOV outbreak [3] , we used the cynomolgus macaque BEBOV NHP model [23] , [25] . In this study , we used four separate vaccination groups of NHPs as shown in Figure 1B . Group 1 ( PBS ) was a negative control group , Group 2 ( BEBOV ) was an internal control group vaccinated with a rVSV-BEBOV-GP vaccine , Group 3 ( Blend ) was vaccinated with an equal blend of rVSV-SEBOV-GP and rVSV-ZEBOV-GP in a single inoculation , and Group 4 ( Boost ) was vaccinated with rVSV-SEBOV-GP first and 14 days later vaccinated with rVSV-ZEBOV-GP . To determine the humoral immune response to the different vaccination strategies ( Fig . 1B and C ) , we tested the pre-challenge serum of the animals for immunoglobulin G ( IgG ) antibody titers that were cross-reactive for SEBOV GP ( Fig . 2A ) , ZEBOV GP ( Fig . 2B ) , or BEBOV GP ( Fig . 2C ) by enzyme-linked immunosorbent assay ( ELISA ) . Mean reciprocal titers of IgG antibodies were calculated and are shown in Figure 2 . As expected , we observed no antibody titers for Group 1 when tested against all three EBOV GPs ( Fig . 2A , B , and C ) whereas we detected only BEBOV GP cross-reactive IgG for Group 2 at day −8 pre-challenge and on the day of challenge ( Fig . 2C , Day −8 and 0 ) . Groups 3 and 4 were vaccinated with the same vaccine vectors but had different vaccination regimens ( Fig . 1B and C ) . While the day 0 IgG titers for SEBOV GP ( Fig . 2A , Day −8 and 0 ) and ZEBOV GP ( Fig . 2B , Day −8 and 0 ) were similar between the groups there was a higher cross-reactive IgG titer for BEBOV GP in the cohort from Group 4 ( Fig . 2C , Day −8 and 0 ) . In addition , although BEBOV GP cross-reactive IgG titers were not as high as those elicited in Group 2 , the vaccination regimen for Group 4 did elicit IgG antibodies which could recognize BEBOV GP . To date , studies have shown that the mortality rate for the cynomolgus macaque model after BEBOV challenge is between 66% and 75% [23] , [25] , whereas the SEBOV and ZEBOV models are 100% lethal [16] . To test whether we could induce cross-protection against BEBOV challenge after vaccinating with heterologous rVSV vaccines expressing SEBOV and ZEBOV GPs , we challenged all four groups of NHPs with a 1 , 000 pfu dose of BEBOV . The animals were closely monitored over the course of 28 days post-challenge for clinical signs of illness . Groups 2 and 4 were 100% protected against BEBOV ( Fig . 3A ) , while Groups 1 and 3 each had two of the three animals succumb to BEBOV infection ( Fig . 3A ) . For Group 1 , animal 6936CQ succumbed on day 11 and animal 6942CQ succumbed on day 10 post-challenge ( Table 1 ) . In Group 3 , animal 98C007 succumbed on day 10 and animal 98C017 from this group expired on day 14 post-challenge . Clinical scores were recorded each day post-challenge for each animal using a scoring system based on dyspnea , depression , recumbency , and rash . The clinical scores for each animal associated with the survival data as seen with animal 98C020 from Group 1 and animal 91670 from Group 3 each scoring lower than the non-surviving animals in their cohort ( Fig . 3B ) . The signs of disease in response to BEBOV infection were more dramatic for the animals in Groups 1 and 3 when compared to the animals in the other two groups ( Table 1 ) . This observation correlates well with the fact that infectious virus was only isolated from the serum of all the animals in these groups after challenge ( Table 1 ) . Though reduced when compared to Groups 1 and 3 , Group 4 had one animal ( Table 1 , 98C027 ) with very mild signs of disease and two of the animals ( Table 1 , 07411 and 98C027 ) were positive for viral genomes in serum as detected by qRT-PCR , whereas Group 2 had no signs of disease nor were the serum samples positive for viral genomes by qRT-PCR ( Table 1 ) . To further address the humoral response to rVSV-filovirus-GP vaccination and BEBOV challenge , we assessed sera for neutralizing activity against BEBOV GP . Neutralizing antibody titers were not detected in any animal before vaccination ( Table 2 , Pre-vaccination ) . None of the 3 animals in Group 4 showed any evidence of neutralizing antibodies after the prime vaccination ( Table 2 , Day −22 ) . By the day of BEBOV challenge there were five animals that had modest neutralizing antibody titers ( PRNT50 of 1∶40 to 1∶80 ) including all three animals from Group 2 and two of three animals from Group 4 ( Table 2 ) . Neutralizing antibody titers were also assessed for all animals at the study endpoint ( day of death for animals that expired or day 28 for surviving animals ) . All animals that survived BEBOV challenge had PRNT50 titers ranging from 1∶40 to 1∶160 against BEBOV GP while with the exception of one control animal all macaques that succumbed had PRNT50 values below 1∶40 ( Table 2 ) .
The emergence of BEBOV in 2007/08 and the recent outbreak in the summer of 2012 [3] , [6] are events which underscore the lack of effective vaccines for combating new species of EBOV during outbreaks . While there are well characterized vaccines against ZEBOV and SEBOV [9] , [10] , [11] , [12] , [13] , [14] , [15] , [16] , [17] , [18] , [19] , [20] , [21] , [22] , [23] , [24] , [25] , [26] , [27] , [28] , [29] , until this report , there were no vaccines directed specifically against BEBOV . Here , we have shown that the rVSV-BEBOV-GP vaccine can perform just as well against homologous challenge with BEBOV as seen in previous studies with the rVSV vaccines against homologous challenge with ZEBOV or SEBOV [12] , [33] . In this study the animals in Group 2 lacked clinical signs of disease , were below the limits of detection for viremia by plaque assay and qRT-PCR ( Table 1 ) , and each animal in the cohort had a score of “0” on the clinical score scale ( Fig . 3B ) for the duration of the study; all results were similar to vaccination against homologous challenge with ZEBOV or SEBOV after vaccination with the appropriate rVSV-filovirus-GP vector . As with previous studies [23] , [25] , we were interested in whether or not we could protect against a newly emerging EBOV species with vaccine vectors that were available at the time of an outbreak . Therefore , we used the BEBOV model and the rVSV-SEBOV-GP and rVSV-ZEBOV-GP vaccines available during the initial BEBOV outbreak . In an emergency intervention scenario , a vaccination schedule requiring a number of boosts over a long period of time is not practical . A rapid , single vaccination strategy is desired in this scenario . However , available data suggest that heterologous rVSV-filovirus-GP vaccines cannot always protect against challenge with a different species of EBOV [12] , [25] , [34] nor can a single , blended vaccination using multiple rVSV-filovirus-GP protect as seen in Group 3 in the current study ( Fig . 3A , Table 1 ) . The ability to cross-protect against BEBOV challenge using DNA/rAd-based SEBOV and ZEBOV GP vaccines [23] , [25] encouraged us to use the rVSV-SEBOV-GP and rVSV-ZEBOV-GP as a vaccine strategy for cross-protection against BEBOV challenge . Data from a blended vaccine study using rVSV-SEBOV-GP , rVSV-ZEBOV-GP , and rVSV-MARV-GP suggest that there is potential for vector interference/competition between the EBOV vaccines in particular in regard to the effectiveness of the SEBOV vaccine in the blend [16] . Based on these data , we tested the blended approach ( Group 3 ) but also employed a prime-boost strategy ( Group 4 ) which allowed the rVSV-SEBOV-GP vaccine to induce an immune response to SEBOV-GP 14 days before vaccination with rVSV-ZEBOV-GP and subsequent BEBOV challenge 21 days after vaccination with the ZEBOV vaccine . We hypothesized that this prime-boost strategy with the rVSV platform would provide similar cross-protection when compared with the DNA and rAd-based approach [23] , but with fewer doses ( 2 versus 5 ) and a much shorter vaccine regimen ( 36 days versus 518 days ) . Indeed , we were able to induce 100% protection against heterologous BEBOV challenge with the Group 4 vaccination regimen ( Fig . 3A , Table 1 ) . While 100% and 75% protection was achieved between our study and the previous BEBOV cross-protection studies [23] , [25] , the immunity without any detectable viremia generally seen against homologous challenge [12] , [15] , as noted in Group 2 , was not achieved by the Group 4 vaccine regimen . This observation was also seen in the previous BEBOV studies [23] , [25] with detectable viremia by qRT-PCR and very mild clinical signs of disease reported ( Table 1 ) . The difference between Group 4 and Groups 1 and 3 is clear with animals in each of the latter groups succumbing to infection ( Fig . 3A , Table 1 ) , showing more severe clinical signs of disease , higher levels of viral RNA , and detectable circulating infectious virus ( Table 1 ) . It is interesting when comparing the differences between Groups 3 and 4 ( where the only deviation was in the regimen used ) that animals in Group 4 were protected from severe disease while the animals in Group 3 experienced similar signs of severe disease as the control animals in Group 1 ( Table 1 ) . While the circulating level of BEBOV cross-reactive GP IgG from Group 4 was not as high as from the homologous vaccine animals ( Group 2 ) ( Fig . 2C , p<0 . 01 at Day −8 ) the circulating cross-reactive BEBOV GP IgG from Group 4 was higher than the level from Group 3 ( Fig . 2C , p<0 . 01 at Day −8 ) . In contrast to the blended strategy , the prime-boost regimen was able to generate a greater cross-protective immunity which was associated with the higher cross-reactive BEBOV GP IgG . This is different from the DNA/rAd vaccine strategy where there were no cross-reactive BEBOV GP IgGs detected , although there was a cellular immune response by CD4+ and CD8+ T-cells [23] . This observation may not be too surprising as the DNA/rAd vaccines have been shown to elicit robust cellular immunity [23] , [35] while it has recently been demonstrated that antibodies correlate with protection against ZEBOV infection using the rVSV-based vaccine platform [28] . In fact , single immunization with either rVSV-ZEBOV-GP or rVSV-CIEBOV was able to generate some cross-reactive BEBOV-GP IgG , although higher for the CIEBOV vaccine group . However , these antibody responses did not correlate with 100% protection [25] . While it was reported that antibodies are necessary for protection using the rVSV-ZEBOV-GP vaccine [28] , the neutralizing antibody titers are not very robust when compared to responses induced by vaccines against other highly pathogenic viruses such as Nipah virus [36] . However , this previous work shows that even low to modest levels of neutralizing antibodies appear to be important for protection of NHPs against ZEBOV as a productive immune response was evidenced by increased titers after virus challenge [28] . While ZEBOV is uniformly lethal in cynomolgus macaques , the mortality rate for BEBOV in cynomolgus monkeys is 66 to 75% with a prolonged time to death compared to ZEBOV [23] , [25] . This difference in disease pathogenesis confounds a definitive conclusion in the current study when using the development of neutralizing antibodies to determine a productive immune response to the different vaccine regimens between Groups 3 and 4 . This is reflected in Table 2 where we observed an increase in neutralizing antibody titer post-challenge for the non-vaccinated Group 1 survivor ( 98C020 ) , the rVSV blend Group 3 survivor ( 91670 ) , and for the animal that succumbed in Group 3 ( 98C017 ) . The Group 3 animal ( 987C017 ) succumbed on Day 14 which may account for the presence of antibodies in this case as the disease course was further delayed allowing for an antibody response , although not sufficient enough for protection . On the surface , the results in Table 2 appear to be the same for animal 98C017 from Group 3 and animal C07426 from Group 4 when looking at the antibody titers , although one animal succumbed and the other animal survived BEBOV challenge . While comparing these two animals is difficult , we believe the Day 0 titers shed light on the differences between responses to the vaccine regimens as there were no neutralizing BEBOV GP antibody titers for any animal in Group 3 at the day of BEBOV challenge while two of three Group 3 animals had neutralizing BEBOV GP antibody titers at the day of challenge ( Table 2 , Day 0 ) . All antibody data taken together , it is clear that a prime vaccination with rVSV-SEBOV-GP and subsequent boost vaccination with rVSV-ZEBOV-GP produces higher levels of cross-reactive BEBOV GP IgG than the blended vaccination approach ( Fig . 2C , Group 3 versus 4 ) and a pre-challenge neutralizing BEBOV GP antibody titer ( Table 2 , Day 0 ) . It is also evident that these anti-BEBOV binding and neutralizing antibodies are associated with protection in a prime-boost regimen versus a single immunization with a single heterologous rVSV-EBOV-GP vector [25] . Comparison between our study and the previous two BEBOV cross-protection studies are similar on the surface as each has shown some measure of cross-protection with animals displaying some mild signs of illness and low level viremia by qRT-PCR . However , there are differences in the challenge doses used in each study . Specifically , our study used a 1 , 000 PFU challenge dose , the DNA/rAd study used 1 , 000 TCID50 [23] , and the rVSV-filovirus-GP single immunization study used 10 , 000 TCID50 [25] . Here , we measured levels of infectious virus and viral RNA where only viral RNA was assessed to determine viremia in the previous studies [23] , [25] . This makes it difficult to compare infectious viremia among the studies . However , we can compare the viremia reported by detection of BEBOV genomes among the studies . In the previous studies , the viremia detected in macaques that succumbed to BEBOV infection by qRT-PCR did not reach 6 log10 genome equivalents [23] , [25] , while all animals in Group 1 and Group 3 of the current study exceeded 6 log10 during the course of disease ( Table 1 ) . Based on these data it appears that our study may have had a higher infectious challenge dose and yet we were still able to provide cross-protection against BEBOV challenge with the prime-boost strategy using vaccines that were available at the time of the original BEBOV outbreak [3] . In this study we have shown that a rVSV-BEBOV-GP vector can protect against homologous challenge with the newest species of EBOV , BEBOV , and that a prime-boost strategy with the rVSV-SEBOV-GP and rVSV-ZEBOV-GP vectors , available at the time BEBOV emerged , is capable of providing cross-protection against BEBOV challenge . We propose that this condensed , prime-boost vaccine regimen of available heterologous rVSV-filovirus-GP vaccines should be considered as a paradigm for controlling newly emerging EBOV species . EBOV has recently re-emerged with a potential new species as filovirus-like RNA was isolated from dead bats in Spain [37] . In addition , an outbreak of REBOV recently occurred in pigs in the Philippines [38] , [39] , [40] and further studies have shown that ZEBOV can infect pigs [41] , [42] . In light of these observations and the increasing number of filovirus outbreaks over the past decade , it would be prudent to have a strategy in place which could be used to immediately respond to an outbreak of a new EBOV species while a new homologous rVSV vaccine vector was being developed and produced . The typical period of time between vaccination and challenge for these vaccines in NHPs is 28 days although recently it has been reduced to 21 days ( TWG , unpublished data ) . From this study we could potentially have the population surrounding an epicenter of a newly emerged filovirus protected within 35 days . In addition , we cannot rule out some post-exposure utility for the population , as these vectors have shown post-exposure potential [33] , [43] . While we used single antigenic rVSV-filovirus vaccines in this study , there are rVSV-filovirus vectors that can express multiple filovirus antigens from different EBOV species which have shown improved cross-protective efficacy in guinea pigs [44] . Ideally , a single vaccination capable of immunizing against all EBOV species known to cause human disease would be preferred for a quick response to an outbreak of EBOV; however , at the moment it appears that a prime-boost strategy may be the best approach for broad coverage . Perhaps an approach using the prime-boost strategy used here with single antigenic vectors plus the multiple antigenic vectors would enhance cross-protection to a point where immunity with a lack of mild clinical signs and detectable low level viremia could be achieved . | Ebola viruses ( EBOV ) , of which there are five species , are categorized as Category A Priority Pathogens and Tier 1 Select Agents by several US Government agencies as a result of their high mortality rates and potential for use as agents of bioterrorism . Currently , there are no vaccines or therapeutics approved for human use . Replication-competent , recombinant vesicular stomatitis virus ( rVSV ) vectors expressing filovirus glycoproteins ( GP ) , in place of the VSV glycoprotein have shown promise in lethal nonhuman primate ( NHP ) models of filovirus infection as both single injection preventive vaccines and as post-exposure treatments . The recent outbreak of the fifth recognized EBOV species , Bundibugyo ebolavirus ( BEBOV ) , demonstrates the need for vaccines that can be rapidly deployed to combat an outbreak of a new filovirus species . To date , rVSV-filovirus GP-based vaccines have only been able to protect against challenge with a homologous species of EBOV . Here , we show that the two heterologous rVSV-based filovirus vaccines available at the time of the original BEBOV outbreak can protect NHPs against BEBOV challenge using a short prime-boost vaccination strategy . While the prime-boost strategy was successful , a single injection blended vaccination strategy with the same vaccine vectors failed to provide protection . These data suggest that an abbreviated prime-boost regimen of 36 days may have utility for quickly responding to outbreaks caused by new species of EBOV . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [] | 2013 | Vesicular Stomatitis Virus-Based Vaccines Protect Nonhuman Primates against Bundibugyo ebolavirus |
Blastocystis is the most prevalent eukaryotic microbe colonizing the human gut , infecting approximately 1 billion individuals worldwide . Although Blastocystis has been linked to intestinal disorders , its pathogenicity remains controversial because most carriers are asymptomatic . Here , the genome sequence of Blastocystis subtype ( ST ) 1 is presented and compared to previously published sequences for ST4 and ST7 . Despite a conserved core of genes , there is unexpected diversity between these STs in terms of their genome sizes , guanine-cytosine ( GC ) content , intron numbers , and gene content . ST1 has 6 , 544 protein-coding genes , which is several hundred more than reported for ST4 and ST7 . The percentage of proteins unique to each ST ranges from 6 . 2% to 20 . 5% , greatly exceeding the differences observed within parasite genera . Orthologous proteins also display extreme divergence in amino acid sequence identity between STs ( i . e . , 59%–61% median identity ) , on par with observations of the most distantly related species pairs of parasite genera . The STs also display substantial variation in gene family distributions and sizes , especially for protein kinase and protease gene families , which could reflect differences in virulence . It remains to be seen to what extent these inter-ST differences persist at the intra-ST level . A full 26% of genes in ST1 have stop codons that are created on the mRNA level by a novel polyadenylation mechanism found only in Blastocystis . Reconstructions of pathways and organellar systems revealed that ST1 has a relatively complete membrane-trafficking system and a near-complete meiotic toolkit , possibly indicating a sexual cycle . Unlike some intestinal protistan parasites , Blastocystis ST1 has near-complete de novo pyrimidine , purine , and thiamine biosynthesis pathways and is unique amongst studied stramenopiles in being able to metabolize α-glucans rather than β-glucans . It lacks all genes encoding heme-containing cytochrome P450 proteins . Predictions of the mitochondrion-related organelle ( MRO ) proteome reveal an expanded repertoire of functions , including lipid , cofactor , and vitamin biosynthesis , as well as proteins that may be involved in regulating mitochondrial morphology and MRO/endoplasmic reticulum ( ER ) interactions . In sharp contrast , genes for peroxisome-associated functions are absent , suggesting Blastocystis STs lack this organelle . Overall , this study provides an important window into the biology of Blastocystis , showcasing significant differences between STs that can guide future experimental investigations into differences in their virulence and clarifying the roles of these organisms in gut health and disease .
Blastocystis is a genus of atypical , nonflagellated , anaerobic stramenopiles commonly inhabiting the intestinal tract of humans and other animals . The majority of stramenopiles , which , along with the Alveolata and Rhizaria , are members of the eukaryotic supergroup known as SAR , are marine biflagellated cells with tubular hairs on their surface—Blastocystis has none of these characteristics . The number of humans infected with Blastocystis globally has been estimated at over 1 billion [1] , with prevalence being higher in developing than in developed countries [2] . The infective stage of Blastocystis is an environmentally resistant cyst , with the most common mode of transmission being the fecal-oral route . The pathogenicity of Blastocystis is controversial , but colonization with this organism has been linked to gastrointestinal symptoms , including diarrhea , abdominal pain , nausea , and irritable bowel syndrome [3] . Blastocystis is capable of becoming established in the gut and is difficult to eradicate via pharmacological interventions [3] . However , a causal link between the presence of the organism and disease symptoms has not been established [1 , 3] , and some authors argue that Blastocystis is a part of a healthy gut microbiota [4] . At the genetic level , Blastocystis is remarkably heterogeneous . Many morphologically similar but genetically distinct lineages of Blastocystis have been identified , based primarily on sequences of their small subunit ( SSU ) ribosomal RNA genes [5] . Seventeen lineages have been isolated from mammals and birds to date and are referred to as “subtypes” ( STs ) ; the inter-ST divergence of the SSU rRNA gene is at least 3% . Blastocystis STs can have a remarkably broad host range and are almost never found exclusively in 1 host [5 , 6] . Only STs 1–9 have been found in humans to date , with STs 1–4 being responsible for around 90% of all human cases examined [6] . The high degree of genetic diversity is a confounding factor in establishing whether Blastocystis is a pathogen . Recent in vitro and in vivo molecular investigations have identified hydrolases and proteases as candidate virulence factors . Blastocystis proteases cause cleavage and degradation of immunoglobulin A ( IgA ) secreted by the host , may disrupt the intestinal epithelial barrier , and increase production of pro-inflammatory cytokines [2 , 7 , 8] . However , there is substantial inter-ST variation in adhesion to enterocytes , disruption of intestinal epithelial tight junctions , activation of pro-inflammatory cytokines , and the ability to scavenge nitric oxide . Virulence factor variability presumably extends to the intra-ST level [9] because the same ST is commonly found in both symptomatic and asymptomatic hosts . The highly variable clinical presentations attributed to Blastocystis could potentially be due to colonization with different STs or strains of the organism and untangling the relationships remains an area of active research and concern [10–12] . Recently , researchers have begun to assess the relationship between Blastocystis colonization and the composition of the prokaryotic gut microbiota . While several studies indicate Blastocystis may be associated with a more diverse and “healthy” microbiota [13 , 14] , others have reported an association between Blastocystis colonization and a decrease in protective bacteria in the gut [15] or no differences in microbiota between Blastocystis-positive and -negative patients [16] . Despite the unanswered questions regarding its potential clinical relevance , studies of Blastocystis genomes are still in their infancy . Currently , several mitochondrial genomes ( STs 1–4 and 6–9 [17–20] ) and 2 high-quality draft nuclear genomes ( STs 4 and 7 [21 , 22] ) are available , as are genome survey data for additional STs ( STs 2 , 3 , 6 , 8 , and 9 [13] ) . The ST7 Blastocystis nuclear genome , obtained using a culture established from fecal matter from a symptomatic human , was the first to be sequenced [21] . The study reported a genome size of 18 . 8 Mb with 6 , 020 protein-coding genes . The authors described , among other things , a complex "mitochondria-like" organelle , effector proteins possibly involved in adaptation to a parasitic lifestyle , and a suite of secretory proteins that could have the potential to alter host physiology . They also detailed the genome architecture of ST7 , finding it to be highly compact and having numerous duplicated blocks of genes . More recently , an ST4 genome obtained from a laboratory rodent in Singapore was published [22]; its size was 12 . 91 Mb and a set of 5 , 713 protein-coding genes was predicted . There were no detailed analyses of the genome structure or of the genes themselves . In 2015 , draft assemblies for the nuclear genomes of STs 2 , 3 , 4 , 6 , 8 , and 9 were released to public databases . None of these assemblies have been annotated with predicted genes and no analyses have been reported to date beyond their use in a microbiome analysis [13] . Here , we report a draft genome sequence and transcriptome analysis of Blastocystis ST1 , NandII strain from a symptomatic human , and compare it to the published genomes of Blastocystis ST7 [21] and ST4 [22] . We first provide a general overview and statistics for the NandII genome and perform high-level comparisons between the 3 genomes . We then present findings derived from extensive manual annotation , with a focus on genes that are potential host effectors , and highlight significant genomic differences between the STs . Our study provides a strong framework for subsequent molecular and cellular investigations of the role of Blastocystis in gastrointestinal health and disease and of its impact on the microbial communities of the gut .
The assembled size of the Blastocystis ST1 genome is 16 . 5 Mb spread across 580 scaffolds ( Table 1 ) . In comparison , the assembly of ST7 is composed of 54 scaffolds and 18 . 8 Mb , while the ST4 assembly is 12 . 9 Mb and has 1 , 301 scaffolds . The substantially lower number of scaffolds in the ST7 assembly reflects the longer reads of the Sanger sequencing technology used versus the shorter reads generated in next-generation sequencing ( 454 and Illumina ) used in the cases of ST1 and ST4 . Although the assembly for ST7 is substantially larger with longer scaffolds , the fold coverage , gene complement , and transcriptome coverage suggest that the difference in size between the strains is real . The substantial genomic differences between ST1 , ST4 , and ST7 are not limited to size . The GC content varies by 15% between the 3 STs ( 39 . 6% in ST4 , 45 . 2% in ST7 , and 54 . 6% in ST1 ) , which is significantly different when compared to GC-content variation in some parasites: the GC content difference among 3 strains of Giardia is only 2% [23] , as is the difference among 4 species of Leishmania [24] . More biologically relevant is the variation among the Blastocystis STs in terms of overall gene numbers and gene structure . ST1 has 524 more protein-coding genes than were identified in ST7 and 831 more than in ST4 ( Table 1 ) . ST1 has the largest average gene size , the highest percentage of genes with introns , the highest average number of exons per gene , and the largest percentage of its genome devoted to encoding proteins , whereas ST7 has the lowest figures for these same attributes ( Table 1 ) . Although ST7 was characterized as being a compact genome [21] , ST1 is demonstrably more compact , with an average intergenic spacer size of 615 bps versus 1 , 687 bps in ST7 . ST4 tends to fall in the middle for these various measures , although it has more genes/kb , 0 . 44 , than ST1 ( 0 . 39 ) or ST7 ( 0 . 32 ) because it has the smallest genome size . While these differences in the number and organization of genes in different STs likely reflect real differences between their genomes , they may to some extent be impacted by the procedures used to predict gene models in the 3 genome assemblies . Particularly noteworthy is that ST7 and ST4 were annotated before the recognition of the phenomenon of polyadenylation-mediated creation of stop codons that occurs in Blastocystis ( see below ) . Hence , the statistics concerning the different genomic elements in ST7 and ST4 ( e . g . , genes , exons , introns , intergenic spaces ) reported in this and subsequent sections must be considered current estimates that may change after revised annotation . A striking structural feature of Blastocystis STs that is unique amongst eukaryotic nuclear genomes is the polyadenylation-mediated generation of termination codons [25] . This process involves the creation of a functional stop codon by 3′-polyadenylation of mRNAs in a subset of genes , whereby the first 1 or 2 adenines of the poly-A tail appended to terminal uracil-adenine/uracil-guanine or uracil respectively , complete a termination codon missing in the actual gene sequence . Importantly , the addition of the poly-A tail occurs upstream of any possible canonical stop site in the underlying genomic sequence in these genes . Because most gene-finding algorithms are predicated on the presence of standard stop codons , this unique feature in Blastocystis can lead to erroneous gene models with problems such as overly long protein sequences , overlapping genes , chimeric genes , and the introduction of false introns to reach the next downstream stop codon . The phenomenon was first investigated in ST7 using expressed sequence tag ( EST ) data [25] . The authors concluded that potentially 15% of ST7 genes have stop codons generated by polyadenylation . They also found some examples of the same process in a preliminary draft genome of ST1 . With the full genome of ST1 available , as well as RNA-sequencing ( RNA-Seq ) data , we investigated the extent of the phenomenon . Appropriate stop codons generated by polyadenylation were found in 1 , 693 protein-coding genes , representing 26% of the protein-coding genes in ST1 . The 15% suggested for ST7 may be an underestimate because mapped RNA-Seq data are far superior to ESTs for confirming whether the stop codon is generated via polyadenylation . Because this phenomenon was not identified prior to the gene-finding process for the Blastocystis ST7 genome , a number of the initial gene models were incorrect [25] . At present , it is unclear whether the gene models for ST4 suffer from similar problems or even whether the same polyadenylation-mediated generation of termination codons occurs . The sites for polyadenylation appear to be linked to a highly conserved motif ( TGTTTGTT ) usually found 5 bases downstream of the nucleotide preceding the poly-A tail . A search for this motif indicates that it is abundant ( Table 2 ) in Blastocystis . All 3 genomes have roughly the same number of motifs when genome size is taken into account . ST1 has 1 site per 3 , 097 bps , ST4 has 1 per 2 , 751 bps , and ST7 has 1 per 2 , 782 bps . A random selection of other stramenopile genomes showed that Thalassiosira pseudonana has 1 motif per 17 , 754 bps , while Phytophthora infestans has 1 site per 37 , 441 bps , suggesting that in these 2 genomes , the presence of the motif is random . Because ST4 has a very similar motif complement to ST1 and ST7 , it is likely that it too uses the polyadenylation process to generate some of its termination codons . Also suggestive are the 95 gene model pairs that overlap in ST4 ( Table 1 ) . An examination of these overlapping coding regions found motifs at locations appropriate to allow separation of the genes in most of the 95 cases . For example , the 3′-ends of KNB46045 and KNB46046 overlap by 37 bps ( S1 Fig ) . Forty-four bps upstream of the annotated stop codon for KNB46045 is the conserved motif TGTTTGTT , which , if used to direct the generation of a new termination codon , would eliminate the overlap between the 2 genes . Future studies of additional Blastocystis STs need to take polyadenylation-mediated stop codons into account . The mechanism has been demonstrated to be active in ST1 and ST7 , suggesting that the mechanism evolved prior to the divergence of these 2 STs . The ST1 genome is substantially more intron rich than that of ST7 or ST4 . It has 35 , 412 predicted introns compared with 18 , 200 in ST7 and 24 , 093 in ST4 ( Table 1 ) . Consequently , ST1 has the highest percentage of genes with introns and the highest number of introns per gene . The size of introns in ST1 is also less variable ( Fig 1 ) , with over half being 30 bps in length . In ST4 and ST7 , the percentage of introns with a length of 30 bps is considerably lower . ST7 shows greater variation in intron size , with a much lower peak at 30 bps and higher percentages of introns in the size range 31–38 bps . ST1's relative intron richness may in part reflect methodological differences in gene calling . Exon/intron boundaries in ST1 were corrected using RNA-Seq transcriptome data , which were not available for ST4 and ST7 . Depending on the parameters used to identify a "typical" gene and its exon/intron structure during the automated gene-calling process , having RNA-Seq data would tend to result in finding more real introns and annotating more introns with the correct boundaries . Of the approximately 35 , 000 annotated introns in the Blastocystis ST1 nuclear genome , the vast majority ( 98 . 3% ) represent the standard ( or U2 type ) spliceosomal introns characterized by GT-AG boundaries and spliced by the so-called major ( U2 ) spliceosome containing U1 , U2 , U4 , U5 , and U6 small nuclear RNAs ( snRNAs ) containing small nuclear ribonucleo proteins . Two additional , much less abundant intron categories are also apparently spliced by the major spliceosome . One has GC instead of GT as the 5′ intron boundary and constitutes about 0 . 5% of all introns in both ST1 and ST7 ( Fig 1 ) , which is on par with the proportion of this intron found in metazoan or plant genomes [26] . The second category , characterized by the GA dinucleotide at the 5′ border , is even more sparse , with 59 such introns ( i . e . , only around 0 . 16% of all introns ) identified in ST1 and only 9 supported by EST data in ST7 . The existence of such introns is not without precedent , as such GA-AG introns were previously identified in the genome of the dinoflagellate Symbiodinium minutum [27] . The final intron category in Blastocystis corresponds to the minor , or U12-type , introns , characterized by boundaries typically exhibiting the dinucleotides AT-AC ( Fig 1 ) and spliced by the minor ( U12 ) spliceosome containing U11 , U12 , U4atac , and U6atac snRNA molecules . A systematic analysis of the Blastocystis ST1 genome revealed 346 U12-type introns in 319 genes . In addition , genes for all 4 snRNAs and for all 7 protein subunits ( 20K , 25K , 31K , 35K , 48K , 59K , and 65K ) specific to the U12 spliceosome were identified [28] ( S1 Table ) . U12-type introns and U12 spliceosome-specific components were not noted in the report on the ST7 genome [21] , but both can be identified in the genome sequence of that ST ( Fig 1 , S1 Table ) . Hence , Blastocystis represents a previously missed eukaryotic lineage that has retained U12-type introns and the associated splicing machinery . In stramenopiles , only oomycetes have so far been known to exhibit both types of spliceosomal introns , whereas only the major “standard” type has been retained in the other lineages [29] . Blastocystis ST1 was found to have 6 , 544 protein-coding genes , in contrast to 6 , 020 in ST7 and 5 , 713 in ST4 ( Table 1 ) . An examination of the amino acid identities between putative orthologs reveals an extraordinary degree of dissimilarity among the Blastocystis STs ( Fig 2 ) . The median sequence identity of aligned regions of orthologs among STs 1 , 4 , and 7 ranges from 59% to 61% ( S2 Table ) . The differences among the 3 STs exceed those observed for orthologs from pairs of species from parasitic protistan genera such as Cryptosporidium ( C . parvum–C . hominis , Alveolata ) , Leishmania ( L . major–L . infantum , Excavata ) , and Theileria ( T . parva–T . annulata , Alveolata ) ( Fig 2 ) and indeed among Giardia strains ( WB , GS , P15; Excavata ) ( Fig 2 , S2 Table ) . The dissimilarity is comparable to that between species of Plasmodium ( P . falciparum–P . knowlesi , Alveolata ) and Trypanosoma ( T . cruzi–T . brucei , Excavata ) . The extent of the dissimilarity among the Blastocystis STs supports the contention that they should be considered at least equivalent to separate species , particularly when placed in the context of other protistan pathogens . The lack of morphologically distinguishing traits and low correlation between ST and host [30] will continue , however , to make the taxonomy of Blastocystis challenging . ST1 and ST4 genes are the least alike at the protein level while ST7 protein-coding sequences are usually more similar to ST1 than to ST4 genes ( Fig 2 ) . This is in line with recent phylogenetic analyses [5 , 20 , 31] that place ST4 in a clade sister to one that includes ST1 and ST7 . There are also marked differences in gene content between STs . The percentage of genes present in ST1 but not in ST4 is significantly lower than the percentage of genes in ST1 not present in ST7 ( Fig 3 ) . Similarly , ST4 has far fewer unique genes when compared with ST1 than with ST7 . The percentage of unique genes in ST7 is virtually identical in comparisons with ST1 and ST4 . The differences in gene complement between the Blastocystis STs greatly exceed the differences between selected parasitic protistan species pairs ( Fig 3 ) . Beyond a common core of genes presumably devoted to common housekeeping tasks , the Blastocystis STs possess a substantial number of genes unique to each ST . The consequences of these major genetic differences between the STs are discussed below . Kinase enzymes modify target proteins via phosphorylation and participate in the regulation of cellular pathways , particularly those involved in signal transduction [32] . Our analysis of Blastocystis ST1 identified 221 kinases , which were classified according to kinase . com [33] . Representatives of most kinase groups are similarly distributed among STs 1 , 4 , and 7 ( Fig 4 ) except for 2 clades showing lineage-specific gene expansion . The first is a small clade of calcium/calmodulin-dependent-like ( CAMKL ) kinases almost exclusive to STs 4 and 7 ( Fig 4 ) . The second and more striking example of ST differences is a clade of STE20/7 kinases . While ST4 and ST7 both encode representatives of the STE family , they appear to completely lack members of a cluster of 58 closely related STE20/7 kinases found only in the ST1 genome . This seems to be a spectacular case of gene family expansion specific to ST1 . The exact roles of the various kinases in Blastocystis STs and , in particular , the large group of STE20/7 kinases exclusive to ST1 are currently unclear . The STE20/7 family includes members of the mitogen-activated protein kinase ( MAPK ) signaling pathway , which regulates responses to extracellular stimuli [34] . The ST1 cluster does not exhibit a particularly close relationship with any specific gene of the STE20/7 family . In Giardia , 2 members of the MAPK family have been implicated in initiation of encystation [35] , while in T . brucei , a MAPK is involved in mediating its interferon-γ-induced proliferation in the host [36] . MAPK pathways have been characterized , at least in pathogenic fungi , as a "functional nervous system" that controls virulence and modulates the outcome of the disease [37] . General “housekeeping” responses to stimuli would presumably be similar across the STs , so the highly developed exclusive MAPK pathway in ST1 must have some unique consequences and could be worthy of future functional studies . A stringent , 4-step approach identified a total of 89 genes confidently predicted to encode secreted proteins in Blastocystis ST1 ( S1 Data ) . The corresponding predicted number for ST7 was 307 [21] , although those results were based only on SignalP predictions , a less stringent approach than the one used here . Most of the 89 secreted proteins predicted in ST1 are also present in ST7 , with the exception of a metallophosphoesterase and GH2 and GH33 glycosyl hydrolases . Several of the 89 genes were not predicted as secreted in ST7 because their signal peptides were not part of the predicted gene models . More than half of these putative secreted proteins have a predicted role in posttranslational modification and protein turnover ( Fig 5 ) . We compared the secretory signal peptides of the different STs and were unable to identify a unifying characteristic apart from a core of 10 hydrophobic amino acids ( S2 Fig ) . The majority of the secreted protein-coding genes in the 3 STs were present in more than 1 copy , with a notable expansion of cysteine proteases ( see below ) . Intriguingly , ST1 has 4 genes that are predicted to code for secreted collagen-like proteins . These 4 genes are very different at the amino acid level , with pairwise percentage identities only in the mid to upper 30s , but all have signal peptides that suggest they are secreted and all have the distinctive collagen-like motif GXY ( glycine , second and third residues can be anything but frequently proline and hydroxyproline ) repeated 63 times . ST4 also has 4 copies of genes encoding these collagen-like proteins , with only 1 predicted to be secreted , whereas ST7 has a single gene encoding a related protein that lacks a signal peptide . Based on sequence similarity and repeat characteristics , the Blastocystis collagen-like proteins are of the bacterial type [38] . A number of bacterial pathogens have collagen-like proteins that are involved in pathogenicity , immune response elicitation , and host–parasite interactions . For example , collagen-like proteins are able to bind to the human extracellular matrix , thereby aiding adhesion and colonization [39–41] . It is possible that collagen is one of the factors mediating adhesion of Blastocystis to enterocytes . Another possibility is that collagen may be part of a mechanism used by Blastocystis to trap bacteria and other microbial eukaryotes for nutritional purposes . This process has been observed by electron microscopy , although an exact mechanism has not been described [42–44] . At present , the roles played by collagen-like proteins in Blastocystis are purely speculative , based on their similarities to bacterial proteins that are clearly involved in pathogenicity . However , it suggests fertile ground for further investigation , specifically into whether differences in both number of genes and potential for secretion are implicated in variable virulence . Another class of proteins demonstrating clear ST differences is proteases or peptidases . Proteases are crucial for many biological processes and constitute potential virulence factors in parasitic protists [45] . Cathepsin B , a cysteine protease , has been linked to increased intestinal cell permeability [8] , while other cysteine proteases have been reported to cleave human secretory IgA [46] and induce up-regulation of interleukin 8 cytokine transcription and secretion [47] in intestinal epithelial cells . Blastocystis has a large number of proteases , with the 3 STs having mostly similar profiles , but with some notable differences . The total number of proteases encoded in ST1 is 243 , with 198 in ST4 and 210 in ST7 ( S2 Data ) ( Fig 6 ) . All 3 STs encode aspartic , cysteine , metallo , serine , and threonine proteases . The most prevalent protease genes are those of the cysteine type , constituting between 39% ( ST7 ) and 47% ( ST4 ) of the degradome . Undoubtedly , many cysteine proteases play roles that are conserved in eukaryotes , such as lysosomal function , autophagy , and ubiquitination , but others have been implicated in host–parasite interactions [48–50] . All 3 STs show extensive gene expansions of the cysteine protease families C1 , C13 , and C19 . A large number of C1 genes is fairly common . Less common is the number of C13 genes found in ST1 ( 16 ) , ST4 ( 11 ) , and ST7 ( 11 ) . Most protist genomes contain fewer than 5 C13 genes and many just have a single version ( see MEROPS database , merops . sanger . ac . uk , [51] ) . The only protist known to approach the number of C13 genes seen in Blastocystis is the sexually transmitted human excavate Trichomonas vaginalis ( 10 ) , in which at least 1 type of C13 protein has been implicated in trichomonal cytoadherence [52] . While Blastocystis appears to have an elevated number of C19 genes compared with other protists ( MEROPS database ) , currently there is no indication that these genes are involved in anything other than standard intracellular removal of ubiquitin molecules . Other protease families that exhibit large differences between STs are C56 , C95 , C97 , and M16 ( S2 Data ) . The biological significance , if any , of these differences is currently unknown . The most striking disparity between the STs is seen in metallo-type proteases ( Fig 6 , S2 Data ) . The divergence in the number of genes belonging to this type is entirely attributable to the complete absence of subfamily M23B genes in the ST4 genome , while both ST1 and ST7 have 29 members . The M23 family is composed of metallopeptidases involved in the lysing of peptidoglycans in bacterial cell walls for either defense or feeding ( MEROPS database ) . What functions they possess in Blastocystis and why they are absent from ST4 are open questions . Differences in the number of the S54 rhomboid serine proteases were also identified , with ST1 having 4 genes , ST4 having 3 , and ST7 having only 1 . These transmembrane peptidases play a role in the invasion of host cells in the alveolates Toxoplasma , Cryptosporidium , and Plasmodium , while in the amoebozoan Entamoeba histolytica , the single rhomboid protease is vital to immune evasion via the cleavage of lectins during receptor capping [53–55] . The membrane-trafficking system enables transport of proteins and lipids between intracellular locations and is an interface with the extracellular environment . Crucial to the healthy working of eukaryotic cells , it underpins the pathogenic mechanisms of diverse eukaryotic parasites through the release of virulence factors as well as the uptake of metabolites from the host . Trafficking is enabled by a suite of proteins involved in vesicle formation and fusion [72] . Comparative genomic and molecular phylogenetic analyses have established that a relatively complex complement of membrane-trafficking machinery was present in the last eukaryotic common ancestor ( LECA ) [73] . Our analyses showed that homologs of nearly all components of the vesicle-formation and -fusion machinery present in the LECA are also encoded in Blastocystis STs 1 , 4 , and 7 ( S3 Fig ) . Indeed , there is evidence for somewhat expanded paralog numbers in several of the complexes , most notably in the Sec24 subunit of the COPII coat; the Arf GAPs; the HOPS and GARP tethering complexes; and the adaptor protein complexes 1 , 2 , and 4 ( S3 Fig ) . This is also where much of the variability between the STs is observed ( along with the numbers of some endosomal sorting complexes required for transport [ESCRT] paralogs ) . A number of Ras superfamily proteins ( or small GTPases ) known to be involved in membrane and protein trafficking have orthologs in Blastocystis ( S4 Data ) . Notable among them are 2 ( ST7 ) or 3 ( ST4 , ST1 ) paralogs of a GTPase that serves as the membrane-anchored β subunit ( SRβ ) of the signal-recognition particle receptor on the ER that is involved in the cotranslational import of proteins into the ER [74] . Most eukaryotes , including other stramenopiles , employ only 1 SRβ protein , so the functional significance of the varying number of SRβ paralogs in Blastocystis is unclear . Blastocystis STs also harbor some Ras superfamily members that appear to have emerged in this lineage through extensive divergence of existing or duplicated genes . These can be considered “novel” genes and may potentially underpin significant evolutionary innovations . For instance , ST1 and ST4 both encode a divergent Rab7-like paralog ( Rab7L in S4 Data ) that is apparently missing from the genome of ST7 . Although the cellular function is difficult to predict for such divergent paralogs , the similarity of Rab7L to standard Rab7 proteins suggests that it may be involved in some trafficking processes associated with lysosomes or vacuoles . The Blastocystis genomes encode 5 additional divergent Rab-related proteins whose evolutionary origin cannot be deduced with confidence and hence are labeled RabX1 to RabX5 ( S4 Data ) . These proteins are presumably involved in specialized membrane-trafficking pathways in Blastocystis , but their functions cannot be predicted from sequence analyses alone . The 2 Rab1 paralogs in Blastocystis , annotated as Rab1A and Rab1B , differ from the previous examples in that their origin is more ancient , as they apparently stem from a Rab1 duplication previously suggested to be an evolutionary novelty of the SAR supergroup [75] . The functional significance of the 2 differentiated Rab1 paralogs in organisms of the SAR clade remains unknown despite an attempt to characterize both paralogs in Toxoplasma gondii cells [76] . Most recently , Rab1A was found in association with rhoptries in schizonts of P . falciparum and suggested to be involved in regulating vesicular trafficking from the ER to the former secretory organelles of the parasite [77] . Control of the cell cycle and cell proliferation is mediated extensively by the anaphase promoting complex/cyclosome ( APC/C ) [78] . It functions as an E3 ubiquitin ligase that coordinates the degradation of specific substrates via the 26S proteasome at specific points in the cell cycle [79] . Typically , the complex is composed of about a dozen subunits with a combined mass of about 1 . 5 MDa . It can be divided into 3 functional parts or subcomplexes [80]: ( i ) a structural complex serving as a scaffold , ( ii ) a catalytic arm , and ( iii ) a tetratricopeptide repeat ( TPR ) arm designed to position the substrate for successful transfer of ubiquitin . Surprisingly , Blastocystis seems to lack genes for all the subunits that make up the scaffold ( i . e . , APC1 , APC4 , and APC5 ) , and to encode a reduced TPR arm composed of only 2 subunits versus typically 4 in other stramenopiles and up to 7 in other eukaryotes [78] . Particularly notable is the loss of the APC3 subunit , which is found in virtually all eukaryotic organisms . However , Blastocystis appears to be unusual in that it has several paralogs encoding the TPR subunit APC8 , which potentially may compensate for the absence of the other subunits in the TPR arm ( S5 Data ) . The APC/C interacts with a number of adaptors and coactivators that modulate its activity and specificity . The most important of these adaptors are cell-division cycle protein 20 ( Cdc20 ) and cadherin-1 ( Cdh1 ) because , at various stages of the cell cycle , they are essential for activation of the complex and selecting which substrates to interact with [81] . In particular , APC/C-Cdh1 plays a role in DNA synthesis during the G1/S phase because it allows the 26S proteasome to degrade several DNA replication inhibitors [80] . Surprisingly , while the vast majority of eukaryotes possess both Cdc20 and Cdh1 , Blastocystis only has a homolog of Cdc20 . How the end of anaphase is regulated in Blastocystis remains an open question . In addition , genes encoding 2 of the main APC/C targets that are involved in the integrity and regulation of the cohesion complex , Scc2/Scc4 and Eco1 , were absent . This protein complex keeps sister chromatids together and regulates their separation during cell division . Scc2/Scc4 aids in loading the complex onto the chromosome , while Eco1 is responsible for the establishment of cohesion between cohesin and chromatin [80] . The absence of these components in Blastocystis and other stramenopiles suggests that an alternative route may exist to achieve a properly functioning cohesion complex and the separation of sister chromatids . Potential homologs of proteins involved in DNA damage response and repair , chromatin structure relevant to repair , and meiosis , a process not previously attributed to the life cycle of Blastocystis , were identified ( S4 Fig , S6 Data ) . Homologs of genes encoding 9 out of 11 meiosis-specific proteins required for the progression of meiosis in other organisms [82 , 83] are found in Blastocystis ST1 and ST4; these include Hop1 , Spo11-2 , Top6BL , Dmc1 , Hop2 , Mnd1 , Msh4 , Msh5 , and Mer3 . Msh5 is absent from ST7 , and Rec8 and Spo11-1 were not identified in any of the STs . Similar to T . vaginalis [84] , the Blastocystis genomes apparently do not encode components of the nonhomologous end-joining machinery , suggesting that homologous recombination is the principal mechanism for the repair of double-stranded DNA breaks . A total of 203 carbohydrate active enzymes ( CAZymes ) were identified in the Blastocystis ST1 genome ( S7 Data ) . The most interesting cases are discussed below and additional details are provided in S1 Text . Unlike the intestinal parasites Giardia and Entamoeba , which rely on the host as a source of purines and pyrimidines [95 , 96] , Blastocystis ST1 possesses complete pathways for the de novo synthesis of these compounds ( S8 Data ) . Its capacity for de novo amino acid biosynthesis is limited to alanine , aspartate , and glutamate , while serine and glutamine can be produced via conversion from other amino acids . Blastocystis ST1 has a mostly complete folate biosynthesis pathway , lacking only the alkaline phosphatase responsible for the dephosphorylation of 7 , 8-dihydroneopterin 3′-triphosphate . However , this enzyme is also lacking in many probiotic bacteria such as bifidobacterial , which appear to use another NUDIX enzyme [97] . While genes encoding homologs of this protein could not be identified , it seems likely that Blastocystis ST1 uses an uncharacterized alkaline phosphatase to complete the pathway . Unlike most parasites , Blastocystis ST1 appears to have a nearly complete de novo thiamine ( vitamin B1 ) biosynthesis pathway . In this pathway , 4-amino-2-methyl-5-hydroxymethylpyrimidine ( HMP ) is pyrophosphorylated by HMP kinase sequentially to form HMP-PP , which is in turn condensed with thiazole to form thiamine phosphate by thiamine-phosphate pyrophosphorylase . Only the thiazole salvage enzyme hydroxyethylthiazole kinase ( ThiM ) could be identified , suggesting thiamine biosynthesis in Blastocystis ST1 is either dependent on exogenous thiazole or synthesizes thiazole by an unknown mechanism . The latter is not unprecedented , as P . falciparum has been shown to synthesize vitamin B1 de novo despite having the same repertoire of thiamine synthesis-related genes as Blastocystis [98] . Cytochrome P450 ( CYP ) is a large and versatile heme-containing protein superfamily containing at least 317 families and numerous subfamilies ( https://cyped . biocatnet . de/sequence-browser ) . These proteins are found in all domains of life and in all major eukaryotic lineages . However , no homologs were found of any of the CYP families in the 3 Blastocystis genomes . Genomes available for other stramenopiles do encode various CYPs . The diatom P . tricornutum has 3 CYP genes , consisting of 2 CYP97 and 1 CYP51 enzymes , whereas T . pseudonana encodes 9 family members [99] . P . falciparum was the first documented eukaryotic species without a CYP-encoding gene [100] . However , it should be noted that Kinetoplastida such as T . brucei have a limited set of CYPs devoted to sterol synthesis ( e . g . , CYP51A ) . Another apicomplexan parasite , T . gondii , has a single CYP gene , encoding a steroid 11-β hydroxylase . Since sterols are essential for eukaryotic membranes , the lack of CYP51 in Blastocystis suggests that it obtains sterols from its host , as does Giardia [101] . The intestinal eukaryote Blastocystis continues to be of significant clinical interest because of it widespread prevalence . Despite years of study , however , its pathogenicity and role in the gut remain controversial . This ambiguity is compounded by the presence of STs that , while morphologically indistinguishable , are nevertheless genetically heterogeneous . The extent of the differences and the degree to which they matter clinically are still unclear , but the increase in genomic data reported here opens up the possibility of answering some of the outstanding questions through comparative analyses . Structurally , there is considerable variation among the 3 Blastocystis ST genomes now available in terms of genome size and GC-content . ST1 also displays sizeable differences compared to ST7 and ST4 with regard to the number of genes identified and general gene characteristics , such as intron numbers and gene size . However , the use of a novel mechanism by which stop codons are generated was examined thoroughly in ST1 and led us to conclude that some of the differences in gene numbers and characteristics could be due to previous annotation efforts in ST4 and ST7 not taking the stop codon issue into account . While the need for reannotation of ST4 and ST7 affects the quality of the protein data sets from these STs to some extent , nevertheless , divergence at the amino acid level between homologous proteins has been demonstrated that is almost an order of magnitude greater than that seen between species within other genera of parasitic protistans . Analyses of protein classes of interest revealed numerous differences . Of interest will be the variation in the number and type of protease genes identified among the STs , since this class of protein has been linked to pathogenicity . Another intriguing area ripe for exploration and experimentation is the kinome . ST-specific expansions of protein kinase genes were identified . In particular , ST1 encodes a large number of novel proteins from the STE20/7 family , which is typically involved in responding to extracellular stimuli . General aspects of the Blastocystis genome were examined that help in understanding its biology and how it relates to other stramenopiles and other parasitic protistans as well . Some of the highlights ( see S1 Text for additional analyses ) include an expanded proteome of the MRO , the absence of peroxisome-related genes , the presence of α-glucan metabolism , and intriguing expansions and additions to the repertoire of membrane-trafficking machinery proteins . The extensive structural and gene complement differences between the genomes of the 3 STs suggests the need for a revised taxonomy of Blastocystis . Since Blastocystis has been linked to disease in humans , the primary focus of any genomics analyses will likely be medically oriented . The availability of Blastocystis genomes will greatly assist future functional studies . In particular , the development and successful implementation of a transformation system , to allow finely tuned control over constructs as well as localization experiments , will depend on mining the genome for appropriate primers , promoters , and genes of interest . Resequencing to evaluate the success of transformations and any off-target responses will be facilitated by preexisting genomes . Pathogenomic studies of Blastocystis will benefit greatly from access to a genome of ST1 . Comparisons with existing and future genomes from other STs and other isolates of ST1 will help in finding genes that are correlated with virulence as well as the development and testing of drug targets . Equally as important to elucidating the roles of Blastocystis STs in the human gut is understanding the biology of these anaerobic protists and their repertoires of metabolic pathways , which allow them to thrive in the gut environment .
Blastocystis sp . NandII ( ST1 ) was obtained from the American Type Culture Collection ( ATCC 50177 ) and maintained in Locke's medium and horse serum egg slants at 35 . 6°C in an anaerobic chamber [102] . Blastocystis ST1 cells were harvested by centrifugation at 850xg for 15 min at 4°C . Total DNA was extracted with the standard CTAB protocol [103] . Subsequently , Hoechst dye cesium chloride ( CsCl ) centrifugation was used to obtain purified nuclear DNA . Extracted DNA was diluted with TE buffer to a final volume of 10 mL and resuspended in 11 . 5 g of CsCl . Ten mg of Hoechst dye was added and the mix was homogenized by shaking for 3 hours . The solution was then transferred into Quick-Seal centrifuge tubes and centrifuged in a fixed-angle Ti-75 rotor at 40 , 000xg for 44 h . The resulting DNA bands were visualized under long-wave UV light . The AT-rich organellar band was removed and the remaining nuclear DNA band retrieved with a 30-gauge needle . Hoechst dye was removed with 3 successive washes of water-saturated butanol . DNA was precipitated with 100% ethanol and suspended in water . Total RNA was isolated using a modified Trizol protocol [104] , in which the supernatant , after the first round , underwent another Trizol extraction . The resulting RNA was used to create a cDNA paired-end library ( Vertis Biotechnologies AG [Germany] ) that was then sequenced on a 4SLX Titanium Platform ( Genome Quebec ) . The reads were filtered for primer , vector , and adaptor sequences . Following filtering , 69 , 155 , 000 read pairs remained . A second cDNA library was constructed ( Beijing Genomics Institute ) and RNA-Seq data was generated using the Illumina HiSeq 2000 system . As above , the reads were filtered for primer , vector , and adaptor sequences . The reads were also filtered by quality scores using FASTX-Toolkit ( version 0 . 0 . 13 ) http://hannonlab . cshl . edu/fastx_toolkit/index . html . To be retained , a minimum of 70% of the bases had to have quality scores of 20 or better [99] . Ultimately , 370 , 469 , 956 reads remained . The miraEST [105] and Trinity [106] assembly software programs were used to assemble the paired-end reads into contigs . 20 , 426 contigs were assembled . Only contigs above 200 bps were considered for further analysis . The longest contig was 18 , 604 bps and N50 was 793 bps . The Blastocystis ST1 genome was sequenced using 454 pyrosequencing and Illumina technologies . Illumina sequencing entailed generation of a mate paired-end library with an insert size of 3 kb . Following the filtering steps described above , 1 , 865 , 740 454 and 55 , 617 , 444 Illumina reads remained . The 454 and Illumina reads were assembled using the Ray genomic assembler software [107] . Only contigs above 200 bps were retained for further analysis . Using different combinations of 454 and Illumina reads , 3 different assemblies were created . 2F was generated from just Illumina reads and contained 13 , 338 contigs . 2E was generated from all reads and contained 29 , 673 contigs . 2D was generated just from 454 reads . The software Minimus2 from the AMOS package [108] was used to merge the 3 into a single assembly with 14 , 304 contigs . The genome assembly for Blastocystis ST1 was interrogated for the presence of contigs composed of bacterial reads . For contigs of less than 500 bps , BLASTx and BLASTn [49] searches were performed . All contigs with a percent identity of 95% and higher against entries in the nr and nt NCBI databases were considered contaminants . In total , 305 contigs were removed . Large contigs were split into 5 , 000-bp pieces while those less than 5 , 000 bps in length were kept intact . All the pieces were compared , using BLASTx , against a specialized database consisting of protein sets from all publicly available bacterial genomes and protein sets from 28 eukaryotic genomes drawn mainly from Chromalveolata and Archaeplastida . Pieces that only had matches for bacterial proteins were putatively designated as derived from bacterial contamination . Because of the possibility of LGT from bacteria , all of the 5 , 000-bp pieces from a large contig were required to be “bacterial” to warrant removal of the contig . In total , 459 contigs were set aside as being bacterial in nature and most probably resulting from bacterial contamination . 150 were also removed as clearly derived from contaminating Saccharomyces cerevisiae and E . histolytica DNA . After the removal of bacterial/yeast/Entamoeba contamination , attempts were made to assemble the ST1 contigs into larger scaffolds . SSPACE [109] , which is specially designed to work with preassembled contigs , was used on those contigs greater than 500 bps . The 2 , 494 contigs were assembled into 693 scaffolds with the N50 increasing from 8 , 801 bps to 50 , 205 . The Blastocystis ST1 genome sequences are available from the NCBI with the accession LXWW00000000 . In the absence of an experimentally determined genome size for ST1 , the extent to which the final assembly represents the full genome was determined roughly using 2 methods . The first method entails taking a set of 679 conserved protein sequences derived from RNA-Seq data and matching them against the genome assembly . Of the conserved transcripts , 5 . 4% were not found in the genome assembly , suggesting that the current assembly encompasses 94 . 6% of the genome or at least the gene space of the genome [110] . The second strategy for estimating genome coverage was based on a method first used in the Conus bullatus genome paper [111] to estimate genome size . A database of 31 . 8 million 90-bp Illumina reads was used in BLASTn searches of the assembled contigs . Read coverage for contigs was estimated as ( number_of_hits*read_length ) /length_of_contig . The mode was determined from a histogram of the read coverages of the contigs and used to estimate genome size: ( number_of_reads*read_size ) /read_coverage_mode ( ( 31 , 826 , 528*100 ) /153 = 18 . 7 Mb ) . The estimated size was then compared with the assembled genome size ( 17 . 4 Mb/18 . 7 Mb = 93% ) . Additionally , the protein data sets for the 3 genomes were used for BUSCO [112] analyses to assess their completeness based on a set of 429 highly conserved single copy eukaryotic genes . An initial BUSCO analysis was done using the ST protein data set . Any conserved genes that were considered missing were then compared against the appropriate ST genomic scaffolds to determine if the gene had been missed during annotation . Given the evolutionary distances between Blastocystis and the organisms used to generate the BUSCO test set , the likelihood of achieving 100% coverage was low . More useful was a comparison of the results for the 3 STs . The average level of heterozygosity for ST1 was determined by aligning trimmed genomic reads to the genomic assembly with Bowtie 2 . 3 . 1 [113] . The program mpileup from SAMtools [114] was used to determine the read depth at each position of the assembly as well as the breakdown of the 4 possible bases . An in-house perl script was used to parse the mpileup data . Positions with less than 10X coverage were ignored . A site was deemed to be heterozygous if at least 2 different bases were present and there were at least 2 ( or 3 ) reads with the different bases . The number of heterozygous sites was then divided by the total number of sites ( with ≥10X coverage ) . A set of 679 protein sequences was generated from the ST1 RNA-Seq clustered transcripts and used to train AUGUSTUS [115] on a repeat masked assembly [116] . The training set contained protein sequences with the following criteria: at least 1 intron , less than 80% identical at the amino acid level to any other sequence in the set , and a start site based on highly similar eukaryotic protein sequences . The script optimize_augustus . pl was also run to create species-specific meta parameters . A set of 32 , 269 Trinity [106] assembled RNA-Seq transcripts were used to generate a hints file for donor/acceptor intron sites . Subsequent gene modeling generated 5 , 637 gene models , which were then corrected with PASA [117] based on RNA-Seq transcripts . Because of the unusual nature of termination codons in Blastocystis [25] , it was necessary to set the parameter stopCodonExcludedFromCDS to true . Of the final gene model set , 58% was examined manually at some point to check various aspects such as intron boundaries , stop codons , and chimeric models . The degree to which intron boundaries were confirmed by RNA-Seq data was determined by mapping RNA-Seq data to the genome assembly using Bowtie2 [113] . Using the intron boundary positions and the alignment positions of individual reads from the SAM file , an in-house perl script calculated that 70 . 3% of the intron boundaries were confirmed by 5 or more RNA-Seq reads ( 60 . 1% with 10 or more RNA-Seq reads ) . To assist in the manual annotation of the gene models , various information was generated for each model . The models were compared against the conserved domain database [118] and domain hits as well as Pfam [119] hits were extracted from the results . The models were also compared against the NCBI protein database as well as the protein data set for ST7 using BLASTp . SignalP results were generated using a locally installed version of SignalP . The genome browser Genomeview [120] was chosen for annotation because of its relative ease in setting up and populating with information via gff files and the ability to alter the information using custom scripts . The genome browser also included tracks for genome reads mapped to the scaffolds ( Bowtie2 [113] and SAMtools [114] ) , genome read map coverage , RNA-Seq reads mapped to the scaffolds , and RNA-Seq map coverage . Because of the possibility of missed genes , the intergenic spacers were identified and the sequences compared against the NCBI protein database ( nr ) as well as the ST7 protein data set using BLASTx . Potentially missed genes were mapped to the scaffolds and displayed in the genome browser for confirmation and correction through manual annotation . The ST7 protein data set was also compared to the gene models for ST1 using BLASTp and the results mapped to the scaffolds as a genome browser track . Automated annotation for eukaryote genomes , particularly those from nonmodel organisms and poorly sampled lineages , are prone to missed genes , i . e . , the gene-finding algorithms fail to find all of the protein-coding regions . While having a transcriptome is invaluable for reducing the level of missed genes , invariably , some are not found . Therefore , as a general rule , the various analyses detailed below that involve presence/absence of genes , particularly when comparing STs , used techniques such as tBLASTn to interrogate the full genome sequence before declaring that a gene was missing from an ST . Additional steps to confirm the presence/absence of a gene are indicated in the individual sections . The genomic scaffolds for STs 1 , 4 , and 7 were ordered according to their lengths . They were further stripped of their Ns and header lines . The program GC-Profile [121] as implemented at http://tubic . tju . edu . cn/GC-Profile/ was used to create GC profile and GC content graphs for the 3 genomes . As recommended by the authors , a halting parameter of 300 and a minimum segmentation length of 3 , 000 bps were used . GC-Profile , a windowless method , was chosen from many different algorithms to avoid some of the disadvantages present in other methods to calculate GC content such as GC patterns dependent on window size , and lack of resolution [121 , 122] . The SAMtools option depth [114] was used to determine the read-depth coverage at each position of the genome scaffolds . These values were used to calculate median read-depth coverage and normalized median read-depth coverage using in-house scripts . SNPs were detected using the SAMtools options mpileup and bcftools . The level of amino acid divergence between Blastocystis protein sequence data sets was analyzed using reciprocal best BLAST hit protein sequence pairs . Because of concerns about retained introns in some of the STs as well as issues with incorrect stop codons , the calculation of amino acid identity scores for gene pairs was restricted to high-scoring segment pairs without the inclusion of gaps . The values were generated using an in-house perl script . For Blastocystis ST1 , the analysis of different intron categories was based on the final annotation of the genome , incorporating data from RNA-Seq and extensive manual curation . To analyze introns in the Blastocystis ST7 genome , the available annotation of the genome could not be relied on because it proved to be inaccurate in many cases . Therefore , analysis was restricted to introns that have direct support from EST sequences . Available ESTs were aligned to the genome sequence using STAR [123] and introns indicated by the alignments were identified and manually corrected if needed . Homologs of protein subunits specific for the minor spliceosome were identified by BLAST ( BLASTp , tBLASTn ) using sequences of previously defined subunits from Homo sapiens and plants . The identity of the candidate orthologs was confirmed by reciprocal BLASTp searches against the nr database at NCBI . Incorrect or missing gene models in Blastocystis ST1 were corrected or added to the annotation of the genome . Most of the identified homologs in Blastocystis ST7 are either represented by incorrect gene models or completely missing from the annotated genes but could be found in the assembly with tBLASTn . Models for snRNA in both STs were identified using the program cmscan from the Infernal package [124] . A multipronged approach was used to predict mitochondrial proteins . ( 1 ) Each gene model and translated transcript was queried for N-terminal mitochondrial targeting sequences using Mitoprot and TargetP [125 , 126] . Sequences that returned a score greater than 0 . 5 were further investigated as MRO candidates . ( 2 ) The top 10 BLAST hits retrieved from the nr database at NCBI were surveyed for any mitochondrial annotation and manually investigated . ( 3 ) All gene models and translated transcripts were queried against a local curated MitoMiner data set [127] and significant hits ( e-value < 1e-10 ) were further investigated . ( 4 ) Finally , a local MRO protein data set from various anaerobic protists including Pygsuia biforma , T . vaginalis , and Mastigamoeba balamuthi was queried against the gene models and translated transcript data sets . Mitochondrial targeting sequence scores and protein annotations are provided in S3 Data . Those proteins previously unrecognized as mitochondrial in Blastocystis species are shown in color in Fig 7 . Homologs of proteins suspected to be involved in cytokinesis were identified by BLASTp against the Blastocystis ST1 predicted proteome ( e-value cutoff: 10e-10 ) , using as queries previously identified mammalian orthologs [128] . Absence of homologs was verified by tBLASTn against the genomic scaffolds and transcriptome . Blastocystis hits were then used as queries to run reversed BLASTp searches against nr and results were manually inspected to confirm assignation to a given protein family . Identification of homologs of the APC/C complex and targets was carried out in a similar fashion , using as queries previously identified homologs in Blastocystis ST1 or , if these did not exist , various other eukaryotic homologs [78] . For calcium-binding proteins , H . sapiens homologs of protein families identified in [129] were used as queries to carry out BLASTp searches against the Blastocystis ST1 predicted proteome but also against all eukaryotic protein data present in GenBank as of February 2016 ( "Eukaryota[Organism]" option of the BLAST+ package ) . A query database including 135 DNA repair genes , meiosis-specific genes , and meiosis-related genes from a wide range of eukaryotes was established using literature and key word searches of the NCBI database . Genes with functional experimental evidence were used as queries to retrieve similar proteins using PSI-blast . After retrieval , protein sequence alignments were created and used to build HMM profiles with HMMER3 . 1b2 [130] to enable gene searches in Blastocystis STs 1 , 4 , and 7 . Retrieval of genes was restricted to e-values below 1e-04 . Phylogenetic trees were constructed with FastTree [131] using closely related paralogs and/or reconstructing entire gene families for each gene of interest . Protein domains were mapped to the trees to facilitate gene recognition . In a few cases , secondary structure predictions were made to confirm the gene identification . 221 kinases were identified from the protein-coding gene set of Blastocystis ST1 using HMMER [130] . The STYKc profile ( SM00221 ) from the SMART database [132] was used , as was the Pfam Pkinase profile ( PF00069 ) [119] . Blastocystis kinases were classified by their positions in phylogenetic trees inferred from the alignments of the HMMER output ( both neighbor joining and maximum likelihood trees were taken into consideration ) . Blastocystis gene models with 2 kinase domains were classified by the phylogenetic position of the domain with a better HMMER score . The kinase classification follows the webite www . kinase . com [33] . The 221 kinases identified from Blastocystis ST1 were used to query the protein-coding gene sets from STs 4 and 7 . The annotations for STs 4 and 7 were also searched for the key word “kinase” and putative protein kinases were checked against the kinase database ( www . kinase . com ) using the BLAST search tool . The resulting sequences from STs 1 , 4 , and 7 were aligned using MUSCLE [133] , followed by manual inspection of the alignment . A maximum likelihood tree was inferred by the RAxML program [134] with the PROTGAMMALG model and visualized using Figtree ( http://tree . bio . ed . ac . uk/software/figtree/ ) . The Blastocystis STs 1 and 7 protein models were searched by BLASTp with CYP protein sequences representing the CYP2 ( DmelCYP307A1 , DmelCYP303A1 , and DmelCYP18A1 ) , CYP3 ( HsCYP3A4 , HzCYP321A1 , DmelCYP6A2 , DmelCYP6A8 , and DmelCYP6G1 ) , and CYP4 ( DmelCYP4G15 and DmelCYP4G1 ) animal clades [135] . CYPs from the mitochondrial clan were searched using DmelCYP314A1 , DmelCYP302A1 , DmelCYP12A1 , DmelCYP301A1 , DmelCYP49A1 , and DmelCYP315A1 . Additional animal and fungi CYP clans [136] , clan 51 , clan 7 , clan 26 , clan 20 , clan 46 , clan 19 , and clan 74 , were also used to search for CYPs in the Blastocystis protein models as well as the CYPome from sequenced genomes of Guillardia theta , T . pseudonana , and Phytophthora sojae . CYP51 clan involved in sterol biosynthesis was further studied using evolutionarily close CYP51s including Galdieria sulphuraria CYP51 , Porphyridium purpureum CYP51R1 , Batrachochytrium dendrobatidis CYP51F1 , Cyanidioschyzon merolae CYP51G1 , Dictyostelium discoideum CYP516A1 , L . major CYP51E1 , Monosiga brevicollis CYP51A1 , T . pseudonana CYP51C1 , and Chlamydomonas reinhardtii CYP51G . CYP searches for other organisms were done as above but using NCBI , CYPED ( https://cyped . biocatnet . de/ ) , and the eukaryotic pathogen database EuPathDB ( http://eupathdb . org/eupathdb/ ) [137–139] . The components of the membrane-trafficking system in Blastocystis STs 1 , 4 , and 7 were identified using both comparative genomic and phylogenetic methods . To identify potential homologs , functionally characterized membrane-trafficking components from H . sapiens and S . cerevisiae were used as queries to search the Blastocystis predicted proteins using BLASTp with default search parameters . Hits with an e-value ≤5e-02 were considered candidate homologs and were used as BLAST queries to reciprocally search predicted proteins of the H . sapiens and S . cerevisiae genomes . The Blastocystis protein was considered homologous to the initial H . sapiens query if it retrieved the query or a clear ortholog as ( i ) the top hit and ( ii ) with an e-value ≤5e-02 . If a component could not be positively identified using BLAST searching , a hidden Markov model ( HMM ) was created using HMMER version 3 [140] , including clear homologs from combinations of the following taxa: H . sapiens ( NCBI , http://www . ncbi . nlm . nih . gov/ ) , S . cerevisiae ( NCBI ) , D . discoideum ( dictyBase , http://dictybase . org/ ) , Arabidopsis thaliana ( NCBI ) , C . reinhardtii ( Phytozome , http://www . phytozome . net/ ) , P . sojae ( JGI , http://www . jgi . doe . gov/ ) , T . pseudonana ( JGI ) , Naegleria gruberi ( JGI ) , Bigelowiella natans ( JGI ) , and Emiliania huxleyi ( JGI ) . The HMMs were used to search the Blastocystis STs 1 , 4 , and 7 genomes ( hmmsearch ) . Hits with an e-value ≤5e-02 were used as BLASTp queries to search the H . sapiens genome , and homologs were identified using the same criteria as were used in the initial BLASTp searches . If BLASTp or HMMer failed to identify a membrane-trafficking component , tBLASTn searches were performed , using the H . sapiens and S . cerevisiae queries to search the Blastocystis scaffolds . The region of the scaffold with a hit that had an e-value ≤5e-02 was excised and used as a BLASTx query to search the H . sapiens genome , and homologs were identified using the same criteria as were used in the BLASTp and HMMer searches . The more permissive e-value cutoff of 5e-02 was intended to reduce the number of false negatives from divergent Blastocystis genes when searching with the experimentally characterized opisthokont queries . Nonetheless , we found in post hoc assessment that over 93% of the orthologs identified had e-values of lower than 1e-05 in either the forward or reverse BLAST searches . In searching for the components of a protein family in Blastocystis , Bayesian and maximum likelihood phylogenetics methods were used to more rigorously determine orthology . For each phylogeny , a combination of homologs from H . sapiens , S . cerevisiae , T . pseudonana and P . sojae , P . infestans , A . thaliana , D . discoideum , and N . gruberi were collected and aligned with the Blastocystis homologs using MUSCLE version 3 . 8 . 31 [133] and manually adjusted . Positions of ambiguous homology were removed ( alignments available upon request ) and ProtTest 3 . 2 [141] was used to determine the best-fit model of sequence evolution , which , in all cases , was the LG model [142] , incorporating rate among site and invariant site corrections where relevant . Phylobayes 3 . 3 [143] was used to produce the optimal topology and posterior probability values . Analyses were run until the average standard deviation of the split frequencies fell below 0 . 1 and the effective sample size was at least 100 . Once convergence occurred , the first 20% of sampled trees were removed . Additionally , RAxML and , in some cases , PhyML version 3 [144] were used to obtain maximum likelihood bootstrap values ( 100 pseudoreplicates ) . Resultant phylogenetic trees were viewed using FigTree v1 . 4 . 0 . To identify putative PEX genes in Blastocystis STs 1 , 4 , and 7 , protein sequences from H . sapiens , S . cerevisiae , and/or Neurospora crassa were used as queries for BLASTp and tBLASTn searches against locally hosted Blastocystis protein sequences and scaffolds , respectively . Candidate homologs with e-values less than or equal to 0 . 05 were subjected to reciprocal BLASTp searches against the locally hosted query genome as well as the locally hosted nr database . The retrieval of the original query as the top reciprocal BLASTp hit with an e-value less than or equal to 0 . 05 was the criteria for identification of a putative homolog . To identify putative PEX genes in A . thaliana and the stramenopile genomes of P . ramorum , T . pseudonana , and P . tricornutum , H . sapiens , S . cerevisiae , and/or N . crassa protein sequences were used as queries for pHMMer [145] searches against locally hosted query genomes . Candidate homologs with e-values less than or equal to 0 . 05 were subjected to reciprocal pHMMer searches against the locally hosted genome and NR database . Thus , the retrieval of the original query as the top reciprocal pHMMer hit with an e-value less than or equal to 0 . 05 was the criteria for identification of a putative homolog . Results were compared to previously listed PEX genes in A . thaliana [146] and P . tricornutum [68] . Newly identified A . thaliana , P . ramorum , T . pseudonana , and P . tricornutum PEX genes were subsequently used as queries for BLASTp and tBLASTn searches against locally hosted Blastocystis protein sequences and scaffolds , respectively . Candidate homologs were subjected to reciprocal BLASTp searches according to the criteria described above . The CAZy annotation pipeline was used to analyze 5 , 966 predicted proteins from the Blastocystis ST1 genome using a 2-step procedure of identification and annotation [147] . Sequences were subjected to BLASTp analysis against the CAZy database , composed of full-length proteins . Hits with an e-value <0 . 1 were then subjected to a modular annotation procedure using BLASTp against libraries of catalytic and carbohydrate binding modules and profile HMMs [148] . The results were complemented with signal peptide , transmembrane , and glycosylphosphatidylinisotol ( GPI ) anchor predictions [149 , 150] . Fragmentary gene models and all models suspected of containing errors were identified and flagged . A final functional annotation step involved performing BLASTp comparisons against a library of protein modules derived from biochemically characterized enzymes [147] . Three other stramenopile species , the diatom T . pseudonana , and the oomycetes , Albugo laibachii Nc14 and P . infestans T30-4 , are present in the CAZy database and were used for comparison . The predicted CAZymes encoded by the genome of the brown alga Ectocarpus siliculosus ( http://bioinformatics . psb . ugent . be/orcae/overview/Ectsi ) were also annotated using the same procedures for comparison with other stramenopiles . To predict various pathways related to intermediary metabolism ( e . g . , nucleotide , amino acid , cofactor metabolism , etc . ) , the predicted gene models and RNA-Seq transcripts were annotated using the KEGG automated annotation server ( KAAS ) using the single-directional best hit approach ( see http://www . genome . jp/tools/kaas/ ) . Each pathway was individually searched for completeness and results are summarized in S8 Data . In some cases , in which KAAS did not predict an ortholog capable of a reaction , query sequences were retrieved from KEGG and manually searched against the scaffold and transcriptome data . All predicted gene models were run through SignalP with a threshold of 0 . 70 . The selected models were then run through TargetP in order to identify a secretory pathway signal peptide with a cutoff of 0 . 70 . Subsequently , the remaining models were run through WoLF PSORT [151] and only those predicted as extracellular were retained . As a final step , TMHMM [152] was used as a final checkpoint to find and remove any model with predicted transmembrane domains . Therefore , the secretome was identified as having a secretory pathway signal peptide , extracellular localization , and absence of transmembrane regions . All gene models were run through the MEROPS database [51] . Only the models with predicted active sites were considered . Annotation and naming of the models follows the MEROPS database terminology ( S2 Data ) . A reciprocal best BLAST analysis was performed for the protein data sets from Blastocystis ST1 and C . reinhardtii using the program orthoparahomlist . pl with default settings ( Stanke M . Orthoparahomlist . pl script . 2011 . https://github . com/goshng/RNASeqAnalysis/blob/master/pl/orthoparahomlist . pl ) . The resulting list of orthologs was compared against the 213 C . reinhardtii conserved ciliary proteins [153] to determine which of the orthologs found in Blastocystis ST1 had matches . Members of the Ras superfamily were searched for in the 3 Blastocystis genomes using well-characterized representative sequences from other eukaryotes ( H . sapiens , P . sojae , and others ) as queries . Both the databases of predicted protein sequences ( using BLASTp ) and the genome sequence assemblies ( using tBLASTn ) were searched to ensure that no gene was missed because of the lack of a corresponding protein prediction . Indeed , a number of genes that were missing in the protein sequence databases were identified in all 3 STs . The missing or inaccurate gene models in ST1 were manually predicted or corrected using various lines of evidence ( transcriptomic data and comparisons to homologs ) and incorporated into the main genome annotation . Models missing for the ST4 and ST7 genomes were constructed only when the phylogenetic relationship to ST1 genes was unclear from the tBLASTn searches . Orthologous and paralogous relationships among Blastocystis Ras superfamily genes were established primarily on the basis of reciprocal BLAST searches , but phylogenetic analyses were required in a few cases . Orthology to conserved eukaryotic groups of GTPases was , in most cases , obvious from BLAST comparisons , except for several highly divergent paralogs related or similar to Rab GTPases , which most likely represent lineage-specific , rapidly evolving genes derived from duplications of some common Rab genes . A more detailed analysis was carried out for the Blastocystis family of Miro proteins . These sequences and Miro homologs from 2 newly analyzed stramenopile lineages ( Labyrinthulea , represented by Aurantichytrium limacisporum , and Placididea , represented by Cafeteria sp . Caron lab isolate ) were added to a selection of Miro sequences previously analyzed by Vlahou et al . ( 2011 ) [63] , excluding the very divergent sequences of N . gruberi , ciliates , and trypanosomatids . The sequences were aligned using MAFFT ( version 7 , http://mafft . cbrc . jp/alignment/server/ , [154] ) and the alignment was trimmed using the Gblocks server ( http://molevol . cmima . csic . es/castresana/Gblocks_server . html , [155] ) with the least stringent parameters , keeping 245 aligned amino acid positions . The tree was calculated using RAxML-HPC ( 8 . 2 . 8 ) run at the Cyberinfrastructure for Phylogenetic Research ( CIPRES ) Portal ( http://www . phylo . org/sub_sections/portal/ ) , employing the LG+Г+F substitution model and rapid bootstrapping followed by a thorough search for the optimal tree . | Blastocystis are unicellular eukaryotic organisms related to algae and some plant pathogens . They are common constituents of the human gut microbial community , colonizing approximately 1 billion humans worldwide . Whether their presence is harmful or not continues to be hotly debated . Part of the uncertainty stems from the fact that at least 17 subtypes have been identified from various mammalian hosts , including 9 from humans . To better characterize and understand Blastocystis , we have sequenced and annotated the genome for subtype 1 and compared it with previous genomic results for subtypes 7 and 4 . The comparisons revealed considerable differences between the 3 sequenced subtypes for a number of genomic features like DNA base composition , size of genome , number of genes , and number of introns . We also examined various biochemical pathways and cellular systems in the context of the full gene complement to better understand the biology of Blastocystis , including some of its more unusual features like a mitochondrion related organelle . We also identified subtype-specific gene family expansions that may be related to virulence . Finally , we showed that Blastocystis appears to have most of the genes necessary for sexual reproduction . This study provides resources and hypotheses for future investigations into the biology and potential pathogenicity of these common gut microbes . | [
"Abstract",
"Introduction",
"Results",
"and",
"discussion",
"Methods",
"and",
"materials"
] | [
"methods",
"and",
"resources",
"enzymes",
"enzymology",
"parasitic",
"protozoans",
"genomic",
"databases",
"protozoans",
"genome",
"analysis",
"energy-producing",
"organelles",
"mitochondria",
"bioenergetics",
"cellular",
"structures",
"and",
"organelles",
"research",
"and"... | 2017 | Extreme genome diversity in the hyper-prevalent parasitic eukaryote Blastocystis |
Yeast RNA polymerase II ( Pol II ) terminates transcription of coding transcripts through the polyadenylation ( pA ) pathway and non-coding transcripts through the non-polyadenylation ( non-pA ) pathway . We have used PAR-CLIP to map the position of Pol II genome-wide in living yeast cells after depletion of components of either the pA or non-pA termination complexes . We show here that Ysh1 , responsible for cleavage at the pA site , is required for efficient removal of Pol II from the template . Depletion of Ysh1 from the nucleus does not , however , lead to readthrough transcription . In contrast , depletion of the termination factor Nrd1 leads to widespread runaway elongation of non-pA transcripts . Depletion of Sen1 also leads to readthrough at non-pA terminators , but in contrast to Nrd1 , this readthrough is less processive , or more susceptible to pausing . The data presented here provide delineation of in vivo Pol II termination regions and highlight differences in the sequences that signal termination of different classes of non-pA transcripts .
Termination of RNA Polymerase II ( Pol II ) transcription plays an essential role in the transcription cycle and has been the subject of several recent reviews [1] , [2] . Disruption of the elongation complex at terminators recycles Pol II maintaining a pool of free enzyme able to compete for unoccupied promoters . Correct termination also prevents Pol II from interfering with expression of downstream genes either by colliding with oncoming Pol II elongation complexes or by dislodging transcription factors from the downstream DNA template [3]–[5] . Termination can also serve a regulatory purpose . Several yeast genes are regulated by premature termination [6]–[8] and genes involved in yeast nucleotide metabolism are regulated by the choice of alternative transcription start sites , one of which leads to premature termination [7] , [9] . More recent studies have shown that correct termination is also necessary for efficient re-initiation at the same gene through the formation of a loop between the 3′ and 5′ ends [10] , [11] . In addition to mRNAs , Pol II transcribes a diverse set of non-coding RNAs including snoRNA , some snRNAs and several classes of ncRNA with unknown functions . In yeast , this pervasive transcription [12]–[14] falls into several classes . Cryptic unstable transcripts ( CUTs ) are turned over rapidly by the nuclear exosome [15]–[20] . In contrast , another class of yeast ncRNA , termed stable uncharacterized transcripts ( SUTs ) are observed in the presence of an active nuclear exosome [21] . Pol II terminates coding and non-coding transcripts by different mechanisms [1] , [2] , [22]–[24] . Coding transcripts and possibly some SUTs are processed at the 3′-end by the cleavage and polyadenylation ( pA ) machinery . This reaction is coupled to termination occurring downstream of the processing site [1] , [2] . In contrast , ncRNAs are terminated and processed by an alternative pathway that , in yeast , requires the RNA-binding proteins Nrd1 and Nab3 and the RNA helicase Sen1 [25]–[29] . Yeast Pol II terminators contain short RNA sequences that bind proteins within large complexes associated with the elongating Pol II . Loosely conserved pA signal sequences downstream of protein-coding genes bind to components of the CF1 complex leading to assembly of the cleavage and polyadenylation machinery [1] . Termination is coupled to cleavage in a manner that has not yet been completely resolved . Non-pA termination components Nrd1 and Nab3 recognize RNA sequence elements downstream of snoRNAs and CUTs [8] , [28] , [30]–[34] and this leads to the association of a complex that contains the DNA/RNA helicase Sen1 and the nuclear exosome [35] . The mechanism of termination of these ncRNA transcripts has also not yet been determined . Several possible mechanisms for Pol II termination have been proposed . The “torpedo” model postulates that cleavage at the pA site exposes an uncapped 5′ end on the nascent transcript that acts as a substrate for the 5′→3′ RNA exonuclease Rat1 in yeast or Xrn2 in metazoans [36]–[38] . The exonuclease degrades the nascent transcript and upon reaching the Pol II elongation complex facilitates termination by an unknown mechanism . Another model postulates that an allosteric change in Pol II occurs upon assembly of the pA complex [39] . A member of this complex , Pcf11 has been shown to dismantle an elongation complex in vitro [40] , [41] and it is possible that Nrd1 plays a similar role . The DNA/RNA helicase Sen1 interacts with the Pol II CTD [42] and has been proposed to act like the bacterial termination factor rho and track along the nascent transcript and pry off the elongating Pol II [1] , [25] , [26] , [43] , [44] . Mutation of Sen1 has been shown to lead to readthrough of both coding and non-coding transcripts in vivo [45] and in vitro can arrest transcription [46] . Part of the uncertainty in delineating termination mechanisms is identifying which factors operate at which terminators in vivo . In this study we have used PAR-CLIP to map Pol II on the yeast genome in living cells [8] , [47]–[49] after depletion from the nucleus of components of the different termination pathways [50] . By comparing the Pol II maps with and without nuclear depletion we are able to map the location of termination to narrow regions of the genome downstream of coding and non-coding genes . This approach avoids the complexity of using transcripts to map terminators as the effect of RNA turnover is eliminated . We show here that depletion of Ysh1 , the protein that cleaves nascent pre-mRNA transcripts at the pA site , enhances accumulation of Pol II at the 3′-ends of protein-coding genes . In contrast , both Nrd1 and Sen1 depletion lead to readthrough transcription of ncRNAs and our data has allowed us to map the position of non-pA termination facilitated by these factors .
To map in vivo termination we have used the anchor-away ( AA ) system [50] to deplete termination factors from the nucleus . In this approach rapamycin ( rap ) induces a complex between a protein tagged with a FKBP12-rapamycin binding ( FRB ) domain and an anchor protein tagged with an FKBP12 domain . In our case the FKBP12 domain is on the ribosomal protein RPL13A leading to rap-dependent depletion of FRB-tagged protein from the nucleus . Previous work has shown that a Pol II subunit RPB1-FRB strain shows greater than 90% depletion of Pol II at the PMA1 gene within 40 min [51] . We show here that in NRD1-FRBGFP or YSH1-FRBGFP strains that also contain , RPB3-TAGRFP , Nrd1 and Ysh1 are depleted from the nucleus with similar kinetics ( Figure 1A–B ) . We were unable to observe Sen1-FRBGFP because it is present only in ∼100–2 , 000 copies per cell . When over-expressed 10–100-fold from a Gal promoter Sen1-FRBGFP is visible and is depleted from the nucleus with similar kinetics ( Figure 1C ) . Given the limitations of background signals in live-cell imaging we cannot rule out the possibility that some small amount of Nrd1 , Ysh1 or Sen1 remain in the nucleus . Growth curves show that all three strains grow normally for several divisions after administration of rap and we carried out PAR-CLIP analysis well before any changes in growth were observed ( Figure S1A , shaded box ) . When cells are plated on media containing rap we observe no growth for YSH1-FRB and NRD1-FRB , while SEN1-FRB grows very slowly ( Figure 1D ) . Growing yeast cultures were cross-linked as previously described [8] , [49] with modifications described in Methods . Briefly , 4-thiouracil ( 4tU ) was added to a growing culture for 15 minutes to allow equilibration of uracil pools before the addition of rap . After thirty minutes of rap treatment cultures were irradiated with 365 nm UV for 15 min ( Figure S1B ) . Incubation of yeast cells in 4tU does not significantly affect growth during the course of the experiment as cells continue to grow beyond the time frame of the cross-linking for at least one more doubling ( 0 . 2 OD600 to 0 . 7 OD600 as measured by BioTek scanner ) indicating that 4tU has a less drastic effect on yeast growth than it does in mammalian cells treated for a longer time [52] . To map the position of elongating Pol II we isolated Pol II-bound RNA using a dual 6×His-biotin tagged Pol II Rpb2 subunit ( Rpb2-HTB ) as previously described [8] . Duplicate libraries were derived from the RPB2-HTB NRD1-FRB strain grown in the presence or absence of rap . We also created libraries of Rpb2-bound RNA from SEN1-FRB and the parental RPB2-HTB strains grown in the presence of rap and a single library from YSH1-FRB . Replicate libraries were multiplexed and sequenced by Illumina Hi-seq . Each library yielded between 15–39 million unique reads ( Table S1 ) . Biological replicates were strongly correlated ( ρ> . 99 ) and thus were added together for analysis ( Figure S2 ) . The WT dataset correlates very well to NET-seq [53] and GRO-seq [54] datasets ( Figure S3 ) . Thus , PAR-CLIP data accurately represents the position of Pol II elongation complexes on the yeast genome . To map pA-dependent termination we carried out PAR-CLIP on Rpb2 after depletion of Ysh1 , the factor that has been shown to be required for cleavage of the nascent mRNA at the GAL7 and CYC1 pA sites [55] . We reasoned that failure to cleave the nascent transcript would prevent Rat1 degradation thus disabling the “torpedo” mechanism . While we cannot rule out the possibility that other factors are depleted along with Ysh1 we do observe a failure to cleave at the pA site for several genes ( Figure S4 ) . The most dramatic effect of Ysh1 depletion is seen in a buildup of reads just downstream of the major pA site of highly expressed protein-coding genes ( Figure 2A and 2C ) . The peak of reads is present in most of the highly expressed protein-coding genes as seen in figure 2B . No similar effect is seen downstream of snoRNAs or CUTS . One surprising aspect of Ysh1 depletion is that readthrough Pol II does not extend further downstream but seems to pause or terminate within 200 bp of the pA site ( Figure 2A ) . This result indicates that Pol II that reads through termination signals fails to elongate but is unable to efficiently terminate . We do , however , observe several instances of Pol II extending 1 kb or more downstream of apparent terminators in response to Ysh1 depletion . The RNA14 and DBP2 genes show an increase of Pol II toward the 3′ end of the coding region indicating the presence of a Ysh1-dependent terminator upstream of the main pA site ( Figure 2D ) . Rna14p is part of the CF1 complex required for recognition of pA sites [1] and previous studies have shown that in addition to the mature 2 . 2 kb mRNA several smaller mRNAs are present under normal conditions [56] , [57] . The terminator we map in RNA14 is located in the same region as the 3′ end of the shorter ( 1 kb ) mRNA ( Figure 2D ) . Dbp2p is a DEAD-box RNA helicase that plays a role in assembly of mRNP complexes and in RNA quality control [58] , [59] . The DBP2 gene is unique in yeast in having a long ( 1 kb ) intron localized toward the 3′ end of the coding region and previous work has shown that the gene is autoregulated through sequences in the intron [60] . In Figure 2D we show that depletion of Ysh1 leads to an increase in Pol II cross-linking downstream of the intron suggesting the presence of an upstream terminator . Consistent with this view polyadenylation sites have been mapped to this location in the Dbp2 intron [61]–[63] . Previous work has shown that Nrd1p is required for proper termination of ncRNAs like snoRNAs and CUTs as well as premature termination of coding transcripts of genes regulated by attenuation [6] , [7] , [9] , [19] , [28] , [35] , [64] . Here we show that depletion of Nrd1 in the presence of rap leads to readthrough transcription of a large number of these ncRNAs . Similar results have recently been obtained by a different protocol [65] . To map the position of termination we used an approach similar to that used to calculate a travelling ratio [66] or Escape Index [38] , [65] . Reads in 500 bp windows upstream and downstream of a fixed point were tallied and a ratio of reads in the downstream window to total reads in both windows were calculated for that point . This ratio was determined for each point in the genome on both strands for both control and rap treated cells . Subtracting the fraction of readthrough of the control data from the rap-treated data resulted in a Readthrough Index that allowed us to rank order the regions of the genome showing the highest level of Nrd1-dependent readthrough ( Table S2 ) . Once these regions were identified , we created plots of the difference between the reads in the control and reads with rap ( Figure 3A ) . Nrd1 depletion leads to both a reduction of Pol II upstream and an increase of Poll II downstream of apparent termination sites . This is consistent with previous reports that Pol II pauses prior to termination [67]–[70] . In the absence of Nrd1 , this apparent pause is eliminated and Pol II is able to transcribe through the termination site . Fitting this difference plot with a spline function allowed us to determine the point at which the difference plot crosses the X-axis . We have called this the termination site but this represents the center point of a narrow ( about 50 nt ) region over which Pol II is released from the template . In Figure 3A we show the difference plots for snoRNA snR34 and the antisense CUT to the LEO1 gene demonstrating how the termination sites were identified . We have carried out this analysis for 49 snoRNAs and for 144 CUTs showing the highest level of readthrough transcription ( Tables S2 and S3 ) . Comparing these termination sites to the sites determined by Pol II ChIP [65] reveals that 70% of the Schulz et al . sites are more than 100 nt downstream of our termination sites and 25% are more than 200 nt downstream ( Figure S5 ) . We do not know the reason for this systematic discrepancy but the PAR-CLIP analysis has greater resolution than Pol II ChIP and unlike ChIP gives strand-specificity . The metagene analysis of the ncRNAs for which we have mapped termination sites is shown in Figure 3B demonstrating a more significant increase in readthrough transcription for Nrd1-depletion when compared to the increase in readthrough transcription due to Ysh1 depletion ( Figure 2A ) . SnoRNA and CUT Nrd1-dependent readthrough extends over 500 bp and 1 kb , respectively . Examining DNA sequences upstream of the termination site for these sets of snoRNAs and CUTs revealed an interesting difference in the occurrence of Nrd1 and Nab3 binding motifs . The top hit in the MEME analysis [71] of the top 27 snoRNA upstream regions ( Figure 3C ) include the GUA[A/G] sequence previously identified as a Nrd1 binding site [8] , [31]–[34] , [72] . This analysis also identifies a loosely conserved sequence AACUA centered about seven nucleotides upstream of the GUA[A/G] sequence that has not previously been reported . The second most significant motif upstream of the snoRNA termination site was the UCUU sequence that binds Nab3 , consistent with the presence of a Nrd1-Nab3 heterodimer [29] , [31] . In sequences upstream of CUT terminators the most significant hit was UCUUG which contains the previously identified Nab3-binding sequence but with a downstream G as has been observed previously [8] , [33] , [34] . Among the 144 CUT terminators the Nrd1 binding motif was not present above background . Together , these observations suggest a significant difference in the manner of recognition of snoRNA and CUT terminators . In Figure 4 we show NET-seq [53] and PAR-CLIP data demonstrating that Pol II levels decline in the SNR13-TRS31 intergenic region . NET-seq reads for several peaks , indicated by bars , are reduced several-fold immediately downstream from the 3′-end of the SNR13 gene and more than 6-fold further than 100 nt downstream . PAR-CLIP reads from cells grown in the absence of rap decline at the same region reaching a minimum at about 150 nt downstream of the SNR13 3′-end . A similar correspondence between NET-seq and PAR-CLIP data is seen for other snoRNA genes ( Figure S6 ) . Taken together , these observations indicate that Pol II normally terminates in a Nrd1-dependent fashion in a narrow region located downstream of the snoRNA gene . The sequence from this region for SNR13 , located from 50–150 nt downstream , is shown below the figure and a series of runs of U residues is indicated in red . This is not unexpected as intergenic regions in S . cerevisiae are AT-rich . Sequences of termination regions of several antisense CUTs are shown below the SNR13 sequence . In these cases we also observe runs of U residues . This is unexpected as these termination regions fall within coding regions ( on the opposite strand ) which in general are not AT-rich . We propose that these U-rich regions constitute part of the non-poly ( A ) terminator . We estimate that SNR13 readthrough transcripts are present at less than one copy per cell in wild-type cells and this is consistent with the low level of readthrough transcripts seen in the absence of rap in the Northern blot shown in Figure 5A . Despite this low level of steady-state readthrough transcripts we observe significant cross-linking to Rpb2 in the region downstream of SNR13 ( Figure 5B ) even in the presence of Nrd1 . Presumably the RNA synthesized by these polymerases is rapidly degraded by the nuclear exosome . Figure 5A shows that snR13 readthrough transcripts detected by Northern blot or RT-PCR are increased more than 10-fold by 15 min and more than 100-fold by 30 min of Nrd1 depletion . However , depletion of Nrd1 results in only a 2–3-fold increase in PAR-CLIP signal for Rpb2 ( Figure 5B ) indicating that depletion of Nrd1 not only allows readthrough of ncRNA transcripts but also results in their stabilization by uncoupling degradation by the nuclear exosome . Our analysis of steady-state RNA levels is consistent with recently published 4tU-seq data [65] . Figure 5C shows the comparison of our data and that of Schulz et al for SNR13 and the upstream RPL27B gene and for the SNR62 and SNR9 genes . 4tU-seq data show a dramatic increase in SNR13 and SNR62 readthrough transcription compared to PAR-CLIP data . The most likely explanation for this difference is that 4tU labeling is not completely restricted to nascent transcripts . Labeling cells for six minutes with 4tU allows for multiple rounds of transcription especially for heavily transcribed genes like snoRNAs and ribosomal protein genes . The earliest synthesized transcripts are subject to processing as is clearly evident in the paucity of reads derived from the intron of RPL27B in the 4tU-seq compared to the PAR-CLIP ( Figure 5C ) and a number of other intron-containing genes ( Figure S7 ) . Depletion of Nrd1 does not affect all snoRNAs as is seen in Figure 5C where neither PAR-CLIP nor 4tU-seq detect appreciable readthrough transcription at SNR9 . A role for Sen1p in termination of both coding and ncRNA transcripts has been proposed [26] , [28] , [45] . Our current data show that depletion of Sen1 results in no change in the localization of Pol II at the 3′-ends of most protein-coding genes ( Figure 6A ) including those previously shown [45] to be dependent on Sen1 ( Figure S8 ) . Thus , despite our previous observation that Sen1 cross-links to these transcripts [8] , there is little evidence for a role for Sen1 in pA termination . Sen1 depletion does result in readthrough transcription at both snoRNAs and CUTS ( Figure 6B–C ) which is consistent with previous observations [15] , [19] , [28] . The pattern of Pol II readthrough in response to Sen1 depletion differs , however , from that of Nrd1 . Rather than an extended region of readthrough downstream of the terminator , Sen1 depletion often results in an increase just downstream by a few hundred nucleotides . This type of pattern is seen in Figure 6D for SNR47 and the CUT antisense to the COG8 gene and in Figure S9 for several protein-coding genes regulated by Nrd1-dependent attenuation . Figure 6E compares the percent readthrough downstream of all 266 non-pA terminators in Sen1 depleted and Nrd1 depleted cells . The percent readthrough after Nrd1 depletion is twice that for Sen1 depletion and these values are not dominated by a subset of terminators . This observation suggests that Sen1 may play a different function than Nrd1 at these genes and that loss of Sen1 results in less processive or more pause prone readthrough transcription .
Our data on Ysh1 depletion supports a two-step pause and release model for termination that invokes both allosteric and torpedo mechanisms [79] . The buildup of Pol II just downstream of the uncleaved pA site suggests that Pol II has paused but is deficient in the subsequent release step . This is consistent with a role for Ysh1 in providing the substrate for the Rat1 exonuclease that facilitates removal of Pol II from the template [80] . Ysh1 apparently is not required for the allosteric change that results in pausing downstream of the pA site as seen in the lack of readthrough at most pA sites . In contrast , mutation of the pA site [81] or mutations in RNA14 , RNA15 , PCF11 , CLP1 and GLC7 result in readthrough transcription more than 500 nt downstream [73] , [82] , [83] arguing for a role for these factors in the allosteric change that renders Pol II immobilized . Depletion of Ysh1 has also revealed several genes that are apparently regulated by premature termination through the pA pathway . RNA14 and DBP2 contain pA sites upstream from the 3′ pA site . Loss of Ysh1 leads to processive elongation ending at the downstream pA site indicating that Ysh1 is required for termination at these upstream sites . Why the putative allosteric step does not function efficiently at these upstream sites is not clear . Perhaps the phosphorylation pattern of the CTD at these sites precludes the assembly of factors like Pcf11 . Alternatively , chromatin structure modification induced by the Pol II that elongates through these terminators may suppress normal termination . Finally , it is possible that use of the downstream terminator by Pol II that ignores the upstream pA site creates a gene loop that favors termination at the downstream site . Nrd1 and Nab3 bind specific RNA sequences and act as sensors to detect non-pA terminator sequences in the nascent transcript . While a number of studies have characterized the short motifs recognized by Nrd1 and Nab3 the relative orientation and abundance of these sequences varies widely among non-pA terminators [8] , [31]–[34] , [49] , [65] . In this study we have localized termination downstream of 49 snoRNAs and 144 CUT transcripts . This has allowed a search for sequences that may define these two sets of terminators . We find that the most significant motif in this set of snoRNA terminators contains the GUA[A/G] motif previously identified for Nrd1 in the context of a longer sequence that indicates the possible involvement of another , unidentified RNA-binding protein . The second most significant motif is the Nab3 binding sequence UCUUG . In the case of the CUT terminators the only significant motif is the Nab3 binding sequence . Thus , it appears that while both Nrd1 and Nab3 binding contribute to recognition of snoRNA terminators Nab3 predominates for CUTs . This does not preclude Nrd1 from playing a critical role in termination at these CUTs . We have previously shown that Nrd1 is necessary for termination of the CUT antisense to FMP40 presumably by acting as an adaptor to couple the Nrd1-Nab3-Sen1 complex to the elongating Pol II through interaction between the Nrd1 CID and the CTD [15] . No other significant motifs were observed upstream of these terminators but we did observe that the sequences surrounding the termination site are U-rich and contain multiple runs of U residues . This is not unexpected for snoRNA downstream sequences as intergenic regions are AT-rich in S . cerevisiae . The U-rich sequences surrounding the termination sites ( Figure 4 ) of antisense CUTs are more significant as these sequences occur in the context of a coding region on the opposite strand . U-rich sequences have been shown to form unstable rU:dA base pairs [84] and in the hybrid binding site of Pol II transcribing the antisense CUT such an unstable hybrid sequence may increase the probability of termination [85] , [86] . In addition to its role in Pol II termination , our data highlight the role Nrd1 plays in the turnover of ncRNAs . Nrd1 depletion by anchor away leads to accumulation of readthrough transcripts in 4tU-seq to a much higher degree than observed by our Rpb2-HTB PAR-CLIP . We show here by Northern and RT-PCR that after 40 min of Nrd1 depletion the level of snR13 readthrough transcripts increases over 100-fold . However , after 40 minutes of depletion the amount of readthrough as observed by PAR-CLIP is less than 10-fold . This difference is likely due to an important role for Nrd1 in coupling termination to turnover of the completed transcript . Nrd1 is found in a complex with components of the TRAMP and exosome complexes [35] and previous studies have indicated that nrd1 mutants are deficient in RNA turnover [15] , [19] , [35] . Pulse labeling RNA with 4tU was recently used to analyze newly synthesized transcripts in a Nrd1 anchor away mutant and this data identified over 1500 Nrd1-dependent transcripts [65] , far more than the number of terminators we describe here . Part of the reason for this discrepancy is that the six-minute pulse used in the 4tU-seq protocol is substantially longer than the time needed to synthesize short RNAs . Thus , 4tU-seq over-estimates the effect on termination and may ascribe termination functions to cases where the role in turnover is independent of termination . Depletion of the DNA/RNA helicase Sen1 has allowed us to place this factor squarely in the Nrd1-Nab3 non-pA termination pathway . We observe readthrough transcription downstream of both snoRNAs and CUTs but not downstream of pA sites . Pol II does not progress far downstream after Sen1 depletion indicating that this factor may act after the allosteric change has occurred . This is consistent with Sen1 acting in place of Rat1 to provide the activity that dislodges the paused Pol II [46] and supports previous work showing the difference in the requirement for these factors in pA and non-pA termination [24] . The results presented here are summarized in a model shown in Figure 7 . We have been able to demonstrate a difference in the global effect of depleting different components of the yeast termination machinery . These results have been interpreted in the context of a two state model for termination in which terminator sequences in the nascent RNA trigger assembly of protein complexes that signal to the elongating Pol II to decrease its elongation rate and processivity . These anti-processivity factors include Pcf11 for pA terminators and Nrd1 for non-pA terminators . Loss of either of these factors leads to runaway processive elongation . In contrast , depletion of Ysh1 or Sen1 does not block the transition to non-processive elongation but does result in a defect in removal of Pol II from the template . For Sen1 this likely occurs through loss of a rho-like function [44] , [46] while Ysh1 depletion leads to a lack of cleavage at the pA site thereby limiting access of Rat1 and preventing the torpedo mechanism of Pol II release [80] . The data presented here demonstrate that PAR-CLIP analysis of Pol II in living yeast cells can define the site of Pol II termination regions and when coupled with depletion of termination factors offers an avenue for examining the roles of different factors in termination genome-wide . Such experiments will illuminate the variety of mechanisms the cell uses to prevent Pol II from transcribing beyond genetically determined 3′ boundaries .
Yeast strains for anchor away were constructed using the parental strain HHY168 [50] ( Euroscarf #Y40343 ) . RPB2 was C-terminally tagged with an HTB ( 6×HIS , TEV , and Biotin ) tag as described previously [8] , [87] . NRD1 , YSH1 , and SEN1 were C-terminally tagged with FRB and FRBGFP as described in Haruki et al . [50] . The SEN1 promotor was replaced with the GAL1 promotor using a cassette from pFA6a-kanMX6-PGAL1-HBH [87] . All strains used in these experiments are listed in Table S4 . Cells were grown in Complete Synthetic Media ( CSM ) supplemented with 2% glucose or galactose and 40 mg/l adenine . For RNA analysis , cells were seeded to an OD600 of 0 . 1 from a 5 ml or 50 ml overnight culture . The cells were then incubated at 30°C to an OD600 of 1 . 1 µg/ml of rap ( LC laboratories ) was then added to the cultures and the cells were allowed to grow for the indicated times in the experiment . For growth curves , a BioTek Infinity 2 ( BioTek ) was used to incubate a 24 well plate containing CSM media with and without 1 µg/ml rap ( added after 6 hrs of growth ) with orbital shaking ( Slow , 1 mm amplitude ) . Cells were diluted to an OD600 of 0 . 1 per well and an OD600 was measured every 10 min for 24 hours . Total RNA was extracted from yeast with hot acid phenol as previously described [8] . Strand specific northern blots were done as described previously [88] with the following modifications . Ultrahyb ( Invitrogen ) was used instead of the described hybridization buffer . Real-time PCR analysis was done on a CFX96 instrument ( Biorad ) in triplicate as described previously [49] . Primers used can be found in Table S5 . The strains containing NRD1-FRBGFP , YSH1-FRBGFP , or GALpSEN1-FRBGFP , RPB3-TAGRFP were grown to an OD600 of 1 at 30°C . Live cell imaging was done on a Deltavision microscope ( Applied Precision ) as described previously [89] with the following modification: 1 µg/ml rapamyacin was added to the agarose/media pad . Galactose was substituted for glucose with the GALpSEN1-FRBGFP strain . Pictures were taken of GFP at 0 mins and 45 mins post rap exposure and 45 mins for RFP . Cells were grown overnight 30°C in 50 ml of YPD ( Yeast Extract Peptone with 2% glucose ) media . Four 500 ml flasks of sterile cross-linking media ( CSM–Ura supplemented with 2% glucose , 40 mg/l adenine , 60 µM uracil , and 1 µM biotin were seeded to an OD600 of 0 . 1 and incubated at 30°C until an OD600 of 1 . Crosslinking was carried out by the protocol described in Figure S1B . 4-thiouracil ( 4tU ) was added to a final concentration of 4 mM and incubated at 30°C for 15 min . 1 µg/ml of rapamycin was then added to the cells and incubated at 30°C for 30 min . All four 500 ml cultures were then pooled into one 2 l beaker and irradiated from a distance of two cm with 365 nm UV ( ∼1 W/cm2 ) from an LC-L5 LED UV Lamp ( Hamamatsu ) for 15 min with gentle stirring . The top ∼2 mm of the culture appears luminescent indicating that the UV light does not penetrate further into the culture . Assuming an even mixture of the culture we calculate that the average cell is cross-linked for only ∼10 seconds within the 15 min crosslinking period . The cultures were then filtered through 0 . 45 micron nitrocellulose filters ( Millipore ) and cells were scraped into 5 ml Buffer 1 ( 300 mM NaCl , 0 . 5% NP-40 , 50 mM NaPO4 pH 7 . 2 , 10 mM immidazole , 6 M Guanidine HCl , Protease inhibitor Cocktail VII [RPI] ) and frozen in liquid nitrogen . HTB tagged Rpb2 purification was adapted from our previously published protocol [49] . Cells were lysed in liquid nitrogen using a SamplePrep 6870 freezer mill ( Spex ) with 15 cycles per second of cracking for 15 cycles of 1 min with 2 min of cooling between each cycle . Lysates were then incubated with 1 ml of Ni-NTA agarose ( Qiagen ) , which was equilibrated in buffer 1 for 2 hours at room temperature . The Ni-NTA agarose was then added to an empty 10 mL plastic column ( Biorad ) and washed with 20 ml of Buffer 1 followed by 10 ml of Buffer 2 ( 20 mM NaPO4 pH 7 . 2 , 300 mM NaCl , 0 . 5% NP40 , 10 mM imidazole , 4 M Urea ) . The protein was eluted off of the Ni-NTA agarose with 5 ml Buffer 2+250 mM immidazole and into 5 ml Buffer 2+120 µl Protease inhibitor Cocktail VII+100 µl of hydrophilic streptavidin magnetic beads ( NEB ) . The slurry was allowed to incubate for 4 hours at room temperature . The strepavadin magnetic beads were then resuspended in 500 ul of Buffer 3 ( 50 mM Tris pH 7 . 4 , 200 mM NaCl , 4 M Urea ) , transferred to a 1 . 5 ml siliconized tube and washed with 3×1 ml of Buffer 3 followed by 3×1 ml of T1 Buffer ( 50 mM Tris pH 7 . 4 , 150 mM NaCl , 2 mM EDTA ) . Beads were resuspended in 200 µl of T1 buffer . 0 . 15 U/µl of Rnase T1 ( Fermentas ) was added to the bead slurry and allowed to incubate at 25°C for 15 min . The beads were then washed 3×1 ml T1 wash buffer ( 50 mM Tris pH 7 . 4 , 500 mM NaCl , 1% NP-40 , 0 . 5% Na deoxycholate ) followed by 3×1 ml washes in PNK Buffer ( 50 mM Tris pH 7 . 2 , 50 mM NaCl , 10 mM MgCl2 ) . The beads were resuspended in 200 µl of TSAP reaction solution ( 0 . 15 U/ul Thermosensitive Alkaline Phosphatase [TSAP , Promega] , 1 U/µl SupeRase Inhibitor [Invitrogen] , 1 mM DTT in PNK Buffer ) and incubated at 37°C for 30 min . The beads were washed 1×1 ml Buffer 3 , 2×1 ml T1 Wash Buffer , and 3×1 ml PNK Buffer . Streptavadin beads were resuspended in 200 ul of PNK reaction solution ( 1 U/µl T4 PNK [NEB] , 32P γ-ATP , 5 mM DTT in PNK Buffer ) and incubated at 37°C for 30 min . 1 mM of cold ATP was added to the reaction and incubated at 37°C for 10 min . The beads were washed 1×1 ml Buffer 3 , 1×1 ml T1 wash , 3×1 ml T4 RNA Ligase Buffer ( 50 mM Tris pH 7 . 4 , 10 mM MgCl2 ) . The beads were resuspended in 44 µl of T4 RNA Ligase 2 reaction solution ( 25% PEG 8000 , 5 µM 3′ adaptor [AppAGATCGGAAGAGCACACGTCTddC , IDT] , 10 U/µl T4 RNA Ligase 2 , truncated K227Q [NEB] , 2 U/µl RNase Inhibitor [Invitrogen] , 1 mM DTT in RNA Ligase Buffer ) and incubated at 25°C for 4 hours . The beads were then washed with 1×1 ml Buffer 3 and 5×1 ml T4 RNA Ligase Buffer . The beads were resuspended in 50 ul of T4 RNA Ligase reaction solution ( 5 uM 5′ Adaptor [GUUCAGAGUUCUACAGUCCGACGAUC , IDT] , 1 . 2 U/µl RNase Inhibitor , 1 mM ATP , 0 . 5 U/µl T4 RNA Ligase [NEB] , 1 mM DTT in RNA Ligase Buffer ) and incubated at 16°C overnight . The beads were then washed 5×1 ml Proteinase K Buffer ( 100 mM Tris pH 7 . 4 , 150 mM NaCl , 12 . 5 mM EDTA ) . The beads were resuspended in 200 µl of Proteinase K Buffer +2% SDS and 12 µl of Proteinase K ( NEB ) . The suspension was incubated for 30 min at 37°C at which point the supernatant was removed and saved and the beads were resuspended in 200 µl of Proteinase K Buffer +2% SDS and 12 µl of Proteinase K . The beads were then incubated 30 min at 37°C and the supernatants were pooled . The RNA was recovered by acid phenol/chloroform extraction followed by two serial ethanol precipitations . The resulting pellet was allowed to dry and resuspended in 15 µl of Nuclease-Free Water [Invitrogen] . The RNA was split into 5 µL aliquots and frozen at −80°C for storage . The Reverse Transcription , PCR , and gel extraction of the cDNA Library were carried out as described previously [49] with the following modifications . For reverse transcription , the primer used was AGACGTGTGCTCTTCCGATCT ( IDT ) . For PCR analysis , biological repeats were multiplexed with the following primers . The forward primer was AATGATACGGCGACCACCGAGATCTACACGTTGAGAGTTCTACAGTCCG*A ( where a * denotes and phosphorothioate bond ) . The first reverse primer was CAAGCAGAAGACGGCATACGAGATATTGGCGTGACTGGAGTTCAGACGTGTGCTCTTCGGATC*T for the first biological repeat and the second reverse primer was CAAGCAGAAGACGGCATACGAGATTACAAGGTGACTGGAGTTCAGACGTGTGCTCTTCCGATC*T for the second biological repeat ( IDT ) . Sequencing and demultiplexing was done at UC Riverside on an Illumina HiSeq . Trimming of the resulting sequences as described previously [49] . Briefly , raw reads were trimmed of the 3′ adaptor using a wrapper for R-bioconductor [90] developed by Sarah Wheelan . Raw trimmed reads were condensed ( defined as no more than one of the same exact sequence ) to eliminate any PCR artifacts within the data . Bowtie 1 . 0 . 0 [91] was run with the following arguments ( -y –best -v 2 ) and aligned to the SacCer3 ( R64 ) genome . Reads mapping to tRNA genes were removed as these likely represent artifactual binding during the affinity purification steps . Reads were then converted to a wig format where each read was multiplied by a factor that was defined as the total number of reads aligned per 107 reads . PAR-CLIP datasets have been submitted to GEO with the accession number GSE56435 . To calculate the best polyA sites per gene we used the sites provided by Moqtaderi et al . [63] . We set a hard cutoff of at least 200 raw reads . If , however , there were multiple very strong polyA sites within 200 base pairs of each other , we calculated the read ratio between the two most significant polyA sites . If the read ratio was less than 0 . 4 , we ignored them . Global readthrough percentage was calculated for each point in the genome as the percentage readthrough ( reads downstream 500 bp/reads downstream 500 bp + reads upstream 500 bp ) in the treated sample minus control sample . All points with a percentage readthrough of greater than 10% , at least a total of 1000 reads in either sample , and greater reads in the treatment vs the control were extracted . The list was further refined by taking the difference of every point 2 kb around non-overlapping termination sites ( approx 500 total ) , graphing the region , and then fitting a spline function over a 25 bp window to find the best point of inflection . These points of inflection were then used as a focal point for the meta gene analysis . | Transcription termination is an important regulatory event for both non-coding and coding transcripts . Using high-throughput sequencing , we have mapped RNA Polymerase II's position in the genome after depletion of termination factors from the nucleus . We found that depletion of Ysh1 and Sen1 cause build up of polymerase directly downstream of coding and non-coding genes , respectively . Depletion of Nrd1 causes an increase in polymerase that is distributed up to 1 , 000 bases downstream of non-coding genes . The depletion of Nrd1 helped us to identify more than 250 unique termination regions for non-coding RNAs . Within this set of newly identified non-coding termination regions , we are further able to classify them based on sequence motif similarities , suggesting a functional role for different terminator motifs . The role of these factors in transcriptional termination of coding and/or non-coding transcripts can be inferred from the effect of polymerase's position downstream of given termination sites . This method of depletion and sequencing can be used to further elucidate other factors whose importance to transcription has yet to be determined . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"molecular",
"machines",
"molecular",
"complexes",
"genetic",
"elements",
"gene",
"expression",
"genetics",
"biology",
"and",
"life",
"sciences",
"molecular",
"genetics",
"molecular",
"biology",
"techniques",
"molecular",
"biology"
] | 2014 | Genome-Wide Mapping of Yeast RNA Polymerase II Termination |
Cells use surface receptors to estimate concentrations of external ligands . Limits on the accuracy of such estimations have been well studied for pairs of ligand and receptor species . However , the environment typically contains many ligands , which can bind to the same receptors with different affinities , resulting in cross-talk . In traditional rate models , such cross-talk prevents accurate inference of concentrations of individual ligands . In contrast , here we show that knowing the precise timing sequence of stochastic binding and unbinding events allows one receptor to provide information about multiple ligands simultaneously and with a high accuracy . We show that such high-accuracy estimation of multiple concentrations can be realized with simple structural modifications of the familiar kinetic proofreading biochemical network diagram . We give two specific examples of such modifications . We argue that structural and functional features of real cellular biochemical sensory networks in immune cells , such as feedforward and feedback loops or ligand antagonism , sometimes can be understood as solutions to the accurate multi-ligand estimation problem .
Cells obtain information about their environment by capturing ligand molecules with receptors on their surface and estimating the ligand concentration from the receptor activity . Limits on the accuracy of such estimation have been a subject of interest since the seminal work of Berg and Purcell [1] , with several substantial extensions found recently [2–8] . Most of these assume one ligand species coupled to one receptor species , and the actual detection in most of these models is rather simple , involving counting the number or the duration of binding / unbinding events over a specific period of time . However , cells carry many types of receptors and have many species of ligands around them . The same ligand can bind to many receptors , albeit with different affinities , and vice versa . This is commonly referred to as cross-talk . At the same time , real cellular sensory systems are incredibly complex , involving many dozens of identified biochemical species downstream of a typical receptor [9] . Functionally many of such signaling motifs are probably related to solving the cross-talk problem [10 , 11] , and are a topic of active research . In traditional deterministic chemical kinetics , one cannot estimate concentrations of more ligands than there are receptor types . Further , even a weak cross-talk prevents determination of concentrations of individual chemical species since the activity of a receptor is a function of a weighted sum of concentrations of all ligands that can bind to it . In contrast , here we argue that , with cross-talk , concentration of more than one chemical species can be inferred from the activity of one receptor , provided that the stochastic temporal sequence of receptor binding and unbinding events is accessible instead of its mean occupancy . This is an important departure from the traditional view of cellular signaling that posits as many receptor types as there are ligand concentrations to be estimated . Indeed , previous works studying temporal sequences of receptor occupancy for ligand detection [11] and concentration estimation [5 , 13 , 12] have only considered the detection/estimation of a single ligand present in a mixture . We argue that the receptor occupancy sequence contains much more information about the mixture . In fact , based on the maximum likelihood techniques , which have been used previously to study receptor occupancy , we show that all components of the ligand mixture can be estimated by just one receptor , at least in principle . This surprising result can be understood by noting that a typical duration of time that a ligand remains bound to the receptors depends on its unbinding rate . Thus observing the statistics of the receptor’s unbound time durations allows estimation of a weighted average of all chemical species that interact with it [5] . Then the statistics of the bound time durations tells how common each ligand is . The result is very general and independent on the choice of a downstream biochemical kinetics scheme that actually performs the estimation . In this article , we derive the result for the simplest problem of this class , namely one receptor interacting with two ligand species . While the exact solution of the inference problem for finding both ligand concentrations is hard to implement using common biochemical machinery , we show that an accurate approximation is possible using simple extensions of the familiar kinetic proofreading mechanism [14 , 15] . We identify examples of such motifs implementing such estimation of multiple concentrations in signaling networks found downstream of many immune receptors [9] , arguing that real biological systems may be implementing such multivariate concentration sensing . The kinetic schemes that we analyze detect rare ligands more accurately than a simple kinetic proofreading does , and we argue that the involved biochemical computation can explain properties like ligand antagonism , commonly observed in receptor signaling . Overall , these different arguments support our main idea , that the temporal sequence of binding and unbinding on a single receptor can provide an accurate estimate of the concentration of multiple ligands that bind to the receptor , and that the involved calculations can be performed reliably by known biochemical networks .
Consider a single receptor interacting with a cognate and a non-cognate ligand ( Fig 1 ) that have the concentrations cc and cnc , respectively . The binding rate of the ligands to the receptor are kc and knc . The binding rates are diffusion limited and hence kc∼knc . It is the unbinding or off-rates , rc and rnc , that distinguish the two ligands: rnc > rc , and a cognate molecule typically stays bound for longer . The binding and unbinding rates ( kα’s and rα’s ) are fixed and can be assumed known for each receptor-ligand pair . Thus we are interested in the estimation of the ligand concentrations only , cc and cnc . Following Ref . [5] , we estimate cc and cnc from the time-series of binding , { t i b } , and unbinding , { t i u } , events of a total duration T using Maximum Likelihood techniques , paralleling a recent similar independent discussion , which focused on detection of a single ligand concentration [12] . The numbers of binding and unbinding events are different by , at most , one , which is insignificant since we consider T → ∞ . Thus without loss of generality , we assume that the first event was a binding event at t 1 b , and the last one was the unbinding at t n u . We write the probability distribution of observing the sequence { t 1 b , t 1 u , … , t n b , t n u } , or alternatively the sequence of binding and unbinding intervals τ i b = t i u - t i b , and τ i u = t i + 1 b - t i u: P ≡ P ( { τ i b , τ i u } | c c , c nc ) = 1 Z ∏ i = 1 n e - τ i u ( k c c c + k nc c nc ) k c c c r c e - τ i b r c + k nc c nc r nc e - τ i b r nc . ( 1 ) Here the first term under the product sign is the probability of the receptor staying unbound for τ i u . The second term , which from now on we denote by D ( c c , c nc , τ i b ) , is proportional to the probability of staying bound for τ i b . D ( c c , c nc , τ i b ) has contributions from binding events from both the cognate and the noncognate ligands , with odds of cc and cnc , respectively . Finally , Z is the normalization , Z = ∑ P ( { τ i b , τ i u } | c c , c nc ) , ( 2 ) where the sum is over all sequences of duration T and n binding-unbinding events . Note that here we define τ n u = t 1 b + ( T - t n u ) , so that the n’th unbound interval includes the “incomplete” unbound intervals before the first binding and after the last unbinding . The log-likelihood of cc and cnc is the logarithm of P , Eq ( 1 ) . Taking the derivatives of the log-likelihood w . r . t . cc and cnc and setting them to zero gives the Maximum Likelihood ( ML ) equations for the concentrations . Denoting by T u = ∑ i = 1 n τ i u , the total time the receptor is unbound , these ML equations are ( see Methods for the derivation ) : - k c T u + ∑ i = 1 n k c r c e - τ i b r c D ( c c * , c nc * , τ i b ) = 0 , ( 3 ) - k nc T u + ∑ i = 1 n k nc r nc e - τ i b r nc D ( c c * , c nc * , τ i b ) = 0 , ( 4 ) where c c * and c nc * denotes the ML solution . Multiplying Eqs ( 3 ) and ( 4 ) by c c * and c nc * , respectively , and adding them gives k c c c * + k nc c nc * = n T u . ( 5 ) As in Ref . [5] , the total on-rate ( the weighted average of the external concentrations ) is determined only by the average duration of the unbound interval , ( n/Tu ) −1 , because no binding is possible when the receptor is already bound . For the special case of kc ≈ knc ≈ k ( for ligands with binding rate determined by diffusion ) , Eq ( 5 ) determines the maximum likelihood estimate of the sum of the two concentrations , similar to the result in Ref . [5 , 12]: c tot * = c c * + c nc * = n T u k . ( 6 ) This shows that the estimates are negatively correlated . For general ki’s , a weighted sum of the concentrations is determined , but the negative correlation persists . To get the individual concentrations , we need to solve the ML equations Eqs ( 3 ) and ( 4 ) . In general , they can only be solved numerically . However , as all ML estimators , they are unbiased to the leading order in n ( we verified this numerically ) . The standard errors of the ML estimates can be obtained by inverting the Hessian matrix , ∂ 2 log P ∂ c α ∂ c β c c * , c nc * = ∑ i = 1 n - 1 D ( c c , c nc , τ i b ) 2 × k c 2 r c 2 e - 2 τ i b r c k c k nc r c r nc e - τ i b ( r c + r nc ) k c k nc r c r nc e - τ i b ( r c + r nc ) k nc 2 r nc 2 e - 2 τ i b r nc , ( 7 ) where greek indices stand for {c , nc} . Each term in the Hessian matrix is a sum of n numbers , each smaller than zero . The inverse of ∂ 2 log P ∂ c α ∂ c β , which scales as ∝ 1/n , sets the minimum variance of any unbiased estimator according to the Cramer-Rao bound . It has straightforward analytical approximations in various regimes . For example , when the noncognate ligand is almost absent ( cc/cnc ≫ 1 ) , and its few molecules do not bind for long ( rc/rnc ≪ 1 ) , one gets σ 2 ( c c * ) ≈ ( ∂ 2 log P / ∂ c c 2 ) c c = c c * - 1 ≈ 1 / n , matching the accuracy of sensing one ligand with one receptor [5] . A regime relevant for detection of a rare , but highly specific ligand [11 , 12 , 16] can be investigated as well . For now , we focus on how the receptor estimates ( rather than detects ) concentrations of both ligands simultaneously , which requires us to explore the full range of on- and off-rates . The estimates of the concentration cc and cnc are obtained by numerically solving ML equations , Eqs ( 3 ) and ( 4 ) . We study the variability of these ML estimators in terms of their posterior variances . Notice that these posterior variances scale as 1/n , so we define the error of the ML estimators , E , as the squared coefficient of variation times the number of binding-unbinding events , n . Hence , we have , E c = n σ 2 ( c c * ) / c c 2 and E nc = n σ 2 ( c nc * ) / c nc 2 for cognate and non-cognate ligands , respectively . These quantities have a finite limit at n → ∞ . Specifically , E = 1 is the accuracy that a receptor that binds only a single ligand can obtain [5] . Thus Ec and Enc compare the performance of our multi-ligand ML estimator to the limit achievable by a single ligand ML estimator . We show log10 Ec and log10 Enc for different concentrations and off-rates in Fig 2 . If the two ligands are readily distinguishable , rc ≪ rnc , then the ligand with the larger concentration has E ∼ 1 . When cc ∼ cnc , Ei ∼ 4…5 , and it grows to 10…30 for a ligand with a very small relative concentration . Emphasizing the importance of the time scale separation , E > 100 if the ligands are hard to distinguish , rc ∼ rnc . Here the correlation coefficient ρ of the two estimates reaches −1 because the same binding event can be attributed to either ligand . Finally , the asymmetry of the plots w . r . t . the exchange of cc and cnc is because the cognate ligand can generate short binding events , while long events from the noncognate ligand are exponentially unlikely . In summary , it is possible to infer two ligand concentrations from one receptor , with the error of only 1…10 times larger than for ligand-receptor pairs with no cross talk , as long as the two off-rates are substantially different . This complements the findings of [12] that a single concentration can be inferred from a time series of “on” and “off” events in a background of noncognate bindings using Maximum Likelihood estimation . We have verified that the analytical expression for the estimation error derived in Ref . [12] for a single cognate ligand matches our numerical results ( see Methods ) . It is not clear if there exist biochemical networks that can solve the ML equations , Eqs ( 3 ) and ( 4 ) , exactly . Luckily , an approximate solution exists . Note that most of the long binding events come from the cognate ligand since the noncognate one dissociates faster . Defining long events as τ i b ≥ T c and using Eq ( 5 ) , we rewrite Eq ( 3 ) as k c n k c c c * + k nc c nc * = ∑ τ i b ≥ T c + ∑ τ i b < T c k c r c e - τ i b r c D ( c c * , c nc * , τ i b ) ( 8 ) Assuming that all long events are cognate , Tc ≫ 1/rnc , gives k c n k c c c a + k nc c nc a = n l c c a + ∑ τ i b < T c k c r c e - τ i b r c D ( c c a , c nc a , τ i b ) , ( 9 ) where nl is the number of long events , and the superscript “a” stands for the approximate solution . If further T is long enough so that there are many short events , and a single binding duration hardly affects k c * , then the sum in Eq ( 9 ) can be approximated by the expectation value: n k c c c a + k nc c nc a = n l k c c c a + ( n - n l ) ∫ 0 T c r c e - τ b r c P ( τ b | c c a , c nc a ) d τ b D ( c c a , c nc a , τ b ) , ( 10 ) where P ( τ b | c c a , c nc a ) is the probability of observing a binding event of duration τb for the given binding rates , P ( τ b | c c a , c nc a ) = D ( c c a , c nc a , τ b ) k c c c a + k nc c nc a . ( 11 ) Plugging Eq ( 11 ) into Eq ( 10 ) , we obtain 1 k c c c a + k nc c nc a = n l n k c c c a + 1 - n l n 1 - e - r c T c k c c c a + k nc c nc a . ( 12 ) Finally , since nl ≪ n , using Eq ( 5 ) , we get ( see Methods for a detailed derivation ) : c c a = 1 k c n l T u e r c T c , ( 13 ) c nc a = 1 k nc n T u - n l T u e r c T c . ( 14 ) In other words , the approximate cognate ligand concentration is proportional to the number of long events . We can estimate the bias and the variance of c c a and c nc a in a limiting case . If rc and rnc are not very different from each other , then one needs to focus on extremely long events in order to identify cognate bindings . This is only possible if Tc is much larger than the inverse of both of the unbinding rates , T c ≫ { r nc - 1 , r c - 1 } . Large Tc ensures that the long binding events get no or minimal contribution from non-cognate ligands . However , since the time for which the receptor stays bound is exponentially distributed , under this condition , the number of “long” events ( such that τb > Tc ) would be very small , nl ≪ n . Thus most of the variance of c c a and c nc a in Eqs ( 13 ) and ( 14 ) comes from the variability of nl , but not Tu ( since Tu ∝ n ) . Thus we write 〈 c c a 〉 ≈ 〈 n l 〉 〈 T u 〉 e r c T c k c . Further , the individual unbound periods are independent , so that 〈Tu〉 = n〈τu〉 = n/ ( kccc + knccnc ) ( notice the use of c rather than ca here ) . Further , 〈 n l 〉 = n P ( τ b > T c ) = n ( k c c c + k nc c nc ) ( k c c c e - r c T c + k nc c nc e - r nc T c ) . Combining these expressions , we get 〈 c c a 〉 ≈ c c + k nc c nc k c e - ( r nc - r c ) T c . ( 15 ) Thus for large Tc , the bias of the approximate estimator , k nc c nc k c e - ( r nc - r c ) T c , grows with the relative number of noncognate long bindings events . In turn , the latter is proportional to cnc , but decreases exponentially with Tc . Within the same approximation , the variance of the estimator is given by σ 2 ( c c a ) ≈ σ 2 ( n l ) 〈 T u 〉 2 e 2 r c T c k c 2 . However , long binding events are rare , independent of each other , and hence obey the Poisson statistics . Thus σ2 ( nl ) = 〈nl〉 , so that σ 2 ( c c a ) ≈ 〈 c c a 〉 c c + k nc c nc / k c n e r c T c . ( 16 ) The variance obviously grows with Tc . Knowing that the bias and the variance of the approximation change in opposite directions with Tc , we can find the optimal cutoff ( T * c ) by minimizing the overall error . We define such error L as the sum of the variance and the squared bias of the estimator , i . e . , L c = σ 2 ( c c a ) + c c - 〈 c c a 〉 2 , ( 17 ) L nc = σ 2 ( c nc a ) + c nc - 〈 c nc a 〉 2 . ( 18 ) The optimal cutoff is obtained by minimizing Lc or , in other words , solving the bias-variance tradeoff: T * c = arg min T c L c . ( 19 ) Near the optimal cutoff , the bias is small , and we use cc instead of c c a for the variance of the estimator , Eq ( 16 ) . Then solving Eq ( 19 ) gives: T * c = 1 ( 2 r nc - r c ) log 2 T u r nc r c - 1 k nc 2 c nc 2 k c c c . ( 20 ) Plugging this into Eqs ( 15 ) and ( 16 ) , we get the minimal error of the estimator , which we omit here for brevity . The optimal cutoff is proportional to 1/rnc if rnc ≫ rc , and it grows with rc , allowing for better disambiguation of cognate and noncognate events . Crucially , the off-rates are dictated by the ligand identities . In contrast , the concentrations , cc and cnc , are what the receptors measures . Therefore , it is encouraging that T * c depends only logarithmically on the concentrations ( and also on the duration of the measurement , Tu ) . Thus even if Tc is fixed as T * c at some fixed values of cc , cnc , it remains near-optimal for a broad range of external concentrations . To illustrate this , we use T c = T * c ( k c c c = k nc c nc = 1 / 2 ) ≡ T 0 and analyze the quality of the approximation in Fig 3 , where we plot the ratio L c ( T 0 ) / σ c c 2 and L nc ( T 0 ) / σ c nc 2 . Notice that σ c c 2 and σ c nc 2 , the variances of the exact ML estimators , are proportional to Ec and Enc , respectively . Since ML estimators are unbiased , the ratios L c ( T 0 ) / σ c c 2 and L nc ( T 0 ) / σ c nc 2 compare the errors of the approximate solution to the errors Ec and Enc . Since these ratios approach 1 when rc/rnc → 0 ( specifically , for rc/rnc = 0 . 1 , L c ( T 0 ) / σ k c 2 ≈ 1 . 47 , and L nc ( T 0 ) / σ k nc 2 ≈ 1 . 21 ) , we conclude that the approximation is accurate even at fixed Tc = T0 when its assumptions are satisfied . This happens even though T * c depends on cc and cnc , but apparently the approximate estimates are as good as the ML estimates even at fixed Tc = T0 and work well for a large range of concentration ratios . This is important , as the molecular mechanisms that sets the delays in the cell does not need to be modified for different ligand concentrations . In contrast , when the ligands are nearly indistinguishable ( rc/rnc ∼ 1 ) , both L c ( T 0 ) / σ k c 2 ∼ 100 and L nc ( T 0 ) / σ k nc 2 ∼ 100 , but here one would not use one receptor to estimate two concentrations since even the ML solution is bad ( cf . Fig 2 ) . Note also that both Lc and Lnc are smaller for rc ∼ rnc if cc ≫ cnc . This is because our main assumption ( that almost all long events are cognate ) holds better when cognate ligands dominate . Finally , the correlation coefficient between the approximate estimates , ρa ( right panel ) reaches -1 earlier than in Fig 2 . This is a direct consequence of Eqs ( 13 ) and ( 14 ) . The approximate solution can be computed by cells using the well-known kinetic proofreading ( KPR ) mechanism [14 , 15 , 17 , 18] . In the simplest model of KPR [19] , intermediate states between an inactive and an active state of a receptor delay the activation . Thus bound ligands can dissociate before the receptor activates , at which point it quickly reverts to the inactive state . Since rc < rnc , cognate ligands dominate among bindings that persist to activation . The resulting increase in specificity in various KPR schemes has led to their exploration in the context of detection of rare ligands [11 , 12 , 16 , 18] . Instead , here we analyze their ability to measure concentrations of both ligands simultaneously . We first consider the case where both the cognate and the non-cognate ligand concentration are comparable , cc ∼ cnc and the dissociation rates are distinct , rc ≪ rnc . In the following sections , we explore another case , cc ≪ cnc and rc ≲ rnc , a situation common in immunology . Consider a biochemical network in Fig 4 ( a ) : the receptor , R , activates two messenger molecules , A and B . The former is activated with the rate kA only if the receptor stays bound for longer than a certain Tc ( with the delay achieved using the KPR intermediate states ) . The latter is activated with the rate kB whenever the receptor is bound . The molecules deactivate with the rates rA and rB , respectively , and all activations/deactivations are first-order reactions . Then the mean concentrations of the messenger molecules are ( see Methods ) : A ¯ = k c c c / r c e - r c T c + k nc c nc / r nc e - r nc T c 1 + k c c c / r c + k nc c nc / r nc k A r A , ( 21 ) B ¯ = k c c c / r c + k nc c nc / r nc 1 + k c c c / r c + k nc c nc / r nc k B r B . ( 22 ) Assuming again that most bindings longer than Tc are cognate ( Tc ≫ 1/rnc ) , Eq ( 21 ) , can be written as: A ¯ = k c c c / r c e - r c T c 1 + k c c c / r c + k nc c nc / r nc k A r A . ( 23 ) Further , it is easy to see that Eq ( 22 ) can be rewritten as: k c c c r c + k nc c nc r nc = B ¯ k B / r B - B ¯ . ( 24 ) Now solving Eqs ( 23 ) and ( 24 ) for the on-rates , we get c c = A ¯ e r c T c r c r A k c k A 1 + B ¯ k B / r B - B ¯ , ( 25 ) c nc = r nc k nc B ¯ k B / r B - B ¯ - A ¯ e r c T c r A k A 1 + B ¯ k B / r B - B ¯ . ( 26 ) The corrections of the form B ¯ / ( k B / r B - B ¯ ) appear because bindings only happen to unbound receptors , as emphasized in Ref . [5] . However , these nonlinear relations are still hard to implement with simple biochemical components . We solve this by further assuming ϵ = B ¯ / ( k B / r B ) ≪ 1 , which is true if the receptor is mostly unbound , which happens at low concentrations . This gives c c KPR ≈ A ¯ e r c T c r c r A k c k A , ( 27 ) c nc KPR ≈ r nc k nc r B B ¯ k B - A ¯ e r c T c r A k A . ( 28 ) These equations are analogous to Eqs ( 13 ) and ( 14 ) . They are easy to realize biochemically ( cf . Fig 4 ( a ) ) : cc is related to the concentration of the proofread species A by a rescaling , and cnc comes from subtracting rescaled versions of B and A from each other . The subtraction can be done by the third species C , activated by B and suppressed by A . Since ϵ ≪ 1 , then A ¯ and B ¯ are small , and many such activation-suppression schemes are linearized as the subtraction [8] . Note that such incoherent feedforward loops ( the receptor activates A and B , which then affect C incoherently by suppressing and activating it , respectively ) are ubiquitous in cellular networks downstream of receptors [9] . The bias of c c a and c nc a due to long , but noncognate binding events , Eq ( 15 ) , carries over to c c KPR and c nc KPR . However , there is an additional contribution since the time to traverse the intermediate states in KPR schemes with multiple intermediate steps is random . Thus Tc has some variance σ T c 2 [19 , 20] . This variability changes the rate of occurrence of long biding events , but they are still rare , nearly independent , and Poisson-distributed . Denoting by 〈⋅〉 the averaging at a fixed Tc , and by · ¯ the averaging over Tc , we get 〈 n l 〉 ¯ = n P ( τ b > T c ) ¯ = n ( k c c c + k nc c nc ) k c c c e - r c T c + k nc c nc e - r nc T c ¯ ≈ n k c c c e - r c T ¯ c + 1 2 r c 2 σ T c 2 k c c c + k nc c nc , ( 29 ) where we have used the approximation T ¯ c ≫ 1 / r nc in the last step . Thus σ T c 2 effectively renormalizes the cutoff to T ¯ c - 1 2 r c σ T c 2 . Replacing Tc in Eqs ( 27 ) and ( 28 ) by its renormalized value , which is an easy change in the scaling factors , removes this additional bias due to the random Tc in the KPR scheme . Since long bindings are rare , the variance of the KPR estimator is dominated again generally by A ¯ , but not B ¯ . The intrinsic stochasticity in the production of molecules of A contributes to the variance . However , this contribution can be made arbitrarily small by increasing kA , and we neglect it here . A larger contribution comes from the random number of long bound intervals and a random duration of each of them . To calculate this , in the limit of rare long binding events , we use well-known results in the theory of noise propagation in chemical networks [21] σ A 2 A ¯ 2 ≈ 1 + k c c c / r c + k nc c nc / r nc e r c T c - 1 2 r c 2 σ T c 2 k c c c ( 1 / r c + 1 / r A ) = e r c T c - 1 2 r c 2 σ T c 2 k c c c ( 1 / r c + 1 / r A ) + O ( ϵ ) . ( 30 ) This is a direct analog of Eq ( 16 ) . In principle , one can measure more than two concentrations similarly , as long as all species have distinct off-rates . For example , to estimate three concentrations , one needs an additional branch downstream of the receptor that proofreads for an intermediate time . Then the branches with the strongest , intermediate , and no proofreading would measure approximately the highest affinity ligand , a combination of the two higher affinity ligands , and all three ligands , respectively . Appropriate activation and inhibition of downstream targets will then allow identifying individual concentrations from these combined readouts . The error ( the variance of the ML estimator , and both the bias and the variance for the approximate and the KPR estimators ) would grow with an increasing number of ligand species , largely because a larger range of off-rates would be required to disambiguate more ligands . However , this would still represent a dramatic increase in the information gained by a receptor that tracks its precise temporal dynamics , rather than just the average binding state . Here we depart slightly from our scenario and show how a KPR-based scheme relying on the entire temporal sequence of activation / deactivation events can estimate the concentration of a single cognate ligand even if the two ligands have very similar off-rates rc ≲ rnc , a situation common in immunology . In such a situation , the KPR branch gets activated not just by the cognate ligand , but also by the non-cognate ligand ( though at a smaller rate ) . When the goal is the accurate estimation of the cognate ligand only , then the contribution to the KPR branch by the non-cognate ligand needs to be removed . To construct a signal transduction network able to do this , we abstract from the existing detailed model of FcϵRI immunological receptor [9] , a well studied eukaryotic signal transduction system mediating many allergic reactions [22] . Here the main signaling branch gets activated through the Lyn-Syk kinase pathway following kinetic proofreading after a ligand binds to the receptor [9] . However , receptor binding excites an additional branch early on , after only one step in kinetic proofreading ( a single phosphorylation on the β chain of the receptor ) . This branch activates Inpp5d ( SHIP ) phosphotase , which later dephosphorylates Phosphatidylinositol 3-phosphate ( PIP3 ) , a key downstream output of the main signaling branch , and sequesters the dephosphorylated product PtdIns ( 3 , 4 ) P2 [9] . The part of this signaling motif relevant for our analysis is summarized in a deliberately simplified signaling diagram in Fig 4 ( b ) , where A stands for PtdIns ( 3 , 4 , 5 ) P3 ( PIP3 ) , I stands for PtdIns ( 3 , 4 ) P2 , and I is produced by SHIP . Further , R is the FcϵRI receptor bound to an antibody , and cognate and noncognate molecules are the antigens specific/nonspecific to the bound antibody . In this network , we consider the main activator branch ( A ) , activated after the usual KPR delay , and hence sensitive to long binding events only ( which now have contributions both from kc and knc ) . The secondary inhibiting branch ( I ) is activated by many more binding events , though the shortest , nonspecific background binding events may be removed from both branches by additional proofreading steps ( an early cross-phosphorylation event in the FcϵRI system ) . The messengers in both branches later form a complex AI , and only A not in the complex activates further downstream signaling . If the production rates of A and I are appropriately matched ( which can be done if the off-rates are known a priori , which they should be for such a molecular signal detection system ) , this sequestration of A by I can effectively remove the contribution to the A branch coming from the non-cognate ligand . The kinetic diagram can be described with the following rate equations ( where , for simplicity , we neglect the first proofreading common to both branches ) : d A d t = β A - r A A - r AI A I , ( 31 ) d I d t = β I - r I I - r AI A I , ( 32 ) where rA/I are the degradation rates of the messengers A and I , rAI is the sequestration rate , and βA/I are the messenger production rates , derived as above: β A = k c c c / r c e - r c T c + k nc c nc / r nc e - r nc T c 1 + k c c c / r c + k nc c nc / r nc k A , ( 33 ) β I = k c c c / r c + k nc c nc / r nc 1 + k c c c / r c + k nc c nc / r nc k I . ( 34 ) Here kA/I are the rates of production of A and I , respectively , when the receptor has been bound for a sufficiently long time to produce either . We assume for simplicity rA = rI . Further , we choose rA = rI ≪ rAIA ∼ rAII , so that sequestration rather than degradation is primarily responsible for the disappearance of the messengers . Then the steady state solution of the rate equations ( Eqs 31 and 32 ) is [23] ( see Methods ) : A ¯ ss = β A - β I 2 r I + β A - β I 2 r I 2 + β A r AI , ( 35 ) I ¯ ss = β I - β A 2 r I + β I - β A 2 r I 2 + β I r AI . ( 36 ) The numerators of both βA and βI are linear combinations of cc and cnc . If the parameters of the biochemical networks are such that the production rate of the proofread branch is k A = k I e r nc T c , then ( βA−βI ) =kccc/rc ( e ( rnc−rc ) Tc−1 ) ( 1+kccc/rc+knccnc/rnc ) kI>0 , which has a cnc-independent numerator . Thus the contribution of non-cognate ligand to the activator branch is largely sequestered . Moreover , for large rAI , we have A ¯ ss ∝ ( 1 + k c c c / r c + k nc c nc / r nc ) - 1 , so that the activation of the A branch decreases as cnc increases . In contrast , if cc = 0 ( no cognate ligands present ) , then A ¯ ss = I ¯ ss = k I r AI k nc c nc k nc c nc + r nc , which grows with cnc . This behavior is reminiscent of the agonist-antagonist picture in FcϵRI receptor activation [24]: a weak ligand by itself can activate the cellular response , but it inhibits ( antagonizes ) activation of the response by a stronger agonist if both are present .
The realization of Refs . [5 , 13 , 12] and others that the detailed temporal sequence of binding and unbinding events carries more information about the ligand concentration than the mean receptor occupancy is a conceptual breakthrough . It parallels the realization in the computational neuroscience community that precise timing of spikes carries more information about the stimulus than the mean neural firing rate [25–30] , and it has a potential to be equally impactful . This extra information when measuring one ligand concentration with one receptor [5 , 12] amounted to increasing the sensing accuracy by a constant prefactor , or , equivalently , getting only a finite number of additional bits from even a very long measurement [31] . In contrast , here we show that two concentrations can be measured with one receptor with the variance that decreases inversely proportionally to the number of observations , n , Eq ( 16 ) , or to the integration time , 1/rB , Eq ( 30 ) , so that the accuracy is only an ( often small ) prefactor lower than would be possible with one receptor per ligand species . Asymptotically , this doubles the information obtained by the receptor [31] . Crucially , such improvement would not be possible without the cross-talk , or binding among noncognate ligands and receptors . Normally , the cross-talk is considered a nuisance that must be suppressed [32 , 33] . Instead , we argue that cross-talk can be beneficial by recruiting more receptor types to measure the concentration of the same ligand . In particular , this allows having fewer receptor than ligand species , potentially illuminating how cells function reliably in chemically complex environments with few receptor types . Further , the cross-talk can increase the dynamic range of the entire system: a ligand may saturate its cognate receptor , preventing accurate measurement of its ( high ) concentration , but it may be in the sensitive range of non-cognate receptors at the same time . Finally , the increased bandwidth may lead to improvements in sensing a time-dependent ligand concentration [11 , 13] . In forthcoming publications , we plan to explore such many-to-many sensory schemes , extending ideas of Ref . [34] to tracking temporal sequences of activation of the receptor and to temporally varying environments . While the exact maximum likelihood inference of multiple concentrations from a temporal binding-unbinding sequence is rather complex , we showed that when the cognate and the non-cognate off-rates are substantially different , there is a simpler , approximate , but accurate inference procedure for joint measurements of cognate and noncognate ligands . And even if the off-rates are close , one can still measure the cognate ligand concentration reliably . Crucially , this inference can be performed by biochemical motifs readily available to the cell . Namely , one needs two branches of activation downstream of the receptor , with at least one of them having a kinetic proofreading ( KPR ) time delay . Then the individual ligand concentrations can be obtained by mutual inhibition between the two branches , or by incoherent feedforward loops . We emphasize again that this allows estimation of multiple concentrations from activity of a single receptor , in contrast to a better estimation of just one concentration [12] . Our simple models only illustrate a wide class of models that can use the temporal structure of the receptor binding sequence to measure more that one ligand concentrations for various ligand combinations , including similar and dissimilar ligands . Additional branches from different points in the proofreading cascade provide additional information about the binding affinities of the mixture of ligands present in the environment , and then algebraic operations on these readouts can be performed by a large diversity of feedforward and feedback loops , competitions for the substrate and the enzyme , and so on . For example , in our simple model , the action of the antagonist is due to the competition for available receptors , while experiments suggest competition for a critical initiating kinase [24] , which would require a straightforward modification of the model . Similarly , antagonists are usually “medium” affinity ligands , while very weak ligands do not antagonize receptors . As illustrated in Fig 4 ( b ) , this can be achieved by having an additional KPR time delay common to both A and I branches , which occurs in practice [9] . The kinetic diagram for the FcϵRI receptor is not unique , and similar ( though not equivalent ) structures exist for other immune cells and receptors as well [9] . Such common structural features result in a similar phenomenology of activation profiles , which are different for pure ligands and ligand mixtures , and depend nontrivially on the details of the binding affinities and concentrations of the ligands in the mixture [10 , 16 , 35–38] . Interestingly , on longer time scales , a potentially related phenomenon in innate immune response is that of endotoxin tolerance ( desensitization to commonly present ligands ) [39] , which also affects ligands of different affinity differently , and in this case also depends on the history of exposure to other ligands [40] . It is mediated by SHIP , a crucial player in our analysis of FcϵRI signaling [41] , whose activity may be interpreted as setting the relative gain on the A and I branches of Fig 4 ( b ) , thus resulting in a more accurate signal estimation . In other words , one interpretation of the known results is that , as various feedback loops increase the activity of SHIP in response to frequent activation of signaling downstream of the receptor , the amount of I increases , thus sequestering more A , lowering its steady-state activity , and inducing tolerance . An important contribution of the understanding developed here is that one can try to interpret these various kinetic diagrams and their phenomenological consequences as implementing estimation of concentrations of potentially many ligands ( rather detection of a single one [11 , 13 , 16] ) , and maybe even doing it in a ( nearly ) Maximum Likelihood optimal fashion , under various assumptions about the number of distinct ligands , their relative abundance , and the ( dis ) similarity of the off-rates . Exploring feasibility of such an interpretation is an additional interesting venue for future research . In summary , monitoring precise temporal sequences of receptor activation/deactivation opens up new and exciting possibilities for environment sensing by cells .
We start with: P ≡ P ( { τ i b , τ i u } | c c , c nc ) = 1 Z ∏ i = 1 n e - τ i u ( k c c c + k nc c nc ) k c c c r c e - τ i b r c + k nc c nc r nc e - τ i b r nc . ( 37 ) The log-likelihood of kc , nc is the logarithm of P: log ( P ) = - log Z - ∑ i = 1 n τ i u ( k c c c + k nc c nc ) + log k c c c r c e - τ i b r c + k nc c nc r nc e - τ i b r nc . ( 38 ) Taking the derivatives of the log-likelihood w . r . t . cc and cnc and setting them to zero gives the Maximum Likelihood ( ML ) equations for the concentrations . These are: ∂ log ( P ) ∂ c c = - ∑ i = 1 n τ i u k c + ∑ i = 1 n k c r c e - τ i b r c D ( c c * , c nc * , τ i b ) = 0 , ( 39 ) ∂ log ( P ) ∂ c nc = - ∑ i = 1 n τ i u k nc + ∑ i = 1 n k nc r nc e - τ i b r nc D ( c c * , c nc * , τ i b ) = 0 . ( 40 ) Here , D ( c c * , c nc * , τ i b ) = ( k c c c * r c e - τ i b r c + k nc c nc * r nc e - τ i b r nc ) , with * denoting the ML solution . Denoting by T u = ∑ i = 1 n τ i u , the total time for which the receptor is unbound , these equations can be rewritten as - k c T u + ∑ i = 1 n k c r c e - τ i b r c D ( k c * , k nc * , τ i b ) = 0 , ( 41 ) - k nc T u + ∑ i = 1 n k nc r nc e - τ i b r nc D ( k c * , k nc * , τ i b ) = 0 . ( 42 ) Multiplying Eqs ( 41 ) and ( 42 ) by c c * and c nc * , respectively , and adding them gives k c c c * + k nc c nc * = n T u . ( 43 ) Here we compare the results obtained from the numerical simulations to the analytical expressions derived in Ref . [12] for detection of the concentration of the cognate ligand in a background of spurious ligands . The variance of the concentration estimation obtained from the simulations matches quite well with integral expression , Eq . ( 7 ) in Ref . [12] , Fig 5a . Note that this expression is inverse of the ( 1 , 1 ) term of the Hessian matrix , Eq 7 . The analytical results obtained for the low concentration of the cognate ligand compared to a non cognate ligand ( cc ≪ cnc ) also match the simulations , Fig 5b . In the biochemical network in Fig 4 ( a ) of the main text , the receptor R activates two messenger molecules , A and B . The former is activated with the rate kA only if the receptor stays bound for longer than a certain Tc ( with the delay achieved using the KPR intermediate states ) . The latter is activated with the rate kB whenever the receptor is bound . The molecules deactivate with the rates rA and rB , respectively , and all activations/deactivations are first-order reactions . The rate equation for the two molecules can be written as: d A d t = k A Θ ( τ b > T c ) - r A A , ( 54 ) d B d t = k B Θ ( τ b > 0 ) - r B B . ( 55 ) The Θ functions represent the fact that A is produced only when the receptor has been bound for longer than the cutoff time Tc , and B is produced only when the receptor is bound . The steady state value of A ¯ can be obtained by equating the average deactivation rate r A A ¯ to kA times the fraction of time the receptor occupancy was larger than the cutoff , Tc , i . e . , r A A ¯ = k A 〈 τ b > T c 〉 〈 τ u 〉 + 〈 τ b 〉 . ( 56 ) Similarly , B ¯ can be obtained as: r B B ¯ = k B 〈 τ b 〉 〈 τ u 〉 + 〈 τ b 〉 . ( 57 ) Therefore , the mean concentrations of the messenger molecules are: A ¯ = k c c c / r c e - r c T c + k nc c nc / r nc e - r nc T c 1 + k c c c / r c + k nc c nc / r nc k A r A , ( 58 ) B ¯ = k c c c / r c + k nc c nc / r nc 1 + k c c c / r c + k nc c nc / r nc k B r B . ( 59 ) The rate equations are: d A d t = β A - r A A - r AI A I , ( 60 ) d I d t = β I - r I I - r AI A I . ( 61 ) Equating the r . h . s . to zero gives the steady state conditions: β A - r A A ¯ ss - r AI A ¯ ss I ¯ ss = 0 , ( 62 ) β I - r I I ¯ ss - r AI A ¯ ss I ¯ ss = 0 . ( 63 ) The latter of these can be rewritten as: I ¯ ss = β I r I + r AI A ¯ ss . ( 64 ) Plugging this in Eq ( 62 ) , we get β A - r A A ¯ ss - r AI A ¯ ss β I r I + r AI A ¯ ss = 0 , ( 65 ) which can be simplified to: A ¯ ss 2 + r I r AI + β I - β A r A A ¯ ss - β A r I r AI r A = 0 . ( 66 ) This quadratic equation has the solution: A¯ss=− ( rI2rAI+ ( βI−βA ) 2rA ) + ( rI2rAI+ ( βI−βA ) 2rA ) 2+βArIrAIrA . ( 67 ) Now sssuming rA = rI and rA = rI ≪ rAIA ∼ rAII , we get: A ¯ ss = ( β A − β I ) 2 r I + ( β I − β A 2 r A ) 2 + β A r AI ( 68 ) One can similarly can get the equation for I ¯ ss as well . | Cells live in chemically complex environments with many different chemical ligands around them . Can cells estimate concentrations of more ligands than they have receptor types ? In this paper , we show that , surprisingly , the answer is “yes” , and the estimation can be implemented with simple biochemical components already present in many cells . Therefore , cells may “know” a lot more about their environment and thus may be able to implement more complex and accurate response strategies than was previously thought . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"immunology",
"enzymology",
"social",
"sciences",
"neuroscience",
"immune",
"receptor",
"signaling",
"membrane",
"receptor",
"signaling",
"sequence",
"motif",
"analysis",
"enzyme",
"kinetics",
"research",
"and"... | 2017 | Simple biochemical networks allow accurate sensing of multiple ligands with a single receptor |
Cell-to-cell transmission of vaccinia virus can be mediated by enveloped virions that remain attached to the outer surface of the cell or those released into the medium . During egress , the outer membrane of the double-enveloped virus fuses with the plasma membrane leaving extracellular virus attached to the cell surface via viral envelope proteins . Here we report that F-actin nucleation by the viral protein A36 promotes the disengagement of virus attachment and release of enveloped virus . Cells infected with the A36YdF virus , which has mutations at two critical tyrosine residues abrogating localised actin nucleation , displayed a 10-fold reduction in virus release . We examined A36YdF infected cells by transmission electron microscopy and observed that during release , virus appeared trapped in small invaginations at the plasma membrane . To further characterise the mechanism by which actin nucleation drives the dissociation of enveloped virus from the cell surface , we examined recombinant viruses by super-resolution microscopy . Fluorescently-tagged A36 was visualised at sub-viral resolution to image cell-virus attachment in mutant and parental backgrounds . We confirmed that A36YdF extracellular virus remained closely associated to the plasma membrane in small membrane pits . Virus-induced actin nucleation reduced the extent of association , thereby promoting the untethering of virus from the cell surface . Virus release can be enhanced via a point mutation in the luminal region of B5 ( P189S ) , another virus envelope protein . We found that the B5P189S mutation led to reduced contact between extracellular virus and the host membrane during release , even in the absence of virus-induced actin nucleation . Our results posit that during release virus is tightly tethered to the host cell through interactions mediated by viral envelope proteins . Untethering of virus into the surrounding extracellular space requires these interactions be relieved , either through the force of actin nucleation or by mutations in luminal proteins that weaken these interactions .
Crossing the membrane of host cells , either during entry or escape , is a major obstacle facing prospective viral pathogens during infection . Enveloped viruses are released from cells by either the acquisition or loss of an outer membrane and both strategies pose unique challenges to the final separation of pathogen from host . Where viruses gain a membrane , for example influenza virus and human immunodeficiency virus , a tight association must be formed between assembling viral complexes and the internal surface of the cell membrane that is loaded with viral envelope proteins [1] , [2] . As the budding virus emerges into the extracellular space , membrane scission must take place , an energetically difficult event [3] . Other viruses , including herpes simplex virus , acquire a double membrane during morphogenesis that is tightly complexed by viral protein interactions across the luminal space [4] . Upon reaching the cell surface , an exocytotic membrane fusion event is followed by the peeling away of the outer vesicle membrane accompanied by the disengagement of virus–cell associations . Mature enveloped virions are then free to diffuse in the extracellular space . The release of extracellular enveloped virus ( EEV ) , the morphological variant implicated in cell-to-cell transmission of vaccinia virus ( VACV ) , operates through a membrane-loss mechanism . Intracellular enveloped viruses ( IEV ) arrive at the plasma membrane where the outer of two early endosome or trans-Golgi derived membranes fuses with the plasma membrane forming cell-associated enveloped virus ( CEV ) [reviewed in 5] . Extracellular CEV remain associated with host cells and this attachment is likely to be mediated by viral envelope proteins [6] , [7] , [8] . For example , the viral proteins B5 , A33 and A34 ( encoded by the B5R , A33R and A34R genes , respectively ) contain significant luminal regions [9] , [10] , [11] , [12] . In support of a role for these luminal domains in virus–host cell adhesion , a number of mutations in these proteins have been documented that lead to increased EEV release ( B5P189S , A34K151E and a C-terminal deletion in A33 ) [6] , [13] , [14] . While CEV are able to maintain a tight affinity to the surface of infected cells , over time they are untethered and free to access the extracellular space , though how the affinity is regulated remains largely unknown . A number of lines of evidence demonstrate that host-signalling pathways affect the dissociation of VACV . For example , cell lines of different origin display significantly altered kinetics of EEV release that is independent of the total amount of virus made [15] . Inhibition of epidermal growth factor receptor reduces EEV from VACV IHD-J strain [16] . Finally , EEV levels are reduced in the absence of the phosphoinositide 5-phosphatase SHIP2 [17] . These data further support the hypothesis that the generation of enveloped virus and the untethering of enveloped virus are regulated independently . Inhibition of Abl tyrosine kinases also results in reduced EEV [18] , [19] , although at this stage it is unclear as to whether this is a specific defect in EEV dissociation or a more general defect in virus morphogenesis . The viral transmembrane protein A36 is currently the only Abl substrate that has been implicated in VACV morphogenesis , although its role in Abl-mediated release is unknown [18] , [19] , [20] . Of the integral viral proteins localised to IEV , A36 is unusual in that it is expressed exclusively in the outer of the two trans-Golgi-derived membranes [21] , [22] . Consequently , it is present beneath CEV once the outer virus membrane fuses and becomes contiguous with the plasma membrane , and accordingly is absent from released EEV . Unlike B5 , A33 and A34 , the bulk of the A36 protein extends into the cytoplasm . Here it mediates interactions with the microtubule cytoskeleton via WE/WD motifs [23] , [24] , [25] , and the actin cytoskeleton via the phosphorylation of tyrosine residues at positions 112 and 132 ( A36Y112 , A36Y132 respectively ) [26] , [27] , [28] . Src- and Abl-family kinases instigate the phosphorylation of A36Y112 and A36Y132 , which generates binding sites for the SH2 domains of Nck and Grb2 adaptor proteins that combine to stabilise N-WASP , a potent activator of the Arp2/3 complex [20] , [26] , [28] , [29] , [30] . Activation of the F-actin branching activity of the Arp2/3 complex beneath CEV promotes virus motility that is restricted to the plane of the cell membrane and results in thin virus-tipped membrane protrusions . Although Src- and Abl-family kinases act redundantly to phosphorylate A36 [19] , [20] , inhibition of Abl-family kinases alone with the specific inhibitor imatinib results in reduced EEV release without affecting actin nucleation [19] , [31] . Localised actin nucleation expedites the dispersal of CEV to neighbouring cells and also facilitates super-repulsion , which has been proposed to account for the rapid cell-to-cell transmission of virus that outpaces replication dynamics [32] . Super-repulsion occurs when either CEV or EEV contact the surface of early-infected cells , are repelled by A36-induced actin-based motility and leapfrog to uninfected cells . This process is dependent on early-stage expression of A36 in the recipient cell membrane , before infectious progeny are produced . Here we show that introducing both Y112F and Y132F mutations into A36 leads to a severe decrease in the production of EEV . In the absence of virus-induced actin nucleation , virus particles remained trapped at the plasma membrane in small invaginations . When parental VACV was examined using three-dimensional structured illumination microscopy ( 3D-SIM ) , it was found that upon reaching the surface , A36 redistributed to a discrete region beneath CEV , reducing contact between extracellular virus and the host cell . This redistribution was dependent on actin-based motility but could also be phenocopied by a mutation in the luminal domain of the envelope protein B5 . In our model , following the fusion of IEV at the plasma membrane , CEV remain tethered to the cell in tight membrane pits through interactions between viral proteins . Tethering can be relieved either by mutations in the luminal domains of viral proteins that disrupt these interactions , or by the force supplied by localised actin nucleation .
Actin-based motility appears to have evolved independently in a diverse range of pathogenic bacteria and viruses , and it is speculated to contribute to virulence in a context-specific manner [33] , [34] . To better understand the role of pathogen-induced actin nucleation in the replication and spread of VACV , we tested the effects of blocking the phosphorylation of A36Y112 and A36Y132 on the release of EEV . These residues are critical to the canonical pathway whereby VACV induces actin nucleation by recruitment and activation of the Arp2/3 complex . Plaque assays performed under semi-solid overlay are used to quantitate replication dynamics and cell-to-cell transmission . When performed under a liquid overlay , these are termed ‘comet’ assays and the efficiency of EEV release can be qualitatively assessed by the formation of satellite plaques that disperse from primary plaques due to the action of convection currents [6] , [35] . Plaque assays were performed in BSC-1 cells with VACV Western Reserve ( WR , parental strain ) , A36Y112F , A36Y132F and A36YdF ( = A36Y112F/Y132F ) to assess cell-to-cell spread ( Figure 1A ) . The A36YdF virus yielded plaques significantly smaller than WR , in broad agreement with previous reports [36] . Mutation of Y112 alone also had a significant effect on plaque size whereas mutation of Y132 alone resulted in plaques not significantly different to those of WR ( Figure 1B ) . Similar results were observed when using NIH3T3 cells ( Figure 1C ) . Reduction in plaque size is likely to be due to the loss of actin-based motility , potentially by its role in facilitating cell-to-cell spread by super-repulsion [32] . A36YdF and A36Y112F viruses do not activate the Nck pathway , which is both necessary and sufficient for actin-based motility , whereas mutation of Y132 leads to a disruption in the dynamics of N-WASP recruitment and a reduction in the efficiency of the initiation of actin-based motility [26] , [28] . We next subjected the A36 mutant strains to comet assays to assess EEV release . A reduction in the extent of comet formation was observed in all strains carrying Y/F substitutions ( Figure 1A ) . The greatest reduction was observed in cells infected with A36YdF , followed by A36Y112F , and only a minor reduction was seen with A36Y132F . To support these findings and quantify the effects of A36 phosphorylation on virus release , EEV release assays were performed in BSC-1 cells . We observed the same trend in EEV release as in the comet assays , with the greatest inhibition of release ( approximately 10-fold ) seen with A36YdF and a significant inhibition seen when Y112 was mutated ( Figure 1E ) . These findings were replicated when release assays were performed in NIH3T3 cells ( Figure 1F ) . Disruption of A36 does not result in early defects in wrapping and morphogenesis and the Y112 and Y132 residues are not required for microtubule-dependent delivery of IEV to the cell surface [21] , [23] , [24] , [37] , [38] , [39] . Single-step growth curves did not reveal significant differences in the replication dynamics between WR , A36Y112F , A36Y132F and A36YdF ( Figure 1D ) . Thus , any differences observed in EEV release for these viruses cannot be accounted for by earlier morphogenesis or transport defects . Abl kinases regulate at least two distinct steps during VACV replication: EEV release and actin-based motility of CEV , in the latter case through A36 phosphorylation [19] , [20] . The effects of abrogating release by mutation of Y112 or Y132 are consistent with A36 fulfilling the role as the Abl-dependent mediator of virus release . To test this role , we examined the effects of imatinib in our various mutant backgrounds . Imatinib is a kinase inhibitor that blocks the ATP binding site on Abl-family kinases and has remarkable specificity displaying no detectable activity against Src-family kinases [20] , [31] . Surprisingly , we found that regardless of the integrity of A36 residues Y112 and Y132 , treatment with imatinib led to a significant reduction in comet formation and an approximately 2- to 3-fold decrease in EEV ( Figure 1A , 1E , 1F ) . These results demonstrate a role for A36 tyrosine phosphorylation in release that is independent of the effects of imatinib . Although we cannot exclude that Abl kinases regulate virus release via A36 phosphorylation , if they do so they act redundantly with Src-family kinases , as they do in the initiation of actin-based motility . Deletion of genes encoding envelope-specific proteins can increase ( A33R , A34R ) or decrease ( B5R , F13L ) production of EEV , often due to earlier defects in morphogenesis [40] , [41] , [42] , [43] . While our results are the first to implicate A36 tyrosine phosphorylation in virus release , it has previously been shown that deletion of A36 results in a dramatic reduction in EEV [37] . This is unsurprising given the essential role this protein plays in delivering IEV to the cell surface through kinesin-1 based transport; however , Y112 and Y132 play no part in this process [23] , [24] , [25] . Inhibition of virus release in an A36 deletion strain can be derepressed , and significantly enhanced , by second-site mutations in genes that encode the envelope-specific proteins B5 ( P189S substitution ) and A33 ( C-terminal truncations ) [14] . A point mutation in A34R ( K151E ) , identified in the IHD-J strain , also significantly enhances EEV release in a WR background [6] , [19] . We tested whether the B5P189S or A34K151E alleles would also promote EEV release in an A36YdF background . Both B5P189S and A34K151E led to substantial increases in comet formation both in a parental and A36YdF background ( Figure 2A ) . EEV release assays revealed that the B5P189S virus generated 5-fold more infectious EEV than the parental WR strain and 40-fold more than A36YdF ( Figure 2B ) . There was no significant difference in EEV release between B5P189S in a WR background or an A36YdF background . The A34K151E virus produced 50-fold more infectious EEV than WR and 400-fold more than A36YdF ( Figure 2B ) . In contrast to the B5P189S virus , a significant reduction ( 10-fold ) in EEV was observed in A34K151E/A36YdF virus compared to A34K151E ( Figure 2B ) . Therefore , in a background where actin-based motility is intact ( WR and A34K151E ) [17] , [44] , the A36YdF mutation potently supresses EEV release . Conversely , in a B5P189S background , which is strongly deficient in actin-based motility owing to a failure of kinase activation , the A36YdF allele has no effect on EEV release [30] . Our results indicate that phosphorylation of A36 is required for efficient EEV release but do not discriminate whether this is due to the induction of localised actin polymerisation or signalling via A36 through another mechanism . We therefore tested the effects of inhibiting actin nucleation on EEV production using two different approaches . Nck-null mouse embryonic fibroblasts do not support virus-associated actin nucleation due to the essential requirement of this adaptor protein in the recruitment of N-WASP and subsequent activation of the Arp2/3 complex [26] , [28] , [45] . Nck-null cells infected with WR or A36YdF released similar levels of EEV ( Figure 3A ) . Nck-null cells infected with B5P189S , B5P189S/A36YdF , A34K151E and A34K151E/A36YdFexhibited similar levels of EEV release , approximately 3-fold greater than WR and A36YdF . We further tested the role of virus-induced actin nucleation by treating infected cells with cytochalasin D ( Cyt D ) , a potent inhibitor of actin polymerisation that caps the fast growing ends of actin filaments . Previous studies have shown that Cyt D significantly reduces WR EEV while having little effect on morphogenesis or formation of CEV [15] , [39] , [46] . Treatment with Cyt D resulted in a 3-fold reduction in EEV release from cells infected with WR or A34K151E ( Figure 3B ) . In contrast , Cyt D had little impact on EEV production in cells infected with A36YdF or B5P189S . Collectively , these data suggest that in the absence of virus-induced actin nucleation , the release of EEV is independent of the status of A36 Y112 and Y132 residues . Hence , the deficiency in EEV release observed for A36YdF can be replicated by Cyt D treatment , and in a cell line where actin-based motility is blocked , inhibition of EEV release due to the YdF mutation is not observed . Our results show a correlation between actin-based motility and release of EEV , with one anomaly . The A36Y112F virus displays a small , but significant , increase in virus release compared to A36YdF ( Figure 1E , 1F ) , yet both strains have been characterised as deficient in actin-based motility [26] . To resolve this paradox , we re-examined A36Y112F-infected cells and compared these with WR- and A36YdF-infected cells . Consistent with previous reports we were unable to detect virus-associated F-actin tails of 3–4 µm in length , which are typical of WR infection , in A36Y112F-infected cells ( Figure 3C ) [28] . However , we did observe F-actin accumulations that colocalised with extracellular A36Y112F virus , albeit infrequently . Small clumps of F-actin accumulation are observed in VACV-infected and uninfected cells at a low frequency , so to determine whether the observed colocalisations were stochastic events or represented viral-induced actin nucleation , we imaged cells with live microscopy ( Figure 3D and Video S1 , S2 , S3 ) . In cells infected with A36YdF ( n = 9 ) , no motile virus particles were associated with F-actin . In cells infected with A36Y112F ( n = 9 ) , six cells had at least one example of virus motility associated with transient F-actin accumulation over the course of two minutes . This was in contrast to actin tails that were associated with parental WR virus , which were robust and longer lived ( Figure 3D ) . Taken together , these results confirm a correspondence between the ability to stimulate actin nucleation , albeit weakly in the case of A36Y112F , and defects in EEV release . In order to disclose the mechanism by which EEV release is disrupted in A36YdF , we examined infected cells by transmission electron microscopy ( TEM ) . The majority of exiting enveloped viruses were found loosely associated with the plasma membrane in WR infected cells ( Figure 4A ) , whereas A36YdF viruses were predominantly contained in membrane pits at the surface of infected cells ( Figure 4B ) . A36YdF viruses that remained at the cell surface were identified as CEV as the outer viral membrane was observed to be contiguous with the plasma membrane [47] , [48] . The proportion of the CEV envelope in contact with the cell membrane was measured and was found to be significantly different between WR and A36YdF . A36YdF CEV displayed greater contact on average with the host membrane than the parental strain ( Figure 4C ) . CEV residing in plasma membrane pits have been previously documented when cells infected with IHD-J , a strain harbouring the A34K151E substitution , were treated with Cyt D ( see Figure 4A in [15] ) . These results suggested that virus-induced actin nucleation might function to expel CEV from membrane pits leading to release of EEV . Despite the high resolution afforded by TEM , the extremely small portion of the total cell volume that is included in an ultrathin section does not provide sufficient data for extensive quantitative comparisons between strains . Fluorescence microscopy of fluorescently-tagged or immunolabelled proteins is a more efficient way to visualise protein localisation and this approach is also amenable to labelling of non-permeabilised cells to distinguish extracellular epitopes . Standard wide-field or confocal imaging is , however , limited by the diffraction of light that poses a limit on resolving power to approximately 250 nm . To exploit the advantages of fluorescence microscopy and simultaneously resolve virus particle morphology we utilised 3D-SIM super-resolution microscopy [49] . We generated recombinant viruses that inserted A36-YFP or A36YdF-YFP into the endogenous locus; these were subsequently combined with B5P189S , A34K151E and B5-mRFP ( monomeric red fluorescent protein ) alleles or transgenes . Initially , cells expressing Lifeact fused to cerulean fluorescent protein ( to visualise the actin cytoskeleton ) were infected with A36-YFP/B5-mRFP virus . Lifeact is a 17 amino acid peptide that binds to filamentous actin [50] . Using 3D-SIM , envelope proteins A36 and B5 resolved as hollow spherical structures that corresponded to IEV [data not shown and 51] . When A36 was localised to virus-tipped F-actin tails , a redistribution was apparent as A36 concentrated to discrete regions adjacent to the virus particles , whereas B5 remained distributed along the circumference of virus particles ( Figure 5A ) . This is consistent with previous electron microscopy studies which have demonstrated that unlike B5 , A36 is excluded from the inner of the two trans-Golgi derived viral membranes [22] . To further support our electron microscopy data , we compared the localisation of A36-YFP with A36YdF-YFP , while distinguishing between extracellular and intracellular enveloped virus with an antibody label ( Figure 5B , 5C , 6C ) . We were able to observe a trend of A36-YFP polar redistribution upon fusion with the plasma membrane that was inhibited in the A36YdF background . Thus , in the absence of A36 tyrosine phosphorylation and actin nucleation , close contact is maintained between extracellular virus ( labelled with anti-B5 ) and the surface of the cell ( labelled with A36-YFP ) . Analysis of individual z-planes in a 3D reconstruction revealed contact between the virus and cell was limited to a few planes in A36-YFP , but in A36YdF-YFP A36 colocalised with exposed B5 over the majority of a CEV particle ( Figure 6C ) , reflecting the TEM phenotype . A36 was frequently polarised at the surface of both B5P189S/A36-YFP and B5P189S/A36YdF-YFP enveloped virus ( Figure 5D ) . Thus introduction of the B5P189S allele restored A36 polarisation in a YdF background . In contrast , the polarisation of A36 in A34K151E/A36-YFP was inhibited by the YdF mutation ( Figure 5D ) . These data were quantified by enumerating single viruses that corresponded to categories of close , intermediate and loose contact ( polar , intermediate and circular A36 distributions ) . This analysis revealed the majority of A36-YFP , B5P189S/A36-YFP , B5P189S/A36YdF-YFP and A34K151E/A36-YFP CEV displayed polar A36 distribution and the majority of A36YdF-YFP and A34K151E/A36YdF-YFP CEV displayed a circular A36 distribution , colocalising with extracellular B5 ( Figure 6A , 6B ) . Collectively these data show that in backgrounds that support actin-based motility , introduction of the YdF mutation leads to viruses getting trapped at the cell membrane with a failure to readily disengage virus–host contact .
Substantial evidence supports the critical role of the enveloped form of orthopox viruses in cell-to-cell transmission [52] , [53] , despite EEV being stoichiometrically the minor infectious form produced during virus replication [54] , [55] . Epitopes derived from envelope proteins are major antigens recognised by host immune defences and neutralising antibodies to these antigens are effective in providing protective immunity [56] , [57] . Mutations in viral genes that encode envelope-specific proteins also lead to attenuation both in vitro , such as reduced cell-to-cell spread measured through plaque assays , and when used to infect animal models [8] , [14] . These findings can be explained by effects on transmission by CEV or the speed at which EEV are released from infected cells . While there is evidence that both IMV and IEV are transported to the cell periphery on microtubules , enveloped virus , using A36-mediated transport via kinesin-1 , is efficiently translocated to the cell periphery at approximately eight hours post infection . This is the earliest time point that infectious progeny are released [39] , [58] and at this time relatively few IMV are dispersed throughout the cell . Additionally , IMV lack a specific mechanism for release from infected cells other than through cell lysis , which occurs only late during infection [56] . Little is known regarding the fusion event between the outer IEV envelope and the plasma membrane leading to the formation of CEV , but this event correlates with a number of changes at the molecular level . These changes include the rapid dissociation of the viral proteins F12 and E2 and the kinesin-1 motor complex [22] , [30] , [59] , [60] , followed by recruitment of Src- and Abl-family kinases and the phosphorylation of A36 [19] , [20] , [27] , [30] . Although the exact mechanism that leads to the recruitment of cellular kinases to CEV is unknown , it is dependent on an intact SCR4 domain of B5 which becomes exposed on the cell surface upon viral fusion [30] . During super-repulsion , the B5 SCR4 must also be present on the surface of EEV to efficiently induce actin-based motility on A33/A36 expressing cells [44] . This provides further support for the ability of this domain to induce signalling events across the extracellular space between enveloped virus and the plasma membrane . We have now shown that in the absence of A36 phosphorylation and subsequent localised actin nucleation , CEV remain trapped in pits at the plasma membrane thereby blocking the liberation of EEV from infected cells ( Figure 6D ) . Our model is supported by the strong inhibition in EEV release observed in the A36YdF strain , which can be replicated through blocking actin nucleation by removing Nck or with drug intervention . Humphries et al . [61] have recently described a role for clathrin and the clathrin adaptor protein AP-2 in the regulation of VACV-induced actin nucleation . In the absence of AP-2 , actin-based motility is delayed and N-WASP turnover during motility is reduced . Concurrent with these effects was a reduction in A36 polarisation , which strongly supports our findings: disrupting actin nucleation correlates with inhibition of the expulsion of enveloped virus at the cell surface . Even slight differences in the potential of actin nucleation , such as between the A36Y112F and A36YdF strains , results in changes to the efficiency of EEV release . Humphries et al . also describe the localisation of N-Wasp , which , like A36 , is less polarised when actin nucleation is disrupted , although the temporal dynamics in relation to virus expulsion are not clear , as they were unable to distinguish between extracellular and intracellular virus . With great prescience , Payne [15] , upon observing electron micrographs of Cyt D-treated IHD-J infected cells , speculated that “the final separation of EEV from the plasma membrane is probably dependent on functioning microfilaments [actin]” but , as actin-based motility was unknown at the time , concluded that the membrane dynamics associated with cell motility was the ultimate cause . Inhibiting the action of Abl-family kinases with the specific inhibitor imatinib leads to a reduction in EEV release , implicating these kinases in the regulation of this process . Despite our previous work demonstrating that A36 was a direct Abl substrate [20] , we believe the mechanism by which imatinib disrupts virus release is independent of the role of A36 based on the following evidence . Firstly , disruption of Y112 and Y132 ( A36YdF ) leads to a far stronger reduction in EEV ( 10-fold ) than treatment with imatinib ( 3-fold ) . Secondly , imatinib inhibits EEV release regardless of the integrity of A36 Y112 and Y132 residues . Therefore , while Abl family kinases may regulate EEV release through phosphorylation of A36 , they do so redundantly with Src-family kinases , as we have previously proposed for actin-based motility [20] . The effects of imatinib on virus release suggests another , Src-independent , role for Abl kinases , the mechanism of which remains unknown . To better characterise the mechanism by which actin nucleation promotes virus release , we examined the effects of recombining the A36YdF mutation into high EEV release backgrounds . These included strains carrying mutations in the envelope proteins B5 ( B5P189S ) and A34 ( A34K151E ) . The B5P189S virus exhibits high levels of EEV release irrespective of the status of A36 Y112 and Y132 residues [6] , [13] , [14] . In contrast , the high level of A34K151E EEV release was potently suppressed by the A36YdF mutation . These data show that in strains that activate actin-based motility ( WR and A34K151E ) the YdF mutation suppresses EEV release , but has no effect in cases where actin-based motility is absent ( B5P189S-infected or Nck null cells ) [13] , [14] , [30] . Although both A34K151E and B5P189S result in high levels of EEV , only the latter bypasses the requirement for actin-based motility to facilitate release . A34K151E appears to exhibit a faster rate of enveloped virus production , but EEV release is nonetheless regulated in the same manner as the parental WR strain [62] . In B5P189S a high rate of EEV release is achieved at a cost to morphogenesis , hence the small plaque phenotype . Collectively , these findings indicate that two pathways can liberate EEV from the surface of cells: localised actin filament nucleation or disruption of viral protein-to-host membrane interactions . The exact nature of these interactions is unclear as few experiments have been performed that are able to distinguish between interactions occurring in cis and trans in the viral envelope [63] , [64] , and the pleiotropic roles of these proteins confounds the interpretation of mutant phenotypes . The best evidence for an interaction between the outer viral membranes being mediated by the luminal domains of viral proteins derives from Perdiguero et al . [65] , who show that the SCR regions of B5 can interact with the ectodomain of A34 in the absence of transmembrane domains . Recent findings on the mechanism of super-repulsion also implicate A33 as a B5-interaction partner . Doceul et al [32] demonstrated that expression of only A33 and A36 in the recipient host cell membrane is required for super-repulsion . It was subsequently reported that the B5P189S mutation on the surface of EEV also disrupts super-repulsion actin-based motility [44] . Recently , an interaction was identified between the isolated luminal domain of A33 and a luminal coiled-coiled domain of B5 that lies adjacent to SCR4 [63] . As this interaction is dependent on the coiled-coiled domain being anchored to a membrane , its role in adhesion across the viral envelope membranes is unclear and may represent an independent role for A33 in incorporation of B5 into the envelope , or vice versa [63] , [64] , [65] . Intriguingly , mutations that increase B5–A33 adhesion result in reduced binding of EEV to host cells suggesting a cis interaction in the membrane between these proteins and raising the possibility that A33 might mask and/or regulate the availability of the adjacent SCR4 [66] . Further characterisation of the nature of these interactions may clarify how adhesion between the outer viral membranes is achieved and regulated . In most cases , a trade-off exists between adhesion and efficient wrapping as a subset of high release mutations in envelope proteins result in small plaque phenotypes and these viruses are attenuated when used to infect mice ( B5 , A33 [8] , [13] ) . Disrupting these interactions via the force of localised actin nucleation offers an elegant solution to unshackling the luminal interactions that , while required for efficient wrapping , need to be released for EEV to be untethered . What is the role of EEV release during pathogenesis ? The literature does not offer a clear consensus as to whether strains that release more EEV are more or less virulent . Mutations isolated in a WR background such as B5P189S and A33 C-terminal truncations lead to increased EEV release but also a reduction in plaque size [13] . Unsurprisingly , viruses carrying these mutations are attenuated when used to infect mice [13] . Alternatively , one can compare orthopox isolates with characteristic EEV release profiles and correlate virulence . A comparison of variola virus strains identified a high correlation between decreased virulence and high EEV release [67] . It has been speculated that the decreased virulence of the high release orthopox strains may facilitate transmission between hosts to the detriment of cell-to-cell transmission within a host [67] . Mice infected with the high-release IHD-J strain , which carries the A34K151 allele , display reduced mortality compared to those infected with WR ( 70% verses 85% lethality over 3 weeks post-infection ) [52] , [54] . However , WR , which has low EEV release , may be the exception rather than the rule; a broader comparison of VACV strains show a positive correlation between EEV release and the ability of the virus to disseminate within a host and cause lethality [54] . All of these studies are limited by the use of animals that have not been verified as a natural host and may therefore not preserve aspects of endemic host–pathogen interactions during in vivo spread and transmission . The first virus characterised as capable of undergoing actin-based motility was VACV , but its role in enhancing infection outcomes has only recently become clear . Virus plaques formed by A36YdF are reduced in size , attesting to the role of these residues in enhancing cell-to-cell spread ( [36] , this study ) and enabling super-repulsion [32] . Here we show that inhibition of A36-induced actin nucleation results in a severe reduction in the release of infectious EEV from infected cells , measured by liquid plaque assay or directly quantified . It should not be surprising that once poxviruses evolved a mechanism to exploit host actin dynamics , this mechanism might be co-opted for other functions . For example , localised actin nucleation may have evolved to untether EEV and then subsequently have been exploited to mediate super-repulsion , or vice versa . In principle , the direction of the co-opting of function could be discriminated by the examination of the promoter activity of A36R homologues; if actin nucleation evolved to facilitate EEV release then distantly related homologues might be expected to lack early stage transcriptional activity . Although A36 is not highly conserved at the sequence level within the Chordopoxvirinae subfamily , it is at the functional level . For example , a diversity of poxviruses encode proteins at a homologous locus that are able to restore actin-based motility in a VACV ΔA36 strain despite exhibiting as little as 9% homology at the amino acid level [68] . Owing to the broad definition of early and late promoter consensus motifs [69] , [70] , a preliminary inspection to identify the type of promoters in the A36R homologues was unsuccessful . Functional studies of the activities of these promoters would potentially resolve this issue . Future research may identify yet further functions for VACV-induced actin nucleation in facilitating virus spread . Pathogen-induced actin nucleation has evolved independently in a number of viral and bacterial lineages , but the role of actin nucleation in pathogenesis varies substantially [71] , [72] . During the colonisation of the intestinal tract by Enteropathogenic Escherichia coli , extracellular bacteria are thought to attach to the epithelium by the formation of F-actin-rich pedestals , utilising a cascade bearing remarkable similarity to VACV-induced actin nucleation [73] . Listeria monocytogenes stimulates polarised actin nucleation at the surface of intracellular bacteria promoting motility within the cell and forming long membrane extensions that may enhance infection of neighbouring cells [74] . Similar structures are observed in VACV infected cells , albeit tipped by an extracellular pathogen , and these are likely to also facilitate cell-to-cell spread during VACV infection . Herpesviruses must also resolve the untethering luminal interactions during de-envelopment from the nuclear membrane and exit at the plasma membrane . Although very little is know regarding these processes there are remarkable similarities to VACV escape [4] . Whether herpesvirus and VACV have evolved similar mechanisms to release virus particles must await further studies , currently F-actin has not been localised to herpesvirus particles during release [75] . A close parallel to the role of actin nucleation that we describe here for EEV release may be clathrin-mediated endocytosis . Here actin filament nucleation at the neck of endocytic vesicles exhibits sufficient force to promote internalisation and membrane scission [76] . This pathway is used during influenza virus entry , which is also sensitive to Cyt D [77] . Vaccinia virus EEV release operates in reverse , the nucleator is localised to the intracellular cargo and the force of actin nucleation peels away the outer envelope , expelling the virus particle to the surface of the cell . Thus a similar intermediate , a recently internalised influenza virus particle or a recently egressed VACV CEV , will resolve in opposite directions depending on where the force is applied .
African green monkey kidney cells ( BSC-1 ) , murine embryonic fibroblasts ( NIH3T3 ) and Nck-null NIH3T3 cells were maintained in Dulbecco's Modified Eagle's Medium ( DMEM; Invitrogen ) supplemented with 5% foetal bovine serum ( FBS ) , 292 µg/ml L-glutamine , 100 units/ml penicillin and 100 µg/ml streptomycin ( DMEM-FPSG ) at 37°C and 5% CO2 . Vaccinia virus strain Western Reserve ( WR ) and the mutant viruses A36YdF , A36Y112F , A36Y132F and B5-YFP have been described previously [26] , [59] . A36-YFP , A36YdF-YFP , A36-YFP/B5-mRFP , B5P189S , B5P189S/A36YdF , A34K151E and A34K151E/A36YdF , A36YdF/B5-YFP and A36Y112F/B5-YFP were prepared as described previously [28] , [59] , [78] , [79] . Briefly , plasmids containing recombination cassettes with either the point mutations or fluorescent protein sequences flanked by left and right arm regions homologous to the desired area of insertion were prepared . BSC-1 cells were infected with VACV WR and transfected with the relevant vector using Lipofectamine ( Invitrogen ) , as described by the manufacturer , for production of A36-YFP , A36YdF-YFP , B5P189S and A34K151E , or infected with A36YdF or A36Y112F and transfected with the relevant vector for production of B5P189S/A36YdF , A34K151E/A36YdF , A36YdF/B5-YFP and A36Y112F/B5-YFP . For the other fluorescent viruses , BSC-1 cells were infected with A36-YFP and transfected with the relevant vector for production of A36-YFP/B5-mRFP , B5P189S/A36-YFP and A34K151E/A36-YFP , or infected with B5P189S or A34K151E and transfected with the A36YdF-YFP vector to produce B5P189S/A36YdF-YFP and A34K151E/A36YdF-YFP , respectively . At 24 hours post infection ( hpi ) cells were scraped and recombinant viruses purified by three rounds of plaque purification . Insertion at the correct locus and correct sequence was confirmed by PCR and sequencing . BSC-1 or NIH3T3 cells were seeded in six-well plates and grown to confluence . The virus strains were diluted in serum free DMEM ( SFM ) and approximately 25 PFU was added to each well . After incubation at 37°C in 5% CO2 for 1 h , the cells were washed twice and overlaid with either 1 . 5% carboxymethyl cellulose ( CMC ) in minimal essential medium ( MEM ) containing 2 . 5% FBS , 292 µg/ml L-glutamine , 100 units/ml penicillin and 100 µg/mL streptomycin for plaque assays , or DMEM-FPSG for comet assays . For experiments with imatinib treatment , imatinib mesylate ( ChemiTek ) at a final concentration of 10 µM or 0 . 01% of the carrier , DMSO , was included in the overlay . Cells were incubated for 3 days and then the overlay removed and cells stained with 1% crystal violet in methanol for visualisation . BSC-1 cells were seeded into 12-well dishes , and confluent monolayers were infected with either WR , A36YdF , A36Y112F or A36Y132F in triplicate at a multiplicity of infection ( MOI ) of 5 for 1 h . The inoculum was removed , and the cell monolayer was washed three times with PBS . Cells and supernatant were collected at various times post-infection ( 4 , 8 , 12 and 24 h ) , freeze-thawed three times and the virus concentration determined by plaque assay of 10-fold serial dilutions in triplicate on BSC-1 cells . Cells were incubated for 3 days and then the overlay removed and cells stained with 1% crystal violet in methanol for visualisation . 12-well dishes were seeded with BSC-1 or NIH3T3 cells and incubated with VACV strains ( in triplicate ) at an MOI of 0 . 1 for 1 h . Cells were then washed twice with PBS and DMEM-FPSG was added . For inhibitor treatments , imatinib mesylate at a final concentration of 10 µM , Cytochalasin D ( Sigma-Aldrich ) at a final concentration of 0 . 1 µg/ml or 0 . 01% of the carrier , DMSO , was added to the overlay media . The supernatants were collected at 16 hpi . To quantify the infectious EEV , plaque assays of 10-fold serial dilutions of the supernatant were performed on BSC-1 cells as described above . After 3 days cells were stained with methanol/1% crystal violet and plaques enumerated . All EEV assays were performed on at least three separate occasions . Confluent monolayers of BSC-1 cells were infected at an MOI of 5 and processed for TEM at 9 hpi . Briefly , cells were rinsed twice with PBS and fixed with 1 . 5% glutaraldehyde in Na-cacodylate buffer ( 0 . 1 M and 0 . 1 M sucrose ) at pH 7 . 4 for 1 h . The samples were subsequently washed with PBS and postfixed with 1% osmium tetroxide in 0 . 1 M Na-cacodylate at pH 7 . 4 for 1 h . The cells were next dehydrated in a graded ethanol series . After dehydration , samples were embedded in Epon and after hardening of the embedding medium , the plastic of the multiwell dishes was removed using liquid nitrogen . Sections of 120 nm under various angles were cut with a diamond knife , stained first with uranyl acetate , subsequently with lead citrate , and examined in a Jeol 2100 ( Jeol , Tokyo , Japan ) at 200 kV . BSC-1 cells were grown to 80% confluency on glass coverslips and infected with WR , A36Y112 and A36YdF and fixed 8 hpi with 3% paraformaldehyde ( PFA ) in cytoskeletal buffer ( CB ) [10 mM 2- ( N-morpholino ) ethanesulfonic acid ( MES ) buffer , 0 . 15 M NaCl , 5 mM ethylene glycol tetraacetic acid ( EGTA ) , 5 mM MgCl2 , 50 mM glucose , pH 6 . 1] for 10 minutes at room temperature . Cells were blocked in blocking buffer ( 1% BSA , 2% FBS in CB ) for 20 minutes then incubated for 40 minutes with 19C2 rat anti-B5 primary antibody ( 1/300 ) [80] . After three washes with PBS cells were incubated with AlexaFluor568 ( Invitrogen ) anti-rat secondary antibody ( 1/200 ) and AlexaFluor488 Phalloidin for 20 minutes . The coverslips were mounted on a glass slide with 0 . 3–1% ( w/v ) P-phenylenediamine ( PPD; Sigma-Aldrich ) in mowiol mounting media [10% ( w/v ) Polyvinyl Alcohol 4–88 ( Sigma-Aldrich ) , 25% ( w/v ) glycerol , 0 . 1 M Tris , pH 8 . 5] and imaged with an Olympus microscope BX51 with filter sets 31001 , 31002 and 31013v2 ( Chroma ) . The resulting images analysed with Photoshop CS3 ( Adobe ) . For 3D-SIM , BSC-1 cells grown to 80% confluency on glass coverslips were infected with the fluorescently-tagged viruses at an MOI of 1 . Transfection with plasmids expressing Lifeact-cerulean [79] was performed using Lipofectamine ( Invitrogen ) , as described by the manufacturer . Cells were fixed at 8 hpi with 3% PFA for 10 min at RT then washed with PBS . If required , cells were blocked with blocking buffer and then probed with 19C2 rat anti-B5 primary antibody ( 1/300 ) and AlexaFluor568 anti-rat secondary antibody ( 1/200 ) , diluted in the blocking buffer . The coverslips were mounted on a glass slide with 0 . 3–1% PPD in mowiol mounting media . Imaging was performed using a DeltaVision OMX 3D Structured Illumination Microscopy System ( OMX 3D-SIM , Applied Precision Inc . , Issaquah , USA ) , as described previously [81] , [82] . General image handling was undertaken with either Image J or Adobe Photoshop CS4 . HeLa cells at 80% confluency were grown on Ibidi glass bottom μ-Dishes coated in fibronectin ( 5 µg/cm2 , Sigma-Aldrich ) . Cells were infected with B5-YFP , A36YdF/B5-YFP or A36Y112F/B5-YFP and transfected with pE/L Lifeact-mRFP ( Lipofectamine-2000 , Invitrogen ) [79] . Immediately prior to imaging on Olympus FV1000 confocal microscope at 6–9 hours post infection , media was changed to Lebovitz's L-15 ( Invitrogen ) with 5% FBS . Cells showing distinct viral B5-YFP localisations and the appearance of defined actin structures were imaged for 60 frames over 2 mins . Images were analysed with Olympus Fluoview Viewer ( Ver . 03 . 1 ) and compiled with Adobe Photoshop CS4 . Plaque diameters were measured using Image J . Differences in EEV and plaque diameters were calculated using an unpaired t-test with Prism 5 . 0 ( Graph Pad Software ) software . To determine A36 distribution in 3D-SIM images , for each virus type 100 virus particles selected from five representative cells were classified as having polarised ( P , less than 20% of virus circumference ) , intermediate ( I , 20–80% of virus circumference ) or circular ( C , greater than 80% of virus circumference ) A36 distribution ( see Figure 6A ) . Classification was performed double-blinded . | Traversing the plasma membrane of the host cell is a significant challenge for many viruses during the infection cycle , and the efficiency of detachment from the host cell and subsequent release can have implications in pathogenesis . Vaccinia virus exits cells through the loss of an outer membrane but remains attached via viral envelope proteins that mediate adhesion between the cell and virus . Here we report that actin filament nucleation by the viral protein A36 promotes the disengagement of virus attachment . Viruses unable to locally induce actin nucleation displayed significantly reduced release and particles were found trapped in small pits at the plasma membrane . Mutations in luminal viral proteins that disrupt attachment identified an alternative route to virus release , bypassing the requirement for actin nucleation . Our results suggest that untethering virus attachment to the cell surface is a rate-limiting step during exocytic release of vaccinia virus . We have elucidated that the force of actin nucleation is the primary mechanism that operates to relieve these interactions . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"viral",
"envelope",
"molecular",
"cell",
"biology",
"viral",
"transmission",
"and",
"infection",
"virology",
"biology",
"microbiology",
"viral",
"structure",
"host-pathogen",
"interaction"
] | 2013 | A36-dependent Actin Filament Nucleation Promotes Release of Vaccinia Virus |
While major inroads have been made in identifying the genetic causes of rare Mendelian disorders , little progress has been made in the discovery of common gene variations that predispose to complex diseases . The single gene variants that have been shown to associate reproducibly with complex diseases typically have small effect sizes or attributable risks . However , the joint actions of common gene variants within pathways may play a major role in predisposing to complex diseases ( the paradigm of complex genetics ) . The goal of this study was to determine whether polymorphism in a candidate pathway ( axon guidance ) predisposed to a complex disease ( Parkinson disease [PD] ) . We mined a whole-genome association dataset and identified single nucleotide polymorphisms ( SNPs ) that were within axon-guidance pathway genes . We then constructed models of axon-guidance pathway SNPs that predicted three outcomes: PD susceptibility ( odds ratio = 90 . 8 , p = 4 . 64 × 10−38 ) , survival free of PD ( hazards ratio = 19 . 0 , p = 5 . 43 × 10−48 ) , and PD age at onset ( R2 = 0 . 68 , p = 1 . 68 × 10−51 ) . By contrast , models constructed from thousands of random selections of genomic SNPs predicted the three PD outcomes poorly . Mining of a second whole-genome association dataset and mining of an expression profiling dataset also supported a role for many axon-guidance pathway genes in PD . These findings could have important implications regarding the pathogenesis of PD . This genomic pathway approach may also offer insights into other complex diseases such as Alzheimer disease , diabetes mellitus , nicotine and alcohol dependence , and several cancers .
Complex diseases occur commonly in the population and are a major source of disability and death worldwide . They are thought to arise from multiple predisposing factors , both genetic and nongenetic , and joint effects of those factors are thought to be of key importance [1 , 2] . Parkinson disease ( PD ) serves as an example of a complex disease [3 , 4] . Other examples include Alzheimer disease , diabetes mellitus , nicotine and alcohol dependence , and several types of cancer [5] . While major inroads have been made in identifying the genetic causes of rare Mendelian disorders , little progress has been made in the discovery of common gene variations that predispose to complex diseases [6 , 7] . The single gene variants that have been shown to associate reproducibly with complex diseases typically have small effect sizes or attributable risks . However , the joint actions of common gene variants within pathways may play a major role in predisposing to complex diseases ( the paradigm of complex genetics ) , and the discovery of susceptibility genes and pathways may have sizeable public health benefits [8 , 9] . As early as 1997 , experimental studies inferred that genetic variability in the axon guidance pathway was a possible factor contributing to the cause of PD [10] . More recently , a high-resolution , whole-genome association study of PD highlighted the semaphorin 5A gene ( SEMA5A ) as containing the single nucleotide polymorphism ( SNP ) most significantly associated with PD susceptibility in that study [11] . SEMA5A maps to the deletion candidate interval for cri du chat syndrome , which is associated with severe abnormalities in brain development [12] . Semaphorin proteins play an important role in axon guidance and in the development of the mesencephalic dopamine neuron system during embryogenesis [13] . They interact with several other proteins from the axon guidance pathway to provide a complex and dynamic set of cues that either repel ( as for the semaphorins ) or attract axons toward their synaptic targets [14 , 15] . Indeed , several axon-guidance pathway proteins ( including ephrin and netrin and slit proteins and their ephrin and deleted in colorectal carcinoma and roundabout family receptors ) have been shown to also be important for dopamine axonal maintenance , regeneration , and target recognition [16 , 17] . Several experimental findings from the neurodevelopment literature and our preliminary genetic findings led us to hypothesize that while the main effects of a single gene such as SEMA5A may be small and of limited significance , the joint effects of multiple axon-guidance pathway genes may predispose to PD .
Of the 1 , 460 SNPs within brain-expressed genes of the axon-guidance pathway , 183 SNPs ( 12 . 5% ) were individually associated with susceptibility to PD . Table 1 contains results for the final model produced by running SNPs through the multistage process to predict PD susceptibility . This model used data from 442 matched PD patients/sibling controls ( one pair was missing data on one or more SNPs ) . The odds ratios ( ORs ) ( 95% confidence intervals [CIs] ) for the groups defined by predicted PD probability of <0 . 25 , 0 . 25–0 . 50 , 0 . 50–0 . 75 , and >0 . 75 were as follows: 1 ( reference ) , 4 . 58 ( 2 . 27–9 . 23 ) , 15 . 42 ( 6 . 19–38 . 45 ) , and 90 . 76 ( 32 . 60–252 . 67 ) respectively . Since we were interested in the significance of the pathway , rather than individual SNPs , the p value for the overall model was of primary importance . In this case , the model had an overall p value of 4 . 64 × 10−38 ( 95% CI 6 . 94 × 10−28 − 5 . 39 × 10−40 ) . This model significantly predicted whether or not an individual was a case or an unaffected sibling . The predicted probabilities of PD were high ( towards 1 ) for most of the cases , and low ( towards 0 ) for most of the unaffected siblings ( Figure 1 ) . Indeed 35% of the cases had predicted probabilities above 0 . 9 , and 34% of unaffected siblings had predicted probabilities below 0 . 1 . However , as shown by Figure 1 , the model did not completely distinguish the two groups; some cases had low predicted probabilities , and some controls had high predicted probabilities . The concordance for the model was about 0 . 70 , indicating good but not complete agreement between predicted and observed case/sibling status . Of the 1 , 460 SNPs , 175 ( 12 . 0% ) were individually associated with survival free of PD ( hazard function ) using Cox proportional hazards models , as detailed in Table S1 . Table 2 contains results for the final proportional hazards model produced by running SNPs through the multistage process to predict survival free of PD . This model used data from 400 PD patients ( 43 patients were missing data on one or more SNPs ) . In this case , the model had an overall p value of 5 . 43 × 10−48 ( 95% CI 3 . 19 × 10−37 − 2 . 36 × 10−61 ) . By contrast , the model was not significant at predicting survival ( age at study ) of the matched sibling controls ( p = 0 . 73 ) . This last finding suggests that the model predicts survival free of PD ( hazard function ) , but not survival in general , and that the model is specific for PD cases . Figure 2 shows a Kaplan-Meier plot to describe the results of the model . The groups were formed by calculating a risk score for each PD patient using the equation from the proportional hazards model , then categorizing the score at the 25th ( Q1 ) , 50th ( Q2 ) , and 75th ( Q3 ) percentiles . The survival curves separated nicely right from the earliest ages of onset . By age 60 , only 9% of PD patients in the predicted highest risk group were still free of PD , whereas 89% of PD patients in the predicted lowest risk group were free of PD . By age 70 , none of the PD patients in the predicted highest risk group were still free of PD , whereas 66% of PD patients in the predicted lowest risk group were free of PD . The median ages at onset for each group , from lowest risk group to highest risk group , were 72 . 1 , 66 . 5 , 59 . 7 , and 51 . 7 , a difference in survival free of PD of more than 20 years from lowest to highest . The concordance for this model was 0 . 76 . The hazards ratios ( HRs ) ( 95% CIs ) for the four groups , from lowest to highest risk , were 1 ( reference ) , 2 . 96 ( 2 . 14–4 . 07 ) , 6 . 91 ( 4 . 87–9 . 81 ) , and 19 . 04 ( 13 . 11–27 . 65 ) . Of the 1 , 460 SNPs , 160 ( 11 . 0% ) were individually associated with age at onset of PD using linear regression models , as detailed in Table S1 . Table 3 contains results for the final model produced by running SNPs through the multistage process to predict PD age at onset . This model used data from 395 PD patients ( 48 patients were missing data on one or more SNPs ) . In this case , the model had an overall p value of 1 . 68 × 10−51 ( 95% CI 3 . 56 × 10−42 − 1 . 16 × 10−51 ) . By contrast , the set of SNPs was not significant at predicting age at study of the matched sibling controls ( p = 0 . 34 ) . This last finding suggests that the model predicts age at onset of PD , not age at the time of the study , and that the model is specific for PD cases . Figure 3 shows a plot of predicted age at onset versus reported age at onset to summarize the results of the model . The plot showed a nice elliptical pattern , reflecting the model R2 of 0 . 683 ( 95% CI 0 . 62–0 . 69 ) . The model explained about 68% of the variability in age at onset of PD . Figure S1 shows the distributions of the test statistics from the models with randomly selected SNPs and the values of the test statistics from our final models for PD susceptibility ( A ) , survival free of PD ( B ) , and age at onset of PD ( C ) . The test statistics from our final models were in every case much greater than those observed from the other models . Other combinations of SNPs from the axon guidance pathway also performed quite well in predicting PD susceptibility , survival free of PD , and age at onset of PD . Although the models reported in this manuscript provided good fits to our data , our results do not preclude other combinations of axon-guidance pathway SNPs as significant predictors of PD . The SNPs in the final models that we selected showed no significant linkage disequilibrium in unaffected siblings . We also mined a second available whole-genome association dataset for PD , to determine whether the genes in each of the predictive genetic models in our primary whole-genome association dataset were also predictive of the same PD outcomes in the secondary dataset ( same genes and outcomes , different SNPs and samples ) . Details regarding the participants and genotyping procedures for that study have been recently reported [18] . The secondary whole-genome association dataset included 1 , 195 SNPs in the 22 genes from the model predicting PD susceptibility in our primary dataset . Of those SNPs , 127 ( 10 . 6% ) were individually associated with susceptibility to PD , as detailed in Table S2 . Table S3 contains results for the final model produced by running SNPs through the multistage process to predict PD susceptibility . This model used data from 528 individuals ( 264 PD patients and 264 unrelated controls; eight individuals were missing data on one or more SNPs ) . The model had an overall p value of 3 . 93 × 10−44 . The ORs ( 95% CIs ) for the groups defined by predicted PD probability of <0 . 25 , 0 . 25–0 . 50 , 0 . 50–0 . 75 , and >0 . 75 were as follows: 1 ( reference ) , 7 . 86 ( 3 . 94–15 . 71 ) , 16 . 14 ( 8 . 13–32 . 05 ) , and 121 . 14 ( 56 . 63–259 . 14 ) , respectively . The predicted probabilities of PD were high ( towards 1 ) for most of the cases and low ( towards 0 ) for most of the unaffected siblings ( Figure S3 ) . Indeed 38% of the cases had predicted probabilities above 0 . 9 , and 39% of unaffected siblings had predicted probabilities below 0 . 1 . However , as shown by Figure S3 , the model did not completely distinguish the two groups; some cases had low predicted probabilities , and some controls had high predicted probabilities . The concordance for the model was 0 . 90 . The secondary whole-genome association dataset included 1 , 411 SNPs in the 26 genes from the model predicting age at onset of PD in our primary dataset . Of those SNPs , 142 ( 10 . 1% ) were individually associated with survival free of PD ( hazard function ) using Cox proportional hazards models , as detailed in Table S2 . Table S4 contains results for the final proportional hazards model produced by running SNPs through the multistage process to predict survival free of PD . This model used data from 263 PD patients ( five patients were missing data on one or more SNPs ) . In this case , the model had an overall p value of 6 . 30 × 10−35 . However , the model was not significant at predicting survival ( age at study ) of the matched sibling controls ( p = 0 . 14 ) . Figure S4 uses a Kaplan-Meier plot to describe the results of the model . The groups were formed by calculating a risk score for each PD patient using the equation from the proportional hazards model , then categorizing the score at the 25th ( Q1 ) , 50th ( Q2 ) , and 75th ( Q3 ) percentiles . The survival curves separated nicely right from the earliest ages of onset . By age 60 , only 41% of PD patients in the predicted highest risk group were still free of PD , whereas 94% of PD patients in the predicted lowest risk group were free of PD . By age 70 , none of the PD patients in the predicted highest risk group were still free of PD , whereas 68% of PD patients in the predicted lowest risk group were free of PD . The median ages at onset for each group , from lowest risk group to highest risk group , were 74 , 68 , 63 , and 60 , a difference in survival free of PD of 14 years from lowest to highest . The concordance for this model was 0 . 78 . The HRs ( 95% CIs ) for the four groups , from lowest to highest risk , were 1 ( reference ) , 4 . 02 ( 2 . 62–6 . 16 ) , 10 . 69 ( 6 . 72–16 . 98 ) , and 21 . 05 ( 12 . 85–34 . 48 ) . The secondary whole-genome association dataset included 1 , 605 SNPs in 28 of the 29 genes from the final model predicting age at onset of PD in our primary dataset . Of those SNPs , 157 ( 9 . 8% ) were individually associated with age at onset of PD using linear regression models , as detailed in Table S2 . Table S5 contains results for the final model produced by running SNPs through the multistage process to predict PD age at onset-squared . This model used data from 265 PD patients ( three patients were missing data on one or more SNPs ) . In this case , the model had an overall p value of 4 . 72 × 10−40 . However , the set of SNPs was not significant at predicting age at study-squared of the matched sibling controls ( p = 0 . 527 ) . Figure S5 shows a plot of predicted age at onset versus reported age at onset to summarize the results of the model . The plot showed a nice elliptical pattern , reflecting the model R2 of 0 . 714 . The model explained about 71% of the variability in age at onset of PD . We also mined an available gene-expression profiling dataset for PD that considered 21 different brain regions . Details regarding the participants , biological samples , and microarray experiments have been recently reported [19] . For this study , we limited our data analyses to the substantia nigra and the striatum ( putamen and caudate nuclei ) , since these are the brain regions contributing most significantly to the nigrostriatal dopamine deficiency that is characteristic of PD and since we defined our three PD outcomes according to the corresponding motor phenotype . There was a total of 45 genes represented by the SNPs listed for the three predictive genetic models ( Tables 1–3 ) , and the gene expression dataset had informative probe sets for 32 of those genes in the substantia nigra , 34 in the putamen , and 35 in the caudate . Table S6 provides a detailed listing of the multiregional expression data for the 45 genes . In each region , there were more differentially expressed genes observed than expected by chance: substantia nigra , seven observed ( 22% ) versus 1 . 6 expected ( 5% ) ; putamen , five observed ( 15% ) versus 1 . 7 expected ( 5% ) ; and caudate , five observed ( 14% ) versus 1 . 8 expected ( 5% ) . Overall , 36 genes had data in at least one of the three regions , and 14 ( 39% ) of those were differentially expressed in at least one region . Figure 4 provides a detailed summary of the multiregional expression data for the 45 axon-guidance pathway genes that were predictive of PD outcomes . We observed no significant differences in the number of probe sets for differentially expressed versus normally expressed genes , except for the putamen where two genes with eight fragments each were both differentially expressed ( unpublished data ) . Removing those two genes ( outliers ) removed the significance ( no differences in the median numbers of informative probe sets ) . We found very similar results from our sensitivity analyses treating accurate-type and cross-reacting type fragments equally . Again , in each region there were more differentially expressed genes than expected by chance . Restricting to the 45 genes represented in the three predictive genetic models , the results were: substantia nigra , seven observed ( 22% ) versus 1 . 6 expected ( 5% ) ; putamen , five observed ( 15% ) versus 1 . 7 expected ( 5% ) ; and caudate , seven observed ( 20% ) versus 1 . 8 expected ( 5% ) . Overall , of the 36 genes with data in at least one of the three regions , 14 ( 39% ) were still differentially expressed in at least one region . Finally , we renormalized and reanalyzed the raw data from the original gene expression profiling study using a different standard software package , to obtain results for all probe sets in all 45 genes and in all three regions . We again observed differential expression of axon-guidance pathway genes in PD , although the findings were more modest . A total of 13 genes were differentially expressed in at least one region ( eight in the substantia nigra , one in the putamen , and four in the caudate ) . For all analyses of differential gene expression , we were able to code the differential expression unambiguously as increased or reduced . For genes with multiple informative and differentially expressed probe sets , the direction of effect was always the same .
In summary , we employed a genomic pathway approach to determine whether polymorphism in a candidate pathway ( axon guidance ) predisposed to a complex disease ( PD ) . We found that multiple SNPs in axon-guidance pathway genes were strong predictors of PD susceptibility , survival free of PD , and age at onset of PD in two independent whole-genome association datasets; and many axon-guidance pathway genes were differentially expressed in PD . Although an exact replication of the predictive genetic models remains to be done ( same SNPs , independent samples ) , we consider these findings timely and potentially important . To date , predictive biomarkers for PD are lacking . Our genetic findings for the axon guidance pathway and PD might suggest a new environmental focus on exposures that occur during intrauterine life ( miswiring hypothesis ) . Our findings would also provide evidence that complex diseases such as PD can be due to the joint effects of many genes that , taken singly , would show only small effects ( additive effects model and epistasis ) [40–43] . While the available datasets were exploratory and may have been underpowered to detect the main effects of single SNPs or probe sets , our findings are consistent with recent studies that report greater statistical power to detect the joint effects of multiple loci than the main effects of single loci [44 , 45] . By contrast to familial aggregation and twin studies , our findings demonstrate that a largely sporadic disease such as PD can in fact have a strong genetic component [4 , 46 , 47] and suggest that a similar genomic pathway approach might provide insights into several other complex diseases . Finally , these findings provide important insights regarding the molecular mechanisms that may control dopamine circuit formation and programmed cell death in healthy versus diseased individuals . This in turn may facilitate the development of treatments for axonal synaptic regeneration and repair and for neuroprotection in PD [17 , 37] . We envision that a systematic study of all defined genomic pathways will yield additional important findings for PD , even in the absence of prior evidence of strong single-gene association .
To formally test our hypothesis , we consulted the Kyoto Encyclopedia of Genes and Genomes ( KEGG ) [48–50] . The KEGG PATHWAY database is a bioinformatics resource that provides wiring diagrams of molecular interactions , reactions , and relations . There are at least 270 pathways in KEGG related to Homo sapiens and diseases . This includes a detailed summary of the axon guidance pathway , updated as recently as October 3 , 2005 ( http://www . genome . jp/dbget-bin/www_bget ? path:hsa04360 ) . We identified all of the genes that encoded proteins within the KEGG axon guidance pathway via Entrez Gene ( http://www . ncbi . nlm . nih . gov/entrez/query . fcgi ? db=Gene ) and consulted the UniGene database ( http://www . ncbi . nlm . nih . gov/entrez/query . fcgi ? db=unigene ) to determine which of the genes were expressed in the human brain ( n = 128 ) . We then mined an available whole-genome association dataset for PD to identify those SNPs that were genotyped in brain-expressed , axon-guidance pathway genes as part of that study [11] . Specifically , in that study , ( which we refer to as the primary whole-genome association dataset ) , we had individually genotyped 198 , 345 genomic SNPs that were uniformly spaced ( one per 12 kb average gap distance ) and informative in 443 sibling pairs discordant for PD . This included 1 , 460 SNPs within 117 axon-guidance pathway genes expressed in the brain . All statistical tests were two tailed and considered significant at the conventional alpha level of 0 . 05 . All statistical analyses were performed in SAS version 9 . 1 ( SAS Institute , http://www . sas . com ) or S-Plus version 7 ( Insightful Corporation , http://www . insightful . com ) . We considered three outcomes of interest: ( 1 ) PD susceptibility , ( 2 ) survival free of PD , and ( 3 ) age at onset of PD . We sought to identify joint action models of SNPs from the axon guidance pathway that predicted each of the three outcomes . For the first outcome , we used conditional logistic regressions stratified on sibship to examine associations of the SNPs with PD susceptibility while adjusting for age and gender [51] . For each SNP , we calculated ORs , 95% CIs , and p values . Goodness-of-fit was assessed through measuring concordance and visually through histograms of predicted probabilities [52] . We estimated overall ORs by categorizing the predicted probability of PD from the model into four groups ( <0 . 25 , 0 . 25–0 . 50 , 0 . 50–0 . 75 , and >0 . 75 ) and then calculating the ORs for each group relative to the <0 . 25 group . We used a likelihood ratio test to assess the significance of the overall model and calculated a 95% bias-corrected bootstrap CI for the associated p value using 10 , 000 resamples . For the second outcome , we used Cox proportional hazards models to test for associations of the SNPs with survival free of PD [53] . For each SNP , we calculated HRs , 95% CIs , and p values . Concordance was again calculated for the proportional hazards models , and Kaplan-Meier plots of categorized scores predicting risk of PD were generated to provide visual gauges for goodness-of-fit [54] . We also calculated HRs for risk groups categorized at the quartiles , using the lowest risk group as reference . We used a likelihood ratio test to assess the significance of the overall model and calculated a 95% bias-corrected bootstrap CI for the associated p value using 10 , 000 resamples . For the third outcome , we predicted the reported age at onset of PD using multiple regression models [55] . Goodness-of-fit was described through the model R2 values and plots of the predicted versus observed ages at onset . We used an F test to assess the significance of the overall model and calculated a 95% bias-corrected bootstrap CI for the associated p value and the R2 using 10 , 000 resamples . Assumptions were tested throughout . Only the linear regression models required a data transformation; age at onset-squared was used as the outcome to meet the required normality assumption . We performed tests of linkage disequilibrium in unaffected siblings for the SNPs in the final models for the three outcomes using LDSELECT version 1 . 0 ( copyright 2004 by Deborah A . Nickerson , Mark Rieder , Chris Carlson , and Qian Yi , University of Washington , United States ) with a threshold R2 of 0 . 80 . Figure S2 summarizes the scheme used to develop models for each outcome . Since the modes of expression of the alleles in the SNPs of interest were not known , we first looked at each SNP using three coding schemes: log-additive , Mendelian dominant , and Mendelian recessive ( Step 1 ) . We simplified subsequent analyses by removing from further consideration those SNPs with no significant main effects in any coding scheme ( Step 2 ) . Removing those SNPs was a conservative approach , potentially biasing our tests towards the null hypotheses , since the SNPs were prevented from possibly entering the joint action models after adjustment for other variables . We generally coded the remaining SNPs using the schemes that produced the smallest p values , since these provided our best estimates of the modes of expression in our data ( Step 3 ) . For each outcome , we then created multiple sets of SNPs , where each set contained only SNPs with at least a certain number of non-missing values ( Step 4 ) . This was done to address issues due to missing values . While most SNPs had fairly complete data , others had missing values from substantial numbers of participants: up to 18% of cases and up to 32% of case-sib pairs . We therefore chose an approach where we constructed candidate models using sets of SNPs with fairly complete data ( effective sample sizes close to the maximum 443 ) to explain as much of the outcomes as possible , then checked to see if adding other SNPs on top of the candidate models would contribute significantly . We constructed the candidate models for each set using standard automated procedures ( Step 5 ) and selected a final candidate model for each outcome based on significance and goodness-of-fit ( Step 6 ) . We then added other SNPs , which were significant given the candidate models ( Step 7 ) and significant pair-wise interactions ( Step 8 ) . To compare the significance of our axon-guidance pathway SNP models to the significance of randomly selected genomic SNP models , we constructed 4 , 000 models for each outcome by randomly selecting the appropriate number of SNPs from the entire available dataset . We then plotted the distributions of the test statistics from those models and the values of the test statistics from our final models . A second whole-genome association study of PD was recently published , including first stage analysis and public release of the data [18] . That study included 276 patients with PD and 276 neurologically normal and unrelated controls . The samples used for that study were derived from the National Institute of Neurological Disorders and Stroke Neurogenetics repository hosted by the Coriell Institute for Medical Research ( https://queue . coriell . org/Q/snp_index . asp ) . There were 408 , 803 SNPs individually genotyped , and call rates and Hardy-Weinberg equilibrium p values were previously reported [23] . We downloaded the individual level data for that study from the Coriell Institute website , and we identified SNPs in that secondary dataset that were assigned to the genes represented by SNPs in each of the predictive genetic models for our primary dataset ( same genes and outcomes , different SNPs and samples ) . We then employed the same statistical methods to construct predictive genetic models for the same outcomes in the secondary dataset , with two exceptions . First , because the secondary dataset employed unrelated controls that were not individually matched , we performed unconditional logistic regression analyses instead of conditional logistic regression analyses for the PD susceptibility outcome . Second , we restricted the age at onset of PD analyses in the secondary dataset to exclude participants with ages younger than those reported in the primary dataset , which resulted in the exclusion of one participant whose age at onset was reportedly 13 years . We explored an available gene-expression profiling dataset to determine if there was convergence of those functional data with the genetic association data and models [19] . That study included multiregional gene expression data from postmortem brain specimens from 22 PD cases and 23 normal aged brain donors and represents the most comprehensive expression profiling study of PD to date ( largest numbers of participants , brain regions studied , and genes assayed ) [56] . Very strict RNA quality control criteria were used . We analyzed data derived from Affymetrix Human Genome U133 Plus 2 . 0 GeneChip arrays , which included probe set data for 126 of the 128 brain-expressed axon-guidance pathway genes that we initially identified ( see Bioinformatic methods ) . We analyzed probe set data for the substantia nigra , putamen , and caudate regions using methods and criteria similar to the published study [19] . Supporting information regarding the study design , expression values calculations , and array normalization methods for the published study are also provided in Text S1 . For the differential expression analyses of this study , we identified probe sets from within the original dataset that were assigned to the axon-guidance pathway genes of interest ( those represented by SNPs in either of the three predictive genetic models in the primary whole-genome association dataset ) . For each probe set we compared expression for cases and controls in each of the three nigrostriatal regions ( Table S6 ) . We considered a probe set informative in a given region if it was expressed in at least 75% of cases or 75% of controls . Although differential expression of a single probe set is most of the times enough to characterize a gene as differentially expressed ( multiple polyadenylation sites are represented on Affymetrix gene chips to account for multiple gene transcripts ) , for this study we employed a more conservative definition of differential gene expression than for the original study [19] . We defined a gene as differentially expressed in a given region if at least one accurate-type ( at ) probe set assigned to the gene had a t-test p value <0 . 05 and absolute value of the fold expression ≥1 . 3 . Alternatively , we defined a gene as differentially expressed in a given region if at least 30% of accurate-type ( at ) or cross reacting-type ( x_at , s_at ) probe sets assigned to the gene had t-test p values <0 . 05 and absolute values of the fold expression ≥1 . 3 . These definitions weigh the significance of cross-reactive type probe sets lower than accurate-type probe sets because they have less specificity , and also account for the possibility that multiple probe sets for the same gene may not provide concordant gene expression measurements [57] . We also performed a sensitivity analysis where a gene was defined as differentially expressed in a given region if at least one probe set of any type assigned to the gene had a t-test p value <0 . 05 and absolute value of the fold expression ≥1 . 3 . This was consistent with the differential expression analyses performed for the original study [19] . Finally , we performed a second sensitivity analysis whereby we employed the same CEL file data as for the original gene expression profiling study [18] , and we renormalized and reanalyzed the data using a second standard software package ( GeneSpring , Agilent Technologies , http://www . home . agilent . com ) . That normalization procedure allowed us to obtain informative results for all probe sets and in all 45 genes . Additional details of this sensitivity analysis are provided as supporting information ( Text S1 ) . For our primary differential expression analyses , we tested for possible bias due to the number of informative probe sets per gene by comparing the distributions for differentially expressed versus normally expressed genes using the Wilcoxon rank sum test . For the primary and two sensitivity analyses , we coded the differential expression of genes as increased or reduced if all probe sets assigned to the gene had the same direction of effect or ambiguous if the probe sets had opposite directions of effect .
The National Center for Biotechnology Information Entrez Gene website ( http://www . ncbi . nlm . nih . gov/entrez/query . fcgi ? db=gene ) accession numbers ( GeneIDs ) for the genes named in the paper include: ABL1 ( 25 ) , ABLIM2 ( 84448 ) , CDC42 ( 998 ) , CHP ( 11261 ) , CXCR4 ( 7852 ) , DCC ( 1630 ) , DPYSL2 ( 1808 ) , EFNA5 ( 1946 ) , EPHA4 ( 2043 ) , EPHA8 ( 2046 ) , EPHB1 ( 2047 ) , EPHB2 ( 2048 ) , FYN ( 2534 ) , GNAI3 ( 2773 ) , GSK3B ( 2932 ) , KRAS ( 3845 ) , MRAS ( 22808 ) , NFATC2 ( 4773 ) , NFATC4 ( 4776 ) , NTNG1 ( 22854 ) , PAK1 ( 5058 ) , PAK3 ( 5063 ) , PAK4 ( 10298 ) , PAK6 ( 56924 ) , PAK7 ( 57144 ) , PLXNA2 ( 5362 ) , PLXNC1 ( 10154 ) , PPP3CA ( 5530 ) , RAC2 ( 5880 ) , ROBO1 ( 6091 ) , ROBO2 ( 6092 ) , ROCK2 ( 9475 ) , RRAS2 ( 22800 ) , SEMA3A ( 10371 ) , SEMA3D ( 223117 ) , SEMA3E ( 9723 ) , SEMA4D ( 10507 ) , SEMA5A ( 9037 ) , SLIT2 ( 9353 ) , SLIT3 ( 6586 ) , SRGAP1 ( 57522 ) , SRGAP3 ( 9901 ) , UNC5A ( 90249 ) , UNC5C ( 8633 ) , and UNC5D ( 137970 ) . | Complex diseases are common disorders that are believed to have many causes . Examples include Alzheimer disease , diabetes mellitus , nicotine and alcohol dependence , and several cancers . This study represents a paradigm shift from single gene to pathway studies of complex diseases . We present the example of Parkinson disease ( PD ) and a complex array of chemical signals that wires the brain during fetal development ( the axon guidance pathway ) . We mined a dataset that studied hundreds of thousands of DNA variations ( single nucleotide polymorphisms [SNPs] ) in persons with and without PD and identified SNPs that were assigned to axon-guidance pathway genes . We then identified sets of SNPs that were highly predictive of PD susceptibility , survival free of PD , and age at onset of PD . The effect sizes and the statistical significance observed for the pathway were far greater than for any single gene . We validated our findings for the pathway using a second SNP dataset for PD and also a dataset for PD that studied RNA variations . There is prior evidence that the axon guidance pathway might play a role in other brain disorders ( e . g . , Alzheimer disease , Tourette syndrome , dyslexia , epilepsy , and schizophrenia ) . A genomic pathway approach may lead to important breakthroughs for many complex diseases . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods",
"Supporting",
"Information"
] | [
"developmental",
"biology",
"cell",
"biology",
"neurological",
"disorders",
"neuroscience",
"homo",
"(human)",
"genetics",
"and",
"genomics"
] | 2007 | A Genomic Pathway Approach to a Complex Disease: Axon Guidance and Parkinson Disease |
Recent work has shown that functional connectivity among cortical neurons is highly varied , with a small percentage of neurons having many more connections than others . Also , recent theoretical developments now make it possible to quantify how neurons modify information from the connections they receive . Therefore , it is now possible to investigate how information modification , or computation , depends on the number of connections a neuron receives ( in-degree ) or sends out ( out-degree ) . To do this , we recorded the simultaneous spiking activity of hundreds of neurons in cortico-hippocampal slice cultures using a high-density 512-electrode array . This preparation and recording method combination produced large numbers of neurons recorded at temporal and spatial resolutions that are not currently available in any in vivo recording system . We utilized transfer entropy ( a well-established method for detecting linear and nonlinear interactions in time series ) and the partial information decomposition ( a powerful , recently developed tool for dissecting multivariate information processing into distinct parts ) to quantify computation between neurons where information flows converged . We found that computations did not occur equally in all neurons throughout the networks . Surprisingly , neurons that computed large amounts of information tended to receive connections from high out-degree neurons . However , the in-degree of a neuron was not related to the amount of information it computed . To gain insight into these findings , we developed a simple feedforward network model . We found that a degree-modified Hebbian wiring rule best reproduced the pattern of computation and degree correlation results seen in the real data . Interestingly , this rule also maximized signal propagation in the presence of network-wide correlations , suggesting a mechanism by which cortex could deal with common random background input . These are the first results to show that the extent to which a neuron modifies incoming information streams depends on its topological location in the surrounding functional network .
Networks of cortical neurons transmit and compute information . Fortunately , recent research has improved our understanding of both of these tasks [1–6] . Progress in these two areas lets us now ask how information modification ( computation ) depends on the number of connections a neuron receives ( in-degree ) or sends out ( out-degree ) . This is an important question because almost nothing is currently known about how a neuron’s topological location in a complex network affects how it computes . Connectivity has been widely studied in the brain at the macroscopic scale ( see [1 , 2] for reviews ) and at the microcircuitry or cellular level ( e . g . [7–12] ) . At the cellular level , several studies have examined clustering in cortical networks and found that the networks are non-random [13–16] . In addition , several other studies have found or discussed the implications of a log-normal distribution of connection weights in networks of cortical neurons [17–22] . Finally , a few studies have provided evidence that the distribution of the number of connections made by a neuron ( degree distribution ) at the cellular level in hippocampus [23] and cortex [15 , 24] is heavy-tailed , which indicates the presence of neurons with many more connections than expected in a random network . These previous studies have provided valuable insights about information transmission between cortical neurons and they emphasize the need to better understand the structure of these networks . Along these same lines , we recently used Transfer Entropy ( TE ) [25]–a well-established information theoretic method for detecting connectivity between neural sources [5 , 12 , 26–43]–to measure the time-scale dependent effective connectivity among hundreds of neurons in cortico-hippocampal slice cultures [44] . This analysis indicated the presence of time scale dependency in hub neurons , physical separation between connected neurons , and network modularity . The activity of these neurons was recorded at high temporal resolution ( 20 kHz ) using a high density 512-electrode array ( 60 μm electrode spacing ) . This preparation and recording method combination yielded high neuron numbers at high temporal resolution , beyond what is currently available in any in vivo recording system . The temporal recording resolution of 50 μs was small enough to resolve synaptic delays of 1–20 ms that are typically found in cortex [45 , 46] . The interelectrode spacing of 60 μm also improved the likelihood of detecting synaptically connected neurons , which most often share contacts within a 200 μm radius [13 , 14] . Using the same data , in this work we sought to go beyond pairwise connectivity to examine the relationship between connectivity and computation among groups of individual neurons using newly developed tools from information theory . It is a widely held position that neurons perform computations [5 , 47–54] . However , the precise definition of “computation” tends to vary throughout neuroscience and other fields that study systems that compute . Intuitively , computation involves combining information from different sources and producing some type of output . In contrast , communication or information transmission ( the basis of many information theoretic neural connectivity studies ) intuitively involves only passing information from a source to a receiver ( see Materials and Methods – Information Theory Terminology ) . As a first step , in this novel analysis , we chose to use a recently introduced multivariate information theoretic tool ( the Partial Information Decomposition ( PID ) [3 , 4] ) to quantify the amount of information computed by a neuron about the firing states of other neurons that send it connections . We are aware of no other studies that have used this new tool to analyze computation among individual neurons . We chose to use the PID to measure computation because it allowed us to quantify in an information theoretic sense the amount of information processed by a neuron about the states of input neurons in distinct parts . One of these parts ( the “synergy” ) quantifies the bonus information processed by the receiver based on the non-overlapping information from both inputs simultaneously . Similar to [55] , we feel it is natural to interpret the PID synergy as a measure of computation . We wish to emphasize that other definitions of “computation” exist and that other researchers have produced valuable and important studies using alternative definitions and analysis methodologies . We wish to make no claims that the PID synergy is the best or only measure of computation . Rather , we feel it is particularly well suited to analyzing computations performed by groups of neurons . Because the PID is an information theoretic analysis tool , it is able to capture linear and nonlinear interactions . Furthermore , unlike mutual information ( which is a measure of information transmission or communication ) or entropy ( which measures the amount of information contained in an individual variable ) [56] , the PID is able to quantify the amount of information a neuron computed based on simultaneous inputs from other neurons . In this analysis , we sought to determine if a neuron’s topological location in a complex network affected how it computes . To do so , we investigated if there were correlations between neuron degree and computation . Specifically , we addressed two main questions . First , did the in-degree of a neuron affect how much information it computed ? For instance , did neurons that received many connections compute more information than neurons that received few connections ? Second , did the out-degree of a neuron affect how much information was computed by the receivers of those connections ? For instance , did neurons that received connections from high out-degree neurons tend to compute more information ? In addition to these two main questions , we also examined the connection strength distribution and the degree distribution in our networks for comparison to previous results . Furthermore , we examined the role higher-order computations played in the network and we developed a simple feedforward network model to explore the relationship between computation and degree . Using advanced tools from information theory and our high quality recordings of individual cortical neurons , we found the following results . ( 1 ) In agreement with previous studies , we found a roughly log-normal distribution of TE connection strengths and a heavy-tailed degree distribution . ( 2 ) The in-degree was independent of the amount of information computed by a neuron . Conversely , neurons that received connections from high out-degree neurons tended to compute more information than neurons that received connections from low out-degree . ( 3 ) Though higher-order computations are difficult to measure , we found evidence that higher-order computations did not dominate high in-degree neurons . ( 4 ) Using a simple feedforward network model as an illustrative example , we found that a degree-modified Hebbian model best matched computation/degree correlation results from the real data and simultaneously maximized signal propagation in the presence of network-wide correlations . Though the simplicity of our model implies it largely serves as a guide to future research , the model results do connect the issues of network topology and computation with the frequently discussed topic of signal propagation in correlated and noisy networks [57–63] . Previous versions of this work were presented at conferences in abstract form [64–66] .
All neural tissue samples from animals were prepared according to guidelines from the National Institutes of Health and all animal procedures were approved by the Indiana University Animal Care and Use Committee ( Protocol: 12–015 ) as well as the Animal Care and Use Committee at the University of California , Santa Cruz ( Protocol: Litka1105 ) . A general overview of the analysis is presented in Fig 1A . The raw spiking data utilized in this analysis are fully described elsewhere [24 , 44] . Briefly , cortico-hippocampal organotypic cultures were produced using postnatal day 6 Black 6 mouse pups ( wild-type C57BL/6 from Charles River ) following the protocol described in [67] . The mice were anesthetized in an ice bath prior to decapitation and brain removal . Each culture was recorded after 2 to 4 weeks . After culturing , spontaneous activity was recorded from each slice using a custom made 512-electrode array system [68] . The array contained 5 μm diameter flat electrodes arranged in a triangular lattice with an inter-electrode distance of 60 μm . In this arrangement , the total recording area of the array was approximately a 0 . 9 mm by 1 . 9 mm rectangle . Though recordings from both hippocampus and cortex were performed [24 , 44] , only recordings from cortex were used in this analysis ( number of recordings: 25 , recording region: somatosensory cortex ) . Action potentials ( spikes ) were then detected and spike-sorted using a well-established method ( PCA using spike waveforms from seven adjacent electrodes ) [24 , 44 , 68] . Duplicate neurons and neurons with many refractory period violations were then excluded from further analysis ( see [68] for additional details ) . After spike sorting , neurons with less than 100 spikes in the 60 minute recording ( firing rate < 0 . 028 Hz ) were removed from the analysis . Spike sorting yielded spike times for each neuron ( 7735 total neurons ) with a resolution of 50 μs . The average firing rate for each neuron was 2 . 10 Hz and each recording had an average of 309 . 4 neurons . One central purpose of information theory is to quantify information in spite of the natural ambiguity we associate with the concept of information . Throughout this paper , we have attempted to clearly articulate–both conceptually and mathematically–the information theoretic quantities we employed . Briefly , we wish to explicitly state and relate several important terms as we will use them . We will provide a full mathematical description of the measures in subsequent sections . Following the detection of action potentials , transfer entropy ( TE ) [25] was used to measure the effective connectivity between each pair of neurons . The complete procedure for the TE analysis is described in [44] , though we provide a brief overview below . Several other methods of measuring information transfer or causality have been proposed in the past , including Granger Causality [69–77] , Dynamic Causal Modeling [78–80] , and directed information [81 , 82] . We chose to use TE because it has been widely used in neuroscience [5 , 26–43] , it is model independent , it is capable of detecting nonlinear interactions , it is well suited to spiking data due to the discrete nature of spike trains and the need for discrete probability distributions in TE , and because it quantifies interactions in the general units of bits , which allows for straightforward comparisons between different systems ( see [44] for more details ) . Herein , we used a subset of TE results produced in the previous analysis ( see below ) . In general , TE measures the amount of information the past state of one time series ( call it JP ) provides about the future state of another time series ( call it IF ) conditioned upon the past state of the receiver time series ( call it IP ) . In our case , the time series were spike trains and the states were spiking or not spiking . Before defining the TE , it is first necessary to define mutual information ( Eq 1 ) and conditional mutual information ( Eq 2 ) [56] . Noting p ( iF , iP , jP ) as the probability for a given state ( e . g . combination of spiking or not spiking ) of the IF , IP , and JP time series , the mutual information between JP and IF , for instance , is given by: MI ( IF;JP ) =∑iF , jPp ( iF , jP ) log ( p ( iF , jP ) p ( iF ) p ( jP ) ) ( 1 ) The mutual information quantifies the amount of information one variable provides about the other . The mutual information is greater than or equal to zero and is symmetric ( i . e . MI ( IF; JP ) = MI ( JP; IF ) ) . Mutual information can be used to quantify the communication between a source and a receiver [56] . As may be predicted by the name , the conditional mutual information is similar to the mutual information , except that it conditions the information shared between the original two variables by a third variable . The conditional mutual information between JP and IF conditioned on IP is given by: MI ( IF;JP|IP ) =∑iF , iP , jPp ( iF , iP , jP ) log ( p ( iF|iP , jP ) p ( iF|iP ) ) ( 2 ) Then , the transfer entropy from J to I is simply defined as the conditional mutual information when the temporal relationship between variables described above is employed: TE ( J→I ) =MI ( IF;JP|IP ) =∑iF , iP , jPp ( iF , iP , jP ) log ( p ( iF|iP , jP ) p ( iF|iP ) ) ( 3 ) In this analysis , the probability distributions were calculated by counting the number of occurrences of a given joint state throughout the hour long recording . Doing so required the assumption that the activity was stationary throughout the recording . Given the fact that spontaneous activity was recorded and that the cultures were isolated from outside stimuli , we feel this is an appropriate assumption . As described in [44] , we normalized the TE by the entropy of the receiver: TE ( J→I ) Norm=TEJ→I−∑ifuturep ( iF ) log ( p ( iF ) ) ( 4 ) In our previous work , we used multiple bin sizes and delays to examine network connectivity at multiple discrete time scales , thus forming so called multiplex networks [44] . In this analysis , we used these same TE networks , but we chose to focus only on short time scales with interactions ranging from 1 . 6 to 6 . 4 ms and 3 . 5 to 14 ms because those time scales correspond well with the reported synaptic delay of 1–20 ms [45 , 46] . We chose to use overlapping time scales to ensure all interactions were captured . Note that a significant strength of this type of information theory analysis is that it is theoretically able to detect both excitatory and inhibitory interactions . However , because the neurons had relatively low firing rates , we predict that excitatory interactions were easier to detect because it is easier to detect a significant increase in an unlikely event than it is to detect a significant decrease in an unlikely event . We made no attempt to identify the excitatory or inhibitory interactions in the networks produced by our analysis . To assess which connections were significant , we used a Monte Carlo approach to generate a null distribution of TE values [44] . This consisted of generating 5000 surrogate data sets for each pair of neurons using spike jittering and calculating their TE values . If less than 5 of the surrogate data sets produced TE values larger than the real data ( i . e . p < 0 . 001 ) , the connection was deemed significant . Finally , for each recording , 500 sub-networks with 50 neurons and average total degree 3 were produced from the full networks to reduce bias associated with average degree and network size [44 , 83] . To move beyond bivariate connectivity and investigate information processing by neurons that receive inputs from two or more other neurons , it is necessary to employ a multivariate information measure . There has been a great deal of debate regarding recent developments in multivariate information theory [5 , 6] , much of which is centered on the Partial Information Decomposition ( PID ) [3] . In addition , the PID has been used to establish a form of multivariate transfer entropy ( Fig 1B , [4] ) . Though research is ongoing in this area and alternate methods have been put forward ( see [5 , 6] for reviews , see [27 , 84–87] for examples ) , it is important to note that the PID multivariate TE possesses two distinct advantages over alternative methods , some of which have been used previously in neural system studies . First , PID multivariate TE can incorporate the necessary four variables ( past states of three neurons plus the future state of one of the neurons ) required to measure multivariate TE , unlike the other recently introduced multivariate information decompositions [84 , 86 , 87] . Second , PID multivariate TE can dissect the interaction into non-overlapping , non-negative terms , unlike the other multivariate TE methods [27 , 85] or previously used multivariate interaction methods , such as mutual information , entropy , or the interaction information [54 , 56] . Based on these advantages , we chose to employ the PID multivariate TE method . Because the PID multivariate TE is rather complicated , we will only present a brief description here . The interested reader is directed to [3 , 4] for further details . Fundamentally , the PID can be thought of as a method for dissecting well defined information terms into relevant parts . For instance , if we have three time series ( I , J , and K ) , we can consider the following decompositions ( Fig 1B ) : TE ( {J , K}→I ) =Synergy ( {J , K}→I ) +Unique ( K;J→I ) +Unique ( J;K→I ) +Redundancy ( {J , K}→I ) ( 5 ) TE ( J→I ) =Unique ( K;J→I ) +Redundancy ( {J , K}→I ) ( 6 ) TE ( K→I ) =Unique ( J;K→I ) +Redundancy ( {J , K}→I ) ( 7 ) In Eqs 5–7 , we note {J , K} as a vector valued combination of time series J and K . In Eqs 5–7 we have used several information terms with intuitive meanings [3 , 4] . The unique terms correspond to the information provided only by that time series ( J or K ) about the future state of I . The redundant term corresponds to the information provided by both time series ( J and K separately ) about the future state of I . In Eq 6 , for instance , note that the unique information from J depends on K . The TE from J to I is independent of K , but the redundant term in Eq 6 is dependent upon K . Therefore , the unique information from J is also dependent upon K . In other words , K influences what portion of the TE from J to I is redundant and what portion is unique . The synergistic term corresponds to the bonus information gained by the simultaneous knowledge of both time series ( J and K together ) about the future state of I . Note that all of the TE terms on the LHS of Eqs 5–7 can be calculated easily via Eq 3 . If a method were found to calculate the redundant term , the unique terms could be calculated by subtracting the redundant term from the TE terms in Eqs 6 and 7 . Then , the synergy term could be found by subtracting the redundant and unique terms from the TE term in Eq 5 . Thankfully , Williams and Beer provide a method for measuring the redundant term in Eqs 5–7 [3 , 4] . They define the redundancy using a quantity called the minimum information Imin: Redundancy ( {J , K}→I ) ≡Imin ( IF;JP , KP|IP ) =∑iFp ( iF ) minR∈{JP , KP}Ispec ( IF=iF;R|IP ) =∑iFp ( iF ) minR∈{JP , KP}[Ispec ( IF=iF;R , IP ) −Ispec ( IF=iF;IP ) ] ( 8 ) where the specific information Ispec is given by: Ispec ( IF=iF;R , IP ) =∑r , iPp ( r , iP|iF ) log ( p ( r , iP , iF ) p ( r , iP ) p ( iF ) ) ( 9 ) and Ispec ( IF=iF;IP ) =∑iPp ( iP|iF ) log ( p ( iP , iF ) p ( iP ) p ( iF ) ) ( 10 ) The minimum information can be thought of as a weighted sum of the common information from time series J and K about each state of the future of time series I ( i . e . the information provided by both J and K individually about each state of the future of time series I ) , conditioned upon the past state of time series I . Thus , it is the shared information provided by time series J and K , and therefore can be viewed as the redundancy . We calculated the redundancy using Eqs 9 and 10 , as well as the transfer entropy using Eq 3 . We then used the relationships given in Eqs 5–7 to calculate the unique information terms and the synergistic information . We normalized the information terms using the entropy of the future state of I , identical to the procedure described by Eqs 3 and 4 for normalizing TE . The terms produced by the PID multivariate TE analysis can be best understood using several example systems ( Fig 2 , Table 1 ) . To aid in the comparison with previously used information theory measures , we also calculated the interaction information II ( JP; KP; IF ) ( Eq 11 ) [6 , 54 , 88] between the transmitting time series and the receiver , as well as the mutual information between one of the transmitting time series and the receiver MI ( JP; IF ) ( Eq 1 ) [56] in these simple examples . In these simple examples ( Fig 2 , Table 1 ) , note that the mutual information does not detect common drive from the history of the receiving neuron ( Hidden Self Interaction Example ) . The interaction information is not able to detect simultaneous synergy and redundancy ( Synergistic and Redundant Interaction Example ) and it is not able to detect unique information ( Single and Redundant Interaction Example ) . The simultaneous synergy and redundancy example is especially relevant in neural networks if we assume that input neurons are usually somewhat correlated and the receiver neuron functions like an integrate-and-fire neuron ( i . e . similar to an AND-gate ) . Thus , the ability of the PID multivariate TE to separate out synergistic and redundant portions is a crucial advantage over previous multivariate methods that are unable to make such a separation . Following the dissection of multivariate TE into synergistic , redundant , and unique components using the PID , we felt it was appropriate to identify the synergistic term as a measure of computation performed by the receiver neuron . This is a natural interpretation of the synergistic term because synergy requires simultaneous knowledge of the states of both input neurons and non-trivial computation intuitively requires the combination of at least two pieces of information . This interpretation can be further explained using the examples in Fig 2 . Synergy is only found in the Synergistic Interaction example and the Synergistic and Redundant Interaction example . These are the only examples where neuron I utilizes simultaneous knowledge of J and K to determine its state . Other examples , such as the Single Interaction example , show cases where I and J share information ( e . g . I in the future predicts J in the past or vice versa ) , but no computation is present . In this sense , the synergy quantifies the degree to which I required simultaneous knowledge of J and K in these complex multivariate temporal relationships and we believe it is reasonable to interpret it as a measure of computation . A similar identification between synergy and information modification has been asserted previously [55] . To be clear , in this work we wish to make no claim that the PID multivariate TE synergy term is the best or only measure of computation . Rather , we feel defining computation using synergy is a natural and reasonable interpretation . In addition , we feel it would not be appropriate to define computation in an information theoretic sense using the joint entropy between the neurons or the time lagged mutual information , for instance . The joint entropy simply calculates the overall variability of the variables , which is different from computation . Time lagged mutual information would provide a description of information flow , but it is not able to quantify the outcome of information being combined from two sources , as is done in the PID . Furthermore , we wish to clarify that our definition of computation is based solely on the spiking activities of neurons . In other words , when we say one neuron is computing information from other neurons , the information being computed is represented by the input neurons’ spike trains , not some sensory stimuli or other non-neuronal signal . A great deal of previous research has focused on computation and coding in the brain related to sensory stimuli ( e . g . [49] ) and we wish to highlight the conceptual difference between our work and those previous works ( see Fig 3 ) . Note that we make no assumptions about the types of computations or operations the receiving neuron performs based on the input neurons’ spike states . Information theory is capable of detecting linear and nonlinear interactions , making it ideal for this type of study . The computation calculated as the PID multivariate TE synergy term for the two-input system ( two input neurons and one receiver neuron ) formed the majority of this analysis . Following the detection of significant bivariate connections using TE ( described above ) , the PID multivariate TE synergy term ( computation ) was calculated for all possible groups of three neurons such that two of the neurons sent significant TE connections to the third neuron . The binning and delay methods for the PID multivariate TE were identical to the bivariate TE analysis [44] . This implied that the previously calculated bivariate TE values were identical to the bivariate TE terms in Eqs 6 and 7 . Using TE and the PID multivariate TE to analyze systems of two or three neurons was possible using the methods described above because each systems was assumed to be isolated . However , calculating higher-order synergy ( computation ) terms for systems with more than two transmitting neurons becomes computationally difficult because the number of PID terms grows very rapidly as the number of variables increases [3] . However , it is relatively easy to calculate a bound on the highest order synergy term . The highest order synergy term must be less than or equal to the information gained by including an additional input: Igain ( IF;IP , J1 , P , … , Jn−1 , P;Jn , P ) =TE ( {J1 , P , … , Jn , P}→IF ) −TE ( {J1 , P , … , Jn−1 , P}→IF ) ≥Synergy ( {J1 , … , Jn}→I ) ( 12 ) The information gain bounds the highest order synergy because the highest order synergy will only be present in the highest order TE term ( left TE term in Eq 12 ) and not in the TE from all but one of the inputs ( right TE term in Eq 12 ) . Subtracting the lower order TE will remove many of the lower order terms , but leave the highest order synergy and other higher-order terms . For a group of n input neurons , we averaged the information gain across the n possible permutations of input neurons . We calculated the information gained for up to six input connections . For each receiver and value of n input neurons , we calculated the information gain for either all possible combinations of inputs or 100 sample combinations , whichever was smaller . We chose these parameters because they are large enough to adequately sample the data and convey general trends in the data , but small enough to allow for a reasonable computation time . We used a simple feedforward network to see if a given synaptic wiring rule could reproduce the patterns of connectivity and synergy we found in the in vitro data . The model contained very few neurons and was meant to represent a small segment of a larger network . The activity of the larger portion of the network was approximated using a simple binary signal . This model was constructed to provide clear connections to the computation results found in the biological networks and to motivate future research with more sophisticated models . The model’s simple structure and illustrative purpose imply that conclusions drawn from it cannot be directly applied to biological networks , but rather should serve as guides to develop new hypotheses . The model had two layers ( referred to as the input and output layers ) , each with 20 binary state neurons . We decided to use only 20 neurons to reduce computation time and because other neurons in the network were approximated with the binary signal ( see below ) . The neurons functioned using a probabilistic rule that used a sigmoid function to define the likelihood p for the neuron to fire given the total amount of input current I: p ( I ) =11+e−αI+β ( 13 ) We utilized a sigmoid function to define the firing probability to introduce nonlinear behavior in the neurons . Furthermore , the sigmoid function produced the desired general behavior that low currents should cause the neuron to spike infrequently , high currents should cause the neuron to spike frequently , and mid-range currents should produce approximately linear changes in firing probabilities . Sigmoid functions have been widely used in the neural network literature in a variety of applications to introduce similar non-linear neuron behavior ( e . g . [89–91] ) . In Eq 13 , the constants α and β were identical for all neurons and set via the following conditions: p ( I=0 ) =0 . 01 ( 14 ) p ( I=Icon*NNeurons ) =0 . 5 ( 15 ) In Eq 15 , Icon represented the amount of current carried by each connection from the input layer to the output layer ( see below ) and NNeurons represented the number of neurons in the input and output layers ( i . e . NNeurons = 20 ) . The probability in Eq 14 was chosen to ensure that neurons with no input current would rarely fire spontaneously . The probability in Eq 15 was chosen to produce separate firing regimes for maximally connected output layer neurons ( see below ) . These conditions produced values of α ≈ 0 . 023 and β ≈ 4 . 6 . To approximate network-wide activity , all neurons received input current from a random binary signal ( b ( t ) = 0 , 1 ) whose states were equally likely . Note that the binary signal was included not to model some type of external stimulus because our experimental system utilized only spontaneous activity . Rather , the binary signal was used to model large scale changes in other neurons that were not explicitly considered in the model . As an example of the type of network-wide changes that the binary signal could be considered to model , many of the cultures in our experiments showed bursts of elevated activity [44] . The relationship between the current from the binary signal was not uniform for all neurons . This produced varying degrees of correlation between neuron activity and the binary signal . The current was varied to produce a linear change in firing probability in the absence of connectivity across neurons based on the binary signal state . Specifically , we used the following relationship for the ith input layer neuron to establish the appropriate current via Eq 13: p ( b ( t ) =0 , i ) =0 . 5−0 . 4i−1NNeurons−1 ( 16 ) p ( b ( t ) =1 , i ) =0 . 5+0 . 4i−1NNeurons−1 ( 17 ) For the oth output layer neuron , we used similar relations: p ( b ( t ) =0 , o ) =0 . 5−0 . 2o−1NNeurons−1 ( 18 ) p ( b ( t ) =1 , o ) =0 . 5+0 . 2o−1NNeurons−1 ( 19 ) The parameter values in Eqs 16–19 were chosen for several reasons . In the absence of connectivity , the first neurons in the input and output layer were uncorrelated with the binary state and equally likely to be active or inactive , while the last neurons were strongly correlated with the binary state . The correlation was reduced for the output layer to allow for influence from connectivity . Connectivity was established in the network by randomly inserting NConnections=0 . 2*NNeurons2=80 binary connections from input layer neurons to output layer neurons . By fixing the connectivity density in the network at 0 . 2 , we insured that , in the randomly connected version of the network , each input layer neuron connected to roughly one-fifth of the output layer neurons . Each run of the network was independent and consisted of the following steps: ( 1 ) Randomly select the binary signal value . ( 2 ) Inject the appropriate current into the input layer neurons based on the binary signal value . ( 3 ) Determine the spiking state of the input layer neurons based on the probabilistic rule . ( 4 ) Inject 10 units of current ( Icon = 10 ) into output layer neurons based on active connections from input layer neurons . ( 5 ) Inject the appropriate current into the output layer neurons based on the binary signal value . ( 6 ) Determine the spiking state of the output layer neurons based on the probabilistic rule and the injected current from the binary signal and the connectivity . For several models , the initial random connectivity was altered using a Hebbian or modified Hebbian rule . The rewiring proceeded by running the network 500 times to gather statistics and then calculating a rewiring score S ( i , o ) for each pair of neurons in the input and output layers . The connected pair with the lowest score was disconnected and the unconnected pair with the highest score was connected . This process of gathering statistics , calculating the rewiring score , and performing one rewiring was performed NConnections times . For each unique type of model , 100 example models were produced . In general , the score was calculated as: S ( i , o ) =A ( i , o ) +a1DegIn ( o ) +a2DegOut ( i ) +a3FR ( o ) +a4FR ( i ) ( 20 ) In Eq 20 , A ( i , o ) was the proportion of network runs that produced identical states between the ith input layer neuron and the oth output layer neuron ( i . e . the agreement between the neurons ) . DegIn ( o ) and DegOut ( i ) were the in-degree of the oth output layer neuron and the out-degree of the ith input layer neuron , respectively . To the best of our knowledge , neuron degree has not previously been explicitly incorporated in Hebbian wiring rules , though the relationship between wiring and degree in many types of networks has been studied in terms of preferential attachment ( e . g . [92] ) . We felt it would be interesting to include these factors given the fact that we investigated the relationship between computation and neuron degree . FR ( o ) and FR ( i ) were the firing rates of the oth output layer neuron and ith input layer neuron , respectively . a1 , a2 , a3 , and a4 were parameters that could be set to produce different types of rewiring rules and thus different types of models . In this analysis , we examined four types of models . The first model utilized the random initial connectivity and no rewiring was performed . The second model used a1 = a2 = a3 = a4 = 0 and , therefore , employed a purely Hebbian rewiring rule . The third and fourth models were restricted to a3 = a4 = 0 and a1 = a2 = 0 to produce degree-modified and firing rate modified Hebbian rewiring rules , respectively . The precise values of the non-zero parameters in these two models were set via a three step manual lattice search of parameter space ( bounds: −3 ≤ a1 , a2 ≤ 3 , −4 ≤ a3 , a4 ≤ 4 ) to find the model that most closely matched the computation vs . degree correlation results seen in the real data . Therefore , the rewiring rules were themselves fits to the data and the resulting parameter values represented important results ( See Results – Feedforward Network Model section below for further details on the final model parameter values ) . To insure that results from the model were not heavily dependent on the number of neurons , we ran the model with 40 neurons each in the input and output layers ( double the original size , NNeurons = 40 ) . All other parameters and equations related to the model were used as defined above . A similar three step manual lattice search of score parameters space was used to find score parameters for the degree-modified and firing rate modified Hebbian rewiring rules .
Before analyzing the computations performed by neurons in the effective connectivity networks using multivariate TE , we first examined the distributions of bivariate TE values and the degree distributions of the networks ( Fig 4 ) . Several previous studies found log-normal distributions of connection weights or distributions of weights that varied widely over several orders of magnitude [17–22] . Similar to these studies , we found a roughly log-normal distribution of significant TE weights ( Fig 4A and 4C ) . We fit the distributions using the following probability mass function: p ( x ) =ασ2πe− ( log10 ( x ) −μ ) 22σ2 ( 21 ) Note that we used a probability mass function because we binned the TE values into 100 logarithmically spaced bins to produce the distributions shown in Fig 4A and 4C . This binning necessitated the use of an additional normalization factor α in the fits . The TE distributions appeared more skewed when considering normalized TE values . This skew may be caused by the bound in the normalized TE at 1 . Regardless of the precise distribution that best fits the TE distributions , both TE and normalized TE values varied broadly over several orders of magnitude . We found the degree distribution from the real data to be markedly different from the degree distribution from random networks with identical numbers of neurons and connections ( Fig 4B and 4D ) . Specifically , the degree distributions indicate that the real data contained many more high-degree neurons than would be expected in a random network . The random networks were created by randomly placing ( uniform probability ) binary connections in networks with numbers of neurons and numbers of connections set to match the real data . This result agrees well with previous studies that have found the degree distribution to be heavy-tailed in hippocampal networks [23] and cortical networks [15 , 24] . The subject of the precise nature of the degree distribution and whether it is scale-free has received a great deal of attention ( e . g . [94] ) , but it has been shown that sub-sampled scale-free distributions are not scale-free [93] . Practically speaking , this implies that a perfect scale-free degree distribution would appear as a straight line in a log-log plot , but that a sub-sampled distribution would not simply be a noisy version of the ideal scale-free distribution . Rather , the sub-sampled distributions would appear bent . Our degree distributions appear to contain linear portions in log-log space , but they curve downwards at the end . We chose not to attempt to assess whether the deviations in these plots could be due to sub-sampling because of the wide range of assumptions that would be required . Rather , we felt it was appropriate to simply remark that the distributions from the real data are clearly heavy-tailed in that they contain more high-degree neurons than were found in random networks . Using recently introduced multivariate TE terms related to the partial information decomposition [3 , 4] , we analyzed the amount of normalized information computed ( multivariate synergy term ) by individual neurons about the states of other neurons . For each neuron in the network that received two or more connections ( defined by significant TE values ) , we calculated the computation ( synergy ) between each possible grouping of two input neurons and the receiver neuron ( see Materials and Methods for details on the information calculations ) . After determining these computation values for all possible groupings of one receiver and two input neurons in the network , we calculated the correlation between the computation values and either ( 1 ) the in-degree of the receiver neuron or ( 2 ) the out-degree of one of the transmitter neurons ( the other transmitter neuron being considered as a distinct data point ) ( Fig 5 ) . We found two primary results . First , though significant positive and negative correlations between receiver in-degree and computation were found , the distribution of these correlation values for each recording was centered near zero with a slight negative skew ( Fig 5A ) . Second , the majority of recordings exhibited positive correlations between transmitter out-degree and computation ( Fig 5B ) . These trends were also observed in distributions of all neuron groups combined across all recordings for the two time scales of interest ( Fig 5C and 5D ) . In general , these results imply that neurons tended to compute the same amount of bivariate information regardless of their in-degree ( Fig 5E ) , but that neurons with high out-degrees tended to contribute more information used in computations by other neurons ( Fig 5F ) . Following the result that bivariate computation was independent of neuron in-degree , we chose to investigate higher-order computation terms . Unfortunately , higher-order computation is difficult to calculate [3] , so we were limited to measuring a bound on the highest order computation using the information gain ( see Materials and Methods ) . For the two time scales we examined in this analysis , we found that the information gain remained constant or decreased with added number of inputs ( Fig 6 ) . Because the information gain did not increase , we interpret these results to indicate that higher-order computation did not dominate high in-degree neurons . In order to better understand the degree vs . computation correlation results discussed above , we used a simple feedforward network model with various rewiring rules to produce degree vs . computation correlation results for comparisons to the results from the real data ( Fig 7A ) . This model network was designed to capture a small segment of a larger network . The network consisted of two layers ( input and output ) of 20 neurons each . We included only 20 neurons in the model to reduce computation time . 80 binary connections linked input layer neurons to output layer neurons . Varying network-wide correlation was added to the network using a random binary signal that approximated the larger portion of the network . The presence of the binary signal further motivated the use of only 20 neurons in each layer of the network ( see Materials and Methods for further details ) . Four distinct models were employed: a model with no rewiring and random connectivity , a model that used a Hebbian rewiring rule , a model that used a firing rate modified Hebbian rewiring rule , and a model that used a degree-modified Hebbian rewiring rule . The rewiring was controlled via a score calculated for all pairs of neurons ( Eq 20 , see Materials and Methods ) . The free parameters in the degree ( a1 and a2 ) and firing rate ( a3 and a4 ) modified Hebbian rules were found using a three stage manual lattice search of parameter space . This search was performed to find models that produced computation vs . degree correlation results that most closely matched the results seen in the real data . Therefore , the models represented different rewiring methods for fitting the computation vs . degree correlation results seen in the real data . The results of this search yielded the following score equations for the degree and firing rate modified Hebbian rules: SDeg ( i , o ) =A ( i , o ) +0 . 05DegIn ( o ) −1 . 75DegOut ( i ) ( 22 ) SFR ( i , o ) =A ( i , o ) +0 . 35FR ( o ) +3 . 1FR ( i ) ( 23 ) The three rewiring rules produced markedly different connectivity patterns ( Fig 7B ) . As expected , the purely Hebbian rule pooled connections between the most strongly correlated neurons ( correlation created via the binary network-wide signal ) . The firing rate modified rule spread connections more broadly to input layer neurons less well correlated with the binary network-wide signal . Finally , the degree-modified rule , primarily by virtue of the strong negative weight associated with the degree of the input layer neurons ( Eq 22 ) , spread connectivity almost uniformly across the input layer . Both the degree-modified and firing rate modified Hebbian rules qualitatively matched the degree vs . computation correlation results seen in the real data ( Fig 7C ) . The degree-modified rule was closer than the firing rate modified rule , but given the wide differences between the model and the real cortex , the significance of the correlation matching results cannot be assessed . That said , the purely Hebbian rule and the random network produced correlation results that were qualitatively very different from the real data . Next , we compared mutual information ( Eq 1 ) between connected neurons in the four models ( Fig 7D ) . Note that we did not examine the locations of the connections ( see Fig 7B ) or whether connections were moved in assessing information transmission along connections . Rather , we examined the amount of information transmitted through network connections regardless of the locations of the connections . Unsurprisingly , we found the largest mutual information in the purely Hebbian rule . This was expected because the Hebbian rule pooled connections among already strongly correlated neurons . However , when we conditioned the mutual information on the binary network-wide correlation signal , we found that the degree-modified Hebbian rule produced the highest mutual information . Because mutual information is a measure of communication [56] , we can view the conditioned mutual information as neuron-to-neuron communication in the presence of network-wide correlations and we can view the unconditioned mutual information as primarily carrying information about network-wide correlations . Therefore , the degree-modified rule–which produced degree vs . computation results that most closely matched the real data–both increased neuron-to-neuron communication and reinforced network-wide correlations , while the other rewiring rules maintained or decreased neuron-to-neuron communication ( Fig 7E ) . Our model was only intended to serve as an illustration of the computation results found in the biological tissue and therefore results from the model cannot be expected to directly relate to cortical networks . Though , to insure that our results were not heavily dependent on network size , we ran the network model with 40 neurons ( double the original size ) to see how the number of neurons affected the overall model results ( S1 Fig ) . This new model produced the same firing rate modified Hebbian rewiring rule ( Eq 23 ) and a similar degree-modified Hebbian rewiring rule given by: SDeg , 40 ( i , o ) =A ( i , o ) +0 . 16DegIn ( o ) −0 . 3DegOut ( i ) ( 24 ) The new model produced connectivity diagrams ( S1B Fig ) qualitatively identical to those seen in the smaller model ( Fig 7B ) . As with the smaller model , the larger model produced computation correlation results similar to the results seem in the data for the degree-modified and firing-rate modified Hebbian rewiring rules , but not the pure Hebbian rewiring rule ( S1C Fig ) . Results for the connection mutual information in the larger model were very similar to the results seen in the smaller model , though the Hebbian and firing rate modified Hebbian rewiring rules produced larger conditioned mutual information than was seen in the small model ( S1D Fig ) . As in the smaller model , the degree-modified Hebbian rewiring rule reinforced network-wide activity and increased neuron-to-neuron communication ( S1E Fig ) .
The main finding of this work is that the amount of information computed by a neuron about the states of other neurons depends significantly on its topological location in the surrounding functional network . More specifically , the neurons that compute the most information tend to receive inputs from high-degree neurons . The in-degree of a neuron , however , has no relationship to the amount of information it computes . Previous studies have found a log-normal ( or at least widely varied ) distribution of synaptic weights in networks of cortical neurons [17–21] . The presence of a wide range of connection strengths dictated by a log-normal distribution can significantly impact several features of a network , including signal propagation in the presence of noise and synaptic plasticity [20] . We found a roughly log-normal distribution of transfer entropy values . It is interesting that both structural connectivity as assessed in these previous studies and the effective connectivity measured herein produced similar distributions of connection weights . Though such experiments are currently difficult to perform , future studies could be conducted to further investigate the relationships between these two types of connectivity ( see [42] as an example of such an analysis conducted in a model ) . In our analysis of the degree distributions , we found them to be heavy-tailed , indicating that some neurons had more connections than would be expected if the network were randomly connected . This result corresponds well with different analyses performed using the same data [15 , 24] and previous studies conducted in hippocampus [23] . The nature of the degree distribution has been widely discussed in the literature , with possible implications including network formation mechanisms and rhythm formation [94 , 99] . Due to issues surrounding sub-sampling , we were unable to determine if the degree distributions were actually scale-free [93] . Still , the presence of high-degree neurons leads to the question of what role degree plays in the network , which we primarily addressed by examining computation ( see below ) . We sought to measure the relationship between the neuron degree and computation performed or contributed . Recall , we defined computation using the synergistic information calculated using the PID because this portion represents the information gained by simultaneous knowledge of all input variables . We found that computation performed , as defined here , was relatively independent of in-degree , while contribution to computation was correlated with out-degree . This indicates that neurons that received more connections did not compute more information than neurons that received a few connections . Conversely , we found that neurons that sent out many connections contributed more information to computations performed by other neurons . In other words , we found that neurons that received connections from high out-degree neurons tended to compute more information from those high out-degree neurons than did neurons that received connections from low out-degree neurons . There are several interesting consequences of these results . First , because we primarily measured bivariate computation ( two neurons sending connections to a third ) , it is possible that neurons with high in-degrees were performing higher-order computations . We addressed this issue by calculating a limit on the highest order synergy term ( highest order computation ) . This analysis showed that the limit of the higher-order computation was constant or decreasing as the number of inputs is increased . We interpret this result to indicate that higher-order computation did not dominate high in-degree neurons . However , our analysis only placed a limit on the highest order computation and , due to the large number of synergy terms for many input variables , it is possible that for some neurons certain terms were maximized for large numbers of input connections . Therefore , we feel these higher-order computations must be investigated further . These higher-order computation effects may be especially relevant for studies of conductance states in high in-degree neurons [18 , 100] . Second , the correlation between neuron out-degree and contribution to computation implies that high out-degree neurons have a special role in the network . It appears that these neurons were broadcasting information the rest of the network was using in computations , so perhaps they were sources of especially relevant or important information . Furthermore , it is possible that these neurons were physiologically different from other neurons ( e . g . excitatory , inhibitory , located in a certain cortical layer , etc . ) and/or that there was some type of interplay between the information the neuron provides and how it formed connections with other neurons . Unfortunately , we were unable to investigate either of these possibilities in this experimental system , though we hope to do so in the future . Third , the lack of correlation between neuron in-degree and computation performed implies that neurons compute a similar amount of information regardless of their in-degree . This even spreading of the computational burden may be the most robust or efficient . Alternatively , this result may be due to spike rate limitations in the neurons themselves or to the presence of higher-order computations ( see above ) . Additionally analyses could be undertaken in the future to assess the role spike rate limits may play in neuron computation . Also , it may be possible that high and low in-degree neurons , while computing the same amount of information , are performing different types of computations . Perhaps some types of computations are best performed by low in-degree neurons , while other types are best performed by high in-degree neurons . Additional studies could be conducted to characterize the types of computations performed by the neurons . Finally , we did not relate the information computed by a neuron to its out-degree . So , while in-degree did not affect the amount of information computed by a neuron , neurons that computed a large amount of information may have sent out more connections than neurons that computed a small amount of information . In the future , we hope to examine this possibility with further analyses . Fourth , in this analysis we related computation to neuron in and out-degree , but it would be interesting to relate computation to other network topology metrics [101] , such as modularity [102] , assortativity [103] , and the clustering coefficient [104] . Perhaps computation primarily occurs in neurons that receive connections from distinct modules . Also , the relationships we found between computation and degree may have a special importance for network assortativity given the emphasis placed on the degrees of connected neurons in the assortativity calculation . Using a toy feedforward network model , we found that it was possible to tune a degree-modified Hebbian rewiring rule to produce computation vs . degree correlation results that qualitatively matched the results seen in the real data . We then compared this tuned degree-modified Hebbian rewiring rule to other possible rules . We found that a purely Hebbian rewiring rule and a random network produced computation vs . degree correlation results that were markedly different from the results found in the real data . Furthermore , we found that the degree-modified Hebbian rule maximized neuron-to-neuron mutual information in the presence of network correlations , while still increasing reinforcement of network-wide correlations . Finally , the specific parameter values that produced the degree-modified Hebbian rule included a negative weight for the out-degree of the input layer neurons . This negative weight had the effect of spreading connections broadly from input layer neurons , which may be relevant for recent studies of bottlenecks in network activity [105] . Obviously , the simplicity of this model prevents us from drawing direct conclusions about networks of neurons in the cortex from our results . We wish to emphasize that in a more realistic model or in the actual cortex , contradictory results could be found . For instance , our model did not contain inhibitory connections , which would probably greatly affect firing rate modified Hebbian rules . In spite of that , the results from our model do motivate several intriguing hypotheses that should be investigated further . First , our result that a degree-modified Hebbian rewiring rule best matched correlation results opens the door to exploring degree-modified Hebbian wiring rules in more accurate models and organic systems . We are unaware of any other analysis that has directly incorporated neuron degree in a Hebbian rewiring rule . Though a similar concept ( preferential attachment [92] ) has been explored in the literature . Because it seems unlikely that neurons have direct access to information about their degree or the degrees of other neurons , experiments should be performed to see if some other physiological or chemical factors are capable of communicating this information . That said , one could imagine possible explanations for why neurons with many connections would tend to gain more connections ( e . g . to spread relevant or important information ) or why neurons with many connections would tend not to gain more connections ( e . g . fault tolerance , resource allocation concerns , etc . ) . Similar hypotheses could be developed for firing rate modified Hebbian rewiring rules . Second , our result that the degree-modified Hebbian rewiring rule also maximized neuron-to-neuron communication in the presence of network-wide correlations while simultaneously increasing the reinforcement of network-wide correlations points to possible relationships between signal propagation in correlated networks , computation , and network topology . Furthermore , it is noteworthy that all three rewiring rules tended to concentrate the end point of connections in neurons that possessed strong network-wide correlation , but the degree-modified and firing rate modified Hebbian rewiring rules spread the start point of connections to neurons that were not strongly correlated with network-wide correlations . Given the wide interest in the topic of signal propagation in correlated and noisy networks [57–63] , this result should be further investigated . Perhaps network topology is determined in such a way that computation and information transmission are routed through certain neurons , while other neurons maintain network-wide correlations . We hope to further investigate these issues in more accurate models and in vivo systems in the future . For the sake of simplicity , we chose to only focus on the synergy term ( computation ) for this study . However , the other PID terms could provide useful information about the cortex . For instance , though unexplored here , the PID multivariate TE redundancy term may prove useful for measuring interactions from correlated inputs . It may be the case that high out-degree neurons tend to broadcast more redundant information than low out-degree neurons , implying that network-wide correlations may be managed by high out-degree neurons . Also , the redundancy term could be used to group neurons into functionally similar groups . It would be interesting to relate these functionally similar groups to topological properties , such as degree and modularity [101 , 102 , 106] , and , if possible , to neuron properties like cell type or physical location in the tissue . Finally , it would be interesting to compare PID multivariate TE unique information terms to redundancy and synergy , especially for higher-order interactions if possible . Doing so would illuminate the contributions from individual neurons in the network . It is possible that the unique and synergistic terms decrease with additional inputs , while the redundancy terms increase , thereby possibly reducing the importance of individual neurons in the network . In the future , we hope to further investigate these terms in the cortex and other systems , and we would like to emphasize that these other terms may be crucial to additional analyses of the topics discussed above . Perhaps the most noticeable potential limitation of this analysis is the fact that it was performed using organotypic cultures [67 , 107] . Although organotypic cultures have been widely used in research [108 , 109] , these cultures have been shown to possess several differences in comparison to the in vivo system using both mice and rats . Such differences in vitro include additional synaptic connectivity [110 , 111] , decreased ease of LTP induction [112] , changes in protein expression [113] , increased excitability [111 , 114] , and changes in cellular organization in mice [115] . Despite these issues , the overall structure and electrical activity of cortico-hippocampal organotypic cultures have been shown to essentially match the in vivo system [110 , 112 , 116] . Furthermore , it has been shown that interneurons in organotypic cultures are physiologically and morphologically identical to interneurons in vivo [117] , cortical layer structure and cell migration are preserved in postnatal organotypic cultures ( as were used in this analysis ) in rats [118] , and that intracortical connection structure is preserved in organotypic cultures when sub-cortical regions are preserved in culturing ( as was done in this analysis ) [119–121] . Based on these previous studies , we concluded that organotypic cultures represent a useful model system for intact in vivo neural systems . Therefore , we believe our results are relevant for the field given the strength of the preparation used , the power of the analysis , and the novelty of the results themselves . Furthermore , at this time , it would not have been technologically possible to achieve the same level of spatial and temporal recording resolution and the same number of recorded neurons in vivo . While some in vivo recording methods are capable of recording from hundreds of neurons , these methods demand trade-offs in terms of temporal or spatial resolution . For instance , in vivo calcium imaging allows for the simultaneous recording of up to approximately 1000 neurons , but the temporal resolution for these recordings is significantly less ( tens of ms ) than we achieved in our recordings ( 50 μs ) [122–124] . Recording methods with lower temporal resolution would have been unable to capture the interactions that we observed . Furthermore , in vivo electrophysiological recording methods that employ planar arrays or shank electrodes are capable of recording hundreds of neurons with high temporal resolution , but these recording methods possess limited spatial resolution in comparison to our array ( inter-electrode spacing of 60 μm ) due to larger inter-electrode spacing in arrays ( e . g . 400 μm in Utah arrays ( Blackrock Microsystems ) ) and larger spacing between shanks ( e . g . 250 μm in [125] ) . Therefore , our use of organotypic cultures and a high density , high temporal resolution multi-electrode array permitted a dramatic improvement in the quality of the data , which improved the strength of the analysis . Our recording method possessed several distinct features that were advantageous especially during the developmental stages of this method . Still , other recording methods could be used with this analysis method to investigate other phenomena . For instance , the use of in vivo calcium imagining , while lacking the temporal resolution to capture short time scale connections , would more easily facilitate the gathering of additional information about the neurons involved in the networks ( e . g . cell type , cell layer , physical location in the tissue , etc . ) . In this analysis , we related computation to the topological locations of the neurons in the functional networks , but we were unable to relate neuron computation to the physical locations of the neurons in the tissue . Also , in vivo calcium imaging would more easily allow for direct cell stimulation or inhibition via optogenetic techniques [126] . Furthermore , in vivo calcium imaging reduces the burden associated with spike sorting because the neurons are visually identified . Spike sorting in any analysis is a significant issue because it typically prevents detection of neurons with very few spikes . In our analysis , for instance , we only recorded ~300 neurons on average from each culture despite the fact that our array covered approximately 2 mm2 of tissue . ( For comparison , we estimate roughly 10 , 000 neurons were covered by the array . ) While the issue of large populations of silent or nearly silent neurons is itself an active area of research [127] , utilizing a method like in vivo calcium imaging could reduce problems with low spike count neurons . Finally , in vivo studies could investigate the relationship between computation and phenomena that can only be studied in vivo , such as behavior and sensory coding ( e . g . [22] ) . We feel these types of analyses could produce novel insights into computation at the cellular level in the brain and we plan to pursue them in the future . In addition to new questions that could be addressed with different types of experimental systems and recording methodologies , improvements can also be made to the information analysis method itself . First , of particular relevance to the analysis method is how it addresses noise . In this analysis , we used randomized data that preserved noise to some extent in the process of generating null information values . We then compared the distribution of null information values to the real values to establish significant TE [44] . We did not characterize how different types of noise or levels of noise affect this method , nor does the method explicitly take noise into account . In the future , we hope to improve the information analysis method to explicitly incorporate noise . Second , though we were able to calculate a limit on higher-order synergy terms ( Fig 6 ) , the PID analysis method is difficult to scale up to more than three neurons . These higher-order interactions are most likely very relevant in neural networks . Though , complex interactions between many variables are notoriously difficult to analyze with any tool , so this limitation of the PID analysis is not unique and future work must be done in general to address the topic of higher-order interactions . Furthermore , two or three neuron interactions ( as were studied herein ) can be successfully analyzed using the PID . Another limitation of this analysis is that we examined two combinations of bin size and delay , but other bin sizes and delays could be used , possible complicating the analysis . In the future , we hope to develop methods to address these issues and they represent important concerns for other analyses . | We recorded the electrical activity of hundreds of neurons simultaneously in brain tissue from mice and we analyzed these signals using state-of-the-art tools from information theory . These tools allowed us to ascertain which neurons were transmitting information to other neurons and to characterize the computations performed by neurons using the inputs they received from two or more other neurons . We found that computations did not occur equally in all neurons throughout the networks . Surprisingly , neurons that computed large amounts of information tended to be recipients of information from neurons with a large number of outgoing connections . Interestingly , the number of incoming connections to a neuron was not related to the amount of information that neuron computed . To better understand these results , we built a network model to match the data . Unexpectedly , the model also maximized information transfer in the presence of network-wide correlations . This suggested a way that networks of cortical neurons could deal with common random background input . These results are the first to show that the amount of information computed by a neuron depends on where it is located in the surrounding network . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"action",
"potentials",
"medicine",
"and",
"health",
"sciences",
"neural",
"networks",
"membrane",
"potential",
"sociology",
"signaling",
"networks",
"social",
"sciences",
"electrophysiology",
"neuroscience",
"network",
"analysis",
"thermodynamics",
"computer",
"and",
"in... | 2016 | High-Degree Neurons Feed Cortical Computations |
Histone H3 di- and trimethylation on lysine 4 are major chromatin marks that correlate with active transcription . The influence of these modifications on transcription itself is , however , poorly understood . We have investigated the roles of H3K4 methylation in Saccharomyces cerevisiae by determining genome-wide expression-profiles of mutants in the Set1 complex , COMPASS , that lays down these marks . Loss of H3K4 trimethylation has virtually no effect on steady-state or dynamically-changing mRNA levels . Combined loss of H3K4 tri- and dimethylation results in steady-state mRNA upregulation and delays in the repression kinetics of specific groups of genes . COMPASS-repressed genes have distinct H3K4 methylation patterns , with enrichment of H3K4me3 at the 3′-end , indicating that repression is coupled to 3′-end antisense transcription . Further analyses reveal that repression is mediated by H3K4me3-dependent 3′-end antisense transcription in two ways . For a small group of genes including PHO84 , repression is mediated by a previously reported trans-effect that requires the antisense transcript itself . For the majority of COMPASS-repressed genes , however , it is the process of 3′-end antisense transcription itself that is the important factor for repression . Strand-specific qPCR analyses of various mutants indicate that this more prevalent mechanism of COMPASS-mediated repression requires H3K4me3-dependent 3′-end antisense transcription to lay down H3K4me2 , which seems to serve as the actual repressive mark . Removal of the 3′-end antisense promoter also results in derepression of sense transcription and renders sense transcription insensitive to the additional loss of SET1 . The derepression observed in COMPASS mutants is mimicked by reduction of global histone H3 and H4 levels , suggesting that the H3K4me2 repressive effect is linked to establishment of a repressive chromatin structure . These results indicate that in S . cerevisiae , the non-redundant role of H3K4 methylation by Set1 is repression , achieved through promotion of 3′-end antisense transcription to achieve specific rather than global effects through two distinct mechanisms .
Packaging of eukaryotic DNA with histones has a generally repressive effect on transcription [1] . Histones themselves are subject to a variety of post-translational modifications , such as acetylation , methylation and ubiquitinylation . These modifications correlate with specific states of transcription , as well as with the activity of other DNA-linked processes , such as chromosome segregation and DNA repair [2] , [3] . Among the epigenetic marks , histone methylation has been extensively associated with both activation and repression of genes in euchromatic and heterochromatic regions respectively [4] . Methylation of histone H3 on lysine 4 ( H3K4 ) for example , has been linked to transcriptional activation in many eukaryotic species . Vertebrates possess several H3K4 methyltransferases related to the SET domain of yeast Set1 and Drosophila Trx ( MLL family ) [5] . These methyltransferases are responsible for mono- ( H3K4me1 ) , di- ( H3K4me2 ) and trimethylation ( H3K4me3 ) of H3K4 [6] . Di- and trimethylation of H3K4 is generally restricted to euchromatin and genome-wide studies in metazoan cells have revealed high levels of histone acetylation and H3K4 methylation in promoter regions of active genes [7] , [8] , [9] , [10] , [11] . H3K4me2 and H3K4me3 are thought to facilitate transcription through the recruitment of general transcription factors [12] and cofactors [13] or by preventing repressors from binding to chromatin [14] . The precise mechanism through which the various H3K4 methylation states contribute to control of gene expression are not fully understood . In Saccharomyces cerevisiae , H3K4 methylation is carried out by the Set1 complex , COMPASS [15] , which is composed of the catalytic subunit Set1 and at least six other components ( Swd1 , Swd2 , Swd3 , Bre2 , Sdc1 and Spp1 ) [16] , [17] , [18] , [19] . Loss or inactivation of individual subunits differentially affects the methylation state of H3K4 . Swd1 , Swd2 and Swd3 are required for COMPASS stability and their disruption affects all three H3K4 methylation states . Bre2 and Sdc1 promote the efficient di- and trimethylation of H3K4 , while inactivation of Spp1 only affects H3K4 trimethylation [20] , [21] , [22] . In addition , monoubiquitylation of Swd2 has recently been shown to mediate the trans-tail process between H2B ubiquitylation and H3K4 trimethylation , by controlling the recruitment of the Spp1 subunit [23] . Set1 has been found to be predominantly associated with the coding regions of highly transcribed RNA polymerase II genes and the presence of trimethylated H3K4 correlates with Set1 occupancy [24] and transcription rate [25] . Genome-wide studies in yeast indicate that active transcription is characteristically accompanied by histone H3K4 trimethylation at the 5′-end of genes and by H3K4 dimethylation and monomethylation at nucleosomes positioned further downstream in the transcription unit [25] . Although H3K4 trimethylation has been linked to transcription initiation and elongation in yeast [6] , [21] , [26] , its precise role in transcription as well as the role of H3K4 mono- and dimethylation remain poorly understood . This is in part because previous genome-wide analyses of the effects of H3K4 methylation loss have yielded conflicting results [6] , [27] , [28] , [29] [30] . While two studies suggested a global reduction in transcription when H3K4 methylation is abolished [27] , [28] , a third study reported and focused on only 480 very marginally down-regulated genes , even though twice as many genes were observably upregulated upon applying the same selection criteria [6] . The most recent study also reported roughly 300 genes up-regulated and 100 down-regulated [30] . A more statistically stringent study that included adequate replicate experiments showed that 200 genes become up-regulated upon loss of SET1 , with virtually no down-regulation observed [29] , suggesting that H3K4 methylation may actually play a more prominent role in repression than in activation of protein-coding genes . Recently , a form of RNA-mediated transcriptional repression has been reported in S . cerevisiae , that is independent of the RNAi machinery which is absent from budding yeast . Ty1 , PHO84 and GAL1/10 expression have been shown to be regulated by antisense RNA transcription [31] , [32] , [33] . For PHO84 , it was found that expression of PHO84 antisense RNA from an ectopic PHO84 gene copy was able to trigger silencing of the endogenous PHO84 gene [34] . Production of the PHO84 antisense RNA was found to be positively regulated by Set1 [34] potentially linking H3K4 methylation to non-coding RNA ( ncRNA ) regulation . Genome-wide analysis has recently revealed the existence of hundreds of previously uncharacterized ncRNAs in mammals [35] , [36] , [37] , [38] and in yeast [39] , [40] , that either stably exist or are rapidly degraded by the RNA surveillance pathway . Strikingly , most of these newly identified transcripts initiate from nucleosome-free regions associated with bidirectional promoters of protein-coding genes or regions in the body or close to the 3′-ends of protein-coding genes [40] . Regulation of ncRNAs is far from understood . Here we present an extensive genome-wide analysis that discriminates between the roles of the different H3K4 methylation states . While preventing H3K4 trimethylation on its own has no effect on mRNA expression of coding genes , 1% of coding genes are derepressed upon combined loss of di- and trimethylation . Further analyses indicate distinct roles for these two marks in repression of coding genes through mechanisms that are mediated through 3′-end antisense transcription .
Previous genome-wide analyses of the effects of losing H3K4 methylation [6] , [27] , [28] , [29] [30] focused on loss of all three H3K4 methylation states simultaneously , either through deletion of the gene that codes for the H3K4 methyltransferase , SET1 or through substitution of H3K4 with alanine or arginine . To investigate whether there are separate roles for H3K4 mono- , di- and trimethylation , we made use of the fact that mutating different components of the Set1 complex , COMPASS , results in different methylation states . First , the methylation status of H3K4 was assessed in strains with deletions of the non-essential members of the complex , in the single genetic background used for this study ( BY4741 ) . An additional strain was included that carries a mutation that prevents monoubiquitylation of the essential subunit Swd2 ( swd2K68 , 69R ) , resulting in a severe reduction of H3K4me3 [23] . Histones were purified from each strain and their H3K4 methylation status was checked with antibodies specific for each methylated state ( Figure 1A ) . As expected from previous results ( see the introduction ) , deletion of SET1 , SWD1 or SWD3 abolishes mono- , di- and trimethylation of H3K4 . Deletion of BRE2 or SDC1 results in a complete loss of H3K4me3 , a significant decrease of H3K4me2 but no change in H3K4me1 , while inactivation of SPP1 or mutating SWD2 ( swd2K68 , 69R ) results in a severe and specific decrease of H3K4me3 ( Figure 1A ) . The same strains were analyzed in parallel by long oligo DNA microarray expression-profiling , targeting the coding strand of virtually all yeast genes . Throughout this study all microarray analyses were performed with four replicates ( two independent cultures , each measured in duplicate , Materials and Methods ) . In addition , controls were included that allow detection of global changes in the entire mRNA population [41] . Such global changes were not detected . In agreement with the most recent studies of SET1 deletion on its own [29] [30] , expression of only a minority of genes is affected in the different COMPASS mutants . Within the entire set of deletion mutants , 89 genes changed significantly in at least two mutants ( p-value lower than 0 . 01 and fold-change versus wild-type more than 1 . 7 ) , with 69 genes showing increased expression and only 20 exhibiting decrease ( Figure 1B ) . Deletion of any of the five core subunits Set1 , Swd1 , Swd3 , Bre2 and Sdc1 leads essentially to the same expression profile ( Figure 1B and Figure S1 ) . It is interesting to compare the changes in gene expression to the H3K4 methylation states observed in the different mutants . Virtually no significant changes in gene expression are observed in spp1Δ or in the swd2K68 , 69R mutant ( Figure 1B ) that both show a specific and severe decrease of H3K4 trimethylation ( Figure 1A ) . Changes in gene expression are observed in bre2Δ and sdc1Δ , where H3K4 dimethylation is significantly diminished on top of the loss of trimethylation , ( Figure 1A , 1B ) . The additional loss of H3K4 monomethylation , as observed in set1Δ , swd1Δ or swd3Δ ( Figure 1A ) , does not lead to additional changes in gene expression ( Figure 1B ) . Because of the correlation between their location and transcription rates [25] , H3K4 methylation marks in yeast have generally been associated with transcription activation . The main effect of mutating COMPASS components in S . cerevisiae is nevertheless derepression ( Figure 1B ) . Furthermore , the effect is only strong upon loss of dimethylation on top of trimethylation loss , which on its own has little effect . To distinguish whether the repressive effect of COMPASS is related to H3K4 methylation or is due to an unidentified methylation target of Set1 , a H3K4 point mutant was analyzed . The predominant effect is up-regulation ( Figure 1C ) and the overlap with the COMPASS-repressed genes is highly significant ( p-value 3 . 1*10−27 , hypergeometric test ) . An apparently lower number of genes is derepressed in the H3K4 point mutant . As analyzed later , this is likely related to the H3/H4 histone dosage effect of the strain used to generate the point mutant . To nevertheless investigate the possibility that Set1 repression is mediated by a target other than H3K4 , SET1 was deleted in the H3K4 point mutant strain . DNA microarray analysis of the double mutant shows a completely epistatic relationship with no additional effect of deleting SET1 in the H3K4 point mutant strain ( Figure S2 ) . This confirms that the repressive effect of COMPASS observed here is mediated through H3K4 . It has been previously shown that H3K4 di- and tri- , but not monomethylation states are controlled by the Rad6/Bre1-mediated monoubiquitylation of histone H2BK123 via a trans-tail pathway involving ubiquitylation of Swd2 [23] , [42] , [43] , [44] , [45] . To investigate whether the repressive effects of H3K4 methylation are mediated by this pathway , a bre1Δ strain was analyzed . Changes in gene expression in bre1Δ matches the COMPASS mutants profiles with a highly significant overlap ( p-value of 1 . 0*10−37 , hypergeometric test ) ( Figure 1C ) . The repressive effects observed here therefore correspond to the action of the entire pathway starting from ubiquitylation of histone H2B and leading to di- and trimethylation of H3K4 . Since the experiments described above deal with steady-state changes in mRNA levels , we next asked whether the absence of H3K4 methylation would affect the kinetics of gene expression changes . This is based on the proposal that H3K4me3 may have a memory function , bookmarking genes that require rapid induction under specific growth conditions , both in mammals [46] and yeast [24] . For this purpose , wild-type ( wt ) , set1Δ ( absence of all three H3K4 methylation states ) and spp1Δ ( lack of H3K4 trimethylation only ) were expression-profiled at multiple time-points during the transition from post-diauxic shift to early log phase , a transition during which a large number of genes change expression levels [47] . During this transition , expression of approximately 3400 genes change significantly in wt cells , covering a broad range of gene expression dynamics ( Figure 2A ) . No major differences in the transcription kinetics between wt and the two mutant strains are observed . This indicates that disruption of H3K4 methylation or H3K4 trimethylation on its own does not have a global effect on the dynamics of transcription ( Figure 2A ) , even though most active genes exhibit H3K4me3 marks [6] , [11] , [27] . These results also agree with the lack of a global effect after removing the H3K4me3 mark under steady-state conditions ( Figure 1B ) . A detailed statistical analysis for genes showing differences in their induction or repression kinetics in the mutants was also performed among the 3400 genes that change significantly during the time-course experiment . In the set1Δ ( loss of all three H3K4 methylation states ) time-course , 220 genes show statistically significant differences in their expression kinetics compared to the corresponding time points in wt ( compare Figure 2B , 2C first and third panel ) . The vast majority of these ( 194 genes - Figure 2B ) exhibit defective repression , observed as delayed repression or faster activation . Only a minority of genes exhibit an activation defect ( 26 genes - Figure 2C ) . To facilitate visualization of these mostly quite subtle changes , the wt time-course was subtracted from each mutant time-course . This results in the right-hand panels of Figure 2B and 2C , showing for each time-point , the difference in expression levels for each mutant relative to the wt at the same time point . For spp1Δ ( loss of H3K4me3 ) , only 15 genes exhibit any differences in their expression kinetics ( Figure 2B , 2C , second panel ) . These all belong to the 220 genes with slightly altered kinetics in the set1Δ time-course . In agreement with the steady-state analysis , the effects detected in the time-course experiments are thus virtually all attributable to the complete loss of methylation observed in set1Δ , rather than to the specific loss of H3K4me3 observed in spp1Δ . The results concur with a repressive role for COMPASS on mRNA expression of a subset of genes , as observed in the steady-state experiments too ( Figure 1 ) with an extremely significant overlap between the affected genes ( p-value 6*10−35 ) , as expected . We next investigated whether there are any particular characteristics shared by the set of genes upregulated upon mutation of COMPASS components ( Figure 1B ) . In agreement with a recent analysis of set1Δ [29] , statistically significant enrichment for location close to telomeres is observed ( Figure S3 ) . Among the 69 COMPASS-repressed genes , 10 are telomere-proximal ( within 15 kb ) Figure S3 and Table S1 ) . Although this enrichment is significant , in most cases the expression of adjacent genes was not found to be affected by the deletion of COMPASS subunits . For instance , PHO11 , SNO4 , MCH2 , SOR2 , YGL258W-A and PHO12 , that are located between 4 and 10 kb from the telomeric DNA on different chromosomes ( Table S1 ) are all flanked by genes that are not affected by the absence of Set1 . This , as well as the small number of all telomere- proximal genes being derepressed in the COMPASS mutants makes it unlikely that the observed derepression of telomere-proximal genes is only caused by loss of the Sir-dependent telomeric position effect [19] , [48] , [49] , [50] , [51] , [52] . As the effect of COMPASS deletions is attributable to H3K4 methylation ( Figure 1C ) , the H3K4 methylation patterns of COMPASS-repressed genes were examined using chromatin immunoprecipitation data from a wt strain from the same genetic background , grown under similar conditions [53] . Intriguingly , the di- and trimethylation patterns of the 69 COMPASS-repressed genes ( Figure 3A ) deviate from the average gene which has enrichment of H3K4me3 around the transcription start site ( Figure S4 ) [25] , [53] . Instead , the majority of COMPASS-repressed genes show enrichment of H3K4me3 at the 3′-end or in the body of the gene . In the minority of cases where 5′-end enrichment is observed , this is accompanied by a second trimethylation peak at the 3′ end . To exclude that the deviating localization of peaks is not due to measurement noise or signal originating from neighbouring genes , the methylation profiles are averaged in Figure 3B only for those genes that have a greater than 2-fold enrichment of H3K4 methylation on any portion of the gene . This average pattern for COMPASS-repressed genes shows a clear enrichment of H3K4me3 at the 3′ end , followed by H3K4me2 enrichment in the gene body , which is in turn followed by H3K4me1 further towards the 5′-end . Genes repressed by COMPASS therefore show abberant H3K4 methylation patterns that are characterized by a reversed orientation of the normal H3K4 methylation pattern observed for active genes [3] . A plausible explanation for the H3K4 di- and trimethylation peaks at the 3′-ends of COMPASS-repressed genes is the presence of antisense transcription initiation at the 3′-end of the coding region , leading to non-coding transcription over the same genomic location but in the opposite direction of the sense transcription . The DNA microarrays used in the previous experiments are coding strand-specific and do not detect anti-sense transcripts . However , two recent genome-wide surveys of non-coding transcripts [39] , [40] , do detect antisense RNAs for more than 85% of the COMPASS-repressed genes ( Table S2 ) . Interestingly , PHO84 belongs to the group of COMPASS-repressed genes identified here ( Figure 3A , marked with P ) and has been shown to be regulated by antisense RNA transcripts originating from its 3′-end both in cis and in trans [32] , [34] . We therefore investigated the manner in which 3′-end antisense transcription may be involved in Set1-mediated repression . One hallmark of the mechanism of repression of PHO84 is the contribution of the antisense transcript itself rather than only the process of antisense transcription . Stabilization of the antisense transcript by deletion of the exosome component RRP6 [32] is sufficient to repress sense PHO84 transcription . To test whether COMPASS repression is mediated by 3′-end antisense transcripts , an rrp6Δ profile was generated and compared to set1Δ . Deletion of RRP6 affects expression of 117 coding genes in total ( p<0 . 01 , fold-change>1 . 7 ) and does not have a general effect on all COMPASS-repressed genes ( Figure 4A ) . In agreement with previous studies however , a significant down-regulation of PHO84 is observed ( marked P in Figure 4A ) . Lack of down-regulation of the other COMPASS-repressed genes in rrp6Δ may be simply due to an already repressed state in wt . Since these genes are derepressed in set1Δ , the possible involvement of antisense transcripts in repressing all COMPASS-affected genes was further tested by analysis of an rrp6Δ set1Δ double mutant ( Figure 4A ) . The double mutant expression-profile reveals two classes of COMPASS-repressed genes . On the smaller group of genes ( Figure 4A , marked with a black bar ) , that includes PHO84 as well as several other phosphate-related genes , an epistatic effect is observed in rrp6Δ set1Δ , whereby the upregulation in set1Δ is lost in the double mutant . This implies that the antisense transcript mediated repression of sense genes is not unique for PHO84 , but is shared with functionally related genes . Such genes are the exception however . The largest group of Set1-repressed genes behaves in a different manner , still showing derepression in the double mutant , similar to their behaviour upon deletion of SET1 on its own . This therefore likely represents a distinct mechanism of COMPASS repression . In order to understand the mechanism by which COMPASS represses coding transcription in an exosome-independent manner , five representative genes from this group , AMS1 , YGR110W , ARG1 , SPR3 and OYE3 ( indicated by 1 to 5 in Figure 4A ) , were analyzed in greater detail . These genes represent different functional categories , different telomeric proximities and different types of antisense transcripts , as suggested by the genome-wide datasets . The first three genes contain antisense stable unannotated transcripts ( SUTs ) , while the other two have antisense cryptic unstable transcripts ( CUTs ) [40] . The location of H3K4 methylation patterns [53] corresponds to the location of the transcription initiation sites of these antisense transcription units ( Figure S5 ) . The effects of different COMPASS mutants on both sense and antisense transcription of these genes were analyzed by quantitative RT-PCR using strand-specific primers ( Figure 4B ) . Sense transcript upregulation of the five genes is observed in set1Δ ( Figure 4B ) , that exhibits loss of all H3K4 methylation marks ( Figure 1A ) , in bre2Δ ( Figure 4B ) , that exhibits loss of all H3K4me3 and most H3K4me2 ( Figure 1A ) and in set1Δ combined with rrp6Δ ( Figure 4B ) , all in agreement with the sense-specific microarray results ( Figure 1B , Figure 4A ) . In bre2Δ and set1Δ , upregulation of sense transcription is accompanied by a decrease in antisense transcription ( Figure 4B , panels 1–3 ) . As expected , changes in antisense CUT transcription are not evident without prior stabilization by the exosome deletion ( Figure 4B , panels 4 , 5 ) . For all five genes , stabilisation of antisense transcripts does occur in rrp6Δ ( Figure 4B ) , but these increased antisense levels do not necessarily result in more repression of sense transcription ( Figure 4B ) as is clearly the case for PHO84 ( [32] and Figure 4A ) . This confirms that an increase in antisense transcript levels through rrp6Δ-dependent stabilisation is not the mechanism of COMPASS repression for these genes . Rather , the data suggest that it is the process of antisense transcription itself that represses the sense transcription . Because sense and 3′end antisense transcription seem coupled [54] , it is difficult to distinguish whether the increased sense transcription in COMPASS mutants is caused , or is followed , by a decrease in 3′-end antisense transcription . One way of addressing this directly is to eliminate 3′-end antisense transcription by other means than through disruption of COMPASS . For this purpose strong terminator sequences were introduced downstream of the five model genes analyzed in Figure 4B , either as insertions between antisense promoters and the end of the ORF , or as replacement of complete intergenic sequences . Neither approach resulted in loss of 3′-end antisense transcription , which agrees with the recent finding that terminators can function as promoters [55] . Disruption of 3′-end antisense transcription was then attempted by removal of all , or a significant part of the intergenic region . Complete loss of all antisense transcription was only observed for the YGR110W intergenic deletion mutant , which we further analyzed in depth ( YGR110W-ingdel , Figure 5 ) . Strand-specific Northern blot analysis of YGR110W-ingdel shows that loss of antisense transcription ( Figure 5B , asYGR110W , lane 1 versus lane 3 ) , is accompanied by derepression of sense transcription ( Figure 5B , sYGR110W ) . This demonstrates that 3′-end antisense transcription results in repression of sense transcription . Furthermore , introduction of SET1 deletion into the YGR110W-ingdel strain , does not result in significant further derepression as is observed in the presence of 3′-end antisense transcription ( Figure 5B , lanes 1 and 2 versus lanes 3 and 4 ) . This agrees with the proposal that the repressive effect of COMPASS on coding genes is a result of promoting 3′-end antisense transcription . The results presented in Figure 4B and Figure 5B imply a positive role for Set1 on antisense transcription . SET1 deletion results in loss of H3K4me1 , me2 and me3 ( Figure 1A ) . SPP1 deletion ( loss of H3K4me3 only ) , has little effect on sense transcript levels ( Figure 1 , Figure 2 , Figure 4B ) . SPP1 deletion does result in decreased antisense transcripts as observed either in the presence or absence of RRP6 ( Figure 4B ) . Our results indicate that H3K4 trimethylation , which is found at the 3′-end of these genes , has a role in promoting 3′-end antisense transcription . This effect is not absolute however . Antisense transcripts are reduced in the SET1 RRP6 double deletion compared to rrp6Δ , but are not completely absent . This indicates that antisense transcription is promoted by , but not fully dependent on , H3K4me3 . Since spp1Δ still exhibits wt levels of H3K4me2 ( Figure 1A ) and virtually no derepression of sense transcription ( Figure 1B and Figure 4B ) , this indicates that it is the H3K4me2 mark which is most important for repression of sense transcription on these genes . Together , the results of these experiments are consistent with a model , whereby the majority of COMPASS-repressed genes are maintained in an inactive state through 3′-end antisense transcription that is in part promoted through H3K4me3 at the 3′-end , and in turn deposits a repressive H3K4me2 mark further into the body of the gene . We next asked what determines the specificity of the effects observed upon mutation of COMPASS . H3K4 methylation marks all active genes and approximately one third of all genes exhibit antisense transcripts [40] , yet only a subset are affected by deleting COMPASS subunits ( Figure 1 ) . It has recently been proposed that the transcription factor Reb1 may drive non-coding transcription , either from neighbouring genes [33] or from the promoter of the antisense transcript itself [56] . Reb1 binding sites are found downstream of only three of the 69 COMPASS-repressed genes . There is also no statistically significant enrichment for Reb1 binding sites in the ORFs or flanking regions of genes up-regulated in the COMPASS mutants . Both observations suggest that the specificity of Set1 repressive effects is not generally linked to Reb1 . In addition , no other putative regulatory motifs could be detected in these regions using different search algorithms [57] . An alternative explanation for the specificity of COMPASS repressive effects is that specificity is dictated by increased sensitivity of specific genes to a particular chromatin structure which is influenced by H3K4 methylation . While profiling strains with altered histone expression levels we noted an interesting correlation with the collection of COMPASS mutants . To investigate this , a strain bearing single copies of the histone H3 and H4 genes under control of their native promoters [58] was analyzed . The two-fold reduction in mRNA levels of H3 and H4 in this strain ( Figure 6B , marked HHT2 and HHF2 ) is accompanied by slightly decreased H3 and H4 protein levels ( Figure 6A ) . Interestingly , this results in upregulation of a specific subset of genes that strongly correspond to the genes upregulated upon SET1 deletion ( p-value 1 . 4*10−23 , Figure 6B ) . Although the overlap is highly statistically significant , it is not complete and does not extend to PHO84 for example , in agreement with the proposal for a distinct repressive mechanism for such genes ( Figure 4A ) . SET1 deletion does not globally affect nucleosome levels ( data not shown ) , and antisense transcript levels are not reduced in the single copy H3 H4 strain ( Figure 6C ) . Besides antisense transcription , a second common property of the genes affected by loss of COMPASS function is therefore sensitivity to histone abundance . Since histone abundance affects nucleosome density [59] , this suggests that Set1 may repress genes by effecting nucleosome density . As is discussed below , one manner in which this may be achieved is through the repressive H3K4me2 mark that is laid down through 3′-end antisense transcription .
The results presented here add to a number of reports that indicate that the major non-redundant role of COMPASS in S . cerevisiae is repression of coding genes [34] , [56] . Early genome-wide analyses of set1Δ yielded conflicting results , in two cases pointing to global positive effects [27] , [28] and in one case ignoring the prevalence of specific repressive effects [6] . Some of the differences between these studies and the current one can in retrospect be attributed to use of double-stranded cDNA arrays , less convenient for discriminating between sense and ant-sense effects , as well as to normalisation issues . The analyses presented here , using strand-specific techniques , with replicate experiments for a variety of different mutants under both steady-state and dynamic conditions , indicates that removal of H3K4me3 , a global mark of active transcription , has no global effect . The repressive effects observed on a specific subset of genes agree with the most recent other genome-wide analyses of set1Δ [29] [30] , as well as with the fact that deletion of SET1 is not lethal . Gene Ontology analysis of the affected genes reveals an overrepresentation of vitamin metabolism ( essentially thiamin biosynthesis ) and spore wall assembly ( Table S3 ) in agreement with the cell wall and stationary phase defects previously observed in set1Δ cells [51] . What is the mechanism of the observed repression ? Despite the fact that COMPASS-repressed genes show a significant enrichment for telomeric-proximal localization ( Figure S3 and [29] ) , Set1-dependent repression of these genes due to a telomere position effect can probably be ruled out since the derepression observed in set1Δ only affects a small percentage of individual genes within these regions . Only a few of the affected genes are close to telomeres and only few telomere-proximal genes are affected . Analysis of methylation patterns ( Figure 3 and [53] ) , non-coding RNA maps ( Table S2 and [39] , [40] ) and the comparison of mutants with different methylation states support a model whereby COMPASS mediates repression of coding genes by promoting the expression of 3′-end antisense transcripts through deposition of H3K4me3 at their 3′-end . An involvement of Set1 in promotion of 3′-end antisense transcription , resulting in a repressive effect on sense transcription has been reported for PHO84 , which is repressed through the presence of antisense transcripts [34] . Our results are consistent with a repressive role for Set1 on PHO84 . The genome-wide nature of our experiments indicates however that the majority of Set1 affected genes are repressed through a different mechanism , independent of the level of antisense RNA transcripts . Rather , for the majority of Set1-regulated genes , repression is caused by the process of antisense transcription itself . This mechanism is therefore related to the recently reported attenuation in GAL10-GAL1 activation which is also facilitated through cryptic transcription [56] . One major difference is that for the mechanism reported here , COMPASS is required to maintain antisense transcription whereas this does not seem to be the case for the cryptic transcription observed at the GAL10-GAL1 locus [56] . The comparison of different COMPASS mutations carried out here , facilitates distinguishing between the roles of the different H3K4 methylation states . Mutants with grossly lowered or completely absent H3K4me3 exhibit decreased antisense transcription . However this only results in derepression of the coding gene if H3K4me2 is also abolished . The positive role of H3K4me3 on antisense transcription fits with the correlation observed between the presence of this mark and promoter activity of coding genes [7] , [8] , [9] , [10] , [11] , [25] . Most non-coding RNAs originate from nucleosome-free regions ( NFRs ) shared with protein-coding transcripts [40] . This is also the case for three of the five COMPASS-repressed genes analyzed here in detail ( AMS1 , ARG1 and OYE3 ) . Interestingly , despite sharing a NFR with the downstream protein-coding gene , loss of H3K4 trimethylation causes reduction of antisense transcription without affecting transcription of the flanking protein-coding gene in each case . This fits with the observation that bidirectional transcription from a single NFR , originates from two distinct preinitiation complex recruitment sites [60] . This may indicate the presence of redundant mechanisms for maintaining protein-coding gene transcription in the absence of H3K4me3 , which are lacking for the antisense non-coding transcription originating from the same NFR . Another , non mutually exclusive mechanism , can be that lack of H3K4 trimethylation in the antisense transcription start site increases the recruitment of corepressor complexes , such as Rpd3S [61] , that repress the expression of the non-coding transcript , but not that of the coding gene [62] . The results also indicate a role for the H3K4me2 mark in facilitating repression . This agrees with several recent studies suggesting mechanisms through which H3K4me2 may play a repressive role . For example , it has recently been reported that H3K4me2 in the body of active genes is recognized by the Set3 complex , leading to histone deacetylation , a repressive chromatin state [63] . A different histone deacetylase , Rpd3 , has been implicated in the repressive role involving Set1 on the GAL10-GAL1 locus [56] . Furthermore , methylation of H3K4 protects against an H3 tail endopeptidase recently described in S . cerevisiae and humans that facilitates transcription initiation and precedes histone eviction [64] , [65] . All these possible mechanisms fit with the observation made here that globally reducing H3 and H4 levels mimics the derepression of COMPASS mutants . The degree of overlap between the COMPASS mutants' profiles and the histone depletion profile also give an explanation for the specificity of the COMPASS repression . Genes repressed by COMPASS have antisense transcription , but are also sensitive to nucleosome density . Sensitivity to histone depletion may not be the only reason for the lack of genome-wide effects upon COMPASS mutation . Functional redundancy may also contribute . One of the prevalent ideas for a general role of H3K4 methylation in S . cerevisiae is that transcription-associated H3K4 methylation , as well as deposition of the histone variant H2A . Z , antagonizes the local spread of Sir-dependent silent chromatin into adjacent euchromatic regions [52] , [66] , [67] . It has recently been shown that H2A . Z deposition and Set1 cooperate to prevent Sir-dependent repression of a large number of genes located across the genome [29] . This functional redundancy between H3K4 methylation and H2AZ deposition may thus buffer transcription from changes in euchromatin , thereby minimizing the observed effects of H3K4 methylation loss . This work offers a plausible explanation for how a transcription factor , previously thought to positively contribute to transcription , can nevertheless exert a negative effect , through promoting antisense transcription . The opposite has previously been shown for regulation of IME4 . Here a repressor complex binds to the promoter of an antisense transcription unit in the 5′-end of IME4 and by repressing the antisense transcription , facilitates the in cis sense transcription activation [68] . Although further work is required to pinpoint the mechanisms further downstream of H3K4 di- and trimethylation , COMPASS exemplifies the growing insight that the roles of histone modifications in gene expression are non-linear [69] and context-dependent [70] .
Strains and primers used in this study are described in Tables S4 and S5 respectively . The YGR110W-ingdel strain was created by first inserting a cassette containing the Sp HIS5 gene in reverse orientation flanked by the Ag TEF promoter , terminator sequences and loxP sites from plasmid pUG27 [71] to replace the intergenic region between YGR110W and YGR111W using YGR110W_HIS5_F and YGR110W_HIS5_R primers . Subsequently the cassette was floxed out by transforming the strain with the plasmid pSH47 and expressing Cre recombinase as previously described [72] . Histones were purified as described [42] , subjected to 16% SDS-polyacrylamide gel electrophoresis , and either Coomassie Blue stained or transferred to 0 . 2 µm ProtranR nitrocellulose . Antibodies used to detect mono- , di- and trimethylated H3K4 and histone H3 were from Abcam . Two independent colonies of each strain were first inoculated and grown overnight in synthetic complete medium with 2% glucose . For the mid-log/steady-state experiment , larger cultures were inoculated the next day at an OD600 of 0 . 15 in fresh medium , allowed to grow at 30°C and harvested at OD600 0 . 6 , ( P-UMCU-36 ) . For the time-course experiment , overnight cultures were used to inoculate 50 ml cultures at an OD600 of 0 . 15 . These were allowed to deplete glucose by growing for 24 hours and were used the next day to start 500 ml cultures at an OD600 of 0 . 15 in fresh medium for the time-course sampling ( P-UMCU-47 ) . Total RNA isolation was by hot acid phenol ( P-UMCU-37 ) and cleaned up using RNeasy ( Qiagen ) . Before amplification , external RNA controls were added to total RNA to check for global shifts in mRNA levels [41] . cRNA amplification and labelling using amino-allyl UTP was performed on a Caliper robot system ( P-UMCU-38 ) . Each sample was generated twice , as independent biological replicates . These were hybridized in dye-swap against a common wt reference RNA ( P-UMCU-39 ) on oligo-arrays that represented each gene twice ( P-UMCU-34 ) . After scanning ( P-UMCU-40 ) , raw data were extracted with Imagene ( Biodiscovery ) ( P-UMCU-42 ) . Since spike-in of external RNA controls revealed no global changes in the mRNA population [41] for the mid-log experiment , non-background corrected data were normalized with print-tip LOESS [73] on gene probes with a span of 0 . 4 ( P-UMCU-41 ) . For the time-course , all features , including negative and external controls ( EC ) except EC 4 , 6 and 8 , were used for the estimation of the LOESS curve ( P-UMCU-46 ) . Probes flagged as absent , or with a nearly saturated signal were not used to estimate the LOESS curve . For differential expression analysis , the LIMMA package [74] was used . Mitochondrial-encoded genes and Ty elements were excluded due to their high biological variation . Genes with an FDR-adjusted p-value less than 0 . 01 and a fold-change of more than 1 . 7 were considered significant . These thresholds are based on systematic analyses of the variation observed in a collection of more than 100 wt expression profiles [75] . For the time-course experiment , changes were considered significant if they fulfilled these criteria for two consecutive time-points . Hierarchical clustering was by MeV [76] , using standard correlation and average linkage . Analysis of overlap between genelists of significantly changing genes of two expression profiles was by hypergeometric test . For the GO and transcription factor enrichment analysis , a right-sided Fisher's exact test was used and the p-values were corrected for multiple testing using Bonferroni . The GO annotations were obtained from SGD . For the H3K4 methylation ChIP-chip analysis , the data are from [53] . For each gene , a region corresponding to the ORF plus 500 bps in both directions was used . The ORF was divided into 30 bins of equal length and the flanking regions in 10 bins each . A loess algorithm [77] with a span of 0 . 2 was used to estimate the enrichment of the methylation marks for every bin . cDNAs of sense RNA or antisense RNA were generated by SuperScript III Reverse transcriptase ( Invitrogen ) from total RNAs using gene and strand-specific primers . For each gene , cDNAs obtained from the reverse transcription of sense or antisense RNA were quantified by a real-time qPCR with gene-specific primers corresponding to a 150 bp fragment ( Figure 4B ) . The same primers were used to quantify sense and antisense cDNA of each gene . The position and the sequence of each primer are indicated in Figure 4B and Table S5 . The strand-specific DNA probes used to detect the presence of sense and antisense transcripts of YGR110W are shown schematically in Figure 5A . First a cold PCR product template was obtained using primers 3′qYGR110W and 5′qYGR110W . Subsequently the hot ssDNA probes for detection of the sense and antisense transcripts were generated from the template using the first or the second primer , respectively , in linear PCR reactions . Quantitation of the radioactive signal was performed using ImageQuant ( Molecular Dynamics ) . | In eukaryotes , DNA is packaged together with histones into nucleosomes . This packaging has a repressive role on gene expression . The N-termini of histones are subject to multiple modifications that affect DNA–dependent processes . The histone modification that has been predominantly linked with active transcription in all eukaryotes is histone H3 lysine 4 ( H3K4 ) methylation . Here we investigate the functional effects of each H3K4 methylation state on transcription . Removal of the mark that is most characteristic for transcription , H3K4 trimethylation , has no effect on coding gene expression , in steady-state or dynamically changing conditions . Combined loss of H3K4 tri- and di-methylation does have an effect and leads to loss of repression of specific genes , the opposite of what is expected for global marks of active genes . The affected genes have antisense transcription . We show that there are two separate mechanisms through which H3K4 methylation represses transcription of protein-coding genes , one through antisense transcripts and one through the process of antisense transcription . In summary , we show how a general mark of active transcription can have specific repressive effects that are themselves also linked to repression through nucleosomes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genome",
"expression",
"analysis",
"functional",
"genomics",
"gene",
"regulation",
"microbiology",
"dna",
"transcription",
"histone",
"modification",
"model",
"organisms",
"epigenetics",
"molecular",
"genetics",
"chromatin",
"gene",
"expression",
"biology",
"molecular",
... | 2012 | Two Distinct Repressive Mechanisms for Histone 3 Lysine 4 Methylation through Promoting 3′-End Antisense Transcription |
Salmonella enterica is a bacterial pathogen that causes enteric fever and gastroenteritis in humans and animals . Although its population structure was long described as clonal , based on high linkage disequilibrium between loci typed by enzyme electrophoresis , recent examination of gene sequences has revealed that recombination plays an important evolutionary role . We sequenced around 10% of the core genome of 114 isolates of enterica using a resequencing microarray . Application of two different analysis methods ( Structure and ClonalFrame ) to our genomic data allowed us to define five clear lineages within S . enterica subspecies enterica , one of which is five times older than the other four and two thirds of the age of the whole subspecies . We show that some of these lineages display more evidence of recombination than others . We also demonstrate that some level of sexual isolation exists between the lineages , so that recombination has occurred predominantly between members of the same lineage . This pattern of recombination is compatible with expectations from the previously described ecological structuring of the enterica population as well as mechanistic barriers to recombination observed in laboratory experiments . In spite of their relatively low level of genetic differentiation , these lineages might therefore represent incipient species .
Salmonella enterica subspecies enterica ( subsequently referred to simply as enterica ) is a major cause of enteric fever in humans and gastroenteritis in humans and animals . Its diversity has traditionally been described on the basis of serological differences following the Kauffmann-White classification [1] , [2] . Certain serovars are linked to particular diseases and hosts . For example , enteric fever is mostly caused by members of serovar Typhi and Paratyphi A , both of which only infect humans [3] . Gastroenteritis on the other hand is most often caused by Enteritidis in humans and Typhimurium in animals [4] , although both serovars can infect a wide range of hosts [3] . However , the usefulness of the serological classification of S . enterica is undermined by the fact that unrelated strains sometimes belong to the same serovar [5] , [6] . In an attempt to shed some new light on the population structure of enterica , a multi-locus sequence typing scheme ( MLST; [7] , [8] ) was developed which relies on the sequencing of 400-500 bp fragments from seven housekeeping genes . This typing technique was originally applied to strains from serovar Typhi [9] , and later to the whole of enterica [10] , [11] . Phylogenies reconstructed from MLST data are highly star-shaped [12] and therefore carry little information about relationships between isolates . This can be traced back to substantial incongruencies between gene trees [13] , [12] , [14] , which are often caused by high levels of homologous recombination [15] . This is in contrast for example with the closely related species Escherichia coli which has a well defined population structure made of several clearly defined clades [16] . The first genomes of enterica to be fully sequenced were those of Typhimurium LT2 [17] and Typhi CT18 [18] , followed by those of Typhi Ty2 [19] , Paratyphi A [20] and Choleraesuis [21] . A comparison of the genomes of Typhi and Paratyphi A revealed that they had exchanged about a quarter of their genes during the course of their adaptation to a human-specific and highly virulent lifestyle [22] . This high level of recombination is , however , exceptional between two distantly related lineages of enterica [22] , and selection is likely to have favoured recombinants between these two types which combined adaptations to their new host [22] . The pattern of recombination of these strains , with a burst of recombination being followed by completely clonal evolution [23] , [24] , appeared to be atypical of gene flow in the species as a whole , but only limited data from a small number of lineages has been analyzed [22] . The number of enterica genomes currently available is insufficient ( only eleven whole published genomes available at the time of writing in the Genomes OnLine Database; [25] ) , and their distribution is too focused on highly virulent types to allow an exploration of the population genetics of enterica . Furthermore statistical methodology to analyze such whole-genome data efficiently is currently lacking [26] , [15] . Reconstructing the clonal relationships between lineages that have evolved under the influence of recombination requires data from a large number of loci [27] . We therefore designed an Affymetrix CustomSeq Resequencing Array to sequence approximately 300Kbp from the core genome of enterica isolates , which represents two orders of magnitude more data per isolate than is provided by MLST . Resequencing arrays are a highly parallel DNA sequencing technology with quick application and low cost , and are based on the principle of sequencing by hybridization [28] . They have been previously applied to a wide diversity of bacterial samples , including monomorphic clones such as Bacillus anthracis [29] or Mycobacterium tuberculosis [30] , relatively clonal species such as Bacillus cereus [31] or Staphylococcus aureus [32] , and species with high rates of recombination such as Neisseria meningitidis [33] or Francisella tularensis [34] . We applied our resequencing array to a global collection of 114 isolates from multiple major lineages of enterica , with the exception of Typhi . Typhi was excluded because extensive studies using a wide range of molecular techniques [23] , [35] , [24] , [36] , [37] have revealed that its population biology differs from that of other lineages of enterica . We therefore excluded Typhi from the present study in order to focus on the remainder of enterica , which has been studied much less thoroughly . The main aims of this study were to provide an improved description of the population structure of enterica and to clarify the role played by recombination during its evolution . To this end , we analyzed our genetic data using the linkage model of Structure [38] , [39] and ClonalFrame [40] with a posteriori attribution of the origin of recombination events [41] .
For each of the 114 isolates under study ( Table S1 ) we resequenced 146 regions of length 2000-2500bp each from the core-genome of enterica ( Table S2 ) . These 295 , 137 bp per isolate represent approximately 10% of the core genome of enterica [42] . Figure 1 illustrates the extent of our resequencing scheme on the genome of Typhimurium LT2 [17] . On average , 85% of nucleotides were called , with variation across isolates ranging from 75% to 95% . A total of 18 , 068 of the resequenced sites ( 6% ) were found to be polymorphic in this sample . Regions overlapping the seven MLST loci were included in our resequencing scheme , and by comparing our results with preexisting MLST sequences we estimated the error rate of our method to be lower than one error per 10 , 000 calls . Only one isolate had more than one error in its MLST gene fragments: isolate 54 ( SARB32; ST82 ) had two errors , one in gene hisD and the other in gene purE . An equivalent error rate was found when comparing the sequence of LT2 reported in [17] with our resequenced sequence of LT2 . The density of errors was therefore sufficiently low enough that errors would be misinterpreted as mutations , and would not affect our results below which are essentially focused on the recombination process . We applied the linkage model of Structure [38] , [39] to our data and identified ancestral populations in our sample ( Figure S1 ) . The proportion of ancestry from each of these sources is shown for each isolate in Figure 2 . The 114 isolates fell into six distinct groups based on the major ancestral source of genetic diversity of each isolate . ( Figure 2 ) . Group 1 ( light blue ) consisted of 14 strains of Choleraesuis , Paratyphi C and Typhisuis , Group 2 ( dark blue ) comprised 12 strains of Typhimurium and Saint-Paul , Group 3 ( orange ) contained 17 strains of Montevideo , Javiana , Decatur and others , Group 4 ( yellow ) consisted of 19 strains of Enteritidis , Gallinarum and Dublin and Group 5 ( red ) comprised 5 strains of Paratyphi A and Sendai . Finally , Group 6 ( cyan ) contained the remaining 47 strains from diverse serovars . These groups showed relatively little admixture between ancestral sources ( Figure 2 ) , with the exception of Group 6 , which seemed to have acted frequently both as a donor and as a recipient of recombinational exchanges ( Figure 2 ) . ClonalFrame is a method designed to reconstruct the clonal relationships between isolates in a sample , while accounting for the effect of non-vertical genetic transfer which would otherwise confuse such a reconstruction [40] . Figure 3 shows the clonal genealogy inferred from our data by ClonalFrame . The first five groups identified by Structure ( Figure 2 ) corresponded to clades on Figure 3 and are represented with corresponding colors . Based on the combined evidence from the Structure and ClonalFrame analyses , these five groups can confidently be called lineages of enterica . On the other hand , the sixth group found by Structure encompassed the remaining isolates in Figure 3 , which did not constitute a clade in Figure 3 and therefore did not represent a true lineage . Instead , seven small groups of two to four isolates formed small clades at this level of analysis according to ClonalFrame , but these were not detected by Structure . The content of the five identified lineages of enterica is summarized in Table 1 . Using Structure and ClonalFrame on MLST data only revealed parts of this population structure , and hardly revealed any relationships within lineages in comparison with the resequencing array data ( Figures S3 and S4 ) . Yet the deep phylogeny of enterica remained largely unresolved when using our resequencing data , and in particular the relationships of the five lineages above with one another and with the rest of the isolates remained unclear ( Figure 3 ) . We estimated the age of the five lineages relative to the time of the most common ancestor of the whole of enterica ( Table 1 ) . The common ancestor of lineage 5 was the most recent , followed by that of lineage 1 . Lineage 3 was found to be particularly ancient , with an estimated age of two thirds of the age of enterica . Widespread recombination has previously been suggested to explain the lack of deep structure in enterica [12] , [14] and we wanted to assess the role played by recombination in the evolution of enterica . Measuring the frequency of recombination is often done relative to that of mutation [43] by forming the ratio of rates at which recombination and mutation occurred in the ancestry of a sample . ClonalFrame estimated that recombination happened less frequently than mutation with ( 95% credibility interval ) . Recombination can however change several nucleotides in a single event . Another measure of recombination is therefore the ratio of rates at which substitutions are introduced by recombination and mutation [44] . ClonalFrame estimated that recombination and mutation had approximately the same effect in introducing polymorphism with ( 95%CI [1 . 06 , 1 . 23] ) . Recombination was found to affect segments of length 1826 bp on average ( 95%CI [1670 , 1980] ) which is comparable to the lengths of recombination tracts estimated when comparing four genomes of Typhimurium [40] as well as the lengths of the regions that were exchanged by Typhi and Paratyphi A [22] . We further studied recombination by looking at its specific role and patterns within each of the five lineages of enterica . The role played by recombination seems to be uneven across these five lineages according to the Structure results in Figure 2 . The isolates in recently diversified populations 1 and 5 showed no admixture ( 1% of material from other populations ) whereas the isolates in population 4 , 3 and 2 had acquired 4% , 11% and 12% respectively of their genetic material from a different population ( Figure 2 ) . To confirm this observation , we extracted from ClonalFrame output the numbers of mutation events , recombination events , and substitutions introduced by recombination for each of the five lineages ( Table 1 ) . Recombination was found to have played a much more important role relative to mutation in lineages 2 and 3 ( = 2 . 17 and 2 . 95 respectively ) than in lineages 1 and 5 ( = 0 . 20 and 0 . 15 respectively ) , and a somewhat intermediate role in lineage 4 ( = 0 . 82 ) . These results are in good qualitative agreement with those of Structure ( Figure 2 ) . Since lineages 1 and 5 are the most recently evolved from a common ancestor , these results point to a possible reduction in the role played by recombination in these two lineages , and maybe even throughout enterica . ClonalFrame estimated that within the regions imported by recombination , an average of of the nucleotides were substituted ( 95%CI [0 . 31% , 0 . 33%] ) . This value of is significantly lower than the average pairwise distance between two members of enterica which is around 1% [12] . The same applies to the distribution of genetic diversity introduced by recombination events ( Figure S5 ) . This observation goes against the natural tendency of ClonalFrame which is to identify more readily events between distantly related types [40] , [41] , and therefore indicates that recombination happened predominantly between related strains during the evolution of enterica , with recombination between distinct lineages being rarer . We attempted to attribute an origin to each recombination event found by ClonalFrame in the five lineages following the method of [41] . Table S3 shows the events for which an origin could be unambiguously attributed , and Figure 4 illustrates the flux of recombination between the five lineages as well as the events coming from other origins within enterica . In lineages 1 , 3 and 5 , the majority of events was found to come from within these lineages even if ClonalFrame is predisposed to underestimate the propensity of such events [40] . In lineages 2 and 4 however , the primary source of recombination events was “External” , i . e . not contained within one of the five lineages ( Figure 4 ) . The origin of these events was not attributed to any isolate or group of isolates in particular , but seemed to come fairly uniformly from all parts of enterica minus the five lineages .
We have sequenced approximately one tenth of the core genome from 114 isolates of enterica from global sources in order to study its population structure . We identified five clear lineages , defined as groups of isolates having the same majority of ancestry in the Structure analysis and representing a clade in the ClonalFrame analysis . It is likely that other similar lineages exist and would be identified using a larger sample of strains . For example , the four strains of serovar Heidelberg ( labelled 44 , 45 , 70 and 81 ) were closely related to each other ( Figure 3 ) and would probably have been called a lineage in our analysis if our sample had contained one or two more similar isolates , since lineage 5 was reconstructed based on only 5 isolates ( Table 1 ) . Our analysis did not include any isolate of serovar Typhi , which has previously been shown based on whole-genome comparisons to be highly monomorphic [19] , [24] , [36] and unrelated to other serovars [22] , [45] . In the context of the enterica data reported here , Typhi would thus constitute a separate and independent lineage , with all current Typhi samples descended from a recent common ancestor on this lineage . One of the five lineages we identified is particularly ancient , estimated to be two thirds of the age of enterica . In the absence of an internal mutation rate for enterica [46] , it is currently not possible to date this age in terms of years . This ancient lineage was designated as “clade B” in a previous study based on MLST [12] , which also noted that it might represent the deepest lineage within enterica but that MLST data was insufficient to confirm this hypothesis . Here we provide such data and confirm the existence of this lineage . The identification of this deep lineage is in sharp contrast with a lack of resolution in the deep ancestry of enterica in general ( Figure 3 ) . A star-shaped phylogeny had also been reconstructed before based on MLST data [12] . Two non-mutually exclusive hypotheses can be proposed to explain this observation: a loss of information about clonal relationships due to extensive recombination [47] , and the fast growth of the effective population size shortly following the birth of the population [48] . It is now clear that recombination plays a driving role in the evolution of many bacteria [15] , including Salmonella [14] . It has been noted that recombination happens more often within the subspecies of Salmonella enterica than between members of separate subspecies [13] , but little is known about the details of the recombination process within subspecies enterica . A recent study based on MLST data hinted at an unusually high rate of recombination between the Newport-II and Newport-III groups [11] . However , the number of recombination events detectable with MLST is generally too small to draw hard conclusions about rates of recombination . Here we sequenced a hundred times more data per isolate than MLST , which allowed us to reconstruct many recombination events , thus revealing clear patterns . We found evidence for recombination that varied over at least an order of magnitude across lineages of enterica ( Table 1 ) . Different recombination rates for individual lineages of a same species have been found previously between the seroresistant and serosensitive clades of Moraxella catarrhalis [49] , between lineages I and II of Listeria monocytogenes [50] , [51] , and between the six hypervirulent lineages of Neisseria meningitidis [27] . It is likely that more examples will be found in future studies as improved methods for detecting recombination are applied to large datasets of whole genomes [52] . Recombination events that occurred between distantly related bacteria are easier to detect than events involving close relatives , because they introduce more polymorphism . ClonalFrame is especially biased against the detection of intra-lineage recombination , because it is based on a model of extra-population recombination [40] . In spite of this , we found that recombination was predominantly between members of a lineage in at least three of the five lineages ( Figure 4 ) . At least three hypotheses can be formulated to explain this general pattern . Firstly , certain serovars of enterica are restricted or associated with specific host species [3] which may result in greater opportunities for recombination between related strains , as previously described in Campylobacter jejuni [53] . For instance , lineage 5 consists of isolates of Paratyphi A and Sendai which are restricted to infecting humans [20] , [22] . However , lineage 1 contains serovars Choleraesuis , Paratyphi C and Typhisuis which share the same antigenic formula but are differentially adapted to infecting swine , humans and swine , respectively [54] . The other three lineages contain isolates from serovars that are usually described as ubiquitous [3] . Secondly , imports from a distant source might reduce the fitness of the recipients and therefore be removed by selection . Thirdly , laboratory experiments have shown that in many bacteria the chances of success of an import decrease exponentially with the genetic distance between donor and recipient due to the DNA mismatch repair system [55] , [56] . This decrease is particularly strong in enterica , with recombination between Typhi and Typhimurium reported to be times less likely than within Typhimurium [57] , [56] . The predominance of recombination events within lineages could thus reflect a fundamental property of recombination rather than ecological structuring or selection . The genus Salmonella is now generally accepted to contain two species , S . bongori and S . enterica , the latter of which consists of six subspecies including subspecies enterica which is the subject of the present study [58] , [59] . Many previously named species that had been defined on the basis of phenotypic differences were regrouped into the single species S . enterica on the basis of DNA hybridization results [60] . The difficulty in defining bacterial species stems from our lack of understanding of the processes involved in their formation [61] . Recombination plays a cohesive role in bacteria , so that lineages can evolve into separate species only if recombination is rare between members of distinct lineages [56] , [62] . Computer simulations have shown that reduced recombination between lineages can lead to patterns of genetic diversity that are similar to those observed in nature [12] , [63] . Our reconstruction of recombination flux within and between the five lineages of enterica ( Figure 4 ) strongly supports the existence of barriers to recombination between members of separate lineages . It is therefore possible that the five lineages we identified in enterica represent incipient species which have already diverged too far from each other for recombination to regroup them . Such incipient species have the potential to eventually become separate species unless an important shift in genetic flow occurred like the one that was recently reported between Campylobacter jejuni and coli [64] . Many biological models of bacterial speciation have been proposed in the literature , and it is interesting although speculative to ask ourselves which ones apply to the diversification pattern we described in enterica . Under a strict host-association , speciation would be expected to happen through the periodic selection model where adaptation to a host progressively drives between-lineages divergence whilst constraining the genetic diversity of each lineage [65] , [66] . This model might apply to lineage 5 which contains serovars restricted to humans , but is unlikely to apply to the other four lineages which can be found in a range of hosts . Alternatively , speciation in enterica could be driven by co-evolution with certain bacteriophages which have been shown to infect some serovars more readily than others [67] . Under the geographic mosaic model [68] , [69] , such uneven adaptive pressures can increase the rate of divergence between populations , and this effect was demonstrated in laboratory experiments on Pseudomonas fluorescens [70] . Future research aimed at testing the geographic mosaic theory will need to investigate whether the underlying process is relevant to the evolution of enterica [71] . The results we have described were obtained using two popular analytical tools: Structure [38] and ClonalFrame [40] , which are based on very different evolutionary models . Structure assumes that each individual in the sample is a mixture from a number of unrelated ancestral populations . ClonalFrame assumes that the individuals are related via a phylogenetic framework , but that clonal relationships are occasionally obscured by recombination events . Clearly the Structure model makes more sense for highly recombinogenic species ( for example H . pylori; [72] ) and the ClonalFrame model for mostly clonal bacteria ( for example Yersinia pestis; [73] ) . However , for many species including Salmonella enterica , recombination occurs but is not sufficiently frequent to completely erase all clonal relationships . Species with such intermediate population structure are eminently suitable for analysis by both models . We have demonstrated that a combined approach using both methods can aid interpretations of population structure and ancestry . In order to study genetic flux , we needed to first define lineages on the ClonalFrame phylogeny ( Figure 3 ) , and Structure allowed us to determine which clades represent meaningful populations . Conversely , the clustering by Structure ( Figure 2 ) could easily have been misinterpreted in the absence of the phylogenetic information provided by ClonalFrame . Structure suggested the existence of a sixth population which seemed to be both a frequent donor and recipient of recombination events ( Figure 2 ) . This sixth population is in fact a random mixture of all “other” strains that did not fall into one of the five true lineages ( Figure 3 ) and therefore does not represent a real evolutionary lineage . We therefore interpret this sixth population as an artifact and do not believe that it represents a true evolutionary lineage . In interpreting the levels of mixed ancestry of these five lineages it is also important to note their different relative ages ( Figure 3; Table 1 ) . Older lineages will have had more opportunities for recombination than recent ones , resulting in greater admixture in some lineages than in others . Once the outputs of the two methods were interpreted correctly in the light of each other , it became clear that they were in good agreement and allowed a more detailed and trustworthy analysis than each approach would have allowed on its own .
We analysed a total of 114 previously described isolates of enterica including nine from the Salmonella reference collection A ( SARA; [74] ) , and 63 of the 72 strains in the Salmonella reference collection B ( SARB; [75] ) . The isolates were chosen to span the global diversity of enterica as measured by serotyping and MLST . Table S1 contains the full list of the 114 isolates , including their serotype and Sequence Type ( ST ) in the MLST scheme of [9] . A database of isolates that have been typed using this MLST scheme is accessible at http://mlst . ucc . ie/mlst/dbs/Senterica . The genome of Typhimurium LT2 [17] was aligned using Mauve [76] , [77] against the following ten publicly available genomes from the Genomes OnLine Database ( accessible at http://www . genomesonline . org; [25] ) : Choleraesuis [21] , Dublin ( University of Illinois , unpublished ) , Pullorum ( University of Illinois , unpublished ) , Paratyphi A [20] , Paratyphi B ( University of Washington , unpublished ) , Typhi CT18 [18] , Enteritidis PT4 [78] , Gallinarum [78] , Hadar ( Sanger Institute , unpublished ) and Infantis ( Sanger Institute , unpublished ) . The black circle on Figure 1 shows the proportion of these ten genomes that aligned to various parts of the LT2 genome . We selected 146 regions of length 2000-2500bp each from the core genome of enterica where at least nine of the ten genomes aligned with LT2 . The regions were selected to be distributed evenly around the genome of LT2 ( Figure 1 ) , and to include the location of the MLST fragments of the scheme of [9] . This allowed an assessment of the accuracy of the sequencing and direct assessment of analysis based on MLST data . Table S2 contains the location and gene content of each region . We designed an Affymetrix CustomSeq Resequencing Array to sequence each of the 114 isolates in Table S1 across the 146 genomic regions listed in Table S2 . The reference genome on the microarray was generated by in silico optimisation of the probability of accurately resequencing the 11 genomes above . Briefly , we started with the genome of LT2 as reference , proposed iterative changes accepted only when they decreased the chance of having two differences within 25 bp between the reference and one of the 11 genomes ( which might make them more difficult to call ) , and repeated the process until convergence . Tests performed on an earlier version of our resequencing array showed that such an optimised reference performed better than using the genome of LT2 as reference in terms of both calling and error rates ( data not shown ) . Base calling was performed using the Affymetrix GeneChip Sequence Analysis Software ( GSEQ ) . We excluded the GSEQ calls of differences from the reference sequence which were within 13 bp of each other . Such calls are unreliable because hybridization at the central position of a probe can be affected by additional differences in the flanking 12 bp . Our resequenced data is available from http://www . stats . ox . ac . uk/lab/salmonella . zip . We used the Bayesian analysis tool Structure version 2 . 3 [38] to identify the populations present in our data . The linkage model of Structure was used; this explicitly accounts for the correlation between nearby sites that arise in admixed populations [39] . Four independent runs were performed for each value of the number of populations ranging from 2 to 10 . Each run consisted of 100 , 000 MCMC iterations , of which the first half was discarded as burn-in . Convergence and mixing of the program were found to be acceptable by manual comparison of independent runs with the same value of . The optimal value was found to be by comparing the posterior probabilities of the data given each value of from 2 to 10 ( Figure S1 ) , and identifying the value of where the posterior probabilities plateau as described in [79] . Applying the method of [80] also resulted in the estimate ( Figure S2 ) . We applied the analysis tool ClonalFrame version 1 . 2 [40] to our data . ClonalFrame is a Bayesian inference method which jointly reconstructs the clonal relationships between the isolates in a sample , as well as the location of recombination events that have disrupted the clonal signal . Four independent runs of ClonalFrame were performed each consisting of 200 , 000 MCMC iterations , and the first half was discarded as burn-in . Convergence and mixing of the MCMC were found to be satisfactory by manual comparison of the runs and using the method in [81] . The genealogies estimated by ClonalFrame have branch lengths measured in coalescent units of time , which are equal to the effective population size times the duration of a generation . We multiplied this by the posterior means of the scaled mutation rate and the scaled recombination rate in order to have branch lengths measured in terms of the expected number of mutation and recombination events ( where and are the per-generation rates of mutation and recombination ) . For each branch of the tree reconstructed by ClonalFrame , we extracted the fragments that had a posterior probability of recombination above 0 . 5 throughout and which reached 0 . 95 in at least one position . Each such recombined fragment was then compared with the homologous sequence of all isolates other than those below the affected branch as described [41] . If a match was found with 0 or 1 difference , the origin of the recombination was attributed to the lineage to which the matching isolate belongs . If no match was found , or if several isolates from different lineages matched , the origin of the recombined fragment was considered unresolved . | Salmonella enterica is a species of bacteria that causes severe diseases in humans and animals . We sequenced about a tenth of the genome from a broadly sampled collection of S . enterica . By comparing these genetic sequences , we were able to partially reconstruct the ancestry of this sample . We identified five lineages within S . enterica , one of which is almost as old as the common ancestor of our sample . We also found evidence for frequent homologous recombination in the ancestry of S . enterica , where fragments of genes from one individual bacterium are acquired by a distinct individual . These recombination events make the ancestry harder to reconstruct in its entirety , but also contain interesting information . We found in particular that recombination had happened more often between strains belonging to the same lineage than across lineage boundaries . This observation is compatible with the lineages of S . enterica becoming progressively isolated from each other , which could lead to their gradual splintering into new species . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bacteriology",
"bacterial",
"evolution",
"genomics",
"biology",
"computational",
"biology",
"comparative",
"genomics",
"microbiology",
"salmonella",
"bacterial",
"pathogens"
] | 2011 | Recombination and Population Structure in Salmonella enterica |
Nuclear receptors were originally defined as endocrine sensors in humans , leading to the identification of the nuclear receptor superfamily . Despite intensive efforts , most nuclear receptors have no known ligand , suggesting new ligand classes remain to be discovered . Furthermore , nuclear receptors are encoded in the genomes of primitive organisms that lack endocrine signaling , suggesting the primordial function may have been environmental sensing . Here we describe a novel Caenorhabditis elegans nuclear receptor , HIZR-1 , that is a high zinc sensor in an animal and the master regulator of high zinc homeostasis . The essential micronutrient zinc acts as a HIZR-1 ligand , and activated HIZR-1 increases transcription of genes that promote zinc efflux and storage . The results identify zinc as the first inorganic molecule to function as a physiological ligand for a nuclear receptor and direct environmental sensing as a novel function of nuclear receptors .
Zinc is an essential nutrient for all life , including plants , animals and microbes , because zinc is involved in many different cellular events . Zn2+ binds tightly to many proteins and thereby contributes to their tertiary structure or catalytic activity [1] , and Zn2+ has been proposed to function as a second messenger signaling molecule during synaptic transmission , development , and immune responses [2–4] . Zinc homeostasis is vital for human health . Inadequate dietary intake is a prevalent cause of human zinc deficiency , whereas genetic disorders that disrupt zinc uptake occur rarely . Zinc deficiency causes pathological changes in a wide range of tissues , reflecting the many uses of zinc [5–7] . Zinc excess also causes human pathology , and it may occur systemically or in specific tissues . For example , ischemic injury has been proposed to cause zinc release that mediates cell death [8 , 9] . Several human diseases , including diabetes , cancer , and neurodegenerative diseases , are correlated with genetic variations that affect zinc metabolism [10–12] . Elucidating zinc homeostasis is important for understanding an ancient biological process and may contribute to improving human health . Because excess zinc is toxic , organisms require mechanisms to sense and detoxify high levels of zinc . One fundamental and evolutionarily conserved mechanism is transcriptional regulation of genes involved in zinc detoxification , such as zinc exporters and zinc sequestering proteins [13–16] . However , the regulation of high-zinc–activated transcription remains poorly understood in animals . Critical questions in this field include the following: What is the direct sensor of high zinc ? And how does this sensor activate transcription of zinc homeostasis genes ? The nematode Caenorhabditis elegans is a useful model system for studies of zinc homeostasis because its simple body plan , its transparency , and the availability of powerful genetic techniques facilitate experimental analysis [13 , 17–21] . Studies of C . elegans are relevant to mammalian biology , since the genome encodes evolutionarily conserved zinc transporters and metallothioneins [22 , 23] . The transcription of these genes is regulated by dietary zinc levels in C . elegans , which is also similar to yeast and mammals [24 , 25] . C . elegans contains two metallothionein genes , mtl-1 and mtl-2 , that are induced at the level of transcription in intestinal cells by high dietary zinc [26] . C . elegans contains 14 cation diffusion facilitator ( CDF ) genes that encode transporters for zinc and possibly other metal ions . The CDF genes cdf-2 and ttm-1b are transcriptionally up-regulated in intestinal cells by high dietary zinc , and these transporters promote zinc storage and excretion [14 , 19] . Transcriptional activation of these genes is mediated by the high zinc activation ( HZA ) element , a DNA enhancer [15] . However , the HZA-binding factor has not been identified . Here we describe an unbiased forward genetic screen used to identify mediators of high-zinc–activated transcription that resulted in the discovery of the HZA-binding factor , which we named the high-zinc–activated nuclear receptor ( HIZR-1 ) . hizr-1 encodes a nuclear receptor transcription factor that has an evolutionarily conserved DNA-binding domain ( DBD ) and ligand-binding domain ( LBD ) . We demonstrated that HIZR-1 is both necessary and sufficient to activate transcription of endogenous zinc-homeostasis genes in response to high dietary zinc . Thus , HIZR-1 is the master regulator of high-zinc homeostasis in C . elegans . We used genetic and biochemical approaches to analyze HIZR-1 function . The LBD directly bound zinc , which promoted nuclear accumulation and activation of the protein , indicating the LBD regulates protein activity and zinc is a physiological ligand; the DBD directly bound the HZA enhancer , which mediates transcriptional activation of multiple genes involved in zinc homeostasis [15] . These findings advance the understanding of zinc biology by identifying a sensor for high zinc in animals and elucidate homeostatic systems by defining a positive feedback loop embedded in a negative feedback circuit . Nuclear receptors ( NRs ) were discovered and characterized as sensors of endocrine signals in mammals . These transcription factors contain a regulatory ligand-binding domain that interacts with hormones and a DNA-binding domain that interacts with target genes . Despite decades of effort , only about half of human NRs have identified ligands , raising the possibility that novel classes of ligands remain to be discovered [27] . NRs exist in simple organisms such as sponges that lack endocrine signaling , suggesting NRs might have primordial function in sensing external molecules [28] . However , external ligands have yet to be identified . The analysis of HIZR-1 expands the understanding of the NR superfamily by ( 1 ) identifying transition metals as a new class of physiological ligand that is distinct from previously described classes such as steroids and lipids and ( 2 ) identifying direct nutrient sensing as a new function that may represent a primordial role of NRs .
Homeostasis in response to high zinc is mediated by transcriptional activation of multiple genes in C . elegans , including the zinc transporter genes cdf-2 and ttm-1b [13 , 14] . To identify genes that mediate this response , we screened for mutant animals that displayed abnormal regulation of cdf-2 . To visualize cdf-2 transcription , we used the method of bombardment to generate transgenic animals with an integrated , multicopy array containing a plasmid with the cdf-2 promoter fused to the coding region for green fluorescent protein ( GFP ) ( cdf-2p::gfp ) ( see Materials and Methods ) . Transgenic cdf-2p::gfp animals displayed low-level fluorescence in standard medium and high-level fluorescence in intestinal cells in medium supplemented with zinc ( Fig 1A and 1B ) . We identified one semidominant mutation ( am285 ) that caused increased fluorescence in standard medium , a phenotype we named zinc-activated transcription-constitutive ( Zat-c ) . We identified five recessive mutations ( am279 , am280 , am286 , am287 , and am288 ) that caused reduced fluorescence in medium supplemented with zinc , a phenotype we named zinc-activated transcription-deficient ( Zat-d ) ( Fig 1C and 1D ) . All six mutant strains displayed growth rates similar to wild type when cultured on standard medium . All five Zat-d mutations failed to complement one another , indicating they affect the same gene . The Zat-c and Zat-d complementation groups were positioned in the center of linkage group X ( Fig 1E ) . To determine how these mutations affect transcription of endogenous genes , we analyzed mRNA levels by quantitative PCR ( qPCR ) . In wild-type animals , mRNA levels of cdf-2 , ttm-1b and the metal binding metallothionein , mtl-1 , are increased significantly by zinc supplementation [14 , 19] . By contrast , in am286 mutant animals cdf-2 , ttm-1b , and mtl-1 transcript levels were significantly lower than wild type when exposed to supplemental zinc ( Fig 2A–2C ) . Furthermore , the strain with the am285 semidominant mutation displayed significantly higher levels of cdf-2 , ttm-1b , and mtl-1 transcripts compared to wild type in the absence of supplemental zinc ( Fig 2D–2F ) . Thus , the gene affected by am286 was necessary and the gene affected by am285 was sufficient for high-zinc–activated transcription of multiple endogenous genes . To test the hypothesis that the gene affected by am286 is functionally important for zinc homeostasis , we analyzed growth in the presence of supplemental zinc . am286 mutant animals grew similarly to wild-type animals in standard medium , demonstrating the strain is healthy in standard conditions , but displayed retarded growth compared to wild-type animals when cultured in high dietary zinc ( Fig 2G ) . This growth defect was specific to zinc toxicity , since am286 mutant animals displayed growth similar to wild-type animals in high dietary copper ( Fig 2H ) . Thus , the gene affected by am286 was necessary for normal growth and development in high dietary zinc . To identify the affected gene , we performed whole genome sequencing . All six mutant strains contained mutations in the gene ZK455 . 6/nhr-33 , which is in the mapping interval . A comparison of the predicted protein sequence to protein databases revealed homology to nuclear receptors , and we named the gene high-zinc–activated nuclear receptor ( hizr-1 ) ( S1 Fig ) . HIZR-1 has a conserved DBD that contains two predicted zinc-finger DNA-binding motifs , which is typical of nuclear receptors , and a conserved LBD . The recessive Zat-d alleles include three encoding substitutions of conserved residues in the DBD and two encoding truncated proteins ( one nonsense and one change in a consensus splicing site ) . The semidominant , Zat-c allele encodes a substitution affecting the LBD ( Fig 1E , S1 Table ) . To confirm this gene assignment , we analyzed an independently derived allele and performed rescue experiments . hizr-1 ( gk698405 ) , a nonsense mutation generated by the C . elegans million mutations project [29] , caused a Zat-d phenotype , as predicted ( Fig 1E , S2A Fig , S1 Table ) . Expression of wild-type HIZR-1 protein fused to GFP rescued the am286 Zat-d mutant phenotype , as predicted ( S2B and S2C Fig ) . The molecular and genetic analyses indicate that the five Zat-d mutations are likely strong loss-of-function or null alleles of hizr-1 , whereas the Zat-c mutation is likely a gain-of-function allele of hizr-1 . We hypothesized that zinc is a ligand for HIZR-1 , leading to the predictions that zinc will directly bind the LBD of purified HIZR-1 and high zinc will promote nuclear accumulation of HIZR-1 in animals . To avoid the complication of zinc binding to the DBD , which contains two predicted zinc finger motifs , we used affinity chromatography to partially purify the LBD of HIZR-1 fused to glutathione-S-transferase ( GST ) . GST alone and GST fused to the LBD of the C . elegans DAF-12 NR , which uses dafachronic acid as a ligand [30 , 31] , were used as specificity controls . The amino acid sequences of the DAF-12 LBD and the HIZR-1 LBD are about 15% identical , and about 47% of the residues are weakly similar . Zinc binding was analyzed using radioactive zinc-65 . The LBD of HIZR-1 displayed saturable , high-affinity binding to zinc ( Fig 3A ) . GST alone and the LBD of DAF-12 fused to GST displayed similar low-level binding , demonstrating the protein specificity of the zinc binding ( Fig 3B ) . Nickel and manganese did not effectively compete with zinc for protein binding , demonstrating the interaction between the LBD of HIZR-1 and zinc was metal-selective ( Fig 3C ) . However , copper did display binding to the LBD of HIZR-1 . These results demonstrate a direct , high-affinity , protein-specific and metal-selective interaction between the LBD of HIZR-1 and zinc . These data can be used to estimate the stoichiometry of binding; however , this estimation is subject to important caveats . The crystal structure of GST indicates the protein binds one molecule of zinc , suggesting the stoichiometry of binding is 1 zinc:1 GST protein molecule [32] . If we assume this stoichiometry in our binding reactions , which has not been demonstrated directly , then we estimate the zinc:HIZR-1 LBD stoichiometry is 3:1 ( Fig 3A ) and 4:1 ( S3 Fig ) . The amino acids that typically coordinate zinc in proteins are histidine and cysteine , and zinc is typically coordinated by four such residues . The predicted LBD of HIZR-1 contains 15 histidine and cysteine residues , suggesting it might have the capacity to coordinate three or four zinc ions with these residues . However , the data presented here do not address the role of these residues in zinc binding . These data can be used to estimate a dissociation constant , although this value is subject to important caveats . The calculated dissociation constant was 2 . 6 +/- 0 . 2 μM when the concentration of zinc was varied ( S3 Fig ) and 1 . 7 +/- 0 . 3 μM when the concentration of protein was varied ( Fig 3A ) , which are in good agreement . However , several caveats must be considered in interpreting these calculated dissociation constant values: ( 1 ) The calculation assumes that all added zinc is available for protein binding; however , the binding reaction contained glutathionine , which is known to have zinc binding affinity and is predicted to reduce the concentration of zinc available to bind the protein . ( 2 ) The calculation assumes that every protein molecule is “active” and able to bind zinc , whereas some might have been inactive or denatured . ( 3 ) The calculation assumes that zinc and protein are bound in a 1:1 stoichiometry—as described above , HIZR-1 appears to bind more than one zinc per protein molecule . These data indicate that zinc directly binds the LBD of HIZR-1 with high affinity and metal selectivity . However , further biochemical studies are necessary to accurately define the dissociation constant , stoichiometry , and specific amino acid residues that mediate zinc binding . To characterize regulation in animals , we analyzed the HIZR-1 ( WT ) ::GFP protein that rescues the Zat-d phenotype , indicating it is functional and expressed in a pattern similar to endogenous protein . In animals cultured with no supplemental zinc , HIZR-1 ( WT ) ::GFP displayed low-level expression in intestinal cells; it was primarily localized in the cytoplasm and occasionally in nuclei . By contrast , culture in high dietary zinc resulted in striking HIZR-1 ( WT ) ::GFP accumulation in most nuclei of alimentary tract cells , primarily in the intestine ( Fig 4A , 4B and 4D ) [33] . High dietary copper did not significantly affect the localization , demonstrating metal specificity of this response ( Fig 4D ) . To analyze the am285 gain-of-function mutation , we generated animals that express mutant HIZR-1 ( D270N GF ) ::GFP protein . When cultured with no supplemental zinc , HIZR-1 ( D270N GF ) ::GFP animals displayed significantly more nuclear accumulation of GFP than HIZR-1 ( WT ) ::GFP animals ( Fig 4A , 4C and 4D ) . Thus , the D270N amino acid substitution is sufficient to promote nuclear accumulation of HIZR-1 and transcriptional activation of zinc-activated genes , highlighting the critical regulatory role of the LBD . Together , these results support the model that zinc binding to the LBD causes nuclear accumulation and transcriptional activation of HIZR-1 . We hypothesized that HIZR-1 directly binds the HZA enhancer to mediate transcriptional activation , leading to the predictions that HZA enhancer DNA will interact directly with the DBD of purified HIZR-1 and that the HZA mediates the transcriptional activation activity of HIZR-1 in animals . An electrophoretic mobility shift assay ( EMSA ) was conducted using partially purified , full-length HIZR-1 protein and fluorescently labeled DNA . The 35 base pair DNA sequence was derived from the cdf-2 promoter and included the 15 base pair HZA with 10 flanking base pairs on each side . HIZR-1 protein retarded the migration of the HZA DNA in the gel , indicating a direct interaction ( Fig 5A and 5B ) . The binding was saturable , and an apparent dissociation constant of 20 . 4 +/- 6 . 8 nM was calculated ( Fig 5C ) . Unlabeled wild-type ( WT ) and mutant HZA were used as specificity controls . Unlabeled WT HZA was identical in sequence to the fluorescently labeled HZA while the unlabeled mutant HZA was identical except for randomizing the order of the central 15 base pair HZA . Unlabeled mutant HZA DNA did not effectively compete for binding to HIZR-1 , indicating the binding activity is sequence specific ( Fig 5D ) . These data demonstrate a direct , high-affinity , sequence-specific interaction between HIZR-1 and the HZA enhancer . To investigate hizr-1 ( lf ) mutations that result in substitutions of highly conserved amino acids in the DBD , we conducted EMSA assays with mutant proteins . All three mutant proteins displayed dramatically reduced DNA binding ( Fig 5A and 5B ) . These biochemical and genetic analyses indicate that the DBD of HIZR-1 mediates the interaction with HZA DNA and DNA binding is necessary for the transcriptional response to high dietary zinc in animals . To investigate the role of the HZA enhancer in animals , we utilized a promoter construct that contains three copies of the HZA enhancer upstream of a basal pes-10 promoter driving expression of GFP with a nuclear localization sequence ( Fig 6A ) [15] . Computational analysis revealed that the basal pes-10 promoter does not contain a recognizable HZA element , and experimental analysis demonstrated that the basal pes-10 promoter is not activated by high zinc [15] . In hizr-1 ( + ) transgenic animals , 2% displayed GFP when cultured with no supplemental zinc , whereas 92% displayed GFP in high dietary zinc , demonstrating the promoter is significantly activated by high zinc . By contrast , 0% of hizr-1 ( am286lf ) mutant animals displayed GFP in high dietary zinc , significantly less than hizr-1 ( + ) ( p < 0 . 001 by Chi-squared test ) . Furthermore , 93% of hizr-1 ( am285gf ) mutant animals displayed GFP induction when cultured with no supplemental zinc , significantly more than hizr-1 ( + ) ( Fig 6B–6D , S4 Fig ) . These data show that hizr-1 was necessary and sufficient for high-zinc–activated transcription mediated by the HZA enhancer in animals . In C . elegans , high zinc homeostasis is mediated by a parallel negative feedback circuit; high levels of cytoplasmic zinc increase expression of CDF-2 and TTM-1B , which detoxify zinc by sequestration and excretion , respectively [13 , 14] . Here we show that HIZR-1 plays a pivotal role in this negative feedback circuit by sensing zinc levels , binding the HZA enhancer , and promoting transcriptional activation of these genes . We noticed that the promoter of hizr-1 contains a predicted HZA element ( Fig 7A ) , leading to the hypothesis that HIZR-1 activates transcription of its own promoter . Consistent with this model , the level of hizr-1 mRNA was significantly increased about 4-fold by high dietary zinc in wild-type animals ( Fig 7B ) . This regulation appears to occur at the level of transcription , since a construct containing the hizr-1 promoter driving expression of GFP ( hizr-1p::gfp ) was also induced by high dietary zinc ( Fig 7C–7E ) . By contrast , hizr-1 ( am286lf ) transgenic animals containing this construct did not display increased GFP expression in response to high dietary zinc , indicating that hizr-1 is necessary for this transcriptional activation ( Fig 7D and 7E ) . These results identify a positive feedback circuit , since HIZR-1 protein increases levels of hizr-1 mRNA , which in turn increases levels of HIZR-1 protein . This positive feedback circuit is embedded in and promotes the parallel negative feedback circuits—increased levels of HIZR-1 protein will enhance the activation of cdf-2 and ttm-1b mRNA , promoting zinc homeostasis ( Fig 8 ) .
Sensing high and low levels of zinc is critical for homeostasis . Here we identify the nuclear receptor HIZR-1 as a high zinc sensor in an animal . The DNA-binding domain of HIZR-1 interacted directly with the HZA enhancer in purified extracts , and hizr-1 functioned through the HZA enhancer in vivo to mediate transcriptional activation in response to high zinc . Thus , HIZR-1 appears to be the HZA-binding factor that was postulated by Roh et al . ( 2015 ) when the HZA was identified as the enhancer that mediates transcriptional activation in response to high dietary zinc [15] . Nuclear receptors are typically regulated by ligand binding , which promotes nuclear accumulation and transcriptional activation [27] . We propose that a ligand for HIZR-1 is zinc , and this model is supported by two lines of evidence . First , HIZR-1 accumulated in the nucleus and activated transcription in response to high dietary zinc . Second , the ligand-binding domain of HIZR-1 interacted directly with zinc with high affinity in purified extracts . This zinc affinity of the ligand-binding domain is specific for HIZR-1 , since the related ligand-binding domain of DAF-12 , which uses dafachronic acid as a ligand , did not bind zinc . Furthermore , our genetic studies demonstrate that the ligand-binding domain of HIZR-1 plays a key regulatory role , since a missense mutation in the domain causes a gain-of-function phenotype characterized by constitutive nuclear accumulation and transcriptional activation . The DNA-binding domain of HIZR-1 contains two predicted zinc finger motifs that are likely to bind zinc and promote DNA binding . While the results indicate the ligand-binding domain plays a critical regulatory role , the data do not exclude the possibility that zinc interactions with the DNA-binding domain also regulate HIZR-1 activity . These results establish the function and mechanism of action of hizr-1 as the direct sensor of high zinc and the effector of high-zinc–activated transcription . Our results raise the possibility that HIZR-1 might respond to multiple metal ions . Metal binding by the HIZR-1 LBD binding was relatively metal-specific , as nickel and manganese did not compete effectively with zinc for binding . Interestingly , copper was able to compete for binding to the HIZR-1 LBD , similar to zinc . However , multiple lines of evidence suggest that copper is not a functional ligand for HIZR-1: ( i ) hizr-1 ( lf ) animals were not hypersensitive to high copper toxicity , ( ii ) high copper did not stimulate nuclear accumulation of HIZR-1 , and ( iii ) high copper did not induce the transcription of cdf-2 , ttm-1b , or mtl-1 [15] . Cadmium is similar to zinc , but it is an environmental pollutant rather than a physiological metal . Cadmium activates gene transcription in worms , including some genes that contain HZA elements and also respond to zinc . Further studies are necessary to determine the role of HIZR-1 in cadmium-activated transcription . Transcription factors that play a role in zinc homeostasis have been characterized in several eukaryotic organisms . In the budding yeast Saccharomyces cerevisiae the response to low zinc is mediated by the zinc-responsive activator protein 1 ( ZAP1 ) . The ZAP1 transcription factor binds directly to a conserved DNA element in promoter regions called the zinc-responsive element ( ZRE ) and thereby induces target gene expression [34] . ZAP1 target genes include the zinc importers ZRT1 and ZRT2 that are induced by ZAP1 to promote zinc uptake; ZRT3 is induced to mobilize zinc stored in the vacuole [35] . ZAP1 activity is repressed by high zinc [24] and activated by low zinc; this transcription factor contains multiple zinc finger domains that may play a regulatory role . In the fission yeast Schizosaccharomyces pombe , gene repression in zinc-replete cells is mediated by the Loz1 transcription factor [36 , 37] . The response to high zinc has been characterized in animals based on studies of the metallothionein genes . In mammals , a zinc-finger containing transcription factor called the metal-responsive-element-binding transcription factor-1 ( MTF-1 ) directly binds the metal response element ( MRE ) in the promoters of metallothionein genes [16 , 25 , 38 , 39] . MTF-1 is necessary for the transcriptional response to a wide range of stresses including high cadmium , high zinc , hypoxia and oxidative stress caused by reactive oxygen species . The mechanisms of MTF-1 regulation are controversial , and it is unclear whether MTF-1 senses zinc directly or is part of a system that includes another high zinc sensor [40 , 41] . HIZR-1 is a new type of zinc sensor , and its discovery and characterization represent important advances in understanding mechanisms of high zinc homeostasis . In response to fluctuating zinc levels , organisms maintain zinc homeostasis by regulating the abundance and activity of metallothioneins and zinc transporters . In C . elegans , exposure to high levels of zinc causes induction of CDF-2 , which sequesters zinc in lysosome-related organelles , and TTM-1B , which excretes zinc into the intestinal lumen . CDF-2 and TTM-1B function in a parallel negative feedback circuit , since single mutants display moderate or undetectable hypersensitivity to high zinc , respectively , whereas double mutants display dramatic hypersensitivity to high zinc [14] . Here we elucidate new aspects of the homeostatic system by showing that hizr-1 mediates the transcriptional response of cdf-2 and ttm-1b . Furthermore , HIZR-1 activates transcription of its own mRNA , thereby establishing a positive feedback loop: high dietary zinc increases the activity of HIZR-1 protein , which increases the levels of hizr-1 mRNA and protein , resulting in a further increase in HIZR-1 activity . Because this positive feedback loop is embedded in a negative feedback circuit , it serves to enhance the overall negative feedback circuit . The transcription factor ZAP1 is a key part of a similar feedback circuit in response to low zinc in yeast . Zinc deficiency causes activation of ZAP1 protein , which binds the promoter and activates transcription of the Zap1 gene; this autoregulation is a positive feedback loop [42] . Activated ZAP1 protein also increases the transcription of key zinc importers such as ZRT1 and ZRT2 , representing the negative feedback component of this system[43 , 44] . The similarities between the circuits controlled by HIZR-1 in response to high zinc in animals and ZAP1 in response to low zinc in yeast highlights the utility and importance of embedding a positive feedback loop within negative feedback circuits to maintain homeostasis . The discovery of HIZR-1 establishes a new intersection between two important fields that were previously separate: nuclear receptors and zinc biology . Nuclear receptors were originally defined as endocrine sensors in humans , including the glucocorticoid receptor and estrogen receptor [45 , 46] . Genome analysis identified a nuclear receptor superfamily consisting of about 49 members in mammals . Despite decades of intensive efforts focused on ligand discovery , about half of the nuclear receptors remain “orphans” with unknown physiological ligands . This raises the possibility that novel classes of ligands remain to be discovered . Indeed , the demonstration that zinc functions as a ligand for the HIZR-1 nuclear receptor represents the first example of a new class of physiological metal ligands . All previously described physiological ligands for nuclear receptors are small hydrophobic molecules; established ligand classes include retinoids [47 , 48] , steroids [49] , sterols [30 , 50] , fatty acid derivatives [51] , and other organic molecules , such as heme [52] . Interestingly , several metals have also been reported to bind to the LBD of the estrogen receptor and appear to alter its activity; however , these metalloestrogens are typically classified as endocrine disruptors rather than physiological ligands for the estrogen receptor [53] . The demonstration that a transition metal ion is the physiological ligand establishes a new structural class of nuclear receptor ligand molecules . Furthermore , this finding raises the possibility that nuclear receptors may be sensors of high levels of other essential metal ions , such as iron , copper , and manganese . Nuclear receptors are not present in single-celled eukaryotes , such as yeast , and appear to have evolved in primitive multicellular organisms . Ancestral nuclear receptor genes exist in sponges , animals of the phylum Porifera , which lack higher-level body organization such as tissues and organs and thus lack hormone signaling [28 , 54] . Therefore , nuclear receptors were proposed to have ancestral roles as environmental sensing proteins that later evolved to sense intraorganismal endocrine signals [28] . However , no such functions have been rigorously demonstrated . Our results document a nuclear receptor that responds to a nutrient , consistent with the theory that the ancestral function of nuclear receptors might have been sensing dietary and environmental stimuli , including metals such as zinc . This represents a distinct paradigm from canonical endocrinology in which a hormone like estrogen is synthesized in endocrine cells , is secreted into the bloodstream , and enters distant cells , where it binds and activates the estrogen receptor [55] . Furthermore , this is a novel demonstration that a dietary nutrient is a direct ligand for a nuclear receptor . This establishes a new paradigm for nuclear receptors as direct sensors of environmental nutrients .
C . elegans strains were cultured at 20°C on nematode growth medium ( NGM ) seeded with Escherichia coli OP50 unless otherwise noted [56] . The wild-type strain was Bristol N2 . The Zat mutations hizr-1 ( am279 ) , hizr-1 ( am280 ) , hizr-1 ( am285 ) , hizr-1 ( am286 ) , hizr-1 ( am287 ) , and hizr-1 ( am288 ) are described here . These mutations were generated by mutagenizing the high-zinc reporter strain WU1391 ( cdf-2p::gfp , see Plasmid DNA construction and transgenic strain generation for details ) with ethyl methanesulfonate ( EMS ) [56] and identified by screening for abnormal patterns of fluorescence , as described below . hizr-1 ( gk698405 ) was identified by the C . elegans million mutations project and obtained from the Caenorhabditis Genetics Center [29] . To position newly identified mutations in the C . elegans genome , we used the following mutations on linkage group X that cause visible phenotypes: unc-115 ( e2225 ) , egl-15 ( n484 ) [57] , and sma-5 ( n678 ) . To generate the high-zinc reporter strain WU1391 , we utilized the unc-119 ( ed3 ) mutation [58] . To generate the transcriptional fusion constructs for cdf-2 ( cdf-2p::gfp ) ( pSC24 ) and hizr-1 ( hizr-1p::gfp ) ( pKW11 ) , we polymerase chain reaction ( PCR ) -amplified DNA fragments positioned upstream of the coding sequence using wild-type C . elegans DNA . These fragments were ligated into pBluescript SK+ ( Stratagene ) containing the green fluorescent ( GFP ) coding sequence and the unc-54 3' untranslated region ( UTR ) . The cdf-2 promoter was amplified from the ATG start codon to 1 , 371 base pairs upstream of the ATG start codon . The hizr-1 promoter was amplified from the ATG start codon to 441 base pairs upstream of the ATG start codon . To analyze the HZA enhancer , we used previously described transcriptional reporter constructs containing the basal pes-10 promoter driving transcription of GFP with a nuclear localization sequence ( NLS ) ( pes-10p::gfp-nls ) ( pPD107 . 94 , a gift from A . Fire ) and the basal pes-10 promoter with three copies of the HZA enhancer inserted into the promoter ( 3XHZApes-10p::gfp-nls ) ( pID24 ) [15] . To generate the translational fusion construct for HIZR-1 , [HIZR-1 ( 1–412 WT ) ::GFP] ( pKW1 ) , we PCR-amplified the hizr-1 genomic locus from the C . elegans fosmid WRM069cE11 . This fragment was ligated into pBluescript SK+ ( Stratagene ) containing the GFP coding sequence and the unc-54 3' UTR . The hizr-1 locus was amplified from 444 base pairs upstream of the ATG start codon to the TAA stop codon . The TAA stop codon was mutated to TAT to allow translation of the C-terminal GFP . To generate the HIZR-1 ( 1–412 D270N GF ) ::GFP construct ( pKW8 ) , we modified plasmid pKW1 using Agilent QuickChange II Site-Directed Mutagenesis Kit according to manufacturer’s instructions . To integrate the cdf-2p::gfp transcriptional fusion construct into the C . elegans genome , we ligated the DNA fragment encoding the cdf-2 promoter driving expression of GFP with the unc-54 3' UTR into the plasmid pMM016 that contains the wild-type unc-119 locus ( unc-119 ( + ) ) [59] ( pSC24 ) . pSC24 was bombarded into unc-119 ( ed3 ) animals [13 , 59] , and nonUnc animals that segregated only nonUnc self progeny were selected . The cdf-2p::gfp unc-119 ( + ) transgene is integrated on the right arm of linkage group IV and was assigned the allele name amIs10 . The following transgenic strains with amIs10 were used in this study: WU1391 ( cdf-2p::gfp outcrossed seven times to N2 ) , WU1518 ( cdf-2p::gfp outcrossed seven times to Hawaiian CB4856 ) , cdf-2p::gfp; hizr-1 ( am279 ) , cdf-2p::gfp; hizr-1 ( am280 ) , cdf-2p::gfp; hizr-1 ( am285 ) . cdf-2p::gfp;hizr-1 ( am286 ) , cdf-2p::gfp; hizr-1 ( am287 ) , and cdf-2p::gfp; hizr-1 ( am288 ) . Transgenic animals containing extrachromosomal arrays were generated by injecting the gonad of worms with a plasmid of interest ( pKW1 , pKW8 , pKW11 , pID24 , or pPD107 . 94 ) and a co-injection marker [60] . The co-injection marker was either myo-3p::mCherry ( pCJF104 ) [61] or the plasmid pRF4 encoding the dominant ROL-6 ( R71C GF ) mutant protein [60] . Transgenic animals were selected by mCherry expression in body-wall muscles or by the Rol phenotype . The following transgenic strains with extrachromosomal arrays were used in this study: hizr-1 ( am286 ) ; HIZR-1 ( 1–412 WT ) ::GFP , hizr-1 ( am286 ) ; HIZR-1 ( 1–412 D270N GF ) ::GFP , hizr-1 ( + ) ; 3XHZApes-10p::gfp-nls , hizr-1 ( am286 ) ; 3XHZApes-10p::gfp-nls , hizr-1 ( am285 ) ; 3XHZApes-10p::gfp-nls , hizr-1 ( am285 ) ; pes-10p::gfp-nls , hizr-1 ( + ) ; hizr-1p::gfp , and hizr-1 ( am286 ) ; hizr-1p::gfp . All transgenic strains contained the pRF4 dominant Rol marker except for hizr-1 ( + ) ; 3XHZApes-10p::gfp-nls which contained the myo-3p::mCherry marker . To generate plasmid constructs for protein purification of full-length HIZR-1 , we PCR-amplified the complete coding sequence of HIZR-1 from synthesized DNA ( IDT gBlocks ) . This fragment was ligated into pTrcHisA ( ThermoFisher ) . This plasmid encodes an N-terminal 6 histidine affinity purification tag ( 6XHis ) fused to amino acids 1–412 of HIZR-1 ( pKW2 ) . The pKW2 plasmid was modified using the Agilent QuickChange II Site-Directed Mutagenesis Kit according to manufacturer’s instructions to generate the plasmids pKW4 , pKW5 , and pKW6 that encode full-length HIZR-1 proteins with the G23E , S30L , and R63C amino acid substitutions , respectively . To generate plasmid constructs for protein purification of the ligand-binding domains of HIZR-1 and DAF-12 , we PCR-amplified the ligand-binding domains of HIZR-1 ( encoding amino acids 101–412 ) and DAF-12 ( encoding amino acids 440–753 of the A isoform ) from synthesized DNA ( IDT gBlocks ) . These fragments were ligated into pGEX-4T-1 ( GE Healthcare ) . These plasmids encode an N-terminal glutathione S-transferase ( GST ) affinity purification tag fused to either the ligand-binding domain of HIZR-1 ( pKW14 ) or DAF-12 ( pKW15 ) . All plasmid constructs were verified by DNA sequencing using standard methods . To position Zat mutations on the genetic map , we utilized single nucleotide polymorphism ( SNP ) markers [18] . To facilitate this strategy , we introduced the integrated cdf-2p::gfp reporter from WU1391 into Hawaiian CB4856 by performing seven backcrosses to CB4856 with selection for zinc-activated GFP fluorescence . All six Zat-c and Zat-d mutations displayed tightest linkage to the SNP amP117 , positioned at approximately the 9 , 552 kilobase pair on linkage group X ( Fig 1E ) . To determine how many genes were affected by the five recessive Zat-d mutations , we performed complementation experiments . All five mutations failed to complement one another for the Zat-d phenotype , indicating that all five mutations affect the same gene . To define intervals that contain the Zat-c mutation ( am285 ) and the Zat-d complementation group ( represented by am286 ) , we conducted three-factor mapping experiments with mutations that cause visible phenotypes [56] . We chose the X-linked genes unc-115 , egl-15 , and sma-5 that are located at approximately the 10 , 147 , 11 , 016 , and 12 , 005 kilobase pairs on linkage group X , respectively . The cdf-2p::gfp reporter was introduced into the double mutant mapping strains to generate cdf-2p::gfp; unc-115 ( e2225 ) egl-15 ( n484 ) or cdf-2p::gfp; egl-15 ( n484 ) sma-5 ( 678 ) using standard genetic techniques . For the am285 Zat-c mutation , 0/13 Egl nonUnc and 5/5 Unc nonEgl recombinants segregated the Zat-c phenotype , indicating am285 is positioned right of egl-15 . 9/20 Egl nonSma recombinants segregated the Zat-c phenotype , indicating am285 is positioned between egl-15 and sma-5 . For the am286 Zat-d mutation , 0/12 Egl nonUnc and 3/3 Unc nonEgl recombinants segregated the Zat-d phenotype , indicating am286 is positioned right of egl-15 . 9/30 Egl nonSma recombinants segregated the Zat-d phenotype , indicating am286 is positioned between egl-15 and sma-5 . The results that 18/50 recombination events ( 36% ) occurred between egl-15 and hizr-1 and 32/50 recombination events ( 64% ) occurred between hizr-1 and sma-5 are consistent with the molecular distances of approximately 302 kilobase pairs between egl-15 and hizr-1 ( 31% ) and 687 kilobase pairs between hizr-1 and sma-5 ( 69% ) . To identify the gene affected by the newly isolated Zat mutations , we performed whole genome sequencing using DNA from the am279 , am280 , am285 , and am286 mutant strains . Candidate mutations within the mapping interval were identified by comparing the mutant DNA sequence to wild-type DNA sequence . am279 , am280 , am285 , and am286 all contained candidate mutations in the gene hizr-1/ ( ZK455 . 6 ) ( Fig 1E , S1 Table ) . Next , the sequence of the hizr-1 genomic locus was determined by standard sequencing techniques using DNA from am287 and am288 mutant strains , which revealed mutations in the hizr-1 locus . RNA isolation and cDNA synthesis were performed as previously described [19] . Mixed-developmental stage populations of animals were cultured on NGM dishes . Animals were washed and cultured on Noble agar minimal media ( NAMM ) dishes with and without supplemental zinc sulfate . NAMM dishes were seeded with concentrated OP50 E . coli . After 16–24 h of culture on NAMM dishes , animals were collected by washing for RNA extraction . RNA was extracted using TRIzol ( Invitrogen ) , treated with DNase I , and cDNA was synthesized using the High-Capacity cDNA Reverse Transcription kit ( Applied Biosystems ) according to the manufacturer’s instructions . Quantitative PCR ( qPCR ) was performed using a 7900HT Fast Real-Time PCR system ( Applied Biosystems ) and the SYBR Green PCR Master Mix ( Applied Biosystems ) following the manufacturer’s instructions . mRNA fold change was calculated using the comparative CT method [62] . For all qPCR experiments , mRNA levels were normalized to rps-23 and error bars indicate standard deviation . Forward and reverse amplification primers were: rps-23 5'-aaggctcacattggaactcg and 5'-aggctgcttagcttcgacac; cdf-2 5'-atagcaatcggagagcaacg and 5'-tgtgacaattgcgagtgagc; ttm-1b 5'-catgggcactcacacacacac and 5'-ctcggcgacccttttgatatttc; hizr-1 5'-tcattttgcggtttcatcgtg and 5'-catcgcgtgtatctacagctac; and mtl-1 5'-ggcttgcaagtgtgactgc and 5'-cctcacagcagtacttctcac . Synchronized embryos were placed on dishes containing NAMM that were seeded with concentrated OP50 E . coli and supplemented with zinc sulfate ( ZnSO4 ) or copper chloride ( CuCl2 ) and cultured for 3 d [18] . Animals were then mounted on 2% agarose pads on microscope slides and imaged with a Zeiss Axioplan 2 microscope equipped with a Zeiss AxioCam MRm digital camera . The length of individual animals ( tip of head to end of tail ) was measured using ImageJ software ( NIH ) . Live hizr-1 ( am286 ) ; HIZR-1 ( 1–412 WT ) ::GFP or hizr-1 ( am286 ) ; HIZR-1 ( 1–412 D270N GF ) ::GFP transgenic animals ( L4 or young adult ) were washed and cultured on NAMM dishes with or without zinc sulfate for 12–16 h . Transgenic animals were then immobilized and mounted onto a microscope slide with a thin pad of 2% agarose . All images were captured using a Zeiss Axioplan 2 microscope equipped with a Zeiss AxioCam MRm digital camera using identical settings and exposure times in paired experiments ( Fig 4A–4C ) . To score alimentary nuclei per animal with detectable HIZR-1::GFP ( Fig 4D ) , we examined live animals using an Olympus SZX12 dissecting microscope equipped with GFP fluorescence and counted GFP-positive alimentary nuclei . Animals were exposed to 200 μM zinc or 300 μM copper because they are approximately equally toxic to C . elegans [63] . Transgenic L4 or young adult animals expressing cdf-2p::gfp ( Fig 1A and 1B ) or hizr-1p::gfp ( Fig 7C–7E ) were cultured for 12–16 h on NAMM dishes with or without supplemental zinc sulfate . Transgenic animals were immobilized , mounted , and imaged as described above . GFP fluorescence intensity was quantified using ImageJ software ( NIH ) . To determine the percent of GFP-positive animals in a population of transgenic animals expressing P3XHZApes-10p::gfp-nls or pes-10p::gfp-nls , we examined live animals using an Olympus SZX12 dissecting microscope equipped with GFP fluorescence . Animals displaying one or more GFP-positive alimentary nuclei were classified as GFP positive ( Fig 6D ) . Representative images were captured using a Zeiss Axioplan 2 microscope equipped with a Zeiss AxioCam MRm digital camera using identical settings and exposure times in paired experiments ( Fig 6B and 6C and S4 Fig ) . Plasmids encoding full-length HIZR-1 ( 1–412 ) proteins with the wild-type ( WT ) or mutant sequence ( G23E , S30L , or R63C ) fused to an N-terminal 6XHis tag were transformed into BL21 E . coli cells . The empty vector pTrcHisA was transformed as a control . Cells were grown in Luria-Bertani media at 37°C , and expression was induced with 5 mM IPTG when the absorbance at 600 nm reached between 0 . 5–0 . 7 . Expression was induced for 16–18 h at 16°C . Cells were then collected by centrifugation and suspended in 50 mM MOPS ( pH 7 . 0 ) . Cells were lysed by sonication in the presence of 0 . 16 mg/mL of lysozyme . Lysed cell material was pelleted by centrifugation and the supernatant was collected . HIZR-1 protein was purified from the supernatant using Clontech TALON Metal affinity resin . GST alone and the ligand-binding domains of HIZR-1 and DAF-12 fused to GST were expressed and harvested using the techniques described above; these proteins were purified using Genscript Glutathione Resin according to the manufacturer’s instructions and eluted with 10 mM L-glutathionine . To analyze zinc binding to proteins , we used zinc-65 radionuclide ( PerkinElmer , stock date 2/12/2015 , specific activity 3 . 29 mCi/mg , concentration 5 . 90 mCi/mL , and radionuclidic purity 99 . 00% ) and the purified proteins GST::HIZR-1 ( 101–412 WT ) , GST::DAF-12 ( 440–753 WT ) , and GST alone . Protein concentrations were quantified by Bradford assay . Equilibrium binding experiments were performed according to established guidelines [64] . Briefly , a constant amount of zinc-65 ( 0 . 01 μCi ) was incubated with variable protein concentrations in 50 mM MOPS buffer with 10 mM L-glutathionine ( pH 7 . 0 ) ( Fig 3A and 3B ) . Based on the specific activity , we calculated that addition of 0 . 01 μCi zinc-65 results in a final concentration of zinc-65 of about 0 . 4 nM and a final concentration of total zinc ( radioactive and nonradioactive ) of about 1 μM . The half life of zinc-65 is 244 d , and it decays to copper-65 . The number of decay half lives between production of the zinc-65 solution and binding experiments was less than one . Therefore , the amount of copper-65 was negligible compared to the amount of nonradioactive zinc in the source . Although purified proteins were eluted in buffer with no added zinc , we cannot exclude the possibility that the purified proteins contributed some nonradioactive zinc to the reaction mixture . Alternatively , a constant amount of protein ( 0 . 77 μM ) was incubated with variable zinc concentrations ( S3 Fig ) . Therefore , reactions consisted of ( 1 ) protein in 50 mM MOPS buffer with 10 mM L-glutathionine ( pH 7 . 0 ) and ( 2 ) zinc dissolved in water . Reactions were allowed to equilibrate for 40 min , protein was vacuum blotted onto nitrocellulose membranes , membranes were briefly washed , and bound zinc-65 was quantified using a Beckman LS600 scintillation counter . At least two technical replicates were performed for each unique protein or zinc concentration . The dissociation constant was determined using GraphPad Prism software , assuming a 1:1 binding stoichiometry between zinc and the protein of interest . To conduct the metal selectivity experiments , the binding interaction between a constant amount of GST::HIZR-1 ( 101–412 WT ) protein and zinc-65 was competed using no competitor or 500 μM of either nonradioactive zinc sulfate ( ZnSO4 ) , copper chloride ( CuCl2 ) , nickel chloride ( NiCl2 ) , or manganese chloride ( MnCl2 ) ( Fig 3C ) . Binding values were normalized by defining the binding interaction between the protein and zinc-65 with no nonradioactive competitor as maximal binding and setting that value equal to 1 . 0 . Fraction maximal binding was calculated as the bound zinc-65 ( CPM ) for a given nonradioactive metal divided by the bound zinc-65 for no competitor ( CPM ) . EMSAs were conducted using the Licor Odyssey EMSA Buffer Kit according to the manufacturer’s instructions . Images were captured with a Licor Odyssey Infrared Imager , and gel bands were quantified using Licor Image Studio software . To test the effect of the Zat-d ( am279 , am280 , and am287 ) mutations on DNA-binding activity , a constant amount of full-length wild-type and mutant ( G23E , S30L , and R63C ) protein was incubated with a constant amount of labeled HZA oligonucleotide ( Fig 5A ) . Protein concentrations were determined by Bradford assay and confirmed to be equivalent by western blot ( Fig 5B ) . Proteins were visualized utilizing a Pierce 6XHis eptiope tag antibody with a Licor IRDye800 goat anti-mouse antibody . Blots were imaged with a Licor Odyssey Infrared Imager . The dissociation constant was determined by incubating a constant concentration of full-length wild-type HIZR-1 ( 1–412 WT ) protein ( pKW2 ) with variable concentrations of labeled HZA DNA ( Fig 5C ) . The dissociation constant was calculated using GraphPad Prism software . To conduct sequence specificity experiments , the binding interaction between a constant amount of HIZR-1 ( 1–412 WT ) protein and labeled HZA DNA was competed using variable concentrations of either unlabeled WT or mutant HZA DNA oligonucleotides ( Fig 5D ) . The IRDye700 labeled HZA oligonucleotide was 5'- tgtgttatcaatcataaactagaacatgtctcgag-3' . The unlabeled oligonucleotides used were wild-type HZA , 5'-tgtgttatcaatcataaactagaacatgtctcgag-3' and Mutant HZA , 5'-tgtgttatcagaacatacaacattaatgtctcgag-3' . These DNA oligonucleotides were double stranded . All data were analyzed utilizing the two-tailed Student’s t test of samples with unequal variance except for data in Fig 6D , which was analyzed by the Chi-squared test . When displayed , error bars indicate standard deviation . p-Values less than 0 . 05 were considered statistically significant . All the statistical comparisons in the current study consist of comparing two values made in different experiments performed in parallel . One type of comparison is wild-type animals compared to mutant animals , a pairwise comparison of two values determined in different experiments that were performed in parallel . A second type of comparison is the same genotype of animals compared in two distinct environmental conditions . | Zinc is an essential nutrient for all life forms , and maintaining zinc homeostasis is critical for survival . However , little is known about how animals sense changes in zinc availability and make adjustments to maintain homeostasis . In particular , logic dictates there must be a mechanism for zinc sensing , but it has not been defined in animals . We discovered that the nuclear receptor transcription factor HIZR-1 is the master regulator of high zinc homeostasis in the roundworm Caenorhabditis elegans . In response to high dietary zinc , HIZR-1 activates transcription of multiple genes that encode a network of proteins that store and detoxify excess zinc . Furthermore , our results suggest HIZR-1 itself is the high zinc sensor , since it directly binds zinc ions in the ligand-binding domain that regulates transcriptional activation . These findings advance the understanding of zinc homeostasis and nuclear receptor biology . Nuclear receptors were initially characterized as receptors for hormones such as estrogen , indicating complex animals use these transcription factors to monitor their internal environment . However , nuclear receptors are present in simple organisms that lack endocrine signaling , suggesting these transcription factors might have a primordial function in sensing the external environment . Our results identify a new class of nuclear receptor ligands , the inorganic ion zinc , and a new function for nuclear receptors in directly sensing levels of a nutrient . We speculate that nutrient homeostasis mediated by direct binding of nutrients to the ligand-binding domain is a primordial function of nuclear receptors , whereas endocrine signaling in complex animals mediated by direct binding of hormones to the ligand-binding domain is a derived function of nuclear receptors that appeared later in evolution . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biotechnology",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"affinity",
"chromatography",
"caenorhabditis",
"messenger",
"rna",
"dna-binding",
"proteins",
"animals",
"dna",
"transcription",
"animal",
"models",
"physiological",
"processes",
"caenorhabditis",
"... | 2017 | The Nuclear Receptor HIZR-1 Uses Zinc as a Ligand to Mediate Homeostasis in Response to High Zinc |
Our previous studies showed that Trichinella spiralis paramyosin ( TsPmy ) is an immunomodulatory protein that inhibits complement C1q and C8/C9 to evade host complement attack . Vaccination with recombinant TsPmy protein induced protective immunity against T . spiralis larval challenge . Due to the difficulty in producing TsPmy as a soluble recombinant protein , we prepared a DNA vaccine as an alternative approach in order to elicit a robust immunity against Trichinella infection . The full-length TsPmy coding DNA was cloned into the eukaryotic expression plasmid pVAX1 , and the recombinant pVAX1/TsPmy was transformed into attenuated Salmonella typhimurium strain SL7207 . Oral vaccination of mice with this attenuated Salmonella-delivered TsPmy DNA vaccine elicited a significant mucosal sIgA response in the intestine and a systemic IgG antibody response with IgG2a as the predominant subclass . Cytokine analysis also showed a significant increase in the Th1 ( IFN-γ , IL-2 ) and Th2 ( IL-4 , 5 , 6 , 10 ) responses in lymphocytes from the spleen and MLNs of immunized mice upon stimulation with TsPmy protein . The expression of the homing receptors CCR9/CCR10 on antibody secreting B cells may be related to the translocation of IgA-secreted B cells to local intestinal mucosa . The mice immunized with Salmonella-delivered TsPmy DNA vaccine produced a significant 44 . 8% reduction in adult worm and a 46 . 6% reduction in muscle larvae after challenge with T . spiralis larvae . Our results demonstrated that oral vaccination with TsPmy DNA delivered by live attenuated S . typhimurium elicited a significant local IgA response and a mixed Th1/Th2 immune response that elicited a significant protection against T . spiralis infection in mice .
Trichinellosis , a serious food-borne parasitic zoonosis and an important public health problem worldwide , is mainly caused by infection with the tissue-dwelling nematode Trichinella spiralis [1 , 2] . People develop this infection through ingestion of raw or undercooked meat contaminated with encapsulated parasite larva . Domestic pork has been the major source of this infection in China and other countries . Due to the increased consumption of pork and other meat , trichinellosis is an emerging or re-emerging disease in many countries [2] . In China , 17 outbreaks of human trichinellosis were reported , with 828 cases and 11 deaths in eight provinces between 2000 and 2003 [3] . The development of vaccine ( s ) has become an urgent need for controlling trichinellosis in human and domestic animals . T . spiralis is an intestinal parasite whereby the adult worm dwells in the small intestine with the head embedding into the mucosa and the epithelial layer . Female worms produce newborn larvae that penetrate into the intestinal wall and migrate through the blood circulation to the muscle tissue where they form cysts . Obviously , the intestinal mucosa becomes the site for parasite host interaction and the first barrier for protecting the host against Trichinella infection [4] . Therefore , the local mucosal immune response is crucial for establishing protective immunity against intestinal parasite such as T . spiralis . Other studies have shown that attenuated Salmonella typhimurium is an effective oral delivery vector for heterologous antigens to induce long-lasting mucosal and systemic immune responses against infections with intestinal pathogens , providing an efficient design for novel vaccination strategies [5 , 6] . This novel delivery system has proven successful in inducing protective immunity against many infections such as Toxoplasma gondii[7] , Giardia lamblia[8] and Trypanosoma cruzi[9] . In our previous study , attenuated Salmonella typhimurium was used to orally deliver a DNA vaccine of Ts87 , an excretory/secretory antigen from T . spiralis , which has shown significant protection against T . spiralis larval challenge in a mouse model [10] . Additional evidence has shown that attenuated bacteria could effectively induce a mucosal immune response and enhance antibody secreting cells ( ASCs ) homing to the epithelium of the intestine . The secretory IgA ( sIgA ) in the mucosal immune response plays important roles in killing or expelling intestinal pathogens [11 , 12] . In this study , we developed a new DNA vaccine targeting TsPmy , the paramyosin protein of T . spiralis that induced protective immunity when recombinant protein was used [13] , that was delivered by attenuated S . typhimurium . Mice orally vaccinated with Salmonella-delivered TsPmy DNA vaccine elicited robust mucosal and systemic immune responses that induced a significantly protective immunity against T . spiralis larval challenge .
Female BALB/c 6–8 weeks old mice were provided by Laboratory Animal Services Centre of Capital Medical University . Mice were raised under specific pathogen-free conditions with suitable temperature and humidity . All experimental procedures were reviewed and approved by the Capital Medical University Animal Care and Use Committee ( approval number: 2012-X-108 ) and complied with the NIH Guide for the Care and Use of Laboratory Animals . The attenuated S . typhimurium SL7207 strain that could deliver heterologous antigens with the virulent gene aroA knockout and was not pathogenic to mice via oral administration was kindly provided by Prof . J . S . He of Beijing Jiaotong University . The full-length DNA encoding for TsPmy ( accession number: EF429310 ) was amplified from T . spiralis total cDNA using the following primers: forward , 5’-CGGGATCCATGTCTCTGTATCG CAGTCCCAGT-3’ and reverse 5’-CGGAATTCATATTCATGTCCTTCT TCCATCAC-3’ . The amplified DNA fragment was cloned into the eukaryotic expression vector pVAX1 ( Invitrogen , USA ) at the BamHI and EcoRI sites . The correct insert sequence and reading frame was confirmed by double-stranded DNA sequencing using the vector flanking primers T7 promoter and BGH reverse primer . The sequence-confirmed recombinant plasmid pVAX1-TsPmy and the empty plasmid pVAX1 were transformed into attenuated S . typhimurium strain SL7207 by electroporation , and the transformants were selected on LB agar plates containing 50μg/ml kanamycin and identified by PCR amplification with TsPmy specific primers and DNA sequencing . The T . spiralis ISS 533 strain was maintained in female ICR mice . Each mouse was orally infected with 500 T . spiralis infective larvae . The adult worms were isolated from the intestines of infected mice at 5 days following larval challenge . The muscle larvae ( ML ) were recovered at 42 days post-infection from the muscle tissue of infected mice using a modified pepsin-hydrochloric acid digestion method [14] . A full-length cDNA encoding TsPmy was cloned into the expression vector pET-28a ( + ) . The recombinant plasmid containing the TsPmy coding gene was transformed into E . coli BL21 . The rTsPmy was expressed as insoluble inclusion body under induction of 1 mM IPTG and the urea-denatured rTsPmy was purified by Ni-affinity chromatography ( Qiagen , USA ) as previously described [15] . A total of 120 mice were randomly divided into three groups with 40 mice each . The first two groups were immunized orally with 1×108 attenuated S . typhimurium SL7207 transformed with pVAX1-TsPmy ( SL7207/pVAX1-TsPmy ) or with empty pVAX1 ( SL7207/pVAX1 ) in 100 μl of PBS . The third group of mice was given 100 μL of PBS only as control . All mice were given 100 μL of 10% NaHCO3 orally to neutralize gastric acids before oral inoculation with bacteria or PBS . All groups of mice were boosted twice with the same vaccine components at two weeks interval . One week after each immunization , 5 mice from each group were sacrificed by CO2 inhalation . The serum , spleen , and mesenteric lymph nodes ( MLNs ) were collected to evaluate the levels of immune responses , and the intestines were collected with lavage fluid for measuring mucosal IgA . All mice were monitored by research personnel on a daily basis for general appearance , hunched posture , rough haircoat , labored breathing , lethargy , lameness , ataxia , diarrhea , abnormal vocalization and abnormal discharge from the eyes or nose . If any animal has bleeding diarrhea , labored breathing , severe leg injuries or become moribund it will be euthanized immediately by CO2 inhalation . One week after the first immunization , TsPmy mRNA was measured in the MLNs , spleen and liver tissues of immunized mice by reverse transcription polymerase chain reaction ( RT-PCR ) with the TsPmy-specific primers listed above . Total RNAs were isolated from the MLNs , spleen and liver tissues of immunized mice using TRIzol ( Invitrogen , USA ) according to the manufacturer’s instructions . First strand total cDNA was reversely transcribed from total RNAs using poly-T primer , and the Tspmy cDNA was amplified from the total cDNA using TsPmy specific primers . Mouse GAPDH cDNA was amplified from the same sample as a positive control . PCR products were detected by electrophoresis on 1% agarose gels . To determine the expression of rTsPmy in vivo , MLNs of mice immunized with SL7207/pVAX1-TsPmy were fixed , frozen and cryosectioned . Tissue sections were washed three times with cold PBS and blocked with 5% normal goat serum ( NGS , diluted in PBS , pH 7 . 6 ) at room temperature for 30 min . After incubation with anti-TsPmy monoclonal antibody 9G3[16] diluted 1:2000 in PBS plus 5% NGS at 4°Covernight , the tissue section was incubated with DyLight TM488-conjugated goat anti-mouse IgG at a 1:200 dilution . The MLN sections from mice receiving PBS only were incubated with the same antiserum as a negative control . The sections were examined and photographed using a fluorescence microscope ( Leica , Germany ) . Enzyme-linked immunosorbent assay ( ELISA ) was used to analyze the levels of antigen-specific IgG , IgG1 and IgG2a antibodies in the sera of the immunized mice as previously described [13] . The optical density ( OD ) at 450 nm was measured using an ELISA reader on sera diluted at 1:200 . To detect total or antigen-specific secretory IgA ( sIgA ) in intestinal washes , the intestinal lavage washes were prepared as described [17 , 18] . Briefly , 10 cm of the small intestine beginning at the gastro-duodenal junction was cut for each sacrificed mouse , and the interior of the small intestine was flushed twice with 2 mL cold PBS . After centrifugation at 800×g for 10 min , the supernatants were harvested and stored at −80°C until use . Intestinal total sIgA was assessed using a sandwich-type ELISA by trapping intestinal mucosal IgA on a plate coated with purified rat anti-mouse IgA ( BD Biosciences , USA ) , and the specific anti-TsPmy sIgA was measured by standard ELISA using rTsPmy coated plates as described[10] . To measure the specific cellular immune responses in the lymphocytes isolated from the spleen and MLNs of immunized mice upon stimulation with rTsPmy , the production of IFN-γ , IL-2 , IL-4 , IL-5 , IL-6 and IL-10 was detected using an ELISPOT assay according to the manufacturer’s instructions ( BDTM ELISPOT , USA ) . Briefly , the mice were sacrificed one week after the third immunization , and the single lymphocyte suspensions were prepared by dissociating the spleen and MLNs through a mesh into the lymphocyte separation medium ( Dakewe , China ) . The wells of plates were coated with the capture antibody ( anti-mouse IFN-γ , IL-4 , IL-6 , and IL-10; BD Biosciences , USA ) at 1: 200 dilutions in PBS and incubated overnight at 4°C . The plates were washed once with RPMI-1640 medium with 10% fetal bovine serum and blocked for 2 h at room temperature . A total of 1 × 106 lymphocytes were added to each well in a total volume of 100 μL . The lymphocyte cells were stimulated with rTsPmy at a final concentration of 1 μg/mL at 37°C for 48 h in a 5% CO2 incubator . A total 100 μL of biotinylated detection antibody at 1: 200 in PBS containing 10% FBS was added into each well for 2 h . The wells were incubated with 100 μL of streptavidin-HRP for 1 h ( BD Biosciences , USA ) and developed with 100 μL of 3-amino-9-ethylcarbazole substrate solution for 30 s–5 min according to the manufacturer’s instruction . The spot-forming units ( SFU ) were counted automatically by a CTL ELISPOT reader and analyzed using ImmunoSpot image analyzer software v4 . 0 . To determine the expression of homing receptors ( CCR9 and CCR10 ) on B cells induced by recombinant Salmonella inoculation , the mononuclear cells were isolated from spleens ( SP ) , MLNs and intestinal lamina propria ( LP ) of mice two weeks after the third immunization as described [19–21] . ASCs were then enriched from these mononuclear cell suspensions using a magnetic bead B cell negative isolation kit ( Invitrogen , USA ) . The enriched ASCs were blocked with rat anti-mouse CD16/CD32 mAb for 15 min ( 4°C ) and incubated with anti-mouse CD19-FITC ( BD Biosciences , San Diego , California ) , CD199 ( CCR9 ) -PE ( BD Biosciences , USA ) , or CCR10-PerCP ( R&D Systems , USA ) mAbs or their isotype controls for surface marker staining for 30 min . After washing twice , cells were resuspended in 300 μL of 1% para-formaldehyde in PBS and analyzed by FACS ( BD Biosciences , USA ) to sort CCR9 and CCR10 expressed ASCs . In order to determine if CCR9 expressing ASCs migrate toward chemokine CCL25 , a total of 1×105 ASCs were added to the upper chamber of a Transwell filter with 5 mm polycarbonate membrane ( Costar Corning , USA ) and allowed to migrate for 4 h at 37°C into the lower chamber containing RPMI-1640 supplemented with 25 ug/ml of CCL25 ( Prospec , Israel ) . The cells that migrated into the lower chamber were counted and collected for detecting the expression of anti-TsPmy IgA using a modified ELISPOT assay . Briefly , wells coated with 1 μg/mL of rTsPmy were incubated with 1×105 ASCs collected from the lower chamber . The cells expressing specific IgA against TsPmy was identified using biotin-conjugated anti-mouse IgA and Streptavidin-HRP . To evaluate the protective immunity , the left 20 mice of each group were each challenged with 500 T . spiralis muscle larvae two weeks after the third immunization . Adult worms were recovered from the intestines of 10 mice on day 5 post-infection , and the muscle larvae were recovered from muscle of ten mice 45 days after the challenge . The reduction evaluation in adult worm and muscle larvae was calculated based on the number of adult worms or muscle larvae collected from the group immunized with SL7207/pVAX1-TsPmy compared with those from the SL7207/pVAX1 control mice . Statistical analyses were performed with one-way ANOVA using SPSS version 17 . 0 software . All data were expressed as the mean ± standard deviation , with differences considered significant when P was less than 0 . 05 .
Total RNAs were isolated from the MLNs , liver and spleen tissues of mice one week after the first immunization for RT-PCR to determine the transcription of TsPmy in these tissues . The results showed that the TsPmy mRNA was transcribed in the tissues of mice immunized with SL7207/pVAX1-TsPmy but not in the mice received with SL7207/pVAX1 only ( Fig 1A ) . Immunofluorescent staining with anti-TsPmy mAb 9G3 revealed that the TsPmy protein was expressed in MLNs of SL7207/pVAX1-TsPmy immunized mice ( Fig 1B ) . No obvious fluorescence was observed in MLNs of mice treated with SL7207/pVAX1 ( Fig 1C ) . Serum samples of mice were collected one week after each immunization and the levels of specific anti-TsPmy IgG and its subclass ( IgG1 and IgG2a ) antibodies were measured by ELISA . A high titer of anti-TsPmy IgG was elicited following a boost with SL7207/pVAX1-TsPmy and reached their peak titer at one week after the third immunization . Nevertheless , none of the mice that received SL7207/pVAX1 or PBS orally showed detectable anti-TsPmy IgG responses ( Fig 2A ) . The levels of anti-TsPmy IgG subclass IgG1 and IgG2a were also increased significantly in mice immunized with SL7207/pVAX1-TsPmy after the first boost and reached a peak after the second boost . The IgG2a level was significantly higher than IgG1 after the first boost , indicating that attenuated Salmonella delivered TsPmy DNA vaccine induced Th1/Th2-mixed type of immune response with Th1 being predominant ( Fig 2B ) . To evaluate the intestinal mucosal sIgA response upon oral SL7207/pVAX1-TsPmy immunization , the total sIgA and anti-TsPmy specific sIgA were measured in intestinal mucosal washings by ELISA . Total sIgA level was significantly ( p< 0 . 05 ) elevated in the intestinal mucosa of mice immunized with SL7207/pVAX1-TsPmy compared with those administered SL7207/pVAX1 or PBS ( Fig 3A ) . The anti-TsPmy sIgA level was measured using a rTsPmy-coated plate . Anti-TsPmy specific sIgA was also significantly increased in the intestinal mucosa of mice immunized with SL7207/pVAX1-TsPmy compared with those treated with vector alone or PBS ( Fig 3B ) . To evaluate the cytokine profiles induced by SL7207/pVAX1-TsPmy immunization , 5 mice were sacrificed at 1 week after the third immunization . Spleen cells and MLN cells were collected and stimulated with 1 ug/ml of rTsPmy . Cytokines secreted by the lymphocytes including IFN-γ , IL-2 , IL-4 , IL-5 , IL-6 and IL-10 were detected by ELISPOT assay . Compared with the SL7207/pVAX1 and PBS control groups , significantly increased levels of secretion of IFN-γ , IL-2 , IL-4 , IL-5 , IL-6 and IL-10 were observed in both TsPmy-stimulated spleen ( Fig 4A ) and MLNs ( Fig 4B ) cells after vaccination , indicating that the Th1/Th2-mixed immune responses were significantly induced by the oral immunization of SL7207/pVAX1-TsPmy . Moreover , it also suggests that the immune response upon SL7207/pVAX1-TsPmy immunization occurred systemically ( spleen ) and locally in lymphocytes around the intestine ( MLNs ) . To assess whether oral immunization of SL7207/pVAX1-TsPmy induces B lymphocytes into antibody secreting cells ( ASCs ) , the total ASCs were isolated from the spleen , MLNs and intestinal lamina propria ( LP ) of mice immunized with SL7207/pVAX1-TsPmy three times . Both intestinal homing receptors CCR9 and CCR10 were highly expressed on ASCs from the spleen , MLNs and LP of mice immunized with both SL7207/pVAX1-TsPmy and SL7207/pVAX1 groups but not in the PBS group . However , the expression level of CCR9 and CCR10 was higher on LP than on MLNs and even less on spleen cells ( Fig 5A and 5B ) . A similar phenomenon was also observed in the chemotaxis assay upon stimulation with CCL25 , the chemokine ligand of CCR9 . ASC cells isolated from the LP and MLNs of mice immunized with SL7207/pVAX1-TsPmy and SL7207/pVAX1 moved significantly more toward CCL25 than ASCs isolated from PBS control mice . The chemotaxis toward CCL25 was not significant in ASCs isolated from spleen immunized with SL7207/pVAX1-TsPmy and SL7207/pVAX1 compared to mice treated with PBS even though the levels of the former groups were higher than PBS ( Fig 5C ) . Although we observed a higher expression of CCR9 and CCR10 on ASCs from mice immunized with a SL7207/pVAX1 empty control , similar to the level in mice immunized with SL7207/pVAX1-TsPmy , the antigen ( TsPmy ) specific IgA was only expressed in ASCs isolated from LP and MLNs of mice immunized with SL7207/pVAX1-TsPmy . No specific IgA was expressed in mice immunized with SL7207/pVAX1 . The expression level of anti-TsPmy was higher in ASCs from LP than in MLNs without significant expression in the spleen ( Fig 5D ) The protective immunity was tested in immunized mice against T . spiralis larval challenge . The challenge results demonstrated that mice orally immunized with SL7207/pVAX1-TsPmy produced 44 . 8% reduction in muscle larvae burden ( Fig 6A ) and 46 . 6% reduction in adult worm burden ( Fig 6B ) after challenge with 500 T . spiralis infective larvae compared with the PBS control . The mice treated with SL7207/pVAX1 did not show any significant worm reduction compared to PBS control . The number of adult worms and ML collected from each group of mice was shown in S1 Table . This result demonstrated that oral immunization with TsPmy DNA vaccine delivered by attenuated Salmonella induced the partial protection against challenge infection with T . spiralis larvae .
Paramyosin is a thick myofibrillar protein found only in invertebrates [22] . TsPmy is the paramyosin expressed by T . spiralis that is not only a structural component of myofilament but also an immunomodulatory protein present on the surface of newborn larva and adult worm . TsPmy enables binding to C8/C9 [23 , 24] and C1q [25] of the human complement , inhibiting classical complement activation and the formation of complement membrane attack complex ( MAC ) , thereby protecting the parasite from attack by host activated complement . Immunization with rTsPmy protein[13 , 23] , immunogenic peptides [26 , 27] , or passive transfer of monoclonal antibody that specifically binds to the TsPmy C9 binding site[16] induced significant protection against T . spiralis larval challenge . Paramyosin in other helminths such as Schistosoma mansoni[28] , Brugia malayi[29] and Taenia solium[30] also showed protective immunity against parasite infections in different animal models . TsPmy is a large protein with 885 amino acids and a 102 kDa predicted molecular weight , which is difficult to express as a soluble recombinant protein [15] . The low expression yield and solubility hurdles its development as a recombinant protein vaccine for large scale production . DNA vaccine is considered an alternative , even optimal approach because of the simplicity of its manufacturing and distribution , biological stability and cost effectiveness [31] . DNA vaccines have been shown to induce protective immunity against abroad range of pathogens such as Dengue virus [32] , intracellular protozoan Leishmania major [33] or helminth parasite Schistosoma japonicum[34] . DNA vaccine contains a eukaryotic expression vector that could improve protein folding , therefore enable surface-exposed epitopes to be correctly presented and enable post-translational modification [35] . Nevertheless , naked DNA vaccine has inefficient immunogenicity compared to protein vaccines , which need an appropriate delivery system to enhance the immune response . The attenuated Salmonella strain has been identified as a suitable vaccine vector that could deliver heterogeneous antigens to the gastrointestinal mucosa and other lymphoid tissues and enhance specific humoral , cell-mediated , and mucosal immune responses [36] . In our previous study , vaccination with Ts87 antigen delivered by attenuated Salmonella led to partial adult worm and muscle larva reduction following challenge with T . spiralis larvae [10] . In this study , we cloned TsPmy coding DNA into the eukaryotic expression vector pVAX1 , and the recombinant pVAX1 containing TsPmy coding DNA was transformed into attenuated S . typhimurium SL7207 strain to form a bacteria-delivered TsPmy DNA vaccine . Mice orally vaccinated with attenuated Salmonella-delivered TsPmy DNA elicited significant protection against T . spiralis larval challenge in a murine model . The immunized mice showed a 44 . 8% reduction in adult worm burden in the intestines and a 46 . 6% reduction in larval burden in muscle tissues , which is the best protection we have ever obtained so far compared to vaccination with recombinant protein [10 , 13] or peptide [26 , 27] of TsPmy in our lab using the same mouse model ( S2 Table ) . Live attenuated Salmonella is an effective vector that can bring the target DNA to internal organs and lymph tissues efficiently . Only one week after oral inoculation of Salmonella carrying TsPmy DNA , TsPmy RNA transcription and protein expression were observed in the MLNs , liver and spleen . The strong deliverability and adjuvanticity of attenuated bacteria induced strong immune responses , including humoral and cellular immunity against targeting pathogen [37] . Indeed , mice orally vaccinated with attenuated Salmonella-delivered TsPmy DNA vaccine induced a high serum titer of anti-TsPmy IgG . A higher level of an IgG2a subclass than IgG1 detected in the sera of mice after the 2nd immunization suggested that the vaccination induced a relatively mixed Th1/Th2 response , which was further confirmed by the cytokine profiles of splenocytes and MLNs that showed significant increases in both Th1 ( IFN-γ , IL-2 ) and Th2 cytokines ( IL-4 , IL-5 , IL-6 and IL-10 ) upon stimulation with rTsPmy . More importantly , live attenuated Salmonella-delivered TsPmy DNA vaccine was able to induce considerable secretion of antigen-specific sIgA in the intestinal mucosa , which plays a crucial role in the mucosal immunity against intestinal pathogens [38] . In addition to the strong Th1/Th2 mixed systemic immune responses , the sIgA-associated mucosa immunity induced by live bacteria-delivered DNA vaccine may contribute to the better protection against T . spiralis larval challenge in this study . Many lines of evidence support the notion that intestinal sIgA are secreted largely by plasma cells derived from B cells initially activated in gut-associated lymphoid tissue ( GALT ) after vaccination . After immunization , most naïve B cells migrate to the Peyer’s patches or MLNs and differentiate into plasma cells and return to the intestinal mucosa ( Lamina Propria ) to produce high-affinity sIgA [39 , 40] . As we know , the epithelial cells of the small intestine express chemokine CCL25 and CCL28 that play important roles in mucosal immunity by recruiting IgA antibody-secreting cells ( ASCs ) that express their CCR9 and CCR10 receptors in the mucosal lamina propria[41] . In this study , we identified that the Salmonella-delivered TsPmy DNA vaccine immunized mice expressed CCR9 and CCR10 , the receptor of chemokine CCL25 and CCL28 , respectively , on the surface of ASCs isolated from the spleen , MLNs and LP . Indeed , the CCR9-expressing ASCs were able to migrate towards CCL25 by chemotaxis . Interestingly , more CCR9 and CCR10 expressing and antigen-specific IgA-expressing ASCs were observed in LP than in MLNs with the least in the spleen , indicating that more ASCs , especially TsPmy-specific IgA expressing ASCs , migrate towards intestinal lymphatic tissues being attracted by intestinal cells expressing CCT25/CCL28 . Although the Salmonella bacteria themselves also stimulate the ASCs expressing CCR9 and CCR10 in the mice administered Salmonella/pVAX1 empty vector only reflected by the total IgA stimulation in this study ( Fig 3A ) , only TsPmy-specific IgA secreted APCs existed in LP and MLNs . In conclusion , our results demonstrated that oral immunization with attenuated Salmonella-delivered TsPmy DNA vaccine induced a mixed Th1/Th2 systemic immune response and a strong mucosal IgA response that protected mice from infection with T . spiralis with a 44 . 8% reduction in adult worm and a 46 . 6% reduction in muscle larva compared with the PBS control group . The expression of the homing receptors CCR9/CCR10 on antibody secreting B cells may be related to recruiting IgA-secreted B cells to local intestinal mucosa . Whether the Salmonella-delivered TsPmy DNA vaccine induced mucosal sIgA-mediated worm killing is under investigation . The attenuated Salmonella-delivered TsPmy DNA vaccine provides a feasible and promising approach for controlling trichinellosis in human and domestic animals . | Trichinellosis is one of the most important food-borne parasitic zoonoses , and a serious public health issue worldwide . Developing a vaccine is an alternative approach to control the disease . TsPmy is a paramyosin expressed by Trichinella spiralis to bind and neutralize human complement and a vaccine antigen . We made a DNA vaccine of TsPmy orally delivered by attenuated Salmonella typhimurium that elicited a robust Th1/Th2 and mucosa IgA responses , and protected mice against T . spiralis infection with significant worm reduction against larval challenge . The attenuated Salmonella-delivered TsPmy DNA vaccine provides a feasible and promising approach for controlling trichinellosis in human and domestic animals . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"enzyme-linked",
"immunoassays",
"spleen",
"immunology",
"animal",
"models",
"preventive",
"medicine",
"developmental",
"biology",
"model",
"organisms",
"vaccination",
"and",
"immunization",
"immunologic",
"... | 2016 | Oral Vaccination with Attenuated Salmonella typhimurium-Delivered TsPmy DNA Vaccine Elicits Protective Immunity against Trichinella spiralis in BALB/c Mice |
Signaling of the cytokine interleukin-6 ( IL-6 ) via its soluble IL-6 receptor ( sIL-6R ) is responsible for the proinflammatory properties of IL-6 and constitutes an attractive therapeutic target , but how the sIL-6R is generated in vivo remains largely unclear . Here , we use liquid chromatography–mass spectrometry to identify an sIL-6R form in human serum that originates from proteolytic cleavage , map its cleavage site between Pro-355 and Val-356 , and determine the occupancy of all O- and N-glycosylation sites of the human sIL-6R . The metalloprotease a disintegrin and metalloproteinase 17 ( ADAM17 ) uses this cleavage site in vitro , and mutation of Val-356 is sufficient to completely abrogate IL-6R proteolysis . N- and O-glycosylation were dispensable for signaling of the IL-6R , but proteolysis was orchestrated by an N- and O-glycosylated sequon near the cleavage site and an N-glycan exosite in domain D1 . Proteolysis of an IL-6R completely devoid of glycans is significantly impaired . Thus , glycosylation is an important regulator for sIL-6R generation .
Interleukin-6 ( IL-6 ) is a pleiotropic cytokine with important functions in many physiological and pathophysiological conditions [1 , 2] . IL-6 activates intracellular signaling cascades through a homodimer of the ubiquitously expressed β-receptor glycoprotein 130 ( gp130 ) [3] but first has to bind to its nonsignaling alpha receptor ( IL-6R ) . The IL-6R is expressed in a cell- and tissue-specific manner and only found on hepatocytes and some leukocytes like neutrophils and T cells [3–5] . Signaling via the membrane-bound IL-6R is mostly regenerative and anti-inflammatory [5 , 6] . The IL-6R is a typical type-I transmembrane protein that consists of an immunoglobulin ( Ig ) -like domain ( “D1” ) , the cytokine-binding module ( CBM ) residing in two fibronectin-type-III domains ( “D2” and “D3” ) , and a 55 amino-acid residue-long flexible stalk region that is followed by a transmembrane and an intracellular region . A minimal length of 22 amino acids of the stalk region is required for efficient IL-6 classic signaling , which corresponds to a stalk length of approximately 83 Å [7] . N-linked glycosylation is an important post-translational modification that can ensure correct folding and stability of a protein [8] . It has been shown that N-linked glycosylation of gp130 is essential for its stability but dispensable for the signaling function [9] . However , other signaling receptors require glycans for the ability to bind their respective ligands , e . g . , the receptors for epidermal growth factor ( EGF ) [10] , granulocyte-macrophage colony-stimulating factor ( GM-CSF ) [11] , or C-X-C motif chemokine 12 ( CXCL12 ) [12] . Although some asparagine residues within the extracellular part of the IL-6R have been identified that are used for N-linked glycosylation in vitro [13 , 14] , no functional role for these modifications has been determined . Addition of carbohydrates to serine or threonine residues , called O-linked glycosylation , has not been described so far for the IL-6R . Soluble forms of the IL-6R ( sIL-6R ) are found in human serum at concentrations of about 30–70 ng/ml . IL-6 binds to membrane-bound and sIL-6R with similar affinity , and signaling of IL-6/sIL-6R has been termed IL-6 trans-signaling and is causative for the proinflammatory properties of IL-6 . Specific inhibition of IL-6 trans-signaling holds the promise to be as effective as total IL-6 blockade , but with reduced side effects like enhanced susceptibility to bacterial infections [15] . One mechanism that contributes to sIL-6R generation is alternative splicing of the IL6R mRNA , which results in a unique C-terminus of ten amino-acid residues [16] , and only this form of the sIL-6R has been detected in human serum [17] . However , only 1%–20% of the total sIL-6R is generated by alternative splicing [18–21] , and the generation mechanism of the majority remains unknown . Recently , we described expression of the IL-6R on circulating microvesicles , but this also accounted only for a minor proportion of the total sIL-6R [22] . In vitro , sIL-6R can be efficiently generated by limited proteolysis , predominantly by the metalloproteases ADAM10 and ADAM17 . ADAM17-mediated shedding can be induced by a variety of stimuli , and the phorbol ester phorbol-12-myristate-13-acetate ( PMA ) is the strongest known activator in vitro . Carboxypeptidase treatment of sIL-6R , which was purified from the supernatant of PMA-treated IL-6R overexpressing COS-7 cells , led to the identification of the ADAM17 cleavage site within the IL-6R between Gln-357 and Asp-358 [14] . However , cleavage site profiling revealed that glutamine and aspartic acid are rather disfavored at the P1 and P1′ positions [23] , and the IL-6R is the only ADAM17 substrate with such a cleavage site that is listed in the MEROPS database [24] . Indeed , cleavage of an IL-6R peptide with recombinant ADAM17 occurred between Pro-355 and Val-356 [25] . Nevertheless , deletion of amino-acid residues Ser-353 to Val-362 within the IL-6R stalk prevented ADAM17-mediated shedding and did not compromise its biological activity [7 , 14] . Proteolysis by ADAM10 , the closest homologue of ADAM17 , still occurred , which suggested the usage of multiple cleavage sites by ADAM10 or different cleavage sites by the two proteases [7] . Besides the cleavage site , which is considered to be the major determinant of substrate/protease specificity , exosites are also known to contribute to the regulation of IL-6R proteolysis [26–28] . In this study , we identify by liquid chromatography–mass spectrometry ( LC-MS ) an sIL-6R form in human serum that is generated via limited proteolysis , map its cleavage site , and characterize the occupancy of all O- and N-glycosylation sites . We confirm that the same cleavage site is used by ADAM10 and ADAM17 in vitro and show that mutation of Val-356 at P1′ is sufficient to completely block proteolysis . We further show that glycosylation is dispensable for IL-6R trafficking and signaling but regulates proteolysis through an N-glycan exosite in domain D1 and a sequon that is both N- and O-glycosylated adjacent to the cleavage site .
Although the existence of sIL-6R in human body fluids has been known for more than 25 y [29 , 30] , the mechanisms of its generation are largely unexplored . Müller-Newen et al . were able to purify an sIL-6R from human plasma that was generated by alternative splicing of the IL6R mRNA ( termed ds-sIL-6R hereafter ) [17] . However , several studies using ELISA revealed that only 1%–20% of the total sIL-6R is generated by splicing [18–21] , which suggests that other mechanisms must exist in parallel . We sought to confirm these findings and generated a polyclonal antibody ( termed ds6R ) against the ten unique C-terminal amino-acid residues of the ds-sIL-6R , which are not present in the full-length IL-6R or a proteolytically cleaved sIL-6R ( Fig 1A ) . In order to be able to quantify total sIL-6R and to distinguish only ds-sIL-6R in human serum , we first expressed and purified both sIL-6R and ds-sIL-6R as recombinant proteins . As shown in Fig 1B and 1C , both sIL-6R forms were biologically active and equally well able to perform IL-6 trans-signaling , because they stimulated proliferation of Ba/F3-gp130 cells in a dose-dependent manner when combined with IL-6 . A sandwich ELISA with a capture antibody that binds to the sIL-6R N-terminus ( 4–11 ) and a polyclonal detection antibody ( Baf227 , [21 , 31] ) measured recombinant sIL-6R and ds-sIL-6R with equal efficiency ( Fig 1D ) and allowed us to measure total sIL-6R levels in the serum of eight healthy donors , which were in a similar range as described before ( 46 . 0 ± 6 . 4 ng/ml , Fig 1E , [21] ) . Using the ds-sIL-6R-specific antibody ds6R , we set up an ELISA that only detects ds-sIL-6R and not other sIL-6R forms ( Fig 1F , [16] ) . When we analyzed the same eight serum samples for ds-sIL-6R , we detected only 7 . 0 ± 2 . 0 ng/ml , which accounts for 15 . 1% ± 2 . 8% of the total sIL-6R ( Fig 1G ) . Thus , our results confirm previous studies that found that a fraction of the sIL-6R in humans is generated via alternative splicing but that the majority must originate from a different mechanism . In order to identify other human sIL-6R forms that exist in vivo , we performed an isotopic labeling strategy that is based on proteolysis in the presence of H218O [32] . We isolated human serum , depleted the endogenous antibodies , and precipitated all sIL-6R forms by the sepharose-coupled 4–11 antibody , which binds to the N-terminal D1 domain ( Fig 2A ) . We separated the precipitated proteins via SDS-PAGE under nonreducing conditions , confirmed sIL-6R presence in the precipitate via western blot , and excised the corresponding region of the coomassie-stained gel ( Fig 2A ) . Using an anti-IL-6R antibody , we detected several proteins of different molecular weights , which most likely are different sIL-6R isoforms and complexes of these with other serum proteins that are not dissociated under nonreducing conditions ( Fig 2A ) . Prior to in-gel proteolysis , the proteins were deglycosylated using peptide:N-glycosidase F ( PNGase F ) . This endoglycosidase hydrolyzes the bond formed between the carbohydrate and the Asn side chain , resulting in the formation of an Asp residue at the former glycosylation site . Subsequently , protein digestion was performed in 50% H218O-containing buffers . By using this method , all proteolytically generated neo-C-termini contain an 18O-isotope incorporated in the carboxyl group , while the original ( canonical or truncated ) protein C-terminus remains unmodified ( Fig 2B ) . We identified a peptide that maps to the unique C-terminus of the ds-sIL-6R ( Fig 2C ) , which further corroborates that ds-sIL-6R contributes to the total amount of sIL-6R in human serum . Furthermore , we identified another C-terminal peptide , which ends with Pro-355 and thus could belong to an sIL-6R that is generated through proteolysis of the membrane-bound IL-6R between Pro-355 and Val-356 ( Fig 2D ) . Both ds-sIL-6R and sIL-6R had an N-linked glycosylation at Asn-350 , which was converted to Asp-350 during the N-deglycosylation with PNGase F ( Fig 2C and 2D , colored in green ) . Furthermore , using electron transfer dissociation ( ETD ) MS/MS , we identified an O-glycosylation site within the Asn-350 sequon on Thr-352 , which was decorated with several different glycan structures ( Fig 2D and S1A–S1E Fig ) . With this MS-based strategy , we were for the first time able to identify an sIL-6R variant in human serum that does not originate from alternative splicing but appears to be generated by proteolysis . Having shown that a protease-derived sIL-6R exists in vivo , we sought to identify the protease that could be responsible for its generation . The IL-6R is a known substrate for three human proteases , the two metalloproteases ADAM10 and ADAM17 [7 , 27 , 33] , and the neutrophil-derived serine protease cathepsin G ( CG ) [34 , 35] . So far , the only protease of the IL-6R whose cleavage site has been mapped is ADAM17 , which has been reported to cleave between Gln-357/Asp-358 , two amino-acid residues downstream of Pro-355/Val-356 [14] . In order to recapitulate this finding , we transiently transfected HEK293 cells with a cDNA encoding human IL-6R and induced ADAM17-mediated shedding with PMA , a strong and well-known activator of ADAM17 . We precipitated sIL-6R from the cell culture supernatant and performed SDS-PAGE and western blot ( Fig 3A ) . The strongest band detected in the western blot corresponded to a cleavage product of the IL-6R , whereas the lower band at around 55 kDa most likely was the heavy chain of the antibody used for precipitation ( Fig 3A ) . Indeed , MS analysis revealed a C-terminal peptide that ended with Pro-355 ( Fig 3B ) , indicating exactly the same cleavage site that was used in vivo ( Fig 2D ) . We could further confirm the N-glycan site on Asn-350 and the O-glycan site at Thr-352 within the same sequon , which was again decorated with the same pattern of glycan structures found in vivo ( Fig 3B and S2A–S2E Fig ) . Thus , we could not verify the originally published cleavage site Gln-357/Asp-358 [14] but instead found an ADAM17 cleavage site between Pro-355/Val-356 that matches the published cleavage preferences of ADAM17 [23] and is in good agreement with other cleavage sites of ADAM17 substrates in the MEROPS database [24] . To identify the cleavage site for ADAM10 , we performed a similar experiment and stimulated transiently transfected HEK293 cells with ionomycin , a calcium ionophore that induces IL-6R cleavage by ADAM10 ( Fig 3C ) [7 , 21 , 27 , 36] . Similar to ADAM17-mediated shedding ( Fig 3A ) , we detected cleaved IL-6R and a band of lower molecular weight , which most likely corresponded to the heavy chain of the antibody used for precipitation ( Fig 3C ) . Surprisingly , we identified the same cleavage site between Pro-355 and Val-356 ( Fig 3D ) and the same pattern of O-glycans at Thr-352 ( S3A–S3D Fig ) , indicating that both proteases share the same cleavage site . We did not determine the cleavage site used by CG , as the resulting sIL-6R was significantly smaller compared to the sIL-6R cleaved by ADAM17 ( S3E Fig ) , indicating that CG cleaved further upstream within the IL-6R stalk and thus cannot be the protease responsible for the generation of the steady-state sIL-6R serum levels . To further characterize the identified cleavage site , we created IL-6R variants with different point mutations at the P1 and the P1′ sites ( Fig 4A and 4B ) . In general , we exchanged Pro-355 and Val-356 into amino-acid residues that have been shown to be disfavored by ADAM17 [23] . When we mutated the cleavage site to Ile-355/Glu-356 ( termed IL-6R_IE hereafter; all other mutants are termed accordingly ) , we could not detect inducible formation of sIL-6R by western blot ( Fig 4C ) or ELISA ( Fig 4D ) after stimulation with PMA . The same was true for the IL-6R_DG mutant . When we calculated the increase in sIL-6R generation , the wild-type IL-6R was shed 3 . 6 ± 1 . 9-fold compared to vehicle-treated cells , whereas both mutants showed no increase in sIL-6R formation ( 1 . 3 ± 0 . 2-fold and 1 . 4 ± 0 . 2-fold , respectively , Fig 4E ) . We obtained similar results when we activated ADAM10 with ionomycin ( S4A–S4C Fig ) . In order to determine if mutation of either P1 or P1′ alone would be sufficient to block ADAM-mediated proteolysis , we created the corresponding four single mutants ( Fig 4B ) . Whereas IL-6R_DV and IL-6R_IV were still shed after PMA treatment , albeit to a minor extent compared to the wildtype ( Fig 4F and 4G ) , mutation of the P1′ site ( IL6R_PE and IL-6R_PG ) completely abrogated proteolysis by ADAM17 ( Fig 4F and 4G ) , and we could not detect any increase in sIL-6R formation ( 0 . 9 ± 0 . 1-fold and 1 . 0 ± 0 . 1-fold , Fig 4H ) . Similar results were again obtained for ADAM10-mediated cleavage ( S4D–S4F Fig ) . In conclusion , our results show that mutation of the P1′ site ( Val-356 ) is sufficient to completely block ADAM-mediated cleavage of the IL-6R . The coding single nucleotide polymorphism ( SNP ) Asp358Ala within the IL-6R is located in close proximity to the cleavage site , and homozygous carriers of this SNP have increased sIL-6R serum levels [37 , 38] . We have shown previously that the Asp358Ala ( termed IL-6R_PVQA in this study ) mutation renders the IL-6R more susceptible to ADAM-mediated cleavage [21] . In order to investigate whether mutation of the cleavage site would be sufficient to block this increased proteolysis , we created three additional IL-6R mutants that contained Asp358Ala combined with single or double mutations of the cleavage site ( termed IL-6R_DGQA , IL-6R_DVQA , and IL-6R_PGQA; Fig 4I ) . First , we verified our previous finding [21] that IL-6R_PVQA was indeed shed more than the wild-type IL-6R_PVQD ( Fig 4J ) . Mutation of both P1 and P1′ in IL-6R_DGQA completely abrogated PMA-inducible shedding , and mutation of P1′ alone , but not P1 alone , was sufficient to render the Asp358Ala mutant unresponsive towards proteolysis by ADAM17 ( Fig 4J ) . Proteolysis by ADAM10 was also significantly reduced , although some inducible shedding by ionomycin could still be detected ( S4G Fig ) . Thus , mutation of Val-356 is also sufficient to prevent the enhanced shedding of the Asp358Ala mutant . Finally , we analyzed whether these mutations influenced the biological activity of the IL-6R . To this end , we stably transduced Ba/F3-gp130 cells with wild-type IL-6R and IL-6R mutants . After transduction , Ba/F3-gp130-hIL-6R cells proliferated in a dose-dependent manner ( 0–100 ng/ml ) in response to IL-6 ( Fig 4K ) . As an internal control , we incubated all cell lines with Hyper-IL-6 , which is a fusion protein of IL-6 and the sIL-6R and thus activates proliferation via gp130 homodimerization independent of membrane-bound IL-6R . Ba/F3-gp130-IL-6R_IE , Ba/F3-gp130-IL-6R_PE , and Ba/F3-gp130-IL-6R_DG proliferated equally well compared to the wild-type cells , indicating that mutation of the cleavage site did not affect the biological function of the IL-6R ( Fig 4L and 4M and S4H Fig ) . Although N-linked glycosylation of the IL-6R has been reported [13 , 14] , no functional role has been addressed so far . We first confirmed the presence of glycans on the full-length IL-6R and the sIL-6R by either removing just the N-linked glycans with PNGase F or by additionally also removing O-linked glycans ( Fig 5A and 5B ) . While determining the cleavage site , we had already successfully detected an N-glycosylation site at Asn-350 and an O-glycosylation site at Thr-352 , both in vitro and in vivo ( Figs 2D and 3B ) . By performing N-deglycosylation in the presence of H218O-containing buffers , we identified further N-glycosylation sites at Asn-55 , Asn-93 , Asn-221 , and Asn-245 on sIL-6R derived from PMA-stimulated HEK293 cells ( Fig 5C–5F ) . With the exception of Asn-245 , we verified all N-glycosylation sites also on sIL-6R isolated from human serum ( S5A–S5C Fig ) . As we could not detect any other O-glycosylation site in addition to Thr-352 , the IL-6R contains two N-glycans located in domain D1 , two N-glycans in domain D3 , and a combined N-/O-glycan site within the stalk region adjacent to the cleavage site ( Fig 5G ) . In order to analyze a functional role for the determined glycans , we created IL-6R mutants that lacked either one , multiple , or all N-glycosylation sites and stably transduced Ba/F3-gp130 cells with these . Flow cytometry analysis revealed that all IL-6R mutants were transported to and expressed at the cell surface , albeit the amount slightly decreased when more N-glycans were deleted ( S6A Fig ) . When we analyzed IL-6-dependent proliferation of the individual cell lines , we could not detect a difference between the wild-type IL-6R ( Fig 4K ) , IL-6R mutants lacking one of the five N-glycosylation sites ( Fig 6A–6E ) , IL-6R mutants lacking two ( termed 2N ) , three ( 3N ) , or four N-glycosylation sites ( 4N , S6B–S6D Fig ) , or the IL-6R mutant that lacked all five N-glycans ( 5N , Fig 6F ) . Similarly , mutation of the O-glycosylation site on Thr-352 had no influence on IL-6 signaling ( Fig 6G ) . Finally , we analyzed protein stability and turnover at the cell surface of the wild-type IL-6R compared to the 4N mutant . For this purpose , we used a pulse-chase approach and labeled the IL-6R on the cells in the cold , shifted the cells back to 37°C for different time periods , and analyzed the remaining IL-6R on the cell surface via flow cytometry . As shown in Fig 6H , the wild-type IL-6R disappeared from the cell surface in a time-dependent manner ( t1/2 = 44 . 1 ± 8 . 1 min ) , which did not differ significantly from the 4N mutant ( t1/2 = 40 . 9 ± 5 . 2 min ) . Collectively , our data show that glycosylation of the IL-6R is not needed for transport to and expression at the cell surface , is dispensable for IL-6-mediated signal transduction , and does not alter the half-life of the IL-6R at the cell surface . Having excluded that glycosylation is important for the signaling capacity of the IL-6R , we sought to analyze a possible role for the glycans in terms of IL-6R proteolysis . We first concentrated on the N-/O-glycan pair on Asn-350/Thr-352 adjacent to the ADAM10/17 cleavage site and employed an in vitro cleavage assay . Previous attempts to cleave an IL-6R peptide with recombinant ADAM17 gave mixed results [25 , 39] . We synthesized ten different glycopeptides containing either the wild-type sequence ( QD ) or the Asp358Ala mutation ( QA ) and incubated them with recombinant ADAM17 . Whereas the unglycosylated peptide 1 ( QD ) was little but still significantly cleaved ( p < 0 . 05 , Fig 7A ) , the corresponding peptide 2 ( QA ) was cleaved completely ( 1 . 0% ± 1 . 0% remaining , p < 0 . 001 , Fig 7A ) . Addition of an N-Acetylglucosamine ( GlcNAc ) on the residue corresponding to Asn-350 reduced proteolysis of peptide 3 ( QD ) ( 91 . 4% ± 9 . 4% intact ) but did not alter processing of peptide 4 ( QA ) ( 0 . 4% ± 0 . 7% , Fig 7B ) . Interestingly , addition of a larger biantennary N-glycan blocked cleavage of the 5 ( QD ) peptide but did not alter processing of the 6 ( QA ) peptide ( Fig 7C ) . Also of interest , addition of an N-Acetylgalactosamin ( GalNAc ) to the amino-acid residue corresponding to Thr-352 decreased cleavage of the 7 ( QD ) peptide ( 91% ± 8% ) but also decreased shedding of the 8 ( QA ) peptide compared to the unglycosylated peptide ( 10 . 8% ± 5 . 4% intact , p < 0 . 05 , Fig 7D ) . Importantly , we have detected GalNAc endogenously on the sIL-6R at Thr-352 , but not GlcNAc ( Fig 2D and S1A–S1D Fig ) . When we combined the biantennary N-glycan with the GalNAc , cleavage of 9 ( QD ) was again blocked , but 14 . 7% ± 6 . 3% of the 10 ( QA ) peptide was intact , suggesting an additive effect of the two glycans ( p < 0 . 05 , Fig 7E ) . Thus , in line with a recent report [25] , we can detect an inhibitory influence of the glycosylation on the capacity of ADAM17 to cleave the IL-6R peptide , but overall , the influence appears to be rather small . In order to analyze the role of the other N-glycosylation sites , we used the stably transduced Ba/F3-gp130 cell lines and activated ADAM17-mediated shedding with PMA . We set the amount of sIL-6R shed from wild-type cells to 100% ( Fig 8A ) and calculated all other values accordingly . Surprisingly , mutation of Asn-55 enhanced PMA-induced shedding to 348% ± 116% , indicating that this N-glycan serves as a protease-regulatory exosite ( Fig 8B ) . Also , unstimulated shedding was significantly increased ( 148% ± 25% , p < 0 . 05 ) , and the amount was even higher than the stimulated shedding of the wild-type IL-6R , despite equal expression of both receptors ( S6A Fig ) . In order to prove that indeed the glycan and not the asparagine residue is responsible for this effect , we mutated Thr-57 , which also leads to the loss of the glycan at Asn-55 , and observed a comparable increase in IL-6R shedding ( 291% ± 64% and 126% ± 9% , both p < 0 . 05 , Fig 8C ) . Mutation of Asn-93 or Asn-221 did not enhance sIL-6R generation ( Fig 8D and 8E ) . We observed slightly enhanced shedding when we mutated Asn-245 ( 215% ± 82% , Fig 8F ) . Combined mutation of Asn-93 and Asn-245 ( 2N ) did not enhance proteolysis ( Fig 8G ) , but the 3N and 4N variants , in which Asn-55 was mutated , showed increased IL-6R shedding by ADAM17 ( Fig 8H and 8I ) . In line with the peptide cleavage assays ( Fig 7 ) , mutation of the N-/O-glycosylated sequon adjacent to the cleavage site only slightly enhanced proteolysis ( Fig 8J and 8K ) . In conclusion , the N-glycan on Asn-55 within the D1 domain of the IL-6R functions as an exosite , which reduces the induced and constitutive proteolysis of the IL-6R . Finally , we analyzed proteolysis of the 5N variant , which lacks all N-glycans . Surprisingly , we detected significantly reduced PMA-induced shedding ( 27 . 2% ± 17 . 2% , Fig 9A ) , and although the 5N mutant showed less expression at the cell surface compared to the wild type ( S6A Fig ) , this alone cannot explain the reduced proteolysis . To verify this observation , we transiently transfected HEK293 cells side by side with cDNAs encoding wild-type IL-6R or the 5N mutant and analyzed shedding by ELISA and western blot . PMA strongly induced shedding of wild-type IL-6R ( Fig 9B and 9C ) , but we could not detect proteolysis of the 5N mutant in HEK293 cells either by ELISA ( Fig 9D ) or by western blotting ( Fig 9E ) , although the IL-6R-5N was transported to the cell surface ( Fig 9F ) . We have previously shown that the membrane-proximal domain ( MPD17 ) and especially the conserved ADAM seventeen dynamic interaction sequence ( CANDIS ) region of ADAM17 are important for substrate recognition [26 , 40] , and we thus wondered whether the IL-6R-5N variant is still able to bind to ADAM17 . As shown in Fig 9G , both IL-6R and IL-6R-5N could be efficiently coimmunoprecipitated with MPD17-CANDIS , indicating that the glycans are not mandatory for the IL-6R/ADAM17 interaction . Thus , although the IL-6R-5N mutant contained the unaltered cleavage site and is able to bind to ADAM17 , the protease is nevertheless not able to shed the IL-6R .
Soluble cytokine receptors play pivotal roles in health and disease . As long as the ligand-binding domains are retained within the soluble protein , they have similar affinities towards their ligands as the membrane-tethered counterparts . Most of them have antagonistic properties , because they compete with the membrane-bound receptors for the same cytokine , and cytokines bound to soluble receptors are no longer able to bind to their cognate target cells in order to activate them . The most prominent examples are the soluble receptors for tumor necrosis factor alpha ( TNFα ) [41] . An Fc-fusion protein of the extracellular part of tumor necrosis factor alpha receptor 2 ( TNFR2 ) /p75 is approved under the name etanercept for the treatment of rheumatoid arthritis and psoriasis [42] . A rare example of an agonistic soluble receptor is the sIL-6R . Surprisingly , despite its importance in terms of disease and therapy , little is known about the mechanisms that lead to sIL-6R generation in vivo . In the present study , we confirm the presence of the ds-sIL-6R in human serum and detect for the first time a second form of the sIL-6R with a novel C-terminus generated by proteolytic cleavage . Interestingly , the cleavage event occurs between Pro-355 and Val-356 within the stalk region of the IL-6R . This cleavage site does not match the one previously described by Müllberg et al . , who reported cleavage two amino-acid residues further downstream between Gln-357 and Asp-358 [14] . However , cleavage site profiling of ADAM17 as well as data in the MEROPS database confirms a strong preference for a valine at the P1′ site and disfavors an aspartic acid at that position [23 , 24] , making the reported cleavage by Müllberg et al . between a glutamine and an aspartic acid in retrospect rather unlikely . Moreover , we also consistently found ADAM-mediated cleavage at Pro-355/Val-356 when we treated HEK293 cells overexpressing the IL-6R with PMA or when we incubated a peptide of the IL-6R stalk with the recombinant catalytic domain of ADAM17 . We cannot absolutely exclude that exoprotease activity led to trimming of the C-terminal peptide in vivo or during sample preparation , which might confound the identification of the cleavage site via LC-MS . The different technique to determine the cleavage site used in Müllberg et al . [14] might also account for the difference . However , our mutational analysis of the novel cleavage site , which suggests that substitution of Val-356 by glycine or glutamic acid is sufficient to render the IL-6R resistant towards ADAM-mediated proteolysis , further corroborates this finding . The major genetic determinant of human sIL-6R serum levels is a single nucleotide polymorphism ( rs2228145 ) , which leads to the exchange of Asp-358 to an alanine residue [38] . Homozygous carriers have strongly increased sIL-6R serum levels , which reduces their risk of suffering from coronary heart disease [43 , 44] . Although differential splicing of the IL6R mRNA is increased in these individuals [45] , the majority of the sIL-6R is nevertheless generated by an alternative mechanism . We have shown previously that an IL-6R containing the Asp358Ala mutation is more efficiently shed by ADAM10 and ADAM17 , which we thought was caused by the exchange of an aspartic acid to an alanine residue at the P1′ position of the cleavage site , which makes the IL-6R a better substrate for both metalloproteases [21 , 23] . Because we have now determined the cleavage site between Pro-355 and Val-356 , this explanation for the observed effects of the Asp358Ala mutation has to be revised . However , the cleavage site profiling by Tucher et al . clearly shows a preference for an alanine over an aspartic acid residue also at the P3′ position , which explains the observed increase in proteolysis of the Asp358Ala mutant [23] . We have further used the LC-MS approach to determine all N- and O-glycans of the sIL-6R and investigated their functional role ( s ) in terms of signaling and proteolysis . Surprisingly , mutation of all glycan sites did not significantly alter the signaling capacity of the IL-6R , as shown by the IL-6-dependent proliferation of stably transduced Ba/F3-gp130 cells . This is in contrast to other cytokine receptors like epidermal growth factor receptor ( EGFR ) and granulocyte macrophage colony-stimulating factor receptor ( GM-CSFR ) [10 , 11] , in which N-linked glycosylation is essential for ligand binding . A mutant devoid of all glycans of the IL-6 β-receptor gp130 was still able to signal , but the transport and the stability of the protein were severely compromised [9] . However , the transport of the unglycosylated IL-6R mutant to the cell surface was only marginally affected , and the protein turnover compared to the wild-type IL-6R was not altered at all . Further experiments will show whether this is also true for the other two α-receptors of the IL-6 family , namely the IL-11R and the ciliary neurotrophic factor receptor ( CNTFR ) [3] . N- and O-linked glycosylation are common post-translational modifications that have been described for a variety of ADAM substrates besides the IL-6R , e . g . , for TNFα [46] or transforming growth factor alpha ( TGFα ) [47] . Substrates with and without glycans have also been used to generate novel ADAM17-specific inhibitors [48] . Recently , it was shown that O-glycosylation near the cleavage site affects ADAM17-mediated proteolysis of small peptides of several ADAM substrates , among them the IL-6R [25] . Here , we show that the IL-6R is indeed O-glycosylated on Thr-352 in vivo and confirm that this glycan reduces cleavage of an IL-6R peptide in conjunction with an N-linked glycan on Asn-350 . Cell-based assays , however , revealed a rather small impact of the N-/O-glycosylated sequon adjacent to the cleavage site on ADAM17-mediated proteolysis . In contrast , an N-linked glycan on Asn-55 , located in the D1 domain of the IL-6R far away from the cleavage site , was surprisingly identified as a protease-regulatory exosite , whose deletion caused increased shedding of the IL-6R . Importantly , this glycan on Asn-55 has no role in IL-6-mediated signaling . IL-6 binds to the CBM of the IL-6R located in domains D2 and D3 , and the D1 domain is not in contact with IL-6 or gp130 . Consequently , Hyper-IL-6 does not contain this domain [49] , and a membrane-bound IL-6R variant lacking D1 is biologically active [50] . Furthermore , the solved crystal structure of the hexameric IL-6/IL-6R/gp130 complex contains an IL-6R that only consists of the domains D2 and D3 [51] . Thus , our data suggest that the glycan on Asn-55 has a unique role in the regulation of proteolysis but is dispensable for signaling of IL-6 . Furthermore , an IL-6R mutant without any N-glycans reached the cell surface and was able to mediate IL-6-dependent signaling , but its shedding was severely impaired . It is currently unclear why this IL-6R variant , which contains the cleavage site and is able to coprecipitate with the protease via CANDIS , is nevertheless largely resistant towards proteolysis . A possible explanation is cooperativity between the individual glycosylation site , and simultaneous loss of all five N-glycans results in structural changes within the IL-6R ectodomain that disturb cleavage by the protease but do not perturb IL-6-dependent signaling . In summary , we identify a soluble form of the IL-6R in human serum that is generated by a protease in vivo and map the cleavage site by mass spectrometry . We mutate this cleavage site to generate IL-6R variants that are resistant towards ADAM-mediated cleavage . Furthermore , we map the occupancy of all N- and O-glycosylation sites of the sIL-6R and find that glycosylation is dispensable for trafficking , stabilization , and signaling of the IL-6R but is an important regulatory mechanism in terms of proteolysis .
Ethic approval for this study was obtained from the institutional review board of the Medical Faculty of Kiel University ( study #D 515/13 ) . Ba/F3-gp130 cells were obtained from Immunex ( Seattle , Washington , United States; [52] ) . The HEK293 cells were from DSMZ GmbH ( Braunschweig , Germany ) , and the Phoenix-Eco cells from U . Klingmüller ( DKFZ , Heidelberg , Germany ) . HEK293-EBNA cells have been described previously [53] . All cells were grown under standard conditions in DMEM high-glucose culture medium ( Sigma-Aldrich , Deisenhofen , Germany ) . The DMEM was supplemented with 10% fetal bovine serum , penicillin ( 60 mg/l ) , and streptomycin ( 100 mg/l ) , and the cells were kept at 37°C and 5% CO2 in a standard incubator with a water-saturated atmosphere . The Ba/F3-gp130 cells were cultured using 10 ng/ml of the recombinant IL-6/sIL-6R fusion protein Hyper-IL-6 . After stable transduction ( see below ) with IL-6R variants , the Ba/F3-gp130 cells were cultured with 10 ng/ml recombinant human IL-6 instead of Hyper-IL-6 . The anti-hIL-6R mAb 4–11 was described previously [31]; the anti-hIL-6R mAb ( sc-661 ) and the anti-β-actin mAb ( sc-47778 ) were from Santa Cruz Biotechnology ( Santa Cruz , California , US ) . The anti-GAPDH mAb was obtained from Cell Signaling Technology ( Frankfurt/M . , Germany ) . The biotinylated pAb goat anti-hIL-6R antibody Baf227 was from R&D Systems ( Wiesbaden-Nordenstadt , Germany ) . The rabbit pAb ds6R that specifically detected the C-terminus of the ds-sIL-6R was generated by Pineda Antikörper-Service ( Berlin , Germany ) . The peroxidase conjugated secondary antibodies were obtained from Pierce ( Thermo Scientific , Perbio , Bonn , Germany ) , and the APC-conjugated anti-mouse monoclonal secondary antibodies for flow cytometry experiments were obtained from Dianova ( Hamburg , Germany ) . PMA and ionomycin were purchased from Sigma-Aldrich ( Deisenhofen , Germany ) . G-418 and puromycin were from Carl Roth ( Karlsruhe , Germany ) . Purified human CG was purchased from Athens Research ( Athens , Georgia , US ) . PNGase F and the Deglyco-Mix were purchased from New England Biolabs GmbH ( Frankfurt am Main , Germany ) . Protein G- and A-agarose-beads were obtained from Merck/Millipore ( Darmstadt , Germany ) , and NHS-Sepharose from GE Healthcare ( München , Germany ) . Concanavalin A-covered sepharose beads were from Sigma-Aldrich ( Taufkirchen , Germany ) . Hyper-IL-6 was produced as described previously [54 , 55] . IL-6 was expressed , purified , and refolded as described previously [56] . The recombinant glycosylated catalytic domain ( aa Pro-18 to Val-477 ) of ADAM17 was purchased from Enzo Life Sciences GmbH ( Lörrach , Germany ) . Recombinant sIL-6R and recombinant ds-sIL-6R were expressed in HEK293-EBNA cells as follows: 2 x 106 cells were seeded in 10 cm cell culture dishes and transfected 24 h later using polyethylenimin ( PEI ) according to the manufacturer’s instructions . Forty-eight h later , expressing cells were selected with 0 . 25 mg/ml G-418 and 1 μg/ml puromycin for 3 wk . Afterwards , proteins were produced by seeding 4 x 106 cells in a 175 cm² cell culture flask in 20 ml DMEM ( supplemented with 10% fetal bovine serum , penicillin [60 mg/l] , streptomycin [100 mg/l] , G-418 [250 mg/l] , and puromycin [1 mg/l] ) . After the cells reached confluency , the medium was replaced by 20 ml DMEM ( supplemented with G-418 [250 mg/l] , puromycin [1 mg/l] , and 5% Ultra Low IgG FCS from PAN-Biotech [Aidenbach , Germany] ) . The medium was harvested two times a week , and the cell culture supernatant filtrated using a vacuum filter unit ( pore size 22 μM ) . After devolatilization , recombinant proteins were isolated with a His-Trap FF Crude column ( GE Healthcare , Munich ) on an Aekta Purifier ( GE Healthcare ) and eluted with 50 mM and 100 mM imidazole in PBS . The proteins were further purified via size exclusion chromatography on a Superdex 200 column ( GE Healthcare ) . Construction of the different expression plasmids encoding IL-6R variants was performed using standard molecular biology techniques with restriction-enzyme-based cloning . The template expression plasmid encoding the human IL-6R in pcDNA3 . 1 has been described previously [27] . Splicing by overlapping extension ( SOE ) -PCR was used to mutate single or multiple N- and O-glycosylation sites . Mutations within the protease cleavage site of the IL-6R were performed similarly . For retroviral transduction of Ba/F3-gp130 cells , constructs were subcloned into the pMOWS plasmid [57] . For the production of recombinant sIL-6R variants in HEK293-EBNA cells , a 6xHis-tag was inserted between the signal peptide and the D1 domain at the N-terminus . The ds-sIL-6R construct had a stop codon after the unique C-terminus GSRRRGSCGL , and the sIL-6R construct had a stop codon after Gln-358 ( see Fig 1A ) . The ADAM17-MPD-CANDIS construct was described previously [40] . Retroviral transduction of Ba/F3-gp130 cells via Phoenix-Eco cell supernatant has been described previously [57–59] . Cells stably expressing IL-6R were selected with puromycin ( 1 . 5 μg/ml ) and subsequently cultivated with 10 ng/ml recombinant IL-6 instead of Hyper-IL-6 . Proliferation of the different Ba/F3-gp130-IL-6R cell lines in response to IL-6 and Hyper-IL-6 and of Ba/F3-gp130 cells in response to IL-6 and the two different recombinant sIL-6R variants was determined using the Cell Titer Blue Cell viability assay reagent ( Promega , Karlsruhe , Germany ) , following the manufacturer’s protocol as described previously [59] . Values were measured in triplicates per experiment , and relative light units ( RLU ) obtained after 60 min were normalized by subtraction of control values obtained after 0 min . The maximum proliferation was set to 100% , and all other values were calculated accordingly . Peripheral blood from healthy volunteers was collected by venipuncture and serum isolated via centrifugation . Nine hundred μl of cell culture supernatant was mixed with 100 μl Concanavalin A beads and incubated overnight at 4°C under constant agitation . The beads were washed three times with 1 ml PBS each and boiled at 95°C for 5 min in 25 μl 3x Laemmli buffer ( 6% SDS , 30% Glycerol , 5% β-mercaptoethanol , 150 mM Tris-HCl [pH 6 . 8] , and 0 . 2% bromphenol blue ) . Samples were analyzed by western blot as described below . For mass spectrometry analysis , sIL-6R was precipitated using 4–11 antibodies covalently linked to NHS-beads ( 25 μl 4-11-beads per 1 ml supernatant ) . The samples were run on a 10% SDS-PAGE under reducing conditions and stained with Coomassie , and the region corresponding to sIL-6R was sliced out and analyzed by mass spectrometry ( see section below ) . Successful precipitation was confirmed by western blot . Twenty μl protein G- and 20 μl protein A-agarose-beads were added per 1 ml serum to remove antibodies and incubated overnight under constant agitation . Afterwards , the beads were removed by centrifugation , and sIL-6R was precipitated with 4-11-beads as described above . SDS-PAGE was performed under nonreducing conditions , the gel was stained with Coomassie , and the region corresponding to sIL-6R was sliced out and analyzed by mass spectrometry ( see section below ) . Successful precipitation was confirmed by western blot . After excision of gel bands and slicing them into small cubes ( ~1 mm3 ) , the gel pieces were washed with water and 50 mM ammonium bicarbonate ( ABC ) and destained ( 3–4 incubations with 30% acetonitrile [ACN]/50 mM ABC ) , followed by dehydration ( ACN ) and drying ( 10 min at 45°C in a SpeedVac vacuum centrifuge [Eppendorf , Hamburg , Germany] ) . Disulfide bond reduction was performed with 10 mM dithiothreitol ( DTT ) for 1 h at 56°C , followed by alkylation of reduced cysteines with 55 mM iodoacetamide ( IAA ) for 1 h at 25°C in the dark . Finally , gel bands were washed once more with H2O and twice with 50 mM ABC followed by dehydration with 100% ACN . Two strategies were applied ( i ) to identify the C-terminus of sIL-6R and ( ii ) to map the N-glycosylation sites . For the identification of the C-terminus of sIL-6R , N-glycans were deglycosylated with 50–100 U PNGase F overnight at 37°C in 50 mM ABC . Afterwards , the gel bands were washed once with 50 mM ABC and three times with H2O . After dehydration of the gel bands , proteins were in-gel digested overnight with 50–100 ng of the endoproteases chymotrypsin ( Sigma-Aldrich , Steinheim , Germany ) , trypsin , or Asp-N ( both from Promega , Madison , WI ) in 25 mM ABC containing 50% H218O ( from Eurisotop , St-Aubin Cedex , France ) . For the assignment of N-glycosylation sites , the same protocol was performed , but H218O was only used during the deglycosylation step , while the proteolysis with chymotrypsin , trypsin , or Asp-N was performed in 25 mM ABC . Samples were acidified with 10% formic acid ( FA ) to pH 3–4 . Peptide extraction was performed in three cycles with ( i ) 1% FA , ( ii ) 60% ACN/1% FA , and ( iii ) 100% ACN by incubating for 15 min at 25°C , with sonication for 2 min in ice-cold water . Between each step , the liquid supernatants were separated from the gel bands and collected in a second reaction tube . Merged supernatants were dried in a vacuum centrifuge , and the samples were reconstituted in 3% ACN/0 . 1% trifluoroacetic acid ( TFA ) . In-gel digested in vitro samples were analyzed on a Dionex Ultimate 3000 nano-HPLC coupled to a LTQ Orbitrap Velos with ETD ( Thermo Scientific , Bremen ) using an analytical setup described before [60] . After injection , samples were washed on a trap column ( Acclaim Pepmap 100 C18 , 10 mm × 300 μm , 3 μm , 100 Å , Dionex ) for 5 min with 3% ACN/0 . 1% TFA at a flow rate of 30 μl/min prior to peptide separation using an Acclaim PepMap 100 C18 analytical column ( 15 cm × 75 μm , 3 μm , 100 Å , Dionex ) . A flow rate of 300 nL/min using eluent A ( 0 . 05% FA ) and eluent B ( 80% ACN/0 . 04% FA ) was used for gradient separation as follows: linear gradient 5%–50% B in 120 min , 50%–95% B in 5 min , 95% B for 10 min , 95%–5% B in 0 . 1 min , and equilibration at 5% B for 10 min . Spray voltage applied on a metal-coated PicoTip emitter ( 30 μm tip size , New Objective , Woburn , Massachusetts , US ) was 1 . 25 kV , with a source temperature of 250°C . Full scan MS spectra were acquired from 5 to 145 min between 300 and 2 , 000 m/z at a resolution of 60 , 000 at m/z 400 ( automatic gain control [AGC] target of 1e6; maximum ion injection time [IIT] of 500 ms ) . The five most intense precursors with charge states ≥2+ were used ( i ) for collision-induced dissociation ( CID ) with fragment ion detection in the ion trap ( parameters: normalized collision energy [NCE] of 35%; isolation width of 2 m/z; resolution , AGC target of 1e4 and maximum IIT of 100 ms ) and ( ii ) for higher-energy collisional dissociation ( HCD ) with fragment detection in the orbitrap with an NCE of 40% ( parameters: isolation width of 3 m/z; resolution , 7 , 500 at m/z 400; AGC target of 1e5 and maximum IIT of 1 , 000 ms ) . The precursor mass tolerance was set to 10 ppm , and dynamic exclusion ( 30 s ) was enabled . For all spectra , a lock mass correction was performed using the polysiloxane contaminant signal at 445 . 120025 m/z . The serum samples were analyzed on a Dionex Ultimate 3000 nano-UHPLC coupled to a QExactive Plus ( both from Thermo Scientific , Bremen ) . Injected samples were loaded and washed as described above . Peptide separation was performed using an Acclaim PepMap 100 C18 analytical column ( 50 cm × 75 μm , 2 μm , 100 Å , Dionex ) with a flow rate and eluent composition as described above: linear gradient 5%–50% B in 120 min , 50%–90% B in 5 min , 90% B for 10 min , 90%–5% B in 0 . 1 min , and equilibrating at 5% B for 10 min . Ionization was performed with 2 . 4 kV spray voltage applied on a noncoated PicoTip emitter ( 10 μm tip size , New Objective , Woburn , Massachusetts , US ) with the source temperature set to 250°C . MS data were acquired from 5 to 115 min with MS full scans between 300 and 2 , 000 m/z at a resolution of 70 , 000 at m/z 200 ( AGC value of 3e6 and maximum IIT of 100 ms ) . The ten most intense precursors with charge states ≥2+ were subjected to fragmentation with HCD with NCE of 28% ( isolation width of 3 m/z; resolution , 17 , 500 at m/z 200; AGC target of 1e5 and maximum IIT of 100 ms ) . Dynamic exclusion for 45 s was applied with a precursor mass tolerance of 10 ppm . Lock mass correction was performed as described above . Acquired spectra were first searched by computer-assisted database searches using Proteome Discoverer 1 . 4 . 1 . 14 with the search engines SequestHT ( Thermo Scientific ) and Mascot ( Matrix Science , Boston , Massachusetts ) . Searches were performed against the full human protein database including isoforms ( downloaded from Uniprot , 17 July 2012 , appended with common contaminants and enzymes used ) using the following settings: semienzyme specificity; three missed cleavage sites; mass tolerances of 10 ppm for precursors and for fragment masses 0 . 02 Da ( HCD ) and 0 . 5 Da ( CID ) ; static modifications: carbamidomethylation on Cys; and dynamic modifications: oxidation of Met , deamidation of Asp , and 18O-incorporation at peptides C-termini . Beside automated database searches , manual MS/MS-spectra interpretation was performed . Analysis of IL-6R ectodomain shedding in transiently transfected HEK293 cells or stably transduced Ba/F3-gp130 cell lines by ADAM10 and ADAM17 has been described elsewhere [7 , 27] . Analysis of IL-6R proteolysis by CG has been described previously [34] . Solid phase peptide synthesis was carried out on a PTI Tribute peptide synthesizer using Trt-ChemMatrix resin ( PCAS Biomatrix , Quebec , Canada ) . The resin ( 500 mg , loading 0 . 42 mmol/g ) was activated with 10% acetyl bromide in dichloromethane ( DCM ) over 3 h and washed with absolute dichloromethane . The activated resin was reacted with the first amino-acid ( 5 eq . ) and DIPEA ( 5 eq . ) in DCM . After 16 h , the resin was washed with NMP and DCM and quenched with a solution of DCM/MeOH/DIPEA ( 17:2:1 ) ( 3x2 min ) . The resin was subsequently washed with DCM and dried . Peptide synthesis was carried out automatically using 5 eq . of Fmoc amino acid , 5 eq . of HCTU and 9 eq . of 0 . 4 M DIPEA in DMF . The Fmoc group was cleaved with 20% piperidine in DMF . The building block Fmoc-Thr ( Ac3GalNAc ) -OH was synthesized according to [61] and incorporated into the peptide manually . Fmoc-Thr ( Ac3GalNAc ) -OH ( 2 eq . ) and HOOBt ( 2 eq . ) were dissolved DCM/DMF ( 1:1 ) , DIC ( 2 eq . ) was added , and after 20 min at ambient temperature , the mixure was reacted with the resin for 2 h . On resin , deacetylations were carried out with 5% hydrazine hydrate in DMF . On resin , deallylations and subsequent couplings with GlcNAc amine or the biantennary nonasaccharide amine were carried out according to [62] . The peptides and glycopeptides were cleaved from the resin with a mixture of TFA/TES/H2O ( 96:2:2 ) for 10 min ( 3x ) . The filtrates were concentrated , and the peptides and glycoeptides were precipitated by addition of diethyl ether ( 10-fold volume ) followed by centrifugation . The precipitate was purified by HPLC ( Supleco Ascentis C-18 , 5μ , 10 x 250 mm , gradient of CH3CN-H2O with 0 . 1% TFA ) . The N , O-glycosylated peptides 9 and 10 were obtained using Fmoc-Asp ( PhiPr ) OH [63] . The O-glycosylated peptide was cleaved from the resin with 20% HFIP in DCM for 30 min . After removal of the solvent , the residue was reacted with 6 eq . of t-butylcarbazate , 4 eq . of Cl-HOBt , and 4 eq . of DIC at 0°C . The mixture was warmed to ambient temperature and was subsequently ( after 24 h ) purified by flash chromatography ( DCM/MeOH ) ( 0%–15% MeOH ) . For removal of the PhiPr group , the protected hydrazide was dissolved in DCM containing 1% of TFA . After 30 min , the solution was extracted with dilute KHCO3 , dried , and concentrated . The aspartylation of the O-glycopeptides was carried out with 1 . 5 eq . of nonasaccharide amine according to [62] in DMF-DMSO ( 1:1 ) using 4 eq . of DIPEA , 2 . 5 eq . of HOAt , and 2 . 5 eq . of HATU . After 24 h , the reaction mixture was deprotected by addition of the 20-fold volume of TFA/TES/H2O ( 96:2:2 ) . The mixture was concentrated to half the volume after 2 h . Diethyl ether ( 10-fold volume ) was added , and the precipitated glycopeptide was treated with hydrazine hydrate ( 5% in CH3CN-H2O 1:9 ) , dried , and purified by HPLC as described above . The yields were as follows: 1 ( 15% ) , 2 ( 15% ) , 3 ( 16% ) , 4 ( 17% ) , 5 ( 12% ) , 6 ( 13% ) , 7 ( 25% ) , 8 ( 22% ) , 9 ( 25% ) , 10 ( 19% ) . Two 1 . 5 ml reaction tubes were filled with 50 μl peptide ( 50 μM ) solved in 25 mM Hepes ( pH 9 . 0 ) , and 0 . 5 μl of the recombinant catalytic domain of ADAM17 was added to one of the reaction tubes , whereas the other sample remained untreated . All samples were incubated for 16 h at 37°C . Afterwards , 1 , 150 μl 0 . 1% TFA was added to each sample , and the peptides were separated on a reverse phase chromatography column ( Multo High Bio-200-C18 ) from Chromatographie Service ( Langerwehe , Germany ) using a linear gradient elution with a binary solvent system and a flow rate of 1 ml/min . Solvent A consisted of 0 . 1% TFA , and solvent B consisted of 100% ACN . After peak integration , peak areas for cleaved and noncleaved peptides were determined . The sIL-6R was precipitated with 4–11 beads from 10 ml supernatant of transiently transfected HEK293 cells after PMA stimulation . Furthermore , about 8 x 106 HEK293 cells , transiently transfected with a cDNA encoding the human IL-6R , were lysed ( 50 mM Tris , pH 7 . 5 , 150 mM NaCl , 1% Triton X-100 , and complete protease inhibitor mixture tablets ) . Full-length IL-6R was precipitated from the lysate with 4–11 beads . The 4–11 beads were washed three times with PBS , and sIL-6R and full-length IL-6R were eluted in 74 μl H2O , 10 μl G7 buffer , 1 μl 10% NP 40 , and 2 . 5 μl β-mercaptoethanol ( 10 min , 95°C ) . The eluted volume was divided into three 1 . 5 ml reaction tubes . The first tube remained untreated , the second was treated with 2 μl PNGase F , and the third was treated with 3 μl Deglyco-Mix . All samples were incubated at 37°C overnight . After addition of 15 μl 5x Laemmli buffer , samples were analyzed by western blot . Analysis of ( s ) IL-6R via SDS-PAGE and western blot has been described previously [34] . The immunoprecipitation of the IL-6R and ADAM17-MPD-CANDIS was performed as described previously [40] . IL-6R cell surface expression on stably transduced Ba/F3-gp130 cell lines and transiently transfected HEK293 cells was analyzed as described previously [7] . Flow cytometry was performed with a BD Biosciences FACS Canto and FCS Express V3 ( De Novo Software , Los Angeles , California , US ) . In order to analyze internalization of the IL-6R , 1 x 107 cells were washed in PBS , seeded onto a 10 cm dish in serum-free DMEM , and starved for 2 h at 37°C . Afterwards , cells were again washed with PBS , and the IL-6R was labeled with the 4–11 antibody diluted in FACS buffer ( 1% BSA in PBS ) . After 1 h incubation on ice , cells were washed and resuspended in 10 ml serum-free DMEM , seeded onto a 10 cm dish , and shifted back to 37°C . At the time points indicated ( 0–180 min ) , 1 ml of the cell suspension was taken from the dish , transferred into a 1 . 5 ml reaction tube , and kept on ice to inhibit further internalization of the IL-6R . After obtaining the last sample , the cells were washed in cold FACS buffer , and the remaining IL-6R on the cell surface was stained with an APC-conjugated anti-mouse mAb . The cells were washed again and analyzed by flow cytometry using the BD Biosciences FACS Canto II ( Becton-Dickinson , Heidelberg , Germany ) . The sandwich ELISA , which detects sIL-6R as well as ds-sIL-6R ( 4-11/Baf227 antibodies ) and can be used to quantify total sIL-6R serum levels , was described previously [21 , 31] . To specifically detect ds-sIL-6R ( either recombinant or in human serum ) , a similar approach was used ( ds6R/Baf227 ) . Both sandwich ELISAs were performed with streptavidin-horseradish peroxidase ( R&D Systems , Minneapolis , Minnesota , US ) and the peroxidase substrate BM blue POD ( Roche , Mannheim , Germany ) . The enzymatic reaction was stopped by addition of 1 . 8 M sulfuric acid , and the absorbance read at 450 nm on a Tecan rainbow reader ( Tecan , Crailsheim , Germany ) . Statistical analyses were performed with GraphPad Prism ( GraphPad Software , La Jolla , California , US ) . Data were analyzed by a Student’s t test , and the differences were indicated ( */§p < 0 . 05; **p < 0 . 01; ***p < 0 . 001 ) . The one-sample t test was used to compare values to normalized data ( e . g . , comparison to 100 . 0 values ) . p-Values were corrected via the Bonferroni method for multiple comparisons . The numerical data used in all figures are included in S1 Data . | Interleukin-6 ( IL-6 ) is a cytokine secreted by our body upon infection or trauma to stimulate the immune system response . IL-6 is partially responsible for fever and triggers inflammation in many diseases . It activates its target cells via the membrane-bound IL-6 receptor ( IL-6R ) , and soluble forms of this receptor ( sIL-6R ) are present in high amounts in the serum of healthy individuals and mediate the inflammatory response in all cells of the human body . However , it remains unclear how the soluble form of this cytokine is generated in humans . In this study , we isolate sIL-6R from human serum and show that the majority is produced via cleavage of the membrane-bound IL-6R by a protease . We identify the exact cleavage site and find that it is identical to a cleavage site used by the metalloprotease ADAM17 . We further show that glycosylation , a post-transcriptional modification , is dispensable for the transport and biological function of IL-6R and map the occupancy of all O- and N-glycosylation sites . However , we find that only a single N-glycan is critically involved in the regulation of proteolysis by ADAM17 and conclude that glycosylation is an important regulator for sIL-6R generation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"flow",
"cytometry",
"medicine",
"and",
"health",
"sciences",
"enzyme-linked",
"immunoassays",
"enzymes",
"metabolic",
"processes",
"enzymology",
"plasmid",
"construction",
"physiological",
"processes",
"dna",
"construction",
"glycosylation",
"molecular",
"biology",
"techni... | 2017 | Proteolytic Origin of the Soluble Human IL-6R In Vivo and a Decisive Role of N-Glycosylation |
During viral infections cellular gene expression is subject to rapid alterations induced by both viral and antiviral mechanisms . In this study , we applied metabolic labeling of newly transcribed RNA with 4-thiouridine ( 4sU-tagging ) to dissect the real-time kinetics of cellular and viral transcriptional activity during lytic murine cytomegalovirus ( MCMV ) infection . Microarray profiling on newly transcribed RNA obtained at different times during the first six hours of MCMV infection revealed discrete functional clusters of cellular genes regulated with distinct kinetics at surprising temporal resolution . Immediately upon virus entry , a cluster of NF-κB- and interferon-regulated genes was induced . Rapid viral counter-regulation of this coincided with a very transient DNA-damage response , followed by a delayed ER-stress response . Rapid counter-regulation of all three clusters indicated the involvement of novel viral regulators targeting these pathways . In addition , down-regulation of two clusters involved in cell-differentiation ( rapid repression ) and cell-cycle ( delayed repression ) was observed . Promoter analysis revealed all five clusters to be associated with distinct transcription factors , of which NF-κB and c-Myc were validated to precisely match the respective transcriptional changes observed in newly transcribed RNA . 4sU-tagging also allowed us to study the real-time kinetics of viral gene expression in the absence of any interfering virion-associated-RNA . Both qRT-PCR and next-generation sequencing demonstrated a sharp peak of viral gene expression during the first two hours of infection including transcription of immediate-early , early and even well characterized late genes . Interestingly , this was subject to rapid gene silencing by 5–6 hours post infection . Despite the rapid increase in viral DNA load during viral DNA replication , transcriptional activity of some viral genes remained remarkably constant until late-stage infection , or was subject to further continuous decline . In summary , this study pioneers real-time transcriptional analysis during a lytic herpesvirus infection and highlights numerous novel regulatory aspects of virus-host-cell interaction .
Herpesviruses are large DNA viruses which cause a broad range of disease ranging from the common cold sore to cancer . They all share the ability to establish a life-long , latent infection , leaving the infected individual at constant risk of reactivation and subsequent disease . The human cytomegalovirus ( HCMV ) poses a severe threat to immunocompromised patients and represents the most common infective cause of congenital disorders affecting about 1 in 1 , 000 newborns [1] . Like all herpesviruses , cytomegaloviruses ( CMV ) have co-evolved with their animal and human hosts for millions of years . During this time , they have mastered host-cell modulation to facilitate their needs and thus provide ideal tools to study many fundamental cellular processes . Numerous signaling events are triggered during the first few hours of infection . As such , binding of CMV particles to the cell membrane and virus entry result in the activation of cellular signaling pathways , some of which , e . g . NF-κB signaling , play an important role in initiating lytic viral infection [2]–[4] . Concomitantly , viral pathogen-associated molecular patterns are recognized by pattern-recognition receptors , resulting in robust activation of an innate immune response . Virion-associated proteins as well as the advent of viral gene expression then counteract intrinsic and arising host cell defense [5] . Several high-throughput studies addressed the transcriptional response of the cell to lytic CMV infection by analyzing temporal changes in total RNA levels [3] , [6]–[10] . These studies revealed lytic CMV infection altered the expression of numerous cellular genes involved in a variety of processes including inflammation , innate immunity , cell cycle progression , cellular metabolism and cell adhesion . One of the earliest events upon entry of the viral DNA into the nucleus is the deposition of viral genomes at nuclear domain ( ND10 ) bodies [11] , [12] . This appears to be part of an intrinsic antiviral defense mechanism suppressing the expression of foreign DNA entering the nucleus [13] . In part , this is mediated by chromatin-remodeling enzymes recruited to these structures [14]–[16] . In HCMV infection , this intrinsic host defense is overcome by the viral tegument protein pp71 [17] , [18] as well as the viral immediate early 1 ( IE1 ) protein [19] , [20] . In lytic murine cytomegalovirus ( MCMV ) infection , dispersion of ND10 bodies seems to be predominately mediated by the IE1 protein [21] ( reviewed in [22] ) . In addition to disruption of ND10 body-mediated antiviral defense , the immediate-early proteins initiate the lytic replication cycle by facilitating the transcription of early genes [23] , [24] . The latter then modulate host cell environment , disarm the arising immune response , and establish the viral replication machinery . Upon viral DNA replication , viral late gene expression is initiated , culminating in the production and release of infectious virus particles [25] . The analysis of de novo early viral gene expression has been substantially hindered by large amounts of so called ‘virion-associated RNA’ , unspecifically bound by the virus particles and delivered to the newly infected cell [26]–[30] . Chromatin immunoprecipitation ( ChIP ) has thus been employed to study the kinetics of viral transcriptional activity by looking at markers of active and inactive chromatin associated with the viral promoters . Immediately upon infection of permissive fibroblasts ( at ‘pre-IE’ times of infection , using low multiplicities of infection ) HCMV genomes become associated with markers of repressed chromatin [31] . As infection progresses , the chromatin status of viral promoters reflects the cascade of viral immediate-early , early and late gene expression [32] , [33] . Standard gene expression analysis ( using total RNA ) to study kinetics of transcriptional regulation has several limitations . Firstly , short-term changes in total RNA levels do not match changes in transcription rates but are inherently dependent on the RNA half-life of the respective transcripts [34] . This strongly favors the detection of up-regulation of short-lived transcripts , commonly encoding for transcription factors and genes with regulatory function . This , in turn , may result in substantial bias in downstream bioinformatics analyses . Secondly , the temporal resolution - particularly for down-regulated genes - is rather low due to the relatively long median RNA half-life ( 5–10 h ) in mammalian cells [35] , [36] . The same is true for detecting ( viral ) counter-regulation of cellular genes induced earlier in infection . Thirdly , alterations in RNA synthesis rates cannot be differentiated from changes in RNA decay rates . Finally , transcriptional activity of the incoming CMV genomes cannot be definitively studied due to the presence of virion-associated RNA introduced to the newly infected cells by the incoming virus particles [28] , [37] . Recently , we developed an approach termed 4-thiouridine- ( 4sU ) -tagging to purify newly transcribed RNA from total cellular RNA [34] . This is applicable to a broad range of organisms including vertebrates , drosophila and yeast [38] , [39] . In short , cells are cultured in presence of 4sU resulting in metabolic thiol-labeling of newly transcribed RNA at a frequency of about one 4sU residue in 50 to 100 nucleotides [34] . After isolation of total cellular RNA , RNA-incorporated 4sU is thiol-specifically biotinylated . Labeled newly transcribed RNA is then efficiently purified from total RNA using streptavidin-coated magnetic beads . All three RNA fractions , i . e . total , newly transcribed and unlabeled pre-existing RNA , are suitable for quantitative RT-PCR ( qRT-PCR ) , microarray analysis and next-generation sequencing [34] , [40]–[43] . In the present study , we employed this approach to lytic murine cytomegalovirus ( MCMV ) infection of fibroblasts to study the real-time kinetics of cellular and viral gene expression using qRT-PCR , microarray analysis and RNA-sequencing ( RNA-seq ) . We show that this approach circumvents all the caveats mentioned above , thereby providing intriguing new insights into cytomegalovirus host-cell modulation and regulation of viral gene expression .
Upon its addition to the cell culture medium , 4-thiouridine ( 4sU ) is rapidly taken up by cells , phosphorylated and incorporated into newly transcribed RNA in a concentration-dependent manner [34] . To establish 4sU-tagging for MCMV infection , we first analyzed the effect of lytic MCMV infection on 4sU-incorporation . NIH-3T3 fibroblasts were infected with MCMV at a multiplicity of infection ( MOI ) of 10 . At different times of infection , 200 µM 4sU was added to the cell culture medium for 1 h . Total RNA was prepared and subjected to thiol-specific biotinylation . 4sU- ( biotin ) -incorporation was quantified by dot blot ( Figure 1A ) . At all times of infection , 4sU-incorporation was at least as efficient as in uninfected cells , ensuring efficient purification of newly transcribed RNA at all times of infection . Interestingly , from 5 to 24 hours post infection ( hpi ) the extent of 4sU-incorporation into cellular RNA was about 20-fold greater than in uninfected cells . By 47–48 hpi this had returned to levels found in uninfected cells . These data are consistent with increased transcriptional activity as well as enhanced nucleoside metabolism during lytic CMV infection [44] . Metabolic labeling of newly transcribed RNA with 4sU has negligible polar effects on eukaryotic cells [34] , [45] , [46] . To exclude gross adverse effects of 4sU-labeling on MCMV replication , we applied 1 h of 200 µM 4sU-treatment to NIH-3T3 cells at different times of infection . No effect of 4sU exposure on virus titers , determined at 48 hpi , was observed ( Figure S1 ) . We therefore decided to use 1 h of 200 µM 4sU in all following experiments . To detail transcriptional changes in host gene expression during early MCMV infection , we infected NIH-3T3 fibroblasts with MCMV at an MOI of 10 and labeled newly transcribed RNA from 1–2 , 3–4 and 5–6 hpi . Three replicates of both total and newly transcribed RNA were subjected to Affymetrix Gene ST 1 . 0 arrays . After Robust Multichip Average ( RMA ) normalization , we identified all genes significantly regulated ( p≤0 . 05 ) by at least 2-fold compared to uninfected cells in any condition . This resulted in the identification of 1 , 674 probe sets showing differential expression ( Table S1a ) . With the exception of 4 genes , all differentially expressed genes were either exclusively up- or down-regulated during the first 6 h of infection . The number of genes with differential expression detectable in total RNA only represented 13% ( at 2 hpi ) , 25% ( at 4 hpi ) and 54% ( at 6 hpi ) of those identified in newly transcribed RNA . As predicted , down-regulation only started to become detectable in total RNA with substantial delay , i . e . at 4 hpi ( Figure 1B ) . Furthermore , a peak of MCMV-induced and rapidly counter-regulated gene expression was apparent in newly transcribed RNA at 3–4 hpi . This was invisible in total RNA . The overlap of differential gene expression detectable at different times of infection was substantially greater for newly transcribed RNA ( Figure 1C ) . Notably , we found all genes induced or repressed by at least 2-fold in total RNA at 6 hpi to show concordant regulation in newly transcribed RNA ( Figure 1D ) . In addition , only 3 probe sets showed more than 2-fold greater regulation in total RNA than in newly transcribed RNA ( genes listed in Table S1b ) . Hence , the vast majority of differential gene expression during the first six hours of MCMV infection is the result of alterations in transcription rates and not due to changes in RNA decay rates . We therefore decided to focus all our subsequent analyses on newly transcribed RNA . Clustering genes based on >2-fold differences in regulation at different times of infection , we identified 5 clusters of genes characterized by distinct kinetic profiles ( Figure 2A; for details and genes represented in each cluster see Table S1c and materials and methods ) . MCMV-induced genes peaked at 1–2 ( Cluster 1 ) , 3–4 ( Cluster 2 ) or 5–6 hpi ( Cluster 3 ) . Of these , Cluster 2 was not detectable in total cellular RNA at all . Rapid and rather constant down-regulation was characteristic of genes in Cluster 4 , while genes in Cluster 5 showed delayed down-regulation . These five clusters also became apparent when unsupervised clustering based on the changes in induction between 1–2 , 3–4 and 5–6 hpi was performed ( see Figure S2 ) . To look for functional characteristics of these five gene clusters , we performed an enrichment analysis of Gene Ontology ( GO ) terms ( biological process ) and KEGG pathways . Interestingly , all five clusters were associated with distinct functional terms ( see Figure 2B; for complete list of over-represented gene ontologies see Table S2a ) . Genes in Cluster 1 were involved in immune and inflammatory processes as well as apoptosis . Genes in Cluster 2 mainly played a role in p53 signaling and cell cycle progression . Delayed induction was observed for genes involved in the ER stress response ( Cluster 3 ) . Rapid and sustained down-regulation was observed for genes involved in cell proliferation and differentiation , focal adhesion as well as actin filament-based processes ( Cluster 4 ) . Finally , delayed down-regulation was characteristic of genes with a role in chromatin assembly and cell cycle processes ( Cluster 5 ) . Due to the delayed visibility of down-regulation in total RNA ( see Figure 1B ) , Cluster 4 and 5 could only be differentiated using newly transcribed RNA . This approach thus allowed dissecting differential gene expression into discrete functional clusters regulated with distinct kinetics . These provided us with ideal templates to elucidate the underlying transcription factors and molecular mechanisms using in silico promoter analysis . We performed promoter analysis on the five clusters to identify cellular transcription factors ( TFs ) involved in their regulation . Proximal promoter regions ( PPR ) ranging from −500 to +100 bp from the transcription start site ( TSS ) were analyzed for over-represented transcription factor binding motifs . While a number of transcription factor binding motifs were over-represented in the five clusters ( for complete list and data see Table S2b ) , we observed distinct transcription factor binding sites to be uniquely over-represented in each of the individual clusters ( see Figure 2C for exemplary TFs and Table S2b ) . These correlated very well with the functional annotations of the associated clusters . In Cluster 1 , uniquely over-represented binding sites were found for NF-κB and IRF-1 . NF- κB is one of the key players of the cellular immune response and it thus has important roles in the antiviral defense [47] . Infection with CMV results in an activation of this transcription factor [48] , however , subsequently it is counter-regulated by the virus . In the case of MCMV this is mediated by the viral protein M45 [49] , [50] . IRF-1 is an important factor in the antiviral IFN response and , like canonical type I interferon signaling , results in activation of promoters containing interferon stimulated response elements ( ISREs ) . HCMV pp65 can inhibit activity of IRF-1 [51] . In addition , the MCMV M27 counteracts the interferon response by targeting Stat2 for degradation [52] . Elk-1 and YY1 are examples of transcription factors for which binding sites were over-represented in the cluster of genes ( Cluster 2 ) showing a peak of induction at 3–4 hpi . Elk-1 is activated by the MAPK/ERK pathway , which is stimulated during HCMV infection [53] . YY1 is a DNA-binding transcription factor that acts as a repressor of some promoters and an activator of others [54] . It has been shown that YY1 can directly bind to the HCMV major IE promoter region and mediates repression of HCMV IE gene expression [55] . In latent MCMV infection , YY1 is also specifically recruited to the major immediate early protein ( MIEP ) promoter and might thus play a role in the control of latency and reactivation [56] . For Cluster 3 , which contained genes induced with delayed kinetics ( Figure 3C ) , we found over-represented binding sites for c-Myc and Ap-4 . c-Myc is a proto-oncogene which drives cell cycle progression and apoptosis , whereas cellular differentiation and cell adhesion are negatively influenced [57] . Both , IE1 and IE2 proteins of HCMV were shown to be able to up regulate the c-Myc promoter and thus to increase c-Myc expression [58] . AP-4 was identified to be a direct transcriptional target of c-Myc [59] . In Cluster 4 , genes had uniquely over-represented binding sites for Mzf1 and AP2 . Mzf1 belongs to the Krüppel family of zinc finger proteins and plays a role in regulating transcriptional events during hemopoietic development and controls cell proliferation as well as tumorigenesis [60] , [61] . Until now , nothing is known about a role of Mzf1 upon CMV infection . AP2 are a family of transcription factors which were shown to play role in apoptosis , cell-cycle control and proliferation as well as tumorigenesis [62] . Little is known about a role of AP2 in CMV infection . Cluster 5 showed uniquely over-represented binding sites for E2F and C/EBP . E2F is a family of TFs which are involved in the regulation of S phase genes as well as DNA damage and apoptosis [63] . For E2F-1 , Song and Stinski showed that HCMV increases its activity [64] . C/EBP belongs to the bZIP transcription factors and has important function in adipocyte differentiation , maintains energy homeostasis and regulates cell differentiation [65] . It can induce growth arrest by interacting with CDK2 and CDK4 and interacts with the heterodimer E2F-DP to inhibit cell growth [66] , [67] . To date , little is known about its function in CMV infection , however , the MIEP promoter of CMV contains a binding site for this transcription factor [68] . Two exemplary TFs were chosen for further validation . As a proof-of-principle TF , we decided to look at NF-κB to see whether it's well-described rapid induction and counter-regulation during CMV infection [48] , [49] would precisely reflect the transcriptional changes we observed in newly transcribed RNA under our experimental conditions . NF-κB-dimers of the NF-κB- ( p105 and p100 ) and Rel-subfamily ( c-Rel , RelB and RelA ) are present in inactive IκB-bound complexes in the cytoplasm . IκK-mediated phosphorylation induces degradation of the inhibitor IkBα , enabling translocation of the NF-κB dimers to the nucleus and enhanced transcription of NF-κB target genes ( reviewed in [69] ) . To look for degradation of the inhibitor IκBαduring the first 12 h of MCMV infection , we performed immunoblotting ( Figure 3A ) . In addition , immunofluorescence analysis was performed to reveal the shift of RelA into the nucleus ( Figure 3B ) . Results from both experiments demonstrated the kinetics of transcriptional regulation of genes in Cluster 1 to precisely mirror NF-κB activation , highlighting the ability of 4sU-tagging to detail real-time transcription factor activity . In addition , we looked at a representative TF of Cluster 3 , namely c-Myc . c-Myc forms a heterodimer with Max , followed by its binding to target genes [70] . Furthermore , phosphorylation of two amino acids at the NH2-terminal domain is important for transactivation of c-Myc [71] . Hagemeier et al . showed that HCMV IE1 and IE2 can transactivate the c-Myc promoter [58] . We performed luciferase assays using a c-Myc-specific reporter construct transfected into NIH-3T3 cells 48 h prior to infection to analyze c-Myc activation . Luciferase activity started to significantly increase at 4 hpi , matching the expression kinetics of genes in Cluster 3 ( Figure 3C ) . We then addressed the role of viral gene expression in the regulation of each cluster using infection with UV-inactivated virus . To provide a more comprehensive picture , we extended the kinetics until 48 hpi . To this end , NIH-3T3 cells were infected with either wild-type ( wt ) or UV-inactivated virus . RNA was labeled for 1 h at different times of infection and newly transcribed RNA was purified . Transcription rates of exemplary genes of each functional cluster were determined in newly transcribed RNA using quantitative RT-PCR ( qRT-PCR ) . This included NF-κB- ( Cluster 1 ) , interferon- ( Cluster 1 ) , DNA-damage- ( Cluster 2 ) and ER-stress- ( Cluster 3 ) induced genes as well as MCMV-repressed genes involved in the regulation of cell differentiation ( Cluster 4 ) and cell cycle/chromatin organization ( Cluster 5 ) . The housekeeping gene Lbr ( Lamin B receptor ) was used for normalization . Cluster 1 contains both NF-κB- as well as interferon-induced genes . We thus chose NF-κBiα ( NF-κB-inhibitor alpha ) , an NF-κB-induced negative regulator of the NF-κB response , as well as Ifit1 ( Interferon-induced protein with tetratricopeptide repeats 1 ) for this analysis . Both NF-κBiα and Ifit1 were rapidly induced and counter-regulated by lytic MCMV infection ( Figure 4A , B ) . Induction of both genes following infection with UV-inactivated virus was comparable to wt-MCMV infection , consistent with previous reports showing that viral gene expression is not required for induction of both NF-κB- and interferon-signaling . In both cases , however , counter-regulation was substantially delayed following infection with UV-inactivated virus . While counter-regulation of the NF-κB response is consistent with the MCMV M45 gene product efficiently targeting NF-κB- signaling [50] , a viral gene product targeting the induction of the interferon response remains to be identified [72] . To monitor DNA-damage response-mediated signaling , we analyzed transcriptional activity of Gadd45a ( Growth arrest and DNA damage-inducible protein A ) , a well characterized DNA damage- induced gene [73] . Consistent with our microarray data , qRT-PCR revealed the same slightly delayed induction at 3–4 hpi , followed by a more protracted counter-regulation than we observed for NF-κBiα and Ifit1 . Interestingly , UV-inactivated virus also triggered the induction of Gadd45a with similar kinetics . This was , however , no longer counter-regulated , but continued to increase until 48 hpi ( Figure 4C ) . For HCMV it has been described that IE1 increases p53 activity by phosphorylation through ATM , an important kinase in the DNA damage response [74] . While our data are indicative of counter-regulation of this response by an MCMV gene product , we cannot exclude that the induction and the enhanced response - at least in parts - reflects increased activation by the UV-damaged viral DNA . For Cluster 3 , expression of Herpud1 ( Homocysteine-responsive endoplasmic reticulum-resident ubiquitin-like domain member 1 protein ) , a gene induced by endoplasmatic reticulum ( ER ) stress [75] , was monitored . Delayed induction was observed , which was rapidly counter-regulated . Induction of Herpud1 was lost upon infection with UV-inactivated virus , consistent with viral gene expression being required for the induction of the ER stress response . In summary , these findings indicate that a so far unknown viral gene product counteracts the ER stress response provoked by viral gene expression ( Figure 4D ) . For HCMV , this function is thought to be performed by the viral pUL38 protein [76] . For Clusters 4 and 5 we chose to monitor the transcription kinetics of Lamb1-1 ( Laminin beta 1 ) , an important extracellular matrix glycoprotein , and Top2α ( Topoisomerase 2 alpha ) , which is involved in the control and alteration of the topologic states of DNA during transcription [77] , [78] . Interestingly , consistent down-regulation of both genes was observed following wt-MCMV , but not UV-MCMV infection , indicating that viral gene expression is required for both their regulation ( Figure 4 E , F ) . In summary , these data highlight that all cellular signaling pathways we identified to be induced during early MCMV infection are rapidly counter-regulated by the virus later on . In contrast , down-regulation of defined cellular signaling pathways prevails and thus most likely represents an intentional action of the virus to facilitate its needs . Herpesvirus particles , like other herpesviruses , unspecifically incorporate and transfer large amounts of so called ‘virion-associated RNA’ to newly infected cells [26]–[30] , [37] . This has substantially hindered detailed studies on the kinetics of viral gene expression during the first few hours of infection and in latency . 4sU-tagging allows the removal of virion-associated RNA and thus , the dissection of the regulation of viral gene expression during the initial phase of infection . To show that 4sU-tagged newly transcribed RNA fraction is indeed free of virion-associated RNA , we labeled newly transcribed RNA in MCMV infected NIH-3T3 cells from 1–2 , 3–4 and 7–8 hpi in the presence and absence of the RNA polymerase II inhibitor Actinomycin D ( Act-D ) . Act-D treatment inhibits RNA synthesis and thus prevents 4sU-incorporation into newly transcribed RNA . Following isolation of total RNA , we included a DNase digest prior to biotinylation to further remove viral DNA . Newly transcribed and total RNA samples were subjected to qRT-PCR analysis for the spliced viral ie1 gene and the cellular housekeeping gene Lbr . In total RNA , ie1 transcripts were detectable even in presence of Act-D , consistent with large amounts of virion-associated RNA delivered to the infected cells . However , in newly transcribed RNA virtually no ie1 and Lbr transcripts ( below detection limit of our qRT-PCR assay ) were detectable in presence of Act-D , consistent with the complete removal of virion-associated RNA ( Figure 5A ) . We then performed a comprehensive time-course analysis of transcriptional activity during lytic MCMV infection . To study relevant time frames , we first determined the kinetics of viral DNA replication in NIH-3T3 cells infected with wt-MCMV at an MOI of 10 . Both qRT-PCR on M54 , the catalytic subunit of MCMV DNA polymerase , and southern blot analysis of concatameric DNA revealed viral DNA replication to start at ∼15 hpi ( Figure S3 ) . We therefore decided to label newly transcribed RNA at 1–2 hpi , 3–4 hpi , 5–6 hpi , 11–12 hpi ( prior to the onset of DNA replication ) , 24–25 hpi ( first infectious virus particles starting to be released ) and at 47–48 hpi ( late stage infection ) . Following DNase digest and purification of newly transcribed RNA , this was subjected to qRT-PCR analysis for ie1 as well as two genes well characterized to be expressed with either early ( m152 and m169 ) or late ( m129/131 and M94 ) kinetics [79]–[81] . For the spliced late gene m129/m131 [82] we designed the qRT-PCR to span exon-exon junctions to further eliminate any residual risk of amplifying viral DNA or transcripts derived from the opposite DNA strand . To our great surprise , we found not only the two early , but also the two late genes to be well expressed during the first few hours of infection peaking at 1–2 hpi followed by a down-regulation at 5–6 hpi ( Figure 5B–F ) . It is important to note that we could not detect any specific signals in any assay when qRT-PCR was carried out using non-reverse-transcribed samples or Act-D-treated samples ( data not shown ) . This demonstrated the complete removal of viral DNA and virion-associated RNA from these samples . Expression of M94 was also observed when strand-specific cDNA synthesis was performed , matching the kinetics shown in Figure 5F ( data not shown ) . In addition , agarose electrophoresis on m129/m131-PCR products confirmed a band of the predicted size , thereby excluding amplification of viral DNA or a transcript expressed from the opposite DNA strand ( data not shown ) . Interestingly , qRT-PCR analysis on total RNA also revealed a transient , less prominent peak in viral transcript levels at 2 hpi for ie1 and m152 and at 4 hpi for m169 and m129/m131 ( Figure S4 ) . For ie1 , Actinomycin-D treatment demonstrated that virion-associated RNA only comprised <5% of total RNA levels at 1–2 hpi ( Figure 5A ) . Similar data were obtained for m152 ( data not shown ) . Therefore , there appears to be at least a fraction of newly synthesized viral transcripts which are rather unstable ( RNA half-life of ∼1 to 4 h ) . A second unexpected finding was the dramatic drop of transcriptional activity of all viral genes starting 3–4 hpi . For ie1 and m169 , transcription dropped >30-fold by 5–6 hpi ( compared to 1–2 hpi ) and then leveled off ( Figure 5B , C ) . In contrast , transcription rates of m152 continued to drop exceeding 500-fold at 47–48 hpi ( Figure 5D ) . Both late genes , i . e . m129/m131 and M94 , showed a substantial increase in synthesis rates with the onset of viral DNA replication , consistent with their kinetic class . To rule out that our observations were simply caused by the high MOI , we repeated the experiment using a low MOI of 0 . 5 ( Figure S5 ) . As observed with high MOI , ie1 transcription had already peaked by 1–2 hpi . In contrast , transcription rates of both the two early and late genes had not peaked and were now peaking at 3–4 hpi ( see Figure S5 ) . Nevertheless , transcription rates of all five genes significantly dropped by 5–6 hpi . These data indicate that transient expression of viral late genes during the first few hours of infection is not an artifact of high MOI . To confirm these observations at the whole transcriptome level , we repeated the experiment described above and subjected newly transcribed RNA samples from 7 time points ( mock , 1–2 , 5–6 , 11–12 , 18–19 , 24–25 and 47–48 hpi ) to next-generation sequencing . In addition , we included samples of total and pre-existing RNA ( mock , 25 hpi and 48 hpi ) . We obtained between 16 and 42 million reads per sample , which were mapped to the mouse transcriptome , mouse genome , MCMV predicted coding sequences and the MCMV genome in the respective order ( for details on read numbers and mapping statistics see Table S4a ) . As expected , introns were substantially over-represented in newly transcribed RNA ( Figure S6 and Table S4a ) reflecting the substantially greater contribution of immature , unspliced nascent transcripts [41] , [43] . When considering only coding sequences , viral transcripts accounted for ∼20% of all reads in total RNA at 48 hpi and in newly transcribed RNA at 47–48 hpi ( Figure 5G ) . Interestingly , the extent of viral gene expression in newly transcribed RNA at 1–2 hpi also accounted for ∼15% of all reads , dropping to around 5% of all reads at 5–6 hpi . This corroborates our qRT-PCR finding of a burst of viral gene expression at 1–2 hpi , but also highlights that not all viral genes expressed at 1–2 hpi are subject to the same massive down-regulation we observed for m152 . A closer look at the distribution of transcription rates across the whole viral genome ( direct and complementary strand ) revealed viral gene expression arising from multiple loci at 1–2 hpi ( Figure 6 ) . By 5–6 hpi , transcription rates of many , but not all , viral genes dropped substantially ( for details see Table S4b ) . With the onset of viral DNA replication , late gene expression was initiated , accounting for the increasing number of viral reads at 24–25 and 47–48 hpi ( although of shifted viral gene subsets compared to 1–2 hpi ) . At late stages of infection , transcription rates of viral genes stabilized , reflected by the continuous accumulation of the respective viral transcripts in both total and unlabeled pre-existing RNA ( Figure S7 ) .
In this work , we first employed 4sU-tagging combined with microarray analysis to study the dynamic changes in transcriptional activity of NIH-3T3 fibroblasts during early MCMV infection . Interestingly , we found virtually all changes in total cellular RNA to be matched by concordant changes in newly transcribed RNA , indicating that alterations in RNA decay rates do not substantially contribute to differential gene expression during early phase of MCMV infection of fibroblasts . This is consistent with previous reports showing that alterations in RNA decay rates do not seem to provide a major contribution during the first 3 h of the response of fibroblasts to type I and II interferons [34] or of dendritic cells to lipopolysaccharides [41] . It is important to note that the high MOI we employed facilitated the initiation of a fast , contemporaneous infection , crucial to dissect the temporal cascade of rapid transcriptional changes during lytic MCMV infection . This approach revealed extensive regulation , which remained undetectable in total cellular RNA . These elements of the host response to lytic infection are of particular interest because they are likely to be subjected to rapid viral counter-regulation . Analysis of newly transcribed RNA combined with the use of UV-inactivated virus detailed such rapid viral counter-regulation for inflammatory- , interferon- , DNA-damage- and ER-stress-induced changes . While numerous MCMV proteins have been shown to counteract the consequences of the induced ER-stress response , i . e . the induction of stress-induced natural killer cell activating ligands ( reviewed in [83] ) , little is known about the counter-regulation of ER-stress signaling itself . The same is true for the very transient DNA-damage response we observed . The rapid counter-regulation in transcriptional activity revealed by newly transcribed RNA implies the existence of novel viral factors targeting these important cellular processes . Furthermore , our approach will now allow screening of large deletion mutants for the responsible viral genes . Analysis of newly transcribed RNA revealed delayed down-regulation of genes involved in chromatin modification as well as down-regulation of genes involved in cell proliferation and actin filament-based processes . Within a few hours of infection , MCMV-infected cells show a profound cytopathic effect . The underlying molecular events remain to be elucidated . It is tempting to speculate that transcriptional down-regulation of genes involved in actin filament-based processes and cell adhesion , which we found to be consistently down-regulated as early as 1–2 hpi , contributes to this phenomenon . As we exemplified for both NF-κB and c-Myc , changes in transcriptional activity ( as detectable in newly transcribed RNA ) directly mirror the changes in the activation status of the involved transcription factors . In addition , the ability to group the large number of differentially regulated genes ( as usually identified when analyzing total RNA changes ) into well-defined functional clusters of genes ( regulated with distinct kinetics ) further aids the subsequent success of in silico promoter analyses . Our approach thus provides an ideal mean to obtain insights into the molecular mechanisms/transcription factors involved . It is important to note that most of the transcription factors , implicated by our promoter analysis , have already been associated with the functional annotations of the respective gene clusters . In addition , many of the transcription factors specifically over-represented in the clusters ( see Table S2b ) were consistent with published data on their functional role in lytic CMV infection . Of interest , E2F-sites were significantly over-represented in genes repressed with delayed kinetics ( Cluster 5 ) . The E2F family consists of both activating and repressing transcription factors ( for review see [84] , [85] ) . The activating E2F family members , E2F-1 , -2 and -3 , are important for the transactivation of target genes involved in G1/S transition and apoptosis . E2F-4 and -5 predominantly have repressive functions , mediating cell-cycle exit and cell differentiation . E2F-6 , -7 and -8 also act as transcriptional repressors , but are less well characterized . A caveat of this is that the DNA-binding sites of the E2F-family members cannot be differentiated from each other by bioinformatics means , highlighting the complexity of this regulation . Therefore , the repression of genes in Cluster 5 could either be mediated by repression of activating E2F-family members ( E2F-1 , -2 and -3 ) or activation of repressive family members like E2F-4 and -5 . Interestingly , the most significantly associated gene ontology for Cluster 5 was not ‘G1/S-phase of the mitotic cell cycle’ , as would have been expected in case of E2F-1-associated regulation , but ‘M-phase of the mitotic cell cycle’ . Recently , the LIN complex ( LINC ) , which involves the repressive E2F-4 family member , was shown to selectively repress genes involved in G2/M phase [86] . A closer look at the genes of Cluster 5 revealed the presence of numerous genes reported by this and other work [87] to be key marker-genes repressed by the LINC complex ( which involves E2F-4 ) . These included Survivin ( Birc5 ) , Cyclin B1 , Aurkb , Espl1 and Bub1 . It is thus tempting to speculate that the repression of genes in Cluster 5 is mediated not by repression of activating E2F-family members , but by activation of repressive E2F-family members ( involving LINC ) . Ongoing work seeks to clarify the role of E2F-family members in this regulation . Lytic HCMV infection has been shown to result in E2F-1 activation by rapid degradation of the under-phosphorylated form of pRB by the HCMV protein pp71 [88] , [89] and to increase E2F-1 responsive genes [64] . Of note , we did not observe over-representation of E2F sites in any of the MCMV-induced gene clusters . While HCMV can only induce lytic infection in G0/G1 phase , MCMV can also efficiently replicate in cells that have passed through S phase by arresting them in G2 [90] . Differences in shifting cell populations may thus account for the lack of E2F-1 over-representation in our MCMV-induced gene clusters . 4sU-tagging provides the unique opportunity to study the regulation of viral gene expression in real-time without the interference of virion-associated RNAs . Employing 4sU-tagging combined with qRT-PCR and RNA-seq to lytic MCMV infection , we report on three surprising findings . 1 ) A peak of viral gene expression including expression of immediate-early , early and even well-characterized late genes at 1–2 hpi at both high and low MOI . 2 ) The rapid suppression of all three classes of viral gene expression by 5–6 hpi . 3 ) Very constant levels of viral gene transcription rates ( e . g . for ie1 and m169 ) or even continuously increasing suppression ( e . g . m152 ) , despite the onset of extensive viral DNA replication . Both qRT-PCR and RNA-seq on newly transcribed RNA revealed a peak of transcriptional activity of viral genes at 1–2 h of lytic MCMV infection , to an extent only reached again at late stages of infection ( 47–48 hpi ) . We were surprised to observe that this even included transcription of well-characterized late genes , e . g . m129–131 and M94 . Rigorous controls excluded DNA contamination , virion-associated RNA and gene expression from the opposing DNA strand . Furthermore , this transcriptional activity was also observed at low MOI . In this case , however , the peak of both early ( m152 and m169 ) and late ( M94 and m129/131 ) gene expression slightly shifted from 1–2 hpi to 3–4 hpi , thus separating this from the peak of immediate-early gene expression , which still had already peaked at 1–2 hpi . Although even our low MOI infection will still have resulted in multiple virus particles entering the cells , this shift strongly argues against ‘leaky’ promoters being responsible for this late gene expression . It rather highlights a role of virus-mediated regulation ( most likely mediated by IE- or viral tegument-proteins ) in facilitating this part of viral gene expression . It is important to note that this early burst in ‘late’ transcripts at 1–2 hpi does not generate defective or partially processed transcripts . Transcripts were both poly-adenylated ( cDNA synthesis for qRT-PCR was performed with oligo-dT primers ) and fully spliced ( m129-m131 ) . Nevertheless , it is still unclear whether the respective transcripts are actually translated . Recent reports indicate regulation occurring at the level of translation to provide a major contribution to overall regulation of mammalian gene expression [42] . It is tempting to speculate that the virus uses additional , so far undisclosed mechanisms , to regulate its gene expression at post-transcriptional level . Of note , previous studies already postulated a role of post-transcriptional regulation in HCMV [91] and expression of some HCMV genes has already been shown to be regulated at post-transcriptional level [92] , [93] . Interestingly , the peak of viral gene expression we observed at 1–2 hpi was followed by a rapid drop in transcription rates by 6–12 hpi . We were surprised to see this rapid down-regulation of early gene expression . Interestingly , this was sustained or even exaggerated at late times of infection . What is the cause of this down-regulation ? Chromatin modifications of the viral genome are well known to play an important role during productive CMV infection ( reviewed in [94] ) . The ND10 body-associated protein Daxx is known to rapidly repress transcription of incoming viral genomes by inducing repressive chromatin modifications around the HCMV major immediate-early promoter ( MIEP ) within 3 hpi . Dependent on the MOI of infection , the virus is able to overcome this repression . This results in the inhibition/reversion of HDAC-mediated repression of the viral genomes present in the nucleus and initiates the expression of early genes [21] , [95] . It is important to note that the peak of early gene expression should thus only occur after ND10 body-mediated repression has already been efficiently disrupted . Our data is fully consistent with this hypothesis . While we observed both immediate-early and early gene expression to peak or already have peaked at 1–2 hpi following high MOI ( consistent with rapid disruption of ND-10 bodies ) , low MOI resulted in a visible delay between the peak of immediate-early and early gene expression . This is consistent with a delayed dispersion of ND-10 body-mediated repression of early gene expression at low MOI . Of note , we observed repression of viral gene expression after the peak of early gene expression , i . e . after ND-10 bodies have already been dispersed . Therefore , this suppression , which we observed by both qRT-PCR and RNA-seq to occur in between 6 and 12 hpi , is unlikely to be due to the intrinsic antiviral defense known to be mediated by ND-10 bodies . Interestingly , transcription of some genes , e . g . ie1 and m169 substantially dropped and then continued at low level despite the onset of viral DNA replication ( i . e . a rapid >100-fold increase in viral DNA load ) . The most likely explanation for this observation is that the DNA architecture of de novo synthesized viral genomes does not support transcription of ( at least ) some viral genes . On the other hand , it is tempting to speculate that these differences in chromatin structure play an important role in initiating viral late gene expression . Interestingly , transcription rates of some viral genes , e . g . m152 , continued to drop ( exceeding 500-fold compared to 1–2 hpi ) until 47–48 hpi . These observations highlight the importance of transcription factors in the regulation of viral gene expression . It is tempting to speculate that the early burst of transcriptional activity is due to activation of specific cellular transcription factors , which are transiently activated following virus entry but can only initiate viral gene expression ( but for the IE genes ) once ND-10 body-mediated repression has been overcome . Most likely , their subsequent de-activation by 5–6 hpi is responsible for the subsequent drop in transcription of viral genes we observed . It may result from cellular or viral mechanisms . With the onset of viral DNA replication these or other transcription factors are again activated culminating in viral late gene expression . Activation and repression of these transcription factors will also influence the expression of the cellular genes they govern . The kinetics of viral gene expression during the first few hours of infection best matches those of cellular genes within the Clusters 1 and 2 . This is consistent with the well described presence of binding sites for NF-κB and even interferon stimulated response elements ( ISRE ) in many viral promoters [96] , [97] . The changes in transcription factor activity following the advent of viral DNA replication are less well understood . 4sU-tagging now allows us to properly study the changes in cellular gene expression following the onset of viral DNA replication . Correlating these with the changes in viral gene expression will substantially enhance our understanding of how these viruses modulate the host-cell machinery for their own needs and pinpoint novel targets for antiviral intervention .
Murine NIH-3T3 fibroblasts were cultured in DMEM ( Gibco ) supplemented with 5% fetal calf serum . Cells were seeded overnight to 80% confluence followed by infection with BAC-derived MCMV Smith strain . Infection was performed at an MOI of 10 using centrifugal enhancement ( 30 min , 2000 rpm ) or an MOI of 0 . 5 . The time point after centrifugation was marked as time point ‘0 min’ in all experiments . To block RNA polymerase II transcription , Actinomycin-D ( Sigma ) was used at a final concentration of 5 µg/ml . UV irradiation of virus stocks was performed with 1500 J/m2 UV light using a UV-Crosslinker ( Vilber Loumart ) . Standard plaque assays were performed as described [98] to analyze the influence of 4sU on productive virus infection and to confirm the efficiency of UV-inactivation . RNA labeling was started by adding 200 µM 4-thiouridine ( 4sU , Sigma ) to cell culture media for 1 h at different times of infection . At the end of labeling , total cellular RNA was isolated using Trizol reagent ( Invitrogen ) . Biotinylation and purification of 4sU-tagged RNA ( newly transcribed RNA ) as well as dot blot analysis were performed as described previously [34] . For all samples subjected to qRT-PCR analysis , DNase I ( Fermentas ) treatment was performed on total RNA according to the manufacturer's instructions before biotinylation . RNA was recovered using the RNeasy Mini Kit ( Qiagen ) . Reverse transcription was carried out in 20 µl reactions using Superscript III ( Invitrogen ) and Oligo-dT primers ( Invitrogen ) following the manufacturer's instruction . Samples were diluted 1∶5 with H2O before performing qRT-PCRs on a Light Cycler ( Roche Molecular Biochemicals ) as described in Dölken et al . [34] . Relative quantification was performed in relation to uninfected controls normalized to the housekeeping gene Lbr ( Lamin B receptor ) . Primers were designed using the online Roche Universal Probe Library primer design tool spanning exon-exon junctions . All primers are listed in Table S3 . For the microarray analysis , 200 ng RNA of each sample was amplified and labeled using the Affymetrix Whole-Transcript ( WT ) Sense Target Labeling Protocol without rRNA reduction . Affymetrix GeneChip Mouse Gene 1 . 0 ST arrays were hybridized , washed , stained , and scanned according to the protocol described in WT Sense Target Labeling Assay Manual . Microarray data were assessed for quality and normalized with RMA . All microarray data are available at Gene Expression Omnibus ( GEO ) record GSE35919 . Data were analyzed using R and Bioconductor [99] . Only „present“ genes ( i . e . expression values greater than 20 in at least 2 out of the total number of arrays for each RNA type ) were included in downstream analysis . In the total RNA data set , n = 9 , 022 genes passed this filter; in the newly transcribed RNA data set , n = 9 , 399 genes passed . To better compare fold-changes between the total and newly transcribed data , the set union of genes ( n = 10 , 071 ) was used . Differentially expressed genes were identified separately for the total RNA and newly transcribed RNA data sets using the LIMMA package . Differential expression was defined as having an estimated fold-change of at least 2 ( calculated as the fold-change of the average expression in the triplicate measurements after infection compared to uninfected status ) and a p-value smaller than 0 . 05 ( adjusted for multiple testing using the Benjamini and Hochberg method [100] ) . Newly transcribed transcripts , which showed differential expression in at least one condition , were grouped into clusters based on their fold-changes in newly transcribed RNA upon MCMV infection . Five clusters were defined ( for gene lists see Table S1c ) . Briefly , cluster 1 contains rapidly induced genes ( >2-fold at 1–2 hpi ) which are >2-fold counter-regulated by 3–4 hpi . Cluster 2 comprises genes induced with slightly delayed kinetics ( <2-fold induction at 1–2 hpi , but >2-fold at 3–4 hpi ) showing >2-fold counter-regulation by 5–6 hpi . Cluster 3 comprised genes which were not induced at 1–2 hpi ( <1 . 41-fold induced = 20 . 5 ) but >2-fold induced by 5–6 hpi and not counter-regulated >2-fold by 5–6 hpi . Consistently down-regulated genes ( >2-fold at 1–2 , 3–4 and 5–6 hpi , with <1 . 41-fold variation in the extent of repression in between 1–2 and 5–6 hpi ) define cluster 4 . Finally , Cluster 5 comprises all genes not repressed >1 . 41-fold at 1–2 hpi but repressed >2 . 82-fold ( 21 . 5 ) at 5–6 hpi . It is important to note that the criteria and cut-offs for these cluster were chosen empirically trying to keep the criteria simple . This resulted in cluster containing 50–100 genes thus providing solid templates for the down-stream bioinformatic analyses . Enrichment analysis of Gene Ontology ‘Biological Process’ terms and KEGG pathways was performed for each cluster using the DAVID bioinformatics analysis suite ( http://david . abcc . ncifcrf . gov/; release 6 . 7 ) [101] . For all probe sets with a mapped EnsEMBL ID , the core promoter sequence ( −500 to +100 bp relative to the transcriptional start site ( TSS ) ) was retrieved using the Regulatory Sequence Analysis Tools ( RSAT; http://rsat . ulb . ac . be/ ) retrieve EnsEMBL seq function . In case of alternative transcripts the most 5′ TSS was chosen . Over-represented TFBSs for each cluster were predicted using Transcription Factor Matrix ( TFM ) Explorer ( http://bioinfo . lifl . fr/TFM/TFME/; release 2 . 0 ) [102] . Weight matrices modeling putative TFBSs were taken from TRANSFAC ( version 6 . 0 public; vertebrate matrices only ) . P-value thresholds to define locally over-represented TFBSs were set to 0 . 0001 and 0 . 00001 for clusters of less or more than 100 genes , respectively . For immunoblotting , NIH-3T3 infected with MCMV were harvested in 500 µl cell lysis buffer ( 62 . 5 mM Tris , 2% SDS , 10% glycerol , 6 M urea , 5% β-mercaptoethanol , 0 . 01% bromophenol blue , 0 . 01% phenol red ) at several time points of infection . Following heat denaturation ( 95°C , 5 min ) , 50 µl of the lysates were subjected to SDS-PAGE . Proteins were transferred to a nitrocellulose membrane ( Schleicher & Schuell ) using a semidry blotter ( Peq-Lab ) ( 2 h , 400 mA ) . Membranes were blocked in PBS +2 . 5% milk powder and subsequently incubated with primary antibody ( o/n , 4°C ) . After incubation with horseradish peroxidase-conjugated secondary antibody ( 1 h , RT ) the proteins were visualized by the ECL system ( Perkin Elmer ) in the Fusion FX device ( Vilber Lourmat ) . For indirect immunofluorescence NIH-3T3 cells were seeded onto fibronectin-coated glass cover-slips in 24-well plates and infected with MCMV . At various time points of infection , cells were fixed with 4% paraformaldehyde in DPBS ( w/v ) for 10 min at 37°C . The fixative solution was replenished twice with DPBS and the cells were permeabilized for 10 minutes with a solution of 0 . 1% Triton X-100 in DPBS . After extensive washing with DPBS , the cells were blocked using 3% ( w/v ) BSA in DPBS ( blocking solution ) for 1 h at room temperature ( RT ) . Primary antibodies were applied in blocking solution and incubated with the cells at RT for 1 h followed by three DPBS washing steps and 1 h incubation at RT with 1∶1 , 000 dilutions of Alexa Fluor-conjugated , specific secondary antibodies ( Invitrogen ) in blocking solution . After a final washing step with DPBS , the preparations were mounted on glass slides with Prolong Gold including DAPI ( Invitrogen ) and analyzed using an LSM 710 ( Zeiss ) confocal laser scanning microscope with 405 nm , 488 nm and 561 nm laser excitation and appropriate filter sets . The following primary antibodies were used for immunoblotting and -fluorescence: mouse anti-IE1 ( CROMA101; kindly provided by S . Jonijic , University of Rijeka , Rijeka , Croatia ) , rabbit anti-GAPDH ( mAbcam 9484 ) from Abcam , rabbit anti-RelA ( A ) and mouse anti-IκB-α ( H4 ) from Santa Cruz Biotechnology . To monitor the activity of c-Myc-regulated signal transduction pathway , cells were transfected in a 6-well format using the Cignal c-Myc-Reporter ( luc ) Kit from Qiagen . 6 hours post transfection cells were seeded in 96-well plates ( 10 , 000 cells/well ) . 48 hours after seeding , the transfected cells were infected with MCMV ( n = 3 ) . At various times of infection , cells were lysed in 100 µl lysis buffer and luciferase firefly activity was determined according to the manufacturer's ( Promega ) instructions . RNA was subjected to WTAK library construction to generate transcriptomic fragment libraries ( 50 bp , SOLiD ( Life Technologies , Foster City , CA , USA ) Total RNA-seq Kit V3 ) that preserve strandedness information of the reads . Molecular barcoding was used in order to pool several libraries in a single sequencing reaction according to the manunfacturer's protocol . Sequencing was performed using the SOLiD 3 system ( Life Technologies ) . Potential sequencing errors were corrected using the SOLiD Accuracy Enhancement Tool ( solidsoftwaretools . com/gt/project/saet ) . All sequencing data are available at Gene Expression Omnibus ( GEO ) at GSE35833 . Between 26 and 42 million 50 nt reads were obtained per sample ( Table S4a ) . Reads were aligned in a 4-step process using the Bowtie alignment program . First , all reads were aligned to mouse transcripts ( Ensembl version 63 ) . Remaining unaligned reads were aligned to the mouse genome ( mm9 , NCBI Build 37 ) . Remaining reads were aligned to MCMV coding sequences and finally to the full MCMV genomic sequence . Reads with ambiguous base calls , non-unique alignment positions or more than 4 mismatches were discarded . Reads were classified as exon-exon and exon-intron junction reads , respectively , if they overlapped an exon-exon or an exon-intron junction by ≥1 nt ( Table S4a ) . MCMV genome coverage was scaled before plotting ( for Figure 6 and S7 ) . Scaling factors were derived using DESeq based on reads aligned to mouse exons . RPKM values of individual MCMV genes are included in Table S4b ) . | Cytomegaloviruses are large DNA viruses , which establish life-long latent infections , leaving the infected individual at risk of reactivation and disease . Here , we applied 4-thiouridine- ( 4sU ) -tagging of newly transcribed RNA to monitor the real-time kinetics of transcriptional activity of both cellular and viral genes during lytic murine CMV ( MCMV ) infection . We observed a cascade of MCMV-induced signaling events including a rapid inflammatory/interferon-response , a transient DNA-damage-response and a delayed ER-stress-response . All of these were heavily counter-regulated by viral gene expression . Besides dramatically increasing temporal resolution , our approach provides the unique opportunity to study viral transcriptional activity in absence of any interfering virion-associated-RNA . Virion-associated-RNA consists of transcripts that are unspecifically incorporated into the virus particles thereby resembling the cellular RNA profile of late stage infection . A clear picture of which viral genes are expressed , particularly at very early times of infection , could thus not be obtained . By overcoming this problem , we provide intriguing insights into the regulation of viral gene expression , namely 1 ) a peak of viral gene expression during the first two hours of infection including the expression of well-characterized late genes and 2 ) remarkably constant or even continuously declining expression of some viral genes despite the onset of rapid viral DNA replication . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"infectious",
"diseases",
"infectious",
"disease",
"modeling",
"viral",
"diseases"
] | 2012 | Real-time Transcriptional Profiling of Cellular and Viral Gene Expression during Lytic Cytomegalovirus Infection |
The rapid evolution of RNA-encoded viruses such as HIV presents a major barrier to infectious disease control using conventional pharmaceuticals and vaccines . Previously , it was proposed that defective interfering particles could be developed to indefinitely control the HIV/AIDS pandemic; in individual patients , these engineered molecular parasites were further predicted to be refractory to HIV’s mutational escape ( i . e . , be ‘resistance-proof’ ) . However , an outstanding question has been whether these engineered interfering particles—termed Therapeutic Interfering Particles ( TIPs ) —would remain resistance-proof at the population-scale , where TIP-resistant HIV mutants may transmit more efficiently by reaching higher viral loads in the TIP-treated subpopulation . Here , we develop a multi-scale model to test whether TIPs will maintain indefinite control of HIV at the population-scale , as HIV ( ‘unilaterally’ ) evolves toward TIP resistance by limiting the production of viral proteins available for TIPs to parasitize . Model results capture the existence of two intrinsic evolutionary tradeoffs that collectively prevent the spread of TIP-resistant HIV mutants in a population . First , despite their increased transmission rates in TIP-treated sub-populations , unilateral TIP-resistant mutants are shown to have reduced transmission rates in TIP-untreated sub-populations . Second , these TIP-resistant mutants are shown to have reduced growth rates ( i . e . , replicative fitness ) in both TIP-treated and TIP-untreated individuals . As a result of these tradeoffs , the model finds that TIP-susceptible HIV strains continually outcompete TIP-resistant HIV mutants at both patient and population scales when TIPs are engineered to express >3-fold more genomic RNA than HIV expresses . Thus , the results provide design constraints for engineering population-scale therapies that may be refractory to the acquisition of antiviral resistance .
Defective interfering particles ( DIPs ) are ‘cheaters’ in a viral population . Rather than carrying a full set of genes essential for their own replication , these deletion mutants require co-infection by replication-competent ‘helper’ viruses to provide their missing components for replication , packaging , and spread [1 , 2] . By stealing essential viral components from wild-type viruses , DIPs act as molecular parasites of viruses . Further , natural DIPs have been observed to arise spontaneously across a range of viruses , and have been predicted to reduce disease virulence by interfering with viral replication processes [3–9] . Consequently , DIPs have been proposed as novel antiviral therapeutics [5 , 10–12] . While natural DIPs have never been documented in HIV infections , HIV-derived DIPs have been engineered artificially [10 , 13–15] and shown to reduce HIV replication [10 , 16 , 17] . Here we quantitatively probe a subset of DIPs that are engineered to have basic reproductive ratios greater than 1 during co-infection with HIV ( i . e . , maintain stable TIP loads ) while suppressing HIV viral loads . These stable and suppressive DIPs are termed therapeutic interfering particles ( TIPs ) . Previous mathematical models predicted that TIPs would substantially outperform current state-of-the-art antiretroviral therapy campaigns [12 , 18–21] . However , the models did not test whether TIP efficacy would be undermined at the population-scale by the evolution and spread of TIP-resistant HIV mutants . Since TIP intervention is designed to reduce wild-type HIV viral loads within individual patients , it may pressure HIV to evolve . Specifically , reductions in HIV load correlate with reduced transmission of HIV [22 , 23] , so any TIP-mediated reductions would create a selection pressure for ‘resistant’ HIV mutants that are not suppressed . Arguably , the most direct way for HIV to evolve TIP-resistance is by reducing the amount of intracellular resource ( e . g . , capsid proteins ) available for TIP parasitism . In that way , HIV may be able to ‘starve’ the parasitic TIP particles of the resources they parasitize . TIP-resistant HIV mutants would then have the potential to outcompete the wild-type HIV strains within a patient and spread through a host population , progressively nullifying the TIP intervention . This is a form of ‘unilateral escape’ , evolutionary escape by mutations affecting a feature encoded by the viral genome but not the ( reduced ) TIP genome . Recently , we analyzed the unilateral escape dynamics of HIV resistance at the level of an individual patient and found that HIV mutants that starve TIPs would be selected against within individual patients [19] . However , given their reduced suppression , these TIP-resistant HIV mutants would likely transmit from infected individuals more efficiently than the wild-type HIV [22] . Thus , even if disfavored in individual patients , TIP-resistant HIV mutants could supplant the wild-type HIV strain at the population-scale , undermining the effectiveness of a TIP therapy campaign . We sought to test whether TIP-resistant HIV strains would spread through a population . Notably , the transmission of HIV strains through a population is primarily driven by small ‘core groups’ of infected individuals ( ~1–2% of the population ) who engage in high-risk behaviors [24–26] . TIPs similarly concentrate within these high-risk groups , because TIPs spread via the same transmission routes and risk factors as HIV [18] . Further , the increased prevalence of TIPs within high-risk groups increases the selective pressure in favor of TIP resistance in these sub-populations . And even if TIP-resistance initially emerges in a disparate sub-population , transit through high-risk sub-populations is critical for the population-scale spread of TIP-resistant HIV strains . Consequently , we developed a mathematical model to quantify whether HIV mutants with increased TIP resistance could stably invade high-risk populations . Strikingly , model results show that as long as TIP genomes are initially engineered to express at least ~3-fold more genomic RNA transcripts than HIV expresses in co-infected cells , TIPs can generally maintain population-scale stability . Further , due to two intrinsic evolutionary tradeoffs , TIPs are shown to be evolutionary stable at the population-scale whenever they are evolutionarily stable at the patient-scale .
The model of HIV and TIP replication and transmission includes three levels of biological organization: a population-scale model , an individual patient-scale model , and a cellular-scale model ( Fig 1A ) . Each scale is represented by a well-studied system of deterministic ordinary differential equations . The population-scale model is an epidemiological Susceptible-Infected ( SI ) model [27] extended to include the spread of TIPs [18] ( Section C in S1 Text ) . The patient-scale model is a variant of the basic model of HIV dynamics [28 , 29] , again extended to account for the presence of TIPs ( Section B in S1 Text ) . Finally , the cellular scale model is a ‘public-goods game’—a well-studied system in game theory and evolutionary biology [30 , 31]—in which HIV and TIP sequences compete for HIV capsid elements within dually-infected cells ( Section A in S1 Text ) . Each scale’s equations are shown below , with all model parameters derived in the Supporting Information and summarized in Tables 1 and 2 . At the single-cell scale , TIPs can only replicate and package by parasitizing essential trans-acting elements from full-length HIV in co-infected cells ( Fig 1A , bottom left ) . In the absence of HIV , TIPs can only enter CD4+ T cells and integrate their genetic material into the cellular genomes—little to no TIP production occurs , since TIPs are engineered to lack essential trans-acting elements required for lentiviral replication and packaging . As a result , TIP genomic RNAs ( gRNAs ) only express after a cell is co-infected by HIV , at which point TIP gRNAs compete with HIV gRNAs for encapsidation by HIV capsid proteins . HIV capsids are thus intracellular ‘public goods’ that both HIV and TIP gRNAs utilize , as shown in the following equations: dGHIVdt=θ︸production−kpckGHIVC︸packaging−αGHIV︸decay dGTIPdt=mPθ︸production−kpckGTIPC︸packaging−αGTIP︸decay dCdt=ηθ︸production−kpck ( GHIV+GTIP ) C︸packaging−βC︸decay The three state variables represent the within-cell concentrations of HIV gRNA ( GHIV ) , TIP gRNA ( GTIP ) , and capsid proteins ( C ) , respectively . As shown in the Supporting Information ( Section A in S1 Text ) , the rate parameters kpck , α , and β can be grouped into a single composite waste parameter , κ , by non-dimensionalizing the model . The parameter m quantifies the number of TIP integrations , and results naturally from multiple TIP infections of a cell in the host scale model ( below ) . The parameter θ is HIV’s gRNA production rate—θ serves to scale the size of all outputs , and is therefore absorbed into the host-scale parameters ( Section B in S1 Text ) . Most importantly , η reflects the ratio of HIV-capsid to HIV-genome production and P corresponds to the ratio of TIP gRNA expression to HIV gRNA expression ( i . e . , the expression asymmetry ) . From previous studies , η and P are known to be two major parameters that determine TIP evolutionary stability within patients [18 , 19] . Thus , we tracked TIP stability at the population scale as a function of η and P . For each value of ( η , P ) , the single-cell scale equations are solved to determine HIV and TIP ‘burst sizes’ ( i . e . , the net numbers of HIV and TIP virions produced in the lifespan of an infected cell ) ( Eqs . 9–11 in S1 Text ) . The burst sizes are passed to the patient-scale equations , where they are used to calculate patient-wide viral set points , through the parameters n , ψm , and ρm ( below ) . At the patient scale , the model incorporates four fundamental asymmetries to determine TIP and HIV viral loads ( Fig 1A , bottom right ) . ( i ) First , while TIPs require the presence of HIV to replicate , HIV can replicate in the absence of TIPs . ( ii ) Second , HIV gene expression ( of the vpr gene ) prevents the division of infected cells [32 , 33] . Conversely , TIPs lack vpr and TIP infection is silent ( absent HIV ) , so TIPs do not block cell division . ( iii ) Third , TIP-infected cells live as long as uninfected cells in the absence of HIV ( due to this replicative silence ) ; HIV infection results in rapid cell death [34 , 35] . ( iv ) Fourth , HIV gene expression suppresses subsequent superinfection of a cell via the nef gene [33 , 36]; TIPs lack nef and do not suppress superinfection . Thus , multiple copies of the TIP provirus can integrate into a cellular genome prior to HIV infecting that cell . These TIP infected cells can further divide to seed a reservoir of cells with a range of TIP proviruses . The resulting numbers of TIP and HIV particles in a patient are tracked in the following host-scale model: dT0dt=b︸production+dhT0︸division−dT0︸death−kVHT0︸HIV infection−kVTT0︸TIP infection dTmdt=dhTm︸division−dTm︸death−kVHTm︸HIV infection+kVTTm−1−kVTTm︸TIP infection , m≥1 dImdt=kVHTm︸HIV infection−δIm︸death , m≥0 dVHdt=nδI0+nδ∑m=1∞ψmIm︸HIV production−cVH︸clearance dVTdt=nδ∑m=1∞ρmψmIm︸TIP production−cVT︸clearance The patient-scale state variables quantify: the numbers of HIV-uninfected cells with m TIP integrations ( Tm , m ≥ 0 ) , the numbers of HIV-infected cells with m TIP integrations ( Im , m ≥ 0 ) , the number of HIV virions ( VH ) , and the number of TIP virions ( VT ) . All host-scale parameters are described in the Supporting Information ( Section B in S1 Text ) , where non-dimensionalization is shown to reduce the number of parameters to four: R0 , d/δ , c/δ , and h ( Table 2 ) . Notably , HIV viral loads are lower in TIP-treated individuals than in HIV-only infected individuals , because cells co-infected with both HIV and TIP produce fewer HIV virions than HIV-only infected cells . The steady-state TIP ( VT ) and HIV viral loads ( VH ) resulting from this model are passed to a population-scale model to calculate virulence and spread across a population , as in [22] . The population-scale model ( Eqs . 1 and Eqs . 59–61 in Section C in S1 Text ) is a standard SI model with a single , well-mixed population , corresponding to high-risk disease spreaders [26] . As is common , a constant influx of susceptible individuals is assumed . These susceptible individuals are converted into HIV-infected individuals upon contact with an HIV-infected patient at a rate dependent on HIV’s viral load in the infected ‘donor’ patient [22] . While HIV can directly infect susceptible individuals , TIPs are ( conservatively ) assumed to only infect patients already infected with HIV . Further , co-transmission of HIV and TIP is neglected , due to the evidence showing that only a single founder virus generally establishes in patients after transmission through mucosal bottlenecks [37 , 38] . Based on epidemiological and patient data [22] , HIV infection progresses to AIDS as a function of HIV’s viral load in the patient , which reduces the effective lifetime of an infected individual and is modeled as removal from the population . Superinfection with TIP slows progression to AIDS ( and reduces transmission of HIV from that individual ) by reducing the HIV viral load ( Fig 1A , top panel ) . The equations describing these epidemiological processes are: dSdt=λ︸input−cNβHISI−cNβHIDSID︸HIV infection−δSS︸death dIdt=cNβHISI+cNβHIDSID︸HIV infection−cNβTIDIID︸TIP infection−δII︸death dIDdt=cNβTIDIID︸TIP infection−δDID︸death The state variables represent the prevalences of: individuals susceptible to HIV ( S ) , HIV-infected individuals ( I ) , and dually infected ( HIV+TIP+ ) individuals ( ID ) . As shown in the Supporting Information ( Section C in S1 Text ) , non-dimensionalization reduces the population-scale system to: dSdt=δI ( 1B ( 1−S ) −R0pop ( SI+μSID ) ) dIdt=δI ( R0pop ( SI+μSID−ϕIID ) −I ) dIDdt=δI ( R0popϕIID−τID ) ( 1 ) The model parameters are defined as follows: δI is the rate of removal of HIV-infected individuals from the population ( i . e . AIDS progression rate ) ; B is the ratio of the removal rates of infected and uninfected individuals; R0pop is the basic reproductive ratio of HIV in a population; μ and ϕ are the respective HIV and TIP transmission rates from TIP-treated individuals relative to the HIV transmission rate from individuals infected with HIV alone; τ is the decrease in the AIDS-progression rate due to superinfection with TIP . As in [22] , these parameters are directly calculated from the HIV and TIP viral loads in the individual-patient model ( Eqs . 62–69 in S1 Text and Table 1 ) . Following a well-established approach [19 , 39 , 40] , the three biological scales ( the single cell , the individual patient , and the host population ) are integrated into a single multi-scale model , by using the steady-state outputs from the lower scale models as inputs into the higher scale models . This separation of timescales approach is possible , because the timescales of the processes that occur on each scale are so disparate that the processes on the lower scales are approximately at steady state relative to those on the higher scales . More specifically , the cell-scale processes reach steady-state in hours , the patient-scale processes require days to months , and the population-scale processes play out over decades . Using this separation of timescales approach , the multi-scale model is simulated by setting values for η and P within single-cells . Inputting these values into the cell-scale model outputs HIV and TIP viral burst sizes . Using these burst sizes as inputs , the patient-scale model outputs set-point viral loads for HIV and TIP ( Section B in S1 Text ) . Finally , the viral set-points are used as inputs into the population-scale model to calculate the parameters in Eq ( 1 ) , specifically: the progression time of infected individuals to AIDS ( τ ) and the relative transmission rates of HIV ( μ ) and TIP ( ϕ ) ( see [22 , 23] and Section C in S1 Text ) . Thus , the final output of the multi-scale model is the prevalence of HIV and TIP across a population as a function of the intracellular design parameters η and P . Once we can calculate HIV and TIP viral loads and prevalence levels as functions of η and P , we can then map the regions of the ( η , P ) parameter plane in which HIV mutants are able to maintain higher steady-state levels in the presence of TIP than is the wild-type HIV strain . We term these HIV mutants to be ‘resistant’ mutants . Given that the model covers multiple scales of behavior , there are multiple scales at which resistance can arise . At the host scale , HIV resistance corresponds to increased viral loads . Thus , HIV mutants that are able to maintain higher viral loads than wild-type HIV in the presence of TIP are termed ‘viral-load increasing’ resistant mutants . At the population scale , HIV resistance corresponds to an increased prevalence of unsuppressed ( HIV+TIP- ) individuals in the population . Thus , HIV mutants that are able to maintain a higher prevalence of HIV+TIP- hosts in the presence of TIP are termed ‘prevalence-increasing’ resistant mutants . Notably , at both host and population-scales , whether or not an HIV mutant is TIP-resistant is defined relative to the wild-type HIV strain—resistance is therefore dependent upon the parameter ( η ) values of both the wild-type and mutant HIV strains . After finding the parameter values that generate TIP-resistant viral phenotypes , the next step is to determine whether or not these resistant mutants can invade established populations of HIV and TIP to overcome a TIP intervention campaign . This invasion analysis is performed by taking the dominant eigenvalue of the Jacobian matrix of the system ( Section D in S1 Text Eqs . 108–109 ) , as is standard [41–43] . Behaviors at the host scale are independent of the population scale , so invasion within hosts can be solved agnostically of the population scale . On the other hand , whether or not a mutant can invade the population is dependent upon its behavior at the individual-host scale . To resolve this , we extend the model to allow for ‘host stealing’: take-over of hosts by the strain better able to propagate at the host scale . Because the time-scales differ greatly between the host and population scales , this host stealing is assumed to occur instantaneously from the perspective of the population . This ‘super-infection’ assumption is standard in multi-scale modeling studies in epidemiology [44] . Standard invasion analysis is then performed on the multi-scale model: a ( TIP ) treatment is termed ‘evolutionarily stable’ if no TIP-resistant HIV mutants are able to invade populations infected by wild-type HIV .
Using the multi-scale model described above , we first sought to determine the intracellular parameter regimes that result in TIP invasion and dynamic stability: i . e . , the requirements for a small amount of TIP to invade a patient or population ( R0TIP > 1 ) and reach a stable nonzero steady-state . At the patient-scale , model simulations show that HIV can only achieve a nonzero steady-state when η > ~0 . 1 , matching earlier predictions [19] . When η < ~0 . 1 , HIV replicates so poorly within an individual that neither HIV nor TIP can propagate ( Fig 1B , left and S1 Fig ) —resulting in dual HIV and TIP extinction . When η ≈ 0 . 1 , HIV achieves stability , and there is a small slice of the ( η , P ) parameter plane at which HIV is stable but TIP is not ( Fig 1B , left ) . Everywhere else in the ( η , P ) parameter plane , TIP co-stability with HIV can be achieved with a sufficient level of TIP gRNA overexpression ( P ) . In particular , when P > Pcritical ≈ 3 , TIPs are stable in the host across all η > ~0 . 1 and even remain stable as the TIP instability regime expands at η ≈ 1 . Thus , engineering TIPs to express greater than ~3X more gRNA than HIV expresses is an essential design constraint at the patient-scale . At the population-scale , P >~ 3 again generates TIP stability across a broad range of the ( η , P ) plane ( Fig 1B , right ) . However , TIP stability also depends on the initial HIV prevalence in the high-risk population ( in which HIV is highly prevalent ) and on whether TIPs can ‘pre-immunize’ HIV-negative individuals ( Fig 1C ) . For the maximally conservative assumption of no TIP pre-immunization , TIPs act as obligate secondary parasites and replicate only within HIV-infected individuals , so both the subpopulation available for TIP infection and the frequency of contacts within this subpopulation depend on the prevalence of HIV . Since the initial HIV prevalence depends on HIV’s basic reproduction ratio , R0pop [45] ( see Eqs . S78 ) , the effective reproduction ratio of TIPs , Reff , is also dependent on the reproduction ratio of HIV within the population . The requirement for TIP spread and stability in a population under the conservative assumption of no TIP pre-immunization is ( see Table 1 and Section C in S1 Text ) : Reff ( η , P ) =R0pop−Bτϕ ( η , P ) >1 ( 2 ) Under this assumption , as the initial prevalence of HIV increases , the region of the ( η , P ) plane in which TIPs remain stable in the population expands and approaches the within-host stability region ( Fig 1B ) . In particular , given a high-risk population with ~60% prevalence of HIV , if TIPs are stable in an individual , they stably coexist with HIV in the population . On the other hand , if pre-immunization is allowed and TIPs can ( silently ) pre-infect susceptible individuals and remain latent until subsequent infection with HIV ( as was assumed in previous analyses [18] ) , the TIP stability region expands and is less sensitive to HIV’s initial prevalence ( Fig 1C ) . Importantly , there is some evidence to suggest that pre-immunization is possible: lentiviruses are able to establish latency in rhesus monkeys even when suppressive antiretroviral therapy ( which prevents replication ) is started as little as three days post-infection [46] . Further , this latent state persists for more than six months [46] . In both cases , whether or not pre-immunization occurs , in regions of parameter space where TIPs coexist with HIV , TIPs will substantially suppress HIV/AIDS prevalence and incidence across a population ( S2 Fig ) . Having determined the critical engineering constraint for the dynamic stability of TIP treatments at both host and population scales—i . e . , P > ~3—we next mapped the regions of parameter space at which HIV can achieve resistance to these stable TIPs . Since the model assumes no specific genotype-to-phenotype map , the difference between HIV strains is modeled phenotypically , as a difference in η . Importantly , the parameter η is likely to be under selection in vivo , as HIV has evolved suboptimal splicing ( splicing increases η ) and a molecular switch to control this splicing efficiency [47] . In contrast , the parameter P is assumed to be constant and independent of the HIV parameters ( i . e . , P depends on the TIP alone; see Discussion ) . As defined in the Methods ( above ) , η mutants can generate two types of resistance relative to the wild-type η strain . HIV mutants that generate increased viral loads in an individual are considered TIP-resistant within the host , and are termed ‘viral-load increasing’ mutants . HIV mutants that generate increased prevalence in the population are considered TIP-resistant within the population , and are termed ‘prevalence increasing’ mutants . In TIP+ individuals , HIV viral loads reach a maximum at a critical value of η near η = 1 ( Fig 2A , left ) . Since this maximal HIV viral load depends on the value of P ( i . e . , the particular TIP variant ) , we denote this critical value as ηc ( P ) . Any mutation in η toward ηc ( i . e . any mutant for which |ηmut − ηc| < |ηwt − ηc| ) is sufficient to increase the HIV load in TIP+ hosts ( Fig 2A , left ) . Additionally , in a thin region of low η ≈ 0 . 1 , TIP is destabilized in HIV+ hosts , so HIV loads again peak . So , any mutant with η ≈ 0 . 1 or with a value of η closer to ηc than the wild-type is a virus-load increasing mutant . In contrast , at the population-scale , mutants that reduce η reduce the population-level coverage of TIPs in the HIV-infected population , whatever the wild-type value of η ( Fig 2A , right ) . Thus , any mutant with ηmut < ηwt is a prevalence-increasing mutant . Intuitively , both modes of TIP-resistance result from decreasing η to starve TIPs of the public goods ( e . g . , capsid proteins ) they require . However , when η is decreased below ηc ≈ 1 , decreasing capsid production begins to harm HIV’s own ability to propagate within hosts ( so ηc ≈ 1 is effectively a ‘sweet-spot’ ) . Both viral-load increasing mutants and prevalence-increasing mutants can lead to either full or partial loss of HIV suppression , and , consequently , full or partial elimination of TIPs from the pertinent scale ( Fig 2A and S2 Fig ) . Full resistance at each scale ( individual or population ) essentially drives the system to regions of P and η where TIPs are unstable ( Fig 1B and 1C and S2 Fig ) . However , if P > 3 , full resistance to TIP treatment only occurs at low η values of ~0 . 1 , near the HIV extinction threshold ( Fig 2A ) . After mapping the regions in which HIV escape mutants arise , the next step is to whether or not these resistant mutants can spread across a population to undermine a TIP campaign . To do so , we examine the introduction of mutant HIV strains into the host population and analyze the competition between the wild-type HIV strain and each new mutant HIV strain . HIV mutants are introduced in small quantities into an individual with steady-state TIP and wild-type HIV viral loads . As in the stability analysis above , we rely on a time-scales separation , since a large number of viral replication events ( i . e . , viral generations ) occur within each individual between inter-individual transmission events . Examining the dominant eigenvalues of the Jacobian , we determined the fitness landscape for HIV mutants at the individual-patient level from the rate of expansion or contraction of a mutant strain with slight differences in η relative to the wild-type ( Section D in S1 Text ) . The Jacobian analysis enables us to calculate the net selective advantage ( or disadvantage ) of any TIP-resistant mutants . There are two scales at which evolutionary fitness must be analyzed: the host scale and the population scale . At the host scale , the relative fitness of an HIV variant reflects the relative growth rate of that clone within a host . For any HIV strain , growth rate increases with its effective reproductive ratio [19] , which depends on the viral burst sizes from individual cells and the distribution of TIP multiplicities among the cells . Critically , within a given cell , a larger η always corresponds to a larger viral burst size , regardless of the TIP multiplicity ( S1 Text Eqs . S9-S13 ) . Consequently , HIV mutants with larger values of η always have higher relative fitness within an individual , regardless of the presence or level of TIP ( Fig 2B , left; and S3 Fig ) . This result can be understood intuitively as a ‘tragedy of the commons’ [48 , 49]: enhanced capsid production favors the HIV strain that can achieve it , despite enabling increased TIP parasitism of all HIV mutants in the host . At the population scale , the effective reproductive ratio of HIV is determined by its ability to transmit between members of the host population [22] . The HIV transmission rates are calculated ( Eq 1 , Table 1 ) from the viral load outputs from the individual-patient model as in [18] ( see Section C in S1 Text ) . Since these patient-level viral loads themselves depend on the single-cell parameters , the transmission rates of HIV mutants are ultimately functions of η and P . Further , these transmission rates differ greatly between individuals only infected with HIV alone individuals infected with both HIV and TIP . In the absence of TIP , more resource production ( i . e . , a higher η ) is always better for HIV , so HIV transmission always increases with η in TIP− individuals ( Fig 2B , top right ) . However , in TIP+ individuals , the transmission rate and the viral load both peak at η = ηc ( Fig 2B , bottom right ) . Intuitively , in TIP+ hosts , there is a balance between producing enough resources to propagate , and producing too many resources , which allows TIPs to establish a larger population . Taken together , the model results capture conflicting selection pressures driving HIV transmission from TIP+ individuals , HIV transmission from TIP− individuals , and HIV viral loads within patients . These conflicting pressures push HIV evolution in different directions along the η-axis , resulting in evolutionary conflicts on the value of η . Overall , two evolutionary tradeoffs emerge from the model: an inter-scale conflict in TIP co-infected individuals between host-level HIV fitness and population-level HIV transmission , and an intra-scale ( population-level ) conflict between HIV transmission from individuals co-infected with TIP and individuals not co-infected with TIP ( Fig 2B ) . The inter-scale conflict arises from the fact that when TIP is evolutionarily stable on the host level , evolution within TIP-treated hosts leads to higher η values . Yet , at the population-level ( i . e . , in TIP+ individuals ) , HIV variants with lower η values are evolutionarily beneficial , since they reduce TIP levels and attendant parasitism . The intra-scale conflict also arises from the benefit to HIV of reducing η in TIP+ populations—the tradeoff is that in TIP− populations , reducing η actually reduces HIV loads and transmission . Thus , the intra-scale conflict exists within the population scale alone and is dependent on the frequency of TIP+ hosts ( i . e . , it is a frequency-dependent effect ) . Given these two evolutionary conflicts on the value of η , we probed which HIV mutants could spread through a population by extending the population-level model to include multiple HIV strains ( Fig 3A and S1 Text Eqs . 100–104 ) . For each HIV strain , relative transmission rates were calculated based on viral loads , as in the one-strain model ( Eq 1 , Table 1 ) . When wild-type and mutant HIV strains co-infect the same host , host fitness comes into play at the population scale , since the within-host infection dynamics occur extremely rapidly relative to the population-scale dynamics [50] . As a result , the more fit HIV strain takes over within a host prior to population-scale transmission events . The rapid host take-over is due to ‘competitive exclusion , ’ which precludes two strains from coexisting at steady state , regardless of the presence of TIP ( dashed vertical arrows in Fig 3A ) . Therefore , we neglect individuals co-infected with multiple HIV strains—the most fit HIV strain rapidly excludes the others ( see the Discussion for an analysis of cases where competitive exclusion does not occur , due to weakened within-host selection ) . The dual-strain model was first used to track whether a representative HIV mutant with ηmut = 1 can invade a TIP-treated population with a wild-type HIV strain in which ηwt = 2 ( Fig 3B ) . Given its decreased η value , this HIV mutant would increase the HIV viral load in HIV+TIP+ ( co-infected ) individuals and increase the prevalence of HIV+TIP- individuals in a population ( Fig 2B ) . However , the mutant would be disfavored in TIP− populations and disfavored within individual hosts . The dual-strain model weighs the conflicting selective benefits at both scales to calculate whether the HIV mutant can spread . In the particular example of ηwt = 2 and ηmut = 1 , the dual-strain model demonstrates that the overall population-level trajectory of the resistant HIV mutant is toward extinction—given both large P ( i . e . , P > 3 ) and the presence of co-infection ( Fig 3B and S4 Fig ) . Effectively , P > 3 drowns out the selective advantage of decreasing η: the modest increases in transmission from TIP+ individuals are matched by decreases in transmission from TIP- individuals . In contrast , when P = 2 . 5 , decreasing η results in a major increase in transmission from TIP+ individuals ( Fig 1B ) , which dwarfs the decrease in transmission from TIP- individuals . Thus , increasing P to a value greater than 3 is required for a robust intra-scale conflict . The trajectory of the mutant strain also depends on whether co-infection can occur , since co-infection results in the out-competition of reduced η mutants within hosts despite their population-level fitness advantage . In other words , co-infection is required for an inter-scale conflict . Competing mutant and wild-type HIV strains in the presence and absence of P > 3 and co-infection demonstrates that both intra-scale and inter-scale conflicts are necessary to prevent the establishment of TIP resistance ( Fig 3B ) . To determine the general conditions across all of ( η , P > 3 ) parameter space under which a mutant virus ( with parameter ηmut ) can spread into a wild-type-infected host population ( with parameter ηwt ) , we performed an invasion analysis [41–43] to examine the initial expansion rates of HIV mutants after introduction ( Fig 3C ) . When ηmut < ηwt , the model shows that HIV mutants never expand ( Fig 3C ) . This result is the key to determining parameter regions of P where TIPs would be safe from HIV escape mutants ( i . e . the design criteria for engineering ‘resistance-proof’ TIPs ) . Indeed , if the small HIV-mutant population shrinks initially , it will never be able to outcompete wild-type HIV . Thus , when P is safely in the TIP stability region of P > ~3 , HIV evolution is constrained to move toward larger η values and away from TIP-resistance . In terms of the two types of resistant mutants discussed above , this invasion analysis ( Fig 3C ) shows the extinction of prevalence-increasing mutants anywhere , and the extinction of virus-load increasing mutants when ηwt > ηc . Virus-load increasing mutants do spread in a host and in a population when ηwt < ηc , in which case all the selection pressures align ( S5 Fig ) , because an increase in η results in an HIV load increase . However , in this range ( ηwt < ηc ) , HIV is pressured toward higher η values regardless of the presence of TIP ( Fig 2B ) . This selection pressure arises not from the presence of TIP , but from the enhanced replication of HIV at the host scale at higher capsid production rates ( regardless of TIP ) . At the population level , the selection pressure towards higher η even decreases when TIP is present ( see the transmission rates in Fig 2B , right ) . In other words , the population-level instability comes from a pre-existing host-level instability , prior to the introduction of TIP . Taken together , the results of the invasion analysis show that—given dynamic and evolutionary stability at the host scale ( i . e . , P > 3 and ηwt > ηc ) —TIP interventions would be both dynamically and evolutionarily stable from unilateral HIV escape mutants at the population scale . While the models used to test TIP evolutionary stability at both host and population-scales are well-established [45] , as in any modeling study , our analysis necessarily utilizes simplifying assumptions . To determine whether these simplifying assumptions impacted model outcomes , we performed a number of sensitivity analyses in which model assumptions were relaxed . For example , a concern in the host-scale model is the function used to model target-cell division ( since cell division enables the vertical transmission of provirally integrated TIPs across a host ) . To keep the uninfected T-cell population bounded , we assumed that the cell division rate ‘shuts off’ at high T-cell concentrations . Yet , the form of the function used to model this homeostatic shutdown could , in theory , affect the model’s outcomes . We thus tested disparate shutdown functions , finding that large changes in the shutdown function only result in small changes in the dependence of the equilibrium target-cell division rate ( heq ) on the maximal target-cell division rate ( h0 ) ( S7 Fig ) . This is because the equilibrium target-cell division rate is mostly driven by the asymptotic T-cell level , which is independent of the form of the shutdown function ( SI Section B ) . A second assumption is that all TIP-immunized hosts have significantly reduced HIV transmission rates , due to a TIP-mediated reduction in HIV viral loads . However , a large fraction of HIV transmission can occur prior to TIP suppression , especially during the acute phase of infection [51–53] . To account for the possibility that a large fraction of HIV transmission by a patient occurs during the acute phase of infection prior to TIP suppression of viral loads , we re-analyzed the model under the strong assumption that TIP therapy does not reduce HIV transmission at all in dually-infected ( i . e . , TIP+ , HIV+ ) individuals . This is equivalent to the ( worst-case ) assumption that all of HIV’s spread occurs during acute infection prior to TIP inoculation . Importantly , the results of the model are virtually unchanged—both TIP dynamic stability ( S8A Fig ) and TIP evolutionary stability ( S8B Fig ) are preserved . Intuitively , the reason for the sustained TIP efficacy despite high HIV transmission is that increased HIV transmission enables increased TIP colonization of the population ( middle columns of S8 Fig ) . Thus , the increased HIV levels only strengthen the evolutionary stability of TIPs to HIV mutants ( last columns of S8 Fig ) . Finally , we examined whether increased death rates reduce the transmission potential of higher η ( i . e . , TIP-susceptible ) mutants and thereby select for lower η ( i . e . , TIP-resistant ) strains . In fact , the increase in death rates due to increasing η only has a minimal effect on the transmission potential ( S9 Fig ) . This robustness occurs because the increase in viral loads saturates as η is increased . Further , this saturation point is at a viral set-point of ~105 , whereas the measured sharp decrease in transmission potential occurs at a set point >105 [22] . Overall , the results of these sensitivity analyses support the earlier model results , showing that TIPs can be engineered to be both dynamically and evolutionarily stable at the population-scale .
The competition of mutant pathogen strains across multiple biological scales has previously been considered in a number of studies , as reviewed in [43] . Notably , these studies often predicted that co-infection would lead to increased pathogen virulence , because more virulent strains are likely to replicate more rapidly ( i . e . , have increased fitness ) [49 , 54 , 55] . This increased virulence result is , in many ways , analogous to our finding that HIV always evolves toward higher η—except that increased η in the context of TIP therapy results in decreased HIV virulence . Importantly , decreased virulence is a predicted outcome in ‘public goods’ models in which selfish , but less virulent pathogens outcompete cooperative , virulent strains [31] . In the public goods framework at the host-scale , TIPs are the public goods shared among the HIV strains co-infecting a host . Critically , an individual HIV strain benefits when there is more TIP production , because increased capsid production increases both TIP production and that strain’s relative fitness . Yet , increased TIP production is deleterious to the overall HIV population , reducing all HIV strains uniformly . Thus , evolution toward increased TIP production can be viewed as evolution toward a cheating HIV strain , explaining the overall virulence reduction as a viral ‘tragedy of the commons’ [48 , 49] . Still , a key question of this study was to determine whether the cheating HIV strain would outcompete cooperative HIV strains that produce fewer TIPs , since TIP production decreases HIV transmissibility at the population-scale . As analyzed in detail in [39 , 56] , whether or not a cheater outcompetes a cooperator in a multi-level evolutionary conflict depends on the relative strength of the within-host evolutionary pressures ( which favor the cheater ) and the between-host evolutionary pressures ( which favor the cooperator ) . Our results capture the dominance of the within-host pressures in the context of HIV-TIP dynamics , because the between-host pressures in favor of decreased TIP production are absent ( and in fact inverted ) in the sub-population that remains TIP- . In the multi-mutant model , new HIV mutants are assumed to arise infrequently relative to the strength of within-host selection—i . e . , a weak-mutation , strong-selection regime is assumed . In fact , the multi-mutant model assumes an extreme weak-mutation regime , with HIV mutants only introduced into patients via co-infection . This neglects the de novo generation of new HIV mutants within hosts , which , in truth , occurs rapidly for an RNA-encoded virus . Fortunately , this minimization of HIV mutation represents a worst-case scenario for demonstrating TIP evolutionary stability at the population-scale . If within-host mutations were to arise more frequently , the effects of host-level selection would only become more pronounced at the population scale . This is because there would be greater numbers of cheater HIV mutants that increase TIP production ( i . e . , HIV mutants with higher η values than the wild-type ) . Further , all TIP-resistant mutants with lower η values than the wild-type would be lost due to their selective disadvantage within hosts . The increased mutation of HIV strains would thus enable the emergence of HIV mutants with higher η values , enhancing the evolutionary stability of TIP treatments . Consequently , once evolutionary stability has been established in a multi-mutant model in which host-scale effects are only exerted through co-infection , stability in a model with de novo generation of mutants follows . In addition to changing the strength of mutation , one could also study how the results hinge on the strength of selection . The assumption that TIP-susceptible ( i . e . , higher η ) HIV strains competitively exclude TIP-resistant ( i . e . , lower η ) HIV strains within hosts depends on the strength of within-host selection . If host-scale selection is too weak , a distribution of mutants with different η values ( i . e . , TIP resistance levels ) could persist within hosts . However , the presence of a distribution of mutants ( rather than a single mutant ) does not obviate either TIP dynamic or evolutionary stability . We begin by considering TIP dynamic stability in the context of decreased within-host selection . Given that HIV’s burst size increases monotonically in η , any within-host distribution of η mutants can be modeled as a single ‘characteristic' η mutant: the ‘characteristic’ HIV mutant whose η value gives rise to the average HIV burst size in the host . As long as the η value of this mutant is within the stability regime derived in Fig 1B for the single-mutant case ( i . e . , η > ~ 0 . 2 ) , TIPs will remain dynamically stable at the host and population scales . For evolutionary stability , the key point is that TIP introduction leaves any distribution of η mutants essentially unchanged ( if anything , it shifts the distribution slightly toward higher η: Fig 2B ) . This is because TIPs suppress all HIV mutants within a host essentially uniformly ( they all feel the same TIP load ) . Thus , both the relative fitness values ( Fig 2B ) and relative transmission rates of the HIV mutants are unchanged by TIP introduction . As a result , there is no change in the patient-scale or population-scale distribution of η mutants and no selection for TIP-resistant strains . These arguments aside , if the strength of mutation were significantly increased , it might be possible for a beneficial mutant to arise that happens to have both a lower η and a net fitness advantage due to secondary beneficial mutations . We neglected these higher-order effects due to factors such as genetic linkage and clonal interference—as well as non-deterministic effects due to factors such as Muller’s ratchet and genetic drift—in our simplified model . Following a number of recent studies in theoretical population genetics [50 , 57–59] , these ideas could be the subject of future models . A more basic assumption of the model is that HIV escapes TIP parasitism by reducing η—i . e . , by unilaterally reducing the production of capsid elements available for TIP parasitism . Since TIPs do not encode trans elements such as capsids , the TIPs would have no ability to restore η to high values ( i . e . , η is an asymmetric parameter ) . Consequently , HIV’s unilateral evolution of η offers the most direct route for HIV to evade TIP-mediated suppression . Yet , as an alternative to this unilateral evolution in η , one could consider mutations in θ , which would alter HIV’s gRNA production rate . Importantly , mutations in θ would be more symmetric , changing the production rates of both HIV and TIP . For example , by increasing HIV’s genomic RNA ( gRNA ) production rate , increasing θ would also increase the production of the HIV protein ( Tat ) that transactivates the LTR promoters of both HIV and TIP and increase the production of the HIV protein ( Rev ) that exports both HIV and TIP gRNAs into the cytoplasm for encapsidation . Thus , TIP gRNA production would be increased in symmetry with HIV gRNA production . A key point is that if the TIP:HIV gRNA overexpression ratio ( i . e . , P ) were still sufficiently high—i . e . , if P were still >3—the TIPs’ evolutionary stability would be expected to remain . A further reason that HIV mutations in θ are unlikely to generate stable TIP-resistance is that increasing θ may not be an evolutionarily stable strategy for HIV—an intrinsic fitness cost may prevent HIV from increasing θ . In particular , a recent study [61] showed that increasing the rate of transcription ( e . g . , by adding transcription factor binding sites to the LTR ) reduces the level of HIV replication , likely by disrupting an evolutionarily-tuned viral replication program . If θ has been optimized over the millennia of lentiviral evolution in primates , then a similar cross-scale evolutionary conflict to the one shown here for η could limit the emergence of HIV mutants with increased θ values , despite their increased TIP-resistance potential . Conversely , if increasing θ were evolutionarily beneficial to lentiviruses , then TIPs could directly co-evolve to match HIV evolution for a ‘symmetric’ parameter such as θ . This is because , unlike trans ( e . g . , capsid ) elements , TIPs encode all cis-acting elements . And given their shared error-prone reverse transcriptase enzyme , TIPs have the same evolutionary capacity and pressure to modify their LTR promoter towards increased gRNA production , should increased gRNA production prove beneficial within a cell . Thus , ‘red queen’ type selection races may arise [60] , with both TIP and HIV particles simultaneously adapting to attempt to gain the upper hand in gRNA production ( i . e . , with both TIP and HIV evolving to modulate the level of gRNA overexpression , P ) . In fact , one numerical simulation study appears to have demonstrated this [20] , although the particular methodologies and assumptions have been questioned [62 , 63] . Taken together , given the potential of TIPs to co-evolve in θ and the simpler possibility that θ is already evolutionarily optimized , this study assumes that θ remains fixed , as in previous studies [12 , 18 , 19] . With θ fixed , P is similarly fixed once the TIP has been engineered . A final non-unilateral escape mechanism involves mutations in the HIV trans elements that the TIP parasitizes . It has been argued that HIV is unlikely to win the resulting cis-trans arms races with TIP , due to an intrinsic mutational asymmetry between HIV and TIP [18 , 19] . To escape TIP parasitism , HIV needs to almost simultaneously adapt both its cis and trans elements in a correlated way , so that the mutant trans still interacts with the mutant cis but not the original cis element ( which remains in the TIP ) . Within this same adaptation timeframe , the TIP only needs to mutate its corresponding cis element to keep pace . Since both HIV and TIP maintain the same mutation rate—TIPs share the same error-prone reverse transcriptase protein that drives HIV mutations—and TIPs need to mutate fewer elements , TIPs would have a built-in evolutionary advantage . Notwithstanding this advantage , the outcomes of these arms races may depend on other factors , such as the ability of an individual TIP to parasitize distinct HIV mutants . Thus , detailed simulations of arms races between HIV and TIP will be carried out in a subsequent study . As a result of the conflicting selection pressures induced by TIP , there are important caveats for modeling the evolutionary behavior of HIV in the presence of TIP . In general , the selective forces at the population level cannot be described by a fitness landscape . In order to express selection acting on a mutant as a fitness landscape , it must be possible to express the relative slope of expansion , s , as the difference in log fitness ( f ) between two strains , s = f ( η1 , η2 ) = f ( η2 ) − f ( η1 ) , where η1 and η2 are initial and final values , respectively . However , this condition is violated in the present model for three reasons: ( i ) in the case of co-infection with different HIV strains ( leading to competitive exclusion ) , the less host-fit mutant experiences a negative-selection pressure at the population level with the strength of this pressure depending on the frequency of contacts with individuals infected with fitter strains; ( ii ) the initial expansion rate of mutant strains has a term that does not depend on the magnitude of the parameter difference from the wild type ( η1 − η2 ) but only on its sign , because of the speed of within-host competitive-exclusion relative to the population time-scale; ( iii ) even in the absence of within-host co-infection , the spread of a strain depends on the balance between TIP+ and TIP− individuals in the population and this ratio adjusts with the prevalence of the strains , causing long-term oscillations ( S6 Fig ) . Given the importance of the intra-scale ( i . e . , population-scale ) conflict and the resultant frequency-dependent fitness effects at the population scale , it could be reasonably expected that similar features would appear at the host-scale . However , these frequency-dependent features can be safely neglected in this model , where we calculated a fitness landscape corresponding to incremental small changes in η ( Fig 2B ) . Although modest frequency-dependent corrections would appear for large jumps in parameter η , these effects only adjust the strength of selection , not the direction . Because we assumed a strong separation of timescales between the host and population levels , only changes in the direction ( not the magnitude ) of host-level selection affect the final results . Within hosts , HIV fitness always increases as η increases regardless of how many TIP copies are in a cell ( i . e . , regardless of TIP and HIV loads in a host ) , since a larger η always results in a larger burst size ( S1 Text Eqs . S9-S13 ) . Hence , regardless of TIP frequency , the HIV strain with largest η always spreads fastest in a host , driving the other strains to extinction . The predicted lack of unilateral evolution of HIV towards resistance to TIP is in striking contrast to HIV resistance to antivirals , which commonly arises in treated individuals due to poor adherence or suboptimal therapy regimens [64 , 65] . HIV strains that are resistant to antivirals can then transmit from host to host , spreading through the population [65] . The critical difference between antivirals and TIPs is that TIPs parasitize HIV trans elements ( i . e . , steal HIV proteins ) , and this parasitism inexorably uses the same biochemical processes as HIV replication . So , to prevent TIPs from interfering , HIV must interfere with its own ability to replicate ( i . e . , ‘shoot itself in the foot’ ) . In other words , the cost of mutation is always directly related to the benefit of the mutation . In contrast , HIV-escape from antiviral pharmaceuticals may produce some disadvantages for the mutant strain relative to the wild-type strain , but the benefit of the mutation and the cost of the mutation are not necessarily related . With each TIP-evading mutation necessarily arising at a cost , our analysis quantifies the net-benefit ( i . e . , evolutionary viability ) of these mutations across a population , where resistance may be beneficial to transmission . By capturing the parameter regimes under which resistance mutations are driven extinct , the analysis offers general guidelines for engineering therapies that obviate the spread of antiviral resistance within populations; in fact , these therapies are likely to direct pathogens toward increased susceptibility . As a result , these design constraints may aid in the engineering of resistance-proof interventions against a range of viral and bacterial pathogens beyond HIV . | A major obstacle to effective antimicrobial therapy campaigns is the rapid evolution of drug resistance . Given the static nature of current pharmaceuticals and vaccines , natural selection inevitably drives pathogens to mutate into drug-resistant variants that can resume productive replication . Further , these drug-resistant mutants transmit across populations , resulting in untreatable epidemics . Recently , a therapeutic strategy was proposed in which viral deletion mutants—termed therapeutic interfering particles ( TIPs ) —are engineered to only replicate by stealing their missing proteins from full-length viruses in co-infected cells . By stealing essential viral proteins , these engineered molecular parasites have been predicted to reduce viral levels in patients and viral transmission events across populations . Yet , a critical question is whether rapidly mutating viruses like HIV can evolve around TIP control by reducing production of the proteins that TIPs must steal in order to replicate ( i . e . , by ‘starving’ the TIPs ) . Here we develop a multi-scale model that tests whether TIP-starving HIV mutants can spread across populations to undermine TIP therapy campaigns at the population-scale . Strikingly , model results show that inherent evolutionary tradeoffs prevent these TIP-resistant HIV mutants from increasing in frequency ( i . e . , these TIP-resistant HIV mutants are continually outcompeted by TIP-sensitive mutants in both patients and populations ) . Maintained by natural selection , TIPs may offer a novel therapeutic approach to indefinitely control rapidly evolving viral pandemics . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"organismal",
"evolution",
"hiv",
"infections",
"medicine",
"and",
"health",
"sciences",
"hiv",
"prevention",
"pathology",
"and",
"laboratory",
"medicine",
"viral",
"transmission",
"and",
"infection",
"parasite",
"evolution",
"pathogens",
"microbiology",
"parasitology",
... | 2016 | Conflicting Selection Pressures Will Constrain Viral Escape from Interfering Particles: Principles for Designing Resistance-Proof Antivirals |
In HIV/SIV-infected humans and rhesus macaques ( RMs ) , a severe depletion of intestinal CD4+ T-cells producing interleukin IL-17 and IL-22 associates with loss of mucosal integrity and chronic immune activation . However , little is known about the function of IL-17 and IL-22 producing cells during lentiviral infections . Here , we longitudinally determined the levels and functions of IL-17 , IL-22 and IL-17/IL-22 producing CD4+ T-cells in blood , lymph node and colorectum of SIV-infected RMs , as well as how they recover during effective ART and are affected by ART interruption . Intestinal IL-17 and IL-22 producing CD4+ T-cells are polyfunctional in SIV-uninfected RMs , with the large majority of cells producing four or five cytokines . SIV infection induced a severe dysfunction of colorectal IL-17 , IL-22 and IL-17/IL-22 producing CD4+ T-cells , the extent of which associated with the levels of immune activation ( HLA-DR+CD38+ ) , proliferation ( Ki-67+ ) and CD4+ T-cell counts before and during ART . Additionally , Th17 cell function during ART negatively correlated with residual plasma viremia and levels of sCD163 , a soluble marker of inflammation and disease progression . Furthermore , IL-17 and IL-22 producing cell frequency and function at various pre , on , and off-ART experimental points associated with and predicted total SIV-DNA content in the colorectum and blood . While ART restored Th22 cell function to levels similar to pre-infection , it did not fully restore Th17 cell function , and all cell types were rapidly and severely affected—both quantitatively and qualitatively—after ART interruption . In conclusion , intestinal IL-17 producing cell function is severely impaired by SIV infection , not fully normalized despite effective ART , and strongly associates with inflammation as well as SIV persistence off and on ART . As such , strategies able to preserve and/or regenerate the functions of these CD4+ T-cells central for mucosal immunity are critically needed in future HIV cure research .
HIV infection in humans and SIV infection in rhesus macaques ( RMs ) is characterized by the establishment of high and persistent levels of immune activation and inflammation , which are strong and independent predictors of disease progression in the natural history of infection and co-morbidities/mortalities in individuals on antiretroviral therapy ( ART ) . While the causes of this sustained immune activation during chronic HIV/SIV infections are complex and not completely understood , the severe depletion of intestinal CD4+ T-cells early after infection and the associated loss of mucosal barrier integrity are commonly regarded as two of the most critical contributors to persistent immune activation and disease progression [1–4] . CD4+ T-cells , the main targets of HIV and SIV infections , can be classified in subsets of Th1 , Th2 , Th17 , Th22 , follicular helper ( Tfh ) , and regulatory T-cells ( Treg ) based on their phenotypes , cytokine production , transcriptional profiles and anatomic localization [5 , 6] . Th17 cells are characterized by the expression of CCR6 and the transcription factor RORγt , as well as by the production of IL-17 [7–13] . Th22 cells are characterized by the expression of the chemokine receptors CCR4 , CCR6 , and CCR10 , as well as the transcription factor aryl hydrocarbon receptor ( AHR ) [14–17] . The main cell targets of IL-22 are mucosal epithelial cells [18–20] . The in vivo effector functions of IL-17 and IL-22 are crucial to maintaining mucosal immunity against specific pathogens and include the recruitment of neutrophils to the sites of bacterial invasion , the enhancement of mucosal barrier repair and maintenance through stimulation of epithelial cell proliferation and tight junction protein production , as well as the induction of antimicrobial proteins , including beta-defensin [18 , 19 , 21 , 22] . Indeed , IL-17 and/or IL-22 associated protection has been described for numerous infections , including Citrobacter rodentium [23] , Klebsiella pneumonia [24] , Toxoplasma gondii [25] , Candida albicans [26] , Bordetella pertussis [27] , Pneumocystis carinii [28] , among others . Intestinal IL-17 and IL-22 producing cells are preferentially depleted in chronically HIV/SIV-infected subjects , with the severity of their depletion correlating with the extent of microbial translocation , chronic immune activation , and disease progression [2 , 29–36] , as well as with effective CD4+ T-cell restoration in gut-associated lymphoid tissue of HIV-infected patients on ART [37] . Further supporting their roles in disease progression , IL-17 and IL-22 producing CD4+ T-cells have been shown to be relatively preserved in the gastrointestinal tracts of chronically SIV-infected sooty mangabeys ( SMs ) and African green monkeys [29 , 34 , 38 , 39] , natural hosts of SIV infection that avoid microbial translocation , chronic immune activation and progression to AIDS , as well as in HIV controllers and long-term nonprogressors [40–42] . In addition , a recent study showed that the size of the Th17 cell compartment prior to infection limited viral replication in SIV-infected RMs [43] . Although several studies have confirmed the loss of intestinal IL-17 producing CD4+ T-cells , the dynamics of IL-22 and IL-17/IL-22 producing T-cells during HIV and SIV infection have not been studied extensively . In addition , very little is known about how pathogenic HIV/SIV infection also impacts the functional ability of Th17 and Th22 cells to co-produce additional cytokines . This is an important issue , since the ability of individual T-cells to simultaneously perform multiple effector functions is critical for protective immune responses against pathogens [22] . Recently it was shown in HIV infected individuals that intestinal Th17 cell function , which was assessed for coproduction of IFNγ , TNFα , and IL-22 , independently predicted immune activation [44] . In addition , there were two very recent studies connecting the initiation of early ART therapy to preserved Th17 function and reversed HIV-induced immune activation [45 , 46] . Since the existing functional studies have been in solely HIV-infected humans , we investigated intestinal Th17 cell function in the context of SIV infection in RMs to potentially support and further these results . In this study , we used SIV-infected RMs to investigate the levels and functions of blood and tissue IL-17 , IL-22 , and IL-17/IL-22 producing CD4+ T-cells during progressive SIV infection , how these features are recovered during effective antiretroviral therapy and affected by structured ART interruption , as well as how they associate with immune activation and viral persistence .
We first aimed to determine the ability of blood and tissue-derived IL-17 and IL-22 producing CD4+ T-cells to co-produce multiple cytokines in RMs by using a five-cytokine flow cytometric panel that included IL-17 , IL-22 , IFN-γ , TNF-α , and IL-2 . Co-production of these cytokines after PMA and Ionomycin stimulation was assessed through Boolean gating in live ( dead cells were excluded based on live/dead staining ) CD3+CD4+ T-cells producing IL-17 , IL-22 or both cytokines ( IL-17/IL-22; Th17/Th22 ) in blood , lymph node ( LN ) and colorectum . Since those cytokines are not produced by naïve CD4+ T-cells , our gating strategies correct for differences in proportions of naïve/memory CD4+ T-cells between PBMC , LN and MMC . A representative staining for the different cytokines is shown in Fig 1B . In determining functionality of these cells , we used two different methods of analysis . First , we used the SPICE program to plot the comprehensive cytokine expression profiles of each cell subset for each time point investigated . Secondly , we assigned each cell subset at each time point a “functional score” , being the mean number of cytokines produced per individual cell . Through this functional score analysis , we could then effectively quantify cell subset function , as well as correlate this with the numerous virologic and immunologic parameters of SIV infection investigated in the study . Throughout this manuscript , the words “cytokine profile” refer to a result of the SPICE analysis , and “functional score” refers to the mean number of cytokines produced by an individual cell . The functional score analysis paired with the SPICE cytokine profile analysis allowed us to assess IL-17 and IL-22 producing CD4+ T-cell polyfunctionality longitudinally before SIV infection ( with the exception of the lymph node ) , during its untreated early phase ( day 58 p . i . ) , and at time points throughout ART treatment , as well as after interruption ( Fig 1A ) . We first investigated if the function of IL-17 and IL-22 producing cells is different at the mucosal level as compared to blood in the 16 SIV-uninfected control RMs included in this first section of the study . As shown in Fig 2 , colorectal IL-17 , IL-22 , and IL-17/IL-22 producing CD4+ T-cell cytokine profiles were significantly different from their blood counterparts . Specifically , all three intestinal CD4+ T-cell subsets showed remarkably higher fractions of cells producing four and five cytokines ( Fig 2A; p<0 . 0001 for all ) and significantly higher functional scores ( 2 . 2 ± 0 . 07; 2 . 9 ± 0 . 04; 2 . 1 ± 0 . 03 for Th17 , Th22 , and Th17/Th22 , respectively ) than the cells in the blood ( 1 . 5 ± 0 . 06; 1 . 9 ± 0 . 04; 1 . 6 ± 0 . 05 for Th17 , Th22 , and Th17/Th22 , respectively ) of the same animals ( Fig 2B; p<0 . 0001 for all ) . Thus , consistent with their important roles in antimicrobial immunity and mucosal integrity [22] , IL-17 and IL-22 producing CD4+ T-cells acquire much higher polyfunctional profiles when present in the gastrointestinal tract . We then investigated how SIV infection in RMs impacts the frequency and function of IL-17 and IL-22 producing CD4+ T-cells . To this aim , the frequencies ( Fig 3 ) and function ( Fig 4 ) of IL-17+ , IL-22+ and IL-17+IL-22+ CD4+ T-cells were compared in the same 16 RMs before ( day -20 ) and post ( day 58 ) SIV infection ( p . i . ) . At day 58 p . i . , the means ± S . E were 694 , 901 ± 244 , 099 for viral load ( copies viral RNA per ml of plasma ) and 549 ± 60 for CD4 T cell counts ( cells for mm3 of blood ) . In SIV-infected RMs the levels of intestinal IL-17 , IL-22 , and IL-17/IL-22 producing cells , expressed as fractions of total CD4+ T-cells , were all severely depleted by day 58 p . i . ( p<0 . 0001 for all three subsets ) ( Fig 3B ) . Significant depletion was also observed in the blood ( p = 0 . 0002 for IL-17+; p = 0 . 0048 for IL-22+; and p = 0 . 0182 for IL-17+IL-22+; Fig 3A ) , although its extent was not as substantial as seen in the colorectum . We then sought to determine if SIV infection also impairs the function of IL-17 , IL-22 , and IL-17/IL-22 producing CD4+ T-cells . The cytokine profiles and functional scores of IL-17 , IL-22 or IL-17/IL-22 producing CD4+ T-cells remained similar between pre and post SIV infection in blood ( Fig 4A and 4B ) , but were severely altered in the mucosa of SIV-infected animals ( Fig 4A and 4C ) . In particular , at day 58 p . i . , we found a significant loss in the proportion of Th17 and Th22 cells co-producing four or five cytokines ( p<0 . 0001 ) , with a concomitant expansion of cells able to produce only their signature cytokine ( IL-17 or IL-22 ) , or their signature cytokine plus a single additional cytokine ( p<0 . 0001 ) . A similar loss of function was found in Th17/Th22 CD4+ T-cells ( p<0 . 0001 ) . Consistently , the functional score fell from 2 . 2 ± 0 . 07 to 1 . 23 ± 0 . 12 for Th17 cells ( p<0 . 0001 ) , from 2 . 86 ± 0 . 04 to 2 . 22 ± 0 . 12 for Th22 cells ( p = 0 . 0002 ) , and from 2 . 1 ± 0 . 03 to 1 . 53 ± 0 . 09 from Th17/Th22 cells ( p = 0 . 0001 ) , ( Fig 4C ) . In summary , pathogenic SIV infection in RMs specifically impacts the critical intestinal Th17 , Th22 , and Th17/Th22 CD4+ T-cell compartments both quantitatively and qualitatively , with a severe numeric and functional loss of these cells . We then sought to determine the efficacy of combined antiretroviral therapy ( ART ) to reconstitute intestinal Th17 , Th22 , and Th17/Th22 CD4+ T-cells . Specifically , number and function of IL-17 , IL-22 , and IL-17/IL-22 producing CD4+ T-cells were longitudinally determined in blood , LN and colorectal mucosa before ART ( day 58 p . i . ) and at four experimental points during ART ( days 84 , 135 , 200 , and 256 p . i . , i . e . , weeks 3 , 10 , 20 , and 28 on ART ) , in eight of the original 16 SIV-infected RMs included in this study ( RbT12 , RGe12 , RCb12 , RJw12 , RPu12 , RKg11 , RKd12 and RPy8 ) . All eight treated animals showed undetectable ( < 60 copies/ml of plasma ) levels of viral SIV-RNA starting from 20 weeks on ART ( day 200 p . i . ) . While the lymph node IL-17 producing cells’ function and cytokine profiles were unchanged throughout treatment ( S1 Fig ) , ART was very effective in improving the cytokine profile of the intestinal Th22 cell subset , specifically expanding the proportion of cells co-expressing 3 and 4 cytokines ( p = 0 . 0005 , S2 Fig ) . In fact , the Th22 functional score significantly rose from 2 . 22 ± 0 . 12 at pre-ART to 2 . 85 ± 0 . 07 at the end of ART treatment ( p = 0 . 0072 ) , thus virtually matching the function level seen before infection ( 2 . 86 ± 0 . 04; p = 0 . 6769; Fig 5B ) . Similar results were found for the intestinal Th17/Th22 cells , whose functional scores increased from 1 . 53 ± 0 . 09 at pre-ART to 2 . 08 ± 0 . 07 at the end of ART ( p = 0 . 0034 ) , matching pre-infection levels ( 2 . 08 ± 0 . 07 vs . 2 . 09 ± 0 . 03; p = 0 . 7321; Fig 5C ) . In contrast , intestinal Th17 cells were not fully restored by ART . Indeed , although the Th17 cell functional score increased from 1 . 23 ± 0 . 12 cytokines at pre-ART to 1 . 64 ± 0 . 17 cytokines at the last experimental point on ART ( p = 0 . 0151 ) , it remained significantly lower when compared to that at pre-infection ( 2 . 20 ± 0 . 07 cytokines; p = 0 . 0193; Fig 5A ) . We also quantified the levels of intestinal IL-17 and IL-22 producing cells , and here ART’s reconstitution was minimal , with all three subsets’ levels remaining significantly lower p<0 . 001 ) than pre-infection levels ( Fig 5D ) . When cell function and subset levels were cumulatively combined , ART’s inability ( at least when limited to a 7 month duration as in this study ) to fully restore the subsets’ functional and numeric levels was clearly demonstrated ( Fig 5E ) , with cumulative scores at the last experimental point on ART still remarkably lower than those at pre-infection for intestinal Th17 , Th22 , and Th17/Th22 CD4+ T-cells ( p<0 . 001 for all three subsets ) . We then investigated how the function and levels of IL-17 and IL-22 producing cells were affected with the discontinuation of ART . At day 180 post structured ART interruption , the levels of all three subsets significantly decreased from those seen on ART ( p = 0 . 0416 for Th17; p = 0 . 0313 for Th22; p = 0 . 0469 for Th17/Th22; Fig 6A ) . Importantly , the Th17 cell cytokine profile changed from that observed late on-ART ( p<0 . 0001 ) and reverted back to the same profile found at pre-ART ( p = 0 . 6600; S2 Fig ) . Functional score analysis confirmed the drastic fall in Th17 cell function after ART discontinuation , with functional score dropping from 1 . 67 ± 0 . 17 to 1 . 24 ± 0 . 18 cytokines per cell ( p = 0 . 0391; Fig 6B ) , and reverting back to pre-ART levels ( p = 0 . 2500; Fig 6B ) . A similar functional decrease after ART interruption was also observed in the Th17/Th22 cells , whose functional score decreased from 2 . 08 ± 0 . 03 ( on ART ) to 1 . 79 ± 0 . 12 ( at d180 post ART interruption ) ( p = 0 . 0257; Fig 6D ) . Interestingly , although the cytokine profile changed from on-ART levels , with a reduced fraction of Th22 cells co-producing three cytokines ( p<0 . 0001 ) , the overall functional score for Th22 cells after ART interruption remained unchanged from ART levels ( p = 0 . 2167; Fig 6C ) . In blood , no significant changes were seen in the cytokine profiles or function after ART discontinuation ( S3 Fig ) . In summary , and despite its inability for full restoration , ART is necessary for at least maintenance of colorectal Th17 , Th22 , and Th17/Th22 cell function and levels , as evidenced by the rapid general regression of both upon ART interruption . We next determined if the functional impairment ( quantified and represented by the functional scores ) of the intestinal Th17 , Th22 and Th17/Th22 cells associated with the main virologic and immunologic markers of disease progression . We found a negative correlation between Th17/Th22 T-cell function and viral loads at d58 p . i . ( p = 0 . 0349; r = -0 . 5353; Fig 7A ) . This same pattern was observed at the intermediate on-ART time point of day 135 , in which Th17 ( p = 0 . 0247 , r = -0 . 7722; Fig 7B ) , Th22 ( p = 0 . 0195 , r = -0 . 7907; S4 Fig ) , and Th17/Th22 T-cells ( p = 0 . 0063 , r = -0 . 8592; S4 Fig ) increased their functional scores while the viral loads decreased , thus suggesting a link between intestinal CD4+ T-cell functionality and viral replication . Moreover , absolute CD4+ T-cell counts positively correlated with intestinal Th22 cell functional score ( p = 0 . 0146 , r = 0 . 5970 ) ( Fig 7C ) . To further explore the association between reduced IL-17 producing CD4+ T-cell function and disease progression , we also examined levels of colorectal CD4+ T-cell proliferation ( as measured by Ki-67 expression ) and immune activation ( as measured by the co-expression of HLA-DR and CD38 ) . At day 58 p . i . ( pre-ART ) , we found a negative correlation between intestinal Th17 cell functional scores and the fraction of HLA-DR+CD38+ CD4+ T-cell levels ( p = 0 . 0305 , r = -0 . 5408 , Fig 7D ) . Similarly , at day 135 p . i . , a negative correlation was also seen between Th17/Th22 cell functional scores and the levels of CD4+ T-cells co-expressing HLA-DR and CD38 ( p = 0 . 0417 , r = -0 . 7254; Fig 7E ) . Furthermore , the intestinal Th17 ( p = 0 . 0212 , r = -0 . 7843; Fig 7F ) , Th22 ( p = 0 . 0285 , r = -0 . 7605; S5 Fig ) , and Th17/Th22 ( p = 0 . 0353 , r = -0 . 7413; S5 Fig ) cell functional scores negatively correlated with the levels of proliferation during ART ( day 135 p . i . ) . We also expanded the analyses to IL-17+IFN-γ+ CD4+ T-cells , as the emergence of this population with a mixed Th17/Th1 phenotype has been proposed to be triggered in pro-inflammatory conditions [47–49] . The frequencies ( within the total CD4+ T-cell population ) of intestinal IL-17+IFN-γ+ cells significantly decreased upon SIV infection ( day 58 p . i . ) as compared to pre-infection ( day -20 p . i: 3 . 303 ± 0 . 3027; day 58 p . i: 1 . 088 ± 0 . 3906; mean ± S . E; p = 0 . 0017 ) , and were significantly increased with ART treatment ( day 256 p . i; 1 . 979 ± 0 . 7622; mean ± S . E ) as compared to day 58 p . i . ( p = 0 . 0156; S6 Fig ) . Interestingly , while the frequencies of RB IFN-γ+ cells associated with levels of RB CD4+ DR+38+ T-cells at the majority of measured time points , including before , throughout , and after ART treatment , IL-17+IFN-γ+ CD4+ T-cells only correlated with levels of RB CD4+ DR+38+ T-cells at day 58 p . i . , and never associated with levels of CD8+ immune activation or with measures of CD4 and CD8 T-cell proliferation . Next , we longitudinally examined several soluble markers of immune activation , such as IP-10 , sCD14 , lipopolysaccharides ( LPS ) , lipopolysaccharide binding protein ( LBP ) , and sCD163 , in order to monitor the inflammation status at pre and post-ART initiation . Levels of sCD14 , which has been shown to be elevated and to predict mortality in HIV-infected individuals [50] , significantly correlated with the functional score fold change between day 256 p . i . and day 58 p . i . in Th22 cells ( p = 0 . 0349 , r = -0 . 7424; Fig 7G ) and strongly trended in the Th17/Th22 ( p = 0 . 0679 , r = -0 . 6721 ) population . In addition , furthering the theory that impaired intestinal IL-17 producing CD4+ T-cell functionality is mechanistically linked to immune activation and disease progression , higher functional scores of Th17 cells ( p = 0 . 0465 , r = -0 . 7144; S7 Fig ) and Th17/Th22 T-cells ( p = 0 . 0228 , r = -0 . 7786; Fig 7H ) during late-ART treatment negatively correlated with levels of sCD163 , whose expansion has been linked to faster AIDS progression and residual inflammation [51 , 52] . We saw similar patterns with IP-10 levels at day 200 p . i . , which negatively trended with both Th22 and Th17/Th22 T-cell functional scores , lending further evidence to an association between restored mucosal homeostasis and reduced levels of inflammation . Of note , there were no correlations between intestinal T-cell function and plasma levels of LPS and LBP . Additionally , the same was found with colorectal expression of myeloperoxidase ( MPO ) , a proinflammatory , neutrophil-associated enzyme whose levels are elevated in HIV-infected individuals . Finally , and despite these mucosal associations , intestinal Th17 , Th22 , and Th17/Th22 T-cell functional scores did not associate with immune activation or disease progression parameters in blood or LN . Taken together , these results further indicate the critical importance of intestinal IL-17 and IL-22 producing CD4+ T-cell function in the pathogenesis of SIV infection , particularly in the gut mucosa . We then investigated whether , by impacting residual activation and inflammation , the Th17 , Th22 , and Th17/Th22 cell number and function loss associates with SIV persistence on ART . To address this question , we measured at various experimental points the amount of total SIV-DNA in the colorectum and in CD4+ T-cells purified from blood . At the last on-ART time point ( day 256 p . i . ) , increasing Th17 ( p = 0 . 0014 , r = -0 . 9167 ) and Th22 cell ( p = 0 . 0115 , r = -0 . 8259 ) function , as well as Th22 cell levels ( p = 0 . 0264 , r = -0 . 7669 ) , strongly correlated with lower levels of SIV-DNA in the colorectum ( copies per 108 cells equivalent ) ( Fig 8A–8C ) . Of note , these associations between Th17 and Th22 cells and SIV-DNA content were independent from and even stronger after controlling for plasma viremia pre-ART ( S1 Table ) . Additionally , this mucosal SIV-DNA content at the latest time point on ART correlated with the fractions of Th17 ( p = 0 . 0165 , r = -0 . 8025 ) and Th22 ( p = 0 . 0154 , r = -0 . 8333 ) cells at pre-ART , thus suggesting that IL-17 and IL-22 producing T-cell numbers before treatment can also predict the level of intestinal cell-associated SIV-DNA that will persist during ART ( Fig 8D and 8E ) . The suggested link between loss of IL-17 and IL-22 producing T-cell function and levels and SIV persistence is furthered by SIV-DNA data from after ART-interruption . Six months after ART had been interrupted ( day 440 p . i . ) , the fraction of Th17 cells ( p = 0 . 0106 , r = -0 . 8312 ) and Th22 cell function ( p = 0 . 0318 , r = -0 . 7509 ) both correlated with blood SIV-DNA content ( copies per 106 CD4+ T-cells ) ( Fig 8F and 8G ) . Further , blood SIV-DNA content at six months after ART interruption correlated with Th17 cell levels ( p = 0 . 0006 , r = -0 . 9365 ) and Th17/Th22 cell function ( p = 0 . 0058 , r = - . 08632 ) before ART initiation ( day 58 p . i . ) ( Fig 8H and 8I ) . Longitudinal values for plasma viremia and blood CD4+ T-cells SIV-DNA contents are shown in S8 Fig for the eight animals that underwent ART interruption . Collectively , these results indicate that animals with higher function and numbers of IL-17 , IL-22 , and IL-17/IL-22 producing cells before and during ART treatment exhibit lower SIV-DNA content throughout and after ART treatment , thus suggesting Th17 , Th22 , and Th17/Th22 cell numbers and function as critical regulators of SIV persistence .
A severe loss of intestinal CD4+ T cells [1–4] , which preferentially involves CD4+ T-cell subsets with anti-microbial properties such as Th17 and Th22 cells [29 , 30 , 32–34 , 53] , has been proposed as one of the most critical factors contributing to the breakdown of mucosal epithelial integrity during chronic HIV and SIV infections . This loss of mucosal integrity is considered a key contributor to the establishment and persistence of high levels of chronic immune activation in HIV infection [34 , 36 , 54] . While several studies have convincingly shown a preferential loss of intestinal IL-17 and IL-22 producing CD4+ T-cells in HIV and SIV infection , it is unclear how exactly the functions of these cells are perturbed during infection . Furthermore , the extent to which the number and function of IL-17 and IL-22 producing CD4+ T-cells are recovered during highly effective ART therapy is still unclear . The only limited insight to these important questions comes from a few HIV studies that documented the loss of IL-17 producing cell function during progressive infection and the challenge in reversing this defect , with Th17 cell function fully recovered only after extremely long-term ART , particularly that which was started very early after infection [45 , 46] . To provide better insights on these important questions , we used the HIV model of SIV infection in RMs to extensively investigate the effects of SIV infection , ART treatment , as well as ART interruption on the number and functional ability of IL-17 and IL-22 producing CD4+ T-cells to simultaneously produce multiple effector cytokines such as IL-17 , IL-22 , TNFα , IFNγ , and IL-2 . While several recent cross-sectional studies have linked decreased Th17 cell function to HIV infection and increased immune activation , none have attempted to follow changing functional dynamics longitudinally . A key advantage of our study was the possibility to follow the same animals longitudinally throughout pre and post SIV infection , ART treatment , as well as after ART interruption , which is virtually impossible in human studies due to ethical , demographic , and time constraints . Another strength in our investigation was that Th22 and Th17/Th22 cell function were also simultaneously examined , as several recent studies have indicated that Th17 cells are not the only subset of CD4+ T-cells preferentially lost from the intestines of HIV and SIV infected individuals . In addition , this quantification of IL-17 and IL-22 producing CD4+ T-cell levels and function was simultaneously investigated in blood , colorectum , as well as in the lymph nodes . This provided us with the unique opportunity to investigate the extent of association between numerous immunologic and virologic correlates of disease progression , as well as SIV persistence and total DNA content , and IL-17 producing cell function . To the best of our knowledge , these longitudinal , in-depth studies on the number and functional abilities of IL-17 and IL-22 producing CD4+ T-cells have never been performed in HIV-infected humans or SIV-infected RMs . We found that intestinal , but not blood or lymph node , IL-17 , IL-22 , and IL-17/IL-22 producing CD4+ T-cells are polyfunctional in SIV-uninfected RMs , with the large majority of the cell types co-producing four or five cytokines . However , SIV infection induced a numeric loss and a severe dysfunction of these intestinal CD4+ T-cells and caused the average numbers of cytokines produced per each cell to drastically drop , while leaving blood and lymph node cytokine expression profiles unchanged , thus supporting the critical importance of intestinal IL-17 and IL-22 producing CD4+ T-cells in maintaining mucosal homeostasis during HIV and SIV infection . We examined how ART treatment at different time points and after its interruption affected cell cytokine expression dynamics and functional levels . Our results give strong evidence that ART partially restores and maintains colorectal Th17 , Th22 , and Th17/Th22 CD4+ T-cell polyfunctionality , especially evidenced by severe decreases in level and function observed when ART was interrupted . However , it is clear that ART alone , at least when used for 7 months as in our study , is not sufficient to bring cell function and numbers back to pre-infection levels , as all examined subsets by the end of ART showed significantly reduced cumulative levels; this was particularly noticeable in the case of the Th17 cells . Importantly , we also found that the loss of intestinal IL-17 and IL-22 producing CD4+ T-cell function associated with the extent of chronic immune activation , pre-ART viral loads , as well as lower CD4+ T-cell counts before and during ART treatment . Furthermore , levels of gut T-cell activation and proliferation were associated with loss of function from all three ( IL-17 , IL-22 , and IL-17/IL-22 producing ) intestinal cell subsets . In addition , cell function negatively correlated with levels of sCD163 and sCD14 , two soluble markers of inflammation , whose levels have been shown to increase during HIV and SIV infection and are linked to microbial translocation , macrophage activation , disease progression , and mortality in ART-treated HIV-infected individuals [50–52] . Remarkably , this loss of function and cell frequency also consistently correlated with higher SIV-DNA content in the colorectum and blood throughout and after ART , thus suggesting IL-17 and IL-22 producing CD4+ T-cell numbers and function as regulators and potential predictors of SIV persistence . While the design of our study has allowed for novel findings and a longitudinal investigation into the nature of IL-17 and IL-22 producing cell frequency and function in the context of SIV infection , there exist several confining factors . Our study utilized a 7 month ART duration , largely dictated by the practicality of using a nonhuman primate model . Based on our data , we cannot exclude that a more prolonged therapy and/or earlier initiation would have resulted in a more significant restoration of Th17 cell number and function . A recent study in HIV-infected humans showed Th17 cell function , after having been drastically ablated after infection , was eventually fully restored to pre-infection levels , but only after very prolonged ART ( median of 13 years ) [44] . Additionally , two other studies showed that intestinal Th17 cell function was preserved when ART was initiated very early during acute infection [45 , 46] . In addition , since our analyses during the natural history of SIV-infection were performed in the early chronic phase of infection ( day 58 p . i . ) , we cannot determine how the function of IL-17 and IL-22 producing CD4+ T-cells are affected in the early acute infection . Further studies focusing on the first days after SIV infection are needed to address this important point . In conclusion , we demonstrated that the polyfunctionality specific to intestinal IL-17 and IL-22 producing CD4+ T-cells is severely compromised upon SIV infection . ART does not fully restore the function and levels in these mucosal Th17 , Th22 , and Th17/Th22 T-cells , yet is necessary for at least partial functional maintenance , as evidenced by the ablative effects of ART interruption . Importantly , this loss of IL-17 and IL-22 producing cell function and frequency associates with and predicts immune activation and disease progression in RMs , as well as SIV persistence in colorectum and blood . As such , our data suggests that therapies able to preserve and/or regenerate the functions of these intestinal CD4+ T-cells subsets central for mucosal immunity should be included in the therapeutic regimen necessary for achieving HIV remission .
All animal experimentations were conducted following guidelines established by the Animal Welfare Act and the NIH for housing and care of laboratory animals and performed in accordance with Institutional regulations after review and approval by the Institutional Animal Care and Usage Committees ( IACUC; protocol #2001973 ) at the Yerkes National Primate Research Center ( YNPRC ) . Appropriate procedures were performed to ensure that potential distress , pain , discomfort and/or injury was limited to that unavoidable in the conduct of the research plan . All the blood and tissue collections were obtained from RMs housed at the Yerkes National Primate Research Center , which is accredited by American Association of Accreditation of Laboratory Animal Care . The sedative Ketamine ( 10 mg/kg ) and/or Telazol ( 4 mg/kg ) were applied as necessary for blood and tissue collections and analgesics were used when determined appropriate by veterinary medical staff . RMs were fed standard monkey chow ( Jumbo Monkey Diet 5037 , Purina Mills , St Louis , MO ) twice daily . Consumption is monitored and adjustments are made as necessary depending on sex , age , and weight so that animals get enough food with minimum waste . SIV-infected RMs are singly caged but have visual , auditory , and olfactory contact with at least one social partner , permitting the expression of non-contact social behavior . The YNPRC enrichment plan employs several general categories of enrichment . Animals have access to more than one category of enrichment . IACUC proposals include a written scientific justification for any exclusions from some or all parts of the plan . Research-related exemptions are reviewed no less than annually . Clinically justified exemptions are reviewed more frequently by the attending veterinarian . Sixteen RMs , all housed at the Yerkes National Primate Research Center , Atlanta , GA , were included in the study . All animals were Mamu-B*08 and B*17 negative , while eight of them were Mamu-A*01 positive ( RLm12 , RBt12 , RJp11 , RCb12 , RVt10 , RKg11 , ROc10 and RPy8 ) . The 16 RM were in average 9 . 1 ± 0 . 6 years old and weighed 7 ± 0 . 28 kg . All 16 animals were infected intravenously ( i . v . ) with 300 TCID50 SIVmac239 ( day 0 ) . Starting at day 58 p . i . , all animals were treated with a five-drug ART regimen comprising of two reverse transcriptase ( RT ) inhibitors ( PMPA 20 mg/Kg and FTC 30 mg/Kg ) , one integrase inhibitor ( raltegravir 100 mg/bid ) , and one protease inhibitor ( darunavir 375 mg/bid with ritonavir 50 mg/bid as a boosting supplementation ) for seven months . Animal ROc10 was euthanized at d140 p . i . due to post-surgery ( lymph node biopsy ) related complications . At day 270 p . i . , ART was interrupted and all animals were monitored for an additional eight months . Peripheral blood ( PB ) , colorectal mucosa ( RB ) and lymph node ( LN ) biopsies were collected at numerous experimental points throughout all the study ( Fig 1a ) . Collections and processing of PB , RB , and LN were performed as previously described [29–31 , 55 , 56] . All samples were processed , fixed ( 1% paraformaldehyde ) , and analyzed within 24 hours of collection . Fourteen-parameter flow cytometric analysis was performed on PB- , LN- and RB-derived cells . Predetermined optimal concentrations were used of the following antibodies: anti-CD3-APC-Cy7 ( clone SP34-2 ) , anti-Ki-67-Alexa700 ( clone B56 ) , anti-IFN-γ-PE-Cy7 ( clone B27 ) , anti-CD8-PE-CF-594 ( clone RPA-T8 ) , anti-TNFα-Alexa700 ( clone MAb11 ) , ( all from BD Pharmingen ) ; anti-IL-17-Alexa Fluor488 ( clone eBio64DEC17 ) , anti-IL-22-APC ( clone IL22JOP ) ( all from eBioscience ) ; anti-CD4-BV421 ( clone OKT4 ) , anti-IL-2-BV605 ( clone MQ1-17H12 ) , ( all from Biolegend ) ; anti-CD8-Qdot705 ( clone 3B5 ) and Aqua Live/Dead amine dye-AmCyan ( all from Invitrogen ) . Flow cytometric acquisition was performed on at least 100 , 000 CD3+ T cells on an LSRII cytometer driven by the FACS DiVa software . Analysis of the acquired data was performed using FlowJo software ( TreeStar ) . Levels of Th17 and Th22 cells were determined as the percentage of CD4+ T-cells that produce IL-17 and IL-22 following in vitro stimulation with PMA & Ionomycin [31] . PBMC , LN and RB derived cells , isolated as described above , were resuspended to 3 × 106 cells/ml in complete RPMI 1640 medium . Cells were then incubated for 4 h at 37°C in medium containing PMA , A23187 , and Golgi Stop . Following incubation , the cells were washed and stained with surface markers for 30 minutes in the dark at room temperature followed by fixation and permeabilization . After permeabilization , cells were washed and stained intracellularly with the antibodies for the cytokines of interest for 1 hour in the dark at room temperature . Following staining , cells were washed , fixed in PBS containing 1% paraformaldehyde , and acquired on an LSRII cytometer . Plasma SIV viral loads were determined by standard quantitative RT-PCR as previously described ( limit of detection 60 copies/ml ) [57] . Quantitative assessment of cell-associated total SIV-DNA within circulating CD4+ T-cells at days 58 p . i . , 105 and 256 on-ART , and day 180 off-ART was performed using a modified version of a recently published quantitative nested PCR assay for cell-associated total HIV-DNA [58] . In a first round of PCR , total SIV DNA was amplified with two primers that anneal within conserved region of the LTR 5’ ( SIV-LF1 ) and at the junction with Gag gene ( SIV-R1 ) . The forward primer SIV-LF1 is extended with a lambda phage-specific heel sequence at 5’ end of the oligonucleotide . Primers targeting CD3 gene ( HCD3OUT 5’ and HCD3OUT 3’ ) were also added to quantify the exact number of cells in the initial samples . Gag-LTR sequence were amplified from 15 μL of lysate in a 50 μL reaction mixture comprising 1X Taq Buffer , MgCl2 , dNTP , SIV- 38 LF1 , SIV-R1 and Taq polymerase . The first round PCR cycle conditions were as follows: a denaturation step of 8 min at 95°C and then 16 cycles of amplification ( 95°C for 1min , 62°C for 40 sec , 72°C for 1 min ) , followed by an elongation step at 72°C for 15 min . In a second round of PCR , the lambda T-specific primer ( Lambda T ) and the LTR primer ( SIV-R2 ) , were used to amplified SIV sequences obtained from the first amplification . Primers targeting CD3 were also used in another second round PCR . Nested PCR was performed on 1/10 of the first round PCR product in a mixture comprising 1x Rotor Gene Master mix , Lambda T primer , SIV-R2 primers and SIVprobe . For CD3 amplification , nested PCR was performed in a mixture comprising 1X Rotor Gene Master Mix , HCD3IN 5’ and MamuCD3IN 3’ and MamuCD3probe . The cycling was performed on the Rotorgene ( Qiagen ) as follow: a denaturation step ( 95°C for 4 min ) , followed by 40 cycles of amplification ( 95°C for 3 sec , 60°C for 10 sec ) . The copy number of total SIV DNA was calculated by using a standard curve as a reference . This standard curve consisted in serial dilution of the 3D8 cell lysates ( carrying one integrated copy of SIV genome per cell ) [59] . Quantitative assessments of SIV-DNA in colorectal tissue at days 50 and 200 on-ART were determined by the quantitative hybrid real-time/digital RT–PCR and PCR assays , as previously described [60] . For each sample , 12 replicate reactions were run with a nominal single copy sensitivity . The clinical sensitivity ( based on the number of cells assessed ) in our samples was as low as 1 copy/850 , 000 cells . Based on sample distribution ( normal or non-normal ) , T-tests or Mann Whitney tests were used to compare the differences of each parameter between two different groups . Statistical tests were two-sided . Pearson product-moment correlation coefficients were utilized to estimate linear associations for normally distributed data and Spearman rank correlation coefficients were used for skewed and other non-normal distributions . A P value ≤ 0 . 05 was considered statistically significant . The mean ± SEM were used for descriptive statistics for each parameter . All linear regression and correlation analyses were performed using Prism version ( 5 ) software ( GraphPad ) . IL-17 and IL-22 producing cell polyfunctionality was obtained by FlowJo Boolean gating analysis . SPICE software version 5 . 33 ( National Institute of Allergy and Infection Diseases/National Institutes of Health ) was utilized to perform Th17 and Th22 ployfunctionality analysis with the Wilcoxon signed rank test . Functional scores were calculated based on the mean numbers of proinflammatory cytokines produced by each Th17 , Th22 , or Th17/Th22 cell ( specifically calculated by multiplying the fraction of cells in each pie slice by the number of cytokines it represented , and then summing together the resulting values and dividing by 100 ) . | Persistent immune activation and inflammation are key features and strong predictors of morbidity/mortality in HIV infection . A specific quantitative loss of Th17 and Th22 CD4+ T-cells , which are crucial to maintaining the mucosal immunity , has been shown to directly associate with microbial translocation , systemic immune activation , and disease progression . Despite this , how HIV infection impacts Th17 and Th22 cell qualitative function remains largely unknown . To address this important question , we investigated Th17 and Th22 cell function and levels longitudinally before , during , and after ART in the rhesus macaque model of SIV infection in the colorectum , blood , and lymph node . We found that mucosal Th17 and Th22 cell function and levels were profoundly ablated upon SIV infection , and only partially restored by ART . Importantly , this loss of IL-17 and IL-22 producing cell function directly correlated with disease progression , immune activation , and SIV persistence . These data strongly support a molecular link between persistent inflammation and viral persistence as well as the importance of preserving intestinal Th17 and Th22 cell function during HIV infection , and urge the need for therapeutic strategies aimed at improving these cells function in future HIV cure research . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"blood",
"cells",
"hiv",
"infections",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"cytokines",
"body",
"fluids",
"immune",
"activation",
"pathogens",
... | 2016 | Loss of Function of Intestinal IL-17 and IL-22 Producing Cells Contributes to Inflammation and Viral Persistence in SIV-Infected Rhesus Macaques |
Multilayered defense responses ensure that plants are hosts to only a few adapted pathogens in the environment . The host range of a plant pathogen depends on its ability to fully overcome plant defense barriers , with failure at any single step sufficient to prevent life cycle completion of the pathogen . Puccinia striiformis , the causal agent of stripe rust ( =yellow rust ) , is an agronomically important obligate biotrophic fungal pathogen of wheat and barley . It is generally unable to complete its life cycle on the non-adapted wild grass species Brachypodium distachyon , but natural variation exists for the degree of hyphal colonization by Puccinia striiformis . Using three B . distachyon mapping populations , we identified genetic loci conferring colonization resistance to wheat-adapted and barley-adapted isolates of P . striiformis . We observed a genetic architecture composed of two major effect QTLs ( Yrr1 and Yrr3 ) restricting the colonization of P . striiformis . Isolate specificity was observed for Yrr1 , whereas Yrr3 was effective against all tested P . striiformis isolates . Plant immune receptors of the nucleotide binding , leucine-rich repeat ( NB-LRR ) encoding gene family are present at the Yrr3 locus , whereas genes of this family were not identified at the Yrr1 locus . While it has been proposed that resistance to adapted and non-adapted pathogens are inherently different , the observation of ( 1 ) a simple genetic architecture of colonization resistance , ( 2 ) isolate specificity of major and minor effect QTLs , and ( 3 ) NB-LRR encoding genes at the Yrr3 locus suggest that factors associated with resistance to adapted pathogens are also critical for non-adapted pathogens .
An integral characteristic of plant-pathogen interactions are the several events that lead to infection of a plant by a pathogen . To successfully complete its life cycle , that is to colonize a plant and reproduce , a plant pathogen needs to overcome several preformed and inducible barriers [1] . Successful life cycle completion relies on compatibility at all of these stages and incompatibility at only one stage prevents pathogen reproduction . Because of this , a plant is generally resistant to the vast majority of potential pathogens in the environment and only susceptible to a small number of adapted pathogens [2] . Additionally , colonization of new plant species by plant pathogens is considered a rare event [3] , as any new pathogen would have to overcome all defense barriers employed by the new plant species . Prior to the arrival of a pathogen , plants have several preformed barriers that will limit infection . Examples include the leaf surface composition , which can prevent germination and differentiation of the plant pathogen , or antimicrobial molecules , such as avenacins of oat that can prevent pathogen growth in leaf tissue [1 , 4 , 5] . Once a plant pathogen evades preformed barriers , recognition of the attempted infection may occur and initiate the deployment of inducible barriers [1] . Examples of inducible barriers include the three PENETRATION ( PEN ) genes of Arabidopsis thaliana , which regulate the structural rearrangements necessary for the formation of papillae , localized reinforcements of the cell wall that prevent pathogen colonization [2 , 6–9] . PEN gene expression is induced upon flagellin perception , a bacterial pathogen-associated molecular pattern ( PAMP ) , by the membrane-localized plant immune receptor FLS2 , a receptor-like kinase [7 , 10] . Recognition at the membrane can be overcome by plant pathogens through the secretion of effector molecules into plant cells [11 , 12] . In turn , plants have evolved nucleotide binding , leucine-rich repeat ( NB-LRR ) proteins that recognize effector molecules or effector modifications of plant proteins . By initiating localized cell death , also called hypersensitive response , this recognition forms a further defense layer [13–15] . These late stages of the plant-pathogen interaction are conceptualized as PAMP-triggered immunity ( PTI ) and effector-triggered immunity ( ETI ) . ETI can be suppressed by additional pathogen effectors , prompting an evolutionary arms race between plant and pathogen [13] . The interactions between NB-LRRs and effectors are commonly genetically observed as a gene-for-gene interaction between the host plant and an adapted pathogen [13 , 16] . Puccinia striiformis , causal agent of stripe or yellow rust , is an agronomically important obligate biotrophic fungal pathogen of wheat , barley , and other domesticated crops , as well as many non-domesticated grasses [17–19] . Following stomatal penetration , the first stage of P . striiformis development involves hyphal differentiation and colonization of host leaf tissue through the formation of haustoria for nutrient acquisition and effector secretion [18] . After substantial colonization of a compatible host , P . striiformis transitions to a reproductive stage through the development of urediniospores , which completes the asexual reproductive lifecycle [18 , 20] . Sexual reproduction involves additional spore stages on the alternative host , Berberis spp . [18 , 21] . P . striiformis isolates adapted to certain host genera are differentiated as formae speciales , including P . striiformis f . sp . tritici with wheat as the main host ( wheat stripe rust , Pst ) and P . striiformis f . sp . hordei with barley as the main host ( barley stripe rust , Psh ) [22] . However , this classification is complicated by the existence of formae speciales with overlapping host ranges . For example , a P . striiformis race emerged on triticale in Denmark and Sweden in 2008 and 2009 , which also infected spring wheat , barley , and rye [18 , 23] . Straib [24] investigated the host range of Pst and Psh isolates on a panel of 227 mainly non-domesticated grass species and observed chlorotic or necrotic flecks as well as pustule formation in some genera . The panel included an accession of the diploid monocot model Brachypodium distachyon , which was completely immune to the isolates studied . Draper et al . [25] identified B . distachyon accessions that produced disease symptoms in the form of “brown flecking” upon Pst and Psh inoculation . These observations were confirmed by Barbieri et al . [26] , who described “large dark flecks” on some B . distachyon accessions in response to infection with Pst and Psh isolates . A comprehensive analysis of B . distachyon–Pst interactions linked these macroscopic flecks with hyphal colonization [27] , which led to the application of a robust and quantitative phenotyping assay to a diversity set of Brachypodium spp . accessions inoculated with two UK Pst isolates [28] . A strong correlation between macroscopic leaf browning and hyphal colonization was observed across 210 Brachypodium spp . accessions . Although host jumps are considered rare events , pathogens are often able to infect or colonize plants other than their adapted host with varying degrees of success [29] . As exemplified by the interaction between P . striiformis and B . distachyon , a range of phenotypes are observed that are difficult to assign to a host ( full compatibility ) or nonhost ( full incompatibility ) state . Therefore , the status of species can be described by the range of colonization and life cycle completion by the pathogen [30] . This classification is based on the diversity observed at the species level for both plant and rust . In the case of intermediate nonhost species , no accession would support life cycle completion by different rust isolates , but some accessions would allow a degree of colonization . The above-mentioned studies established B . distachyon as an intermediate nonhost of Pst and Psh . In contrast , rice is considered a nonhost of rusts , as no accessions have been identified that allow extensive colonization or further disease progression [31–33] . P . brachypodii is an adapted rust pathogen of B . distachyon and the related B . sylvaticum [26 , 34] . Therefore , unlike for rice , fully compatible interactions exist between B . distachyon and a rust pathogen . Consequently , the B . distachyon–P . striiformis interaction provides a unique system to study the genetic architecture underlying defense responses against non-adapted rust pathogens . Using three differential B . distachyon mapping populations and a quantitative microscopic assay , we investigated colonization resistance to P . striiformis . We found that the ability of four diverse P . striiformis isolates ( three Pst and one Psh ) to colonize B . distachyon leaves is governed by a simple genetic architecture , with resistance largely provided by two major effect QTLs ( Yrr1 and Yrr3 ) . Yrr3 is functional against all Pst and Psh isolates tested , while Yrr1 mediates resistance to the Pst isolates only . These findings show that although plant defense responses to non-adapted pathogens are multilayered , the genetic basis of individual layers of resistance resemble the complexity of host resistance .
In the B . distachyon–Pst interaction , macroscopic infection symptoms manifest as leaf browning ( Fig 1 ) . In a survey of 210 Brachypodium spp . accessions , strong correlation was found between macroscopic leaf browning ( Fig 1A ) and hyphal growth ( percent colonization , pCOL; Fig 1B ) of the Pst isolate 08/21 [28] . While leaf browning and hyphal colonization are correlated traits in diverse germplasm , it is unknown whether a shared genetic architecture controls these phenotypes . We hypothesized that leaf browning and pCOL have a shared genetic architecture . To test this , we assessed these phenotypes in the interaction of Pst isolate 08/21 and three segregating B . distachyon populations . The ABR6 x Bd21 F4:5 population was derived from a cross between accessions collected from geographically distinct regions , i . e . Spain ( accession ABR6 ) and Iraq ( reference accession Bd21 ) [35] . These two accessions differ substantially at the genomic level [35 , 36] . ABR6 does not develop any macroscopic symptoms following Pst infection , whereas Bd21 displays leaf browning and allows hyphal growth ( Fig 1 ) . Leaf browning and pCOL phenotypes in the ABR6 x Bd21 F4:5 population were not normally distributed and heavily skewed towards resistance ( S1A and S1B Fig ) . The segregation pattern for pCOL phenotypes displayed a broader distribution than leaf browning and transgressive segregation for more colonization than Bd21 was observed . Leaf browning and pCOL showed strong correlation ( ρ = 0 . 85; S1C Fig ) . Upon infection with Pst , B . distachyon accessions collected in the western Mediterranean ( predominantly Spain ) displayed greater phenotypic diversity than accessions derived from the eastern Mediterranean ( Turkey to Iraq ) , ranging from large macroscopic lesions to complete microscopic immunity [28] . Three Spanish accessions were selected to generate F2 populations: Foz1 does not develop any infection symptoms and Jer1 only displays very small browning and colonization sites , whereas Luc1 leaves become heavily colonized after infection ( Fig 1 ) . For the Foz1 x Luc1 and Luc1 x Jer1 populations , 188 F2 individuals were evaluated for leaf browning at 14 days post inoculation ( dpi ) and for both leaf browning and pCOL at 23 dpi . Similar to observations on the ABR6 x Bd21 F4:5 population , the leaf browning and pCOL phenotypes were not normally distributed . All three phenotyping results for the Foz1 x Luc1 F2 population were skewed towards resistance ( S2A–S2C Fig ) , as were the phenotyping results for the Luc1 x Jer1 F2 population at 14 dpi ( S2E Fig ) . Interestingly , at 23 dpi leaf browning phenotypes were distributed uniformly and the pCOL phenotypes were almost normally distributed in the Luc1 x Jer1 F2 population ( S2F and S2G Fig ) . At 23 dpi the leaf browning and pCOL phenotypes were correlated with correlation coefficients of 0 . 86 and 0 . 76 for the Foz1 x Luc1 and Luc1 x Jer1 F2 populations , respectively ( S2D and S2H Fig ) . Transgressive segregation towards increased colonization was observed in the Foz1 x Luc1 F2 population and towards increased resistance and colonization in the Luc1 x Jer1 F2 population . Strong correlation of leaf browning and pCOL in segregating populations indicates that these macroscopic and microscopic phenotypes share a similar genetic architecture . This is further supported by the overlapping physical localization of these phenotypes ( Fig 1 ) , suggesting that fungal development contributes to the macroscopic physiological response of infected B . distachyon leaves . To explore the complexity of the genetic architecture of this interaction , SNP-based genetic maps were created for the Foz1 x Luc1 and Luc1 x Jer1 F2 populations . A genetic map was previously developed for the ABR6 x Bd21 F4:5 population [35] . The Foz1 x Luc1 genetic map is based on 179 genotyped F2 lines , contains 101 non-redundant markers , and has a cumulative size of 1 , 430 cM ( S3 Fig ) . The Luc1 x Jer1 genetic map is based on 188 genotyped F2 lines , contains 107 markers , and has a cumulative size of 1 , 446 cM ( S4 Fig ) . Both genetic maps have five linkage groups , corresponding to the five chromosomes of B . distachyon . The quality and integrity of these genetic maps were confirmed by assessing two-way recombination fraction plots for all markers ( S5 Fig ) and by analyzing all chromosomes for segregation distortion and missing data ( S6 Fig ) . Linkage analyses using composite interval mapping were performed on all three mapping populations in order to determine the genetic architecture underlying resistance to the UK Pst isolate 08/21 . For the ABR6 x Bd21 F4:5 population , linkage analyses were performed with phenotypic scores from averaged and individual replicates . Linkage analyses were performed for 179 and 188 genotyped F2 lines in the Foz1 x Luc1 and Luc1 x Jer1 F2 populations , and further validated with 95 F2:3 derived families from the Luc1 x Jer1 F2 population . Both leaf browning and pCOL were assessed for all three populations . Loci that significantly contributed to resistance were designated Yrr ( Yellow rust resistance ) , based on the naming convention for resistance loci in B . distachyon [37] . Two major effect QTLs were found to control leaf browning and pCOL for Pst isolate 08/21 in all three populations . In the ABR6 x Bd21 F4:5 population , a QTL at 328 . 0 cM on chromosome Bd2 controlled 17 . 8% of the phenotypic variation for leaf browning and 24 . 0% of the phenotypic variation for pCOL ( Fig 2 and Table 1 ) . A second QTL with peak markers located around 140 cM on chromosome Bd4 controlled 10 . 9% of the variation for leaf browning and 18 . 3% of the variation for pCOL . These QTLs on chromosomes Bd2 and Bd4 were designated Yrr3 and Yrr1 , respectively ( Fig 3 ) . Only one additional minor effect QTL was detected for Pst isolate 08/21 , which explained 4 . 5% of the phenotypic variation for pCOL in the first replicate ( S7A Fig and S1 Table ) , but was not detected in the second replicate or the averaged dataset . All statistically significant QTLs were contributed by the resistant parent ABR6 . Two-dimensional QTL analysis only uncovered Yrr1 and Yrr3 , which have a non-additive interaction for leaf browning ( p = 8 . 9e-8 ) and pCOL ( p = 6 . 3e-8 ) ( Fig 3 and S1 File ) . The same two major effect QTLs , Yrr1 and Yrr3 , also segregated in the Foz1 x Luc1 F2 population ( Fig 4 ) . However , unlike the ABR6 x Bd21 F4:5 population , Yrr1 was the sole major effect QTL that controlled leaf browning , whereas both Yrr1 and Yrr3 controlled pCOL ( Table 2 and S8A Fig ) . Yrr1 accounted for 37 . 4% and 48 . 9% of the variation observed in the population ( peak markers near 100 cM ) and Yrr3 controlled 28 . 2% of the variation observed for pCOL . A minor effect QTL on chromosome Bd4 contributed to the pCOL phenotype , accounting for 8 . 2% of the variation observed ( Table 2 ) . All three QTLs were contributed by the resistant parent Foz1 . Similar to the ABR6 x Bd21 F4:5 population , Yrr1 and Yrr3 had a non-additive interaction in the Foz1 x Luc1 F2 population ( Fig 4 and S1 File ) . In contrast to the ABR6 x Bd21 F4:5 and Foz1 x Luc1 F2 populations , only one major effect QTL conferred resistance in the Luc1 x Jer1 F2 population ( Figs 4B and S8 ) . Yrr3 explained between 27 . 2% and 46 . 5% of the variation observed for the four phenotypes ( Table 2 ) . The physical location of the QTL corresponds to the same locus observed in the ABR6 x Bd21 F4:5 population . Several minor effect QTLs were only detected in individual replicates . With the exception of the minor effect QTLs on the long arm of chromosome Bd1 and the short arm of chromosome Bd4 , all QTLs were contributed by the resistant parent Jer1 . Therefore , only two major effect QTLs were identified in the three mapping populations in response to Pst isolate 08/21 . To investigate the conservation of Yrr1 and Yrr3 in colonization resistance to diverse Pst isolates , the ABR6 x Bd21 F4:5 population was inoculated with the UK Pst isolates 08/501 and 11/08 . These isolates are genetically distinct to isolate 08/21 and have differential infection outcomes on wheat accessions with various Yr resistance genes [38] . Similar to the phenotypic distributions observed for Pst isolate 08/21 , the infection phenotypes were heavily skewed towards resistance and a strong correlation between leaf browning and pCOL was observed ( S1D–S1I Fig ) . Linkage analyses with the leaf browning phenotype identified Yrr1 and Yrr3 as the two major effect QTLs for both isolates ( Figs 2B , 2C , 3B and 3C and Table 1 ) . Yrr1 accounted for 21 . 7% and 15 . 6% of the phenotypic variation observed upon infection with Pst isolates 08/501 and 11/08 , whereas Yrr3 was responsible for 24 . 6% and 15 . 6% of the phenotypic variation for these two isolates . No additional QTLs were identified in individual replicate experiments ( S7C–S7F Fig and S1 Table ) . These two QTLs also had major effects on pCOL , with Yrr1 contributing 17 . 2% and 14 . 9% and Yrr3 contributing 19 . 4% and 23 . 0% of the phenotypic variation for Pst isolates 08/501 and 11/08 , respectively . The greater resolution obtained with the pCOL phenotype enabled the identification of two additional minor effect QTLs that exhibited isolate specificity . A QTL on the short arm of chromosome Bd4 accounted for 4 . 5% of the variation for Pst isolate 11/08 and 11 . 1% of the variation for Pst isolate 08/501 ( Fig 2B and 2C and Table 1 ) . As this QTL was statistically significant for more than one Pst isolate tested , it was designated Yrr2 . A QTL on chromosome Bd5 was only statistically significant for Pst isolate 11/08 and explained 5 . 3% of the phenotypic variation ( Fig 2C and Table 1 ) . Two-dimensional QTL analysis using the pCOL phenotype for Pst isolates 08/501 and 11/08 found significant pair-wise non-additive interactions between Yrr1 , Yrr2 , and Yrr3 ( S1 File ) . Collectively , these results indicate that Yrr1 and Yrr3 contribute to colonization resistance to all Pst isolates tested , whereas Yrr2 exhibited isolate specificity in its detection . Using three diverse Pst isolates in the ABR6 x Bd21 F4:5 population , we found no evidence of isolate specificity for Yrr1 or Yrr3 . To determine if these major effect loci are specific for Pst or also provide broader resistance to other P . striiformis formae speciales , the mapping population was challenged with the barley-adapted Psh isolate B01/2 . Similar to Pst , phenotypes obtained for Psh were not normally distributed and skewed towards resistance for both leaf browning and pCOL ( S1J and S1K Fig ) . Transgressive segregation was observed with some F4:5 families displaying increased leaf browning and pCOL compared to Bd21 . In contrast to the three Pst isolates tested , ABR6 displayed some macroscopic Psh infection symptoms with an average leaf browning score of 0 . 3 and very limited hyphal colonization ( pCOL of 2% ) . Leaf browning and pCOL phenotypes were correlated with a correlation coefficient of 0 . 63 ( S1L Fig ) . Unlike resistance to Pst in this population , resistance to Psh was predominantly due to a single locus that colocalized with Yrr3 ( Fig 2D ) . This locus explained 28 . 3% and 27 . 3% of the phenotypic variation for leaf browning and pCOL , respectively ( Fig 3D and Table 1 ) . No statistically significant QTLs were observed on chromosome Bd4 using averaged data ( Fig 2D ) or individual replicates ( S7G and S7H Fig ) . Chromosome Bd4 harbors the major effect locus Yrr1 and the minor effect locus Yrr2 , which both confer resistance to Pst isolates . While Yrr3 possesses greater recognition capability towards another P . striiformis forma specialis , Yrr1 and Yrr2 appear to specifically recognize Pst isolates only . Similarity with host systems in the form of major effect loci and isolate specificity prompted us to check the gene content of these QTLs . Several classes of plant immune receptors confer resistance to adapted pathogens , including NB-LRR , kinase-kinase , and LRR-kinase encoding genes [39 , 40] . To date , the majority of cloned resistance genes from host pathosystems encode NB-LRR proteins [41 , 42] . To determine whether NB-LRR are associated with Yrr1 and Yrr3 loci , we evaluated the one-LOD and two-LOD support intervals defined by interval mapping with the pCOL phenotypes ( S2 File ) and examined the gene content of these regions ( S3 and S4 Files ) . Annotated gene models in the Yrr1 and Yrr3 intervals were characterized by comparing the Bd21 reference genome with resequencing data from ABR6 , Luc1 , and Jer1 [36] . Gene expression was assessed by aligning RNAseq reads to the updated gene models from each accession ( S3 and S4 Files ) . Of the 677 genes that were predicted in the Yrr1 interval , most have a resequencing read alignment coverage above 90% ( Table 3 ) . Most gene models present in ABR6 ( 58% ) , Luc1 ( 76% ) , and Jer1 ( 71% ) encoded nonsense or non-synonymous substitutions when compared with the Bd21 reference sequence . No transcripts were detected for 124 ( ABR6 ) to 152 ( Jer1 ) of these gene models , while 74 gene models were not expressed in any of the four accessions . A similar situation was observed for the 789 gene models predicted in the Yrr3 interval . Again , high sequence coverage was obtained for most gene models ( Table 3 ) and many gene models in ABR6 ( 44% ) , Luc1 ( 44% ) , and Jer1 ( 42% ) contained nonsense or non-synonymous substitutions when compared with Bd21 . Transcripts were identified for the vast majority of the gene models in the Yrr3 interval . Both the Yrr1 and Yrr3 loci contained clusters of NB-LRR encoding genes ( Fig 5 and S6 File ) . However , no strong linkage was observed between the NB-LRR encoding genes at the Yrr1 locus ( S6 File ) and subsequent fine-mapping confirmed the lack of NB-LRR candidates for this locus ( see Gilbert et al . ) . In contrast , the combined maximal two-LOD support interval for the pCOL phenotypes at Yrr3 contains five NB-LRR encoding genes and one NB domain encoding gene . These were expressed in all accessions and with the exception of one also possess structural variation in the other three B . distachyon accessions ( S6 File ) . Crucially , the Yrr3 peak markers center around a cluster of two NB-LRR genes ( Bradi2g52437 and Bradi2g52450 ) and a gene encoding an NB domain only ( Bradi2g52430 ) ( Fig 5 and S6 File ) . These data suggest the involvement of NB-LRR encoding genes in Yrr3 resistance , whereas their involvement in Yrr1 mediated resistance remains unclear . The significance of broad-spectrum or isolate-specific effectiveness of resistance is dependent on the genetic diversity of the P . striiformis isolates used . The three Pst isolates used in our study are known to have different avirulence specificities on wheat accessions with various Yr genes [38] . In addition , the two isolates identified in 2008 ( 08/21 and 08/501 ) are more closely related , whereas the 2011 isolate ( 11/08 ) represents a later incursion of Pst into the UK [38] . To understand the genetic relationships of P . striiformis isolates and formae speciales used in this study , we set out to develop a phylogenetic tree based on transcriptomic and genomic resources . We sequenced the transcriptome of barley leaves infected with Psh isolate B01/2 and used publicly available transcriptome or whole genome sequencing datasets for the three UK Pst isolates [38] , the Australian Pst isolate 104E137A- [43] , and the reference genome of the US Pst isolate 78 [44] . Of the annotated genes in the Pst 78 reference genome , 546 genes spanning 562 , 662 bp had sufficient coverage in all datasets . Pairwise sequence comparisons of these genes showed that the two 2008 UK Pst isolates , 08/21 and 08/501 , are almost identical in sequence for these genes and highly similar to the Australian Pst isolate 104E137A- ( Table 4 ) . The US Pst isolate 78 and the UK Pst isolate 11/08 are more diverged from these three isolates . However , consistent with the formae speciales divide , Psh isolate B01/2 is the most divergent isolate in our analysis . These pairwise sequence comparisons were supported by calculating the substitution rates and building a phylogenetic tree of the six P . striiformis isolates using maximum likelihood ( Table 4 and S9 Fig ) . These analyses demonstrate that Yrr3 is a broad-spectrum QTL that recognizes highly divergent P . striiformis isolates .
Nonhost resistance is defined as all accessions from a plant species being resistant to all isolates of a particular pathogen [45 , 46] . For example , rice is considered a nonhost of rusts , as no naturally susceptible rice accessions have been identified [31–33] . A rice mutant that allows some Pst colonization has recently been described [47] , but in the absence of natural or induced variation interspecific crosses may be the last genetic approach at dissecting nonhost resistance . Such experiments are often prevented by interspecies sexual incompatibility and limited by our ability to cross plants [48] . To dissect resistance in phylogenetically more distant species , it is therefore necessary to study resistance within species that fall onto the continuum from host to nonhost , i . e . species in which some accessions allow a degree of infection or colonization , but other accessions are resistant [30 , 48 , 49] . While most B . distachyon accessions possess barriers against P . striiformis colonization , a subset of accessions allows leaf colonization , before additional barriers prevent further disease progression [28] . Several researchers have proposed that the genetic architecture and molecular basis of resistance to non-adapted pathogens are fundamentally different from the gene-for-gene interactions observed in host systems [48 , 50 , 51] . In our study , the two major effect QTLs Yrr1 and Yrr3 control colonization in response to the Pst isolates , whereas only Yrr3 was detected in response to Psh . Our findings therefore highlight a genetic architecture that relies on major effect loci . In the case of Yrr1 the major effect locus also displays isolate specificity . Both of these characteristics are commonly attributed to resistance against adapted pathogens . Barbieri et al . [34] studied the interaction between B . distachyon and the adapted rust P . brachypodii . In a mapping population derived from the accessions Bd3-1 and Bd1-1 the authors identified the loci preventing pustule formation . Analyses of the F2 population and F2:3 families found three QTLs , two of which govern resistance at the seedling stage and one which governs resistance at the seedling stage and an advanced growth stage . Ayliffe et al . [27] studied the inheritance of resistance to the Australian Pst isolate 104E137A- in an F4 population ( BdTR13k x Bd21 ) and an F2 population ( BdTR10h x Tek-4 ) . The authors assessed the extent of pathogen growth based on macroscopic lesions and occasionally also observed pustule formation in the segregating progeny . The segregation ratios of infection symptoms suggested a simple genetic architecture of two genes and one gene restricting pathogen growth in these populations . Subsequent linkage analyses have identified these loci as Yrr1 and Yrr2 ( see Gilbert et al . ) . Taken together , our results challenge existing assumptions about the genetic basis of resistance [48 , 50–55] and support a genetic model of an overlap between resistance to adapted and non-adapted pathogens [3 , 56 , 57] . Extensive diversity exists within barley for the entire range of resistance and susceptibility symptoms following Pst infection [28] . These include complete immunity , varying degrees of chlorosis associated with hyphal colonization , and pustule formation in the absence of chlorosis ( as observed in the adapted interaction between Pst and wheat ) . In contrast , only complete immunity and hyphal colonization were observed in B . distachyon . In a diversity panel of 210 Brachypodium spp . accessions , pustule formation was largely limited to the close allotetraploid relative B . hybridum [28] . Our study of three mapping populations incorporated phenotypically and genetically diverse B . distachyon accessions [36] and diverse P . striiformis isolates . The parental accessions never exhibited pustule formation in our experiments and we only very rarely observed pustule formation in the progeny . Consequently , no phenotypic assay was developed to assess life cycle completion . The multiple barriers to successful pathogen life cycle completion are highlighted by the lack of regular life cycle completion in the transgressively segregating B . distachyon mapping populations . The absence of regular pustule formation shows that even in plants with extensive colonization an additional layer of incompatibility prevents life cycle completion . As the interaction between plant and pathogen is complex , this could be due to the inability of the pathogen to modify the plant in the same manner as the adapted host . During colonization of an adapted host , pathogens secrete effectors to alter the host environment and facilitate infection [11 , 12] . In this scenario , lack of pustule formation could be due to the inability of P . striiformis to create conducive conditions for the transition from growth to reproduction . Alternatively , the pathogen may lack appropriate host plant signals or cues to initiate life cycle progression . Life cycle progression could also be prevented by an active , induced defense response , which would hint at an absence of variation in the gene or genes limiting pustule formation among the B . distachyon accessions studied . In barley , natural variation exists for resistance to P . striiformis that limits the pathogen at pustule formation , but not hyphal colonization [28] . Therefore , it is possible that a conserved gene in B . distachyon may limit the lifecycle completion of P . striiformis . The arms race between host plant and adapted pathogens has resulted in the evolution of numerous resistance genes that often only confer resistance to particular pathogen isolates . Historically , this allowed Biffen to demonstrate that resistance to P . striiformis in wheat follows Mendel’s laws [58] . Many wheat and barley genes that confer resistance to Pst and Psh isolates have been mapped ( see Chen ( 20 ) for a review of Pst resistance loci in wheat ) . These single resistance genes in host systems have often been identified as NB-LRR encoding genes and act in an isolate-specific manner towards the pathogen [41 , 59] . While the role of NB-LRRs in resistance to adapted pathogens is accepted , it remains unclear ( 1 ) how resistance to non-adapted pathogens is maintained in the absence of selection in plants phylogenetically distant to the adapted host and ( 2 ) the capacity of plant immune receptors to contribute to resistance against non-adapted pathogens . Remarkably , we observed characteristics typical for resistance to adapted pathogens in resistance to non-adapted pathogens . Namely , these included the identification of major effect genes , isolate specificity for both major and minor effect QTLs , and NB-LRR gene clusters associated with the identified QTLs . Yrr1 is a major effect QTL controlling leaf browning and hyphal colonization in response to all three Pst isolates tested . However , in the ABR6 x Bd21 F4:5 population this QTL does not control resistance in response to Psh isolate B01/2 . Additionally , all of the minor effect QTLs detected in the ABR6 x Bd21 F4:5 population in response to the three Pst isolates displayed isolate specificity , although this may be associated with limits of statistical detection . Isolate specificity is a common feature in host-pathogen interactions , due to the gene-for-gene interaction in host systems [16] . ETI exerts considerable selection pressure on pathogen populations , which leads to effector loss or modification to avoid detection by the host plants [13] . The emergence of new isolates with an altered effector repertoire consequently leaves the plant with isolate-specific resistance genes [13] . In line with this , candidate genes encoding NB-LRRs and a phosphatase have been identified for two of the B . distachyon loci providing resistance to the adapted pathogen P . brachypodii [60] . As resistance towards non-adapted pathogens is commonly thought to be governed by many , minor effect QTLs reminiscent of basal host resistance [48] we did not expect isolate-specific major effect genes to control the interaction between B . distachyon and Pst and Psh isolates . Of particular interest is the observation that fine-mapping of Yrr1 did not uncover a known class of plant immune receptor , despite exhibiting isolate specificity ( see Gilbert et al . ) . In contrast , tight linkage observed between peak markers at Yrr3 and an NB-LRR cluster opens the possibility that these canonical host immune receptors may contribute to Pst and Psh resistance in B . distachyon [61] . While it has been proposed that resistance to adapted and non-adapted pathogens is inherently different , the genetic architecture of colonization resistance in this intermediate nonhost system is reminiscent of a host system . Moreover , the isolate specificity observed for major and minor effect QTLs and the associated NB-LRR encoding candidate genes suggest that the genetic architectures of resistance to adapted and non-adapted pathogens are structurally coupled and share conserved components . Emphasis has been placed on the intrinsic differences between resistance to adapted and non-adapted pathogens , whereas resistance to non-adapted pathogens may reflect a complete form of resistance , which can draw on a wide range of barriers to limit pathogen ingress and life cycle progression . In the highly-specialized interaction between a host plant and an adapted pathogen , most of these have been overcome and plant and pathogen are left in an evolutionary arms race where the predominant mechanisms of resistance exhibit major effect and isolate specificity .
The ABR6 x Bd21 F4:5 population has been described previously [35] . Seeds for the B . distachyon accessions Luc1 , Jer1 , and Foz1 were kindly provided by Luis A . J . Mur ( Aberystwyth University ) , and F1 plants were confirmed with CAPS markers ( S2 Table ) . To increase F2 seed yield , F1 plants were grown in a prolonged vegetative state to increase biomass before vernalization and flowering [62] . F2 lines were grown from a single cross for both Luc1 x Jer1 and Foz1 x Luc1 . Tissue for DNA extraction and genetic map construction was collected after phenotyping . P . striiformis isolates were collected in the United Kingdom in 2001 ( Psh B01/2 ) , 2008 ( Pst 08/21 and 08/501 ) , and 2011 ( Pst 11/08 ) . Isolates were maintained at the National Institute of Agricultural Botany on susceptible barley and wheat cultivars , respectively , and urediniospores were stored at 6°C after collection . Resequencing data was obtained from the Joint Genome Institute Genome Portal ( http://genome . jgi . doe . gov/ ) for the projects 1000598 ( Luc1 ) , 404166 ( Jer1 ) , and 404167 ( Foz1 ) [36] . These sequence data were produced by the US Department of Energy Joint Genome Institute ( http://www . jgi . doe . gov/ ) in collaboration with the user community . De novo assemblies were created from the raw reads using default settings and parameters of the CLC Assembly Cell ( version 4 . 2 . 0 ) . To ensure an equal genetic distribution across the whole genome , marker positions were selected based on the ABR6 x Bd21 genetic map [35] . A BLAST search was performed with Bd21 sequence based on desired position against the Luc1 , Jer1 , and Foz1 de novo assemblies . The contig sequences for the respective top hits were aligned in Geneious ( version 7 . 1 . 8 ) . SNPs without additional sequence variation in a 160 bp window were selected for KASP marker development ( S7 File ) . To confirm the relative position of the Luc1 x Jer1 and Foz1 x Luc1 markers in the Bd21 reference sequence , a BLAST search was performed with the sequences used for KASP marker development . Markers were named according to the relative SNP position in the Bd21 reference sequence ( version 3 . 1 ) . DNA was extracted from leaf tissue of the phenotyped F2 lines using a CTAB gDNA extraction protocol modified for plate-based extraction [63] . The final Foz1 x Luc1 genetic map is based on 179 genotyped F2 lines and contains 101 non-redundant markers ( S8 File ) . The final Luc1 x Jer1 genetic map is based on 188 genotyped F2 lines and contains 107 markers ( S9 File ) . The quality of the genetic maps was confirmed by analyzing recombination fractions in R/qtl ( version 1 . 33–7 ) and segregation distortion was assessed using chi-square tests with Bonferroni correction for multiple comparisons . For the ABR6 x Bd21 population , 114 F4:5 families were sown in groups of four in 1 L pots containing peat-based compost . For the Foz1 x Luc1 and Luc1 x Jer1 F2 populations , 188 F2 individuals were sown individually in 24-hole trays containing peat-based compost . Plants were grown at 18°C day and 11°C night in a 16 h photoperiod in a controlled environment room . Seedlings were inoculated four weeks after sowing at the four to five leaf stage as described previously [28] . In the ABR6 x Bd21 F4:5 population , leaf browning and pCOL phenotypes were scored at 14 dpi [28] . Phenotypes were scored for each individual in a family and then averaged ( S10 File ) . The two Pst 08/501 replicates consisted of 20 and five plants per F4:5 family , respectively . The two Pst 08/21 replicates consisted of 10 and five plants per F4:5 family , respectively . All replicates of Pst 11/08 and Psh B01/2 consisted of five plants per F4:5 family . In the Foz1 x Luc1 and Luc1 x Jer1 F2 populations , F2 plants were phenotyped individually at 14 dpi for leaf browning and at 23 dpi for leaf browning and pCOL ( S11 File ) . Additionally , 95 Luc1 x Jer1 F2:3 families were phenotyped by growing and inoculating 16 F3 plants in a 1 L pot . Leaf browning phenotypes were scored at 14 dpi for each individual in a family and then averaged . All experiments were performed using a random complete design . Phenotypes were assessed for normality using the Shapiro-Wilk test [64] and Pearson rank correlation coefficients ( ρ ) between leaf browning and pCOL phenotypes were determined using the cor command in R ( v3 . 2 . 2 ) . For the ABR6 x Bd21 F4:5 population , composite interval mapping was performed using an additive model ( H0:H1 ) due to the extensive homozygosity observed at the F4 stage ( ~87 . 5% ) . For the Foz1 x Luc1 and Luc1 x Jer1 F2 populations , composite interval mapping was performed using the model H0:H3 , which includes both additive and dominance effect estimates . QTL Cartographer ( version 1 . 17j ) was used for composite interval mapping with the selection of five background markers , a walking speed of 2 cM , and a window size of 10 cM [65–67] . Statistical significance for QTLs was determined by performing 1 , 000 permutations with reselection of background markers and controlled at α = 0 . 05 [68 , 69] . For the ABR6 x Bd21 F4:5 population , QTL analyses were performed with the phenotyping data from the individual replicates , as well averaged replicates for each isolate . For the Foz1 x Luc1 and Luc1 x Jer1 F2 populations , QTL analyses were performed with the individual phenotyping scores from the F2 individuals and the averaged phenotyping data from the Luc1 x Jer1 F2:3 families . One-LOD and two-LOD support intervals were estimated based on standard interval mapping [70] . Two-dimensional QTL analysis was performed using R/qtl ( 1 . 40–8 ) using scantwo with parameters of step size of 2 . 0 cM , error probability of 0 . 001 , and 128 number of draws for calc . genoprob and sim . geno , with the Haley-Knott method [71] , and significant QTLs identified based on 1 , 000 permutations with α = 0 . 05 . Transcriptome sequencing for Luc1 and Jer1 was performed as described for ABR6 and Bd21 previously [35] . Briefly , plants were grown in a controlled environment room with 16 h of light at 22°C , and fourth and fifth leaves were harvested as soon as the fifth leaf was fully expanded ( approximately four weeks after sowing ) . RNA was extracted using TRI-reagent ( Sigma-Aldrich; T9424 ) according to the manufacturer’s specifications . TruSeq libraries were generated from total RNA and mean insert sizes were 253 bp and 248 bp for Luc1 and Jer1 , respectively . Library preparation and sequencing was performed at The Genome Analysis Centre ( Norwich , UK ) . Sequencing was carried out using 100 bp paired-end reads on an Illumina HiSeq 2500 . Luc1 and Jer1 yielded 134 , 975 , 912 and 136 , 308 , 576 raw reads , respectively . Resequencing data for ABR6 ( project 1079483 ) was obtained from the Joint Genome Institute Genome Portal ( see above for details on Luc1 and Jer1 ) [36] . An identical quality trimming , read alignment , and SNP/InDel calling strategy was applied to the B . distachyon loci that was used for the P . striiformis data set . Assessment of structural variation at the Yrr1 and Yrr3 loci was made by converting the Bd21 reference genome to the alternate genotype through the identification of single nucleotide and small insertion/deletion variation ( S12 File ) . QKgenome_conversion . py was used with threshold requirements for read coverage was set at 20 reads and allelic variant frequency of greater than 95% . Tophat ( version v2 . 0 . 9 ) was used for splice alignment of RNAseq datasets for Bd21 , ABR6 , Luc1 , and Jer1 to their respective converted reference genomes . FeatureCounts ( version v1 . 5 . 1 ) with commands “-M -O -t exon” was used to identify the number of RNAseq reads mapping to individual gene models . To identify canonical NB-LRR encoding resistance genes , the most recent Bd21 reference genome annotation ( version 3 . 1 ) was searched for genes annotated as encoding NB-ARC domains ( Pfam PF00931 ) and/or belong to the LRR gene family ( PANTHER PTHR23155 ) . The identified genes were largely consistent with annotations of previous B . distachyon reference genome versions [72 , 73] . The susceptible barley accession Aramir ( PI 399482 ) was inoculated with Psh B01/2 as described previously [28] . Plants exhibited a McNeal score of 8 ( abundant sporulation with chlorosis ) [74] . Infected leaves were harvested 12 dpi and flash frozen in liquid nitrogen . RNA was extracted using TRI-reagent ( Sigma-Aldrich; T9424 ) according to the manufacturer’s specifications . TruSeq libraries were generated from total RNA and mean insert sizes were 280 bp . Library preparation and sequencing was performed at The Genome Analysis Centre ( Norwich , UK ) . Sequencing was carried out using 150 bp paired-end reads on an Illumina HiSeq 2500 and yielded 38 , 636 , 376 raw reads . The Pst 78 reference sequence assembly and raw sequencing reads were obtained from the Broad Institute ( GenBank BioProject PRJNA41279 ) [44] . The Pst 08/21 , Pst 08/501 , and Pst 11/08 genome and transcriptome raw sequencing reads were obtained from the GenBank BioProjects PRJNA256347 and PRJNA257181 [38] . The Pst 104E137A- Illumina RNAseq reads from germinated spores and haustoria were obtained from GenBank BioProject PRJNA176472 [43] . Illumina reads were quality controlled using Trimmomatic ( version 0 . 33 ) with the following parameters: ILLUMINACLIP:TruSeq3-PE . fa:2:30:10 LEADING:5 TRAILING:5 SLIDINGWINDOW:4:15 MINLEN:80 . Alignments to the Pst 78 reference assembly were performed with bwa mem ( version 0 . 7 . 5a-r405 ) with default parameters for gDNA samples and Tophat ( version 2 . 0 . 9 ) with default parameters was used for splice alignment mapping of RNAseq samples . Samtools ( version 0 . 1 . 19-96b5f2294a ) was used to convert sam into bam files ( samtools view ) with the requirement that reads mapped in a proper pair ( -f2 ) , to sort the bam file ( samtools sort ) , to remove duplicate reads ( samtools rmdup ) , and to generate an mpileup file ( samtools mpileup ) . Coverage of reads was determined using bedtools ( version v2 . 17 . 0; bedtools genomecov -d -split ) . SNPs and InDels were called using VarScan ( version 2 . 3 . 8 ) with default parameters . The QKgenome suite ( version 1 . 1 . 2 ) of Python scripts were used to identify SNPs from diverse gDNA and RNA sequenced Pst and Psh isolates . QKgenome_conversion . py was used with the requirement of a read depth of 20 across the entire gene model for all isolates studied . In addition , only the first gene model for each gene was used to avoid duplication of polymorphic sites by including splice variants . SNPs and InDels were called based on a frequency threshold of 90% ( i . e . only homokaryotic polymorphisms were included ) . All genes with InDels that disrupted the coding sequence were not included in the analysis . A multiple sequence alignment of polymorphic sites was generated using QKgenome_phylogeny . py . The phylogenetic tree was constructed with the GTR CAT nucleotide model , rapid hill-climbing algorithm , and 1 , 000 bootstrap replicates using RAxML ( version 8 . 2 . 9 ) . Sequencing data were deposited in NCBI under BioProjects PRJNA376485 ( B . distachyon ) and PRJNA376252 ( barley/Psh ) . Individual RNAseq reads include accession numbers SRR5279889 ( Luc1 ) , SRR5279890 ( Jer1 ) , SRR5279891 ( Foz1 ) , and SRR5277779 ( Psh B01/2 ) . De novo genome assemblies of B . distachyon were deposited in figshare ( https://figshare . com/projects/The_genetic_architecture_of_colonization_resistance_in_Brachypodium_distachyon_to_non-adapted_stripe_rust_Puccinia_striiformis_isolates/29752 ) . The QKgenome suite of Python scripts described in this manuscript has been deposited on GitHub ( https://github . com/matthewmoscou/QKgenome ) . | Plants are constantly exposed to a multitude of potential pathogens but remain immune to most of these due to a multilayered immune system . Pathogens have specialized by adapting to certain host plants and their defense barriers . Most of our understanding of plant-pathogen interactions stems from these highly specialized interactions , because they are characterized by qualitative interactions ( resistant or susceptible ) . It has generally been assumed that the genetic and molecular basis of resistance to non-adapted pathogens is fundamentally different , as either no variation exists in a species ( complete immunity ) or variation encompasses only early pathogen invasion ( colonization ) , but not full susceptibility . We have studied the interaction between the agronomically important fungal stripe rust pathogen ( Puccinia striiformis ) of wheat and barley with the wild grass species Brachypodium distachyon . Rust infections consist of two stages: colonization of plant tissues followed by a reproductive phase . We identified natural variation for the degree of P . striiformis colonization in different B . distachyon accessions and dissected the genetic architecture controlling resistance at this infection stage . QTLs conferring resistance possessed several characteristics similar to adapted host systems , indicating that resistance to adapted and non-adapted pathogens are not intrinsically different . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"plant",
"anatomy",
"quantitative",
"trait",
"loci",
"population",
"genetics",
"developmental",
"biology",
"plant",
"science",
"plant",
"pathology",
"molecular",
"biology",
"techniques",
"population",
"biology",
"research",
"and",
"analysis",
"methods",
"gene",
"mapping... | 2018 | The genetic architecture of colonization resistance in Brachypodium distachyon to non-adapted stripe rust (Puccinia striiformis) isolates |
Throughout evolution , cytomegaloviruses ( CMVs ) have been capturing genes from their hosts , employing the derived proteins to evade host immune defenses . We have recently reported the presence of a number of CD48 homologs ( vCD48s ) encoded by different pathogenic viruses , including several CMVs . However , their properties and biological relevance remain as yet unexplored . CD48 , a cosignaling molecule expressed on the surface of most hematopoietic cells , modulates the function of natural killer ( NK ) and other cytotoxic cells by binding to its natural ligand 2B4 ( CD244 ) . Here , we have characterized A43 , the vCD48 exhibiting the highest amino acid sequence identity with host CD48 . A43 , which is encoded by owl monkey CMV , is a soluble molecule released from the cell after being proteolytically processed through its membrane proximal region . A43 is expressed with immediate-early kinetics , yielding a protein that is rapidly detected in the supernatant of infected cells . Remarkably , surface plasmon resonance assays revealed that this viral protein binds to host 2B4 with high affinity and slow dissociation rates . We demonstrate that soluble A43 is capable to abrogate host CD48:2B4 interactions . Moreover , A43 strongly binds to human 2B4 and prevents 2B4-mediated NK-cell adhesion to target cells , therefore reducing the formation of conjugates and the establishment of immunological synapses between human NK cells and CD48-expressing target cells . Furthermore , in the presence of this viral protein , 2B4-mediated cytotoxicity and IFN-γ production by NK cells are severely impaired . In summary , we propose that A43 may serve as a functional soluble CD48 decoy receptor by binding and masking 2B4 , thereby impeding effective NK cell immune control during viral infections . Thus , our findings provide a novel example of the immune evasion strategies developed by viruses .
Natural killer ( NK ) cells are circulating lymphocytes that play a pivotal role in the rapid recognition and control of viral infections . NK functions are regulated by a repertoire of specific receptors that , upon engagement with their respective ligands on target cells , transmit stimulatory or inhibitory signals [1] . The net balance of activating/inhibitory signals determines whether the NK cell will initiate its cytolytic activity through the degranulation of specialized secretory lysosomes into the immune synapse , ultimately causing the destruction of the target cell . One such receptor is 2B4 ( or CD244 ) , a member of the signaling lymphocyte activation molecule ( SLAM ) family of the immunoglobulin ( Ig ) superfamily [2] . In human NK cells , 2B4 predominantly provides co-stimulatory signals , activating NK cytotoxicity and cytokine production [3] . 2B4 interacts with CD48 , another member of the SLAM family that is broadly expressed on the surface of most hematopoietic cells [4–6] . Both receptors contain an ectodomain composed of an N-terminal Ig membrane-distal variable ( IgV ) domain followed by an Ig constant-2-set domain , characterized by conserved cysteines . However , while CD48 is a glycosyl-phosphatidylinositol ( GPI ) -anchored protein , 2B4 is a type I transmembrane molecule that contains four copies of the immune receptor tyrosine-based switch motif ( ITSM ) in its cytoplasmic tail [7 , 8] . 2B4 engagement by CD48 occurs through their N-terminal IgV domains , resulting in the recruitment of specific adaptor molecules by the ITSM motifs followed by signaling transduction events that ultimately modulate immune responses [9] . In addition to NK cells , 2B4 is expressed at lower levels on other cytotoxic cells , including CD8+ T cells , γδ T cells , basophils and eosinophils [10 , 11] . Therefore , the 2B4:CD48 interaction also contributes to the regulation of additional aspects of the innate and adaptive immune responses . NK cells are crucial for the successful control of infections by cytomegaloviruses ( CMVs ) . Consequently , these pathogens have evolved a wealth of strategies to hinder or abrogate NK functions [12–15] . Most of these strategies are based on mechanisms designed to avoid recognition of infected cells by activating NK cell receptors or to trigger inhibitory NK cell signaling . To this end , within their large and densely packed genomes , CMVs encode multiple immunosubversive proteins , meticulously shaped for these purposes [16 , 17] . Some of these immunoevasins are of cellular origin , having been captured from their hosts at different times during their co-evolution [18–20] , and often employed to mimic or interfere with the original host function . Given the relevance of 2B4 for the regulation of NK cell activity , it is not surprising that among their immune evasion strategies CMVs have developed mechanisms to counteract CD48:2B4 interactions . In this regard , we have reported that CMVs reduce the expression of the CD48 receptor on the surface of infected target cells [21] . Indeed , our group identified , in murine CMV , the mucin-like m154 protein as the viral molecule responsible for reducing CD48 macrophage surface expression . We showed that m154 targets CD48 for degradation and helps to impair antiviral NK-triggered cell responses , thereby meliorating viral growth in vivo . Moreover , we have recently identified several CD48 homologs ( vCD48s ) among CMVs [22 , 23] . In particular , two primate cytomegaloviruses , SMCMV that infects squirrel monkeys and OMCMV that infects owl monkeys , encode copies of vCD48s , which arose from a unique gene-capture event in an ancestor CMV genome and subsequent gene duplication episodes . One of these vCD48s , A43 , which shows the highest homology to host CD48 , is a secreted molecule that preserves ligand-binding properties , being able to interact with host 2B4 [23] . However , to date , the biological significance of A43 and the rest of these vCD48s remains unknown . Thus , the potential of these molecules to function in immune evasion mechanisms warrants further investigation . Here , we have characterized the OMCMV encoded CD48 homolog A43 . We show that this viral protein can act as a soluble decoy CD48 receptor , protecting cells against NK cell-mediated cytotoxicity . By binding with high affinity to 2B4 , this viral protein is capable of preventing CD48:2B4 interactions . In addition , we present that A43 limits human NK cell adhesion to CD48-expressing target cells , inhibiting the formation of the NK cell immunological synapse . Thus , our findings point to a novel class of immunoevasins .
We first sought to analyze the expression of A43 in the context of the infection . To determine the kinetic class of this viral protein , owl monkey kidney ( OMK ) epithelial cells were mock-infected or infected with OMCMV at a moi of 1 in the absence or presence of two chemical inhibitors , cycloheximide ( CHX ) or phosphonoacetic acid ( PAA ) , which prevent translation and viral DNA replication , respectively . Total RNA was extracted from the cultures and reverse transcriptase-mediated PCR ( RT-PCR ) analysis was performed using specific primers for this vCD48 . As illustrated in Fig 1A , A43 was detected under all conditions tested , including in the presence of CHX , when the immediate-early OMCMV gene IE1 was abundantly expressed , but not the early UL54 viral polymerase or the late UL73 virion envelope glycoprotein N genes . Hence , A43 can be considered an immediate early gene . We then examined the presence of the viral protein in infected cells . OMK cells , mock-infected or infected with OMCMV , were analyzed at different days after infection by flow cytometry using a polyclonal antibody that we raised against the viral protein . As shown in Fig 1B , at the three time points examined , 1 , 3 and 5 days post infection , A43 could be detected at very low amounts at the surface of the infected cells . To directly investigate whether A43 was shed during infection , culture supernatants were collected at different times post-infection over a 9 day period and examined by ELISA using the A43 polyclonal antibody . Remarkably , and depicted in Fig 1C , A43 was found in the extracellular milieu since the first 6 h after infection and accumulated over time , thereby indicating that the A43 protein is released as a soluble molecule soon after infection . Ectodomain shedding of transmembrane proteins primarily occurs by proteolytic cleavage at sites in close proximity to the cell surface . In this process , the length of the proximal region and the structure of the cleavage site region of the protein tend to be more relevant than the specific amino acid sequence [24] . A43 contains a 12 aa-long membrane proximal ( MP ) region ( Fig 2A ) . Therefore , to gain insight into the cleavage site of A43 , and assess whether the integrity of the MP region was fundamental in this process , different mutants of the viral protein containing deletions or point mutations in this region were engineered . To this end , we used an HA-tagged A43 plasmid , encoding the viral protein with the hemagglutinin ( HA ) tag positioned immediately after its signal peptidase cleavage site . In two of the variants the length of the MP region was shortened , deleting either the six aa immediately juxtaposed to the transmembrane domain ( aa 214–219 , in plasmid HA-A43 ΔT-R ) , or the six aa more distal ( aa 220–225 , in plasmid HA-A43 ΔG-A ) to this region ( Fig 2A ) . In addition , the lysine at position 216 within the A43 MP region was substituted by a proline , which would be expected to disrupt its secondary structure , or alternatively by an alanine as a control , in mutant plasmids HA-A43 K216P and HA-A43 K216A , respectively ( Fig 2A ) . COS-7 cells were transiently transfected with wild type HA-A43 or the constructed plasmids and the effects of the removed amino acid stretches or the mutated lysine residue were quantitatively assessed by measuring HA-A43 surface expression by flow cytometry . Surprisingly , there were no substantial differences in cell surface expression between the deletion mutants HA-A43 ΔT-R or HA-A43 ΔG-A and the wild type HA-A43 protein ( Fig 2B ) . In contrast , a pronounced increase of surface staining was observed in cells transfected with the HA-A43 K216P mutant , but not with HA-A43 K216A . Consistent with these results , when we directly evaluated the release of soluble A43 by a sandwich ELISA of supernatants from COS-7 cells transfected with HA-A43 or HA-A43 K216P , at various time points , constitutive shedding of HA-A43 , but not of HA-A43 K216P , was evident at every time point analyzed , from day 1 to 9 ( Fig 2C ) . Altogether , these results indicate the absence of key particular MP sequences required for the proteolytic cleavage of A43 , and suggest that mutation of lysine 216 to a proline may induce a conformational change resulting in impaired shedding of A43 . To study in more detail the interaction of A43 with 2B4 , we generated an A43-Fc fusion protein , composed by the extracellular region of A43 fused to the Fc domain of the human IgG1 . We first confirmed by flow cytometry the capacity of A43-Fc to bind host ( Aotus trivirgatus ) 2B4 by incubating increasing amounts of the fusion protein with COS-7 cells transiently transfected with host 2B4 , using an expression vector encoding an N-terminal HA-tagged version of this receptor . As shown in Fig 3A ( left graph ) , A43-Fc efficiently interacted , in a dose-dependent manner , with cell surface host 2B4 . Indeed , the observed number of 2B4-transfected COS-7 cells binding A43-Fc ( 80 . 90% ) was comparable to the percentage of COS-7 cells expressing 2B4 ( 85 . 19% ) at the highest dose of A43-Fc evaluated ( Fig 3A , compare upper and bottom panels in the right ) . Next , we carried out surface plasmon resonance ( SPR ) kinetic analyses to quantify in real time the biochemical interaction of A43 and host 2B4 . To this end , host 2B4-Fc fusion protein was immobilized in the sensor chip and the capacity to bind different concentrations of soluble A43-Fc was examined . In addition , host CD48-Fc was employed as a control , allowing us to compare the interaction of A43:host 2B4 with that of host CD48:host 2B4 , as illustrated in Fig 3B . The binding kinetics revealed that A43 interacts with host 2B4 with an association rate ( Ka ) of 1 . 16 x 105 M-1 s-1 and a dissociation rate ( Kd ) of 1 . 88 x 10−4 s-1 , yielding an equilibrium dissociation constant ( KD ) of 1 . 3 nM . These results evidenced a high affinity and stable interaction between A43 and host 2B4 . When compared to control host CD48 , the kinetic measurements were quite similar ( KD of 3 . 3 nM ) , with A43 binding host 2B4 with a 2 . 4-fold higher affinity , and dissociating 2 . 6-times slower . We next explored whether soluble A43 was able to block the host CD48:2B4 interaction . To directly prove this , we employed an ELISA assay , in which the host 2B4-Fc protein immobilized on the plate was exposed to increasing amounts of A43-Fc and subsequently incubated with biotinylated host CD48-Fc . As shown in Fig 4A ( left graph ) , the A43-Fc protein markedly decreased the interaction of host CD48 to its countereceptor . Similar results were also obtained when the assays were performed using supernatants of COS-7 cells transfected with HA-A43 ( Fig 4A , middle graph ) . Moreover , we assessed whether this viral protein produced during infection was also capable to prevent the interaction of host CD48 with host 2B4 . Notably , we found that A43 containing supernatants from infected cells also resulted in an efficient and dose-dependent inhibition of the binding of biotinylated host CD48-Fc to the host 2B4-Fc protein ( Fig 4A , right graph ) . We further studied the ability of the A43-Fc fusion protein to compete the host 2B4:CD48 interaction when both receptors were expressed at the cell surface , by carrying out cell binding assays . We employed host EBV-transformed B lymphocytes , which express CD48 ( Fig 4B ) , and COS-7 cells transiently transfected with host 2B4 . The binding of these B lymphocytes to host 2B4-expressing COS-7 cells or untransfected control COS-7 cells was assessed in the presence or absence of A43 , by analyzing a number of fields under the light microscope . B lymphocytes showed a specific binding to host 2B4-expressing COS-7 cells ( Fig 4C , compare panels a and b ) . Importantly , when the host 2B4+ COS-7 cells were incubated with A43-Fc , a substantial decrease of the interaction between these cells and B lymphocytes was observed ( Fig 4C , panel d ) , whereas this interaction was unaltered after treatment with an unrelated Fc protein ( CTL-Fc; Fig 4C , panel c ) . To quantitatively measure these effects , cell binding assays were performed using host B lymphocytes previously labelled with calcein , allowing the determination of the number of bound B cells by flow cytometry . As observed before , while the binding of B cells to COS-7 cells transfected with host 2B4 was not significantly affected by the presence of the control fusion protein , it markedly decreased after addition of A43-Fc , yielding an inhibition of around 60% of the host 2B4-induced binding ( Fig 4D ) . Taken together , these findings demonstrate that soluble purified A43 can impair the interaction between cells expressing host 2B4 and host CD48+ cells . We then decided to investigate whether A43 displayed cross-species receptor binding properties , by analyzing its interaction with human and mouse 2B4 ( h2B4 and m2B4 , respectively ) . To this end , COS-7 cells transiently transfected with the plasmid HA-A43 K216P , which allows high expression levels of A43 at the cell surface , were tested by flow cytometry in binding assays performed with the h2B4-Fc or m2B4-Fc fusion proteins . Fig 5A shows that while m2B4-Fc did not interact with A43 , a strong binding activity of h2B4-Fc to the viral protein was observed . Human 2B4 is mainly expressed in NK cells , although it is also present at lower levels on other cytotoxic cells , such as CD8+ T cells , γδ T cells , basophils , and eosinophils . Taking this in consideration , we evaluated the capacity of A43-Fc to bind different human cell lines constitutively expressing h2B4 , either derived from NK cells ( NK-92 and YT ) or T lymphocytes ( HSB2 ) ( Fig 5B ) . In these assays , COS-7 cells transiently transfected with h2B4 were used as a positive control , whereas the human T lymphoblast cell line MOLT-4 that does not express h2B4 and untransfected COS-7 cells were employed as negative controls . As illustrated in Fig 5C , A43 displayed a clear reactivity with the four cell lines tested that expressed 2B4 ( NK-92 , YT , HSB2 , and h2B4-transfected COS-7 cells ) , staining them at nearly the same extension than the h2B4 specific monoclonal antibody ( mAb; compare MFI values in corresponding panels in Fig 5B and 5C ) . In contrast , no interaction of the viral protein was detected for the human cell line MOLT-4 or the untransfected COS-7 cells . We also carried out SPR kinetic analyses to quantify in real time the A43:h2B4 interaction in a similar way as described for the interaction of A43 with host 2B4 , but in this case immobilizing h2B4-Fc in the sensor chip . Thus , we assessed the ability of h2B4 to bind different concentrations of soluble A43-Fc , using hCD48-Fc as a control . Fig 5D , shows an example of the sensorgrams fittings generated to obtain the affinity constants . The binding kinetics indicated that A43 binds to h2B4 with an association rate ( Ka ) of 1 . 69 x 105 M-1 s-1 and a dissociation rate ( Kd ) of 4 . 05 x 10−4 s-1 , resulting in an equilibrium dissociation constant ( KD ) of 3 . 6 nM . As we previously found for the binding of A43 to host 2B4 , these results evidenced an interaction of high affinity and stability between A43 and h2B4 . Remarkably however , and in contrast to the findings obtained in the host context , A43 was found to bind h2B4 with a 6-fold higher affinity and dissociate 55-fold slower than hCD48 . The observation that A43 interacts with human 2B4 does not only represent an important finding by itself , but in addition enabled us to pursue functional studies using 2B4+ human NK cells . Thus , in the next set of experiments we sought to determine whether soluble A43 was a functionally active molecule . Since 2B4 is an important co-activating receptor for NK cell function , we analyzed the capacity of A43 to interfere with NK cell activities . First , we confirmed the potential of A43-Fc to block the hCD48:h2B4 interaction , performing cytofluorimetric assays using the human NK-92 cell line and biotinylated hCD48-Fc . As shown in Fig 6A ( left graph ) , A43-Fc was able to efficiently compete in a dose-dependent manner the binding of hCD48 to h2B4 . Comparable results were obtained by sandwich ELISA , when the ability of different doses of A43-Fc to block the binding between the h2B4-Fc protein ( that coated the plates ) and biotinylated hCD48-Fc was assessed ( Fig 6A , right graph ) . Target cell adhesion is an indispensable event for NK cell activation , and the 2B4:CD48 interaction has been shown to contribute to this process [25] . To evaluate the effects of A43 on NK cell adhesion , we performed cell binding assays using COS-7 cells transiently transfected with hCD48 and NK-92 cells . The binding of NK-92 cells to hCD48-expressing COS-7 cells or untransfected control COS-7 cells was determined in the presence or absence of A43 , by examining a number of fields under the microscope . As expected , NK-92 cells exhibited a specific binding to hCD48-expressing COS-7 cells ( Fig 6B , compare panels a and b ) . Rosettes of NK cells adhering to the cell surface of the COS-7 cells transfected with hCD48 could be observed . The specificity of this CD48-induced adhesion was further confirmed using a blocking anti-h2B4 mAb . In this case , binding abrogation to almost background levels ( observed in COS-7 cells and shown in Fig 6B , panel a ) was found upon preincubation of the NK-92 cells with the anti-h2B4 mAb , but not with an isotype control mAb ( see panels c and d in Fig 6B ) . Notably , when NK-92 cells were exposed to A43-Fc , a marked reduced interaction between these cells and hCD48-transfected COS-7 cells was observed , with barely any visible rosette of NK-92 cells in the cultures ( Fig 6B , panel f ) . In contrast , cell adhesion remained unaffected after incubation with an unrelated Fc protein ( CTL-Fc; Fig 6B , panel e ) . We then proceeded to quantitatively measure these effects by flow cytometry using NK-92 cells labelled with calcein . As observed before , while the binding of NK-92 cells to COS-7 cells transfected with hCD48 was not altered significantly after incubating the NK-92 cells with an unrelated fusion protein , it substantially decreased after addition of A43-Fc , resulting in an inhibition of around 50% of the hCD48-induced binding ( Fig 6C ) . Again , the specificity of the binding , abrogating nearly 80% of the hCD48-induced adhesion , was proven with the mAb directed to h2B4 . These data demonstrate that soluble A43 can function interfering with h2B4-mediated NK cell adhesion to target cells . Formation of conjugates between NK and target cells is an important step required for efficient NK cell activation and subsequent cytolytic functions . We therefore examined the impact of the binding of soluble A43 to NK cells on conjugate formation . To this end , we transfected HEK cells with hCD48 ( see Fig 7A ) and employed a flow cytometry-based assay that enabled to determine the frequency of the stable conjugates formed between these target cells and YT cells after being labelled with two distinct fluorophores ( cell tracker for YT cells and CFSE for target cells ) . Conjugates were identified after co-incubating for 10 min these labelled cells in suspension as double fluorescent events . As illustrated in Fig 7B , after this incubation time , the expression of CD48 on the surface of HEK cells conferred to these cells the ability to form conjugates with YT cells . Notably , exposure of the YT cells to A43-Fc , but not of an unrelated fusion protein , provoked a pronounced reduction of conjugate formation with hCD48-expressing HEK cells . As a control , the anti-h2B4 blocking mAb was also included in the assay and shown to drastically diminish the proportion of conjugates to levels comparable to those observed after incubation with HEK cells ( Fig 7B ) . Identical results were observed when the assays were conducted with NK-92 cells instead of YT cells , further confirming that the binding of A43 to h2B4 in NK cells severely affects their ability to form conjugates with target cells ( Fig 7C ) . Following the interaction of the NK cell with the target cell , the establishment of a highly stable immunological synapse is needed to ensure the directed release of perforin-granzyme granules toward the target cell . 2B4 has been shown to rapidly accumulate at the cell-cell contact site and participate in the formation of the lytic NK cell immune synapse [26] . Thus , we next assessed the potential of A43 to interfere with the establishment of the mature cytotoxic NK cell immune synapse . To investigate this , we performed immunofluorescence microscopy assays on conjugates established between YT cells and target HEK cells transfected with hCD48 or the corresponding empty plasmid and labelled with the cell tracker , using phalloidin and an anti-perforin mAb to visualize NK polymerized actin and cytotoxic granules , respectively . We examined the occurrence of three different stages in the progression of the formation of the functional NK cell lytic immune synapse ( Fig 8A ) . Stage 0 , corresponding to conjugates in which actin and cytotoxic granules were not polarized toward the synapse . Stage 1 , corresponding to conjugates in which polymerized actin , but not cytotoxic granules , accumulated at the synapse . And stage 2 , corresponding to conjugates in which both polymerized actin and cytotoxic granules had reached the synapse . In this latter case , co-localization of phalloidin ( actin ) and perforin could be clearly visualized ( right bottom panel in Fig 8A ) . We first evaluated the ability of the YT cells to form an immune synapse with hCD48-transfected HEK cells as compared to control HEK cells , analyzing the same number of conjugates . Fig 8B ( upper panel ) shows an augmented proportion of conjugates exhibiting fully polarized granules ( stage 2 ) upon expression of hCD48 on the surface of the target cells . This effect was associated with a concomitant decrease of conjugates in stage 0 , lacking acting polymerization and granule polarization . These data are consistent with a role of hCD48 promoting the progression of the NK immune synapse . Remarkably , when YT cells were pre-treated with the A43-Fc protein , a clear decrease of conjugates with hCD48-transfected HEK cells in stage 2 was found as compared to those established by YT cells pre-incubated with an unrelated Fc fusion protein ( Fig 8B , bottom panel ) . In conclusion , A43 inhibits the formation of the mature cytotoxic NK cell immune synapse . We then investigated whether soluble A43 had functional consequences on NK cell-mediated lysis of target cells . We performed cytotoxicity assays using as effectors the YT cell line at various E/T ratios , and untransfected or hCD48-transfected HEK cells stained with calcein , as targets . As seen in Fig 9A , in these assays the killing displayed by the YT cells toward untransfected HEK cells was minimal . In contrast , the expression of hCD48 substantially stimulated killing of HEK cells by the YT cells at all the different ratios tested ( Fig 9A ) . We then assessed the effects of pre-treating YT cells with the A43 fusion protein or the anti-h2B4 mAb as a control . As expected , a drastic reduction of the cytotoxicity was observed when the NK cell receptor was masked with the anti-h2B4 mAb , confirming that NK cell-mediated lysis of hCD48-transfected HEK cells is h2B4 dependent . Remarkably , preincubation of the YT cells with the A43-Fc protein , but not with an unrelated Fc fusion protein , significantly inhibited NK-cell-mediated lysis at every E/T ratio tested in 3 independent experiments ( Fig 9A ) . Similar results were obtained when employing the NK-92 cell line as effector cells , although in this case the base-line killing of these cells toward untransfected HEK cells was higher , slightly reducing the window in which to visualize the contribution of the h2B4-NK dependent lysis ( Fig 9B ) . Of note , the blockade exerted by the viral protein did not require the pre-treatment of the NK cells with the A43-Fc protein , since when the viral protein was simultaneously added with the hCD48-expressing target cells to the effector cells , the h2B4-dependent NK cytotoxicity was inhibited at the same levels ( Fig 9C ) . To assess if reduced NK cytolytic functions by A43 were also associated with diminished NK cytokine production , we examined the ability of the YT cells to secrete IFN-γ during the 4h cytotoxicity assay . As shown in Fig 9D , the expression of hCD48 in HEK cells significantly enhanced IFN-γ secretion by YT cells as compared with the untransfected HEK cells . Furthermore , A43-Fc , but not the CTL-Fc protein , caused a profound decrease of the CD48-stimulated YT cell production of IFN-γ . We also examined the ability of soluble A43 to functionally block primary human NK cells . To this end , and to avoid potential complications derived from the use of human Fc fusion proteins and the Fc receptors present on the surface of the primary NK cells , we constructed a new version of the A43-Fc fusion protein , named A43-Fc* , containing six specific residues of the Fc region mutated that abrogate its binding to Fc receptors [27] . When examined by flow cytometry , A43-Fc* , but not an unrelated control Fc* fusion protein ( CTL-Fc* ) , was capable to interact with the isolated primary blood NK cells ( Fig 10A ) . As illustrated in Fig 10B , hCD48-transfected HEK cells were more susceptible to lysis by primary NK cells than untransfected HEK cells . Moreover , as it occurred in YT and NK-92 cells , a pronounced inhibition of the cytotoxicity was observed when the primary NK cells were preincubated with the A43-Fc* protein , but not with the CTL-Fc* protein , at the two different E/T ratio tested . In addition , treatment with A43-Fc* resulted in a significant and specific reduction of the hCD48-dependent NK cell production of IFN-γ ( Fig 10C ) . Due to the impossibility of obtaining host NK cells , and therefore in the absence of a cell system that allowed us to assess the activity of A43 on host NK cell functions , we analyzed if A43 was able to inhibit human NK cell cytotoxicity towards target cells expressing host CD48 . For this purpose , we first confirmed by flow cytometry that YT cells could bind host CD48 ( Fig 11A ) . Thereafter , we performed cytotoxicity assays employing as targets either HEK cells transfected with host CD48 or the immortalized host B lymphocytes . As shown in Fig 11B , YT cell lysis was higher in host CD48-expressing HEK cells than in HEK cells . Importantly , a substantial and specific block of the h2B4-dependent NK cell killing was observed when the YT cells were treated with A43-Fc at the two E/T ratios analyzed . Moreover , comparable results were observed when the host B lymphocytes were used as targets ( Fig 11C ) . Thus , altogether the results indicate that soluble A43 may act protecting CD48-expressing cells from 2B4-dependent NK cell production of IFN-γ and cytotoxicity . Lastly , we explored whether the viral A43 protein secreted during infection could also interfere with NK cell-mediated lysis . To this end , we measured YT or NK-92 cell-mediated cytotoxicity against hCD48-expressing HEK cells in the presence of the supernatant from OMCMV infected cells , using the supernatant from mock infected cultures as controls . As shown in Fig 12A , the supernatant from the OMK infected cells , but not from OMK mock-infected cells , led to a marked reduction of the 2B4-dependent YT and NK-92 cell-mediated killing at the two different E/T ratios evaluated . These observations were in corcordance with the specific ability of the infected culture supernatant to substantially inhibit the binding of hCD48-Fc to the two NK cell lines , in particular at the highest concentration evaluated , which was the one used in the cytotoxicity assay ( Fig 12B ) . Furthermore , to demonstrate that the viral A43 protein present in the supernatant fom the OMCMV infected cells was responsible of the blockade of the h2B4-dependent NK cell cytotoxicity , we depleted A43 from the supernatant of the infected cultures by pre-adsorbing it with the anti-A43 polyclonal antibody attached to beads . As shown in Fig 12C , the supernatant from OMK infected cells depleted of A43 was not longer capable to decrease the h2B4-dependent YT cell-mediated killing . Thus , these results demonstrate that the A43 produced during infection is capable to impair NK cell cytotoxicicty .
Acquisition of host defense genes via horizontal gene transfer constitutes an important process by which CMVs develop new immunomodulatory strategies [28–30] . The existence of different versions of several SLAM family receptors , including CD48 , among CMV genomes indicates that the targeting of SLAM-mediated immune responses is evolutionarily advantageous for these pathogens [22 , 23] . Here , we have performed an in-depth functional analysis of the OMCMV encoded CD48 homolog A43 , showing that it serves as a decoy molecule blocking 2B4-mediated NK cell cytotoxic responses . Our assays demonstrate that A43 is an immediate early gene that gives rise to a protein released from the infected cells within hours after infection . SPR assays indicated that soluble A43 binds to host 2B4 with high affinity and stability , exhibiting a KD of 1 . 3 nM , in a comparable way than its natural ligand ( KD of 3 . 3 nM ) . In this sense , we must take into account that the N-terminal IgV-like domains of host CD48 and A43 are highly conserved ( 91% amino acid identity ) . All consensus residues of Ig superfamily members and those characteristic of SLAM family receptors are preserved in A43 . Moreover , only one of the 14 amino acids of CD48 involved in the CD48:2B4 interaction [31] has been substituted by a different residue in A43 . This is especially relevant since it has been estimated that the CD48 gene from which A43 derives was captured 19 million years ago [23] . This extraordinary degree of sequence preservation of the Ig domain , which contrasts with the substantial divergence of the rest of the A43 molecule , indicates a strong evolutionary pressure for mantaining its ligand binding specificity and affinity , in order to retain the properties most beneficial for viral survival . We show that both , recombinant A43-Fc as well as the viral protein shed to the medium of tranfected or infected cells , prevent host CD48:2B4 interactions . Moreover , we present evidences that purified A43 can substantially reduce the interaction of host 2B4-transfected cells and host B lymphocytes that endogenously express CD48 . Altogether , and taking into consideration the soluble nature of A43 , these findings indicate that the viral protein serves as a decoy molecule , establishing a stable , long-lasting complex with 2B4 that prevents cytotoxic cells from recognizing its counter receptor CD48 on the surface of infected cells . Mammalian decoy receptors are soluble versions of cell surface receptors structurally incapable of signaling . They are able to down-modulate key biological functions by binding with high affinity and specificity to their ligands [32] . Viruses , such as poxviruses and herpesviruses , encode in their genomes a number of secreted decoy receptors that target mainly cytokines and chemokines [33–35] . The function of A43 described here illustrates that secreted viral decoys can also target additional key molecules involved in the antiviral response . Importantly , we report that A43 is also capable of binding to human 2B4 . This is remarkable since the N-terminal Ig domains of human and host CD48 share only 63% identity . Indeed , we show that the purified A43-Fc fusion protein specifically binds to h2B4-transfected COS-7 cells , as well as to human NK and T cell lines that endogenously express h2B4 . In addition , SRP assays revealed that soluble A43 interacts with surface-immobilized h2B4 with high affinity ( KD 3 . 6 nM ) , in a similar way that it is observed for the interaction of the viral protein to host 2B4 . However , in this case , when compared with hCD48 , A43 was found to bind with a higher affinity ( around 6-fold ) to h2B4 . Moreover , of particular relevance is the difference in the dissociation rates between A43:h2B4 and hCD48:h2B4 interactions , with A43 dissociating from h2B4 much slower than hCD48 ( 55-times ) . It should be pointed out , that the affinities reported here for hCD48:h2B4 interactions diverge with preexisting affinity data , which were in the low μM range [4] . This discrepancy may be due to differences in the experimental conditions and/or in the biosensor platforms used . Due to the impossibility of obtaining NK cells from the OMCMV host , the owl monkey A trivirgatus , the finding that A43 is capable of binding to human NK cells allowed us to assess A43 at the functional level . NK cell cytolytic responses against virally infected cells are determined by the balance of inhibitory and activating signaling pathways , with the 2B4 receptor contributing to this process [36 , 37] . Virtually all cytolytic NK cell responses depend on several coordinated steps , requiring a very first stage of contact between the NK cell and the potential target cell , with the implication of integrins , cytoskeletal proteins as well as NK cell receptors , followed by the formation of the immune synapse , sustained signaling , and directed delivery of lytic granules onto the target cell [38] . Engagement of 2B4 by CD48 on the NK cell leads to the initial scanning of target cells and the synapse formation , inducing its recruitment and clustering into lipid rafts [26] . Subsequently , the phosphorylation of the ITSMs in the 2B4 cytoplasmic tail induces its association with the adaptor molecule SAP , which enhances NK cytotoxicity [9 , 39–41] . Here , we explored the function of A43 by employing a number of different assays . We provide evidence that NK cell-mediated cytotoxicity against hCD48 expressing target cells is pronouncedly impeded when NK-92 , YT cells or primary human NK cells are incubated with soluble purified A43 or supernatants from infected cells . In addition , we show that in the presence of A43 , NK cell production of IFN-γ is severely impaired . Our data also indicate that by masking h2B4 in NK cells , the viral protein disrupts the formation of conjugates and the establishment of the mature immunological synapse between effector and target cells . Lastly , we also present data that demonstrate that soluble A43 can interfere with human NK cell-mediated lysis of host B lymphocytes , which endogenously express CD48 , or host CD48-expressing HEK cells . Taking into account the IE kinetics of A43 , we propose that this viral protein may provide a mechanism for rapid counteraction of innate responses soon after infection , reducing host control of viral replication . The results obtained here further emphasize the important function of the CD48/2B4 axes in the antiviral response . We must consider that in addition to OMCMV , other CMVs and large DNA viruses encode vCD48s , and therefore have the potential to act as CD48 decoys or to block 2B4 responses in alternative ways . Interestingly , some E3 proteins of human adenoviruses , reported to share a very low homology with SLAM receptors , show a promiscuous binding to several cell-surface molecules , including 2B4 [22 , 42] . In addition , viral proteins can have the ability to interfere the CD48:2B4 axis by different mechanisms without interacting with 2B4; for example , by downregulating CD48 at the surface of the infected cell . Indeed , we have reported that MCMV , which encodes the early m154 protein responsible for driving proteolytic degradation of CD48 , can lead to defective NK cell responses during infection [21] . Consequently , an MCMV mutant lacking m154 results in an attenuated phenotype in vivo , which can be substantially restored after NK cell depletion in mice . Moreover , human CMV ( HCMV ) also uses in part this strategy , reducing CD48 surface levels on human macrophages , although in this case the viral protein causing these effects is not yet known [43] . However , it must be noted that HCMV does not completely eliminate CD48 from the surface of infected cells , and consistent with this , blockade of 2B4 with a mAb inhibits only partially the human primary NK cell degranulation triggered by the infected macrophages . While HCMV does not encode a recognizable CD48 homolog , we cannot discard that it expresses a soluble protein that , in a similar way to A43 , could contribute to block NK cell functions . This might not be surprising , taking into account that CMVs use diverse and redundant strategies to interfere with some of their immune targets . In this context , it seems that viruses have adopted different , and perhaps complimentary , tactics to modulate CD48:2B4 mediated effects . Given that 2B4 is expressed on other immune cytotoxic cells , such as CD8+ T cells , it is more than likely that A43 also has the capacity to interfere with their functions via a similar blocking mechanism to that reported here for NK cells . Thus , expression of A43 could be of particular importance for the viral evasion of both innate and adaptive immune surveillance . Interestingly , CD48 has also been identified as a soluble form in human serum , and shown to be at increased levels in patients with arthritis , mild asthma , and advanced lymphoid malignancies , potentially acting as an antagonist or decoy receptor capable of blocking CD48:2B4 binding [44–46] . Thus , considering the exceptional binding kinetic features of A43 compared to human CD48 , one key aspect derived from this work is the potential of using A43 to develop novel therapeutic tools to manipulate aberrant immune responses , such as autoimmune diseases . In summary , our work contributes a novel example of the potency of viral proteins to achieve immune modulation , in this case by counteracting the CD48/2B4 axis through a previously undescribed mechanism involving the direct blockade of 2B4 .
All procedures involving animals and their care were approved ( protocol number CEEA 308/12 ) by the Ethics Committee of the University of Barcelona ( Spain ) and were conducted in compliance with institutional guidelines as well as with national ( Generalitat de Catalunya decree 214/1997 , DOGC 2450 ) and international ( Guide for the Care and Use of Laboratory Animals , National Institutes of Health , 85–23 , 1985 ) laws and policies . Human blood was obtained from healthy volunteer donors through the Blood and Tissue Bank of the Catalan Department of Health ( Barcelona , Spain ) . Utilization of blood products for the experiments conducted was approved by the Ethics Committee of the Hospital Clinic of Barcelona ( Barcelona , Spain ) , and according to the principles of the Declaration of Helsinki . The cell lines COS-7 ( green monkey fibroblast ) , HEK-293T ( human embryonic kidney ) , NS-1 ( mouse myeloma ) , YT ( human NK ) , HSB2 ( T lymphocyte ) , and MOLT4 ( T lymphocyte ) were obtained from the American Type Culture Collection . The owl monkey kidney cell line OMK ( 637–69 ) was from Sigma-Aldrich , and the human NK-92 cell line was kindly provided by T . Bellón ( La Paz University Hospital Health Research Institute , Madrid , Spain ) . COS-7 and HEK-293T cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 2 mM glutamine , 1 mM sodium pyruvate , 50 U of penicillin per ml , 50 g of streptomycin per ml , and 10% fetal bovine serum . NS-1 , YT , HSB2 , and MOLT4 cells were cultured in RPMI-1640 medium , and OMK and NK-92 cells in Alpha Minimum Essential medium , supplemented as indicated above . In addition , 12 . 5% horse serum and 200 U/mL of rhIL-2 ( Immunotools ) were added to the medium of the NK-92 cells . B lymphocytes from A . trivirgatus owl monkeys were immortalized by infection with Epstein-Barr virus ( EBV; provided by Montse Plana [Institut d’Investigacions Biomèdiques August Pi i Sunyer , Barcelona , Spain ) ] ) , following standard protocols . Briefly , fresh blood samples from A . trivirgatus were obtained from Faunia ( Madrid , Spain ) . Erythrocytes were lysed by hypotonic buffer for 15 min , and 1x106 cells were resuspended in 500 μL of RPMI-1640 medium supplemented as indicated above , with the addition of 20% of fetal bovine serum and 1μg/mL cyclosporin A ( Sigma-Aldrich ) . Then , cells were mixed with 500 μL of supernatant from the marmoset B cell line ( B95-8 ) that produces EBV , incubated overnight at 37 °C , washed , and grown during 4 weeks to obtain the A . trivirgatus polyclonal B lymphocyte cell line . Human PBMCs were isolated by Ficoll density-gradient centrifugation from fresh blood samples obtained from healthy human donors , as described previously [47] . Human NK cells were isolated from PBMCs by negative selection using the human NK Cell Isolation kit ( Miltenyi Biotec ) according to the manufacturer’s instructions by employing an autoMACS column on the autoMACS Pro separator ( both Miltenyi Biotec ) . The efficiency of the process was confirmed by flow cytometry with anti-human CD56 , CD16 , and CD3 antibodies . The owl monkey cytomegalovirus ( OMCMV ) was provided by A . Davison ( Medical Research Council , University of Glasgow Center Virus Research , Glasgow , United Kingdom; [48] ) . Viral stocks were prepared by infecting at a low moi OMK cells with OMCMV . Cell supernatants were recovered when maximum cytopathic effect was reached , and then cleared of cellular debris by centrifugation at 1700 g for 10 min . Viral titers were determined by standard plaque assays on OMK cells . Except for viral stock preparations , infections included a centrifugal-enhancement-of-infectivity step [49] . The supernatants from mock-infected or infected cells used in blocking and cytotoxicity assays were further clarified by ultracentrifugation at 34500 g for 90 min to remove viral particles and concentrated 20-fold using the Amicon Ultra-15 Centrifugal Filter Unit with an Ultracel-30 membrane ( Millipore ) . The anti-human 2B4 ( clone 2B4 . 69 ) , anti-HA ( employed for flow cytometry ) , and anti-human IgG ( clone 29 . 5; Fc specific ) mAbs were generated in our lab and have been previously described [23 , 50] . The anti-human CD48 ( 99A ) was provided by R . Vilella ( Hospital Clinic , Barcelona , Spain ) . The rabbit anti-HA ( clone c2974 ) used for western ELISA and the anti-perforin mAb employed for immunofluorescence were purchased from Cell Signaling MP , Biomedicals , and Immunotools , respectively . The anti-human CD3-FITC ( clone 17A2 ) , CD16-Brilliant Violet 421 ( clone 3G8 ) , and CD56-PerCP-Cy5 . 5 ( clone B159 ) used to phenotype primary NK cells were obtained from BioLegend and BD Biosciences . The anti-human IgG-POD ( Fc specific ) and the streptavidin-POD conjugate were from Roche . The anti-mouse IgG-PE was purchased from Jackson ImmunoResearch , the goat anti-mouse IgG ( H+L ) -Alexa Fluor 555 was from Life Technologies , and the streptavidin-PE conjugate from BD Biosciences . The polyclonal Ab against A43 was obtained from two BALB/c mouse immunized four times with the A43-Fc fusion protein . Mice were bled , serum collected , and the mice IgGs purified using an Affi-Gel Protein-A MAPS II kit ( Bio-Rad ) . The anti-A43 antibody coupled to protein G Sepharose resin ( GE Healthcare ) was used to pre-absorb ( for 1h at room temperature in a rotating shake ) the A43 protein present in the OMCMV infected supernatants . When required , purified antibiodies were biotinylated using biotinamidocaproate N-hydroxysuccinimide ester ( Sigma-Aldrich ) . HA-CD48-Tm , containing the N-terminal HA-tagged ectodomain of A . trivirgatus CD48 without its signal peptide fused to the platelet-derived growth factor receptor ( PDGFR ) transmembrane domain ( Tm ) , was constructed as follows: first , PCR products were generated using as template DNA extracted from OMK cells and primer sets based on regions flanking the second and third exons of Callihtrix jacchus and Saimiri boliviensis CD48s , corresponding to the first and second Ig domains , respectively . The transcript annotations of CD48 considered were those of GenBank XP_008982993 . 1 of C . jacchus and GenBank XP_003938006 . 1 of S . boliviensis . The resulting PCR products were inserted into the pGEM-T vector ( Promega ) and sequenced . The newly identified nucleotide sequence was deposited in GenBank under the following accession number: MH663530 ( exon2 and exon3 of A . trivirgatus CD48 ) . Splicing by overlap extension ( SOE ) -PCR was then performed to join sequences coding for the first and second Ig domains of this molecule . The generated plasmids were used as templates and , based on their sequencing , two internal sequence-complementary primers , annealing within the Ig domains , and two external primers were employed for PCR amplification . These last two primers contained restriction sites at 5’ and 3’ ends , which were subsequently used to clone the SOE-PCR amplified products in the pDisplay vector ( Invitrogen ) . HA-A43 , HA-A43-Tm , A . trivirgatus 2B4-Fc ( 2B4-Fc , named aoCD244-Fc before ) , full-length human CD48 ( hCD48 ) , and human 2B4 ( h2B4 ) were previously described [23 , 50] . HA-2B4 , expressing the N-terminal HA-tagged ectodomain of A . trivirgatus 2B4 fused to a region of human 2B4 ( from residue 219 in the stalk segment to stop codon ) , was constructed by generating two independent PCR products that were subsequently linked making use of a natural SalI restriction site present in the stalk region of both 2B4s . The first amplified fragment , containing the A . Trivirgatus 2B4 ectodomain was obtained using 2B4-Fc as template and a specific primer set , with the reverse primer containing the SalI restriction site at the 3´end . The second PCR product , encompassing the stalk , Tm and cytoplasmic regions of human 2B4 was generated using the h2B4 construct as template and a specific primer set , with the forward primer containing the SalI restriction site at the 5´end . These PCR products were independently cloned into pGEM-T , and then inserted consecutively into the pDisplay vector , using the shared SalI restriction site to join both fragments . HA-A43 ΔG-R and HA-A43 ΔT-R expressing full length HA-A43 with a deletion in residues 214–219 ( HA-A43 ΔG-R ) or 220–225 ( HA-A43 ΔT-R ) , and HA-A43 K216P and HA-A43 K216A with the lysine in position 216 exchanged by proline ( K216P ) or alanine ( K216A ) , were constructed by SOE-PCR as described above , using HA-A43 as a template and internal primers bearing the corresponding deletions or mutations . PCR-amplified products were inserted into the pGEM-T vector and transferred to pDisplay . A43-Fc , A . trivirgatus CD48-Fc ( CD48-Fc ) , hCD48-Fc and h2B4-Fc fusion proteins , expressing the two Ig domains of these molecules ( with the CD33 leader peptide replacing their own signal peptide ) fused to the Fc region of human IgG1 , were obtained by PCR using as templates HA-A43 or HA-CD48-Tm and hCD48 or h2B4 constructs , respectively , and specific set of primers with restriction sites . The PCR-amplified products were inserted into pGEM-T and finally cloned into pCI-neo Fc vector , as described before [23] . A mutant version of the Fc region of human IgG1 , named here Fc* , containing mutations L234F , S235Q , K322Q , M252Y , S254T , and T256E , was obtained by chemical synthesis ( Genscript ) . The mutations introduced abrogate the binding of the Fc region to Fc receptors [27] . pCI-neo A43-Fc* , expressing the two Ig domains of A43 fused to the Fc* , was obtained by replacing the Fc region present in plasmid A43-Fc by the Fc* region . All PCR reactions were performed under the following conditions: 1 cycle at 94°C for 5 min; 30 cycles of 1 min at 94°C , 1 min at 51°C , and 1 min at 72°C; and 1 cycle at 72°C for 10 min . For the annealing reactions , the conditions were: 6 cycles of 5 min at 94 °C; 1 cycle at 51 °C for 1 min; 1 cycle at 72 °C for 1 min; and 1 cycle of 10 min at 72 °C . The identification of all recombinant plasmids was confirmed by DNA sequencing . OMK cells were mock infected or infected with OMCMV at an moi of 1 . The inhibitory chemicals CHX ( 100 μg/ml; Sigma-Aldrich ) or PAA ( 250 μg/ml; Sigma-Aldrich ) were added to some cultures to assess selective expression of viral immediate-early genes or early genes , respectively . The cultures were treated with CHX 30 min prior to infection , whereas phosphonoacetic acid was added at the time of infection , and both inhibitors were maintained until RNA was harvested . Total RNA was isolated at different time points after infection ( 13 hours post infection for CHX samples and 72 hours post infection for the rest ) by the TRIzol method ( Invitrogen ) . RT-PCR was then carried out using the SuperScript III First-strand Synthesis System for RT-PCR ( Invitrogen ) according to the manufacturer’s protocol . Briefly , RNA samples were treated with RNase-free DNase I ( Promega ) for 30 min at 37°C , and the DNase was inactivated at 65°C for 10 min . The RNA was reverse transcribed using oligo ( dT ) primers at 50°C for 50 min , and reactions were terminated by heating at 85°C for 5 min . The reverse-transcribed products were treated with RNase H for 20 min at 37°C and amplified by PCR as indicated above using specific primer sets . Amplified products ( a 599-bp fragment for A43; a 590-bp fragment for OMCMV IE1; a 570-bp fragment for OMCMV UL54; a 166-bp fragment for OMCMV UL73; and a 101-bp fragment for GAPDH ) were separated on a 1% agarose gel and visualized by RedSafe nucleic acid staining solution ( iNtRON Biotechnology Inc . ) . COS-7 cells were transiently transfected with 5 μg of the indicated plasmid using the Amaxa Cell Line Nucleofector Kit R according to the manufacturer’s protocol . HEK cells were transiently transfected with 0 . 2 μg/cm2 of the indicated plasmid mixed with 6 μL/μg DNA of polyethylenimine ( 1mg/mL , Sigma-Aldrich ) in 0 . 1 mL/cm2 of OPTIMEM medium ( Gibco ) for four hours . Then cultures were washed and used either one day later when performing functional assays or 6 days later to collect A43-Fc , hCD48-Fc , CD48-Fc or host 2B4-Fc fusion proteins from supernatants . The NS-1 stable transfectant secreting m2B4-Fc fusion protein has been previously described [21] . The NS-1 h2B4-Fc stable cell line was obtained by the same procedures . Briefly , eight million of NS-1 cells were electroporated ( 280V , 950 mF ) with 8 μg of linearized DNA using the Gene Pulser II Apparatus ( Bio-Rad ) , selected with 1 . 2 mg/mL of geneticin ( G418 , Invitrogen ) , and further subcloned . The clone producing the highest amounts of fusion protein was cultured in INTEGRA CL 1000 flasks ( Integra Biosciences AG ) . The supernatants containing the fusion proteins were purified using the Affi-Gel Protein-A MAPS II kit ( Bio-Rad ) . To analyse A43 shedding , supernatants of COS-7 cells transfected with HA-A43 or HA-A43 K216P were used after being cleared to remove cellular debris . HA-A43 supernatants employed in blocking experiments were further concentrated 20-fold as indicated above for the supernatants of infected cells . Detection of the HA-A43 protein in cell culture supernatants was performed by sandwich ELISA using the rabbit anti-HA mAb to coat the ELISA plates and the biotinylated anti-A43 polyclonal antibody , followed by incubation with streptavidin-POD conjugate . A similar sandwich ELISA was performed to detect A43 in the supernatants from infected cells , except that in this case the anti-A43 polyclonal antibody was also used as the capture antibody . The capacity of the A43 protein to block host 2B4:CD48 or h2B4:hCD48 interactions was assessed by sandwich ELISA using host 2B4-Fc or h2B4-Fc coated ELISA plates and biotinylated host CD48-Fc or hCD48-Fc for detection , respectively , followed by streptavidin-POD . The concentration of IFN-γ in the supernatants of the co-cultures of NK cells and HEK cells non-transfected or transfected with hCD48 , was measured using the high-sensitivity human ELISA set ( ImmunoTools ) . ELISA measurements of Optical densities ( ODs ) were done at 450 and 570 nm ( Thermo Scientific Multiskan FC ) . Flow cytometry was performed using standard procedures [51] . To minimize non-specific staining , all incubations were carried out in the presence of 20% rabbit serum ( Linus ) and 1% of fetal bovine serum in PBS . Cells were stained with the corresponding antibodies , followed by anti-mouse IgG-PE . Fc fusion protein stainings were performed using the indicated amount of the Fc fusion protein followed by incubation with the anti-human IgG ( Fc specific ) and by anti-mouse IgG-PE . An irrelevant Fc fusion protein ( CTL-Fc ) was always used as a negative control . When performing assays to block the interaction of hCD48 and h2B4 , cells were incubated with biotinylated hCD48-Fc fusion protein , followed by streptavidin-PE . SPR was performed on a Biacore X100 instrument ( GE Healthcare ) as described [52] . Recombinant host 2B4-Fc or h2B4-Fc proteins were immobilized at low density on CM4 chips by amine coupling . Increasing concentrations of host CD48 , hCD48 or A43 recombinant proteins were injected in HBS-EP buffer ( 10 mM HEPES , 150 mM NaCl , 3 mM EDTA , 0 . 005% [vol/vol] surfactant P20 [pH 7 . 4] ) at 30 μl/min during 2 min and a 5-min dissociation was recorded . A 10–1500 nM concentration range of analyte was used . Between analyte injections , surface was regenerated with 10 mM glycine-HCl pH 2 . 0 . Kinetic data were globally fitted to a 1:1 Langmuir model using the Biacore X100 Evaluation software version 2 . 0 . 1 ( GE Healthcare ) . Bulk refractive index changes were removed by subtracting the responses recorded in the reference flow cell , and the response of a buffer injection was subtracted from all sensorgrams to remove systematic artifacts . COS-7 cells , transiently transfected with host HA-2B4 or hCD48 , or the corresponding empty plasmids , were cultured on glass coverslips in 24-well tissue culture plates . After 24 hours , cells were washed with PBS and maintained at 4°C for 20 min . A . trivirgatus B lymphocytes or NK-92 cells were added to wells for 1 hour at 4°C at a concentration of 1x106 or 5x105 cells/well , respectively . Subsequently , coverslips were washed by repeated immersion ( 20X ) in culture medium , and the adhesion of A . trivirgatus B lymphocytes or NK-92 cells to the COS-7 transfected cells was examined using an inverted Leica DMI600 B microscope ( Leica microsystems ) , where contrast images were obtained . When indicated , adhesion assays were performed with A . trivirgatus B lymphocytes or NK-92 cells previously stained with calcein-AM ( 4 μM , Life Technologies ) for 30 min , followed by two washing steps , as previously described [53] . Cells were trypsinized and analyzed by flow cytometry using a FACSCalibur flow cytometer ( BD Biosciences ) and FlowJo software ( TreeStar Inc ) . Cells were gated based on forward and side scatter , examined for calcein labelling , and the percentages of calcein stained A . trivirgatus B lymphocytes or NK-92 cells versus unlabelled COS-7 cells were calculated . Cellular adhesion-blocking experiments were performed by pre-incubation of COS-7 HA-2B4 or NK-92 cells with 10 μg/mL of A43-Fc , an unrelated CTL-Fc protein , and in the case of NK-92 cells with the anti-human 2B4 mAb or an isotype control antibody , for 30 min at 37°C . NK-92 and YT cells were labelled with CellTracker Blue CMAC ( 10 μM , Invitrogen ) for 30 min at 37°C . HEK cells , transfected with hCD48 or the corresponding empty vector , were stained with CellTrace CFSE ( 0 . 5 μM , Invitrogen ) for 15 min at 37°C according to the manufacturer´s instructions . After labelling , cells were washed extensively . Then , 1x105 NK-92 or YT cells were co-cultured at 37°C for 0 or 10 min with HEK cells at a ratio of 1/1 , in a final volume of 150 μL . Reactions were stopped by adding 50 μL of ice-cold PBS . Conjugates were detected by flow cytometry as double positive events . When indicated , labelled NK cells were pre-incubated with 10 μg/mL of A43-Fc , an unrelated CTL-Fc protein or anti-2B4 mAb , at 37°C for 30 min . HEK cells transfected with hCD48 or the corresponding empty vector were first labelled with CellTracker Blue CMAC . After washing , labelled HEK cells were mixed for 10 min at 37°C and a final volume of 500 μL with 1x105 YT cells at a ratio of 1:1 to allow conjugate formation . The suspension was placed on coverslips coated with poly-L-lysine ( Sigma-Aldrich ) for 10 min at 37°C , washed with PBS , fixed with 4% formaldehyde for 10 min , and permeabilized with 0 . 05% Triton X-100 for another 10 min . Subsequently , samples were blocked with 20% rabbit serum and 6% fetal bovine serum in PBS and incubated with Alexa Fluor 488 Phalloidin ( Invitrogen ) and the primary anti-perforin mAb , followed by a secondary antibody goat anti-mouse IgG ( H+L ) -Alexa Fluor 555 . Coverslips were mounted in ProLong Gold antifade reagent ( Invitrogen ) . Fluorescence images were acquired using a Nikon Optiphot-2 microscope ( Nikon Corp . ) . Synapse stages were defined as: 0 , conjugates lacking actin polymerization and perforin polarization; 1 , conjugates with polymerized actin , but missing perforin polarization; 2 , conjugates with actin polymerization and perforin clustered at the immune synapse . When indicated , NK cells were pre-incubated with 10 μg/mL of A43-Fc or CTL-Fc fusion proteins , at 37°C during 30 min prior to being mixed with the target cells . The cytotoxic activity of NK-92 or YT effector cells was determined in a calcein-AM release assay [54] . The assay was performed in V bottom 96-well microtiter plates . Briefly , transfected target HEK cells or A . trivirgatus B lymphocytes were labelled with calcein ( 4 μM ) for 30 min at 37°C . After two washes , 2x104 target labelled cells per well were mixed at 37°C with effector cells at E/T ratios ranging from 20/1 to 0 . 5/1 , in triplicate . The assay also included six replicate wells with only target cells ( spontaneous release ) , only target cells lysed with the addition of 1% Triton X-100 ( maximum release ) , or medium alone ( background ) . Four hours later , 100 uL of supernatant of each culture was collected and transferred into new plates . When indicated , effector cells were pre-incubated with 10 μg/mL of A43-Fc or A43-Fc* , unrelated CTL-Fc or CTL-Fc* proteins , anti-2B4 mAb or an isotype control antibody , or supernatants from mock-infected or OMCMV infected cells , for 30 min at 37°C . Fluorescence was measured using Spark multimode microplate reader ( Tecan ) . The background fluorescence was substracted from all samples . The percentage of specific lysis was calculated according to the formula: ( experimental release—spontaneous release ) / ( maximum release—spontaneous release ) x 100 . Signal peptides and transmembrane regions were predicted by using SignalP 4 . 1 and TMHMM 2 . 0 , respectively . These bioinformatics prediction tools are available at http://www . cbs . dtu . dk/services/SignalP/ ( SignalP 4 . 1 ) , and http://www . cbs . dtu . dk/services/TMHMM-2 . 0/ ( TMHMM 2 . 0 ) . Ig domains were determined from annotations in CDD [55] . To calculate the percentage of amino acid identity of the N-terminal Ig domains of A43 and A . trivirgatus CD48 , protein sequences were paired aligned ( MAFFT version 7 . 402; [56] ) and positions containing gaps were discarded . Analyses were performed with GraphPad Prism software ( v . 7 . 03 ) and Microsoft Excel ( 2010 ) . Results are given as means ± standard deviations ( SD ) or as means ± standard errors of the means ( SEM ) , and statistical significances were determined with the Student’s t-test ( two-tailed ) . P-values less than or equal to 0 . 05 ( * ) , 0 . 01 ( ** ) and 0 . 001 ( *** ) were considered statistically significant . | In order to evade detection and destruction by cytotoxic lymphocytes and successfully persist within their hosts , cytomegalovirus ( CMVs ) have evolved a number of genes dedicated to block immune recognition . Certain CMVs and other large DNA viruses encode homologs of the cell-surface molecule CD48 , a ligand of the 2B4 receptor involved in regulating the function of cytotoxic lymphocytes . Here , we have investigated for the first time the immunomodulatory potential of one of these viral molecules . We show that A43 , a CD48 homolog encoded by owl monkey CMV , is a soluble molecule that exhibits exceptional binding kinetics for 2B4 , and is furthermore capable of blocking the interaction with its counter-receptor CD48 . Moreover , we reveal how this viral protein interferes with human NK cell-mediated cytotoxicity by inhibiting the immune synapse between human NK cells and target cells . Thus , these findings not only underscore the importance of 2B4-mediated immune responses in controlling CMV infections , but also unveil the shedding of a virally-encoded soluble variant of CD48 as a new viral counteract mechanism for subverting immune surveillance . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"methods"
] | [
"blood",
"cells",
"flow",
"cytometry",
"cell",
"physiology",
"cell",
"binding",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"hek",
"293",
"cells",
"nervous",
"system",
"immunology",
"biological",
"culture... | 2019 | Subversion of natural killer cell responses by a cytomegalovirus-encoded soluble CD48 decoy receptor |
Our understanding of persistence and plasticity of IL-17A+ memory T cells is clouded by conflicting results in models analyzing T helper 17 cells . We studied memory IL-17A+ CD8+ T-cell ( Tc17 ) homeostasis , persistence and plasticity during fungal vaccine immunity . We report that vaccine-induced memory Tc17 cells persist with high fidelity to the type 17 phenotype . Tc17 cells persisted durably for a year as functional IL-17A+ memory cells without converting to IFNγ+ ( Tc1 ) cells , although they produced multiple type I cytokines in the absence of residual vaccine antigen . Memory Tc17 cells were canonical CD8+ T cells with phenotypic features distinct from Tc1 cells , and were Ror ( γ ) thi , TCF-1hi , T-betlo and EOMESlo . In investigating the bases of Tc17 persistence , we observed that memory Tc17 cells had much higher levels of basal homeostatic proliferation than did Tc1 cells . Conversely , memory Tc17 cells displayed lower levels of anti-apoptotic molecules Bcl-2 and Bcl-xL than Tc1 cells , yet were resistant to apoptosis . Tc1 cells required Bcl-2 for their survival , but Bcl-2 was dispensable for the maintenance of Tc17 cells . Tc17 and Tc1 cells displayed different requirements for HIF-1α during effector differentiation and sustenance and memory persistence . Thus , antifungal vaccination induces durable and stable memory Tc17 cells with distinct requirements for long-term persistence that distinguish them from memory Tc1 cells .
Memory Th1 and Tc1 ( IFNγ+ CD8+ T cells ) cells have been well characterized . Following the expansion phase of an immune response , a pool of memory precursor cells ( ~10% ) survives the contraction phase , and slowly differentiates into long-lasting memory cells[1] . Memory Th1 and Tc1 cells are maintained stably for years , often lifelong , even in the absence of cognate antigen [2–4] . Maintenance of memory Th1 and Tc1 is chiefly regulated by IL-7 for survival and IL-15 for intermittent homeostatic turnover[5–7] . Type 1 memory precursors are CD127hi , KLRG-1lo , CD27hi , CD122hi , Bcl-2hi , T-betlo , Eomeshi and TCF-1hi[8 , 9] , and the expression of CD62L and CCR7 distinguish “central memory” vs . “effector memory” T-cell subsets[10] . In contrast to Th1 and Tc1 cells , attributes of Th17 cells are less well understood and the persistence of memory Th17 cells is debated . In vitro polarized Th17 cells that are adoptively transferred persist and portray stem-cell like features[11–14] . Conversely , in vivo induction and persistence of Th17 cells depends on route and type of infection . Th17 cells induced upon mucosal Listeria monocytogenes infection are short-lived due to their CD27lo phenotype[15 , 16] , whereas subcutaneous injection of Mycobacterium tuberculosis antigens induces long-lasting Th1 and Th17 memory cells[16] . Acute cutaneous infection with Candida albicans induces Th17 cells that quickly “switch off” IL-17 production[17] , while oropharyngeal infection leads to induction of Th17 cells that persist for at least weeks[18] . Th17 cells are known to persist for a longer time under chronic inflammatory conditions , implying a role for lingering antigen and inflammation[19] . Unlike Th1 and Th2 cells , Th17 cells are poised to be highly plastic , presumably due to their epigenetic instability[20–22] . In vitro polarized Th17 cells either convert to Th1 cells or express both Th1 and Th17 cytokines after adoptive transfer into mice , possibly due to deprivation of signals from polarizing cytokines[23] . Host immune status also may impact Th17 plasticity since lymphopenia promotes expansion of Th1 cells[24] . In some models , immunization of mice with adjuvanted antigen drives Th17 responses , but the cells fail to maintain their phenotype and convert into Th1 cells . For example , in EAE models , Th17 cells show plasticity towards IFNγ and GM-CSF production and augment severity of the disease[25 , 26] . Similarly , Candida-specific human memory Th17 cells co-express IFNγ , and staphylococcus-induced Th17 cells express IL-10 upon re-stimulation[27] . Similar to Th17 cells , Tc17 cells form a distinct subset with widely ranging responses implicated in immunity , immunopathology and systemic autoimmunity [28] . Effector Tc17 cells mediate potent anti-viral , anti-fungal , and anti-tumor activity[29–31] , 33 . Tc17 effectors also potentiate autoimmune actions of Th17 cells during EAE[32] and exacerbate other autoimmune disorders due to their ability to express multiple cytokines e . g . IL-22 , GM-CSF , M-CSF , IFNγ and IL-3[28] . Like Th17 cells , in vitro polarized Tc17 show plasticity towards IFNγ production[33] , and donor Tc17 cells become plastic “inflammatory” iTc17 in recipients with GVHD[34] . Vaccination in the absence of CD4+ T cells induces Tc17 cells that confer resistance against lethal experimental fungal pneumonia[30] , a feature that can be translated to immune-compromised individuals with CD4+ T cell lymphopenia that are susceptible to opportunistic fungal infection[35] . While memory Tc17 cells are required in this type of immunity , the persistence and plasticity of these cells have not been investigated . In this study , we investigated the persistence , fidelity , plasticity and functionality of memory Tc17 cells by using an experimental fungal vaccine that is effective against lethal pneumonia in CD4+ T cell deficient hosts . We found that antifungal memory Tc17 cells are durable and persist long term . Memory Tc17 cells also showed little conversion into IFNγ producing cells , although the cells expressed multiple cytokines . We also found that memory Tc17 cells retain expression of RORγt and portray a phenotypic profile distinct from memory Tc1 cells . Memory Tc17 cells displayed higher proliferative renewal than did Tc1 cells , and lower levels of the anti-apoptotic molecule Bcl-2 , which was not required for maintenance of these Tc17 cells . Memory Tc17 cells required HIF-1α for their homeostasis , whereas memory Tc1 cells did not require this factor . Our work provides new insight into distinct features that characterize and promote persistent , stable anti-fungal Tc17 cells induced upon vaccination .
We first investigated the fate and longevity of memory Tc17 cells after vaccination . We used “fate-mapping” mice ( IL-17AcreRosa26eYFP ) [17] where IL-17A-induced cells become indelibly marked with eYFP irrespective of the fate of IL-17A expression . Similar to a published report[17] , ~30–50% of the eYFP+ cells expressed IL-17A ( S1A Fig ) . Likewise , ~30–40% of total IL-17A+ cells also expressed eYFP[17] . To assess the persistence of memory Tc17 cells , we vaccinated IL-17A fate-mapping mice and enumerated the percent and absolute numbers of CD8+ eYFP+ cells in tissues ( Fig 1A & 1E; S1B Fig ) . Effector CD8+ eYFP+ cells persisted as memory cells maintaining their phenotype up to 11 months post vaccination ( Fig 1A ) . We previously noted that live vaccine yeast persist for ~7 weeks[36] . To exclude that persistent vaccine antigen accounts for the longevity of memory Tc17 cells , we purified effector CD8+ T cells and adoptively transferred them into naïve wild-type ( WT ) and TCRα-/- recipient mice . At serial time points , we enumerated CD8+ eYFP+ T cells in draining lymph nodes ( dLNs ) and spleens ( Fig 1C & 1F; S1C Fig ) . The intake of CD8+ eYFP+ cells was ~10% as expected ( at day 1 post-transfer ) and the remaining cells persisted stably even after 3 months rest in both immune-sufficient ( WT ) and -deficient ( TCRα-/- ) recipients ( Fig 1F ) . Thus , antifungal memory Tc17 cells induced by vaccination persist stably even in the absence of vaccine antigen . One of the cardinal features of memory T-cell subsets is their ability to retain expression of signature cytokines . Effector Th17 cells are known to either cease IL-17A expression or convert into an IFNγ+ subset[17 , 23] . However , Th17/Tc17 cells are essential for immunity against many fungal and extracellular bacterial infections[37] , and we have shown that loss of IL-17A signaling ablates vaccine-immunity[30 , 38] . Here , we investigated whether persistent , memory CD8+ eYFP+ T cells retain the ability to produce IL-17A . We found that ~30% ( dLNs ) to ~45% ( spleen ) of CD8+ eYFP+ cells expressed IL-17A ( Fig 1B ) following vaccination and , with a long period of rest after vaccination , the IL-17A expressing eYFP+ cells increased to 50%-60% , suggesting continuous maturation of memory Tc17 cells with loss of non-producers . We also observed similar frequencies of IL-17A+ cells among adoptively transferred CD8+ eYFP+ cells in WT and TCRα-/- recipient mice ( Fig 1D ) , suggesting that persistent vaccine antigen is dispensable for maintenance and fidelity under lymphopenic conditions . Thus , IL-17A expression in vaccine-induced memory Tc17 cells not only endured , but also increased during the memory phase . In many studies involving infection , tumor , and autoimmunity , Th17 cells are either short-lived or convert into IFNγ producing cells[23] . Given the requirement that we previously observed for effector Tc17 cells following vaccination[30] , we investigated whether antifungal memory Tc17 cells portray such plasticity . We detected significant numbers of eYFP+ IFNγ+ cells immediately after vaccination , especially in the spleen ( Fig 1B ) , most of which did not produce IL-17A . However , the percentage of eYFP+ cells expressing IFNγ actually diminished over time . By day 346 post-vaccination <1% of eYFP+ cells in the dLN and ~3% of eYFP+ cells in the spleen were IFNγ+ ( Fig 1B ) . Similar results were found in adoptively transferred CD8+ eYFP+ T cells , with <10% of them producing IFNγ in both WT and TCRα-/- mice ( Fig 1D ) . Next , we evaluated the effect of IL-12 and/or IFNγ on plasticity of memory Tc17 cells in an in vitro experiment . As expected , IL-12 stimulation significantly enhanced the frequency of IFNγ+ cells among eYFP- CD44hi ( Tc1 ) cells ( S2A Fig ) . However , memory Tc17 cells were resistant to IL-12- or IFNγ-induced signals for plasticity towards IFNγ production ( S2B Fig ) Thus , our data suggest that vaccine-induced anti-fungal memory Tc17 cells do not display plasticity towards IFNγ , even in the absence of persistent vaccine antigen or in a lymphopenic environment . We and others have shown that effector Tc17 cells contribute to fungal resistance in the absence of CD4+ T cells[18 , 30 , 39] . Previously , we showed that effector Tc17 cells are obligatory for anti-fungal immunity , whereas type I cytokines ( IFNγ , TNFα and GM-CSF ) can be compensated[30 , 40] . Here , we investigated whether vaccine-induced memory Tc17 cells retain their recall responses and function in antifungal resistance over an extended period after vaccination . To test this issue , mice were rested for ~5 months after vaccination before we challenged them with a lethal strain of yeast . Upon pulmonary challenge , memory Tc17 cells efficiently recalled into the lung and expressed IL-17A ( Fig 2A ) . However , memory eYFP+ Tc17 cells did not express IFNγ , suggesting a stable Tc17 lineage commitment without plasticity . To analyze the role of IL-17A in resistance , we neutralized soluble IL-17A with anti-IL-17A mAb throughout the infection after challenge . Unvaccinated mice had a higher fungal burden than vaccinated controls . Among vaccinated mice , the IL-17A neutralized group had a ~100-fold higher fungal burden than the mice treated with control antibody ( Fig 2; P≤0 . 001 ) , suggesting that memory Tc17 cells contributed significantly to vaccine-immunity in the absence of CD4+ T cells . Thus , vaccine-induced antifungal memory Tc17 cells persist and maintain the capacity to mediate protective immunity . We next asked whether CD8+ eYFP+ cells can produce cytokines other than IFNγ . Previously , we showed that type I cytokines such as GM-CSF and TNFα augment antifungal vaccine immunity[40] . Here , using cells portrayed in Fig 1 , we assessed multi-cytokine production by memory CD8+ eYFP+ cells by analyzing the percentage of mutually exclusive single , double , triple and quadruple cytokine producing CD8+ eYFP+ cells ( e . g . IL-17A , IFNγ , GM-CSF , and TNFα ) ( Fig 3 ) . Effector eYFP+ cells were largely single IL-17A producers , while most memory IL-17A+ eYFP+ cells expressed more than one cytokine with the largest pool in the spleen being triple producers by 137 days post vaccination ( Fig 3A ) . Notably , single or multicytokine IFNγ producing eYFP+ cells were lost during transition to the memory phase . Adoptively transferred effector eYFP+ cells were more often double producers than triple producers ( Fig 3B ) , suggesting that prolonged antigen exposure may bias towards multi-cytokine producing memory Tc17 cells ( Fig 3A ) . About 30% of eYFP+ cells produced none of these cytokines . We asked whether these cells had higher PD-1 expression , an indicator of a dysfunctional or exhausted phenotype . Only 12% of non-producers were PD-1+ ( S3A Fig ) similar to the proportion of producers expressing PD-1 . We also looked at other T cell phenotypes: ≤5% of CD8+eYFP+ cells produced IL-22 , a cytokine often associated with Th17 responses , and ≤1% expressed FoxP3+ , which is associated with regulatory T cells ( S3B Fig ) . We also evaluated whether memory IL-17 producing CD8+ T cells , similar to Th17 cells , co-express IL-21[41] along with IL-1R , IL-23R and Stat3 . We found that Tc17 cells did not express IL-21 , but they did express higher levels of IL-1R , IL-23R and Stat3 compared to Tc1 cells ( S3C Fig ) suggesting a distinct differentiation program in contrast to Th17 cells . Thus , vaccine induced memory Tc17 cells have the ability to produce multiple type I cytokines that confer fungal immunity , yet without undergoing conversion towards IFNγ production . We evaluated the lineage specificity of CD8+ eYFP+ cells by looking at expression of CD8β , TCRβ and CD3ε on CD8α+ eYFP+ cells . CD8α+ eYFP+ cells equivalently co-expressed a CD8β chain ( Fig 4A ) , a cardinal feature of conventional CD8+ T cells . Similarly , CD8α+ eYFP+ cells were TCRβ+ and CD3ε+ , suggesting that memory Tc17 cells are indeed classical CD8+ T cells . We characterized memory Tc17 cells for their phenotypic attributes and contrasted them with Tc1 cells ( CD8+eYFP- ) ( Fig 4B & S4 Fig ) . Tc17 ( eYFP+ ) cells expressed high levels of CD127 ( IL-7Rα , essential for memory homeostasis ) ; however , they were CD62Llo and Ly6Clo , suggesting they are “effector” memory not “central” memory cells[10 , 42] . As expected for effector memory cells , none of the antifungal memory cells expressed the terminal differentiation marker KLRG-1[36 , 43] . Further , memory Tc17 cells retained high expression levels of chemokine receptor CCR6 and costimulatory molecule , CD43 , similar to effector Tc17 cells[30] . In concordance with other studies[13 , 44] , many of the Tc17 cells were CD27lo [13 , 15 , 30] . Interestingly , most of the activated ( CD44hi ) T cells ( both eYFP+ and eYFP- ) expressed integrin CD103 ( αE ) , the alpha chain of integrin αE β7 that selectively marks tissue “resident” memory ( TRM ) cells[45] . We also compared the transcription factor profile of memory Tc17 cells with Tc1 cells ( Fig 4C ) . Memory Tc17 cells retained high expression of the prototypical transcription factor Ror ( γ ) t , and expressed low levels of transcription factors T-bet and Eomes typically associated with Tc1 cells . Notably , the expression levels of TCF-1 , a LEF/TCF family member associated with memory and stem-cell like activity[13 , 46] , was similar or higher in memory Tc17 cells vs . memory Tc1 ( CD44hieYFP- ) cells . Likewise , the frequencies of many phenotypic attributes of Tc17 cells differed significantly from eYFP-CD44hi cells ( Tc1 ) and naïve CD8+ T cells ( S4 Fig ) . Collectively , our data suggest that memory Tc17 cells are canonical CD8+ T cells that display distinct phenotypic attributes and constitute an “effector memory” population . Th17 cells generally display more proliferative renewal than Th1 cells , but homeostatic renewal of Tc17 cells is less understood[47] . We have shown that antifungal effector Tc17 cells undergo significantly higher proliferation than effector Tc1 cells during the expansion phase after vaccination[39] . Here , we assessed the proliferative ability of early and late memory Tc17 and Tc1 cells ( Fig 5A & S5 Fig ) . Consistent with other studies[48] , memory Tc1 cells became quiescent , and the proliferation of memory Tc1 cells ( ≈20% ) was ~3 . 7 times lower than effector Tc1 cells ( ≈75% ) as measured by BrdU uptake over a 12-day pulse . Although memory Tc17 cells also showed decreased basal homeostatic proliferation compared to effector Tc17 cells ( ~1 . 8-fold reduction ) , proliferation remained significantly higher in memory Tc17 cells than memory Tc1 cells ( 49% vs . 20% ) . Thus , vaccine-induced memory Tc17 cells appear to have a higher proliferative renewal than memory Tc1 cells . Homeostatic turnover of CD8+ T cells involves a similar number of cells undergoing apoptosis in order to maintain constant numbers[48] . Based on our BrdU data , we expected that memory Tc17 cells would display higher levels of apoptosis than Tc1 cells . Surprisingly , eYFP+ cells did not display staining for active caspase3/8 ( Fig 5B ) , albeit the cells were tested at a single time-point ( 76 days after vaccination ) . Likewise re-stimulation with anti-CD3 and -CD28 to enhance apoptosis revealed that memory Tc17 cells were indeed resistant to TCR signal-induced apoptosis ( S6 Fig ) . We also measured the levels of anti-apoptotic factors ( e . g . Bcl-2 , Bcl-xL , Mcl-1 ) and observed that memory Tc17 cells expressed significantly lower levels of Bcl-2 and Bcl-xL and similar levels of Mcl-1 compared with memory Tc1 cells ( Fig 5C ) . Memory Tc17 cells tended to express greater levels of these anti-apoptotic factors than naïve CD8+ T cells . Thus , antifungal memory Tc17 cells are maintained with higher levels of proliferation renewal than Tc1 cells , and although memory Tc17 cells express lower levels of the anti-apoptotic factors Bcl-2 and Bcl-xL , they have greater resistance to apoptosis . Memory Th17 cells express higher levels of Bcl-2 than memory Th1 cells[12 , 13] , linked to resistance to cell death[12] . However , our data revealed that anti-fungal memory Tc17 cells expressed lower levels of Bcl-2 than memory Tc1 cells; the Bcl-2 expression levels in memory Tc17 cells were nevertheless on par with naïve T cells ( Figs 4C & 5C ) , which require Bcl-2 for survival[49] . To investigate the role of Bcl-2 for memory Tc17 cell homeostasis in vivo , we inhibited Bcl-2 chemically with ABT-199 in vaccinated mice and analyzed memory CD8+ T cells . The total numbers of CD8+ T cells , both activated and naïve , were significantly reduced in lymph nodes but not spleens ( S7A Fig ) , indicating that Bcl-2 inhibition reduced lymph node size . Bcl-2 inhibition also significantly decreased the frequency and total numbers of IFNγ+ ( Tc1 ) cells in the lymph nodes ( Fig 6A ) , but not in the spleens . Central memory T cells are enriched in the lymph nodes suggesting that ABT-199 preferentially affects survival of central memory IFNγ+ ( Tc1 ) cells as previously described[50] . Despite the significant reduction in the total CD8 T-cell population and memory Tc1 cells in the lymph nodes upon Bcl-2 inhibition ( S7A Fig ) , the total number of IL-17A+ CD8 T-cells was unchanged in the draining lymph nodes ( Fig 6A ) . The frequencies and total numbers of CD8+eYFP+ ( Tc17 ) cells were unaffected in the spleens , as with Tc1 cells . Although Tc1 cells in the lymph nodes responded to Bcl-2 inhibition with increased proliferation , as assessed by Ki67 staining , Tc17 cells did not ( S7B Fig ) . Collectively , these results suggest that Bcl-2 is required for the maintenance and survival of anti-fungal central memory Tc1 cells , but not Tc17 cells . Hypoxia-inducible factor 1α ( HIF-1α ) plays a role during the induction and survival of Th17 cells[12 , 51] . HIF-1α also regulates the expression of Bcl-2 indirectly through Notch signaling in Th17 cells[12] . To our knowledge , the role of HIF-1α for generation of antifungal effector Tc17 ( or Tc1 ) cells has not been investigated . We first investigated the expression levels of HIF-1α in CD8+ eYFP+ vs . eYFP- cells . Basal expression of HIF-1α in both effector and memory CD8+ eYFP+ cells was low , but increased upon re-stimulation ( S8A & S8B Fig ) . We next investigated the functional role of HIF-1α on Tc17 and Tc1 cells by using the chemical inhibitor Echinomycin to block HIF-1α during the expansion phase , begun 4 days after vaccination . The proportion of effector Tc1 cells was significantly reduced by inhibition of HIF-1α , whereas the proportion of effector Tc17 cells was unaffected , indicating that HIF-1α is required for effector Tc1 cells but not for the sustenance or function of effector Tc17 cells ( Fig 7A ) . In a complementary approach , we generated bone-marrow chimera mice using CD4creHIF-1αfl/fl bone marrow cells . Following vaccination , we harvested spleens to assess the expansion of Tc17 and Tc1 cells . In contrast to Echinomycin treatment , we found a significant reduction in Tc17 cells that lacked HIF-1α intrinsically and a significantly augmented Tc1 response ( Fig 7B ) . To reconcile discrepant findings ( Fig 7A & 7B ) , we began Echinomycin treatment before vaccination and continued it afterward to mimic HIF-1α deficient chimeric mice . We found that Tc17 responses were reduced when HIF-1α was blocked both before and after the vaccination , in contrast to results when HIF-1α was blocked only after vaccination ( Fig 7C ) . These results reconcile the disparate findings of HIF-1α inhibitor treatment and HIF-1α-/- chimeric mice ( Fig 7A & 7B ) and suggest HIF-1α is required for differentiation but not sustenance of effector Tc17 cell responses . We showed above that differentiation but not sustenance of effector Tc17 cell responses requires HIF-1α . To our knowledge , the role of HIF-1α during homeostasis of memory Tc17 ( or Tc1 ) cells has not been investigated . We asked here whether HIF-1α is required for memory homeostasis of Tc17 and Tc1 cells . We took two pharmacological approaches . First , we gave Echinomycin for 10 days at day 90 after vaccination to temporally block HIF-1α activity ( Fig 8A ) . The proportion of memory Tc17 cells was significantly reduced by inhibitor treatment , whereas the proportion expressing IFNγ was not affected , indicating that HIF-1α is required for IL-17A expressing memory cells . We further analyzed the requirement for HIF-1α in memory Tc17 cells by gating on CD8+ eYFP+ cells ( S8D Fig ) . The frequency of memory CD8+eYFP+ cells was similar for the vehicle and inhibitor-treated groups ( S8C Fig ) , suggesting that HIF-1α chiefly affects cytokine expression . In a second pharmacological approach , we administered the HIF-1α agonist , Mimosine[52] , for 14 days starting at day 90 after vaccination ( Fig 8B ) . In contrast to the HIF-1α inhibitor , the agonist enhanced memory Tc17 cells , but again did not affect memory Tc1 cells , buttressing evidence for the selective requirement of HIF-1α for memory Tc17 homeostasis . Above , using chimeric mice that lack HIF-1α ( Fig 7B ) , we observed that T cell intrinsic HIF-1α is required for the differentiation of Tc17 cells . To assess the fate of effector Tc17 cells generated under these conditions , we rested these vaccinated mice for 90 days before analysis of Tc17 memory . Memory Tc17 ( and Tc1 cells ) were unexpectedly similar in all groups ( Fig 8C ) , suggesting a limited intrinsic role for HIF-1α during memory CD8+ T-cell homeostasis generated in the absence of HIF-1α . Our data together argue that while temporal blockade of HIF-1α negatively impacts memory Tc17 cell homeostasis , some memory Tc17 cells may be generated and persist independent of intrinsic HIF-1α or alternatively extrinsic sources of this factor may compensate for loss of intrinsic HIF-1α .
The long term persistence and plasticity of Th17 cells varies in different experimental models . The importance of these features carries particular significance in the setting of vaccine immunity where maintenance and fidelity of the phenotype is often required for durable resistance to infection . Here , we studied vaccine-induced Tc17 cells that are essential for resistance against lethal fungal pneumonia in hosts that lack CD4+ T cells . We report that vaccine-induced antifungal memory Tc17 cells are highly durable over a prolonged period of nearly one year in a murine model . The cells stably persist as memory cells , retain the ability to express IL-17A , mediate immunity upon challenge , do not undergo plasticity towards IFNγ ( yet do also express other type I cytokines important for fungal immunity ) , undergo high proliferative renewal , portray phenotypic markers that are consistent with “effector memory” cells , and are dependent on functional HIF-1α for homeostasis but not on Bcl-2 for survival . Whereas Th17 cells induced by systemic or mucosal bacterial infection are short-lived memory cells characterized by loss of CD27 expression[15] , we observed long-term persistence of antifungal Tc17 memory cells despite their low expression of CD27 . CD27 is a costimulatory molecule known to augment Tc1 cell proliferation and function , but it may be nonessential for Tc17 cell functions and compensated by other co-stimulatory molecules/cytokines such as CD43 , TLR2 , IL-1β , and IL-23[11 , 30 , 53 , 54] . Previous studies have reported the persistence of CD27lo Th17 cells following their transfer into animals after in vitro polarization[12 , 13 , 17] . The anti-fungal memory Tc17 cells investigated herein expressed high levels of TCF-1 , which is a marker of a stem cell-like signature[13] . It is not clear if mucosal-induced Th17 cells can become in situ memory cells in the absence of cognate antigen . Nevertheless , several in vivo and ex vivo studies in mice and humans have shown that Th17 cells can become long-term memory cells[12 , 13 , 47] . Taken together , several lines of evidence argue that Th17 cells are able to persist despite their low expression of CD27 . We extend these observations to include Tc17 cells . Studies of antifungal T cell responses have yielded conflicting results concerning persistence and fidelity of Th17 cells . A murine model of oropharyngeal ( mucosal ) Candida infection revealed that Th17 cells persist for weeks[18] . In contrast , intradermal acute Candida infection in mice induced Th17 cells that quickly lost IL-17A production[17] . Herein , we used a vaccination model in mice to track the persistence and fidelity of Tc17 cells for nearly one year . We observed that Tc17 cells were robustly induced and they durably maintained their IL-17-producing phenotype for the entire period . Th17 cells induced during a cutaneous Candida infection might migrate to secondary lymphoid organs , but the transient exposure to antigen and unique microenvironment during acute cutaneous Candida infection may not be enough to stabilize a Th17 cell phenotype for memory T cell formation . Nevertheless , our adoptive transfer studies demonstrated that vaccine-induced CD8+ eYFP+ T cells persist and produce IL-17A even in the absence of vaccine antigen . In many models , Th17 cells convert into IFNγ producing cells[23] . In the current study , however , we observed little conversion of Tc17 cells into IFNγ producing Tc1 cells , although the cells expressed TNFα and GM-CSF ( Fig 3 ) . The vaccine-induced effector CD8+ eYFP+ donor cells that were adoptively transferred produced comparable levels of IL-17A in both WT and immunodeficient TCRα-/- recipients , and little plasticity towards IFNγ , implying the importance of a distinct initial microenvironment for phenotype stabilization[24 , 55] . Interestingly , the percentage of antifungal Tc17 cells producing multiple cytokines increased when the cells were exposed to vaccine antigen for an extended period ( Fig 3 ) . One possible explanation is that prolonged exposure to fungal ligands and an inflammatory milieu may stabilize the polycytokine-producing phenotype of Tc17 cells[56] . Further studies are needed to elucidate how and what microenvironment shapes the stable Tc17 responses following fungal vaccination . Memory Th1/Tc1 cells are subdivided into central or effector memory cells distinguished by the surface markers CD62L and CCR7 , respectively[10] . Our data showed that antifungal Tc17 cells are effector memory cells as observed in human memory Th17 cells[12] . Notably , antifungal memory Tc17 cells retained expression of the prototypical transcription factor Ror ( γ ) t , while expressing lower levels of T-bet . Our data also showed that the majority of memory Tc17 cells expressed surface CD103 , an integrin marker of tissue resident memory cells , although the significance of this marker for persistent antifungal immunity needs further study . In general , homeostasis of memory T cells is governed by two key cytokines , IL-7 and IL-15 , which are part of the common γc cytokine family . We found that memory Tc17 cells display high levels of CD127 ( IL-7Rα chain ) and CD122 ( IL-2Rβ chain ) , implying similarity with memory Tc1 cell homeostasis . Nevertheless , we found that memory Tc17 cells belong to the class of “effector memory” , and in contrast to effector memory Tc1 cells , Tc17 cells undergo higher proliferation renewal[5] . The cytokines , IL-7 and IL-15 , are required for intermittent proliferation and also promote survival by enhancing anti-apoptotic factors such as Bcl-2 and Bcl-xL , and downregulating apoptotic factors[57 , 58] . Our studies using a specific Bcl-2 inhibitor revealed that Tc17 cells were unaffected , while central memory Tc1 cells were reduced in the draining lymph nodes and accompanied by increased cell proliferation . Our findings suggest that memory Tc17 cell homeostatic survival is independent of Bcl-2 and that Tc17 cells may depend on the other anti-apoptotic factors for their survival . While central memory Tc1 cells require Bcl-2 [50] , the low levels of Bcl-2 in Tc17 cells probably makes them resistant to Bcl-2 inhibition and may promote their high basal proliferative renewal . HIF-1 family members are transcription factors that regulate survival and function of many cells , including T cells under hypoxic conditions , while also promoting carcinogenesis[59] . Abrogation of HIF-1α inhibits differentiation of Th17 cells , and blockade leads to apoptosis of Th17 cells[12 , 51] . Ablation of the HIF-1 negative regulator , VHL , boosts cytolytic CD8+ T cell responses , enhancing viral clearance and suppressing tumor growth . Here , we found disparate requirements for HIF-1α during effector and memory Tc17 responses and also between Tc17 and Tc1 cells . After vaccination , during the expansion phase , antifungal Tc1 ( IFNγ+ ) cells required HIF-1α for their sustenance of effector cell responses , a feature consistent with prior work on Tc1 cells exposed to persistent antigen[60] . Antifungal Tc17 cells were different: while differentiation of effector Tc17 cells required HIF-1α , in line with Th17 cells in a prior study[51] , sustenance of effector Tc17 during the expansion phase did not required HIF-1α . Of note , genetic ablation of HIF-1α intrinsically in T cells or extended use of an HIF-1α inhibitor begun prior to vaccination enhanced the Tc1 cell responses . This is likely due to a reciprocal increase of Tc1 cells over Tc17 cells[61] , partly caused by poor induction of RORγt in HIF-1α-/- T cells[62] . Memory Tc17 cells required HIF-1α for their homeostasis and maintenance of an IL-17A expression phenotype , and this requirement distinguished memory Tc17 and Tc1 cells . This assertion is supported by the results of temporal intervention with Echinomycin in vaccinated mice . Conversely , our bone-marrow chimera studies revealed that a population of effector Tc17 cells lacking HIF-1α-/- intrinsically became memory cells without loss of phenotype or numbers , suggesting either HIF-1α independent generation of memory Tc17 cells or compensation from the extrinsic compartment . We do not know the oxygen level in the microenvironment of antigen presentation and CD8+ T cell expansion in vivo; however , effector Tc17 cells may overcome the requirement of HIF1-α following differentiation while memory Tc17 cells may have a distinct immunometabolism[63 , 64] , accounting for the HIF-1α requirement for maintenance of an active IL-17A locus in addition to survival during the memory phase . Further studies using cell-specific inducible HIF-1α are required to dissect its role in effector and memory Tc17 homeostasis . It is notable that memory Tc1 cells were relatively unaffected by inhibition of HIF-1α , suggesting that anti-neoplastic drugs such as Echinomycin might be used without affecting some tumor-specific memory CD8+ cell functions . In conclusion , we have shown that antifungal Tc17 cells display a set of appealing and unexpected features that are relevant to their function in a vaccine setting . These vaccine-induced Tc17 cells persist as long-lasting memory cells without undergoing plasticity towards IFNγ production , while they express other type I cytokines and have high levels of proliferative renewal for homeostasis . Tc17 cells expressed low levels of Bcl-2 compared to Tc1 cells , and Bcl-2 was not required for the maintenance of memory Tc17 cells . Memory Tc17 cells here were uniquely dependent on functional HIF-1α for the expression of the signature cytokine IL-17A . These results may inform the rational design of vaccines that are dependent on long-lived IL-17A producing cells and perhaps guide the development of immune modulators to alleviate immunopathology and autoimmunity , or enhance immunity against cancer .
Wild type C57BL/6 and B6 . SJL-PtprcaPepcb/BoyCrl mice were obtained from National Cancer Institute/Charles River Laboratories . Breeding pairs of IL17atm1 . 1 ( ( icre ) Stck/J ( Stock 016879 ) and B6 . 129X1-Gt ( ROSA ) 26Sortm1 ( EYFP ) Cos/J ( Stock 006148 ) were purchased from Jackson Laboratories and were intercrossed to maintain a heterozygous IL-17Acre locus in order to analyze both eYFP and functional IL-17A protein expression[17] . Homozygous breeding pairs of B6 . 129S2-TCRatm1Mom/J and B6 . PL-Thy1a/CyJ also were purchased from Jackson Laboratories . Conditional HIF-1α KO ( CD4CreHIF-1αfl/fl ) mice were a generous gift from Dr . Fan Pan ( School of Medicine , John Hopkins University ) . All mice were 7–8 weeks of age at the time of vaccination . Mice were bred , housed and cared for following strict guidelines of the University of Wisconsin Animal Care Committee , who approved all aspects of this work . All animal procedures were performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health . Care was taken to minimize animal suffering . The work was done with the approval of the IACUC of the University of Wisconsin-Madison who approved the relevant animal protocol number MOO969 . The wild-type virulent strain of Blastomyces dermatitidis was purchased from American Type Culture Collection ( ATCC; strain #26199 ) . The isogenic , attenuated mutant lacking BAD1 ( strain #55 ) was used for all vaccinations[38] . Isolates of B . dermatitidis were cultured and maintained as yeast on Middlebrook 7H10 agar with oleic acid-albumin complex ( Sigma-Aldrich ) at 39°C . Mice were vaccinated with B . dermatitidis strain #55 ( ~5 x 105cfu ) subcutaneously at two sites , dorsally and at the base of tail . For challenge studies , the virulent B . dermatitidis strain #26199 ( ~104 cfu ) was inoculated intratracheally . To ennumerate fungal burden , lung tissue was homogenized and plated on brain heart infusion ( BHI; Difco ) agar . All antibodies were purchased from BD Bioscience except antibodies directed against CD43 ( clone 1B11 ) , KLRG-1 and CXCR3 obtained from Biolegend; CD44 , CD127 , IL-22 , T-bet , Ror ( γ ) t , Eomes , IL-21 , FoxP3 from eBioscience; Bcl-2 , Bcl-xL , Mcl-1 active Caspase 3 , active Caspase 8 , TCF-1 from Cell Signaling; HIF-1α from Novus Biologics , and GFP from Life Technologies . All experiments were done in mice depleted of CD4+ T-cells; GK1 . 5 mAb was given in a weekly dose of 100μg/mouse by the intravenous route . The dose and interval was enough to deplete CD4+ T cells with an efficiency of 95%[36] . For in vivo IL-17A neutralization , mice were given 300 μg of anti-IL-17A mAb intravenously every other day . Both mAbs were purchased from Bio X Cell , West Lebanon , NH , while control IgG for IL-17A neutralization studies was obtained from Sigma-Aldrich . Single cell suspensions were re-stimulated with anti-CD3 ( 0 . 1 μsg/ml ) and anti-CD28 ( 1 μg/ml ) antibodies in the presence of Golgi Stop at 37°C for 5 hrs . Cells were surface-stained in 2% BSA/PBS buffer , fixed with Fix/Perm buffer ( BD Biosciences ) and stained for intracellular cytokines in 1X Perm/Wash buffer ( BD Biosciences ) . Cytokine-producing CD8+ T cells were analyzed by flow cytometry using BD LSRII and FACS Aria . Single cell suspensions from lymph nodes and spleens were subjected to a CD8+ T-cell magnetic bead-enrichment kit ( BD Bioscience ) to purify CD8+ T cells . Numbers of CD8+ eYFP+ cells were enumerated by flow cytometry , and equal numbers of CD8+ eYFP+ cells were adoptively transferred into naïve WT and TCRα-/- recipient mice by the intravenous route . Single cell suspensions were re-stimulated , surface-stained and fixed with Phosflow Lyse/Fix buffer and Phosflow Perm/Wash buffer ( BD Biosciences ) . Cells were then stained for intracellular cytokines and transcription factors ( TCF-1 , Ror ( γ ) t and T-bet ) simultaneously . Staining of FoxP3 , HIF-1α and EOMES was done using a FoxP3 buffer kit ( eBioscience ) . In some experiments , cells were first fixed with Perm/Fix buffer ( BD Bioscience ) followed by addition of antibodies to stain cytokines and eYFP ( anti-GFP antibody ) , then subjected to transcription factor staining . Cells were first surface-stained followed by intracellular staining for apoptosis factors using the BD Perm/Fix buffer kit . In some experiments , cells were subjected to intracellular staining for cytokines along with apoptosis factors following ex vivo re-stimulation . We performed BrdU pulse treatment ( 0 . 8mg/ml drinking water , DW ) for twelve days . Cells were surface-stained followed by intracellular staining for cytokines using the BD Perm/Fix kit . Later , cells were subjected to anti-BrdU antibody staining using a BrdU kit according to manufacturer’s instructions ( BD Pharmingen ) . BrdU incorporation in DNA of proliferating cells was analyzed by flow cytometry . Bcl-2 specific inhibitor , ABT-199 , was purchased from APExBIO ( Houston , TX ) and was re-suspended in 60% Phosal 50G , 30% Polyethylene Glycol 400 and 10% ethanol as described[65] . The inhibitor was administered ( 20 mg/kg body weight ) daily by oral gavage for 10 days . On day 11 , tissues were harvested and CD8+ T cells were analyzed by flow cytometry . HIF-1α inhibitor , Echinomycin , was purchased from Cayman Chemical Company and re-suspended in 100% methanol . Echinomycin was administered intraperitoneally at the concentration of 20–30 μg/kg body weight in sterile 1X PBS every other day . A total of 5 doses were administered during the expansion and memory phases over 10 days , unless indicated . The HIF-1α agonist , Mimosine , was purchased from Sigma Aldrich and used at the concentration of 70 mg/kg body weight by s/c route every other day for 14 days during memory phase . Bone marrow cells from congenic donor mice ( WT/Thy1 . 1+ and CD4Cre-HIF-1αfl/fl/Ly5 . 2+ ) were transferred into lethally irradiated Ly5 . 1+ recipient mice . For generation of mixed-bone marrow chimera mice , the donor cells were mixed in 1:1 ratio and were co-transferred into Ly5 . 1+ recipient mice . After 2 months rest , mice were vaccinated to assess the role of HIF-1α in CD8+ T cells for expansion and generation of memory Tc17 cells . All statistical analysis was performed using a two-tailed unpaired Student t test except for analysis of fungal CFUs , which was measured by the non-parametric Kruskall-Wallis ( one-way ANOVA ) . Prism 5 ( GraphPad Software , Inc . ) software was used to analyze all statistics . A two-tailed P value of ≤0 . 05 was considered statistically significant . | CD4+ T-cell deficient patients such as those with AIDS and idiopathic CD4+ T-cell lymphopenia are vulnerable to systemic fungal infections . We previously showed that CD8+ T cells can be exploited in CD4+ T cell deficient hosts for vaccine immunity against lethal fungal pneumonia in mice and that IL-17A production by these cells ( Tc17 ) is essential . Existing dogma holds that IL-17A producing CD4+ T cells ( Th17 ) are highly plastic , unstable , and convert into IFNγ producing cells , losing the capacity to produce IL-17A , which is the signature feature of Tc17 cells . Here , we show that vaccine-elicited antifungal Tc17 cells are maintained as stable and long-lasting memory cells that resist conversion into IFNγ cells ( Tc1 ) and protect CD4+ T cell deficient hosts against lethal pulmonary fungal infection . Antifungal Tc17 cells displayed features that define classical memory cells . However , memory Tc17 exhibited different requirements than Tc1 cells in the factors that promote T cell survival , including anti-apoptotic molecules Bcl-2 and Bcl-xl , and HIF-1α , which aids survival of cells in lower oxygen conditions found during inflammation . Thus , our study reveals that fungal vaccination elicits a durable , stable population of Tc17 cells with distinct features of survival needed for preventing infection in immunocompromised hosts . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"blood",
"cells",
"antimicrobials",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"immune",
"physiology",
"cytokines",
"drugs",
"immunology",
"microbiology",
"antifungals",
"neuroscience",
"learning",
"and",
"memory",
"prevent... | 2017 | Antifungal Tc17 cells are durable and stable, persisting as long-lasting vaccine memory without plasticity towards IFNγ cells |
Characterizing the spatial distribution of proteins directly from microscopy images is a difficult problem with numerous applications in cell biology ( e . g . identifying motor-related proteins ) and clinical research ( e . g . identification of cancer biomarkers ) . Here we describe the design of a system that provides automated analysis of punctate protein patterns in microscope images , including quantification of their relationships to microtubules . We constructed the system using confocal immunofluorescence microscopy images from the Human Protein Atlas project for 11 punctate proteins in three cultured cell lines . These proteins have previously been characterized as being primarily located in punctate structures , but their images had all been annotated by visual examination as being simply “vesicular” . We were able to show that these patterns could be distinguished from each other with high accuracy , and we were able to assign to one of these subclasses hundreds of proteins whose subcellular localization had not previously been well defined . In addition to providing these novel annotations , we built a generative approach to modeling of punctate distributions that captures the essential characteristics of the distinct patterns . Such models are expected to be valuable for representing and summarizing each pattern and for constructing systems biology simulations of cell behaviors .
Fluorescence microscope images can provide important information about the subcellular location of proteins , and automated systems can be used to assign these proteins to major subcellular location classes with accuracy at or above that of human annotators [1 , 2] . However , assigning higher resolution annotations to proteins is more difficult , especially for punctate or vesicular patterns . Punctate subcellular localization patterns may arise either from membrane-bound organelles ( e . g . , transport vesicles ) or from macromolecular complexes of sufficient size ( e . g . , ribonucleoprotein ( RNP ) bodies ) , and they may be quite visually similar . We refer to individual components of these patterns collectively as puncta , to encompass both types of structures . These are important for various cellular tasks such as endocytosis , exocytosis and RNA recruitment , storage or degradation . A critical factor for accomplishing many of those tasks is the association of the vesicles or bodies with cytoskeletal components such as microtubules for intracellular transport . Although microtubules are not necessary for short-range transport , they are required for rapid transport of vesicles [3] . The extent to which the distributions of specific puncta are related to that of microtubules remains unclear , as is the extent to which the distributions vary across different cell lines . Our understanding of cell behavior and the sources of cellular variation can be significantly aided and tested using cell modeling and simulations [4–6] . For this , we need a mechanism to capture the spatiotemporal behavior of cellular substructures , both as a starting point for simulations and to compare against results . Towards this end , we have previously described systems for building image-derived , 2D or 3D generative models of the distributions of either punctate organelles [7 , 8] or microtubules [9] within cells . These models are conditional ( dependent ) on models of cell and nuclear membranes , but they are independent of each other; that is , they do not consider the relationship between puncta and microtubules . Here we describe a new computational method that allows us to model this relationship . Our method requires images in which both punctate proteins and microtubules are visualized . The Human Protein Atlas ( HPA , http://proteinatlas . org ) is a rich source of such images , containing high-resolution images of subcellular location patterns for thousands of proteins in several cell lines [10] . To analyze the patterns of punctate proteins in the HPA , we designed a generative model consisting of compact and interpretable features to characterize the population of puncta within a cell , including measurements of microtubule association , relationship to cell geometry , density , intensity and appearance . We have used the features of these models to discover the major modes of variation among punctate patterns , and to assign subclasses of punctate patterns to unannotated proteins .
We began by creating an image processing pipeline that identified individual puncta and microtubules in 2D confocal microscopy images from the HPA . As illustrated in Fig 1A , an input image ( Fig 1C ) is processed to create images of puncta and microtubules ( shown as a composite in Fig 1D ) and of the remaining background protein fluorescence ( Fig 1E ) . One of our major goals was to generate a model of the distribution of puncta that captures their relationship to microtubules . This would presumably reflect the extent to which puncta were bound to microtubules to accomplish transport to or retention in particular regions of the cell . As a simple measure of this association , we computed the distance ( d ) between each punctum and the nearest microtubule ( Fig 1B ) . We would expect puncta that are bound to microtubules to have a small distance compared to those that are not bound , and perhaps also that the distribution of distances would reflect the extent to which released vesicles diffuse away before being bound again . We added this measure to our previous vesicular object distribution model [8] , which included dependence on fractional distance between the nucleus and plasma membrane ( r , calculated from L1 and L2 ) and the angle ( α ) to the major axis of the cell ( see Methods ) . We also created a model for background intensity that was similarly dependent on microtubules and cell shape ( see Methods ) . We combined the estimated parameters from these models with five parameters that describe puncta size and shape and two parameters that measure the amount of fluorescence in puncta and background . This resulted in twenty-two parameters ( S1 Table ) that can be readily determined from each image of a protein’s subcellular distribution in an individual cell . We used these parameters both as features to describe protein patterns and , later , to construct generative models of punctate patterns . A number of proteins in the HPA are assigned annotations of “vesicles” or “cytoplasm” . We considered whether we could use HPA images to assign these proteins to a more specific organelle or structure . By examining UniProt annotations and primary literature for proteins whose subcellular location has been reasonably well characterized , we selected eleven proteins that are found in eleven specific types of punctate patterns ( Table 1 ) ( we refer to these proteins as “founders” since they enabled us to define specific subtypes ) . We chose these patterns due to the fact that the proteins showed a similar pattern across all three cell types in the HPA and they represent a wide range of membrane and non-membrane bound compartments ( although there are of course additional punctate patterns for which we did not find appropriate founders ) . In particular , they cover all main compartments of the endomembrane system . We calculated the feature values for all cells for each combination of the eleven proteins and three cell lines . We verified that the features accurately reflect the relationship between vesicles and microtubules by comparing the cumulative distribution of the experimentally measured distance between puncta and microtubules with that calculated from the model; the distributions were very similar for all eleven patterns ( S1 Fig ) . We then asked whether these patterns could be distinguished from each other in HPA images . To provide a visual basis for illustrating how the proteins differed in the features , we calculated the first three principal components . Fig 2 shows the position of each antibody-cell line combination in two projections of this three-dimensional space , as well as representative images along each principal axis . For a given cell line , the eleven patterns are roughly separable , although the position of a given protein sometimes varies from cell line to cell line . For example , proteins 2 , 3 , 6 and 7 are close together in pc1 and pc2 but separated by pc3 . From inspection of the projection of each numerical feature onto the three most significant principal component axes , as well as the example images , it appears that the first component primarily represents variation in features 12 , 13 and 5 , which capture relationship to microtubules and variation in intensity . The second primarily represents variation in features 21 , 22 , 2 , and 8 , which capture intensity and distance from the nucleus , while the third principal component represents variation in features 1 , 3 , and 4 , which capture puncta size and variation in size . This figure does not permit accurate assessment of the overlap between patterns , but is presented to give a visual overview of the major modes of variation with the patterns . These results suggest that the feature set may be a reliable basis for measuring variation in punctate patterns , and we therefore sought to determine whether we could use them to predict the compartmental localization of other proteins to one of the eleven patterns . To do this , we first used the features to construct a classification accuracy-derived separability statistic to compare two collections of cells ( see Methods ) and assessed the extent to which the eleven patterns could be distinguished . We used a classification approach based on Bayes error rate in order to avoid problems with imbalance between the numbers of proteins in each class and to allow for class-specific differences in scale for different features ( see Methods ) . For each cell type separately , we classified each image as belonging to one of the eleven patterns using hold-out-image cross validation: for each held-out image , we calculated the separability between the cells contained in that image and the cells of each of the founder patterns . The image was given the label of the pattern that was least separable from it . Using this method for each cell type we achieved an average class accuracy of 86 . 9% ( Table 2 ) . We compared these results to those using the same classification procedure but excluding the features relating to microtubule distribution , which resulted in 82 . 8% average accuracy . This demonstrates that the relationship to microtubules provides information that improves our ability to distinguish punctate patterns . Further examination of Table 2 reveals that the coated pits pattern is the only one that is consistently difficult to distinguish . This may in part be due to the fact that 2D confocal images were used , and thus the features cannot easily distinguish whether puncta are on the surface or inside the cell ( for the other surface puncta pattern , caveolae , their distribution or size must allow them to be distinguished ) . We next asked whether the classification approach could be used to assign a punctate subpattern annotation to an image of proteins other than the founders . We did not want to simply assign the subcellular location of the class that a protein was most similar to ( since the protein might not actually be from any of our classes ) , but wanted to ensure that we only assigned annotations for proteins with a high degree of similarity to one of the founders . For each cell type , we determined a threshold on the separability statistic that could be used to determine whether or not a new protein should be assigned to a particular class . This threshold was determined as the optimal point of the receiver operating characteristic curve ( see Methods and S2 Fig ) for each cell type . To assign subcellular location to a new image , we measured the separability between it and each founder pattern . If the value for one of the patterns was below the threshold , we assigned the corresponding pattern label to that image . In the rare case of an image being below the threshold of multiple patterns , we assigned it the label “ambiguous . ” This classification procedure was applied to the remainder of images in the HPA dataset; the results are contained in S1 Dataset . One hundred and twenty-five proteins were identified as belonging to one of the eleven classes in A-431 , 60 in U-2OS , and 365 in U-251 MG . The list of the most confident assignments is shown in Table 3 . With the goal of providing improved annotations for protein databases , we also generated an XML file that can be used to update those databases . The file ( S2 Dataset ) contains information on HPA antibody IDs , gene targets and proposed annotation . Due to the nature of immunofluorescence tagging , a sequence-specific tag may be present on more than one protein isoform , each of which may show a condition-specific localization pattern . With that in mind , we also report the known protein gene products provided by ENSEMBL 79 , and the percentage of matching peptides after alignment between the gene-product and antigen sequences in the region spanned by the antibody . We also provide annotations to all protein isoforms that match the antibody sequence . For those proteins and isoforms that have a high confidence location assignment , we also provide an XML file for updating their UniProt record ( S3 Dataset ) . In order to provide an independent assessment of the accuracy of the annotation procedure , we searched for literature describing the localization of the most confident annotations . We were able to find literature supporting our proposed labeling for many of the proteins ( although they had often only been analyzed in other cell types ) . For example , of the top hits for A-431 cells , BRD4 , has been suggested to be involved in the lysosome protolytic pathway [11] . For U-2OS , top hit RAB5C is a classic early endosomal protein [12] , and prohibitin ( PHB ) is a multifunctional membrane protein [13] one of whose roles is in regulation of degradation of PAR1 [14] . For U-251MG cells , the top hits include cathepsin H ( CTSH ) , a lysosomal enzyme , DTX3L , which regulates endosomal sorting [15] , and LY6K , which , like other Ly6 antigens , is associated with glycosylphosphatidyl inositol-anchored glycoproteins ( such as TEX101 [16] ) that are typically found in caveolae . These findings increase our confidence in the proposed annotations . Many of the proteins analyzed ( which were all proteins assigned “vesicles” or “cytoplasm” annotations ) were not assigned with high confidence to any of the 11 patterns . There are at least three potential reasons for this . First , the staining may be of low enough intensity or quality that foreground cannot be adequately identified . Second , the unassigned proteins may be cytoplasmic proteins without a discernible punctate pattern , or vesicular proteins from an organelle that we have not considered . Third , they may be present in more than one of the eleven patterns , such that their pattern does not match well enough to any of them . Our models allow us to ask whether different punctate subclasses differ in their relationship to microtubules . We performed a simple characterization of this relationship by calculating the average actual distance of each punctum from microtubules , as well as the average distance from microtubules predicted by our fitted model . S3 Fig shows a comparison of these two distances for each pattern across all cell types and for each combination of pattern and cell type . A confidence interval on the average distance from microtubules was determined via the Tukey-Kramer method after two-way ANOVA [17] ( across proteins and cell types ) . All of the symbols are quite near the diagonal , indicating that the model is in high agreement with the measurements . When averaged across all three cell types , retromer , recycling endosomes , and early endosomes show the closest association with microtubules , and RNP bodies , COPI vesicles and coated pits show the least . When each combination of protein and cell type is considered separately , we see greater variability in the distances ( perhaps due to differences in microtubule-binding proteins or cell size or shape ) . COPII , lysosomes and COPI show the least variation across the three cell types , and coated pits and recycling endosomes show the greatest . Another way in which we can compare the different patterns is by examining the differences in the model features among them . A simple visualization of this is shown in Fig 3 , in which the relative values of each feature are shown for each pattern . In U-2OS , for example , the first four features ( relating to size and intensity ) clearly distinguish the group of RNP bodies , late endosomes , recycling endosomes , lysosomes and COPII from the others , and a high value for mx5 ( number of puncta ) separates RNP bodies from this group . Other distinguishing features or feature combinations can also be identified , such as retromers having the lowest value for mx12 ( consistent with their close association with microtubules ) . These differences provides a interpretable rationale for the ability of the classifiers to distinguish the patterns . A difficult question that frequently gives rise to controversy is how to best describe the subcellular pattern of a given organelle or structure ( especially a novel one ) . Descriptions using unstructured text or Genome Ontology terms defer the question by assuming that the words will be sufficient for the reader to be able to mentally construct the pattern . An alternative is to show an example image , but this does not give an idea of the variation in the pattern ( one can find differences between any two example images , but this does not address whether those differences are statistically significant ) . Unfortunately these two methods of conveying information about the distribution and variation in protein pattern do not provide a quantitative , or much less a probabilistic or statistical representation of the observed pattern . Alternatively , one can give values for a descriptive feature vector or matrix for each pattern ( which can be used for a classifier ) but this allows one only to recognize new examples but not to produce an example of the pattern . Feature vectors also do not necessarily allow an explicit model of the relationship between cell components . Of course , none of the approaches above are helpful if we desire an in silico representation of the cell geometry and expressed patterns ( i . e . , the consumer of the representation is a computer rather than a cell biologist ) . For example , information about subcellular patterns is needed for accurate mathematical simulations of cell biochemistry and behavior [4–6] . As a solution , we have introduced the building of generative models of cell organization directly from images [7 , 8 , 18–20] . The intent is for these models to capture the underlying properties of a particular pattern; in statistical terms , to capture the distribution from which all examples of that pattern are drawn . Such a model can be used to synthesize new cell images from that distribution . We therefore constructed a generative model of punctate patterns whose structure is shown in Fig 4 . The model starts with models of nuclear and cell shape ( dn , dc ) and microtubule distribution ( dm ) and links them to models of puncta distribution using mx7 through mx11 to capture dependence on cell shape and mx12 and mx13 to capture dependence on microtubules ( see Methods ) . Additionally the size , shape and intensity of vesicles are modeled independently of the cell shape and microtubules with mx1 through mx6 . The background intensity is similarly modeled dependent on cell shape and microtubules ( mx14 through 20 ) and scaled to match the fraction of intensity with mx21 and mx22 . We illustrate that the images generated from the models learned for each of the pattern classes are similar to real images in Fig 5 and S4 Fig . Assuming that the distributions of the eleven punctate patterns are independent of each other , we can combine the models and synthesize cells containing all eleven . Fig 6 shows an example of a “typical” cell under this assumption ( using the average values of all model parameters ) .
With the development of systems to fluorescently tag and acquire images of thousands of subcellular protein patterns , a need arose for automated methods to analyze and model the patterns in these images [21] . The goals of such analyses include , but are not limited to , determining the organelles to which different proteins localize and studying the statistical dependency between different protein patterns . However , previous methods have not been able to recognize subpatterns of the major organelle types . Furthermore methods are needed to describe the relationships between cellular components in a way that is not only human-interpretable , but allows us to generate new examples of these patterns for future use in cell simulations [22] . Here we have described a new framework to build models of subcellular punctate patterns conditional on cell geometry and microtubules . These models use interpretable features that capture specific ways in which punctate subpatterns differ between cell types ( such as the differences noted at the beginning of the Results ) and can generate synthetic cell instances representative of the modeled population . We demonstrated the value of this framework by learning models directly from images of eleven well-characterized punctate protein patterns in three cell types . We showed that the major variation in these patterns corresponded to dependence on microtubules , total intensity , and puncta size and shape . Given the model parameters we constructed a pipeline demonstrating both the high discriminative ability of this model across patterns of the same cell type and the ability to automatically assign annotations to 550 proteins ( many of which had been poorly characterized previously with respect to subcellular location ) . High-content screening and analysis have become increasingly frequent , including subtle analysis of location changes induced by chemical compounds or inhibitory RNAs and proteome-scale analysis of patterns . The features we have described should be useful for refining the ability to distinguish different vesicular and punctate patterns , and , most importantly , to provide an interpretable and portable basis for comparing them . The work presented here represents an important step towards bridging detailed models learned from large collections of images for proteins contained in discrete objects with models of microtubule network growth learned by inverse modeling [9 , 18] . It serves as an important component of our CellOrganizer project ( http://cellorganizer . org/ ) [20] , which aims at capturing a detailed model of the spatial organization and relationships between different subcellular location patterns . We plan to extend this work by merging it with models of subcellular pattern dynamics , as well as extend the model to capture further dependency between components . It is hoped that approaches like this will enable the construction of models that capture essential cell behaviors without requiring the simultaneous measurement of the thousands of different proteins in the same living cell , something that is infeasible with current technology .
The data used here were confocal immunofluorescence microscopy images of fixed cells from A-431 , U-2OS and U-251MG cell lines from HPA [10] . All antibodies whose subcellular pattern was annotated as “vesicles” or “cytoplasm” were chosen ( a total of 2357 , 3038 , and 1730 proteins for each line; S1 Dataset contains the complete list of proteins analyzed ) . The images were analyzed as 8-bit TIFF images with three channels each obtained using a different emission wavelength of fluorescence from a single image field . The three channels show the locations of a specific punctate protein , a nuclear stain , and microtubules . Each of the images is 1728 × 1728 pixels and the pixel size corresponds to 0 . 08 microns in the sample plane . Founder proteins for eleven patterns were chosen as described in the Results . After segmenting the image fields for these proteins into single cell regions using a seeded watershed method [2] , the set of founder images was found to contain 1099 cells , 333 from A-431 , 327 from U-2OS and 439 from U-251MG ( the number of cells for each of the 33 combinations of antibody and cell line varied from 12 to 85 ) . In cell images , due to variation in fluorescence intensity in the cytoplasm , segmentation of puncta and microtubules from protein pattern images poses a difficult problem where global threshold-based methods may over-threshold regions of the cytoplasm containing low-intensity structures . The input cell image was de-noised by blurring with a Gaussian filter with standard deviation of 0 . 75 . We isolated high spatial-frequency foreground and low spatial-frequency background intensity images by low pass filtering the smoothed image with a Gaussian filter of 4-pixel standard deviation , and subtracted this background image from the smoothed image , resulting in an image of high-frequency foreground signal ( i . e . puncta ) . The negative-valued pixels of the foreground signal were removed , and the foreground image was subtracted from the first smoothed image , to get the background image ( both of which sum to the total image intensity ) . To increase the speed at which a Gaussian mixture model could be fit over the foreground image , we excluded all pixels below the Ridler-Calvard threshold [23] and all single-pixel objects . We used the skeletonized foreground signal of the microtubule image to model the distances of objects from microtubules . This approach resulted in reasonable definition of both puncta and microtubules and was sufficient to capture variation across the founder patterns analyzed in this paper . The centroids of all puncta were computed by fitting a mixture of Gaussians to distinguish overlapping puncta [7] . The distance between the centroid of each punctum and its nearest microtubule was found using a distance transform of the skeletonized microtubule image . A probability density function ( PDF ) for the position of puncta ( pp ) relative to the cell geometry and microtubules was estimated by extending the model previously described [8] by adding a terms describing the distance from microtubules , d: P ( r , a , d ) = eβ0+β1r+β2r2+β3sinα+β4cosα+β5d+β6d21+eβ0+β1r+β2r2+β3sinα+β4cosα+β5d+β6d2 ( 1 ) The terms β1 through β4 describe the dependency of objects on radial and angular coordinates in relation to the shape of the cell [2 , 8] , and β5 and β6 describe the dependency of objects to be localized in relation to the microtubules . We similarly constructed a PDF for the background intensity ( which presumably results from soluble , non-punctate protein ) . The Bayesian hierarchical framework for the generative model for puncta is shown in Fig 3 as a graphical model . A multivariate statistical model was constructed from the independent distributions of values of the following statistics from each cell: puncta size ( sp ) , puncta per cell ( np ) , and intensity ( ip ) . Synthetic cell instances were created starting from the cell and nuclear boundaries and microtubule image of a randomly-selected cell . ( They can also be created by first generating cell and nuclear boundaries and microtubule distributions using models learned previously for the three cell lines [18] . ) To add puncta to a cell , values were sampled for the number of puncta per cell ( np ) and the size ( sp ) and fluorescence intensities ( ip ) ) for each punctum from distributions learned from 2D HPA data . These were used to generate puncta using the Gaussian object based generative model [8] . Positions for them were sampled from the vesicle position PDF from the model above after morphing to the specific cell geometry . Background fluorescence was added using the learned PDF from the background images , scaled to match a draw from the total background intensity distribution learned from images . The assignment of subcellular annotations to images of cells is a classification task with complications found in many biological contexts; specifically being the structured nature of data ( cells with the same antibody should all be assigned the same label ) , the inseparability of class data ( proteins with different biochemical properties may have similar localization patterns ) , and imbalanced number of observations ( some images may contain many cells while others have few ) . We designed a classification method to specifically address the above complications . Given pattern parameterizations corresponding to cells of two collections ( all cells contained in two images ) , we perform a balanced classification task to determine how distinguishable the two collections are . For each pair of images , we hold out a subset of cells and train an SVM by weighting the training data such that there is a uniform prior across the classes . We then classify the hold-out and count the frequency at which the hold-out was assigned the correct collection , approximating the Bayes Error rate [24] . This approach is similar to other methods used in genomics [25] . We take the average classification accuracy across all cell classification tasks ( whether or not the cells belonging to the two images are assigned the same subcellular pattern ) as a measure of how distinguishable the two collections are , resulting in a possible range of values from 1 ( totally separable ) to 0 ( completely inseparable ) . In virtually all cases , the measure of difference lies between 0 . 5 and 1 . We will refer to this measure as “dissimilarity” . To determine a threshold on dissimilarity , at which we can say two collections belong to the same or different patterns , the pipeline treats images of each of our basis patterns as their own collection ( with multiple images of each pattern ) and performs the above classification task using cells contained in each image . An ROC curve is constructed , indicating the true and false positive classification rates as a function of increasing dissimilarity . For each cell type we constructed an upper-bound of dissimilarity ( above which is considered “not the same annotation” ) by the cutoff determined at the location where the upper-left-most point of the ROC curve intersects with a slope of TN+FPTP+FN , where TN , FP , TP and FN are the counts of true negative , false positive , true positive and false negatives respectively . When comparing our basis set to images containing cells of unknown protein localization , we assign the unknown pattern the label of any basis pattern that is within the similarity threshold . These thresholds were 0 . 78846 , 0 . 70588 and 0 . 72093 for A-431 , U-2OS , and U-251MG , respectively . All software and data used for this work is available as a reproducible research archive ( http://murphylab . web . cmu . edu/software ) . The software will also be available as part of the open source CellOrganizer system ( http://CellOrganizer . org ) . The segmentation and feature calculation pipeline can be used separately . | Determining the subcellular location of all proteins is a critical but daunting task for systems biologists , especially when variation between different cell types is considered . Fluorescence microscopy is the main source of information about subcellular location , but large collections of fluorescence images for many proteins are frequently annotated visually and result in assignment only to broad categories . In this paper , we describe automated methods for analyzing images from the Human Protein Atlas to identify nine specific punctate patterns and assign these more specific annotations to 550 proteins many of which previously had little information about subcellular location . We also describe building models of these patterns that will be useful for carrying out systems biology simulations of cellular reactions using accurate spatial distributions . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Automated Learning of Subcellular Variation among Punctate Protein Patterns and a Generative Model of Their Relation to Microtubules |
Acetylcholine is the canonical excitatory neurotransmitter of the mammalian neuromuscular system . However , in the trematode parasite Schistosoma mansoni , cholinergic stimulation leads to muscle relaxation and a flaccid paralysis , suggesting an inhibitory mode of action . Information about the pharmacological mechanism of this inhibition is lacking . Here , we used a combination of techniques to assess the role of cholinergic receptors in schistosome motor function . The neuromuscular effects of acetylcholine are typically mediated by gated cation channels of the nicotinic receptor ( nAChR ) family . Bioinformatics analyses identified numerous nAChR subunits in the S . mansoni genome but , interestingly , nearly half of these subunits carried a motif normally associated with chloride-selectivity . These putative schistosome acetylcholine-gated chloride channels ( SmACCs ) are evolutionarily divergent from those of nematodes and form a unique clade within the larger family of nAChRs . Pharmacological and RNA interference ( RNAi ) behavioral screens were used to assess the role of the SmACCs in larval motor function . Treatment with antagonists produced the same effect as RNAi suppression of SmACCs; both led to a hypermotile phenotype consistent with abrogation of an inhibitory neuromuscular mediator . Antibodies were then generated against two of the SmACCs for use in immunolocalization studies . SmACC-1 and SmACC-2 localize to regions of the peripheral nervous system that innervate the body wall muscles , yet neither appears to be expressed directly on the musculature . One gene , SmACC-1 , was expressed in HEK-293 cells and characterized using an iodide flux assay . The results indicate that SmACC-1 formed a functional homomeric chloride channel and was activated selectively by a panel of cholinergic agonists . The results described in this study identify a novel clade of nicotinic chloride channels that act as inhibitory modulators of schistosome neuromuscular function . Additionally , the iodide flux assay used to characterize SmACC-1 represents a new high-throughput tool for drug screening against these unique parasite ion channels .
Flatworms of the genus Schistosoma are the causative agents of the debilitating parasitic infection schistosomiasis , afflicting over 230 million people in 74 endemic countries [1] . The majority of human schistosomiasis can be attributed to three species- S . mansoni , S . japonicum and S . haematobium- which cause a wide spectrum of chronic pathology , including hepatosplenomegaly , portal hypertension and squamous cell carcinoma [1] . Currently , praziquantel ( PZQ ) is the only drug used to treat schistosomiasis and there is no vaccine available . Widespread and exclusive use of PZQ has led to concerns of emerging drug resistance . Laboratory strains of PZQ-resistant S . mansoni have been successfully generated and there are now several reports of reduced PZQ cure rates in the field [2] , [3] . Moreover , PZQ is ineffective in killing larval schistosomulae [4] . The stage-limited efficacy of PZQ and looming prospect of drug resistance signal the importance of exploring novel therapeutic targets for the treatment of schistosomiasis . An area of interest for the treatment of helminth parasites is the neuromuscular system , which is targeted by the majority of currently approved and marketed anthelminthics [5] . Inhibition of neuromuscular activity provides two modes of treatment . First , motor inhibition may interfere with parasite maturation , which is closely tied with migration during the larval stage [6] . Second , a loss of muscle function would disrupt essential activities , including attachment to the host , feeding , mating and others [7] , ultimately causing the parasite to be eliminated from the host . The cholinergic system has proved especially successful as a neuromuscular anthelminthic target . Common antinematodal drugs such as levamisole , pyrantel and monepantel [5] , [8] , and the antischistosomal drug , metrifonate [9] , all disrupt neuromuscular signaling by interacting with proteins of the worm's cholinergic system . Acetylcholine ( ACh ) is an important neurotransmitter in both vertebrate and invertebrate species . The neuromuscular effects of ACh are typically mediated by postsynaptic nicotinic acetylcholine receptors ( nAChRs ) , so named because of their high-affinity for nicotine . Structurally , nAChRs are members of the Cys-loop ligand-gated ion channel ( LGIC ) superfamily . They form homo- and heteropentameric structures , which are organized in a barrel shape around a central ion-selective pore [10] . Vertebrate nAChRs are invariably cation-selective ( Na+ , Ca2+ , K+ ) and mediate excitatory responses . Invertebrates , on the other hand , have both cation and anion-selective ( Cl− ) ACh-gated channels . The latter mediate Cl− - driven membrane hyperpolarization and therefore are believed to play a role in inhibitory responses to ACh . One example of these unique invertebrate receptors is the acetylcholine-gated chloride channel ( ACC ) of the snail , Lymnaea , which is structurally related to nAChRs , yet is selective for chloride ions [11] . In addition , nematodes have an unusual type of ACC , which is a functional acetylcholine-gated chloride channel but is more closely related to other chloride channels ( GABA and glycine receptors ) than nAChRs [12]–[13] . A defining feature of the ACCs is the presence of a Pro-Ala motif in the pore-lining M2 domains of the constituent subunits . This motif , which has been shown to confer anion-selectivity to other LGICs , replaces a Glu residue normally found in the cation-selective channels [14] . ACCs have not been identified in any of the flatworms , free-living or parasitic . However , there is experimental evidence supporting an inhibitory role for ACh in the parasites , which could be mediated by this type of receptor . Early studies in the 1960s observed that addition of exogenous cholinergic agonists to parasite cultures caused flaccid paralysis of adult trematodes and cestodes [15]–[16] . Flaccid paralysis indicates muscular relaxation and is in direct contradiction to the excitatory response of tonic contraction expected from cholinergic stimulation . Later research established a causal relationship between activation of a nicotinic-like receptor in S . mansoni muscle fibers and the flaccid paralysis caused by ACh in whole worms [17] . However , this work was performed in the pre-genomic era and no attempt was made to clone or characterize the receptors involved . More recently , the publication of the S . mansoni genome [18] has provided cause to revisit the unusual inhibitory activity of ACh in schistosomes . Several candidate genes have been annotated as nAChR subunits [18]–[19] and the present work aims to confirm the presence of and functionally characterize cholinergic chloride channels in S . mansoni . One strategy that has been used for assessing the therapeutic value of candidate genes in parasites , particularly helminths , is RNA interference ( RNAi ) [20]–[22] . A strength of this reverse genetics strategy is the ability to screen living animals for phenotypic and behavioral changes as a result of abrogation of a particular gene's function , as demonstrated by the large-scale screens in the free-living platyhelminth cousins of schistosomes , the planarians [23] . The RNAi pathway is conserved in S . mansoni [20]–[21] and has previously been used to probe the neuropeptidergic system of the parasite [24] and , more recently , the serotonergic system as well [25] . However , the effects of silencing other important neuroactive pathways , such as the cholinergic system , are not known . Here we describe a novel clade of anion-selective nAChR subunits ( SmACCs ) that appear to be invertebrate-specific . The ion channels formed by these subunits play an inhibitory role in the neuromuscular activity of the parasite , as suggested by the results of RNAi and pharmacological behavioral assays , their tissue distribution and pharmacological properties .
A Puerto Rican strain of S . mansoni-infected Biomphalaria glabrata snails were kindly provided by Dr . Fred Lewis ( Biomedical Research Institute and BEI Resources , MD , USA ) and used for all experiments . To obtain larval schistosomula , 6–8 week-old snails were exposed to bright light for 2 hours at room temperature . The resulting cercarial suspension was mechanically transformed in vitro by vortexing , washed twice with Opti-MEM ( Gibco ) containing 0 . 25 µg/ml fungizone , 100 µg/ml streptomycin and 100 units/ml penicillin and cultured in Opti-MEM/antibiotics supplemented with 6%FBS ( Gibco ) [26] . To obtain adult worms , 40-day old female CD1 mice were injected intraperitoneally with 250 mechanically transformed schistosomula [26] . After 8 weeks , adult worms were collected by perfusion of the mouse hepatic portal vein and mesenteric venules , as previously described [26] . Animal procedures were reviewed and approved by the Facility Animal Care Committee of McGill University ( Protocol No . 3346 ) and were conducted in accordance with the guidelines of the Canadian Council on Animal Care . To generate a target list of putative nicotinic acetylcholine receptor ( nAChR ) subunits , the S . mansoni Genome Database was searched using the keywords “nicotinic” and “acetylcholine receptor” [18]–[19] . A BLASTp homology search was also performed using the Torpedo nAChR ( AAA96704 . 1 ) as a query . The resulting list of nAChR subunit sequences was used as a query against the general NCBI protein database and aligned with other Cys-loop receptor superfamily proteins by CLUSTALX [27] . The alignments were analyzed manually to identify the presence of the vicinal C motif , indicative of nAChR α-subunits , and key amino acids involved in ion-selectivity . Phylogenetic trees were built in PHYLIP using the neighbor-joining method and bootstrapped with 1 , 000 replicates [28] . Trees were visualized and annotated using FigTree3 . 0 [29] and manually inspected to ensure that bootstrap values for each node were above a 70% threshold . Five putative nAChR subunits were targeted by RNA interference ( RNAi ) : Smp_157790 , Smp_037960 , Smp_132070 , Smp_176310 ( SmACC-1 ) and Smp_142690 ( SmACC-2 ) . For each target sequence , we amplified a unique 200–300 bp PCR fragment by RT-PCR . Total RNA was extracted from pooled adult male and female S . mansoni , using the RNeasy Micro Kit ( Qiagen ) and reverse-transcribed with MML-V ( Invitrogen ) and Oligo-dT ( Invitrogen ) . PCR amplification was performed with a proofreading Phusion High Fidelity Polymerase ( New England Biolabs ) , according to standard protocols . PCR primers ( Table S2 ) were designed using Oligo6 . 2 [30] and the unique fragment sequences were identified by BLAST analysis . Amplicons were ligated to the pJET1 . 2 Blunt Vector ( Fermentas ) and verified by sequencing of multiple clones . For synthesis of double-stranded RNAs ( dsRNA ) , the T7 promoter sequence ( 5′-TAATACGACTCACTATAGGGAGA-3′ ) was added to both ends of each target fragment by PCR . Long dsRNAs were generated from the resulting T7-flanked PCR products by in vitro transcription of both DNA strands , using the MegaScript T7 Transcription Kit ( Ambion ) , according to the kit protocol . The dsRNAs were subsequently digested with RNAseIII , using the Silencer siRNA Kit ( Ambion ) , to generate a mixture of siRNAs for each target . The siRNA was quantitated and assessed for purity using a Nanodrop ND1000 spectrophotometer . Larval schistosomula were obtained by the standard protocol ( see above ) with some modification . After the final wash , freshly transformed schistosomula were re-suspended in Opti-MEM without antibiotics or FBS and plated at a concentration of 100 animals/well in a 24-well plate . Animals were transfected using siPORT NEO FX Transfection Agent ( Ambion ) and either an irrelevant scrambled siRNA ( Ambion ) or nAChR subunit-specific siRNA at a final concentration of 50 nM . Transfections were performed blind to rule out selection bias during analysis . Opti-MEM containing antibiotics and supplemented with 6%FBS was added to transfected schistosomula 24 hours post-treatment . A previously described larval motility assay was performed 6 days post-transfection [31] . Briefly , schistosomula were filmed for 45s using a Nikon SMZ1500 microscope equipped with a digital video camera ( QICAM Fast 1394 , mono 12 bit , QImaging ) and SimplePCI version 5 . 2 ( Compix Inc . ) software . Three distinct fields were recorded for each well . ImageJ ( version 1 . 41 , NIH , USA ) software was then used to quantitate worm motility using the Fit Ellipse algorithm in ImageJ , as described [25] . The data shown here are derived from three independent experiments in which a minimum of 12 animals was measured per experiment . Pharmacological motility assays were carried out with 6-day old schistosomulae in the same manner , but without the transfection with siRNA . Baseline measurements of schistosomula motility were recorded prior to drug addition . Compounds of interest ( arecoline , nicotine , mecamylamine , D-tubocurarine ) were subsequently added at a final concentration of 100 µM and larval motility was measured again after 5 minutes exposure . Viability of drug-treated and siRNA-treated schistosomula was routinely monitored by a dye exclusion assay , according to the method of Gold [32] . Six-day old siRNA-treated schistosomula were washed twice with 1X PBS , re-suspended in the lysis buffer provided with the RNEasy Micro RNA Extraction Kit ( Qiagen ) and sonicated with 6 pulses of 10 s each . Total RNA was then extracted from the lysate following the manufacturer's instructions . RNA was quantified and assessed for purity using a Nanodrop ND1000 spectrophotometer . 100 ng total RNA was used for each 20 µl MML-V ( Invitrogen ) reverse transcription ( RT ) reaction , which was performed according to standard protocols . A negative control lacking reverse transcriptase was also prepared in order to rule out contamination with genomic DNA . Quantitative real-time PCR ( qPCR ) was performed using the Platinum SYBR Green qPCR SuperMix-UDG kit ( Invitrogen ) in a 25 µl reaction volume . Primers located in a unique region of each gene and separate from those regions used to generate siRNA were designed using Oligo6 . 2 and may be found in Table S2 . Primers targeting the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase ( GAPDH , Accession #M92359 ) were used as an internal control and are as follows: forward 5′-GTTGATCTGACATGTAGGTTAG- 3′ and reverse 5′-ACTAATTTCACGAAGTTGTTG-3′ . Primer validation curves were generated to ensure similar efficiency of target and housekeeping gene amplification . Cycling conditions were as follows: 50°C/2 min , 95°C/2 min , followed by 50 cycles of 94°C/15 s , 57°C/30 s , 72°C/30 s . Cycle threshold ( Ct ) values were normalized to GAPDH and then compared to the scrambled siRNA control , as well as an off-target gene ( another nAChR subunit ) to ensure transcript-specific silencing . All expression data was analyzed using the comparative ΔΔCt method [33] and was generated from three separate experiments done in triplicate . Two putative anion-selective subunit sequences , Smp_176310 ( SmACC-1 ) and Smp_142690 ( SmACC-2 ) were chosen for further study and cloned by conventional RT-PCR ( see above ) using primers targeting the beginning and end of each cDNA . For SmACC-1 we used primers: forward 5′-ATGGATCTAATATACTTG-3′ and reverse: 5′-TTAGGTAGTTTCTTCTG-3′ . PCR conditions were as follows: 98°C/30 s , 30 cycles of 98°C/10 s , 55°C/60 s , 72°C/90 s and final extension of 72°C/5 min . In the case of SmACC-2 , the full-length cDNA was amplified with primers 5′-ATGGAAAAATCACTTATTCG-3′ ( forward ) and 5′-TTATTGTAGATCAACTACG-3′ ( reverse ) , using the following cycling conditions: 98°C/30 s , 30 cycles of 98°C/10 s , 54°C/60 s , 72°C/60 s and a final extension of 72°C/5 min . The 5′ end of SmACC-2 was further verified by 5′ RACE ( rapid amplification of cDNA ends ) , using a commercial kit ( Invitrogen ) and a gene-specific primer for the reverse transcription [5′-GCAGGTACATAATCTGAG-3′] , according to manufacturer's instructions . All PCR products were ligated to the pJet1 . 2 Blunt cloning vector ( Thermo Scientific ) and verified by DNA sequencing of at least two independent clones . Peptide-derived polyclonal antibodies were generated in rabbits against subunits SmACC-1 and SmACC-2 ( 21st Century Biochemicals – Marlborough , MA ) . Animals were injected with a mixture of two specific peptides per target . For SmACC-1 , the two peptides 1 ( NAKVNRFGKPHGNKFC ) and 2 ( CSKKALSAANAKWNSPLQY ) are located in the third intracellular loop of the protein . For SmACC-2 , peptide 1 ( TDGEAERHIRHEDRVHQLRSVC ) and peptide 2 ( LQNINMKQIKLEYKNSLGC ) are located at the N- and C-terminal ends , respectively . All peptides were conjugated to the carrier protein ovalbumin and were BLASTed against the S . mansoni genome database and the NCBI general database to ensure specificity . Whole antisera were tested for specificity and titer against both immunogenic peptides by ELISA . The anti-nAChR-specific IgG fractions were affinity-purified , using beads that were covalently attached to a mixture of the two peptide antigens added in equal amounts . Peptide conjugation to the beads and subsequent affinity purification were performed with the Pierce Sulfolink Kit for Peptides ( Thermo Scientific ) , according to manufacturer's instructions . ELISA was performed to determine the titer of affinity-purified antibody fractions . Protein was quantified by the Bradford assay , using a commercial kit ( BioRad , USA ) . A mouse monoclonal anti-FLAG M2 antibody was purchased from Sigma-Aldrich . Parasites were prepared for confocal microscopy according to previously described protocols [34] , [35] . Briefly , 6-day old in-vitro-transformed schistosomula or freshly collected adult worms were washed two times in 1X PBS and fixed in 4%PFA for 4 hours at 4°C . Parasites were washed twice , each for 5 minutes in 1X PBS containing 100 µM glycine and then permeabilized with 1%SDS in 1X PBS for 25 minutes [36] . After permeabilization , animals were incubated overnight at 4°C in antibody diluent ( AbD ) containing 0 . 1%Tween-20 , 1% BSA in PBS to block non-specific binding . After 3 washes of 10 minutes each in the AbD , animals were then incubated with affinity-purified anti-SmACC-1 or anti-SmACC-2 ( 1∶100 ) for three days at 4°C . Samples were then washed 3 times in AbD and incubated in secondary antibody ( 1∶1000 ) conjugated to Alexa Fluor 488 or 594 ( Invitrogen , USA ) . In some experiments , tetramethylrhodamine B isothiocyanate ( TRITC ) -conjugated phalloidin ( 200 µg/ml ) was added with secondary antibody and used to visualize the musculature . Secondary antibody incubation lasted for 2 days and animals were again washed three times before mounting for microscopy . Slides were examined using a Zeiss LSM710 confocal microscope ( Carl Zeiss Inc . , Canada ) equipped with the Zeiss Zen 2010 software package . The lasers used for image acquisition were an Argon 488 nm and a HeNe 594 nm , with the filter sets adjusted to minimize bleed-through due to spectral overlap . Negative control slides were prepared by incubating samples in pre-immune serum , secondary antibody only ( primary antibody was omitted ) or primary antibody preadsorbed with 0 . 5 mg/mL of mixed peptide antigen ( 0 . 25 mg/ml of each peptide ) . At least 5 independent samples were examined for each peptide-derived antibody . Membrane-enriched protein fractions were extracted from adult S . mansoni using the ProteoExtract Native Membrane Protein Extraction Kit ( Calbiochem , USA ) and following the manufacturer's instructions . Protein was quantified by the Bradford Assay ( BioRad , USA ) and used for SDS-PAGE and Western blot analysis . Approximately 20 µg of membrane extract was loaded on a 4–12% Tris-Glycine gel ( Invitrogen , USA ) and resolved by SDS-PAGE , then transferred to a PVDF membrane ( Millipore , USA ) . A standard Western blot protocol was followed to visualize proteins . Primary antibodies used were peptide-purified anti-SmACC-1 or anti-SmACC-2 ( both 1∶1000 ) . Secondary antibody ( 1∶5000 ) was goat-anti-rabbit conjugated to horseradish peroxidase ( Invitrogen , USA ) . Membranes were also probed with peptide antigen-preadsorbed primary antibody ( 1∶1000 ) as a negative control . For mammalian expression studies , a human codon-optimized construct of SmACC-1 was synthesized ( Genescript , USA ) and inserted into the pCi-Neo ( Promega ) expression vector , using NheI and SmaI restriction sites . A C-terminal FLAG tag was also included in the SmACC-Neo construct to aid in the monitoring of expression . HEK-293 cells were grown to 50% confluence in Dulbecco's Modified Essential Media ( DMEM ) supplemented with 20 mM HEPES and 10% heat inactivated fetal calf serum . Cells were transiently transfected with the humanized SmACC-1 construct or empty vector , using XtremeGENE 9 transfection reagent ( Roche ) , as recommended by the manufacturer . 24 hours post-transfection , cells were transduced with Premo Halide Sensor ( Invitrogen ) , a halide-sensitive fluorescent indicator used to assess ligand-gated chloride channel function [37]–[38] . Following transduction , cells were incubated at 37°C , 5% CO2 overnight and seeded onto a 96-well plate at a density of 50 , 000 cells per well . After an 8 hour incubation at 37°C , 5% CO2 , the cells were equilibrated with iodide assay buffer provided with the Premo Halide Sensor assay kit for at least 30 minutes at 37°C in the reading chamber of a FlexStation II scanning fluorometer ( Molecular Devices ) . YFP fluorescence was measured for 10 s before and up to 2 minutes after the addition of test agonists . Agonists were added at a final concentration of 100 µM , or as indicated , in a total sample volume of 200 µl . Water was used as a vehicle-only negative control . Antagonist assays were performed the same way , except the cells were pre-incubated with 100 µM cholinergic antagonist ( mecamylamine , D-tubocurarine , atropine ) for 30 min at 37°C prior to addition of 100 µM nicotine . Receptor activity was calculated by measuring the reduction in YFP fluorescence ( YFP quench ) due to iodide influx over the time of measurement . Briefly , a fluorescence measurement was taken 10 s after the addition of drug ( Relative fluorescence ( RF ) initial ) and again after a period of 120 s ( RFfinal ) . The RFfinal was subtracted from the RFinitial to generate ΔRF . ΔRF was then divided by the RFinitial and multiplied by 100 , resulting in a measurement of %YFP quench , as described [38] . Readings were normalized to water-treated controls and reported as Fold-Change in YFP Quench [39] . Receptor activation was also calculated by the linear-regression slope method [40] with similar results . The minimum quench threshold for all experiments was set at zero [41] . Dose response curves were fitted using the non-linear regression function of Prism 6 software ( Graphpad Software , USA ) . Student's t-tests were performed to determine statistically significant differences at P<0 . 05 . Calcium assays were performed using the Calcium 4 FLIPR Assay Kit ( Molecular Devices , USA ) with a FlexStation II fluorometer ( Molecular Devices ) , according to the kit protocol and as described previously [42] . Briefly , HEK-293 cells expressing SmACC-1 were preloaded with a cell-permeable fluorescent calcium indicator 48 hr post-transfection , as per the kit protocol , and treated with 100 µM nicotine , 100 µM acetylcholine or water vehicle . The concentration of calcium in the extracellular medium was ≈2 mM . Intracellular calcium was measured before addition of agonist to obtain a baseline and immediately following agonist addition at 1 . 52 s intervals for a total of 120 s . Calcium responses were calculated as peak fluorescence levels after subtraction of the baseline , as described [42] , and experiments were repeated twice ( two independent transfections ) , each with six replicates . In situ immunofluorescence assays in transfected HEK-293 cells were performed according to standard protocols , using either affinity-purified anti-SmACC-1 antibody ( 1∶500 ) or a commercial monoclonal anti-FLAG ( M2 ) antibody , as described previously [43] .
A combination of BLAST and keyword searches were used to generate a list of potential nAChR subunits in the genome database of S . mansoni [18] . In total , nine putative receptor subunits were identified . All sequences were predicted to have the defining features of a nAChR subunit , including a Cys-loop motif and four transmembrane domains [44] and all subunit genes identified are predicted to contain full-length coding sequences . A structural alignment of the putative schistosome nAChR subunits with two previously characterized human nAChR alpha subunits , the Lymnaea nicotinic chloride channels and the crystal structure of the Torpedo nAChR suggests the presence of both cation and anion-selective schistosome nAChR subunits . Figure 1 shows the M2 domain of the structural alignment in which the Torpedo , human and two of the schistosome receptor subunits contain a conserved glutamate at the M2 interface , which is the hallmark of cation-selective Cys-loop channels . In contrast , the remaining schistosome nAChR subunits , including SmACC-1 ( Smp_176310 ) and SmACC-2 ( Smp_142690 ) and the Lymnaea subunits display a Pro-Ala motif at this position . The Pro-Ala motif is associated with anion-selectivity in Cys-loop receptors [14] . Previous mutagenesis studies have shown that replacing the M2 glutamate of a vertebrate nAChR with Pro-Ala is sufficient to convert the ion-selectivity of the channel from cationic to anionic [45 , 46 , see 47 for review] . The predicted schistosome nAChRs were then aligned with cation and anion-selective Cys-loop receptor subunits from other representative vertebrate and invertebrate species , including the acetylcholine-gated chloride channel ( ACC ) subunits from C . elegans [12] . A phylogenetic tree of the alignment ( Figure 2 ) shows the unique clade formed by the Pro-Ala motif-containing schistosome nAChR subunits is located firmly in the larger group of cation-selective nAChR subunits . Also present in this clade are the nicotinic chloride channel subunits of the snail Lymnaea [11] and putative homologs from fellow flatworms Clonorchis and Dugesia . This is in contrast to the C . elegans ACC subunits , which group more closely to the anion-selective GABA/glycine receptors and have low affinity for nicotine [12] . Thus , the nAChR subunits in schistosomes are all structurally related to cation-selective nicotinic receptors but those carrying the Pro-Ala motif appear to have diverged and may have acquired selectivity for anions . The structural relationship of the schistosome sequences to known chloride-selective nAChRs of Lymnaea reinforces the notion that these are nicotinic anion channels . Moreover , the presence of putative homologs in closely related flatworms and their apparent absence in host species indicate that these receptors may be good targets for broad-spectrum antiparasitics . Two of the predicted anion-selective subunits , SmACC-1 and SmACC-2 were selected for full-length cloning . SmACC-1 contains a predicted ORF of 2415 bp distributed over 9 exons , encoding a protein of 92 kDa . SmACC-1 contains an N-terminal signal peptide and an N-terminal double cysteine motif ( YxCC ) that is the defining characteristic of nAChR alpha-type subunits [48] . Full-length SmACC-1 was successfully amplified by PCR and sequencing of multiple SmACC-1 clones verified the predicted ORF ( GenBank accession # KF694748 ) . The coding sequence of SmACC-2 was predicted to be 2745 bp . However , further sequence analysis by BLAST predicted a large ( ∼1 kb ) N-terminal nucleotide-binding domain ( NBD ) , a feature not normally present in Cys-loop receptors . This excess sequence may have been a result of the concatenation of two distinct proteins during annotation . To identify the correct start codon of SmACC-2 , 5′RACE experiments were performed and an alternative start site downstream of the predicted start codon was identified , removing the NBD sequence . New PCR primers were designed and full-length SmACC-2 was amplified , resulting in a product of 1528 bp and a corresponding protein of 60 kDa ( GenBank accession # KF694749 ) . The new SmACC-2 coding sequence was in frame with the predicted ORF and retained both its Cys-loop and transmembrane domains but does not contain a signal peptide . SmACC-2 also lacks the vicinal cysteine motif , suggesting that it is a non-alpha-type nAChR subunit . A previously described behavioral assay [25] , [31] was used to evaluate the effect of cholinergic compounds on S . mansoni larval motility . Animals were treated with either cholinergic agonists ( arecoline , nicotine ) or antagonists ( mecamylamine , D-tubocurarine ) alone at a concentration of 100 µM and the frequency of body movements ( shortening and elongation ) was calculated as a measure of motility [25] , [31] . Treatment of 6-day old schistosomula with cholinergic agonists caused rapid , near complete paralysis when compared to the water-treated controls ( Figure 3A ) . Conversely , the nicotinic antagonists caused a 2-3 . 5-fold increase in larval motility . These results are consistent with previous studies [reviewed in 49] and support the hypothesis that cholinergic receptors inhibit neuromuscular function in S . mansoni . To examine the role of the predicted anion-selective nAChR subunits in larval motor behavior , we targeted individual nAChR subunits by RNA interference ( RNAi ) , using pooled sequence–specific siRNAs . A mock–transfected sample ( lipid transfection reagent only ) and a nonsense scrambled siRNA control were included as negative controls; there was no significant decrease in motor behavior in either control compared to untransfected larvae . In contrast , animals treated with nAChR siRNAs all showed a significant ( P<0 . 05 ) hyperactive motor phenotype ( Figure 3B ) . Depending on the subunit , the increase in larval motility ranged from 2-4-fold when compared to the negative scrambled control . The two subunits generating very strong hyperactive phenotypes were SmACC-2 ( ∼6-fold ) and SmACC-1 ( ∼4 . 5-fold ) . The hyperactivity in the nAChR RNAi-treated animals is consistent with the phenotype seen in animals where nAChR activity has been pharmacologically abrogated by receptor antagonists ( Figure 3A ) . Knockdown at the mRNA level was confirmed by quantitative qPCR for SmACC-1 and SmACC-2 ( Figure 4A ) . SmACC-2 expression was reduced 60% at the transcript level and SmACC-1 expression was reduced by 90% . In both cases the knockdown was observed only in RNAi-suppressed larvae , indicating the effect was specific . Transfection with SmACC-1 siRNAs had no effect on the expression level of the other subunit , SmACC-2 , or vice-versa ( Figure 4A ) . Knockdown at the protein level was confirmed by western blot analysis of SmACC-1 , using a specific antibody ( Figure 4B ) . The siRNA-treated animals show a drastic reduction in protein expression , as evidenced by the absence of the expected 92 kDa band in the treated sample lane , whereas no difference was seen in the loading control . In order to determine the tissue localization of SmACC-1 and SmACC-2 , we obtained custom commercial antibodies against each target . Polyclonal antibodies were generated using two unique peptide antigens for each gene of interest , each peptide being conjugated to ovalbumin . The antibodies were peptide affinity-purified and tested by ELISA and western blotting . Adult worm membrane fractions probed with anti-SmACC-1 antibody showed a predominant band at ≈100 kDa . Probing with antibodies specific for SmACC-2 resulted in a single band of ≈ 65 kDa . These bands are slightly larger than the predicted sizes ( 92 kDa and 60 kDa , respectively ) , possibly due to glycosylation of the native proteins . Control samples in which the antibody was pretreated with an excess of peptide antigen ( preadsorbed control ) showed no immunoreactivity , indicating specificity of binding for the intended protein . For the immunolocalization study , adult and larval schistosomes were stained with either anti-SmACC-1 or anti-SmACC-2 and an Alexa-488 conjugated secondary antibody . Some animals were counterstained with TRITC-conjugated phalloidin to label muscle and cytoskeletal features . The results suggest that SmACC-1 and SmACC-2 are both localized to the peripheral nervous system ( PNS ) of the worm ( Figure 5 ) , a region that is rich in cholinergic neurons [50] , [51] . Cholinergic neurons are also present in the brain and main nerve cords of the central nervous system ( CNS ) [50] , [51] but we did not observe significant labeling in these regions , either with anti-SmACC-1 or anti-SmACC-2 antibodies . Within the PNS , SmACC-1 immunoreactivity can be seen in fine varicose nerve fibers , resembling beads on a string , which are repeated along the length of the body ( Figure 5 A ) . Close inspection of the confocal stacks suggests these are minor nerve cords of the vast submuscular plexus that innervates the body wall muscles [52] . Similarly anti-SmACC-2 staining revealed numerous varicose nerve fibers in the peripheral innervation of the body wall ( Figure 5B ) . Some of these nerve fibers can be seen criss-crossing the length of the body , where they come into close contact with the musculature . However there was no visible overlay between the antibody labeling ( green ) and the phalloidin-stained muscles ( red ) , either for SmACC-2 ( Figure 5B ) or SmACC-1 , suggesting these receptors are expressed in nerve tissue rather than the muscle itself . Other regions where specific immunoreactivity was detected included the nerve plexuses of the suckers , which were labeled by both anti-SmACC-1 and 2 antibodies , and the surface of the worm . Surface labeling was observed only with the anti-SmACC-2 antibody and it occurred in both males and females , though it was particularly enriched in the male tubercles ( Figure 5C ) . It is unknown if this labeling is associated with the tegument itself or possibly sensory nerve endings that are present on the surface of the worm . No comparable fluorescence could be seen in any of the negative controls tested , including a peptide-preadsorbed antibody control ( Figure 5E , F ) and therefore the labeling is considered to be specific . Immunolocalization studies were repeated in larval schistosomula and the labeling patterns of SmACC-1 and 2 were found to be similar . In both cases , immunoreactivity occurred in a network of fine varicose nerve fibers that run just below the surface and along the entire length of the body ( Figure 5D ) . This resembles the expression pattern seen in the adults and suggests the receptor is expressed in the developing PNS of the larvae . As with the adults , we were unable to detect specific labeling in the CNS of the larvae with either antibody . HEK-293 cells were transfected with codon-optimized ( humanized ) SmACC-1 and protein expression was monitored by in situ immunofluorescence . SmACC-1 was selected for these studies because it is a predicted alpha-like subunit and therefore it is capable , in principle , of forming functional homomeric channels [10] . Initial attempts to express the native ( non-humanized ) SmACC-1 proved unsuccessful . The codon-optimized sequence , however , expressed significant levels of protein in the HEK-293 cells . The transfected cells were immunoreactive for SmACC-1 when probed either with specific antibody ( Figure 6A ) or anti-FLAG antibody targeting the C-terminal FLAG epitope . No immunofluorescence was noted in the negative control cells transfected with empty plasmid ( Figure 6B ) . Cells expressing codon-optimized SmACC-1 were transduced with a YFP sensor ( Premo Halide Sensor ) and seeded on a 96-well plate for the iodide ( I− ) flux assay . The principle of the assay has been described in detail [37]–[40] and is shown schematically in Figure 6C . Cells expressing a chloride channel of interest are bathed in an iodide buffer , which serves as a surrogate for chloride ( Cl− ) anions . After a period of equilibration , test compounds are added and if the chloride channel of interest is activated , an influx of I− occurs , quenching the fluorescence of the YFP sensor . Channel activity was quantified by measuring either the slope of the curve or the decrease in fluorescence following drug addition , as described [39] . Figure 6D shows representative tracings of cells expressing SmACC-1 and mock-transfected cells , each treated with 100 µM nicotine . Activation of SmACC-1 ( red circles ) by nicotine caused a significant decrease in YFP fluorescence compared to nicotine-treated mock-transfected cells ( black circles ) . No significant reduction in fluorescence was seen in SmACC-1 expressing cells treated with water , suggesting the YFP quench was agonist-dependent . In separate experiments , we also tested whether SmACC-1 was able to transport calcium in the HEK-293 cells , using a kit-based calcium fluorescence assay . This was done in part to verify the ion selectivity of the channel and also to address the possibility that the YFP quench might be due to indirect activation of an endogenous calcium-sensitive chloride channel . However these experiments showed no evidence of calcium influx through SmACC-1 . Cells expressing SmACC-1 were treated with 100 µM nicotine or 100 µM ACh and there was no effect of either agonist on intracellular calcium levels ( data not shown ) . Thus we rule out an indirect effect of calcium on I− transport and conclude that SmACC-1 is a cholinergic anion channel , as predicted from the bioinformatics analysis . The I− flux ( YFP sensor ) experiments were repeated with different test substances and the results are shown in Figure 7 . None of the compounds used stimulated a significant influx of I− in the mock control . In contrast the cells expressing SmACC-1 were responsive to several cholinergic agonists , particularly nicotine . Treatment with nicotine ( 100 µM ) caused a significant ( P<0 . 05 ) ≈6-fold increase in YFP quench in cells expressing SmACC-1 . Smaller but statistically significant responses were also seen with other cholinergic agonists ( ACh , choline chloride , carbachol and arecoline ) . Non-cholinergic substances , including biogenic amines ( serotonin ( 5HT ) , dopamine ) and glutamate , had no effect on the cells ( Figure 7 ) . These data suggest that SmACC-1 is capable of forming a functional homomeric chloride channel that displays a preference for nicotine and related cholinergic substances . Furthermore , SmACC-1 was activated by nicotine in a dose-dependent manner with an EC50 = 4 . 3±1 . 4 µM ( Figure 7 , inset ) . To test if the channel is sensitive to inhibition by cholinergic antagonists , SmACC-1 – expressing cells were treated with nicotine ( 100 µM ) in the presence and absence of “classical” ( mammalian ) nicotinic antagonists ( D-tubocurarine , mecamylamine ) or the muscarinic ( GAR ) antagonist , atropine , each at 100 µM . Of the drugs tested , only D-tubocurarine was able to significantly block the activation of SmACC-1 by nicotine ( Figure 8 ) . The other two drugs , mecamylamine and atropine were ineffective at this concentration .
Acetylcholine ( ACh ) has long been known as the quintessential excitatory neurotransmitter of the vertebrate neuromuscular system . Signaling through cation-selective nAChRs , ACh mediates muscular contraction via membrane depolarization due to an influx of Na+ or Ca2+ . More recently , a distinct class of anion-selective nAChRs and other types of acetylcholine-gated chloride channels ( ACCs ) has been reported in several invertebrate organisms , including mollusks and nematodes [11] , [12] . These chloride-permeable channels initiate membrane hyperpolarization , causing an inhibition of action potentials . However , none of these invertebrate channels has been directly implicated in the control of motor function . The effects of ACh on invertebrate neuromuscular activity vary depending upon the organism in question . As in vertebrates , ACh has excitatory neuromuscular effects in many invertebrate phyla , including some helminths such as nematodes and planarians [53] , [54] . In trematodes , however , ACh appears to act in exactly the opposite manner . Exogenous application of cholinergic agonists onto trematodes in culture causes a rapid flaccid paralysis due to relaxation of the body wall muscles [15] , [55] . A similar type of paralysis was observed in tapeworms ( cestodes ) treated with exogenous ACh [16] . This inhibitory response to cholinergic drugs appears unique to parasitic flatworms ( trematodes and cestodes ) , and the receptors mediating this activity may therefore hold value as a therapeutic target . Earlier electrophysiology studies of S . mansoni tentatively identified these receptors as nAChR-like based on their pharmacological properties [17] but the receptors themselves were not identified . The sequencing of the S . mansoni genome [18]–[19] led to the annotation of several candidate nAChR subunit genes , which are the focus of the present work . Using a combination of BLAST and keyword searches , a total of nine nAChR subunit genes were found in the genome of S . mansoni . A structural alignment of the schistosome nAChR subunits with the Torpedo nAChR was then performed to identify peptide motifs associated with ion-selectivity . Cation-selective ion channel subunits have a negatively charged intermediate ring , formed by the presence of Glu residues in the M1-M2 linking region [56] . Anion-selective Cys-loop receptor subunits replace the Glu in this region with a Pro-Ala motif , disrupting the electrostatic interactions in the intermediate ring and conferring anion-selectivity to the channel [14 , 45 , 46 see 47 for review] . The results of our structural alignment indicate that 5 of the schistosome nAChR subunits ( SmACC-1 , SmACC-2 , Smp_157790 , Smp_037910 and Smp_132070 ) contain this anion-selectivity determinant and they were tentatively identified as S . mansoni SmACCs . Furthermore , a dendrogram analysis suggests that the SmACCs are evolutionarily distinct from the ACCs found in C . elegans . Unlike the C . elegans ACCs [12] , the schistosome subunits are structurally related to vertebrate and invertebrate nAChRs , suggesting that the SmACCs are descended from ancient nicotinic channels but have evolved selectivity for chloride . This allies the SmACCs more closely with the anion-selective nAChRs of the snail Lymnaea [11] , with which they share more than 40% identity at the protein level . Interestingly , certain species of Lymnaea are permissive intermediate hosts of schistosomes . However , it is unclear if the presence of anion-selective nicotinic channels in both organisms is due to horizontal gene transfer , common ancestry or convergent evolution . There is also evidence of closely related , putative nAChR chloride channels present in the genome of the trematode Clonorchis sinensis [57] , suggesting a unique clade of platyhelminth-specific nicotinic chloride channels . The next step after identifying the SmACCs was to study their role in the motor function of the parasite . The flaccid paralysis of adult schistosomes caused by treatment with cholinergic compounds is well characterized . However , very little is known about the role of cholinergic receptors in the motor activity of larval schistosomula . Given that larval migration is vital to parasite development and survival [6] and the cholinergic system is a major regulator of motor function in adult worms , we hypothesized that SmACCs play an important role as inhibitory modulators in larval neuromuscular function . To test this , two types of behavioral assay were employed- pharmacological and RNAi . The results of the pharmacological motility assay agree with previous studies implicating ACh as an inhibitor of schistosome movement [15] , [17] . Treatment of 6-day old schistosomula with the cholinergic agonists arecoline and nicotine caused nearly complete paralysis whereas classical antagonists , mecamylamine and D-tubocurarine stimulated movement by 3–4 fold over water-treated control animals . These results suggest that the schistosome cholinergic system mediates inhibitory neuromuscular responses , possibly via an influx of chloride generated by SmACC activation . Although the results of the pharmacological motility assay agree with previously published studies , motor phenotypes elicited by treatment of worms with exogenous compounds are not necessarily of biological or behavioral relevance . Drug permeability across the tegument , non-selective targeting and toxic effects may all induce motor behaviors that obscure the role of the receptors in question . Silencing of receptor function by RNAi mitigates these issues by targeting receptors individually and by measuring effects on basal motor activity in the absence of added drugs . The results of our RNAi assay show that the ion channels formed by the SmACC subunits act as inhibitory mediators of motor activity in schistosomula . Knockdown of each of the 5 identified SmACC subunits resulted in a 3-6-fold hypermotile phenotype , mirroring the hyperactivity seen in antagonist-treated schistosomula . It is unclear why the individual subunits all produced similar hypermotile RNAi phenotypes . It is possible these are all components of the same inhibitory channel , such that the loss of any one subunit results in loss of channel function and hyperactivity . As discussed below , our immunolocalization studies show that two of these subunits , at least ( SmACC-1 and SmACC-2 ) have similar distribution patterns , suggesting they could be components of the same channel in the worm . Alternatively these could assemble into different channels that have similar inhibitory effects on movement . To identify the possible mechanisms by which the SmACCs mediate inhibitory motor responses , immunolocalization studies were performed by confocal microscopy . The tissue distribution of two SmACCs in which silencing elicited large hypermotile phenotypes , SmACC-1 and SmACC-2 , was examined in adult and larval stages of the parasite . The most significant expression was observed in the peripheral innervation of the worm's body wall , both for SmACC-1 and SmACC-2 . Counterstaining with phalloidin suggests that neither subunit is expressed directly on the musculature . Rather , SmACC-1 and SmACC-2 were detected in minor nerve fibers of the submuscular nerve net that innervates the somatic muscles . This suggests that SmACC-1 and SmACC-2 mediate their inhibitory motor effects in an indirect manner , perhaps by modulating the release of other neurotransmitters or by acting as autoreceptors . In flatworms , as well as vertebrate model systems , nicotinic receptors are well known to mediate the release of other neurotransmitters , including neuropeptides and dopamine [58]–[60] . In schistosomes , the cholinergic and neuropeptidergic system ( which is excitatory in flatworms ) , are in very close proximity [50] , [61] . The balance between these systems may , therefore , be an important factor in the regulation of motor behavior . It would be of interest to determine if ACh inhibits neuropeptide release through these receptors , and whether this inhibition might explain the flaccid paralysis and other motor effects of ACh in these parasites . SmACC-2 immunoreactivity was also seen on the surface of the parasite . Discreet , punctate staining is present along and in between the tubercles of adult male worms and along the surface of adult females . This marks the second time a nAChR has been localized to the schistosome tegument [62] . Surface nAChRs in schistosomes have previously been linked to modulation of glucose uptake and are postulated to act through tegumental GLUT-1 like transporters [63] . The possibility also exists that tegumental SmACC-2 may provide sensory cues affecting motor function . The tubercles are known to contain innervated sensory structures [64] , which interface with the peripheral nerve net below and ultimately the CNS . The presence of SmACC-2 at both of these locations points to a potential role for ACh and this receptor in mediating host-parasite interactions affecting worm motor behavior . While behavioral assays and microscopy serve to elucidate the behavioral role of the SmACCs , they provide only limited insight into receptor function at the molecular level . Therefore , functional expression analysis of a SmACC receptor was carried out in a heterologous expression system . A previous study cloned and expressed two cation-selective nAChR subunits from S . haematobium in Xenopus oocytes [65] . However , neither subunit was able to form a functional ion channel either alone or when co-expressed . Our initial attempts to express SmACC-1 and SmACC-2 failed to produce functional channels , either individually or in combination and in two different expression environments , HEK-293 cells and Xenopus oocytes ( data not shown ) . SmACC-2 lacks the YxCC motif of nAChR alpha-subunits and therefore is not capable of forming functional homomeric channels . Further examination with appropriate antibodies of cells transfected with the SmACC-1 subunit determined that the level of protein expression was low , which could explain the apparent lack of activity . It has been shown that differences in codon-usage can significantly decrease the expression of recombinant schistosome proteins in heterologous systems [66] . Thus we obtained a codon-optimized ( humanized ) cDNA for SmACC-1 and repeated the analysis in HEK-293 cells . The humanized construct produced higher levels of protein expression and some of this protein appeared to be correctly targeted to the cell surface , as determined by immunofluorescence analysis . Subsequent functional studies showed that human codon-optimized SmACC-1 produced a functional homomeric ion channel in HEK-293 cells . Several nAChR subunits are known to form functional homomeric channels in vivo . Examples of this include the vertebrate alpha-7 nAChR and the ACR-16 of C . elegans [67]–[68] . However , the expression of functional homomeric nAChRs is limited to neuronally expressed channels [69] . Moreover , only alpha-type nAChR subunits are capable of forming homopentameric channels . Thus , the formation of a functional homomeric SmACC-1 channel , together with its neuronal expression pattern in the worm , both suggest that SmACC-1 is a neuronal-type alpha nAChR subunit . Activity assays were performed using a relatively novel , fluorescence-based assay , the Premo Halide Sensor ( Invitrogen ) . The results of the activity assay show that SmACC-1 is activated by cholinergic agonists but not other biogenic amines . Nicotine and ACh induced the largest response ( ≈ 6-fold and 2 . 5-fold , respectively ) when compared to water-treated control cells . An EC50 of 4 . 3 µM was calculated for nicotine , which falls within the reported range for vertebrate neuronal nAChR response to nicotine , as well as an nAChR characterized in the parasitic nematode A . suum [70]–[72] . Subsequent pharmacological studies showed that the response to nicotine was virtually abolished by D-tubocurarine , suggesting the drug effects on movement are mediated , at least in part , by this subunit . In contrast , mecamylamine had no effect on the recombinant channel and therefore it must be acting through nAChRs that do not involve SmACC-1 . Interestingly , the closely related Lymnae ACh-gated chloride channel was also reported to be insensitive to mecamylamine [11] . Functional analysis of SmACC-1 in a mammalian expression system represents a departure from the more classical electrophysiological method in Xenopus oocytes . Although electrophysiological characterization is the gold standard for measurement of ion channel activity , this method is technically demanding , labor-intensive and generally unsuitable for screening large numbers of compounds . In order to mitigate these issues , researchers have turned to mammalian cell-based ion channel functional assays . Expression of target ion channels in mammalian cells still allows direct measurement of ion flux and membrane potential , however it does so in a high-throughput format . Assays exist for a variety of ion channel types ( Ca2+ , Na+ , Cl- ) and many are commercially available [reviewed in 73] . Moreover , the data from these HTS assays generally correlate well with results generated by traditional electrophysiological methods [73] . The Premo Halide Assay employed in this study is based upon technology used to identify small molecule inhibitors of CFTR chloride channels [37] . The high-throughput format of the assay allows for the possibility of screening large chemical libraries against parasite receptors that may have highly divergent pharmacology . Given the major effects the SmACCs exert over worm motor function , this is an option worth pursuing . The work described here adds to the mounting evidence of acetylcholine's role as a major inhibitory transmitter in schistosomes . We have described a novel clade of nicotinic acetylcholine-gated chloride channel subunits ( SmACCs ) that are phylogenetically distant from the C . elegans ACCs and play a major role in inhibitory neuromuscular modulation as it pertains to larval motor behavior . The localization of the SmACCs to the peripheral nervous system points to their broad , indirect role in this modulation . Functional studies in mammalian cells indicate that the SmACC subunits are capable of forming functional nicotinic chloride channels in vitro . Finally , the use of a fluorescent , mammalian cell-based functional assay for a helminth ion channel represents a new tool in the search for new anti-schistosomal drugs . | Schistosomiasis is a widespread , chronic disease affecting over 200 million people in developing countries . Currently , there is no vaccine available and treatment depends on the use of a single drug , praziquantel . Reports of reduced praziquantel efficacy , as well as its ineffectiveness against larval schistosomula highlight the need to develop new therapeutics . Interference with schistosome motor function provides a promising therapeutic target due to its importance in a variety of essential biological processes . The cholinergic system has been shown previously to be a major modulator of parasite motility . In this study , we have described a novel clade of schistosome acetylcholine-gated chloride channels ( SmACCs ) that act as inhibitory modulators of this pathway . Our results suggest that these receptors are absent in the human host and indirectly modulate inhibitory neuromuscular responses , making them an attractive drug-target . We have also validated a new functional assay to characterize these receptors , which may be modified for future use as a high-throughput drug screening method for parasite chloride channels . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"biochemistry",
"signal",
"transduction",
"infectious",
"diseases",
"medicine",
"and",
"health",
"sciences",
"neurochemistry",
"helminth",
"infections",
"schistosomiasis",
"cell",
"biology",
"membrane",
"receptor",
"signaling",
"neurochemicals",
"mechanisms",
"of",
"signal"... | 2014 | Functional Characterization of a Novel Family of Acetylcholine-Gated Chloride Channels in Schistosoma mansoni |
Melioidosis , caused by bioterror treat agent Burkholderia pseudomallei , is an important cause of community-acquired Gram-negative sepsis in Southeast Asia and Northern Australia . New insights into the pathogenesis of melioidosis may help improve treatment and decrease mortality rates from this dreadful disease . We hypothesized that changes in Von Willebrand factor ( VWF ) function should occur in melioidosis , based on the presence of endothelial stimulation by endotoxin , pro-inflammatory cytokines and thrombin in melioidosis , and investigated whether this impacted on outcome . We recruited 52 controls and 34 culture-confirmed melioidosis patients at Sappasithiprasong Hospital in Ubon Ratchathani , Thailand . All subjects were diabetic . Platelet counts in melioidosis patients were lower compared to controls ( p = 0 . 0001 ) and correlated with mortality ( p = 0 . 02 ) . VWF antigen levels were higher in patients ( geometric mean , 478 U/dl ) compared to controls ( 166 U/dL , p<0 . 0001 ) . The high levels of VWF in melioidosis appeared to be due to increased endothelial stimulation ( VWF propeptide levels were elevated , p<0 . 0001 ) and reduced clearance ( ADAMTS13 reduction , p<0 . 0001 ) . However , VWF antigen levels did not correlate with platelet counts implying that thrombocytopenia in acute melioidosis has an alternative cause . Thrombocytopenia is a key feature of melioidosis and is correlated with mortality . Additionally , excess VWF and ADAMTS13 deficiency are features of acute melioidosis , but are not the primary drivers of thrombocytopenia in melioidosis . Further studies on the role of thrombocytopenia in B . pseudomallei infection are needed .
The soil-dwelling intracellular bacterium Burkholderia pseudomallei is an important cause of community-acquired Gram-negative sepsis in Southeast Asia and Northern Australia [1 , 2] , and the causative agent of melioidosis . Recently , it has been predicted that the annual burden of melioidosis is much higher than previously thought , with 165 . 000 human cases from which 89 . 000 patients die worldwide [3] . Over half of patients are bacteraemic at presentation [1] and despite appropriate antibiotic therapy , melioidosis has a mortality rate of 14–40% [1] . There is currently no vaccine available . The high mortality rate and the emerging antibiotic resistance of B . pseudomallei [4] highlights the need to better understand the pathogenesis of melioidosis . The interaction between innate immunity and blood coagulation contribute to the host defense against bacteria , in attempt to contain the infectious agent at the site of infection and prevent further dissemination [5 , 6] . Ample evidence has shown that severe melioidosis is characterized by strong activation of the coagulation system ( as reflected by high plasma levels of soluble tissue factor , the prothrombin fragment F1+2 and thrombin–antithrombin complexes ) , a downregulation of anticoagulant pathways ( as shown by decreased levels of protein C , protein S , and antithrombin ) and both activation and inhibition of fibrinolysis ( as reflected by elevated concentrations of tissue-type plasminogen activator ( tPA ) , plasminogen activator inhibitor type 1 and plasmin-a2-antiplasmin complexes ( PAPc ) ) [7–11] . Concurrently , a consumption of coagulation factors results in a prolonged prothrombin time and activated partial thromboplastin time [8] . Von Willebrand factor ( VWF ) , a circulating multimeric glycoprotein , is intimately involved in hemostasis and platelet activation and aggregation [5 , 12] . VWF excess is therefore associated with platelet consumption and thrombocytopenia [13–15] . VWF is constitutively expressed by endothelial cells and stored in Weibel-Palade bodies , but can also be released following stimulation by endotoxin , cytokines or thrombin and is consequently detectable in its native , ultralarge isoform ( ulVWF ) . VWF dysregulation might lead to microvascular thrombosis [16] . ADAMTS13 ( A Disintegrin and Metalloproteinase with a Thrombospondin type-1 motif member 13 ) , is a plasma protease primarily synthesized and secreted from hepatic stellate cells ( HSCs ) [17] and is known to be the main regulator of VWF activity by cleavage of the A2 domain within shear activated VWF [13 , 17] . ADAMST13 plasma activity below 10% ( <5% depending on the assay used ) goes along with thrombotic microangiopathies and bleedings known as thrombotic thrombocytopenic purpura ( TTP ) [18] . VWF , ADAMTS13 and platelets have been suggested as possible biomarkers for microangiopathic diseases such as sepsis [19 , 20] . We hypothesize that since endothelial stimulation by endotoxin , pro-inflammatory cytokines and thrombin all occur in melioidosis [8 , 21 , 22] , these would result in derangements of VWF in the host defense against septic melioidosis . First of all , we found that thrombocytopenia is a feature of melioidosis and is correlated with mortality . Additionally , study results showed that excess VWF and ADAMTS13 deficiency are features of acute melioidosis , but are not the primary drivers of thrombocytopenia in melioidosis .
The study was approved by the Ethics Committee of the Faculty of Tropical Medicine , Mahidol University ( MUTM 2008-001-01 ) and the Oxford Tropical Research Ethics Committee ( OXTREC 018–07 ) . Written informed consent was obtained from all subjects or next-of-kin by a native Thai speaker . All procedures were performed in accordance with the Helsinki Declaration of 1975 ( revised 1989 ) . Eligible patients were aged 18–75 years , had culture-proven melioidosis , had received active antimicrobial chemotherapy for less than 48 hours ( ceftazidime , co-amoxiclav , meropenem or imipenem ) , and had ≥ two out of four criteria for systemic inflammatory response syndrome ( SIRS ) [23] . This cohort has been previously described [9] . Controls were seen once and not followed-up; patients were seen daily until death or discharge and then seen at the first follow-up outpatient clinic . Plasma samples were collected at admission , seven days after and at the first outpatient clinic ≥28 days after discharge . We excluded pregnant women , and patients on anticoagulants or immunosuppressive therapy . Melioidosis patients were classified as diabetic if they had a diagnosis of diabetes prior to the onset of illness or an admission HbA1c ≥7 . 8% [24] . The study was restricted to patients with diabetes only for the following reasons: diabetes itself has effects on coagulation , the majority of patients with melioidosis have diabetes , and we were not interested in the effect of diabetes on coagulation during melioidosis , which has been investigated extensively elsewhere [9] . Melioidosis patients who do not have diabetes as a risk factor commonly have other risk factors such as corticosteroid immunosuppression , cancer , renal failure , and so forth [1] , many of which are themselves associated with endothelial stimulation and abnormalities of coagulation , making it very difficult to identify an appropriate control group . Restriction is a well-established design technique in epidemiology [25] . Blood samples were collected once only from controls , and up to three times from patients ( at recruitment , seven days later and at the first follow-up clinic >28 days from discharge ) . No samples were collected at any other time points . HbA1C was measured by high performance liquid chromatography ( Bio-Rad D-10 , Bio-Rad Laboratories , Hercules , California ) . Hemoglobin ( Hb ) , white blood cell count ( WBC ) , neutrophils , lymphocytes , thrombocytes , creatinine , alanine aminotransferase ( ALT ) , aspartate aminotransferase ( AST ) , alkaline phosphatase ( ALP ) and bilirubin were routinely available as part of the initial assessment of all participants . Blood for coagulation assays was collected in citrated tubes ( Becton-Dickinson Vacutainer 369714 ) and plasma was removed after centrifugation at 1000 ×g for 10 minutes . The plasma was stored at –70°C pending assay in The Netherlands . VWF antigen ( Dako , Glostrup , Denmark ) and VWF propeptide ( Sanquin , Amsterdam , The Netherlands ) were assayed by enzyme-linked immunoassay as described previously [15] . ADAMTS13 levels and prothrombin time ( PT ) were measured on an automated blood coagulation analyzer ( BCS XP , Siemens Healthcare Diagnostics , Marburg , Germany ) [26 , 27] . Fibrinogen levels were derived from the change in optical signal in the PT . VWF antigen , VWF propeptide and ADAMTS13 results were expressed in U/dL , where 1 unit is the activity of 1 ml of pooled normal plasma . PT was expressed in seconds and fibrinogen levels were expressed as g/L . Statistical analyses were performed and plots generated on GraphPad Prism 5 . 0b ( Graphad Software , San Diego , CA ) . Quantile-quantile plots were checked for normality and to select appropriate transformations . The distribution of age , PT , and ADAMTS13 levels were Gaussian . HbA1c levels of VWF antigen and propeptide were log-normal . An inverse square-root transform was applied to fibrinogen . Hb , WBC , neutrophils , lymphocytes , thrombocytes , creatinine , ALT , AST , ALP and bilirubin could not be transformed to Gaussian and were therefore analyzed non-parametrically . Other continuous variables were compared using the Student t-test with Welch’s modification applied when appropriate . Thrombocytopenia was defined as platelet count <150 × 109/l . Categorical data were compared by Fisher’s exact test . Strength of correlation was reported using Pearson’s coefficient . P-values were interpreted as recommended by Stern and Davey Smith [28] .
We recruited 52 controls and 34 culture-confirmed melioidosis patients at Sappasithiprasong Hospital in Ubon Ratchathani , Thailand . All patients were septic ( see inclusion criteria ) and had diabetes . Controls were , therefore , otherwise healthy diabetics attending a routine out-patient diabetes clinic . Their baseline characteristics are presented in Table 1 and their laboratory findings are depicted in S1 Table . This cohort has been previously described elsewhere [9] . In the melioidosis group , 12 patients died ( 35% ) before the first follow up ( ≥28 days after enrolment ) . An important function of VWF is to mediate platelet-platelet interactions and platelet adhesion to sub-endothelial collagen and is associated with platelet consumption and thrombocytopenia [13–15] . The median platelet count in patients with melioidosis was 189 × 109/l compared to 299 × 109/l in controls ( p = 0 . 0001 , Fig 1A ) . There were 14 melioidosis patients ( 41% ) with thrombocytopenia ( defined as a platelet count <150 × 109/l ) and no cases of thrombocytopenia among controls ( p<0 . 0001 ) . Among patients , the lowest admission platelet count observed was 13 × 109/l . Platelet counts were lower in non-survivors ( median 138 × 109/l ) compared to survivors ( 247 × 109/l , p = 0 . 02 , Fig 1B ) . Of the 12 patients who died , eight ( 67% ) had thrombocytopenia compared to six of the survivors ( 27% , p = 0 . 04 ) . Having seen thrombocytopenia in acute melioidosis , we predicted that this would be driven by high levels of circulating VWF . We observed that VWF antigen levels were higher in patients ( geometric mean , 478 U/dl ) compared to controls ( 166 U/dL , p<0 . 0001 , Fig 2A ) . However , the level of VWF antigen at recruitment was not associated with mortality ( geometric mean 445 U/dL in survivors versus 540 U/dL in non-survivors , p = 0 . 08 , Fig 2B ) . Next , we looked at whether the excess VWF antigen might be explained by increased secretion . VWF propeptide is a marker for recent secretion of VWF from the Weibel-Palade bodies ( WBD ) of endothelial cells and the dense granules of platelets [29] . We found VWF propeptide concentrations were higher in patients ( 460 U/dL ) compared to controls ( 159 U/dL , p<0 . 0001 , Fig 3A ) . Furthermore , VWF propeptide levels correlated well with VWF antigen levels ( Pearson’s r = 0 . 54 , p = 0 . 003 , Fig 3B ) , supporting our hypothesis that the excess in circulating VWF was due to excess secretion of VWF . VWF propeptide concentration and survival were not correlated , and the range of values obtained in non-survivors fell within the range obtained for survivors ( p = 0 . 21 , Fig 3C ) . ADAMTS13 is a metalloprotease secreted by the liver known as VWF cleaving protease [14 , 17] . Deficiencies of ADAMTS13 results in the accumulation of VWF in the circulation and , consequently , thrombocytopenia [13 , 14] . Previous studies have found an association between sepsis and reduced levels of ADAMTS13 [30 , 31] . The mean ADAMTS13 concentration was 31 U/dL in patients and 90 U/dL in controls ( p<0 . 0001 , Fig 4A ) . ADAMTS13 levels and VWF antigen were negatively correlated ( r = 0 . 53 , p = 0 . 002 , Fig 4B ) , which supports our hypothesis that decreased levels of ADAMTS13 contribute to high concentrations of VWF in melioidosis . However , there was only weak evidence for an inverse correlation between ADAMTS13 deficiency and mortality in melioidosis ( p = 0 . 05 , Fig 4C ) . Although the mean ADAMTS13 level in non-survivors ( 26 U/dL ) was lower than that in survivors ( 34 U/dL ) , the range of values obtained in non-survivors ( 14 to 43 U/dL ) fell entirely within the range of values obtained in survivors ( 11 to 57 U/dL ) . Thrombocytopenia was a feature of acute melioidosis and correlates with mortality . We also found high levels of VWF in melioidosis , which were both explained by increased secretion of pre-formed VWF and by reduced clearance of VWF . However , if VWF were the main driver of thrombocytopenia in melioidosis , then it is surprising that VWF antigen , VWF propeptide and ADAMTS13 levels do not correlate with mortality . We therefore re-examined the relationship between VWF antigen levels and platelet count , and found that although both were deranged in melioidosis , their levels were not correlated ( r = 0 . 28 , p = 0 . 12 , S1 Fig ) . In those patients who survived , a follow-up sample was taken seven days following enrollment and at the first follow-up clinic ( ≥28 days after discharge ) . In all patients , perturbations of platelet counts as well as abnormalities in levels of VWF antigen , VWF propeptide and ADAMTS13 all resolved completely following recovery from melioidosis ( S2A–S2D Fig ) .
In conclusion , thrombocytopenia is a feature of sepsis caused by B . pseudomallei and is correlated with mortality . Excess VWF is a feature of acute melioidosis and is likely driven by both by increased secretion of VWF propeptide in endothelium and by reduced clearance by ADAMTS13 . However , the thrombocytopenia of melioidosis is likely not driven by excess VWF: other possible drivers include diffuse intravascular coagulation ( DIC ) and hemophagocytosis . In the past years , tremendous progress has been made toward our understandings of the protective roll of platelets in sepsis and the possibility in the use of thrombocytopenia as biomarker [35 , 37] . More animal and human studies are necessary to understand the reason of thrombocytopenia in septic melioidosis patients and to translate this to clinical practice . | Melioidosis , caused by bioterror threat agent Burkholderia pseudomallei , is an important cause of community-acquired sepsis in Southeast Asia and Northern Australia . Recently , it has been predicted that the annual burden of melioidosis is much higher than previously thought , with 165 . 000 human cases from which 89 . 000 patients die worldwide . Melioidosis has a mortality up to 40% despite appropriate antibiotic treatment and there is currently no vaccine available . Therefore , it is of importance to better understand the pathogenesis of this debilitating disease . There is extensive cross talk between the innate immune system and blood coagulation , which contributes to the host defense against invading bacteria . One of the hallmark features of melioidosis is extensive abnormalities in the coagulation system . Therefore , we hypothesized that , since endothelial stimulation by endotoxin , pro-inflammatory cytokines and thrombin all occur in melioidosis , these would result in derangements of Von Willebrand factor ( a protein involved in hemostasis and platelet aggregation ) . In a cohort of culture-confirmed patients with severe melioidosis , we found that thrombocytopenia is a key feature of melioidosis and is correlated with mortality . Additionally , our study showed that excess VWF and ADAMTS13 deficiency are features of acute melioidosis , but are not the primary drivers of thrombocytopenia in melioidosis . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"blood",
"cells",
"death",
"rates",
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"pathology",
"and",
"laboratory",
"medicine",
"demography",
"melioidosis",
"diabetes",
"mellitus",
"bacterial",
"diseases",
"physiological",
"processes",
"sepsis",
"endocrine"... | 2017 | Increased Von Willebrand factor, decreased ADAMTS13 and thrombocytopenia in melioidosis |
Paracoccidioidomycosis ( PCM ) is a neglected disease present in Latin America with difficulty in treatment and occurrence of serious sequelae . Thus , the development of alternative therapies is imperative . In the current work , two oxadiazole compounds ( LMM5 and LMM11 ) presented fungicidal activity against Paracoccidioides spp . The minimum inhibitory and fungicidal concentration values ranged from 1 to 32 μg/mL , and a synergic effect was observed for both compounds when combined with Amphotericin B . LMM5 and LMM11 were able to reduce CFU counts ( ≥2 log10 ) on the 5th and 7th days of time-kill curve , respectively . The fungicide effect was confirmed by fluorescence microscopy ( FUN-1/FUN-2 ) . The hippocratic screening and biochemical analysis were performed in Balb/c male mice that received a high dose of each compound , and the compounds showed no in vivo toxicity . The treatment of experimental PCM with the new oxadiazoles led to significant reduction in CFU ( ≥1 log10 ) . Histopathological analysis of the groups treated exhibited control of inflammation , as well as preserved lung areas . These findings suggest that LMM5 and LMM11 are promising hits structures , opening the door for implementing new PCM therapies .
Paracoccidioidomycosis ( PCM ) is an endemic fungal disease in Latin American countries , which presents high prevalence in South America . The lung is the most affected organ , mainly during chronic form , presenting pulmonary architectural distortion , which can lead to hypoxemia and hypercapnia in 90% of patients with PCM [1] . In the last 30 years , the presence of pulmonary damage ranged from 63 . 8 to 100% in the patients [2 , 3 , 4 , 5] . Furthermore , this injury remains even after the treatment and promotes pulmonary fibrosis with loss of respiratory function in 50% of patients [6 , 7] . Considering this worrying scenario , the current available antifungal drugs are limited . In Brazil , only three therapeutics options are available for PCM treatment , such as polyenes , sulfanilamide and triazoles . The azoles action on the sterol biosynthetic pathway leads to many side-effects . Amphotericin B ( AmB ) , a polyene , is the antifungal of choice in severe and acute cases . The treatment time should be as short as possible , between two and four weeks , due to its high toxicity [8] . The sulfanilamide is treatment options according to the severity of the disease; however , several disadvantages have been reported such as hypersensitivity reactions , gastrointestinal symptoms , hemolytic anemia , agranulocytopenia and crystalluria [9] . On the other hand , the most commonly used antifungal agent for treating mild and moderate forms of PCM is itraconazole ( ITZ ) , but the time of therapy may reach 18 months and presents some collateral effects [10] . The major therapeutic challenges of this disease are the long period of continuous use of systemic antifungals , the possibility of relapses and the appearance of sequelae in the lung [1] . This , associated with the limited antifungal arsenal , evidences the necessity of the emergence of a new antifungal class . Thus , the development of a drug that selectively acts on the target pathogenic fungi without producing collateral damage to mammalian cells is a pharmacological challenge . Biotechnological methods have become an important approach in pharmaceutical drug research and development . For example , the in silico methodologies not only reduce the cost associated with drug discovery , but they may also reduce the time it takes for a drug to reach the market [11] . This is a modern strategy to explore the interaction of compounds with a specific target [12] . By comparative genomics , ten potential targets for drugs occurring in eight human pathogenic fungi—Candida albicans , Cryptococcus neoformans , Aspergillus fumigatus , Blastomyces dermatitidis , Coccidioides immitis , Histoplasma capsulatum , Paracoccidioides brasiliensis and Paracoccidioides lutzii—were described [13] . One of these targets is thioredoxin reductase ( Trr1 ) , a flavoenzyme that acts primarily on resistance to oxidative stress , and it is essential to cell growth [14] . The trr1 mutation may result in hypersensitivity to hydrogen peroxide and to high temperatures [15] . In addition , this Trr1 isoform is found only prokaryotes and fungi [14] . Therefore , Trr1 a good target for the development of new anti-PCM therapies [16] . By molecular modeling and virtual screening , several compounds were selected as Trr1 ligands . Preliminary results showed that two compounds , which belong to the oxadiazole class , present antifungal activity against important pathogenic fungi such as Candida spp . , Cryptococcus neoformans and Paracoccidioides spp . [17] . For this purpose , the antifungal activity of two oxadiazole compounds selected by in silico methods was tested both in vitro and in vivo against Paracoccidioides spp .
All the procedures were performed according to the regulations of the Ethical Committee for Animal Experimentation , State University of Maringá , Brazil ( approval no . CEUA 9810191015 , 22/04/2016 ) . The animal’s experimentation were conducted according to the Guideline for the Care and Use of Laboratory Animals ( CONCEA ) . The compounds selected by virtual screening against thioredoxin reductase were commercially purchased from Life Chemicals Inc . ( Burlington , ON , Canada ) . These compounds were named by LMM5 is 4-[benzyl ( methyl ) sulfamoyl]-N-[5-[ ( 4-methoxyphenyl ) methyl]-1 , 3 , 4-oxadiazol-2-yl]benzamide , and the chemical name of LMM11 is 4-[cyclohexyl ( ethyl ) sulfamoyl]-N-[5- ( furan-2-yl ) -1 , 3 , 4-oxadiazol-2-yl]benzamide ( Fig 1 ) . The stock solutions were prepared in dimethyl sulfoxide ( DMSO ) at concentration 100 μg/mL for LMM11 and 50 μg/mL for LMM5 . Nine isolates of Paracoccidioides spp . were used , three P . brasiliensis ( Mg0113 , Mg0213 and Pb18 ) , three P . lutzii ( Pb01 , 8334 and Mg0114 ) and three isolates not identified yet at species level ( Mg0116 , Mg0216 , Mg0115 ) . Candida parapsilosis ( ATCC 22019 ) and Candida krusei ( ATCC 6258 ) were included for quality control . The isolates are part of the collection from Laboratory of Medical Mycology of the State University of Maringá , Brazil . The yeast phase was maintained by weekly passaging at 37°C in Fava Netto's solid medium . For each experiment , the viability of Paracoccidioides spp . was determined by counting viable cells in a Neubauer chamber by the trypan blue method . The assays were performed with ≥80% of viable cells [18] . Balb/c male mice , approximately six weeks old , with an average weight of 20 g , were raised at animal facilities of the State University of Maringá , Brazil . The animals were divided in groups and maintained in ventilated cages , with free access to tap water and food , in a controlled animal facility having a constant temperature of 23°C and a 12 h light/dark cycle . The minimum inhibitory concentration ( MIC ) was determined by the broth microdilution method , following the standard methodology by the Clinical Laboratory Standards Institute ( CLSI ) published in document M-27A3 , with modification for Paracoccidioides spp . [19 , 20] . The oxadiazoles compounds’ concentrations ranged from 1 to 512 μg/mL . The inoculum was adjusted to 2 × 104 yeast cells/mL and diluted 1:2 into a 96-well plate with RPMI-1640 medium . Negative controls were only medium without inoculum , and positive controls were medium plus inoculum . The incubation time was 5 days at 37°C . Interpretation of the growth cutoff point was performed visually based on the comparison of growth in the positive control wells . The MIC values was defined as the lowest oxadiazoles concentration that resulted in at least an 80% reduction in growth relative to the positive control [21] . For AmB , it was considered to be the concentration causing 100% inhibition compared to the control without the antifungal drug . The drug controls were performed with AmB against C . parapsilosis ( ATCC 22019 ) and C . krusei ( ATCC 6258 ) , according to the CLSI ( document M27-A3 ) [19] . The minimum fungicidal concentration ( MFC ) of each compound was determined by transferring aliquots of 5 μL of each well from MIC microplates to brain-heart infusion ( BHI ) agar plates and incubating at 37°C for 7 days . The fungicidal activity was considered the lowest drug concentrations at which no colonies were able to grow . The following assays were performed with isolate Pb18 . The P . brasiliensis isolate Pb18 was cultivated in McVeigh Morton Chemically Defined Culture Medium ( MMcM ) for 7–10 days at 37°C under agitation at 150 rpm , to obtain yeast cells with typical multiple budding [22] . This culture was adjusted to 2 . 5 × 104 CFU/mL and treated with different concentrations of LMM5 and LMM11 ( 8 and 16 μg/mL ) for 1 , 3 , 5 , 7 and 14 days at 37°C . The untreated yeasts were used as controls . At each time interval , yeasts of each group were diluted in phosphate-buffered saline ( PBS ) , and 100 μL was plated on Brain Heart Infusion ( BHI ) agar medium supplemented with 5% of Pb18 culture filtrate and 4% of Fetal Bovine Serum and incubated at 37°C for at least 14 days . The CFU were counted . The effect was considered fungicidal only when the CFU reduction was 3 log10 ( ≥99 . 9% ) ; otherwise , it was considered fungistatic [23] . The metabolic activity of yeast cells of Pb18 was analyzed after exposure to the MIC concentrations of LMM11 and LMM5 ( both 8 and 16 μg/mL , each ) . The assay was performed using FUN-1 and FUN-2 stains according to the manufacturer's protocol ( Molecular Probes ) . Yeasts were suspended in MOPS buffer containing 2% glucose . The fungal cell activity was estimated with 0 . 5 μM FUN-1 ( 100 mM stock solution , dissolved in DMSO ) and expressed as a change in the ratio of red fluorescence ( k = 575 nm ) to green ( k = 535 nm ) . The viability of fungal cells was determined from examination of at least 200 cells in a biological replicate by fluorescence microscopy . A dead control was done using 70% ethanol to kill Pb18 . Metabolically active cells fluoresce as red in their structures , while dead cells or cells with little or no metabolic activity exhibit bright diffuse green cytoplasmic fluorescence and lack of intravacuolar fluorescent inclusions [24] . AmB was chosen to test in combinations with LMM5 and LMM11 against Pb18 isolate . The compounds ( starting at 4× MIC ) were distributed vertically while AmB ( 4× MIC ) was added horizontally as described by Bagatin et al . [25] . A 2 × 104/mL yeast cell suspension was added to 96-well plates and incubated at 35°C for 7 days . Inhibition was read visually and confirmed by XTT viability ( 492nm ) . The fractional inhibitory concentration ( FIC ) was determined by calculating ΣFIC = FICA + FICB = ( CombAmB/MICAmB ) + ( CombLMM/MICLMM ) . For a strongly synergistic effect , FIC < 0 . 5; a synergistic effect , FIC < 1; an additive effect , FIC = 1; no effect , 1 < FIC < 2; and an antagonistic effect , FIC > 2 [26] . The Bliss-independent interactions were analyzed by Combenefit software [27] . Male Balb/c mice at 6 weeks old were divided into four groups: Control group treated with vehicle ( PBS , DMSO 1% , and Pluronic F-127 0 . 2% ) ; LMM5 group treated intraperitoneally with LMM5 at 25 mg/kg; and LMM11 group treated intraperitoneally with LMM11 at 50 mg/kg . The animals were monitored by Hippocratic screening at times 0 , 15 , 30 , 60 , 120 , 240 and 480 minutes . After the 14th day , the mice were anesthetized for blood collection and euthanized . The biochemical examinations were performed , and the liver , heart and kidneys were weighed , on the day of euthanasia . The assay was performed in accordance with Salci et al . [28] . After inoculation with 106 Pb18 yeast ( intratracheally ) , animals were randomly divided into experimental groups: LMM5 , LMM11 , ITZ ( group treated with itraconazole ) and control . The treatment started after 24 hours of infection . The compounds and ITZ were administered at 5 mg/kg , once per day for 14 days , intraperitoneally . The animals were euthanized by isoflurane vaporizer , and the number of CFU/g of the lung tissue was determined [20] . Mice were euthanized 15 days post-infection , and lungs were collected . The organs were fixed in 10% formalin and embedded in paraffin . Five-micrometer sections were stained with Grocott's methenamine silver ( GMS ) and counterstained with hematoxylin–eosin ( H&E ) . From the histological sections of the lung , the area was determined , and CFU/mm2 were counted . The calculation consists of the total number of fungal cells divided by the lung area [29] . The lung sections were analyzed about the cellular changes , and presence of fungi and inflammatory cells , using a Motic model BA310LED microscope , Moticam 5 . 0 MP digital camera ( 100 , 400 and 600x magnification ) and Motic software . Thus , 20 fields of at least two histological sections were classified according to the presence of inflammatory infiltrates categorized as severe ( 3+ or more ) , moderate ( 2+ ) , mild ( 1+ ) and non-inflammatory ( 0+ ) [30] . Statistical analysis of the different experimental groups was performed by GraphPad Prism software ( GraphPad Software , San Diego , CA , USA ) . Reduction of fungal burden from in vivo treatment was reported as log10 of mean ± standard deviation using unpaired Student’s T-test . The significance of differences in histopathological score was determined by Student’s T-test . The level of significance was set as p< 0 . 05 .
LMM5 was able to inhibit the growth of all isolates of Paracoccidioides spp . . 77 . 8% of isolates presented MIC values ranging between 8 and 32 μg/mL ( Table 1 ) . The Mg0114 isolate was the most sensitive ( MIC = 1 μg/mL ) . All isolates presented the minimum fungicidal concentration ( MFC ) values similar to the MIC values . Otherwise , the MIC values of LMM11 was 8 μg/mL for most of the isolates ( 88 . 9% ) . The MFC were 8 and 16 μg/mL , corresponding to 66 . 7 and 33 . 3% of the isolates , respectively ( Table 1 ) . The susceptibilities of isolates to AmB are shown by MIC values of 2 μg/mL ( 66 . 7% of the isolates ) and 1 μg/mL ( 33 . 3% ) . The change in growth over time was evaluated by time-kill curves during 14 days ( Fig 2 ) . The yeast treated with LMM5 or LMM11 exhibited 80% reduction in the Pb18 cell viability from the 5th day post-treatment . The fungicidal profile was determined by CFU reductions of ≥3 log10 as compared with control growth . The LMM5 fungicidal profile can be observed from the 7th day ( Fig 2A ) . For LMM11 , the fungicidal effect was detected on the 7th day ( 16 μg/mL ) as shown in Fig 2B . In addition , the largest difference between groups was observed on the 14th day , for both compounds . The time-kill curve results were corroborated by LIVE/DEAD assay , in which the cellular viability was evaluated by fluorescence microscopy . For this evaluation , Pb18 cells were treated with LMM5 ( Fig 2C ) or LMM11 ( Fig 2D ) . Both compounds were able to produce a diffuse bright green fluorescence profile indicating cell death or yeast with little metabolic activity . This fluorescence profile is quite different from what was observed in the control group ( not treated ) , in which live cells presented yellow-orange intravacuolar structures . AmB , when combined with LMM5 or LMM11 , showed better antifungal activity than alone , reducing the MIC value from 2 to 0 . 5 μg/mL . The new compounds’ interactions with AmB reduced the three-fold MIC value of LMM5 ( from 32 to 4 μg/mL ) and the two-fold MIC value of LMM11 ( from 16 to 4 μg/mL ) . These results indicate a synergistic effect of AmB with LMM5 or LMM11 ( Table 2 ) . The synergic effect revealed by FIC values was validated by the result of the Bliss independence surface analysis . In this way , the AmB combination with each of the oxadiazoles showed predominance of blue areas , indicating a positive ΔE and thus confirming the synergic capacity ( Fig 3 ) . In vivo toxicity parameters showed mild behavioral changes , such as abdominal contortion and motor impairment , within 30 minutes after intraperitoneal administration of the compounds in all groups evaluated . After this period , no alterations were observed . For both compounds , it is important to note that there were no differences in the body weight of animals , in the hematological profile and in the macroscopic analysis of the organs after 14 days . Regarding the biochemical parameters , although the serum levels of amino transferase aspartate ( AST ) from mice receiving LMM5 were significantly higher than that from the control ( p <0 . 05 ) , their values were within those expected for normal mice ( Table 3 ) . Similarly , the LMM11 group showed no statistical differences in the AST values compared to that from the control group ( Fig 4A ) . Both amino transferase alanine ( ALT ) and creatinine levels did not present a statistical difference between the groups evaluated ( Fig 4B and 4C ) . According to Fig 4D , the liver did not exhibit changes in its weight , while in the kidney , although there was a slight reduction in the kidney weight of animals receiving LMM5 and LMM11 , no statistical difference was found ( p >0 . 05 ) ( Fig 4E ) . There were no significant changes in heart weight ( Fig 4F ) . Since LMM5 and LMM11 presented promising in vitro antifungal activity and no toxicity in vivo , the next step was to evaluate them through an in vivo experimental PCM treatment . Daily therapy with the new compounds for 14 days showed a significant reduction of pulmonary fungal burden in relation to the control ( p <0 . 05 ) for LMM5 ( 1 . 2 Log10 CFU/g ) and LMM11 ( 1 . 0 Log10 CFU/g ) , as well as for the group treated with ITZ ( 1 . 5 Log10 CFU/g ) as shown in Fig 5 . There was no statistical difference among the groups treated ( LMM5 , LMM11 and ITZ ) ; all were equally efficient in reducing fungal burden in mice ( p>0 . 05 ) . Because pulmonary fibrosis is the main sequelae of PCM , even after treatment it is essential to evaluate the therapy effect of the new compounds on the inflammatory response triggered by P . brasiliensis . A quantitative analysis of the histological sections allows determination of the number of fungal cells present in each histological lung section . Fig 6A demonstrates that conventional antifungal treatment with ITZ was as effective as the new compounds in reducing the number of yeast cells/mm2 when compared to a control ( p<0 . 05 ) , but no statistical difference between compounds and ITZ was found ( p>0 . 05 ) . These results corroborate the significant reduction of fungal burden presented previously ( Fig 5 ) . The inflammation level , indicated by the presence of inflammatory infiltrates in lung tissue , was significantly reduced in the three treatments tested , in relation to the control ( p<0 . 05 ) ( Fig 6B ) . A qualitative analysis of the histological sections of groups treated was performed . The lungs of the infected mice that received only vehicle ( DMSO 1% and Pluronic 0 . 2% ) showed a predominance of necrotic areas , indicated by black arrows ( Fig 7A ) . In contrast , animals treated with ITZ , LMM5 or LMM11 revealed large areas of preserved lung tissue , indicated by the blue arrows in Fig 7D , 7G and 7J , respectively . In the necrotic areas , it was possible to observe a total loss of pulmonary architecture , leading to no alveolar wall visualization ( black arrows ) . Severe lesions , characterized by the presence of diffuse inflammatory exudate , were also observed . An intense recruitment of mononuclear cells was detected in the control group , as indicated by red arrows ( Fig 7B ) . In the treated groups , the presence of inflammatory infiltrates was lower ( Fig 7B , 7C and 7I ) . The histopathological evidence showed rounded and multi-budding fungal cells presenting viable protoplasm ( white arrows ) and nonviable protoplasm ( green arrows ) in all groups analyzed ( Fig 7C , 7I and 7L ) . Therefore , these results demonstrated that treatment with both compounds can be associated with infection control and maintenance of pulmonary architecture .
The key to a good prognosis is immediate treatment [32] . However , the limited number of antifungal drug classes , the need for long-term treatment , and the high toxicity and adverse effects of the drugs reduce adherence to the treatment by patients [33] . All these concerns indicate a major gap that must be filled in antifungal therapy , especially for severe mycoses treatment . Undoubtedly , the pace of discovery of new antifungals is far from reaching the current needs . However , some groups have sought new therapeutic options through in silico approaches . Virtual screening based on the ligand or structure target is being performed in the identification of new compounds for PCM treatment [20 , 29 , 34] . The thioredoxin reductase ( Trr1 ) has been shown as an important target for the development of molecules with antifungal activity [16 , 35] . This protein is a flavoenzyme that catalyzes the reduction of NADPH-dependent thioredoxin , protecting cells against oxidative stress [36] . Thus , Trr1 comprises the three main parameters to be a potent candidate for antifungal development: it is an essential gene for fungus survival , it is absent in humans , and it is conserved in several pathogenic fungi . Therefore , it will possibly allow a broad spectrum of action [16] . LMM5 and LMM11 were selected by virtual screening as potential inhibitors of Trr1 . Initial tests showed in vitro activity against various invasive fungal infections ( IFIs ) and absence of toxicity [17] . This work reports the antifungal effects of these two oxadiazoles against Paracoccidioides spp . , presenting an inhibition profile with MIC values between 1 and 32 μg/mL . Several studies have used the in silico approach to identify compounds for PCM treatment [16 , 20 , 25 , 30 , 35 , 37] . The MIC values for these compounds was always similar to oxadiazoles . Three inhibitors of thioredoxin reductase showed MIC values ranging from 8 to 32 μg/mL [16] . An inhibitor of chorismate synthase presented antifungal activity with MIC values between 2 and 32 μg/mL [20] . Furthermore , two homoserine dehydrogenase inhibitors demonstrated antifungal activity ( MIC values 32–64 μg/mL ) [25] . This group also synthetized 4-methoxy-naphthalene derivatives with antifungal activity against Paracoccidioides spp . ( MIC values 8–32 μg/mL ) [38] . A thiosemicarbazone derivative tested against 14 isolates of Paracoccidioides spp . showed MIC values between 3 . 90 and 62 . 50 μg/mL [39] . In addition , new chalcone derivatives presented antifungal activity with MIC values between 2 . 9 and 42 . 2 μM [34] . Drugs with fungicidal profiles are more promising than fungistatic [40] . Thus , our findings indicate that the two oxadiazoles compounds analyzed in this study are promising , because both presented fungicidal profile , especially after 7th day . Fluorescence microscopy results confirmed this fungicidal profile , resulting in cell death and not only growth inhibition . A thiosemicarbazone derivative of lapachol was tested against isolate Pb18 by de Sa et al . [39] , and it showed a reduction of 90% in fungal growth after the 5th day; no synergistic effect with conventional drugs was detected . The antifungal effect of this commercially available drug combined with the novel compounds was evaluated . The synergistic interaction between candidate compounds and conventional antifungal agents may reduce the need for high doses , minimizing adverse effects and providing beneficial attributes for new therapeutic strategies against PCM [41] . In this way , LMM5 exhibited a strongly synergistic effect with the most potent antifungal for PCM , and LMM11 also interacted synergistically . It suggests a possible interaction pathway with AmB , increasing its fungicidal effect [42] . It is possible to suppose that the pores opened by AmB could facilitate the access of the new compounds to the intracellular target , the thioredoxin system , leading to the synergistic interaction observed . Although chalcone derivatives have low MIC values for several isolates of Paracoccidioides spp . , no synergistic effect was observed with AmB or another antifungal that was tested [40] . Whereas AmB is the choice drug in the most severe PCM cases , nephrotoxicity affects more than 80% of patients [43] . Recent findings reveal that hepatobiliary changes during treatment with ITZ in patients with PCM are irreversible even if they are not as frequent compared to AmB [44] . The biochemical parameters of in vivo toxicity assays for the new oxadiazoles were analyzed based on the references values for male Balb/c mice suggested by Araujo and collaborators [31] . Therefore , the AST , ALT and creatinine values of mice treated with high doses of oxadiazoles are within normality patterns . These findings reveal that these compounds do not present nephrotoxicity or hepatotoxicity in the murine model . An important validation of anti-Paracoccidioides activity is to extrapolate to in vivo analysis . The experimental PCM model showed the ability of the oxadiazoles to reduce the fungal lung burden of infected mice . It is suggested that the intraperitoneal treatment with LMM5 and LMM11 reached the lungs and controlled the fungal burden as well as for itraconazole . Comparable results were found for chalcone derivatives in treatment of the PCM experimental . In which the fungal reduction was similar to itraconazole treatment [45] . Cyclopalladated treatment also demonstrated fungal burden reduction and decrease of the damages caused by infection [46] . The major challenge for patients undergoing PCM treatment is sequelae triggered by aggressive pulmonary inflammatory response , which may lead to loss of function [1] . This work evaluated how much of the lung was preserved with the different treatments . Representative images of the lung histopathology ( Fig 7 ) demonstrated untreated animals with large areas of necrosis , filled with Pb18 yeast cells throughout the tissue . Thus , the ability of LMM5 and LMM11 to reduce fungal burden and inflammatory response in the lungs of mice infected with P . brasiliensis seems to be very promising for controlling pulmonary sequelae in PCM . In conclusion , we have successfully demonstrated that two new oxadiazoles selected by virtual screening presented promising antifungal activity against Paracoccidioides spp . , opening perspectives for implementing alternative PCM therapy strategies . Both in vitro and in vivo results indicated that LMM5 and LMM11 could be used as lead structures to new antifungal compounds , with fungicidal profiles and leading to reduced tissue damage caused by fungal infection . | Paracoccidioidomycosis ( PCM ) is a granulomatous fungal infection with clinically severe forms and serious pulmonary sequelae . The current limited arsenal and prolonged treatment regimen demonstrate the need for new antifungals . This study reveals two fungicidal oxadiazole compounds for PCM treatment . The in vitro assay showed the antifungal activity and the synergic effect with Amphotericin B . The high doses administration in mice showed absence of toxicity , which allowed to demonstrate the in vivo antifungal activity . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"antimicrobials",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"drugs",
"fungicides",
"tropical",
"diseases",
"microbiology",
"light",
"microscopy",
"immunology",
"antifungals",
"fungal",
"structure",
"toxicology",
"toxicity",
"fu... | 2019 | Antifungal activity of two oxadiazole compounds for the paracoccidioidomycosis treatment |
Candida albicans can stochastically switch between two phenotypes , white and opaque . Opaque cells are the sexually competent form of C . albicans and therefore undergo efficient polarized growth and mating in the presence of pheromone . In contrast , white cells cannot mate , but are induced – under a specialized set of conditions – to form biofilms in response to pheromone . In this work , we compare the genetic regulation of such “pheromone-stimulated” biofilms with that of “conventional” C . albicans biofilms . In particular , we examined a network of six transcriptional regulators ( Bcr1 , Brg1 , Efg1 , Tec1 , Ndt80 , and Rob1 ) that mediate conventional biofilm formation for their potential roles in pheromone-stimulated biofilm formation . We show that four of the six transcription factors ( Bcr1 , Brg1 , Rob1 , and Tec1 ) promote formation of both conventional and pheromone-stimulated biofilms , indicating they play general roles in cell cohesion and biofilm development . In addition , we identify the master transcriptional regulator of pheromone-stimulated biofilms as C . albicans Cph1 , ortholog of Saccharomyces cerevisiae Ste12 . Cph1 regulates mating in C . albicans opaque cells , and here we show that Cph1 is also essential for pheromone-stimulated biofilm formation in white cells . In contrast , Cph1 is dispensable for the formation of conventional biofilms . The regulation of pheromone- stimulated biofilm formation was further investigated by transcriptional profiling and genetic analyses . These studies identified 196 genes that are induced by pheromone signaling during biofilm formation . One of these genes , HGC1 , is shown to be required for both conventional and pheromone-stimulated biofilm formation . Taken together , these observations compare and contrast the regulation of conventional and pheromone-stimulated biofilm formation in C . albicans , and demonstrate that Cph1 is required for the latter , but not the former .
Candida albicans is a prevalent pathogen of humans that colonizes and infects multiple niches in the mammalian host . To achieve such extreme adaptability , this pathogen has evolved genetic and epigenetic mechanisms that modulate cell behavior and morphology in response to environmental signals . Epigenetic variation in C . albicans is perhaps best exemplified by the white-opaque phenotypic switch . This is a heritable and reversible switch in which cells transition between white cells that are round and give rise to dome-shaped , shiny colonies , and opaque cells that are elongated and give rise to flatter , darker colonies [1] . Switching is regulated by a core circuit of transcription factors that operate within a network of positive and negative feedback loops [2] , [3] . Similar transcriptional networks are found in many biological systems and act to regulate developmental programs from yeast to mammals [3] , [4] . White and opaque cells exhibit striking behavioral differences , including their contrasting ability to undergo sexual reproduction . Opaque cells are the mating competent form of C . albicans and secrete sex-specific pheromones that induce mating responses in cells of the opposite mating type [5] . Pheromone signaling in opaque cells leads to the upregulation of genes required for cell and nuclear fusion , as well as the formation of polarized mating projections [6]–[8] . In contrast , white cells are refractory to mating , undergoing a-α cell fusion at least a million times less efficiently than opaque cells [5] . However , white a or α cells become adherent in response to pheromones secreted by opaque cells , leading to enhanced biofilm formation [9] . It is speculated that such pheromone-stimulated biofilms could increase mating between opaque cells by stabilizing pheromone gradients and promoting chemotropism between rare mating partners [9] . Biofilms also represent a significant threat for the development of clinical infections by C . albicans . These surface-associated communities can form on implanted medical devices and host surfaces , and are resistant to antifungal treatment , while also promoting the seeding of serious bloodstream infections [10] , [11] . “Conventional” biofilms are formed when C . albicans yeast cells adhere to a surface followed by maturation due to pseudohyphae and hyphae formation and production of extracellular matrix material [10] , [12] . Hyphae formation is an important feature of biofilms as mutants blocked in filamentation are often impaired in biofilm development [12] . The core transcriptional network regulating conventional biofilms has recently been elucidated and , similar to the white-opaque switch , involves interacting transcriptional feedback loops [13] . Six transcription regulators were shown to operate the biofilm regulatory network including Bcr1 , Brg1 , Efg1 , Rob1 , Ndt80 , and Tec1 [13] . Loss of any one of these regulators significantly compromised biofilm formation in vitro , and these factors were also necessary in two in vivo animal models of biofilm formation [13] . This work was carried out in a/α cells , and these biofilms were formed by exposing C . albicans to a solid surface ( bovine serum-coated polystyrene or silicone substrates ) and allowing the biofilm to form over the course of 24 to 48 hours , with gentle shaking of the samples at physiological temperature ( 37°C ) . Recent studies have begun to address the regulation of biofilm formation in pheromone-stimulated ( or sexual ) biofilms , and to compare mechanisms of pheromone signaling in white and opaque cells . In this case , the biofilms are formed from white a or α cells without shaking or on a slow rocker at 25–29°C . Under these conditions , significantly thicker biofilms are formed by a cells in the presence of α pheromone , or by α cells in the presence of a pheromone [9] , [14] , [15] . Pheromone responses in diverse yeast species are mediated by a conserved G-protein coupled MAPK cascade that culminates in transcription factor activation [16] . Studies have established that the C . albicans transcription factor Cph1 ( ortholog of S . cerevisiae Ste12 ) is activated by MAPK signaling and mediates expression of mating genes in opaque cells [17]–[19] . However , in contrast to this paradigm ( which also holds true for S . cerevisiae , Kluyveromyces lactis , and Candida lusitaniae [20]–[22] ) , pheromone signaling in C . albicans white cells was proposed to activate a different transcription factor , Tec1 , with Cph1 dispensable for signaling in this cell type [14] . In this manuscript , we compare and contrast the genetic requirements for conventional and pheromone-stimulated biofilms , and re-address the role of Cph1 in these processes . We show that four of the six transcriptional regulators of conventional biofilm formation ( Bcr1 , Brg1 , Rob1 , and Tec1 ) are also necessary for pheromone-stimulated biofilms . However , in contrast to previous reports , we demonstrate that Cph1 is the master transcription factor mediating MAPK signaling in white and opaque cells of C . albicans . Thus , Cph1 is essential for pheromone-stimulated biofilm formation in white cells as well as sexual mating in opaque cells . Transcriptional profiling of pheromone-stimulated biofilms was also performed and provides the first genome-wide picture of this developmental program . Gene expression profiles of wildtype , Δcph1/Δcph1 , and Δtec1/Δtec1 strains were compared , and confirmed that Cph1 is essential for the transcriptional response to pheromone . Downstream targets of Cph1 were identified including Hgc1 , which is shown to play a significant role in both pheromone-stimulated and conventional biofilms . Overall , our data reveals that several components of biofilm regulation are shared between conventional and pheromone-stimulated biofilm models , but that other transcription factors operate specifically in only one program of biofilm development .
In order to compare the genetic requirements for different types of biofilms formed under different conditions , we performed an experiment where we directly compared two distinct biofilm models using isogenic strains . These experiments were carried out in two different laboratories and used multiple independent mutants to confirm the findings . Figure 1A and 1B shows the results of a series of isogenic white a strains tested under the set of biofilm conditions described in Nobile et al . [13] . These biofilms were formed at 37°C in Spider medium with shaking . The results show , by cell number , dry weight and confocal scanning laser microscopy ( CSLM ) , that deletion of any one of the six core transcription regulators ( Bcr1 , Brg1 , Efg1 , Tec1 , Ndt80 , or Rob1 ) , severely reduced biofilm formation ( Figure 1A and B , and data not shown ) . These results are consistent with those of Figure 1 in Nobile et al . , the only difference being that the experiments described here were performed with white MTLa cells while Nobile et al . used white MTLa/α cells . Addition of pheromone under these conditions did not produce any apparent differences either in the dry weights or in the appearance of the biofilm by CSLM ( Figure 1A and B ) . These experiments were conducted in the C . albicans SC5314 strain background and we will refer to this protocol as the “conventional” biofilm assay . Figure 1C–F shows the same set of strains subjected to a different type of biofilm assay , first described by Daniels et al . [9] . Here , biofilms were formed at room temperature in Lee's medium without shaking . Under these conditions , wildtype white a cells formed a very weak biofilm in the absence of pheromone , and α pheromone treatment significantly increased biofilm formation ( Figure 1C and D ) . We will refer to this protocol as the “pheromone-stimulated” biofilm assay . We note that biofilms produced by SC5314-derived strains under these conditions are more fragile than those produced under the “conventional” biofilm assay and that they adhere to the plastic surface less tightly than do conventional biofilms . These results show that , rather than being dependent on all six of the transcription factors regulating conventional biofilms , the pheromone-stimulated biofilms show dependencies on only four regulators ( Bcr1 , Tec1 , Rob1 , and Brg1; Figure 1C ) . One interpretation of this result is that because the pheromone-stimulated biofilms are less adherent they require only a subset of the conventional biofilm circuit . We note that deletion of NDT80 shows opposite effects in conventional and pheromone-stimulated biofilms ( deletion of NDT80 compromises the former and enhances the latter , Figure 1D ) , and we return to this point later in the paper . Experiments in which a wild-type allele of the deleted gene was reintroduced into the homozygous deletion mutant significantly complemented all of the mutant phenotypes ( Figure 1C and 1D ) . Complete complementation was not expected as these addback strains contained one functional gene copy compared to two gene copies in the wildtype strain . Confocal images of pheromone-stimulated biofilms revealed them to be relatively patchy compared to conventional biofilms , although there was still a relevant correlation between CSLM images ( Figure 1E and F ) and measurements of adherent cells ( Figure 1C and D ) . Having established the basic requirements for general biofilm production , we now turn to the components specific for pheromone-stimulated biofilm formation . Previous studies have proposed that Tec1 is the master transcriptional regulator of pheromone-induced biofilms , while Cph1 is dispensable for their formation [14] . This result was surprising given that Cph1 or its orthologs are essential for pheromone signaling in multiple yeast species . We therefore directly compared the role of Cph1 and Tec1 in pheromone signaling in C . albicans white cells from multiple strain backgrounds . For these experiments , α pheromone was first used to stimulate biofilm formation in white cells of strain P37005 that is a natural MTLa/a isolate , and like SC5314 belongs to clade I , a major clade of C . albicans strains [23] . In contrast with published reports , we found that pheromone-stimulation of biofilm formation in white cells was strictly dependent on Cph1 , as deletion of this factor abolished formation of biofilms ( Figure 2A and C ) . Mutant strains missing Cph1 therefore resembled Δste2/Δste2 mutants that are lacking the pheromone receptor and also failed to form biofilms ( Figure 2A ) . Reintegration of the CPH1 gene into Δcph1/Δcph1 mutants restored biofilm formation close to wildtype levels , confirming the essential role of Cph1 in the white cell response to pheromone . To account for strain background differences , cph1 deletion mutants were also constructed in SC5314 and these mutants were also found to be completely deficient in pheromone-stimulated biofilm formation ( Figure S1 ) . In contrast , loss of CPH1 had no effect on conventional biofilm formation in either SC5314 or P37005 strain backgrounds ( Figure S2 ) . We similarly re-examined the contribution of Tec1 to pheromone signaling in P37005 white cells . As shown in Figure 1 , deletion of TEC1 has a significant effect on both pheromone-stimulated and conventional biofilms in SC5314 . This is also true in the P37005 background , as tec1 mutants were defective in both models of biofilm formation ( Figure 2B and Figure S2 ) . However , unlike cph1 mutants , pheromone treatment still promoted substantial biofilm formation in tec1 mutants , while biofilm responses were abolished in the cph1 strain . Together , these results indicate that Tec1 does not have a selective effect on pheromone-stimulated biofilms , but that it plays a general role in biofilm formation . Pheromone-stimulated biofilms in P37005 were imaged by confocal scanning laser microscopy ( CSLM ) . These assays demonstrated that biofilms were increased upon pheromone addition , with wildtype biofilms ∼125 µm in depth ( Figure 2C ) . As expected , biofilms were greatly reduced in pheromone-treated Δcph1/Δcph1 mutants ( ∼25 µm thick ) , whereas pheromone-treated Δtec1/Δtec1 mutants produced a biofilm of intermediate thickness ( ∼100 µm ) ( Figure 2C ) . We note that pheromone-stimulated biofilms were substantially more fragile in the SC5314 strain background ( Figure 1 ) compared to P37005 ( Figure 2 ) , and therefore more easily disturbed by washing . Nonetheless , we conclude that Cph1 is the master regulator of the white cell pheromone response in both SC5134 and P37005 strains of C . albicans . In contrast , Tec1 appears to play a more general role in biofilm formation and is not specifically required for the response to pheromone . Transcriptional regulators at the bottom of a signaling cascade are often upregulated in response to the signal . To address whether CPH1 or TEC1 genes are induced upon pheromone treatment of white cells , northern analysis of gene expression was performed . Increased CPH1 gene expression was observed in white cells of both P37005 and SC5314 strains when challenged with pheromone ( Figure 3A ) . Expression of PBR1 , a gene previously reported to be induced by α pheromone [15] , was also increased in white cells treated with pheromone ( Figure 3B ) , whereas TEC1 expression was not detected by northern analysis ( data not shown ) . Gene expression of TEC1 and PBR1 was also examined using quantitative RT-PCR . While PBR1 was highly induced in white cells responding to pheromone ( ∼45-fold ) , TEC1 expression levels were not induced by pheromone in any of the media conditions tested ( Figure 3C ) . These results further support our finding that Cph1 , and not Tec1 , mediates transduction of the pheromone signal in C . albicans white cells . In contrast to white cells , opaque cells efficiently upregulate the entire repertoire of mating genes and undergo a-α cell fusion in response to pheromone . We addressed the roles of Cph1 and Tec1 in opaque cell signaling by quantifying morphological responses ( elongated projections ) in response to pheromone as well as mating frequencies , under standard ( non-biofilm ) conditions . Consistent with previous reports [14] , [17] , [18] , Cph1 was essential for the pheromone response in opaque cells , as Δcph1/Δcph1 mutants lacked detectable projection formation and did not undergo a-α mating ( Figure 4 ) . Reintegration of the CPH1 gene into the mutant strain restored these phenotypes to wildtype levels . In contrast , Δtec1/Δtec1 mutants displayed normal mating projection formation ( 97% ) when challenged with α pheromone , as well as normal a-α mating efficiency ( 59% ) ( Figure 4 ) . These results establish Cph1 as the master regulator of pheromone signaling in both white and opaque cells of C . albicans . Transcriptional profiling of the pheromone-stimulated response previously showed that more than 300 genes are induced in C . albicans opaque cells , while only 30 genes are induced in white cells [6] , [24] . These responses are media dependent , with opaque cells exhibiting the strongest response in Spider medium , while white cells are most responsive in Lee's medium [24] . In addition , previous profiling experiments were performed under planktonic conditions and cellular responses in biofilms were not examined . To determine the gene expression profile of white cells undergoing pheromone-stimulated biofilm formation , we performed profiling of P37005 cells induced to adhere to the plastic surface or grown under planktonic conditions ( Figure 5 ) . For each data point , the transcriptional response in the presence of pheromone was compared to the response in mock-treated controls . White cells examined under both planktonic and biofilm conditions showed pheromone-induced expression of many genes related to mating and pheromone MAPK signaling ( Figure 5A , lanes 1–2 and 7–9 , Figure 5B , and Table S3 ) . Thus , despite the fact that white cells are mating incompetent , genes involved in pheromone sensing ( STE2 ) , pheromone secretion ( HST6 ) , and pheromone modification ( RAM2 ) are upregulated , as well as genes associated with mating and karyogamy ( FIG1 , FUS1 , and KAR4 ) . Biofilm conditions resulted in an enhanced response to pheromone; 52 genes were induced by pheromone after 4 hours in biofilm conditions compared to 23 genes under planktonic conditions ( Figure 5B ) . Multiple genes were also repressed under biofilm conditions ( 14 genes ) while no genes were repressed >4 fold in planktonic conditions , and repressed genes were associated with DNA replication and the cell cycle ( Figure 5B ) . Many of the ‘biofilm-specific’ genes were also expressed under planktonic conditions but did not pass the 4-fold cutoff ( data not shown ) . Overall , the data indicates that the transcriptional response to pheromone in planktonic cells is primarily a subset of the response under biofilm conditions . These findings establish that the mode of growth , in addition to the culture medium , can markedly influence the strength of the transcriptional response to environmental signals . We similarly performed profiling on Δcph1/Δcph1 and Δtec1/Δtec1 mutants to determine the contribution of these factors to the transcriptional response to pheromone in white cells ( Figure 5A , lanes 3–6 and 10–12 ) . Notably , loss of CPH1 essentially abolished the entire transcriptional response to pheromone under both planktonic and biofilm conditions ( lanes 5 , 6 , and 12 ) . In contrast , deletion of TEC1 only slightly compromised the transcriptional response , as 44 genes were induced in Δtec1/Δtec1 cells compared to 52 genes in the wildtype strain ( Figure 5A , lanes 3–4 and 10–11 , and Figure 5C ) . Consistent with our northern blot and RT-PCR data , CPH1 was itself induced ( ∼7-fold ) in white cells exposed to pheromone , whereas the TEC1 transcript was not induced at 4 hours and was only weakly induced ( <3-fold ) at 24 hours ( Figure 5A and Table S3 ) . These observations support our conclusion that Cph1 , and not Tec1 , is the transcriptional mediator of the white response to pheromone . The switch between white and opaque forms occurs approximately once every 104 generations , although switching is also highly dependent on environmental factors [25]–[29] . To ensure that our profiling analysis accurately reflects gene expression from white cells and not from a contaminating minority of opaque cells , we also performed profiling on cells locked in the white developmental state . To this end , a Δwor1/Δwor1 mutant was constructed in the P37005 strain background , as loss of WOR1 prevents cells switching from white to opaque [30]–[32] . Expression profiles of wildtype and Δwor1/Δwor1 strains were similar , with pheromone signaling components and mating genes induced in both strains ( Figure 5A and 5D , and Table S3 ) . While most profiling experiments were performed at 4 hours post-pheromone treatment , expression profiles were also compared at 24 hours following pheromone addition in wildtype and “white-locked” cells ( Figure 5A , lanes 15 and 16 , respectively ) . Gene induction was increased in wildtype ( 196 genes ) and Δwor1/Δwor1 ( 258 genes ) strains at the 24-hour time point ( Figure 5E and Table S3 ) . This data reveals that white cells can mount a substantial response to pheromone exposure , and that the response is significantly stronger at 24 hours than at 4 hours . Finally , the transcriptional response to pheromone was compared between white and opaque cells , both grown under biofilm conditions ( Figure 5A , lanes 15 and 17 , and 5F ) . We note that opaque cells form very weakly adherent pheromone-induced biofilms under these assay conditions [9] , [33] . Overall , the number of the genes induced in opaque cells at 4 hours ( 188 genes ) was similar to the number induced in white cells at 24 hours ( 196 genes , see Table S3 ) . These results indicate that opaque cells in biofilms are generally more responsive to pheromone challenge than white cells and are consistent with previous results obtained under planktonic conditions [24] . Many of the genes induced in white and opaque biofilms were shared ( 87 genes ) , and this overlap was significant ( p<5×10−254 ) ( Figure 5F ) . With a 4-fold cutoff , it appears that 109 genes were induced only in white cells and 101 genes were induced only in opaque cells , however , a number of these genes were induced at least 2-fold in both white and opaque cells . After removal of these genes , there remain 76 white-specific and 59 opaque-specific genes , indicating that there is a unique transcriptional program acting in each cell type . A comparative table showing genes regulated by pheromone in white and opaque biofilms is provided ( Table S4 ) . Overall , our data indicates that phase-specific genes may play an important role in the different phenotypic outputs of pheromone signaling in C . albicans; biofilm formation in white cells and mating in opaque cells . We also compared the transcriptional program in pheromone-induced biofilms to that recently described in conventional biofilms [13] . Using a 2-fold cutoff for up- and down- regulated genes , we found that 662 genes were induced in conventional biofilms , while 486 genes were induced in pheromone-induced biofilms in white cells ( Figure S3 ) . 128 genes were shared between these two transcriptional programs ( p = 2×10−30 ) . The significance of this overlap is lost when using a more stringent cutoff ( p = 0 . 3 for a 4-fold cutoff ) . Similarly , gene overlap between repressed genes in the two biofilm models was significant using a 2-fold cutoff ( p = 9×10−3 ) , but not when using a 4-fold cutoff ( p = 0 . 6 ) ( Figure S3 ) . Gene Ontology analysis revealed that the 128 genes upregulated more than 2-fold in both datasets are enriched for genes involved in adhesion ( p<1×10−5 ) , including HWP1 , HXK1 , XOG1 , SUN41 , PHR1 , RFX2 , SAP4 , SAP5 , SAP6 , ALS1 , TEC1 and PBR1 . These results indicate that the transcriptional changes occurring during conventional and pheromone-induced biofilms are partially overlapping , but that the genes undergoing the highest fold changes in transcription are generally unique to each program . Transcriptional profiling of wildtype , Δcph1/Δcph1 and Δtec1/Δtec1 strains revealed a number of potential downstream targets of Cph1 and Tec1 . In total , we observed 13 genes that exhibited decreased induction by pheromone ( >2-fold ) in cph1 and tec1 mutants ( Table S3 and data not shown ) . Of these genes , six candidates were chosen for further analysis due to their dependence on CPH1 and TEC1 for pheromone-induced expression , and also because they were not induced in pheromone-treated opaque cells ( Figure 6A ) . The lack of induction in mutant white strains suggested these genes may play important roles in biofilm development in C . albicans . The six candidate genes were PBR1 , CFL11 , HGC1 , ORF19 . 7167 , ORF19 . 7170 , and ORF19 . 7305 . PBR1 has previously been implicated in pheromone-stimulated biofilms [15] , CFL11 is induced during the early development of conventional biofilms [34] , and HGC1 is a G1 cyclin-related protein required for hyphal formation and virulence [35] . Little is known about ORF19 . 7167 , ORF19 . 7170 and ORF19 . 7305 , although ORF19 . 7167 is a predicted adhesin-like protein [36] . Each of the six candidate genes were deleted in the P37005 strain background and tested for pheromone-stimulated biofilm formation ( Figure 6B ) . Only the Δhgc1/Δhgc1 mutant showed a significant reduction in biofilm formation when challenged with pheromone , while loss of the other five genes did not impact biofilm formation . Adding a functional copy of HGC1 back into the mutant strain restored biofilm formation , confirming that this gene is necessary for efficient pheromone-induced biofilm formation ( Figure 6B ) . We also tested the role of the six candidate genes in conventional biofilm development . Once again , only deletion of HGC1 had a significant effect on biofilm formation ( Figure 6C ) . Thus , while the average mass of wildtype biofilms was ∼11 µg , biofilms formed by the Δhgc1/Δhgc1 mutant were only ∼1 µg . Given the key role of Hgc1 in hyphal formation [35] , it is likely that the reduced filamentation of hgc1 mutants compromises their ability to form conventional biofilms . However , Hgc1 may play additional roles in adherence and/or biofilm maturation in light of CPH1-dependent upregulation of Hgc1 in pheromone-stimulated biofilms . The role of Hgc1 in sexual mating was also addressed . Opaque a strains lacking HGC1 were found to undergo efficient sexual mating with a wildtype α partner ( Figure S4A ) . The response to α pheromone was also normal in mutant hgc1 a cells ( Figure S4A and B ) . These experiments indicate that , in contrast to Cph1 , Hgc1 does not have a detectable role in the mating program . Taken together , these results identify Hgc1 as an important regulator of biofilm formation in both conventional and pheromone-stimulated biofilm models , but that this factor is dispensable for mating . Surprisingly , we note that deletion of PBR1 did not alter biofilm formation in white cells responding to pheromone , in contrast with a previous report [15] . Deletion of PBR1 also did not have a visible effect on conventional biofilm formation in our assays . One of the most surprising aspects of our comparison of conventional and pheromone-stimulated biofilms is the role of the transcription regulator Ndt80 . Ndt80 is required for conventional biofilms ( Figure 1 and [13] ) , yet its deletion results in increased pheromone-stimulated biofilm formation ( Figure 7 ) . Ndt80 plays a pleiotropic role in C . albicans including regulation of the glycosidase gene SUN41 and the endochitinase gene CHT3 [37] . In the absence of NDT80 , expression of SUN41 and CHT3 genes is compromised , leading to a defect in cell separation and growth as chains of cells [37] . We therefore tested whether enhanced biofilm formation in the ndt80 mutant was due , at least in part , to its cell separation defect . To examine this possibility , overexpression of cell wall degradation genes was carried out in the ndt80 background . Indeed , overexpression of either SUN41 or CHT3 in the ndt80 mutant resulted in a ∼30% reduction in biofilm formation ( Figure 7 ) . These results indicate that increased formation of pheromone-stimulated biofilms in ndt80 mutant strains can be attributed , at least in part , to a defect in cell separation .
We demonstrate that the transcription factor Cph1 ( ortholog of S . cerevisiae Ste12 ) is the master regulator of pheromone signaling in C . albicans , as deletion of this gene abolished both pheromone-stimulated biofilms by white cells and sexual mating by opaque cells . This result was surprising , as previous reports had indicated that Cph1 was critical for the opaque response to pheromone but dispensable for the white response [19] . Instead , the Tec1 transcription factor was proposed to be the downstream target of the pheromone MAP kinase cascade in white cells [14] , [38] . These studies led to a model of pheromone signaling whereby Cph1 directed mating gene expression in opaque cells , while Tec1 regulated biofilm formation in opaque cells [14] , [38] . On the basis of the results described here , we propose a new model in which the same MAP kinase components and Cph1 transcription factor are responsible for signal transduction in both white and opaque cells ( Figure 8 ) . Consistent with Cph1 , and not Tec1 , mediating white cell signaling , the CPH1 gene was highly induced by pheromone under both planktonic and biofilm culture conditions . In contrast , TEC1 was only weakly induced after 24 h in pheromone-induced biofilms . Deletion of Cph1 also abolished the genome-wide transcriptional response to pheromone in white cells and completely inhibited pheromone-induced biofilm formation . By comparison , loss of Tec1 resulted in the altered expression of only a subset of pheromone-induced genes and pheromone treatment still significantly enhanced biofilm formation in the tec1 mutant . Moreover , deletion of Tec1 compromised biofilm formation under both conventional and pheromone-stimulated conditions , indicating that Tec1 has a general effect on biofilm formation that is not specific to pheromone stimulation . Results with Cph1 and Tec1 were similar when compared between C . albicans P37005 and SC5314 strains , confirming that mutant phenotypes were similar in different strain backgrounds . Transcriptional profiling was used to uncover factors that act downstream of Cph1 ( either directly or indirectly ) in biofilm formation . Six genes were identified that are pheromone-induced in wildtype white cells under biofilm conditions but are not induced in either tec1 or cph1 mutants ( Figure 6 ) . Several of these candidates had previously been implicated in cell adhesion and/or biofilm formation [15] , [34]–[36] . Deletion of five of the six genes , however , failed to result in a defect in pheromone-stimulated biofilm formation . Either these genes do not play a role in biofilm development or their roles are masked by redundancy with other genes . Functional redundancy has been observed in other biofilm studies so that deletion of multiple factors is often necessary to observe a biofilm defect [39] , [40] . However , it was surprising that loss of PBR1 ( Pheromone stimulated Biofilm Regulator 1 ) did not affect biofilm development in any of our assays , as it was previously reported to be critical for this process [15] . One gene product shown to significantly influence biofilm formation was Hgc1 . Deletion of HGC1 resulted in decreased biofilm formation in white cells responding to pheromone , and also abolished formation of conventional biofilms . Hgc1 is therefore important for cell adhesion and biofilm development in both models of biofilm formation . Hgc1 is a G1 cyclin-related protein involved in hyphal morphogenesis and virulence [35] , [41] . Cells lacking Hgc1 exhibit a marked defect in hyphal formation , which may explain the inability of hgc1 mutants to form conventional biofilms . Presumably , Hgc1 also contributes to biofilm formation by other mechanisms , as hyphal growth was rarely observed in pheromone-stimulated biofilms and is unlikely to play an important role in this process . Recent studies have identified the core transcriptional network regulating conventional biofilm formation . Loss of any one of six transcription factors ( BCR1 , BRG1 , EFG1 , NDT80 , ROB1 , or TEC1 ) compromised biofilm formation both in vitro and in vivo [13] . The transcriptional changes during conventional and pheromone-induced biofilms show partial overlap ( Figure S3 ) , and we found that mutants lacking Bcr1 , Brg1 , or Rob1 , in addition to Tec1 , were also deficient in pheromone-stimulated biofilm formation . Four of the six master transcription factors therefore play a general role in mediating cell adherence and/or biofilm maturation . Surprisingly , although NDT80 is necessary for conventional biofilm formation , we observed that loss of NDT80 resulted in significantly thicker pheromone-induced biofilms than those formed by wildtype cells . Previous studies have established that C . albicans Ndt80 plays diverse roles in drug resistance and conventional biofilm formation [13] , [37] , [42] , and is also required for expression of cell separation genes ( e . g . , SUN41 and CHT3 ) whose gene products enable the separation of mother and daughter cells [37] , [43] . We found that pheromone-induced hyper-biofilm formation in the ndt80 mutant was significantly suppressed by overexpression of SUN41 or CHT3 . Our results therefore demonstrate that the cell separation defect in ndt80 mutants contributes to the formation of hyper-biofilms . Moreover , regulation of cell separation could play a general role in fungal biofilm formation , either by promoting the aggregation of cells within a biofilm or by increasing cell accumulation on the substrate surface . Despite utilizing the same signaling cascade , white and opaque cells exhibit very distinct phenotypes upon pheromone challenge . To dissect these differences , we performed transcriptional profiling on pheromone-treated white and opaque cells under both planktonic and biofilm conditions . White and opaque cells exhibited significant overlap in their transcriptional responses , although the overall response was weaker in white cells than in opaque cells . In fact , the response at 24 hours in white cells was closest to that at 4 hours in opaque cells ( Figure 5F ) . We also note that differences in gene expression were observed between pheromone-treated cells under planktonic and biofilm conditions . In general , pheromone responses were stronger under biofilm conditions , with increased expression of mating genes and stronger inhibition of DNA replication and cell cycle genes . These results establish that planktonic and biofilm cells experience different microenvironments with direct consequences for gene expression and function . Given that Cph1 mediates pheromone signaling in both white and opaque states , how do these cell types produce distinct biological outputs ? Presumably , white and opaque specific components are responsible , at least in part , for mediating these different responses . In S . cerevisiae , Ste12 ( the Cph1 ortholog ) can activate different signaling pathways through selective interactions with different transcription factors [44] , [45] . Pheromone signaling induces Ste12 homodimers that induce expression of mating genes , whereas a Ste12/Tec1 complex mediates activation of filamentous growth [44] , [45] . By analogy , it is possible that C . albicans Cph1 cooperates with different co-factors in white and opaque cells , thereby directing biofilm formation and sexual mating , respectively ( Figure 8 ) . In support of this model , different subsets of genes were induced by pheromone in white and opaque cells , indicating transcriptional activation of distinct pathways . White and opaque cells may also exhibit differences in biofilm formation due to inherent structural differences . In addition to differences in cell shape , white and opaque states exhibit marked differences in cell wall morphology , phase-specific antigens , and actin motility [46] , [47] . Additional studies will now be necessary to further characterize the physical differences between white and opaque cells and to reveal the roles of these two cell types in biofilm proficiency . Finally , we note that studies in related species will also help shed light on the mechanisms regulating white- and opaque-specific responses . The white-opaque switch has been described in the related species C . dubliniensis and C . tropicalis , where opaque cells again represent the mating-competent form [48] , [49] . It is therefore likely that the white-opaque switch evolved in the ancestor to C . albicans , C . dubliniensis and C . tropicalis . Future studies will compare white and opaque responses in these related pathogens to determine if mechanisms of pheromone signaling have been conserved between species , or if they have accrued different functions since they last shared a common ancestor . These approaches will further define the properties of each cell type that underlie the ability to generate distinct phenotypic outputs .
Media and pheromone used in these experiments were prepared as described previously [50]–[52] . C . albicans strains and oligonucleotides used in this study are listed in Tables S1 and S2 , respectively . To generate Δtec1/Δtec1 strains , the 5′ flanking and 3′ flanking regions of TEC1 were PCR amplified using primers 1167/1168 and 1169/1170 , respectively . The 5′ and 3′ PCR products were digested with ApaI/XhoI and SacI/SacII , respectively , and cloned into the plasmid pSFS2a [53] to generate the plasmid pSFS-Tec1 KO . The plasmid was digested with ApaI/SacI and transformed into either P37005 or an MTLa/a derivative of SC5314 ( CAY716 or RBY717 ) to generate heterozygous Δtec1/TEC1 mutants . The SAT1 marker was recycled [54] and the strains re-transformed with the deletion construct to generate Δtec1/Δtec1 strains CAY2506 and CAY2504 . A TEC1 complementation construct was made by amplification of the promoter and ORF using oligos 1334/1335 . The PCR product was digested with ApaI/XhoI and cloned into pSFS2a to generate pSFS2a-TEC1 AB . The plasmid was linearized with EcoRI and transformed into CAY2504 and CAY2506 to create CAY2748 and CAY2750 , respectively . For cph1 mutants , primers 1336/1337 and 1338/1346 were used to amplify the 5′ and 3′ regions of the CPH1 gene . 5′ and 3′ PCR products were digested with KpnI/ApaI and SacI/SacII and cloned into pSFS2a to generate pSFS2a-Cph1 KO . The construct was linearized with KpnI/SacI and transformed into CAY716 and RBY717 to generate heterozygous deletions . The SAT1 marker was recycled and strains again transformed with the deletion construct to generate Δcph1/Δcph1 strains CAY2899 and CAY2895 , respectively . The CPH1 complementation plasmid was constructed by PCR using primers 1419/1420 , and the PCR fragment digested with NotI/SacI and cloned into pSFS2a to create pSFS-CPH1 AB . The construct was then digested with HpaI and transformed into CAY2899 and CAY2895 to generate CAY3025 and CAY3028 , respectively . To generate gene deletions of ORF19 . 7167 , ORF19 . 7170 , ORF19 . 7305 , PBR1 , CFL11 and HGC1 , 5′ and 3′ flanking regions of each gene were amplified using primers 1226/1227 and 1228/1229 , 1194/1195 and 1196/1197 , 1187/1188 and 1189/2115 , 1218/1219 and 1220/1221 , 1202/1203 and 1204/1205 , 1210/1211 and 1212/1213 , respectively . The PCR products were digested with ApaI/XhoI and SacI/SacII and cloned into pSFS2a to generate pSFS-7167 KO , pSFS-7170 KO , pSFS-7305 KO , pSFS-PBR1 KO , pSFS-CFL11 KO and pSFS-HGC1 KO , respectively . These constructs were linearized with ApaI/SacI and transformed into CAY716 to generate CAY3445 , CAY3447 , CAY3693 , CAY3689 , CAY3687 and CAY3465 , respectively . The HGC1 complementation plasmid was constructed by PCR using primers 1845/1875 to amplify the promoter and ORF of HGC1 . The PCR product was cloned into pSFS2a using ApaI/XhoI to generate pSFS2a-HGC1 AB . This construct was linearized with SnaI and transformed into CAY3488 to create CAY3702 . Gene deletions of ROB1 , BRG1 , and BCR1 were achieved using an established fusion PCR approach [55] . 5′ and 3′ ORF flanking regions for ROB1 , BRG1 , and BCR1 were amplified using oligos 1773/1774 , 1775/1776 , 1781/1782 , 1783/1784 , 1765/1766 , and 1767/1768 , respectively . These PCR products were then combined with a selectable marker ( HIS1 or LEU2 ) by fusion PCR , as described [55] . Fusion PCRs were used to delete target genes in RBY1132 to generate homozygous deletions in ROB1 ( CAY3670 ) , BRG1 ( CAY3583 ) and BCR1 ( CAY3672 ) . For the ROB1 gene addback construct , a PCR fragment was amplified using primers 1875/1876 , digested with ApaI/SalI and cloned into pSFS2a to generate pSFS-Rob1 AB . The construct was then linearized with ApaBI and transformed into CAY3670 to create CAY3805 . BCR1 and BRG1 complementation plasmids were cloned using primers 1878/1879 and 1881/1882 , respectively . PCR products were digested with ApaI/XhoI and cloned into pSFS2a to generate pSFS-BCR1 AB and pSFS-BRG1 AB , respectively . The constructs were digested with EcoRI and BamHI and transformed into CAY3672 and CAY3583 to create CAY3858 and CAY3802 , respectively . To delete the MTLα locus and generate MTLa-type cells , plasmid pJD1 ( GenBank accession #JX486681 ) was digested with XmaI , and transformed into OHY13 , TF22 , TF95 , TF110 , TF115 , TF137 , and TF156 to create OHY13a , TF22a , TF95a , TF110a , TF115a , TF137a , and TF156a , respectively . pJD1 was created using primers MBL 660/661 ( 3′ flank ) and MBL 662/663 ( 5′ flank ) to amplify ∼500 bp regions flanking the MTL loci such that the regions were homologous to both MTLa and MTLα . The flanking regions were fused to the Candida dubliniensis ARG4 marker , which was amplified with primers UP2 and UP5 from pSN69 [56] , using primers MBL 661/663 that introduced the XmaI site to each end of the fusion product . The amplified product was digested with XmaI and ligated into pUC19 ( New England Biolabs ) . C . albicans white cells were grown in Spider medium at room temperature overnight . 5×107 cells were added to 12 well dishes ( Costar , Corning Inc . ) and mixed with 1 ml Lee's medium [51] in the presence of 0 . 01% DMSO or 10 µM C . albicans synthetic α pheromone . Cultures were mixed and incubated without shaking at room temperature for 24 h . Supernatants were removed and wells washed with phosphate-buffered saline ( PBS ) and photographed . Each experiment was performed using at least two independent isolates with three experimental replicates . We used a previously established protocol to measure dry weight of biofilms in a silicone model of conventional biofilm formation [55] . Pre-weighed sterile silicone squares ( Cardiovascular Instruments Corp . , PR72034-06N , 1 . 5 cm×1 . 5 cm ) were pre-incubated in bovine serum ( Sigma B-9433 ) overnight at 37°C while shaking ( 150 rpm ) in a 12-well plastic plate . The treated silicone squares were washed with 2 ml PBS , placed in 12-well culture dishes , and 2 ml Spider medium added . C . albicans strains were grown overnight in YPD medium and approximately 2×107 cells were added to each well . The inoculated plate was incubated at 37°C for 90 min with gentle agitation ( 150 rpm ) , either in the presence ( 10 µM ) or absence of α pheromone , or in the presence of a 0 . 01% DMSO control . Squares were washed with 2 ml PBS and incubation continued in 2 ml fresh Spider medium ( +/− pheromone ) for 24 h or 60 h at 37°C with gentle shaking . Supernatants were removed and silicone squares allowed to dry overnight , before weighing to determine biofilm mass . In addition , biofilm dry weight was measured from biofilms grown directly on the bottom of 12-well plates , as described in Nobile et al . [13] . Following biofilm growth , media was replaced with PBS , biofilm cells collected by pipet , and cells filtered over a pre-weighed filter . Filters were dried overnight and weighed . Four experimental replicates were performed and significance was determined using the student's one-tailed t-test . Pheromone-stimulated or conventional biofilms were grown on the bottom of polystyrene plates , as described above . The supernatant was removed and adherent biofilm cells obtained by scraping the well , and quantified by measuring the OD600 . Opaque MTLa and MTLα cells expressing different selection markers were grown in SCD medium at room temperature . Approximately 2×107 MTLa and MTLα strains were mixed together on a nitrocellulose filter on Spider medium . Cells were incubated for 48 h incubation at 25°C , then resuspended in water and plated onto selective media to quantitate mating frequency , as previously described [57] . C . albicans cells were grown as described in “Pheromone-Stimulated Biofilm Assays” . Cells were collected after pheromone treatment at room temperature for 4 h . Total RNA was isolated and cDNA synthesized as previously described [51] . Quantitative PCR was performed in a 7300 Real Time PCR System ( Applied Biosystems ) . Signals from experimental samples were normalized to the PAT1 gene expression level , as previously described [24] . Procedures and conditions for pre-hybridization , hybridization , washing and immunological detection of the probe with a CSPD chemofluorescent substrate for alkaline phosphatase were performed following the manufacture's recommendations ( Roche Applied Science ) . Probes for northern blot analyses were labeled with digoxigenin-11-dUTP ( Roche Applied Science ) by PCR . Probes for PBR1 , CPH1 and TEC1 were amplified using primers 1098/1099 , 1173/1174 and 1341/1342 , respectively . The transcriptional profiles of white cells responding to pheromone were performed on cells grown in planktonic and biofilm culture conditions . C . albicans white or opaque cells were grown in Spider medium overnight at room temperature . To harvest planktonic cells , overnight cultures were added into 12 ml Lee's medium at OD600 = 0 . 3 with DMSO ( control ) or synthetic α pheromone ( final concentration of 10 µM ) and incubated at 25°C for 4 h with gently shaking ( 150 rpm ) . For biofilm cells , cells were collected from the biofilm mat after treatment with DMSO or pheromone following the method described in “Pheromone-Stimulated Biofilm Assays” . Total RNA was isolated from cells using the Ribopure-Yeast Kit ( Ambion ) and treated with Turbo DNase ( Ambion ) . cDNA was synthesized from 10 mg of total RNA using SuperScript 3 reverse transcriptase ( Invitrogen ) with oligos ( dT ) 19V and pdN9 in reaction mixtures containing 0 . 5 mM DTT and 0 . 5 mM deoxynucleoside triphosphates ( aminoallyl-dUTP and deoxynucleoside triphosphates [3∶2] ) . RNA was hydrolyzed with 0 . 3 M sodium hydroxide and 0 . 03 M EDTA and neutralized with 0 . 3 M HCl to pH 7 . 0 . cDNA was purified and recovered using a Zymo kit ( Zymogen , DNA Clean & Concentrator ) . Samples were dried in a speed vacuum and resuspended in 9 µl of RNase-free water . Coupling of cDNA and hybridization to microarrays was performed as previously described [48] . cDNA from cells treated with 10 µM pheromone was hybridized against cDNA from matched cells treated with a mock DMSO control . Arrays were scanned on a GenePix 4000 scanner ( Axon Instruments ) . Profiles were quantified by using GENEPIX PRO version 3 . 0 and normalized using Goulphar ( http://transcriptome . ens . fr/goulphar ) . Pairwise average linkage clustering analysis was performed using the program CLUSTER and visualized by using TREEVIEW . The Candida genome database ( http://www . candidagenome . org/ ) was used to facilitate further analysis . All microarray data has been deposited into the NCBI Gene Expression Omnibus ( GEO ) portal under the accession number GSE44449 . Gene expression changes greater than 4-fold are displayed in Figure 5 . Gene expression changes greater than 2-fold or 4-fold are displayed in Figure S3 . The chi squared statistical test was used to determine the significance of overlapping array data . For pheromone-stimulated biofilms , 6 well plates ( BD Falcon ) containing 2 ml Lee's medium were inoculated with 1×108 cells from overnight cultures grown in Spider medium . Biofilms were grown in static conditions at room temperature , for 24 h . Conventional biofilms were grown as in “Conventional Biofilm Assays” for 24 h . For both types of biofilm assay , α pheromone ( final concentration of 10 µM ) or a DMSO control was added at the beginning of the 24 h timeframe . Biofilms were stained with 50 µg/ml concanavalin A conjugated to Alexa Fluor 594 ( Molecular Probes , C11253 ) for 1 h in the dark . Biofilms were gently washed with 1 ml PBS , then covered with water and imaged on a Nikon Eclipse C1si upright spectral imaging confocal microscope using a 40x/0 . 80W Nikon objective . For conA-594 visualization , excitation was at 561 nm and emission detection was at 605/60 nm . Images were acquired in a Z-stack series at 0 . 8 µm intervals , using Nikon EZ-C1 Version 3 . 80 software , and assembled into maximum intensity Z-stack projections using Nikon NIS Elements Version 3 . 00 software . | Candida albicans is the predominant fungal pathogen afflicting humans , where many infections arise due to its proclivity to form biofilms . Biofilms are complex multicellular communities in which cells exhibit distinct properties to those grown in suspension . They are particularly relevant in the development of device-associated infections , and thus understanding biofilm regulation and biofilm architecture is a priority . C . albicans has the ability to form different types of biofilms under different environmental conditions . Here , we compare the regulation of biofilm formation in conventional biofilms , for which a core transcriptional network has recently been identified , with pheromone-stimulated biofilms , which occur when C . albicans white cells are exposed to pheromone . Our studies show that several regulatory components control biofilm formation under both conditions , including the network transcriptional regulators Bcr1 , Brg1 , Rob1 , and Tec1 . However , other transcriptional regulators are specific to each model of biofilm development . In particular , we demonstrate that Cph1 , the master regulator of the pheromone response during mating , is essential for pheromone-stimulated biofilm formation but is dispensable for conventional biofilms . These studies provide an in-depth analysis of the regulation of pheromone-stimulated biofilms , and demonstrate that both shared and unique components operate in different models of biofilm formation in this human pathogen . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genome",
"expression",
"analysis",
"microbiology",
"gene",
"function",
"epigenetics",
"molecular",
"genetics",
"mycology",
"cell",
"adhesion",
"gene",
"expression",
"microbial",
"pathogens",
"biology",
"pathogenesis",
"cell",
"biology",
"genetics",
"genomics",
"molecular... | 2013 | Genetic Control of Conventional and Pheromone-Stimulated Biofilm Formation in Candida albicans |
Intrinsic immunity relies on specific recognition of viral epitopes to mount a cell-autonomous defense against viral infections . Viral recognition determinants in intrinsic immunity genes are expected to evolve rapidly as host genes adapt to changing viruses , resulting in a signature of adaptive evolution . Zinc-finger antiviral protein ( ZAP ) from rats was discovered to be an intrinsic immunity gene that can restrict murine leukemia virus , and certain alphaviruses and filoviruses . Here , we used an approach combining molecular evolution and cellular infectivity assays to address whether ZAP also acts as a restriction factor in primates , and to pinpoint which protein domains may directly interact with the virus . We find that ZAP has evolved under positive selection throughout primate evolution . Recurrent positive selection is only found in the poly ( ADP-ribose ) polymerase ( PARP ) –like domain present in a longer human ZAP isoform . This PARP-like domain was not present in the previously identified and tested rat ZAP gene . Using infectivity assays , we found that the longer isoform of ZAP that contains the PARP-like domain is a stronger suppressor of murine leukemia virus expression and Semliki forest virus infection . Our study thus finds that human ZAP encodes a potent antiviral activity against alphaviruses . The striking congruence between our evolutionary predictions and cellular infectivity assays strongly validates such a combined approach to study intrinsic immunity genes .
Recent discoveries have highlighted the role of intrinsic immunity genes in primate host defense against viral infections [1–3] . These genes are predicted to be locked in ancient , ongoing genetic conflicts with an ever-changing repertoire of viral infections [4–6] . Consistent with this prediction , the primate genes that encode for intrinsic immunity have been found to be evolving under positive selection , wherein they accumulate an excess number of non-synonymous substitutions ( protein-altering , dN ) compared to synonymous substitutions ( no effect on protein , dS ) . This kind of selective pressure is seen in cases where innovation in protein sequence can result in a selective advantage and rapid fixation , as is the case for a host immunity gene where a single mutation might improve its ability to recognize and destroy a pathogen . In fact , genome-wide scans for positively selected genes in primates reveal that adaptively evolving genes fall primarily into three functional categories: immune defense , chemosensory perception , and reproduction , with the majority of these genes being involved in immunity [7] . Four known groups of intrinsic immunity genes have been preserved over a broad taxonomic range; APOBEC3G and other APOBEC3 genes as well as TRIM5 are conserved across many mammalian orders [8 , 9] , while Fv1 is conserved in many Mus species [10] . All three of these groups of these intrinsic immunity genes have been shown to evolve under positive selection [4–6 , 10] . A signature of positive selection has not only provided information about the antiviral activity and age of these “restriction” genes , but has also helped to identify protein domains at the direct interface of the host–virus interaction [4] . The fourth intrinsic immunity gene that is present over a broad taxonomic range is the zinc-finger antiviral protein ( ZAP ) . ZAP from rat cells ( referred to as rZAP ) was identified due to its ability to significantly impair the replication of the mouse retrovirus , murine leukemia virus ( MLV ) , through a mechanism that involves binding and degrading viral RNAs in the cytoplasm [11] . Viral RNA recognition by rZAP is mediated by the 4 CCCH zinc finger motifs , which have been shown to directly bind viral RNA with high specificity [12] . ZAP-dependent recruitment of an RNA processing exosome in the cytoplasm then leads to viral RNA degradation [13] . Subsequent studies have found that rZAP can inhibit both alphaviruses [14 , 15] and filoviruses [16] by inhibiting the translation of incoming viral RNA . We undertook a combined approach using an evolutionary analysis of ZAP orthologs in primates as well as functional tests of ZAP isoforms from humans . The human ortholog of the ZAP gene encodes 2 protein isoforms that result from alternative splicing of a carboxy-terminal poly ( ADP-ribose ) polymerase ( PARP ) -like domain . This PARP-like domain is present in ZAP ( L ) and absent in ZAP ( S ) . Two shorter rat ZAP isoforms have previously been described , one corresponding to human ZAP ( S ) and the other to just the first 254 amino acids of the N-terminus ( rat NZAP ) , and neither contain the PARP-like domain[11] . Our evolutionary analysis of ZAP revealed strong evidence of positive selection throughout primate evolution . By examining the pattern of evolutionary change in this gene , we found , surprisingly , no evidence for positive selection in ZAP's CCCH RNA-binding domain . This implies that rapid alterations of viral mRNA binding have not driven ZAP evolution . Instead , the signature of positive selection in primate ZAP is confined to the PARP-like domain , implicating this domain as an uncharacterized and evolutionarily important interface for ZAP–virus interactions . Using infectivity assays , we show that human ZAP is capable of restricting expression of the retrovirus , MLV , as well as infection by the alphavirus , Semliki Forest virus ( SFV ) , in human cells . Moreover , we show that the presence of the PARP-like domain in the longer human ZAP isoform significantly enhances this restrictive capability against both viruses , making this the first demonstration of a PARP-like domain being implicated in an immunity role . Our results demonstrate that a combination of evolutionary and virological analyses can identify and validate specific protein domains involved in host–pathogen interactions , thereby uncovering previously unknown antiviral activity .
The human ZAP gene comprises 13 exons spanning approximately 60 kb ( Figure 1A ) . The two reported alternatively spliced isoforms of human ZAP code for proteins that either lack or contain a carboxy-terminal PARP-like domain ( Refseq NM_020119 . 3 and NM_024625 . 3 ) . We refer to the short and long isoforms as ZAP ( S ) and ZAP ( L ) , respectively ( Figure 1A ) . The ZAP ( L ) isoform encodes a protein with an N-terminal CCCH domain ( four CCCH motifs ) , a TPH , or TiPARP Homology domain ( conserved among ZAP paralogs and containing a fifth zinc finger motif ) , a WWE domain ( predicted to mediate specific protein–protein interactions in ubiquitin and ADP–ribose conjugation proteins [17] ) and a C-terminal PARP-like domain . The antiviral activity of rZAP was first discovered using a truncated rat ZAP protein ( rat NZAP ) , which consisted only of the 4 CCCH motifs that mediate RNA binding [11] ( Figure 1A ) . Subsequent analyses revealed that a longer rZAP protein also has antiviral ability . However , even this rat ZAP , presumed at the time to be “full length , ” corresponds to the human ZAP ( S ) isoform and does not include the C-terminal PARP domain . To gauge the tissue-specificity of the two human isoforms relative to each other , we used RT-PCR from a human tissue cDNA panel and primers specific to the ZAP ( S ) and ZAP ( L ) isoforms ( Figure 1B ) . We found that ZAP ( S ) is expressed has a broader expression pattern compared to the ZAP ( L ) isoform . However , the ZAP ( L ) isoform is expressed in tissues where it may mediate an intrinsic immunity function , including germline tissues and peripheral blood lymphocytes ( PBLs ) . Indeed , the tissue range of ZAP ( L ) is comparable to some of the APOBEC3 genes that we have analyzed previously using the same cDNA panel [6] . Next , we wanted to address whether ZAP ( L ) is limited to the primate lineage . The Refseq entries for both mouse and rat ZAP proteins only refer to a protein equivalent of human ZAP ( S ) . On examining the genomic sequence context for both rat and mouse ZAP genes , we found that they both have a downstream set of exons that could encode a putative PARP domain that would be orthologous to that from primate ZAP ( L ) . In order to determine whether these exons are spliced onto the remainder of the ZAP gene , we performed RT-PCR using RNA from rat liver and primers designed to the WWE domain ( shared between both ZAP ( S ) and ZAP ( L ) isoforms ) and the PARP domain . Indeed , we found that the rat ZAP gene can encode a ZAP ( L ) isoform that includes a C-terminal PARP domain ( Figure 1A ) . The reason this isoform appears to have been missed until now is because no ESTs corresponding to this PARP domain have been reported , suggesting that ZAP ( L ) may be more weakly expressed than ZAP ( S ) in rodents , as is the case in humans ( Figure 1B ) . In terms of overall architecture and paralogous genes , ZAP is most closely related to 3 other PARP-containing genes: PARP11 , PARP12 , and TiPARP [18] as well as ZRP2 , which only consists of a CCCH domain ( Figure S1 ) . Putative orthologs for all five genes can be found in genome sequences from a variety of mammals , as well as chicken and fish , suggesting that this gene family is ancient [19] and that the PARP-containing ZAP isoform dates back to the origin of vertebrates . To investigate whether the ZAP gene has evolved under positive selection during primate evolution , we sequenced the ZAP ( L ) coding sequence ( 2 . 7 kb ) via RT-PCR from 13 primates representing 33 million years of evolution . Our analysis included 6 hominoids , 3 Old World monkeys ( OWM ) , and 4 New World monkeys ( NWM ) . The phylogeny constructed from the primate ZAP sequences was congruent with the generally accepted primate phylogeny confirming that the sequences are orthologous ( Figure 2A ) . Using the free-ratio model in the PAML suite of programs [20] , which allows an independent assignment of dN/dS ratios to each evolutionary branch , we found that several branches of the phylogeny show dN/dS > 1 ( bold numbers in Figure 2A ) . Furthermore , when we compared the likelihood of ZAP evolution under codon models that prohibit ( Nsites models M1 , M7 , or M8a ) or permit positive selection ( M2 and M8 ) , we found that models permitting positive selection fit ZAP evolution significantly better than those that disallow it ( p < 0 . 0003 , Table 1 ) . This indicates that ZAP has evolved under positive selection during primate evolution , suggesting it has a long history of actively participating in host–pathogen interactions . To determine which domains were responsible for the signature of positive selection in ZAP , we performed three separate PAML analyses using the N-terminal CCCH domain ( amino acids [aa] 1–240 ) , the central domain containing the TPH and WWE motifs ( aa 241–700 ) , or the C-terminal PARP domain ( aa 701–902 ) . Significantly , our PAML analyses did not reveal any evidence for positive selection acting on the N-terminal CCCH domain . Indeed , we found a high degree of conservation of the CCCH domain throughout mammalian evolution ( Figure S1 ) , arguing that rapid alteration of viral RNA-binding has not had a significant impact on ZAP evolution . Sliding window dN/dS analyses based on different pairwise alignments of primate ZAP genes suggests that positive selection may have episodically acted on the central domain containing the TPH and WWE motifs ( aa 241–700; Figure S2 ) . However , since no codons show a recurrent signature of positive selection , these domains are not highlighted by the PAML analyses ( Table 1 ) . Rather , we found a robust signature for positive selection in the PARP-like domain ( aa 701–902; Table 1 ) . We found that a model of episodic positive selection ( Free Ratio ) was significantly more likely than a model of constant positive selection ( Model 0 , one fixed dN/dS; p = . 01 , df = 21 ) , suggesting that ZAP has been engaged in episodic conflicts with exogenous infectious agents . Three codons were identified as having evolved under recurrent positive selection in the PARP domain , using both PAML and REL analyses ( Table 1 ) [20 , 21] . Surprisingly , the three residues for which we obtained high confidence of positive selection ( Figure 2B ) are found in close proximity to residues that are thought to mediate the contact residues for NAD+ binding ( cd01439 . 2 from the CDD database [22] largely modeled from the crystal structure of the chicken PARP-1 catalytic domain [23 , 24] ) . This is highly unexpected because it would seem that these residues should be highly constrained as part of PARP or PARP-like function , for which NAD+ binding is obligatory . It is possible that these residues could be rapidly evolving because of ZAP ( L ) interactions with viral proteins that may involve the same protein domains that interact with NAD+ . Thus , contrary to expectation , the only signature of recurrent positive selection we found in primate ZAP evolution is limited to a domain that is missing from the rZAP gene , which is the only version that has been previously tested in any antiviral assays . Since protein innovation has been primarily favored exclusively in the PARP domain , this would predict that ZAP ( L ) is the more evolutionarily relevant antiviral isoform . This prediction can be directly evaluated by testing whether the region under the most intense positive selection , the PARP-like domain , enhances the antiviral activity of ZAP . The original identification of rat NZAP was based on its ability to inhibit MLV infection and MLV long terminal repeat ( LTR ) –based expression . We created HA-tagged human ZAP ( L ) and ZAP ( S ) expression constructs in parallel with the previously described rat NZAP clone [11] and then tested their restrictive capabilities by co-transfecting increasing amounts of ZAP plasmid with a MLV-based vector , in which the firefly luciferase reporter gene is inserted between the 5′ and 3′ LTRs of MLV . A similar MLV-based luciferase reporter construct was previously used to show that the 3′ LTR sequence of MLV is the target of the CCCH domain of rat NZAP [12] . In repeated luciferase assays ( n = 6 ) , we found that all ZAP expression plasmids significantly suppressed MLV LTR-driven luciferase expression ( Figure 3A ) . Our experiments also show that ZAP ( L ) is a 2- to 3-fold stronger suppressor of MLV than ZAP ( S ) at the highest tested levels of ZAP ( 200 ng; Figure 3A–3B ) . In addition , Western analysis of lysates from the comparable transfections reveals that the ZAP ( S ) expression level is significantly higher than ZAP ( L ) ( Figure 3C ) , despite equal amounts of transfected plasmid . This suggests a secondary control of ZAP ( L ) expression , which we have not explored further . However , these results do imply that the 2- to 3-fold higher inhibition of MLV expression seen with ZAP ( L ) is likely to be a significant underestimate . As a control , we also tested whether the ZAP isoforms were effective at inhibiting a similar construct that utilized the LTR from HIV to drive luciferase gene expression . We found that human ZAP ( L ) , ZAP ( S ) and rat NZAP had no effect on HIV LTR-driven luciferase ( Figure 3D ) , suggesting a strong specificity for retroviral inhibition , as was previously reported for rat NZAP [11] . These results demonstrate human ZAP has potent antiviral function , and that ZAP ( L ) is a more effective inhibitor than ZAP ( S ) . Previous results have also demonstrated that rZAP can inhibit alphaviruses [14] . Therefore , we tested whether the two human ZAP isoforms were also capable of restricting an alphavirus , SFV , using a previously described DNA-based recombinant SFV vector system that expresses the β-gal gene [25] . We performed single-round infectivity assays in HeLa cells lines stably-expressing HA-tagged human ZAP ( L ) , human ZAP ( S ) , or rat NZAP . The expression level of ZAP ( L ) was about equal to that of ZAP ( S ) , while rat NZAP appeared to be more highly expressed in our cell lines ( Figure 4A ) . While ZAP ( L ) , ZAP ( S ) and rat NZAP all demonstrate antiviral activity against SFV infection , ZAP ( L ) was significantly more potent than the other isoforms , with almost 10-fold inhibition of SFV compared to only 2-fold inhibition by either ZAP ( S ) or rat NZAP ( Figure 4B ) . To assess whether the antiviral activity of human ZAP is a general antiviral effect ( and to rule out the possibility that our observations are due simply to greater cell death in ZAP ( L ) -expressing cells ) , we performed single-round infectivity assays with HIV expressing luciferase in HeLa cells stably expressing the ZAP isoforms . We found that none of the isoforms of ZAP were capable of significantly inhibiting HIV infection , demonstrating that the antiviral activity of ZAP is virus-specific ( Figure 4C ) . Intriguingly , the expression of human ZAP isoforms in baby hamster kidney ( BHK ) cells did not confer resistance to SFV ( unpublished data ) , suggesting that the rapid evolution of ZAP has likely resulted in species-specific restriction and that ZAP restriction of alphaviruses is at least partially dependent on the context of other host proteins .
Our evolutionary analyses of primate ZAP ( L ) reveals that the C-terminal PARP-like domain encoded only by the ZAP ( L ) isoform is in fact the only domain that is under recurrent positive selection . Based on this insight from our evolutionary analysis , we tested the abilities of both human ZAP ( L ) and ZAP ( S ) isoforms to inhibit SFV and HIV infection in human cells . Our results clearly demonstrate that human ZAP can inhibit SFV , but not HIV , and that the presence of the PARP domain significantly enhances the ability of ZAP to suppress SFV infection . While previous work also implicated the rat ZAP as an inhibitor of the murine retrovirus , MLV , in rodent cells [11] , we do not observe such an effect against MLV infection ( unpublished data ) at the level of ZAP expressed in our stable human cell lines , even though we do see an effect on SFV infection in the same cell lines . Therefore , we believe that the viral antagonist ( s ) that drove ZAP evolution in primates was likely a member of the Togaviridae family [14 , 15] rather than the Retroviridae family . The rZAP CCCH domain can be very specific in its RNA-binding and antiviral activity [12] . Our experiments on MLV and HIV ( Figure 3 ) suggest that the RNA-binding specificities of rodent and primate ZAP are not significantly different . Our analyses cannot rule out the possibility that some subtle alteration of RNA-binding may have occurred during the course of mammalian evolution . However , the absence of positive selection in the CCCH domain unambiguously rules out an evolutionary conflict scenario in which the restrictive ability of ZAP was shaped by repeated episodes of selection for dramatic alterations in RNA-binding activity . It is not yet clear what role a PARP-like domain could play in ZAP function . PARP function has been primarily characterized in the nucleus where it is believed that the addition of ADP-ribose moieties to chromatin proteins results in their looser association with DNA , thereby allowing greater access to transcription and DNA repair machineries [19 , 26–28] . A similar mechanism of protein modification by ZAP's PARP domain could be imagined to disrupt binding by viral RNA-binding proteins ( Figure S4 ) . However , the ZAP PARP-like domain lacks what is believed to be an essential catalytic glutamic acid ( position 988 in PARP-1 ) and therefore may not be catalytically active as a canonical PARP [29] , although its NAD+ binding site is conserved . While the mechanism by which the PARP-like domain mediates its antiviral effect is unknown , we can still address why the PARP-like domain has been evolving under positive selection . Using the model for antagonistic host–pathogen interactions , there are several possible scenarios for the adaptive evolution of the PARP-like domain in ZAP: either the PARP domain evolves towards recognizing the virus , or it evolves away from being recognized by the virus . In the first scenario , the PARP-like domain may increase ZAP's affinity for viral mRNA binding proteins ( analogous to APOBEC3G associating with nucleocapsid [30] ) , in which case the PARP-like domain is under selective pressure to increase binding to viral proteins . The alternate possibility is that viruses may encode specific antagonists to bind ZAP's PARP domain and sequester or degrade ZAP ( analogous to Vif degradation of APOBEC3G [31] ) , in which case the selective pressure on ZAP PARP would be to decrease binding with viral antagonists . We favor the latter possibility because the residues that show a recurrent signature of positive selection also fall proximal to the NAD+ binding sites , which may be essential for any PARP-like function . Given the recent re-emergence of alphavirus epidemics in Africa and Asia and the reports of alphavirus disease in Europe [32 , 33] , our discovery that human ZAP encodes a potent anti-alphaviral activity that depends on the PARP domain may guide future strategies for therapeutic drug design to treat alphavirus-related disease .
Human , chimp , and rhesus ZAP sequences were obtained from the respective genome projects . We amplified ZAP genes from additional primates ( Figure 2A ) by RT- PCR using RNA isolated from individual cell lines ( obtained from Coriell and E . Eichler [gibbon] ) . RT-PCR was done in two overlapping fragments using the Invitrogen One-Step RT-PCR with Platinum Taq kit ( Invitrogen ) . To amplify the 5′ half of the gene , we used the following primers to generate a 1 . 4 kb fragment: Forward-5′ ATGGCGGACCCGGAGGTG , Reverse-5′ CTCGGGAAGCAGGTCCAGCATCC . The 3′ half of the gene was amplified as a 1 . 4 kb fragment with the following primers: Forward-5′ AATGCTGATGGAGTGGCCACAG , Reverse-5′ GACAACTAACTAATCACGCAGGCTTTGTC . To demonstrate the existence of a PARP-containing rat ZAP ( L ) , the following primers were used to generate a fragment spanning exons 7–13 of ZAP: Forward-5′ TCTGACTCCTACCCCATCCGA , Reverse-5′ GCAACCTTTCTCTTTCTCTGATTCCAC . All sequencing was done using ABI BigDye version 3 . 0 . Sequences obtained in this study were deposited in Genbank and assigned accession numbers EF494425–EF494434 . To detect ZAP expression in different human tissues , we used human cDNA panels in which 5 ng of first-strand cDNA from various tissues were preloaded ( Primgen ) and performed PCR using primers specific to either the ZAP ( S ) or ZAP ( L ) isoform . We used Clustal_X [34] to generate a multiple alignment for ZAP from all primate species sequenced . Maximum likelihood analysis was performed with codeml in the PAML 3 . 14 . 1 software package [35] . To detect selection , multiple alignments were fitted to the NSsites models that disallow positive selection ( M1 , M7 , M8a ) or to models that permit positive selection ( M2 , M8 , M8b , respectively ) assuming the f61 model of codon frequencies . Simulations were run with multiple starting values for dN/dS . Likelihood ratio tests were performed to assess whether permitting codons to evolve under positive selection gives a significantly better fit to the data . To identify specific codons that are evolving under recurrent positive selection , we used the NSsites model from codeml in the PAML 3 . 14 . 1 software . To confirm the sites identified by the codeml approach , we also implemented the random effects method ( REL ) from the online DataMonkey package , with a Bayes significance factor of 50 as the cutoff [21] . The REL approach differs from NSsites in that it allows the synonymous substitution rate to vary among codons . Our pairwise sliding window analyses were performed using the K-estimator program [36] , with a 300 bp window and a 50 bp slide . N-terminal hemagglutinin ( HA ) -tagged human ZAP ( L ) and ZAP ( S ) were cloned by reverse transcription ( RT ) -PCR from RNA isolated from 293T cells using the HA-specific 5′ primer , 5′-CAGGCGAATTCGCCACCATGTATCCATACGATGTTCCAGATTACGCTGCGGACCCGGAGGTGTGC-3′ , and isoform-specific 3′ primers as follows: HA-ZAP ( L ) -5′ TTCAGGATATCCTAACTAATCACGCAGGC-3′ and HA-ZAP ( S ) -5′-TTCAGGATATCCTATCTCTTCATCTGCTGCAC-3′ . The rat N-terminal HA-tagged NZAP was cloned by RT-PCR from RNA isolated from Rat2 fibroblasts using the HA-specific 5′ primer , 5′-CAGGCGAATTCGCCACCATGTATCCATACGATGTTCCAGATTACGCTGCAGATCCCGGGGTA-3′ and the 3′ primer , 5′-TTCAGGATATCTCAGTGAAGGAAGCGGTCTCT-3′ . Each gene was transferred into the same mammalian expression vector ( pcDNA4; Invitrogen ) which drives gene expression under the cytomegalovirus IE94 ( CMV ) promoter . All constructs were confirmed by sequencing . HeLa cells were cultured in Dulbecco's modified Eagle medium ( DMEM ) with 10% fetal bovine serum and were maintained at 37 °C in 5% CO2 . To create stably expressing cell lines , the ZAP constructs were individually transfected into HeLa cells using Fugene6 Reagent ( Roche ) and stably expressing lines were selected and maintained with Zeocin ( 0 . 1 mg/ml ) . The expression levels for clonally derived lines were checked by Western analysis using an HA-specific antibody ( HA . 11; Covance ) and the highest expressing lines were chosen for the infectivity assays . Transient co-transfections were performed using the Mirius TransIT-LTR Transfection reagent according to the manufacturer's recommendations . All cells were transfected with a total of 300 ng DNA comprising 100 ng LlucSN plus empty pcDNA4 vector ( No ZAP ) , HA- ZAP expression plasmid ( HA-hZAPL , HA-hZAPS , or HA-rNZAP ) , as indicated , and the total DNA transfected was equalized with empty pcDNA4 vector . After 48 h , the cells from triplicate tranfections were lysed using 150 μl of Cell Culture Lysis reagent ( Promega ) . To quantify the luciferase activity , 20 μl of lysate was analyzed with a Luciferase Assay kit ( Promega ) and luciferase activity ( light intensity ) was measured with a luminometer . For Western blot analysis , 293T cells were plated and transfected as described above using different amounts of transfected HA-huZAP ( L ) or HA-huZAP ( S ) , with the total amount of transfected DNA equalized using empty pCDNA4 . After 48 h , the lysates from these cells were harvested with NP-40-doc buffer ( 20 mM Tris [pH 8 . 0] , 120 mM NaCl , 1 mM EDTA , 1% NP-40 , 0 . 2% Na-deoxycholate , and protease inhibitors [Roche complete cocktail tablets] ) , and incubated on ice for 5 min and then frozen at −20 °C . Lysates were resuspended in SDS loading buffer , boiled for 5 min , and then loaded and run on a 12% NuPAGE Novex Bis-Tris Gel and transferred to nitrocellulose membrane ( Pierce ) . For HA-tagged ZAP protein detection , membranes were probed with a HA-specific antibody ( HA . 11; Covance ) at a 1:2 , 000 dilution , followed by horseradish peroxidase ( HRP ) -conjugated goat anti-mouse secondary antibody ( Amersham Biosciences ) at a 1:3 , 000 dilution . For Actin protein detection , a rabbit anti-Actin Ab ( Sigma ) was used at 1:10 , 000 dilution , followed by HRP-conjugated donkey anti-rabbit secondary antibody ( Amersham Biosciences ) at 1:3 , 000 . Detection was performed with the ECL Plus Reagent ( Amersham Biosciences ) . SFV virus was made using the DNA-based Semliki Forest Virus vectors ( pSMARTlacZ and pSCAHelper ) , which were a kind gift from Dr . Rod Bremner [25] . To make HIV virus stock we used HIV-luc2Δenv pseudotyped with VSV-G [37] . To make the virus stocks , the constructs were transiently transfected into 293T cells with FuGene6 Reagent ( Roche ) . Virus released into the cell culture supernatant by 48 h was harvested , clarified by centrifugation , and stored at −80 °C . SFV infections were performed as previously described [25] . Briefly , cells were plated in 96 well plates at 2 × 104 cells per well the day before infection . Virus was thawed and treated with chymotrypsin to generate active virus and HeLa cell lines ( normal and stably expressing ZAP isoforms ) were challenged with 3-fold serial dilutions of active virus . After 24 h , the cells were lysed and the β-galactosidase activity was measured using Galacton ( Tropix ) as a substrate with 1-s measurements using a luminometer ( Thermo Fluoroskin Ascent , Thermo ) . For the HIV infections , cells ( normal or stably expressing ZAP isoforms ) were plated in 96-well plates at 2 × 104 cells per well and after 24 h were challenged for 48 h with serial dilutions of HIV virus . After 48 h , the cells were lysed using the Cell Culture Lysis reagent ( Promega ) . To quantify the luciferase activity , the lysate was analyzed with a Luciferase Assay kit ( Promega ) and luciferase activity ( light intensity ) was measured with a luminometer . | Host–virus interactions are a classic example of genetic conflict in which both entities try to gain an evolutionary advantage over the other . This “back-and-forth” evolution is predicted to result in rapid changes of both host and viral proteins , which results in an evolutionary signature of positive selection , especially at the direct interaction interface . Recent studies have demonstrated that host proteins can target intracellular stages of the viral life cycle to potently inhibit viruses . Collectively , these host proteins are referred to as “intrinsic immunity” proteins . One such protein , zinc-finger antiviral protein ( ZAP ) , was previously described from rats and shown to inhibit retroviruses and alphaviruses . We queried the primate orthologs of ZAP to ascertain both whether they have evolved under positive selection , and whether they have antiviral activity . We found that the signature of positive selection was restricted to a poly ( ADP-ribose ) polymerase–like domain in a longer isoform of primate ZAP . The longer human ZAP isoform has increased antiviral activity against both retroviruses and alphaviruses . Thus , positive selection correctly predicted the more potent antiviral isoform of this protein . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"primates",
"viruses",
"virology",
"evolutionary",
"biology",
"homo",
"(human)",
"rattus",
"(rat)"
] | 2008 | Positive Selection and Increased Antiviral Activity Associated with the PARP-Containing Isoform of Human Zinc-Finger Antiviral Protein |
The multifunctional NS1 protein of influenza A viruses suppresses host cellular defense mechanisms and subverts other cellular functions . We report here on a new role for NS1 in modifying cell-cell signaling via the Hedgehog ( Hh ) pathway . Genetic epistasis experiments and FRET-FLIM assays in Drosophila suggest that NS1 interacts directly with the transcriptional mediator , Ci/Gli1 . We further confirmed that Hh target genes are activated cell-autonomously in transfected human lung epithelial cells expressing NS1 , and in infected mouse lungs . We identified a point mutation in NS1 , A122V , that modulates this activity in a context-dependent fashion . When the A122V mutation was incorporated into a mouse-adapted influenza A virus , it cell-autonomously enhanced expression of some Hh targets in the mouse lung , including IL6 , and hastened lethality . These results indicate that , in addition to its multiple intracellular functions , NS1 also modifies a highly conserved signaling pathway , at least in part via cell autonomous activities . We discuss how this new Hh modulating function of NS1 may influence host lethality , possibly through controlling cytokine production , and how these new insights provide potential strategies for combating infection .
The multifunctional Non-Structural-1 protein ( NS1 ) is one of 14 proteins encoded by influenza A virus [1–3] . NS1 interacts with several viral and host components to subvert cellular defense mechanisms and promote viral replication via two domains: an N-terminal RNA-binding domain ( RBD ) and an effector domain ( ED ) , both of which can multimerize . The RBD binds dsRNA , a common viral replication intermediate , thereby preventing dsRNA-mediated activation of a key interferon-induced antiviral activity [4] . The ED sequesters multiple host proteins required for cellular mRNA maturation as well as export factors controlling apoptosis [2] . The functions of the NS1 protein benefit the virus by inhibiting the production and activation of host antiviral factors , establishing preferential viral mRNA translation , and extending host cell viability to allow sufficient viral maturation . Most interactions between the influenza NS1 protein and its host targets have been identified in cells grown in culture for which the assays are typically cell autonomous ( e . g . , viral replication , altered host gene expression ) . We sought to identify novel targets of NS1 that might be involved in cell-cell communication using Drosophila melanogaster , a highly developed genetic system for analyzing well-conserved signaling pathways [5 , 6] and for elucidating interactions between host and pathogen encoded proteins [7 , 8] , including viral proteins from influenza [9] . Using this system , we have identified a novel activity of the NS1 protein involved in regulating the Hedgehog ( Hh ) signaling pathway . The Hh signaling pathway plays a central role in growth and development [10 , 11] , tissue repair [12] , and adult immune responses [13–15] of vertebrates and invertebrates . In the Drosophila wing primordium ( or wing imaginal disc ) , Hh is secreted from cells in the posterior compartment and binds to the Patched ( Ptc ) receptor , resulting in phosphorylation and surface accumulation of the seven pass transmembrane domain protein Smoothened ( Smo ) in a stripe of cells in the anterior compartment ( referred to as the central organizer ) [16] . Activated Smo , in turn , recruits Costal-2 ( Cos-2 ) to the plasma membrane , disrupting an inhibitory complex with the transcription factor Cubitus interruptus ( Ci; Gli in mammals ) , thereby stabilizing and activating the full-length Ci-155 protein . In the absence of Hh signaling , microtubule associated Cos-2 promotes Ci-155 phosphorylation via cAMP-dependent Protein Kinase A ( PKA ) and other kinases , resulting in partial proteolysis of Ci-155 to a N-terminal repressor ( Ci-75 ) that silences a subset of Hh target genes [16] . In the current study , we report that NS1 alters expression of Hh target genes by directly modulating the specific activity of the transcriptional effector , Ci/Gli . This novel signaling activity remains unaltered by previously defined mutations in NS1 that block its interactions with known host effectors . We identified a novel point mutation in a surface residue of NS1 ( A122V ) , however , that does abrogate this signaling function . Incorporation of the A122V mutation into a mouse-adapted influenza virus increased expression of some Hh targets and cytokines , accelerated lethality , and increased host morbidity relative to the parental virus . These effects of NS1 are at least in part due to direct cell autonomous effects of NS1 since transfection of NS1 alone into human lung cell lines altered expression of BMP2 , the mammalian homologue of Drosophila dpp , in an A122V-dependent fashion . These findings reveal a new signaling function and binding domain of NS1 that modulates influenza virulence and potentially provides new therapeutic avenues for treating infection .
We expressed an NS1 cDNA from the A/Vietnam/1203/04 H5N1 viral strain ( referred to as NS1 ( Vn ) ) , double tagged with HA ( N-terminus ) and Myc ( C-terminus ) epitopes in the Drosophila wing by placing it under the transcriptional control of the yeast upstream activating sequence ( UAS ) [17] . Flies carrying this construct were crossed to strains expressing the yeast GAL4 transactivator protein in wing-specific patterns to conditionally activate expression of the UAS-NS1 ( Vn ) transgene in the wings of progeny ( Fig 1B–1D ) . Localized expression of NS1 ( Vn ) in the central organizer increased the distance between wing veins L3 and L4 1 . 34X compared to wings with no transgene ( Fig 1A and 1B , n = 5–7 , p<1 . 6x10-5 ) . Similarly , ubiquitous expression throughout the wing increased the distance between the L3 and L4 veins 1 . 3X ( Fig 1C , n = 3–7 , p<7 . 4x10-5 ) in the presence of one copy of NS1 and 1 . 47X ( Fig 1D , n = 6–7 , p<2 . 35x10-6 ) in the presence of two copies , a phenotype indicative of spatially broadened Hh signaling [18 , 19] . Notches along the edge of the wing were also observed ( arrows in Fig 1C and 1D ) indicating that NS1 has additional non-Hh related effects , which may be mediated by the Wg or Notch signaling pathways ( see S1 Text ) . Consistent with its adult wing phenotype , expression of NS1 ( Vn ) in ubiquitous ( Fig 1F ) or organizer-specific ( Fig 1K ) patterns increased expression of a minimal synthetic dpp-lacZ reporter [20] , a known Hh target , 7 . 48X ( n = 3 , p<2 . 8x10-3 ) along the A/P border in late larval wing discs . In contrast , expression of other Hh target genes , such as Ptc , Collier ( Col ) , and Engrailed ( En ) , did not appear appreciably altered by NS1 ( Vn ) ( S1A , S1B , S1E , S1F , S1I and S1J Fig ) , suggesting that this effect was limited to a specific subset of Hh targets genes ( see below , however , regarding the extended effects of the more active NS1 protein from the PR8 strain of influenza ) . Dpp secreted from the central organizer diffuses to form a long range gradient extending both anteriorly and posteriorly [21] , which can be detected in situ by antibody staining for the phosphorylated ( and active ) form of the cytoplasmic signal transducer Mothers against decapentaplegic ( pMAD ) [22] or Dpp target genes such as spalt [23 , 24] . Both ubiquitous ( Fig 1G–1J ) and central-organizer specific ( Fig 1K and 1L ) expression of NS1 ( Vn ) in the wing disc intensified pMAD staining 2 . 36X ( n = 4–6 , p<0 . 025 ) and broadened the domain of Spalt staining ( 1 . 3x distance from A/P , n = 3–5 , p<0 . 03 ) indicating an expanded range of Dpp signaling . Consistent with NS1 ( Vn ) acting non-autonomously via Dpp on adjacent cells , preventing Dpp diffusion out of the central organizer abolished this induction of BMP signaling ( S1 Text and S2 Fig ) . Since NS1 is a multifunctional effector protein that interacts with several well-characterized host factors , we next examined the role of known mutations that disrupt such interactions . Surprisingly , none of the mutations in NS1 residues that abrogate its established interactions with host effectors compromised its Hh-modulating activity in the Drosophila wing ( see S1 Text , S3 Fig , and Table A in S1 Text for further analysis ) , nor did co-expression of NS1 with candidate effectors such as PI3K or polyA binding protein ( see Table A in S1 Text ) . We , therefore , conducted an unbiased mutagenesis screen to isolate new mutant alleles of NS1 that could no longer perform this novel signaling function . One revertant allele we recovered greatly reduced the NS1 wing phenotype as well as reduced NS1-dependent expression of dpp-lacZ 2 . 65X ( n = 5–6 , p<1x10-3 ) in the imaginal disc ( Fig 2C , 2D vs . 2A , 2B ) without affecting NS1 protein levels ( Fig 2M ) . This NS1 mutant carried an alanine to valine substitution at position 122 ( A122V ) —a highly conserved residue mapping to the surface of the ED ( Fig 2N ) . ( Note: NS1 ( Vn ) lacks 5 amino acids after position 79 that are present in almost all other strains . So , in order to remain consistent with the majority of NS1 variants , the NS1 ( Vn ) amino acids were numbered with respect to their alignment with other NS1 sequences , effectively inserting blank sequence into the 5 amino acid deletion . ) Substitutions of other , larger amino acids at position 122 also greatly reduced NS1 activity indicating that a key functional feature of the alanine residue is likely to be the small size of its side chain ( see S1 Text and Table A in S1 Text for further structural requirements of this residue ) . Because different strains of influenza can vary dramatically in virulence , we examined the Hh modulating activity of NS1 from several viral strains including the standard seasonal Udorn virus , NS1 ( Ud ) , the more recently emerged swine flu NS1 ( Sw ) , and the murine adapted PR8 virus , NS1 ( PR8 ) . NS1 transgenes from these strains were expressed with a strong wing specific driver so that even low levels of Hh inducing activity could be detected . NS1 ( Ud ) produced wing phenotypes similar to those of NS1 ( Vn ) ( Fig 2E and 2F ) , while NS1 ( Sw ) was 2 . 24X weaker as judged by relative dpp-lacZ expression at the A/P border of the wing imaginal disc ( Fig 3G and 3H , n = 4 , p<3 . 8x10-3 ) . In contrast , NS1 ( PR8 ) was 7 . 9X stronger than NS1 ( Vn ) ( Fig 3I and 3J , n = 6 , p<1 . 4x10-3 ) . However , this greater activity of NS1 ( PR8 ) was also reduced 3 . 53X ( n = 6 , p< , p<6 . 4x10-3 ) by the A122V mutation ( Fig 2K and 2L ) . Since A122 is highly conserved between strains ( not limited to just those we tested ) , we generated mutations in other surface amino acids close to A122 that differed between high ( PR8 ) and low ( Sw ) activity forms of NS1 to identify residues that may contribute to NS1 activity ( Fig 2N and Table A in S1 Text ) . These experiments revealed that no individual or tested combinations of residues contributed significantly to the variable activity between strains , suggesting that multiple amino acids , possibly encompassing different regions of the protein , contribute incrementally to the differential Hh-modulating activities of NS1 . As described above , among the viral strains tested , NS1 ( PR8 ) was the most active in enhancing dpp-lacZ expression ( Fig 2J ) . In addition , NS1 ( PR8 ) altered expression of Hh target genes that appeared unaffected by NS1 ( Vn ) ( e . g . , Col , Ptc , and a narrow anterior En domain ) ( S1C , S1G and S1K Fig ) . These NS1 ( PR8 ) -specific effects included increased ( e . g . , dpp-lacZ: Fig 2J , Ptc: S1G Fig ) as well as decreased ( e . g . , Col and En: S1C and S1K Fig ) target gene expression . The differences in activity between NS1 ( PR8 ) and NS1 ( Vn ) notwithstanding , incorporation of the A122V mutation into NS1 ( PR8 ) reversed its effects on all Hh target genes , indicating that this single amino acid is critical for both the positive and negative effects of NS1 across strains ( Fig 2K and 2L and S1D , S1H and S1L Fig ) . We next examined the effect of NS1 on Hh pathway components . Overexpressing the Ptc receptor , which sequesters Hh and blocks signaling by preventing it from diffusing into the anterior compartment [19 , 25] , abrogated the ability of NS1 ( Vn ) to upregulate dpp-lacZ expression ( Fig 3A and 3B ) . Conversely , ectopic activation of Hh signaling in the anterior compartment , induced by ubiquitously expressing a constitutively active phosophomimetic form of Smoothened , potentiated NS1 ( Vn ) activity throughout this region ( Fig 3C and 3D ) . These results indicate that NS1 ( Vn ) activity is dependent upon Hh signaling . To determine whether NS1 requires the presence of the full-length form of the transcriptional effector Ci ( Ci-155 ) , we reduced Ci-155 levels either by RNA-interference ( Fig 3E ) , or by promoting proteolysis of Ci-155 to the Ci-75 repressor form by expressing the active catalytic subunit of PKA ( PKAact ) ( Fig 3F and 3G ) or Cos-2 ( S4A and S4B Fig ) . In all cases , NS1 ( Vn ) activity was abolished , demonstrating a requirement for Ci-155 in this process . Full length Ci-155 requires additional modification ( s ) from the Hh signaling pathway for full activity ( Ciact ) [26] . To determine whether NS1 ( Vn ) required such Ci activation , we co-expressed it with the inhibitory regulatory subunit of PKA ( PKArep , Fig 3H ) or in clones of cells lacking function of Cos-2 ( Fig 3I and 3I’ ) , both of which block Ci phosphorylation/degradation and thereby stabilize Ci-155 . Under these conditions , however , Hh signaling is also disabled due to required positive roles of both PKA and Cos-2 in transducing the Hh signal [27] . In these experiments , despite the presence of abundant Ci-155 , NS1 ( Vn ) was unable to fully activate dpp-lacZ expression . Thus , Ci-155 is required , but not sufficient , for NS1 ( Vn ) activity and additional positive input from Hh signaling is essential for its full effect ( see S1 Text and S4 Fig for further analysis ) . Consistent with NS1 affecting the specific activity of Ci rather than its expression or stability , levels and expression pattern of full-length Ci were unaltered by NS1 ( Vn ) ( Fig 3J and 3K ) nor by the stronger variant , NS1 ( PR8 ) ( S1M–S1P Fig ) . In further support of NS1 acting in conjunction with Ciact , co-expression of NS1 ( Vn ) with a non-cleavable variant of Ci-155 , CiS849A [28] , which is active independently of Hh signaling , strongly enhanced dpp-lacZ expression throughout the disc ( Fig 3L and 3M ) . This effect of NS1 extended well beyond the domain of Hh signaling , and did so even in the absence of cos-2 function ( Fig 3N and 3N’ ) . We conclude that NS1 interacts with fully-activated Ciact , which is normally present along the A/P border where Hh signaling is high . We next examined whether Gli1 , the mammalian transcriptional activator homologous to Ci [27 , 29] , would similarly mediate a response to NS1 ( Vn ) . Unlike endogenous ci which is transcriptionally repressed in the posterior compartment of wing discs [30] , both the mammalian and Drosophila effectors inserted into the Drosophila genome on UAS-transgenes can be expressed throughout the disc and subsequently activated in posterior and A/P border cells where the Hh ligand is abundant [31 , 32] . When we expressed Gli1 in Drosophila wing discs , we observed that it exhibited an opposing activity to that of Ci , as it nearly abolished the low basal level of expression of the synthetic dpp-lacZ reporter ( Figs 3O vs . 1E ) and also , unexpectedly , blocked the ability of NS1 ( Vn ) to augment its expression ( Fig 3P vs . 3A ) . Similarly , Gli1 prevented NS1 ( Vn ) from fully activating expression of the more strongly-expressed dpp-lacZEP enhancer trap reporter along the A/P border ( Fig 3T vs . 3Q and 3R ) . [30] . Likewise , Gli1 activity in posterior and A/P border cells was significantly impeded upon co-expression with NS1 ( Fig 3T vs . 3S ) . Taken together , these results demonstrate an interaction between NS1 and Gli1 . Futhermore , the A122V mutation eliminated this interaction such that NS1 ( Vn ) -A122V could no longer prevent Gli1-mediated activation of dpp-lacZEP expression ( Fig 3U vs . 3T ) . This indicates that the A122 residue is critical for interacting with the Hh pathway signaling effector across species . NS1 was also found to regulate Notch ( N ) signaling in wing discs ( see S1 Text and S4 Fig for further analysis ) . Commensurate with NS1 altering the activity of the transcriptional effector of the Hh signaling pathway , an interaction was also detected between NS1 and the transcriptional effector of the Notch ( N ) signaling pathway , N-ICD , which could also be relieved by the A122V mutation ( S1 Text and S4 Fig ) . Similar to Hh signaling , Notch ( N ) signaling is also involved in tissue repair and immunity [12 , 33] , suggesting that this novel binding domain of NS1 may be involved in various types of cell non-autonomous tasks that regulate the broader environment of the tissue during infection . Consistent with the strong genetic interactions observed between NS1 and Ci , we found that full-length Ci and NS1 colocalize in A/P border cells of NS1-expressing discs , although no obvious differences in the staining patterns were observed between the wild type and mutant NS1 proteins ( Fig 4F–4I and 4K–4N ) . In order to determine whether NS1 directly interacts with Ci in this functionally relevant region , wing imaginal discs were incubated with Ci-155 and NS1 antibodies , processed for immunofluorescence , and examined by fluorescence resonance energy transfer-fluorescence lifetime imaging ( FRET-FLIM ) analysis over a zone encompassing the A/P organizer ( Fig 4E , 4J , 4O , and 4Q ) . Discs expressing NS1 ( Vn ) -WT displayed decreased lifetime values ( Tau ) of the fluorescent secondary antibody bound to Ci-155 at the A/P border ( 1 . 9 to 2 . 2 ns ) compared to discs with no transgene ( 2 . 1 to 2 . 4 ns ) , representing a range of 10–15% FRET efficiency [34] and , thus a direct interaction between NS1 and Ci-155 ( Fig 4J vs . 4E and 4Q ) . Furthermore , expression of NS1 ( Vn ) -A122V showed a significant reduction in this interaction , represented by lifetime values between 2 . 0 to 2 . 3 ns , resulting in 2–5% FRET efficiency and a 50% reversal in comparison to NS1 ( Vn ) -WT expressing discs ( Fig 4O vs . 4J and 4Q ) . These data provide strong evidence for NS1 interacting directly with Ci-155 along the A/P border and that mutation of A122 significantly impedes this interaction , implicating this residue in selective Ci binding . We next tested whether NS1 also exerted Hh-modulating activity in mammalian cells . Plasmids containing NS1 ( PR8 ) -WT- , NS1 ( PR8 ) -A122V- , or GFP-encoding sequences were transfected into the human lung epithelial cell line , NL20 , and analyzed for expression of the Hh target gene BMP2 , the mammalian homologue of Drosophila dpp . Compared to surrounding cells or cells expressing a GFP-expressing control transgene , cells expressing NS1 ( PR8 ) -WT proportionally induced BMP2 protein expression , as observed in flies ( Fig 5A , 5B and S6A Fig ) . NS1 ( PR8 ) -A122V protein levels appeared generally lower than NS1-WT in NL20 cells ( S6B Fig ) . Nonetheless , when cells with the highest levels of NS1 ( PR8 ) -A122V ( A122V-high in S6B Fig ) were compared to those expressing NS1 ( PR8 ) -WT , a clear reduction in BMP2 induction was still observed in the mutant-expressing cells ( Fig 5A and 5B ) . These findings indicate that the Hh-modulating activity of NS1 is conserved between flies and humans , and moreover that this activity depends on the same A122 residue of NS1 . Interestingly , in NL20 cells ( Fig 5A and 5C ) as well as in mouse lungs infected with the PR8 strain of influenza virus ( Fig 5D ) , we observed a higher proportion of NS1 ( PR8 ) -A122V in the nucleus as compared to NS1 ( PR8 ) -WT . In addition , mutant NS1 often formed nuclear accumulations or aggregates that were not observed with the wild type protein ( Fig 5A ) . This supports a model in which the A122V mutation alters the specific activity of NS1 and/or its distribution and proximity to host binding partners . We next tested whether Hh signaling was altered by influenza in a murine infection model , and whether the A122V mutation of NS1 had any effect on such activity . We used the PR8 virus for these experiments since NS1 ( PR8 ) is highly active for this function of NS1 and because PR8 virulence is independent of CPSF30 binding by NS1 , a requirement for suppression of cellular mRNA maturation in other strains [35 , 36] . This latter fact was important since the A122V mutation in NS1 ( Ud ) , presumably due to its proximity to the CPSF30 binding site at M106 , reduced this binding interaction in vitro ( S7 Fig ) . Nonetheless , mutations of M106 or G184 which selectively eliminate CPSF30 binding did not alter NS1 ( Vn ) activity in Drosophila ( Table A in S1 Text ) , implying that CPSF30 binding is not involved in Hh signaling modulation by NS1 . Sibling mice were infected with either the PR8-WT or PR8-A122V virus and lung airway epithelium were analyzed for Hh target gene expression 2 days post-infection . Compared to uninfected lungs , the canonical Hh-target genes , Ptch1 ( Fig 6A and 6B ) , and BMP2 ( Fig 6C and 6D ) were both strongly up-regulated in infected lungs in a cell-autonomous fashion at the apical plasma membrane , a ciliated region of the cell where Hh components assemble [37] . Local non-autonomous effects were also occasionally observed in cells immediately adjacent to infected cells suggesting that virally infected cells exert effects on surrounding , ostensibly healthy , cells ( asterisks in Fig 6C ) . Similar to NL20 cells , NS1 expressed from the PR8-A122V mutant virus was found at slightly but statistically significantly lower levels compared to NS1 expressed from the WT virus ( Fig 6E ) . Despite the lower levels , and in contrast to the diminished activity of NS1-A122V when expressed in flies and in transfected cells , the mutated virus induced higher expression of the Hh target Ptch1 compared to PR8-WT ( Fig 6A and 6F ) . However , PR8-A122V did not discernibly alter the expression of BMP2 compared to PR8-WT . Taken together , NS1-A122V has altered activity compared to NS1-WT in flies , human cells , and in vivo in mice . Potential reasons for differences between the activity of NS1 assayed in isolation versus in the context of viral infection in vivo are considered below in the discussion section . Having observed an effect of PR8-A122V in the context of in vivo infection in mice , we next examined its physiological effects . Both the PR8-WT and PR8-A122V viruses established comparable levels of infection as assessed by viral titers in lungs ( Fig 7A ) and the range of tissues infected in mice at either two , four , or seven days post infection ( influenza was below the limit of detection in the brain , spleen , and liver ) ) , and displayed similar temporal regulation of viral genes in cell culture ( see S1 Text and S8A Fig ) . However , the PR8-A122V virus significantly hastened lethality of mice compared to the parental PR8 ( PR8-WT ) strain in three independent experiments ( Fig 7B ) and produced greater signs of morbidity by 3 days post-infection ( e . g . , reduced mobility , hunched posture , labored breathing , and pilo-erectus , although insignificant change in weight loss–S8B Fig ) , indicating that the mutant virus is considerably more pathogenic than the parental strain . This observation suggests that lower levels of Hh target gene induction by NS1-WT during infection in vivo may serve a protective function for the host whereas unrestrained signaling , as occurs during infection with PR8-A122V , may be more detrimental . To determine whether the increased pathogenicity might be linked , at least partially , to the induction of cytokine storms ( an uncontrolled positive feed-back loop between immune cells and cytokine production thought to have caused many fatalities during the past influenza pandemics ) cytokine levels were measured in infected lung extracts [38–40] . Indeed , levels of the proinflammatory cytokines IL6 ( Fig 7C ) and CXCL-10 ( S9 Fig ) were significantly higher in extracts ( lavages ) from lungs infected with PR8-A122V compared to PR8-WT viruses , while other factors such as TNF-α and IL-1α were not significantly altered ( S9 Fig ) . It has recently been shown that Hh signaling can directly modulate the immune response by activating expression of cytokines , such as IL6 [13 , 41 , 42] . Thus , we sought to determine whether the upregulation of IL6 observed in infected lungs ( Fig 7C ) might be a consequence of the direct activation of Hh signaling by NS1 , or rather derives from a general immune response . We detected both a cell-autonomous effect of influenza inducing an increase in IL6 expression in infected cells , as well as a general non-autonomous increase in mouse lung tissue ( Fig 7E ) . Furthermore , IL6 expression was higher in PR8-A122V infected cells compared to PR8-WT ( Fig 7D and 7E ) . Thus , the hastened lethality occurring in mutant-virus infected animals may be due , in part , to a direct interaction between NS1 and Hh signaling causing unrestrained expression of some cytokines .
In this study , we identified a novel effect of the influenza virulence factor NS1 in modulating Hh signaling . This Hh altering activity was initially identified using the Drosophila wing as a discovery model system . Then , we demonstrated a conservation of this activity in vertebrate systems such as transfected cultured human lung cells , as well as in influenza-infected mice . Detailed mechanistic analysis in Drosophila established that NS1 acts at the level of the transcriptional effector , Ci/Gli1 , and FRET data strongly suggest that this interaction is direct . The effect of NS1 on Hh target gene expression in Drosophila was either positive or negative depending on the particular target gene , NS1 strain , and co-expression of Drosophila Ci or vertebrate Gli partner . The positive versus negative effects of NS1 on particular target genes may reflect promoter-specific interactions in which the presence of NS1 can either stabilize or disrupt critical interactions between transcription factors , co-factors , and other transcriptional machinery . Importantly , the single amino acid substitution , A122V , which was recovered by a forward genetic screen in Drosophila , dramatically altered most NS1 responses in flies , human lung epithelial cells , and airways of infected mice . Furthermore , mutations abrogating interactions with several known host effectors involved in viral replication or suppression of the interferon response had little , if any , effect on the Hh signaling function of NS1 assayed in flies . These observations suggest that a novel , highly specific interaction surface of NS1 mediates these modulatory effects on Hh signaling . The above results strongly implicate NS1 as a modulator of Hh signaling , however , differences in its activity could be observed depending on the cellular context . Most notably , NS1-A122V , when expressed independently of other viral proteins ( i . e . , in flies and cultured human cells ) had reduced activity compared to NS1-WT . However , this same mutation resulted in a more pronounced inflammatory response in vivo in mice where we observed higher expression of some Hh targets , and greater virulence when assayed in the context of viral infection ( i . e . , in PR8 infected mouse lungs ) suggesting that other viral factors contribute to determining the net effect of the NS1-A122V mutation on the Hh response . One model that could account for these differences involves specific interactions between NS1 and nuclear factors . We have shown that NS1-A122V localizes more to the nucleus than NS1-WT . Thus , during viral infection , the spectrum of interactions between NS1 , viral factors , host defense components , and Hh targets , occurring specifically in the nucleus may be altered preferentially by the A122V mutation . Increased nuclear levels of NS1-A122V coupled with a reduced interaction with Ci/Gli1 ( as was detected in flies ) might favor alternative NS1-effector interactions that lead to a more pronounced inflammatory state . Viral factors that shuttle between the nucleus and cytoplasm , such as the matrix protein ( M1 ) , nucleoprotein ( NP ) , and nuclear export protein ( NEP ) , may be key in mediating these interactions [43] . Hh signaling is a plausible target for manipulation by pathogens , as this pathway plays an integral role in cell survival and proliferation [10 , 11 , 16] . Indeed , the Hh target gene Ptch1 was identified in an independent study as a critical host gene involved in influenza infection [44] . Therefore , our study provides a mechanistic interpretation for this observation , and reinforces the idea that controlling Hh signaling during influenza infection may favor viral dissemination . One way in which constraining Hh signaling by NS1 may be beneficial to the virus is by limiting damage to the lung in order to preserve the viral habitat . Hh signaling has been shown to be induced in the lung by damage caused by chemical agents and many types of ailments [45–48] . Furthermore , the Hh receptor , Ptc , is expressed in infiltrating and circulating lymphocytes suggesting that immune cells are primed to respond to the Hh ligand secreted from the inflamed area [45] . These studies indicate a key role for Hh signaling in repairing damaged lung tissue by remodeling the epithelium through , or in conjunction , with activated immune cells . Similarly , in fruit flies , Hh signaling has been shown to play a critical role in remodeling the epithelial barrier in response to pathogen infection [15] . However , too much remodeling by overactive signaling has also been shown to lead to the formation of detrimental fibrotic tissue [49] . In the case of influenza infection , Hh signaling may be activated indirectly to help repair the damaged lung epithelium . However , activation in an uncontrolled manner , such as that induced by PR8-A122V , may result in more fibrotic tissue damage . Thus , controlling the Hh response may help the host avoid such deleterious effects to the infected tissue and improve viral survival as well as dissemination . Unrestrained Hh signaling could also be detrimental to the host by excessively activating expression of cytokines , such as IL6 [13 , 41 , 42] . Indeed , we found that influenza infection strongly induced CXCL-10 and IL6 expression , the latter of which was partially enabled by a direct interaction between Hh signaling and NS1 . As IL6 was present at significantly higher levels in animals infected with PR8-A122V compared to PR8-WT , we speculate that the hastened lethality by the mutant virus may be caused , at least partially , by a direct induction of cytokine storms by NS1 . Other cytokines regulated by the Hh pathway such as IL-8 , Mcp-1 , and M-csf could also contribute to this phenotype as well [41] . Furthermore , the fact that no mutation at position 122 of NS1 ( or the analogous position in other viral strains ) has been identified previously in any influenza strain , may reflect the critical role of NS1 in dampening the level of cytokine activation , resulting in optimized host survival and/or viral spread . Interestingly , our initial screen in Drosophila revealed that NS1 from Swine flu ( NS1 ( Sw ) ) behaved similarly to the A122V mutant of other strains despite the lack of such a mutation . This may reflect an inability of this particular NS1 protein to regulate Hh signaling and restrain cytokine induction which may be directly linked to the augmented mortality rate of this virus observed in several animal models [50–53] . Indeed , many pro-inflammatory cytokines , including IL6 , were detected at higher levels in mouse and cynomolgus macaque lungs infected with this particular strain [51] . NS1 proteins from other influenza stains may act similarly to NS1 ( Sw ) , thus initial screening in Drosophila may prove a useful method of defining the virulence of emerging viral strains . There is a precedent for other viruses interacting with the Hh pathway , such as Epstein Barr Virus [54] , and Hepatitis B and C [55] . At least in the case of Hepatitis B and C , treatment of cells and tissue with a potent Hh antagonist appears to significantly limit viral outputs [56 , 57] . Additionally , blocking Hh signaling with the conventional antagonist , Cyclopamine , can often limit the extent of inflammation and fibrosis in other types of distressed tissue as well [58–61] . Therefore , further restricting Hh signaling with small molecule antagonists , such as Vismodegib ( GDC-0449 ) , which has been US-FDA approved as a treatment for basal cell carcinoma , represents a promising avenue to explore as a treatment for influenza [62] . In contrast to the currently available therapies such as the seasonal vaccines and antivirals that target strain-specific and rapidly-mutating viral proteins , treatments that target highly-conserved viral host targets may ultimately provide superior and continual protection across a broader spectrum of viral strains . In conclusion , our study demonstrates a novel activity for NS1 in modulating Hh signaling during influenza infection which elicits , to some extent , protection to the host . This effect is dependent upon a direct interaction with the Ci/Gli transcription factor , but the precise mechanism by which NS1 exerts it effects on the Hh pathway during infection remains to be addressed . Potential new therapies involving Hh inhibitory compounds could derive from these findings and demand further investigation .
Animal care and breeding are performed in the AAALAC accredited vivarium ( Vertebrate Animal Assurance No . A3194-01 ) at The Scripps Research Institute . For mouse experimentation and euthanasia , the AAALAC and NIH guidelines were followed and approval was given by the TSRI animal committee . Mice were euthanized when moribund to prevent undue suffering , and were sacrificed when they showed a failure of locomotion/activity and stayed huddled in a corner of the cage . These mice were unable to eat or drink and showed respiratory difficulties . Pilot studies showed that such mice would die within 24 hours . Generation of mitotic clones and immunohistochemistry was performed as previously described in [28] . All antibodies used are listed in S1 Text . Lungs were harvested at 2 dpi and placed in PBS-buffered formalin , blocked in paraffin , 10-μm tissue sections were cut , and placed on glass slides . Lung sections where deparaffinized in Xylenes ( 3X 10’ ) , and washed in 100% Ethanol ( 2X 5’ ) . After progressive rehydration ( 95% Ethanol 2X 5’ , 70% Ethanol 1X 5’ ) , slides were rinsed 3 times in 1X PBS . Antigen unmasking was performed using a reagent from Vector ( H-3300 , 2 . 5ml in 250ml H2O ) , and microwaving 2X for 5 minutes , followed by a 20 minute cool down , and 2 PBS1X rinses at RT . Lungs sections were outlined using an Elite Mini PAP Pen ( Diagnostic BioSystems K 042 ) , and incubated in blocking buffer ( 1XPBS with 1% BSA ) for 30’ . Incubation with primary antibodies primary antibodies was performed overnight at 4°C in a humid chamber followed by 3X PBS rinses . Secondary incubation was performed in blocking buffer followed by 3X rinses in PBS . Slides were mounted using VectaMount AQ ( Vector , H 5501 ) . NL20 human lung epithelial cells ( ATCC ) were maintained in a 5% CO2 atmosphere at 37°C in Ham's F12 medium ( Gibco ) supplemented with 1 . 5 g/L sodium bicarbonate , 2 . 7 g/L glucose , 2 . 0 mM L-glutamine , 0 . 1 mM non-essential amino acids ( Lonza , 13–1146 ) , 0 . 005 mg/ml insulin ( Sigma-Aldrich ) , 10 ng/ml epidermal growth factor ( Corning ) , 0 . 001 mg/ml transferrin ( Lonza ) , 500 ng/ml hydrocortisone ( Sigma-Aldrich ) and 4% fetal bovine serum . Transfections were performed on monolayers grown on Collagen-coated ( Gibco , A10483-01 ) coverslips and transfected for 24 hours with 0 . 5 μg of the indicated plasmids using X-tremeGENE 9 DNA Transfection Reagent ( Roche ) , according to the manufacturer’s instructions . Cells were washed with PBS , fixed with 4% formaldehyde and stained with primary antibodies . After incubation with Alexa fluor-conjugated secondary antibodies in blocking buffer for 1 hour at RT , cells were mounted with ProLong Diamond antifade mountant with DAPI ( Molecular Probes ) for microscopy analysis . The influenza A/PR/8/34 ( PR8; H1N1 ) wild type virus was rescued as previously described [63] . Alanine 122 in the PR8 NS1A protein was changed to valine using site-directed PCR mutagenesis of the PR8 NS gene , and the resulting DNA was cloned into a pol-lI plasmid , pHH21 . This mutation does not change the sequence of the NS2 protein . Viruses were amplified as described previously [36] . All viral segments were sequenced in their entirety . C57BL/6 mice were infected intra-tracheally with A/PR/8/34; H1N1 viruses at 104 PFU and monitored daily for morbidity for 14 days . Alternatively , mice were anesthetized with isoflurane and lungs dissected . Cytokine production was measured from lung homogenates by ELISA assay using DuoSet kits from R&D systems ( R&D , Minneapolis MN ) . All infected mice were housed in biocontainment at the animal facility of TSRI . Quantification of viral titers in mouse lungs is described in S1 Text . Details of fly genetics , Western blotting , antibodies , cloning constructs and techniques , confocal and FRET-FLIM imaging , quantification , and statistics can be found in S1 Text . | The NS1 protein produced by influenza A viruses alters host cellular defense mechanisms . We report here on a new role for NS1 in modifying cell-cell communication via the Hedgehog ( Hh ) signaling pathway . Genetic and microscopy studies in flies indicate that NS1 alters the transcriptional read-out of Hh targets by interacting directly with the transcriptional effector , Ci/Gli . Infected mouse lungs and human lung cells transfected with NS1 also revealed an upregulation of Hh target genes , including the non-canonical target IL6 . We identified a point mutation in NS1 ( A122V ) that impairs this target activation in Drosophila . The A122V mutation was incorporated into a mouse-adapted influenza A virus and was found to cell-autonomously increase expression of some Hh targets and hastened lethality in mice . NS1 also exhibited A122V-dependent effects on BMP target gene expression in a human lung cell line . These results indicate that , in addition to its multiple intracellular functions , NS1 also modifies communication between host cells . This new activity of NS1 appears to influence lethality caused by the virus and may facilitate viral dissemination by extending host survival . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"invertebrates",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"microbial",
"mutation",
"immune",
"physiology",
"cytokines",
"pathology",
"and",
"laboratory",
"medicine",
"influenza",
"pathogens",
"immunology",
"microbiology",
"orthomyxoviruses",
... | 2017 | Influenza NS1 directly modulates Hedgehog signaling during infection |
Motivation: Recently , copy number variation ( CNV ) has gained considerable interest as a type of genomic variation that plays an important role in complex phenotypes and disease susceptibility . Since a number of CNV detection methods have recently been developed , it is necessary to help investigators choose suitable methods for CNV detection depending on their objectives . For this reason , this study compared ten commonly used CNV detection applications , including CNVnator , ReadDepth , RDXplorer , LUMPY and Control-FREEC , benchmarking the applications by sensitivity , specificity and computational demands . Taking the DGV gold standard variants as a standard dataset , we evaluated the ten applications with real sequencing data at sequencing depths from 5X to 50X . Among the ten methods benchmarked , LUMPY performs the best for both high sensitivity and specificity at each sequencing depth . For the purpose of high specificity , Canvas is also a good choice . If high sensitivity is preferred , CNVnator and RDXplorer are better choices . Additionally , CNVnator and GROM-RD perform well for low-depth sequencing data . Our results provide a comprehensive performance evaluation for these selected CNV detection methods and facilitate future development and improvement in CNV prediction methods .
Copy number variation ( CNV ) is a type of genomic structural variation that contains segmental duplications or deletions of a DNA fragment; the CNV size usually ranges from 1 kb to 3 Mb[1] . CNVs are found widely in individual human genomes , and they seldomly lead to genetic diseases[2] . CNVs can change the number of copies of a gene present in cells , thus affecting the coding sequences of genes , and they are associated with complex phenotypes [3] . CNVs also play an important role in the susceptibility or resistance to human diseases , such as cancer [4] , Alzheimer disease [5] , autism [6] and psoriasis [7] . Previously , researchers developed several experimental methods to explore CNVs , such as fluorescence in situ hybridization ( FISH ) and array comparative genomic hybridization ( aCGH ) [8] , but the low resolution of these methods ( approximately 5~10 Mbp for FISH and 10~25 kbp for aCGH ) [9] presents a bottleneck for the detection of short CNVs [10] . In the last decade , Next Generation Sequencing ( NGS ) technology has enabled precise detection of CNVs , making it possible to identify small variants as short as 50 bp[11] . Many CNV detection algorithms were developed by NGS platforms . The Read Depth ( RD , or Read Count ( RC ) ) [12] and Pair-End Mapping ( PEM , or Read Pair ( RP ) ) [13] algorithms are the most popular methods for CNV detection , and they use statistical models and clustering approaches for CNV detection[14] , respectively . RD-based methods are good at detecting exact copy numbers , large insertions and CNVs in complex genomic region classes , whereas PEM-based methods can efficiently not only identify insertions and deletions but also locate mobile element insertions , inversions , and tandem duplications[14] . Many CNV detection methods have been developed based on the RD or PEM algorithms ( Table 1 ) . CNVnator is based on a statistical MSB model . It provides not only high sensitivity ( 86–96% ) and genotyping accuracy ( 93–95% ) but also a low false-discovery rate ( 3–20% ) [15] . ReadDepth is based on a statistical CBS model , and it can interpret overdispersed data for better breakpoint estimation[16] . Control-FREEC is one of the most widely used RD-based CNV detection software programs , and it uses matched case-control samples or GC content to correct copy number[17] . CNVrd2 computes segmentation scores by integrating the linear regression algorithm[18] into a Bayesian normal mixed model; thus , it has the highest paralog ratio[19] . cn . MOPS decomposes variations in the depth of coverage across multiple samples into integer copy numbers and noise by means of its mixture components and Poisson distributions[20] . RDXplorer is based on the Event-Wise Testing ( EWT ) algorithm , which is a method based on significance testing , and the median size of detected CNVs is much longer than that using PEM methods[9] . Canvas is a favored tool for both somatic and germline CNV detection in large-scale sequencing studies , and it implements all steps of the variant calling workflow[21] . GROM-RD is a control-free CNV algorithm combining excessive coverage masking , GC bias mean and variance normalization[22] . iCopyDAV is a modular-framework based on DoC approaches[23] . RSICNV detects CNVs using the robust segment identification algorithm with negative binomial transformations[24] . LUMPY integrates the CNV detection methods of RD and PEM and allows for more sensitive CNV discovery[25] . Previous studies have surveyed CNV detection software with regards to specificity , sensitivity and computational demands , and they have evaluated their advantages and shortcomings . For example , Fatima et al . evaluate CNV detection software based on analysis of whole-exome sequencing ( WES ) data[26] , and Junbo et al . evaluate six RD-based CNV detection software programs based on analysis of whole genome sequencing ( WGS ) data[27] . However , previous studies neither consider the impact of varied sequencing depth on the software performance nor use a standardized CNV dataset for evaluation based on analysis of real sequencing data . Our study not only adds several newer , untested software programs such as RSICNV , iCopyDAV and GROM-RD but also uses Database of Genomic Variants ( DGV ) as the gold standard so that our test results are more extensive and reliable[28] . Here , we surveyed ten frequently used methods of CNV detection in WGS data ( Table 1 ) , including CNVnator , ReadDepth , RDXplorer , LUMPY and Control-FREEC , and evaluated not only the detected CNV number , length distribution and result coincidence between different CNV methods but also the accuracy , sensitivity and computational demand under the conditions of different sequencing depths . Our study also compares the advantages and shortcomings of such CNV detection methods , providing useful information for researchers to select a suitable method .
The sequencing data ( 94x ) of the individual NA12878 were downloaded from the website of the 1000 Genomes Project[29] as evaluation data to compare the performance of CNV detection methods using real sequencing data . The DGV Gold Standard Variants for NA12878 were download from the Database of Genomic Variants ( DGV ) [28] , and a previously published SV benchmark of NA12878[30] was also fetched from the FTP site ( ftp://ftp . 1000genomes . ebi . ac . uk/vol1/ftp/phase3/data/NA12878/ ) [31] . After removing sequencing adapters and trimming consecutive low-quality bases from both the 5' and 3' end of the reads using an in-house Perl script , clean reads were subsampled by the sequencing depth of 5x , 20x , 10x , 30x , 40x and 50x using seqtk ( https://github . com/lh3/seqtk ) [32] . Then , the six datasets were mapped to the human reference genome ( hg19 ) using BWA ( V0 . 7 . 12 ) ( http://bio-bwa . sourceforge . net/ ) [33] with default parameters . The Picard program ( https://broadinstitute . github . io/picard/ ) [34] was used to sort mapping results to the BAM format . For CNV identification of NA12878 , ten methods were used with default or recommended parameters , including CNVnator , ReadDepth , RDXplorer , LUMPY and Control-FREEC . The CNVs with lengths of more than 1 kb were kept as detected CNVs . The main parameters for each software program used are listed in S1 Table . In the two datasets of the DGV Gold Standard Variants and the SV benchmark , the CNVs longer than 1 kb were merged by location overlap of more than 50% and were taken as the standard CNV dataset for performance evaluation ( S1 Table ) . The identified CNVs of each method were regarded as true positive results if there was more than 50% overlap on chromosome locations compared with the standard CNV dataset; otherwise , they were regarded as true negative CNVs . Then , the true positive rates ( TPRs ) and the false discovery rates ( FDRs ) were calculated and compared . The formulas to calculate TPR and FDR are shown in Table 2 . For computing time estimation , each application was run five times , and the average running times were recorded for the related standard deviation computation . To compare the memory usage of the applications , each application was run five times , and the average memory sizes were recorded for the related standard deviation[35–38] computation . The process used for performance evaluation is shown in S1 Fig .
With sequencing data with depths from 5X to 50X , ten methods were used to identify CNVs in NA12878 ( shown in Table 1 ) , and the tested CNVs were listed in the supplementary files ( S1–S11 Files ) . As shown in Fig 1a , due to different CNV detection algorithms , the numbers of detected CNVs varied greatly . CNVnator and RDXplorer identified the most CNVs , whereas Canvas and cn . MOPS identified the fewest . In most cases , the number of CNVs identified were positively correlated with the sequencing depth . However , RDXplorer detected the most SVs at 30X depth , probably because the method was tested and optimized at a 30X sequencing depth[9] . On the other hand , each software program tended to detect CNVs of different sizes , ranging from less than 1 kb to several hundred kbp . As shown in Fig 1b , most methods identified many small CNVs shorter than 10 kb , whereas LUMPY and ReadDepth predicted more CNVs longer than 200 kb . The detected CNVs for each method at a 30X sequencing depth were also compared in Fig 1c . Generally , CNVs identified by more than one method are more specific than those called by only one method[39] . As shown in Fig 1c , 98 . 27% of CNVs identified by Canvas were also identified by four other methods; the program with the next highest level of consistency with other methods was ReadDepth ( 87 . 00% ) , whereas CNVnator and RDXplorer identified the most CNVs that were only called in a single method . As shown in Fig 2a , the TPR curves of the ten methods were plotted at six sequencing depths from 5X to 50X . At a low sequencing depth of 5X , the TPR of LUMPY reached 0 . 432 , followed by CNVnator ( 0 . 370 ) and GROM-RD ( 0 . 359 ) , which was much greater than other methods ( 0 . 021 to 0 . 254 ) , implying that these three methods have greater sensitivity at low sequencing depth . At high sequencing depths of 30X and 50X , CNVnator also showed the highest TPR of 0 . 725 and 0 . 800 , followed by LUMPY ( 0 . 711 , 0 . 753 ) and RDXplorer ( 0 . 678 , 0 . 621 ) , implying higher sensitivity than other methods . Overall , at each sequencing depth from low to high , CNVnator and LUMPY had the best performance with respect to the sensitivity of CNV detection . At increasing sequencing depths , the trends of the TPR curves were different from one another . For CNVnator , LUMPY and ReadDepth , the range with varying TPR was much wider ( Fig 2a ) , and the TPR curve visibly increased , which indicates that the sensitivity of CNV detection is positively correlated with the sequencing depth . The TPR curve of RDXplorer also significantly increased with sequencing depth from 5X to 30X but reached a plateau at a 30X depth . This may result from the algorithm design as mentioned above . Considering the sensitivity of detecting CNVs and sequencing costs , a sequencing depth of 30X provides the best value for CNV detection , as is indicated by the trends in the TPR curves ( Fig 2a ) . However , the TPR curves were independent from sequencing depth for FREEC , cn . MOPS and Canvas ( Fig 2a ) . With regards to the specificity of CNV detection methods , the FDR curves of Canvas and LUMPY were lower than the others , implying that the specificities of these two methods are better than those of the other methods , i . e . , they predicted the least false positive results ( Fig 2b ) . The FDR value of iCopyDAV reached a peak value at a 30X depth ( 0 . 878 ) , followed by CNVnator ( 0 . 767 ) and RDXplorer ( 0 . 731 ) , but these three methods also predicted the most CNVs ( Fig 2b ) .
This study surveyed the performance of ten CNV detection applications with regards to sensitivity , specificity and computational demands over a range of sequencing depths . We found that most CNVs detected by Canvas and ReadDepth could be explored by other methods , but CNVnator and RDXplorer identified many specific CNVs ( Fig 1c ) . Of all the CNV detection methods , LUMPY showed the best performance in terms of both sensitivity and specificity , probably because LUMPY integrates two different algorithms of PEM and RD for CNV prediction[25] , and the PEM algorithm can provide better mapping accuracy on highly repetitive genomic regions than RD-based methods in some cases . Since TPR values for most methods were below 0 . 8 and the FDR values for most methods were above 0 . 3 ( Fig 2 ) , we believe that the sensitivity and specificity for CNV detection are not likely to be improved in the future . Limiting the CNV detection algorithms studied , our results are consistent with a previous report[39] . For all the ten methods , including RD-based algorithms , the read depth distribution is affected by the following three major causes . First , the GC-content in genomes leads to PCR bias during the construction of sequencing libraries , and the genome regions with ultrahigh or ultralow GC-contents are difficult to sequence , so the read depths on these regions are uneven . Second , because the genome sequencing was performed using short reads and it is difficult to correctly map short reads to genome regions with highly repetitive sequences , false positive CNV results arise in most studies . Lastly , the valuation results for cn . MOPS fall short of expectations . Since the cn . MOPS method was designed for input data from multisamples , the sensitivity and specificity are both very low when inputting single samples . The high FDR of CNV detection was also likely caused by the imperfectness of the standard CNV dataset . We also conducted the evaluation with another set of gold standard CNVs used in a previous study[40] , but the evaluation results were similar . A possible explanation is that it is difficult to identify all the CNVs on real experimental data , in spite of the fact that many platforms were used to confirm the detected CNVs on DGV Gold Variations . Therefore , the standard CNV dataset may not comprise all the true CNVs in NA12878 , and it may include some incorrect CNVs . For example , of all the CNVs in the standard CNV data set , 623 CNVs were not detected by any of the ten methods; these are most likely false positive detection results . The benchmarking above was based on single subsampling on each sequencing depth . To avoid subsampling bias , we evaluated the effect of subsampling on CNV prediction using multiple random subsampling . As shown in S2 Fig , we calculated TPR and FDR using five times subsampling for each CNV program on 30X depth ( S2a & S2b Fig ) , which is a typical depth for whole genome resequencing studies , and also subsampled five times on each depth for one program LUMPY ( S2c & S2d Fig ) . Most CNV prediction results of multiple subsampling are steady and the trends of TPR and FDR curves of each program were consistent with previous benchmarking conclusions ( Fig 2a & 2b ) . The aim of this survey is to help researchers choose appropriate CNV detection methods according to their specific purposes and the features of their data . We suggest that ( 1 ) when low FDR is preferable , LUMPY and Canvas are better choices ( Fig 2 ) ; ( 2 ) when high sensitivity is preferable , LUMPY , CNVnator and RDXplorer are better choices ( Fig 2 ) ; and ( 3 ) if the speed/computation demand is the first priority , CNVnator and ReadDepth should be considered ( Fig 3 ) . In this study , we employed the default or recommended parameters of each application for performance comparison . We plan to compare the best performance for each application by fine tuning the parameters and to include more recently published CNV applications in the future . Considering the limitations of sequencing data comprised of short reads , we are also preparing to evaluate CNV detection methods using long sequencing reads , such as PacBio or Oxford Nanopore , which may further improve the CNV prediction performance with regards to sensitivity and specificity . | As an important type of genomic structural variation , CNVs are associated with complex phenotypes because they change the number of copies of genes in cells , affecting coding sequences and playing an important role in the susceptibility or resistance to human diseases . To identify CNVs , several experimental methods have been developed , but their resolution is very low , and the detection of short CNVs presents a bottleneck . In recent years , the advancement of high-throughput sequencing techniques has made it possible to precisely detect CNVs , especially short ones . Many CNV detection applications were developed based on the availability of high-throughput sequencing data . Due to different CNV detection algorithms , the CNVs identified by different applications vary greatly . Therefore , it is necessary to help investigators choose suitable applications for CNV detection depending upon their objectives . For this reason , we not only compared ten commonly used CNV detection applications but also benchmarked the applications by sensitivity , specificity and computational demands . Our results show that the sequencing depth can strongly affect CNV detection . Among the ten applications benchmarked , LUMPY performs best for both high sensitivity and specificity for each sequencing depth . We also give recommended applications for specific purposes , for example , CNVnator and RDXplorer for high sensitivity and CNVnator and GROM-RD for low-depth sequencing data . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"sequencing",
"techniques",
"applied",
"mathematics",
"genomic",
"library",
"construction",
"simulation",
"and",
"modeling",
"algorithms",
"genome",
"sequencing",
"genomic",
"databases",
"mathematics",
"genome",
"analysis",
"copy",
"number",
"variation",
"molecular",
"bio... | 2019 | Comprehensively benchmarking applications for detecting copy number variation |
How do we use our memories of the past to guide decisions we've never had to make before ? Although extensive work describes how the brain learns to repeat rewarded actions , decisions can also be influenced by associations between stimuli or events not directly involving reward — such as when planning routes using a cognitive map or chess moves using predicted countermoves — and these sorts of associations are critical when deciding among novel options . This process is known as model-based decision making . While the learning of environmental relations that might support model-based decisions is well studied , and separately this sort of information has been inferred to impact decisions , there is little evidence concerning the full cycle by which such associations are acquired and drive choices . Of particular interest is whether decisions are directly supported by the same mnemonic systems characterized for relational learning more generally , or instead rely on other , specialized representations . Here , building on our previous work , which isolated dual representations underlying sequential predictive learning , we directly demonstrate that one such representation , encoded by the hippocampal memory system and adjacent cortical structures , supports goal-directed decisions . Using interleaved learning and decision tasks , we monitor predictive learning directly and also trace its influence on decisions for reward . We quantitatively compare the learning processes underlying multiple behavioral and fMRI observables using computational model fits . Across both tasks , a quantitatively consistent learning process explains reaction times , choices , and both expectation- and surprise-related neural activity . The same hippocampal and ventral stream regions engaged in anticipating stimuli during learning are also engaged in proportion to the difficulty of decisions . These results support a role for predictive associations learned by the hippocampal memory system to be recalled during choice formation .
Model-based decisions stand in contrast to a simpler sort of learned decision making whose neural instantiation is better understood: simply learning to repeat rewarded behaviors [4]–[6] . To explain the former , more knowledge-driven path to decisions , researchers have long argued that the brain maintains internal representations of the contingency structure of a task — a “world model” or , in spatial tasks , a “cognitive map” — that can be adaptively applied to drive behavior . Like a map of space , these representations describe the relationships between situations and actions , separate from any ties to reward . The reliance on these representations is a defining characteristic of goal-directed decisions [1] , [2] . Therefore , to identify the neural mechanisms of these decisions , researchers must first identify the representations that guide them . Here , to examine in detail the process by which contingency representations are learned and inform action choice , we combined a sequential learning task [7] with an interleaved decision task in which rewards depended on contingencies learned in the first task . In the learning task , participants were presented with one of four photograph images at a time , and asked simply to press the key corresponding to that image . Which of the four images appeared next depended , probabilistically , on the image currently being viewed . The sequential learning task allowed us to measure the gradual , trial-by-trial , acquisition of these probabilistic contingencies linking the four image stimuli . Participants' responses provided two observable measurements of learning: reaction time to identify each image , and image-specific BOLD activity in the ventral stream visual cortex . Reaction times to identify an image indicated the degree to which subjects expected it , given the previous one — a classic and relatively direct measure of the learned predictive association [8]–[12] — and category-specific BOLD also reflected engagement of the neural representation of each image in anticipation of its presentation [13] . By fitting computational models to this progression of subject expectations , we extracted a computational signature of the learning process , the learning rate , and used it to generate timeseries of decision variables based on these learned contingencies . This enabled us to quantitatively characterize the influence of these associations when participants were asked , in the interleaved decision probes , to draw on them to make decisions . Specifically , participants were told that one of the four images was , for a short period of time , to be associated with a reward . They were then asked which of two other images would lead to that rewarded image as quickly as possible . This manipulation has a form similar to a latent learning paradigm [14] , [15] , in which contingencies are learned separately from their link to reward . By requiring subjects to use knowledge of the contingencies to guide their decisions , this design allows us to probe how and whether the contingencies are used to seek trial-specific goals — contingencies that are exclusively the realm of model-based decision processes . Comparing the learning rates fit to behavior and BOLD responses we observed a striking match between hippocampal correlates of sequential learning and the learning underlying the reaction times , choices , prediction errors , and ventral visual stream activity , during both simple identification responses and deliberative decisions for reward . These results suggest that regions involved in sequential learning , including hippocampus and ventral cortical areas , indeed provide the necessary contingency representations to support model-based choice — and , critically , demonstrate the use of particular associations learned by these regions during model-based decision making .
We next identified neural correlates of each learning process .
We are able to compare learning across different task phases ( learning and choice ) and sorts of measurements ( reaction times , choices , and BOLD correlates of different quantities ) by treating them all as different windows on a computational learning process . We fit each sort of data with a standard computational model of how predictions are learned from recent experience , and compare the learning rate parameters that best explain these measurements . The pattern of data in Figures 3 and 5 and Table 1 shows a striking consistency in these estimated learning rates between the different measurements . However , there are a number of caveats to keep in mind about these analyses . First , it is in principle not possible to conclude that any two of these learning rate estimates are “the same” as one another — only that they are not statistically distingishable . But this pattern of negative findings is supported by positive ones , for instance that the differences between the various manifestations of “slow” and “fast” learning rates are significant ( Table 1 ) . Also , our findings that apart from exhibiting similar learning rates , neural activity during choice and decisions implicate common neural structures support the interpretation that all this activity relates to a common underlying learning process . Ultimately , however , establishing a definitive link between activity during learning and choice will require additional work using methods that can probe causal relationships between brain function and behavior . A related point is that the estimates of learning rates from BOLD in Figure 5 consistently tend to be less extreme than their behavioral counterparts , i . e . slightly slower relative to the fast learning rate and faster relative to slow . In a couple of cases , this difference between BOLD and behavioral estimates is significant , seeming to contradict the interpretation that all these measurements reflect a common learning process . We believe this relates to another important set of caveats with this study , which is that it is methodologically challenging to estimate learning rates from BOLD data due to the nonlinear relationship between the learning rate and the decision variables that have BOLD correlates ( entropy , etc . ) . To permit estimation , we approximate this relationship as linear using a first-order Taylor expansion [7] , [25] , [39] . This allows us to estimate the learning rate in the context of the same standard fMRI analysis ( using a general linear model ) as the rest of our results , and in turn means these analyses cope in the standard ways with the many methodological complications of fMRI ( including for instance intersubject random effects , temporal and spatial autocorrelation , hemodynamics , and regressor colinearity ) . This method appears to perform robustly in this and our previous study [7] and other closely related analyses of parametric brain-behavior relationships [38] , [40] , [41] , but there has not yet been a formal simulation study quantifying the error introduced by this approximation . One key sort of approximation error that we have examined [7] arises from our choice of the midpoint between fast and slow learning rates as the point around which to linearize . We choose this point to minimize the distance between the linearization point and the hypothetically relevant learning rates , since the error from linear extrapolation is expected to accumulate with distance . However , this choice interacts with the way we identify voxels of interest for fitting the learning rate , by identifying peaks in activity assuming this midpoint learning rate . Intuitively , this selection biases the estimated learning rates toward this midpoint ( see our previous study using this approach for a more thorough technical explanation [7] ) . Although this effect is innocuous with respect to the conclusions in this article , it may account for some of the observed difference between neural and behavioral estimates in Figure 5 . Our choice task has one of the key features of a latent learning task [15]: sequential contingency learning precedes the introduction of a new and unpracticed rewarding goal . In particular , given the sparse occurrence of the choice probes , and the different combinations of rewarded and starting images , these decisions implicate a model-based response strategy requiring participants to evaluate options' chances of reaching the new goal based on the predictive associations being continually learned in the sequential image presentation trials . Conversely , choices of this sort leave little room for model-free reinforcement learning based only on the success of particular choices at earning money in previous choice trials . Consistent with this , a key neural player in both the learning and decision phases in our results is the hippocampus . The hippocampal system is associated with flexible memory for stimulus-stimulus relations [42]–[44] and is a longstanding candidate for maintaining contingency structure in the service of goal-directed decisions [2] , [19] , [45]–[48] . In part , these suggestions are based on the analogy with spatial tasks , in which it has long been argued that the hippocampus implements a cognitive map [49] , [50] . A suggestive connection of these ideas to nonspatial tasks is ubiquitous findings that the the hippocampal system is implicated in acquired equivalence , transitive inference , and sensory preconditioning effects [41] , [51]–[53] , as well as the flexible use of conceptual [54] and structured [55] knowledge . All of these effects demonstrate a bias in novel choice probes caused by previously learned stimulus-stimulus relations . Model-based decision making relies on a similar ability to flexibly chain together or recombine associations in novel ways , as exercised in latent learning tasks like our choice probes here . Accordingly , we hypothesized that participants would draw on hippocampally-linked contingencies to make decisions . Indeed , the learning rates that best explained both choices and BOLD signals during the decision trials were not distinguishable from those seen in hippocampus and nearby ventral stream visual cortex during sequential responding , while differing significantly from those seen in BOLD activity in caudate and the fast process in reaction times . This quantitative convergence between learning processes examined during different tasks and through the lens of different observables substantiates the idea that model-based decisions and incidental stimulus-stimulus learning , like other sorts of relational learning and transfer [41] , [53]–[55] are supported by the same hippocampal memory system . Interestingly , the literature concerning these tasks suggests what appear to be two distinct ( but potentially complementary ) mechanisms supporting the flexible transfer of relational knowledge to novel probes . Some studies have demonstrated that better performance on transfer probes is predicted by hippocampal BOLD activity at learning but not test time [53] , [56] suggesting that transfer is somehow supported by processes that occur already during encoding . One hypothesis is that such activity reflects the immediate transfer of learning , when information is first obtained , to other related associates by a process of spreading activation . In other studies [54] , [55] , neural activity at probe time also related to correct performance or with the relational information itself . This suggests the importance of processes occurring at the time of retrieval , and is consistent with theories ( as in the standard account of model-based RL ) that transfer is supported by some sort of active inference , planning or search at the time of the novel choice . Our result ( discussed further below ) that hippocampal activity tracked the difficulty of the decision probes speaks to the latter mechanism , providing relatively direct evidence that the hippocampal system engages in more computation for harder transfer problems ( see also Simon & Daw [57] ) . Altogether , these two distinct but complementary mechanisms appear to be each well supported across the literature , and could plausibly both contribute in different circumstances . The type of model-based decision making studied here contrasts with “model-free” habit learning , of the sort associated with dorsolateral striatum [58] , predominant temporal-difference learning accounts of reward prediction error signal seen in dopamine neurons [6] , and the striatal BOLD response [29]–[31] . That said , parts of striatum are clearly necessary for model-based decision making in rodents as well [59] , [60] . Perhaps related , in human neuroimaging , even reward prediction errors observed in ventral striatum — though often characterized as reflecting the teaching signal for model-free stimulus-response learning — have recently been shown to report information about the state-state or relational structure of a task that would be known only to a model-based system [38] , [41] . This may suggest some crosstalk between model-based and model-free learning in the brain . The reward prediction errors in the decision phase of the present task are consistent with these results , in that they reflect stimulus-stimulus predictions combined with trial-specific rewards to which a purely model-free reinforcement learner would be blind . The present results also extend these findings by showing that the stimulus-stimulus learning rate driving these prediction error effects matches that from the hippocampal system during the sequential response task , suggesting all these are indeed driven by a common learning process . During the sequential response task , activity was not observed in the ventral striatal region commonly associated with reward prediction errors . This may reflect the lack of overt reinforcement in this more implicit association task . Instead , activity in a more dorsal/posterior region of striatum reflected a transient ( high learning rate ) adaptation process , which also had separate correlates in reaction times . We speculate that this activity ( and the associated component of the reaction times ) may reflect a second process of response learning , which did not carry over into the decision task . Indeed , the stimulus sequence in serial reaction time tasks of the sort we use is accompanied by an equivalent motor sequence ( of button presses ) , leading previous authors to suggest [61]–[63] that participants might learn either or both of two distinct types of sequential associations: stimulus-stimulus and response-response . That these processes then are uniquely tied to separate brain systems — hippocampus and striatum — suggests that they reflect learning of information specialized to each of those systems . Given the broader functional roles of both structures , it is tempting to hypothesize that hippocampus is associated with stimulus-stimulus associations and striatum with response-response [64]–[66] . While we did not explicitly dissociate response-response and stimulus-stimulus associations , the weight of the literature tying each of these types of information to each brain structure suggests this hypothesis and encourages us to carry it forward throughout the below discussion . Importantly , by asking participants to seek a particular stimulus given another , our decision probes isolate only stimulus-stimulus associations and cannot be solved on the basis of response-response associations . Thus , the finding that the hippocampal activity ( and its learning rate ) contributed to these choices , but not the striatal one , is consisistent with these structures' hypothesized involvement in stimulus and response prediction . Further , the exclusive use of the slow-process associations in forward-looking , model-based choice suggest that these associations are of a type that may be flexibly recombined , a property long associated with hippocampal representations and not those of striatum [48] , [52] , [67] . That this learning was ‘slow’ in the hippocampus may at first seem to run counter to the notion that this structure supports flexible , rapidly bound learning , as in episodic memory . Model-based decisions are also characterized similarly , for instance because they tend to dominate behavior during initial learning but not following overtraining . However , it is important to emphasize that the theoretical ‘flexibility’ of the model-based system is in its ability to recombine the learned associations , applying them in novel contexts to novel goals: it is fundamentally about what is learned ( e . g . , a world model rather than a fixed policy ) rather than how quickly . The question over what timescale any associations are learned is distinct from this issue – indeed , much previous work [57] , [68] implies that the learning rate should normatively be controlled by factors such as the volatility of the environment and the reliability of observations . In this context , the learning rate measures the degree to which the model-based system can draw on experiences learned from the far past , in applying them to these novel contexts . A low learning rate indicates a long memory; a higher learning rate indicates a shorter memory . The mechanisms which might give rise to these learning dynamics are an interesting topic for further research . Here , we have provided evidence that hippocampally-learned information is used in behavior via fetching memories of past transition events . That these candidate transition events might be drawn from memories stretching over tens of trials ( spanning under a minute ) into the past is well within understood capacity limitations of the hippocampal memory system . ( For a further treatment of these issues , see the discussion provided in our previous paper using this task [7] . ) In category-selective regions of the ventral visual cortex , we observed reinstatement of stimulus-stimulus associations in a manner that was modulated by task demands , across our two different tasks . Over the sequential response trials , we observed that BOLD activity correlated with stimulus expectations in category-selective regions of the ventral visual stream . Specifically , activity in face- ( or house- ) selective regions of extrastriate visual cortex were also preferentially modulated by the expectation that the face ( or house ) image would appear next . The finding that activity parametrically fluctuates with stimulus predictions in both hippocampus and the ventral visual areas — and that the learning rates explaining these effects match one another — provides evidence that both areas are participating in a common associative learning process . At a more mechanistic level , it may be possible to interpret both entropy-related activity in hippocampus and probability-related activity in the ventral visual areas in terms of associative spreading that activates the representations of likely successors to the currently observed image . On its face , the finding that anticipatory activity in the ventral areas decreases with conditional probability might seem to run counter to such a mechanism . That is , one might expect that , if probability is attributed largely to a single image , then the representation of that image should be more strongly activated . The contrary observation could be explained by a similar mechanism to the one that has been offered to explain ‘repetition suppression’ of BOLD ( and spiking ) responses [69] , [70] . Here , a more narrowly tuned population could be recruited for more strongly expected stimuli . However , this explanation is insufficient to explain the parallel anticipatory activation we observe during choice trials , which are presumably the result of a common mechanism for anticipatory retrieval in the service of behavior . A different interpretation of the effect is suggested by envisioning stimulus prediction as an active process of accessing memories . In particular , previously observed successors might be stochastically retrieved in a likelihood-weighted fashion to build up a statistical profile of the subsequent image , with this mnemonic evidence accumulated in a manner analogous to diffusion-to-bound models of perceptual discrimination [21] , [71] . This idea is consistent with suggestions that anticipatory activity in category regions is driven by evidence accumulation [72] . If such a process terminates when evidence reaches some threshold , then spiking activity would be elevated only over a shorter interval of time and , thus , on trials with strong evidence observed signal would be lower when integrated over the length of the hemodynamic response [73] . The activity of these same category-selective regions during the decision trials could be understood in a similar manner , in terms of retrieving memories to evaluate candidate actions . Here , activity in the face ( and house ) areas of ventral visual cortex correlated with our measure of the difficulty of deciding whether the choice of that stimulus would lead to reward . This observation supports a model where evaluation of decision options occurs by bounded accumulation of evidence — memories stochastically sampled to evaluate the likely consequences of a choice ( here , the successor image and its reward status ) . Our aggregate ( as opposed to stimulus-specific ) choice difficulty measure was also positively correlated with activity in the anterior MPFC and posterior cingulate cortex . Activations under our reporting threshold were also observed in dorsal MPFC and anterior and posterior hippocampus . These regions together comprise the fronto-temporal memory component of the well-known “default network” [28] . Although originally characterized by its increased , coherent , activity during periods of rest , a role in deliberative evaluation is consistent with functional hypotheses for this network , in which activity is modulated by prospective or constructive memory . Tying together experimental data from multiple levels of observation and across task and rest modalities , Buckner & Carroll [26] suggest the default network “enables mental exploration of alternative perspectives based on our past experiences” , a proposal they expanded on in later discussions [27] . Burgess [74] offers a complementary suggestion for one component of the network , proposing that BA10 in particular acts as a ‘gateway’ between a focus on internal ( e . g . , mnemonic ) and external ( e . g . , sensory ) representations . These proposals — along with observations of hippocampus and default network activity during look-ahead planning [75]–[77] — concord with our interpretation of the choice difficulty correlate as reflecting reinstatement of prior experiences . Finally , by offering a closer look at how the brain employs associations in the service of model-based decision making , our study suggests a route toward addressing one key puzzle in this area . To wit , whereas simple reward learning has a straightforward neural implementation ( embodied in model-free temporal difference theories and relatives [6] , [78] , [79] ) , and the inference that these be accompanied by model-based choice is well established [3] , the mechanism by which the brain actually implements such computations remains opaque . The idea we have advanced above , that successor states are retrieved stochastically ( see also [45] ) , and their values integrated , connects directly with known neural mechanisms . In particular , although the idea of model-based planning as a mnemonic version of evidence accumulation differs at least superficially from more abstract conceptualizations based on tree search [3] , [80] , [81] or Bayesian inference [82] , [83] , sampling from successor states provides a more realizable process-level account of model-based evaluation in circumstances ( such as chess ) when the full set of future trajectories is too large to explore systematically . Moreover , it connects closely with evidence accumulation mechanisms that are well studied in the context of perceptual decision making , and comports with other suggestions that sampling or diffusion models apply to value-based decisions as well [28] , [84]–[87] . It also joins those ideas with a literature suggesting that episodic memories can influence decisions [46] , [56] , [88] .
Twenty-four right-handed individuals ( twelve female; ages 18–40 years , mean 28 ) participated in the study . All had normal or corrected-to-normal vision . All participants received a fixed fee of $40 unrelated to performance , for their participation in the experiment , plus additional compensation of between $0 and $40 depending on their performance in one pseudorandomly-selected decision round . Participants were recruited from the New York University community as well as the surrounding area and gave informed consent in accordance with procedures approved by the New York University Committee on Activities Involving Human Subjects . Participants performed a serial reaction time ( SRT ) task in which they observed a sequence of image presentations and were instructed to respond using a pre-trained keypress assigned to that image . The experiment was controlled by a script written in Matlab ( Mathworks , Natick , MA , USA ) , using the Psychophysics Toolbox [90] . The stimulus set consisted of four grayscale images that were matched for size , contrast , and luminance . The images were chosen because they represent categories known to preferentially engage different areas of the ventral visual stream — bodies [32] , faces [33] , houses [34] , and household objects [35] . Each participant viewed the same four images . During behavioral training , the keys corresponded to the innermost fingers on the home keys of a standard USA-layout keyboard ( D , F , J , K ) . Participants were instructed to learn the responses as linking a finger and an image , rather than a key and an image ( e . g . left index finger , rather than ‘F’ ) . For the MRI sessions , the same fingers were used to respond on two MR-compatible button boxes . The mappings between the four images and four responses were one-to-one , pseudorandomly generated for each participant prior to their training session , trained to the criterion prior to the fMRI session , and fixed throughout the course of training and experiment sessions . Participants were informed that the key-to-image mapping was fixed , and that they were not being evaluated on the correctness of responses . At each trial , one of the pictures was presented in the center of the screen , where it remained for three seconds , plus or minus uniformly distributed pseudorandom jitter , up to 474 ms in increments of 59 ms ( the length of one slice in the MRI session ) . Participants were instructed to continue pressing keys until they responded correctly or ran out of time . Correct responses triggered a gray bounding box which appeared around the image for the lesser of 300 ms or the remaining trial time ( Figure 1 ) . Thus , each image presentation occurred for the programmed amount of time , regardless of participant response . The inter-trial interval consisted of 237 ms of blank screen . The test phase of the scanning session proceeded with three blocks of 250 trials: 210 sequential response trials , 20 reward display screens ( see Choice trials , below ) and 20 choice trials . The first two blocks were followed by a rest period of participant-controlled length . During the rest period , participants were presented with a screen that was blank except for a fixation cross . Scan blocks after the first were initiated manually by the operator only after the participant pressed any of the relevant keys twice , to alert the operator that they were prepared to continue the task . Total experiment time — inclusive of training , practice and test periods — was approximately 1 . 5 hours , conducted continuously . Our analysis proceeded in several steps meant to first characterize the associative learning process , and then use this characterization to test behavioral and neural predictions about choices . Each participant's trial-by-trial RTs for correct identifications were regressed on explanatory variables including the estimated conditional probability of the picture currently being viewed given its predecessor — defined , in separate models ( described below ) , in a number of different ways representing different accounts of learning — together with several effects of no interest . Trials on which the first keypress was not correct were excluded from behavioral analysis . Effects of no interest included stimulus-self transitions , image identity effects and a linear effect of trial number . Stimulus-self transitions were included to account for variance due to motor response readiness for the same keypress appearing twice in a row , above and beyond the preparation implied by any effect of the variables of interest . Image identity effects were included to account for any differential response time by each finger . Trial number effects were included to account for any monotonic shift in response time over the course of the experiment . These nuisance effects were identical across all models considered; the models differed in how they specified the explanatory variable of interest , the conditional probability of each image . In our initial analysis , the conditional probabilities were specified as the ground-truth contingencies: the probabilities actually encoded in the transition matrix . Having established that RT reflected such learning by demonstrating a significant correlation with these idealized probabilities ( Figure 2 ) , subsequent analyses used computational models to generate a timeseries of probability estimates such as would be produced by different learning rules with the same experience history as the participant ( see Learning models for details ) . Similarly , the learning rules for conditional probability were fit ( separately ) to choices in the decision trials , estimated so as to maximize the likelihood that the model would have selected the same options as did the participant , given the same series of experience ( see Choice models for details ) . The learning models involved additional free parameters controlling the learning and decision processes ( e . g . learning rates ) , which were jointly estimated together with the regression weights by maximum likelihood . For behavioral analysis , models were fit and parameters were estimated separately for each participant . At the group level , regression weights were tested for significance using a t-test on the individual estimates across participants [91] . To generate regressors for fMRI analysis ( below ) we refitted the behavioral model to estimate a single set of the parameters that optimized the RT and choice likelihoods aggregated over all participants ( i . e . treating the behavioral parameters as fixed effects ) . This approach allowed us to characterize baseline learning-related activity separate from individual variation in neurally implied learning rates relative to this common baseline . For the former , in our experience [22] , [25] , [38] , [92]–[95] , enforcing common model parameters provides a simple regularization that improves the reliability of population-level neural results . Our neural model characterizes between-subjects variation in the learning rate parameter over this baseline , because it includes ( as additional random effects across participants ) the partial derivatives of each of the regressors of interest with respect to the learning rate . | We are always learning regularities in the world around us: where things are , and in what order we might find them . Our knowledge of these contingencies can be relied upon if we later want to use them to make decisions . However , there is little agreement about the neurobiological mechanism by which learned contingencies are deployed for decision making . These are different kinds of decisions than simple habits , in which we take actions that have in the past given us reward . Neural mechanisms of habitual decisions are well-described by computational reinforcement learning approaches , but have not often been applied to ‘model-based’ decisions that depend on learned contingencies . In this article , we apply reinforcement learning to investigate model-based decisions . We tested participants on a serial reaction time task with changing sequential contingencies , and choice probes that depend on these contingencies . Fitting computational models to reaction times , we show that two sets of predictions drive simple response behavior , only one of which is used to make choices . Using fMRI , we observed learning and decision-related activity in hippocampal and ventral cortical areas that is computationally linked to the learned contingencies used to make choices . These results suggest a critical role for a hippocampal-cortical network in model-based decisions for reward . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Cortical and Hippocampal Correlates of Deliberation during Model-Based Decisions for Rewards in Humans |
Protein or DNA motifs are sequence regions which possess biological importance . These regions are often highly conserved among homologous sequences . The generation of multiple sequence alignments ( MSAs ) with a correct alignment of the conserved sequence motifs is still difficult to achieve , due to the fact that the contribution of these typically short fragments is overshadowed by the rest of the sequence . Here we extended the PRALINE multiple sequence alignment program with a novel motif-aware MSA algorithm in order to address this shortcoming . This method can incorporate explicit information about the presence of externally provided sequence motifs , which is then used in the dynamic programming step by boosting the amino acid substitution matrix towards the motif . The strength of the boost is controlled by a parameter , α . Using a benchmark set of alignments we confirm that a good compromise can be found that improves the matching of motif regions while not significantly reducing the overall alignment quality . By estimating α on an unrelated set of reference alignments we find there is indeed a strong conservation signal for motifs . A number of typical but difficult MSA use cases are explored to exemplify the problems in correctly aligning functional sequence motifs and how the motif-aware alignment method can be employed to alleviate these problems .
Sequence motifs are commonly described as relatively short conserved regions within a protein or DNA sequence [1] . These regions are of functional importance: they serve as binding sites for ligands or transcription factors , and as catalytic sites or structural elements . The presence of sequence motifs represents an additional conservation signal [2] , in addition to the conservation of the amino acid sequence . Accounting for motif conservation during sequence analysis is difficult . A conventional multiple sequence alignment ( MSA ) program will weigh each sequence position equally , scoring matches according to a substitution matrix such as BLOSUM [3] or PAM [4] . In a typical protein sequence , only a small fraction of amino acids are associated with a motif , which results in an underrepresentation of the conservation signal encoded by the motif . Cases even exist where traditional amino acid conservation is almost non-existent , such as with hypervariable regions . In these instances only the presence or absence of motifs is conserved . Fig 1 exemplifies an instance where hypervariability causes problems with an alignment of the HIV-1 envelope protein ( ENV , also known as gp120 ) [5] . Two sequence properties of this viral protein family are key to its function . Firstly , it contains several ‘variable’ regions that are hypermutated to avoid detection by the host’s immune system . Secondly , in vivo , gp120 is richly decorated with glycans , hence N-linked glycosylation motifs are abundant in the sequence . The alignment in Fig 1 is generated using the state-of-the-art Clustal Omega [6] program; one can appreciate in the overview at the bottom that in general it does a great job , certainly in the constant regions; one , C3 , is shown in detail as an example . ( Note that the full alignment contains over a hundred sequences , but here we show a representative subset for clarity . ) However , focusing on the variable regions ( marked in red ) with the glycosylation motifs ( in yellow ) , it is equally obvious that these are generally poorly aligned . The detailed illustration of V1 shows that many of the motifs in this region are not aligned ( Fig 1 top left ) . This is a typical case , which can be seen in many sequence data sets . In this gp120 study , the solution was to redress the misaligned regions by hand , which took the better part of two weeks to complete to satisfaction . We will return to the HIV ENV use case as an example in the results section . As is the case with HIV ENV , the presence of motifs in a sequence may provide important clues about the function of that particular protein . Column-wise analysis of motif positions in a MSA may reveal information about motif conservation , implying selective pressure , which in turn suggests a functional role . This also means it is possible to use motif conservation across species to filter out motifs occurring by chance , only considering the motifs that are likely biologically active [7] . Obtaining a more accurate alignment through the inclusion of motif information will further help many downstream analyses; e . g . mutation impact scoring , residue specificity prediction or phylogenetic analysis . We aim to tackle the motif alignment problem through our novel multiple sequence alignment strategy , Motif-Aware PRALINE ( MA-PRALINE ) . MA-PRALINE receives motif patterns in PROSITE pattern syntax , matches them against the input sequences and biases the substitution scoring towards giving the motifs greater significance . This means that MA-PRALINE is not a de novo motif identification method . Motif patterns with significant matches in an input should first be identified through other means; for example , by database searching or running a motif discovery program . The strength of the bias towards motif alignment is controlled by a parameter , α . Larger values of α result in a stronger bias towards motif alignment , whereas α = 0 is equivalent to normal sequence alignment . MA-PRALINE has been implemented on top of the existing multiple alignment program PRALINE [8] . PRALINE is a popular multiple alignment toolbox , with existing functionality to improve alignment quality by incorporating information about transmembrane regions ( TM-PRALINE ) [9] , homology ( PSI-PRALINE ) [10] and secondary structure [11] . Key to the motif-aware alignment algorithm is the support for multiple sequence tracks in PRALINE; these tracks can contain multiple sources of data for every sequence position . Other sequence data could thus be incorporated in a similar manner , such as information about membrane-spanning segments or secondary structure . Several related approaches to improve alignment quality have been attempted in the past . Db-Clustal [12] uses highly conserved fragments of sequences as anchor points to improve the quality of a multiple sequence alignment . COBALT [13] anchors the alignment using a consistent subset of constraints derived from domain information or from PROSITE [14] patterns . FMALIGN [15] allows the user to specify special conserved regions . These regions are then fixed in the resulting alignment; it is also possible to identify new conserved regions in an iterative manner . A key difference in the approach taken by MA-PRALINE , as opposed to these other methods , is the use of soft constraints . By assigning a score bonus , rather than restricting or anchoring the alignment , problems with false positives or spurious motifs can be mitigated more effectively . In this work , we first developed a motif-aware alignment method . Secondly , we show , through a benchmark , that there exists a range of α values where motif information optimizes the alignment of motif-rich regions , while not compromising the overall alignment quality . We further validate our method by deriving an estimate of the motif conservation signal on another data set of reference alignments . We find that these two , largely orthogonal , estimates of the permitted range of α are in agreement . Finally , we illustrate the advantages of using a motif-aware alignment strategy , by considering the nitrate reductase , HIV ENV , cupredoxin protein families , all three containing conserved functional motifs . Previously , we have shown that a similar approach is useful for aligning transcription factor binding motifs in DNA sequence regions [7] . In this work we show that a motif-aware approach can be equally useful for protein sequence motifs . To demonstrate MA-PRALINE in a practical context , we explore a number of real-world use cases . These include several difficult families from the alignment benchmark BAliBASE , as well as the HIV gp120 use case [5] introduced above . MA-PRALINE is available for Windows , Mac and Linux systems and , as open source software , can be found on GitHub at https://github . com/ibivu/MA-PRALINE .
Motif information is included in the pairwise alignment step of the algorithm . Multiple sequence alignment is performed using the progressive multiple alignment strategy on a pre-generated guide tree . The scores used to generate the tree dictating the join order are obtained by global pairwise alignment . The merging step of the progressive multiple alignment algorithm is also performed by global alignment of blocks . This strategy , while considerably less advanced than the standard PRALINE strategy ( creating a guide tree on the fly ) , was chosen because it performed well in our earlier work on motif alignment [7] . It should also be noted that MA-PRALINE can be configured to use the full suite of enhancements that regular PRALINE employs to increase the alignment quality . Information about the run time performance of MA-PRALINE is given in S1 Appendix . To generate the multiple sequence alignment a pairwise score matrix P is first obtained by aligning all pairs of input sequences . This matrix is then transformed into a dissimilarity matrix by shifting the values such that the highest scoring pair receives the value of zero and the lowest scoring pair receives the value of −min ( P ) + max ( P ) . D x y = D y x = - P x y + max ( P ) ( 1 ) Here Dxy is the dissimilarity between sequences x and y; Pxy is the alignment score between input sequences x and y . A guide tree is subsequently built from the full dissimilarity matrix D through hierarchical clustering . The linkage method is set to UPGMA by default , but single and complete linkage are also available as an option . Finally , a multiple sequence alignment is extended by pairwise alignment in the order given by the guide tree . The PRALINE alignment method [8] was adapted for this work to perform the motif-aware multiple sequence alignment , as further detailed in the next section . In the progressive multiple alignment strategy a multiple sequence alignment is grown iteratively by merging new sequences into it by pairwise alignment . In PRALINE these pairwise alignments are performed with a dynamic programming algorithm; both global ( Needleman-Wunsch [22] ) and semi-global ( PRALINE and others ) merge strategies are supported . The algorithm guarantees that the optimal alignment will be found . This optimality is defined as the most probable alignment according to the provided probability model . This model describes the likelihood of changing one kind of symbol ( such as a type of amino acid ) into another kind or into a gap . Inserting a gap means that a symbol in one of the two sequences has no corresponding symbol in the other . The algorithm works in two steps . In the first ( or forward ) step the solution matrix F is obtained by iteratively solving a recursive equation . In the second ( or traceback ) step a maximally scoring path is reconstructed from F . This path then corresponds to ( one of ) the most probable alignment ( s ) between the two sequences . In the global strategy the path is restricted to start in the bottom right corner of F and to end in the top left corner of F . In the semi-global strategy this requirement is relaxed: paths may start in the maximally scoring cell in the last row or column of F and may end in any cell of the first row or column . This has the effect of making gaps at the beginning and end of an alignment free , which in turn improves the quality of an alignment between sequences of strongly varying lengths . Information about the presence of one or more motifs is incorporated into this standard algorithm through the addition of a scoring term to the general recursive equation used to calculate the dynamic programming matrix F . Alternatively , one may think of this as extending the alphabet of the sequence ( and the substitution matrix ) to account for the possible presence of a motif [7] . The formulation of the modified recurrence relation is given below . Note that this is a general definition; MA-PRALINE additionally implements an optimized version that improves the execution time at the cost of restricting the gap penalty function g ( l ) to be linear ( g ( l ) = dl ) or affine ( g ( l ) = e + d ( l − 1 ) ) . F i j = max { F k , j + g ( i - k ) for k = 0 , . . . , i - 1 F i , k + g ( j - k ) for k = 0 , . . . , j - 1 F i - 1 , j - 1 + S ( A i , B j ) + S motif ( A i , B j ) ( 2 ) Fij is the value of the dynamic programming matrix F in row i and column j , g ( l ) is the gap penalty associated with a gap of length l , and S ( Ai , Bj ) is the match score between the symbols at position i and j in sequences A and B . Smotif ( Ai , Bj ) is the motif scoring term , defined as follows . S motif ( x , y ) = { α if both x and y are part of the same motif 0 otherwise ( 3 ) α is a parameter that controls the strength of the bias towards motif alignment . If α is set to a large value , the algorithm will have a very strong preference to align motifs over the maximization of the traditional amino acid substitution score . In the case of α = 0 the behavior reverts to that of the conventional dynamic programming alignment algorithm . An example showing the influence of α on an alignment is shown in Fig 2 . Protein motif annotations are provided to the program in the form of PROSITE regular expression pattern definitions . Patterns in this format can be found in a number of places , such as in PROSITE [14 , 23 , 24] itself and in the Eukaryotic Linear Motif ( ELM ) [25] database . If the structural family of a set of proteins is known , it could be possible to include structural motifs from a database like SMoS [26] . Finally , because the PROSITE pattern format is simple and widely used , it is also possible for an end user to encode a previously undocumented pattern by hand . PROSITE pattern syntax is an example of a regular expression ( or regex in short ) . Regular expressions are a compact way to encode a sequence pattern in a manner that allows for efficient searches through large biological databases . For example , the N-linked glycosylation motif introduced earlier is encoded by the PROSITE pattern N-{P}-[ST]-{P} . This pattern matches any subsequence of amino acids starting with an asparagine ( N ) , followed by any amino acid but a proline ( P ) , followed by either a serine ( S ) or a threonine ( T ) and terminated by any amino acid but a proline ( P ) . The motifs are used to annotate matching regions of input sequences . Motif patterns , however , often contain information about the spacing between matching symbols . For example , the pattern N-x ( 8 ) -N can be read as “match any sequence consisting of two asparagines separated by 8 amino acids of any type” . Here , only the two asparagine positions can reasonably be considered part of the motif; the other positions merely offer information about spacing . We thus only consider a position as part of a motif if it matches against an informative rule in a pattern . A rule in a pattern is considered informative if it excludes more types of amino acids from matching than it includes . For example , the rule “match any amino acid that is not proline” is uninformative because it only excludes one type from matching . However , the rule “match either alanine or tryptophan” is informative because it excludes all but two types . A typical PROSITE motif pattern is described in Fig 3 , together with an example sequence matched against it . If the above set of rules erroneously considers a motif position as uninformative , MA-PRALINE can be configured to treat all motif matches as informative . This may especially be desirable in the case of motif patterns which are too short to contain information about spacing . More information regarding the different ways in which motif information can be provided to MA-PRALINE is given in the Supporting Information , S2 Appendix . In order to obtain an orthogonal estimate of allowed values for the motif score parameter α , we apply a knowledge-based approach on a reference data set . The methodology used is similar to the way BLOSUM matrices [3] are determined . HOMSTRAD was chosen as the reference because it is strictly based on structural alignments . In the other reference sets ( such as BAliBASE ) , expert manual adjustments to the sequence alignment may have introduced a bias towards aligning known motifs . To construct our input set , all reference alignments in HOMSTRAD are annotated with all of the motif patterns present in PROSITE . This yields gapped alignments with both an amino acid sequence and one or more motif annotations . Reference alignments without a single motif match are excluded from the input set . Using this procedure , we find 34568 motif matches in 3102 sequences , over a total of 974 HOMSTRAD alignments . The minimum , maximum and mean amount of matches per alignment are 0 , 619 and 35 . 49 , respectively . Per sequence there is a minimum number of matches of 0 , a maximum of 69 and a mean of 11 . 14 . Additional statistics regarding motif redundancy in HOMSTRAD are given in S3 Appendix . We define α* as the logarithm of the probability of observing the alignment of a motif annotated residue , divided by the expected ( or naive ) probability of such an event . In other words , α* expresses how many times more likely the alignment of motifs is than would be expected purely by chance . α* is given by the logarithm of the ratio between the observed motif alignment probability P ( Omotif ) and the expected , or background , probability P ( Emotif ) . α * = f log b ( P ( O motif ) P ( E motif ) ) ( 4 ) The scaling factor f = 2 and logarithm base b = 2 are chosen so that the parameter scale is equivalent to that of standard BLOSUM matrices [3] , and to that of α . This allows for direct comparison between values of α and α* . In order to obtain P ( Omotif ) we count , for every column in an alignment , the number of symbol pairs in which both symbols are a match symbol . The probability then becomes this frequency divided by the total number of pairs over all columns in the same alignment . If an alignment contains gaps , then these are treated as a non-match symbol . The expected probability P ( Emotif ) is estimated by the square of the fraction of positions containing match symbols ( qm2 ) . This estimate is reasonable as long as motifs are distributed evenly across the different sequences in an alignment . The quality of an MSA is measured in terms of the Sum-of-Pairs score ( SP score ) [27–29] versus a reference alignment . The motif score is defined as the number of motif annotated symbol pairs in an alignment column , divided by the total amount of pairs . The motif score is counted over all columns in an alignment containing at least a single motif match . If multiple patterns are included , their matches are collapsed into a single motif annotation . Consequently , if a certain sequence position matches multiple patterns the pairs are only contributing to the motif score once . BAliBASE [30] contains a large number of manually curated reference alignments for the purpose of benchmarking MSA programs; for an in-depth discussion of the difference between the sets we refer to earlier work [29] . The BAliBASE reference alignments and program outputs were used as-is; no changes were made by hand . All families with significantly matching motifs in the most recent release of BAliBASE 4 were used to assess motif and overall alignment quality for various MA-PRALINE α values—please see S3 Table . for a full list of families and the motif patterns that were annotated in them . We will discuss two families from BAliBASE 3 , BB20035 and BB30015 , and one HIV envelope protein family , in more detail . These represent typical examples of difficult to align sequence families with known functional motifs . For the alignments in the results section we use alignment parameters that one would typically use for these specific families , as described below . Note that non-default parameter choices—the substitution matrix , semiglobal merging and the use of preprofiles—do not affect the quality of the aligned motif regions . A full overview of the parameters is given in S4 Appendix . For the two BAliBASE use cases the additional alignment settings that were used to generate the MA-PRALINE alignment are: semiglobal alignment merging , because the sequences vary greatly in size; the BLOSUM40 and BLOSUM30 substitution matrices , because a lower-than average sequence conservation is expected in these families; and finally global preprofiles , because this is a commonly enabled quality improvement , which only incurs a mild penalty in terms of run time .
In order to test performance and the effect of α on the overall alignment quality , we ran MA-PRALINE over a range of α values on alignments from BAliBASE 4 containing preserved motifs . Preserved motifs were identified by matching sequences against all of the motif patterns documented in PROSITE . A motif is considered to be preserved if at least 50% of the sequences in a BAliBASE alignment contain at least one instance of it . All of the 218 alignments in BAliBASE 4 have at least one motif that meets this threshold . In order to keep the computational run time feasible , 22 very large alignments were excluded from the benchmark set . For a full list of BAliBASE 4 alignments and the motif patterns that were used per alignment , please see S3 Table . Table 1 shows the average SP and motif benchmark scores of alignments generated by MA-PRALINE using a range of α values . These results show that , for small values of α , the overall SP score remains stable while the motif score increases , indicating an improvement in the quality of the motif alignment . If α is increased past a limit of around 15 , however , the overall quality starts to suffer . These findings show that it is possible to improve the quality of the motif alignment without strongly affecting the overall alignment quality . According to these results , α should be set between 5 and 20 , depending on the degree of motif conservation . The standard deviation across different BAliBASE alignments also grows for both the SP score and the motif score; this may indicate that the inclusion of α is changing the order in which sequences are added to the alignment by the progressive alignment algorithm . Changes in join order may result in large changes to an alignment , since progressive multiple sequence alignment algorithms are very sensitive to the order in which sequences are added to the alignment . Given the viable range of α found by benchmarking MA-PRALINE ( 5 < α < 20 ) , we want to see whether this corresponds to the motif conservation signal we observe in reference alignments . Statistically derived through a similar methodology as BLOSUM matrices [3] , we obtain an estimate of the motif conservation signal , α* . The derivation of α* was calculated over the HOMSTRAD data set [37] , because it contains fewer biases than BAliBASE . Fig 4 shows the value of α* for every motif pattern found in PROSITE with at least one match in HOMSTRAD . qm is the fraction of amino acids which match a motif pattern versus the total number of amino acids in all sequence sets with at least a single match of the same pattern . Almost all values of α* fall within a decaying envelope imposed by qm; this is because α* is a measure of ( logfold ) additional conservation over the background probability . The scenario of perfect conservation , where every occurrence of a motif is aligned exclusively against other occurrences of the same motif , translates to an upper bound on α* . The bound is dependent on qm because qm2 is used as the estimate of the background probability P ( E ) . This is roughly analogous to the reason why rare amino acid types generally receive higher scores in a BLOSUM matrix . The data points which do fall outside of the envelope correspond to perfectly conserved motifs in HOMSTRAD; additionally , these patterns match multiple families of strongly varying sizes . In such cases qm2 underestimates the background probability , which in turn allows the α* value of these patterns to transgress the boundary . A list of all PROSITE patterns and associated qm , α* values is available in S4 Table . Most of the longer motifs are of higher complexity , and thus have a lower chance of matching a subsequence spuriously . For these motifs it would be possible to choose a suitable value of α using qm alone , which is known to the user in advance ( after annotation ) . For the shorter ( or lower complexity ) motifs this does not seem to be true , however , with an observed range of α* between 6 and 10 . Nonetheless , these results show a range of α* values largely in accordance with the acceptable range of α , independently obtained through the benchmark and the use cases . A value of α around 10 appears to be a reasonable setting , giving good alignment quality ( as shown in Table 1 ) as well as being in accordance with the observed motif conservation signal for sequence evolution ( as shown in Fig 4 ) . The sequences of the HIV-1 envelope glycoprotein ( ENV ) contain many occurrences of the N-linked glycosylation motif ( pattern N-{P}-[ST]-{P} with PROSITE identifier PS00001 ) . Generating a globally acceptable alignment is not especially difficult as , outside of the variable regions , there is strong sequence conservation . The alignments produced by various alignment methods give very similar SP scores ( S2 Table ) , indicating high alignment similarity . It is when a high-quality alignment for the variable regions is also required that one will run into difficulties; this is shown for a number of commonly used alignment programs in the Supporting Information ( S8 , S9 and S10 Figs ) . In order to study the mechanism by which HIV evades an immune response , it is crucial to have a proper alignment of the glycosylation motifs , many of which are found in the variable regions [5] . Fig 5 shows how MA-PRALINE allows a user to optimize such alignments in various scenarios . When the full protein sequences are aligned , most MSA methods struggle to align the variable regions correctly , as shown in Fig 1 ( for Clustal Omega ) and in Fig 5 , panel A ( for regular PRALINE ) . With extra weight on the glycosylation motifs , MA-PRALINE is able to generate a reasonable alignment for these regions , as shown in Fig 5 , panel B . If we only use the sequence region comprising V1 as input instead of the full sequence ( and cutting V1 out afterwards ) , MA-PRALINE obtains even better results , as shown in Fig 5 , panel D . Regular PRALINE still has difficulties ( Fig 5 , panel C ) .
In this study we explored a number of real-life multiple sequence alignment use cases to demonstrate how MA-PRALINE could be used to improve the biological fidelity of a sequence alignment by aligning sequence motifs more accurately . Motif alignment quality depends strongly on the value of the motif match score parameter ( α ) . We also conclude that the correct value of α is dependent on the expected conservation of the motifs . Care should be taken not to harm the overall alignment quality by setting α to an unrealistically high value . A safe , default setting is α = 10 , but depending on the sequence identity , the length of the motifs and possible motif redundancy , it may be possible to set α to a larger value . This may even be required to gain the full benefits of MA-PRALINE motif scoring . However , values larger than α = 30 lead to poor results . This α seems to indicate a threshold beyond which the motif boosting becomes strong enough to overcome the traditional substitution scoring for non-conserved motifs . We recommend trying out a few possible values within the range given above and to look at how the resulting alignment changes at each increment of α . | The most important functional parts of proteins are often small—but very specific—sequence motifs . Moreover , these motifs tend to be strongly conserved during evolution due to their functional role . Nevertheless , when trying to align protein sequences of the same family , it is often very difficult to align such motifs using standard multiple sequence alignment methods . Aligning functional residues correctly is essential to detect motif conservation , which can be used to filter out spuriously occurring motifs . Additionally , many downstream analyses , such as phylogenetics , are strongly reliant on alignment quality . We have developed a sequence alignment program named Motif-Aware PRALINE ( MA-PRALINE ) that incorporates information about motifs explicitly . Motifs are provided to MA-PRALINE in the PROSITE pattern syntax; it then scans the input sequences for instances of the pattern and provides a score bonus to matching sequence positions . Our method provides a reproducible alternative to editing alignments by hand in order to account for motif conservation , which is a tedious and error-prone process . We will show that MA-PRALINE allows the alignment of motif-rich regions to be fine-tuned while not degrading the rest of the alignment . MA-PRALINE is available on GitHub as open source software; this allows it to be easily tailored to similar problems . We apply MA-PRALINE on the HIV-1 envelope glycoprotein ( gp120 ) to get an improved alignment of the N-terminal glycosylation motifs . The presence of these motifs is essential for the virus in evading the immune response of the host . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"split-decomposition",
"method",
"pathogens",
"microbiology",
"retroviruses",
"viruses",
"immunodeficiency",
"viruses",
"multiple",
"alignment",
"calculation",
"rna",
"viruses",
"protein"... | 2018 | Motif-Aware PRALINE: Improving the alignment of motif regions |
In the event of a new infectious disease outbreak , mathematical and simulation models are commonly used to inform policy by evaluating which control strategies will minimize the impact of the epidemic . In the early stages of such outbreaks , substantial parameter uncertainty may limit the ability of models to provide accurate predictions , and policymakers do not have the luxury of waiting for data to alleviate this state of uncertainty . For policymakers , however , it is the selection of the optimal control intervention in the face of uncertainty , rather than accuracy of model predictions , that is the measure of success that counts . We simulate the process of real-time decision-making by fitting an epidemic model to observed , spatially-explicit , infection data at weekly intervals throughout two historical outbreaks of foot-and-mouth disease , UK in 2001 and Miyazaki , Japan in 2010 , and compare forward simulations of the impact of switching to an alternative control intervention at the time point in question . These are compared to policy recommendations generated in hindsight using data from the entire outbreak , thereby comparing the best we could have done at the time with the best we could have done in retrospect . Our results show that the control policy that would have been chosen using all the data is also identified from an early stage in an outbreak using only the available data , despite high variability in projections of epidemic size . Critically , we find that it is an improved understanding of the locations of infected farms , rather than improved estimates of transmission parameters , that drives improved prediction of the relative performance of control interventions . However , the ability to estimate undetected infectious premises is a function of uncertainty in the transmission parameters . Here , we demonstrate the need for both real-time model fitting and generating projections to evaluate alternative control interventions throughout an outbreak . Our results highlight the use of using models at outbreak onset to inform policy and the importance of state-dependent interventions that adapt in response to additional information throughout an outbreak .
The responsibilities of policymakers during infectious disease outbreaks include the difficult task of choosing between multiple control interventions based on an uncertain future . Mathematical and simulation models have proved a useful tool to aid decision-making during disease outbreaks by both generating forecasts of outbreak severity and comparing different control strategies [1–6] . Using mathematical and simulation models , however , requires estimation of model parameters , and in the early stages of an outbreak , when decisions are most critical and data are most scarce , significant parametric uncertainty may limit the ability of models to provide informed advice . Recent research has outlined frameworks that combine Bayesian parameter estimation and a mathematical model for generating real-time forecasts of outbreak severity throughout the course of an epidemic , aiming to assimilate available surveillance data into model estimates as rapidly as possible [7–10] . The efficacy of these frameworks has typically been evaluated using the accuracy of forecasts . For decision-makers , however , the accuracy of model forecasts per se , are not the best measure of success , it is the selection of the optimal control interventions in the face of uncertainty that counts , and ultimately what they are judged on . However , projections of the impact of alternative control interventions are not always performed alongside projections of the burden of infection . For several infectious diseases it is not possible , or relevant , to include projections of control interventions , for instance when interventions are directly related to estimates of the burden of disease or when control interventions are related to the timing of the peak of an epidemic ( such as in influenza ) . Including projections of interventions does allow disease control problems to be phrased as optimization problems and therefore allows the determination of whether optimal control choice is dependent upon the underlying state of the outbreak . In contrast to real-time forecasting approaches , we demonstrate real-time decision-making , which requires integration of all information until now plus the potential future impact of candidate control interventions . In this work , we simulated the process of real-time decision-making by fitting a dynamic epidemic model to the observed ( confirmed ) , herd-level , infection data at weekly intervals throughout two historical outbreaks of foot-and-mouth disease ( FMD ) , a viral disease of economically important livestock , and compared forward simulations of the impact of alternative culling and vaccination interventions under a management objective of minimizing total culls so as to gain disease-freedom . We repeated these forward simulations at each time point using parameter estimates from a model fitted to the complete outbreaks , thereby comparing the best we could have done at the time with the best we could have done in retrospect . Forward simulations predicted the final total culls ( number of animals culled ) from having taken a single control intervention from a particular date into the future . Control interventions included culling of infected premises only ( IP ) , culling of infected premises and dangerous contacts ( IPDC ) , culling of infected premises , dangerous contacts and contiguous premises ( IPDCCP ) , ring culling in areas surrounding infected premises at 3 and 10 km radii ( RC3 and RC10 respectively ) , and vaccination in areas surrounding infected premises at 3 and 10km radii ( V3 and V10 respectively ) ( see Materials and Methods for further details of the control interventions ) . This set of intervention strategies includes those that governments have implemented , or considered , in the past and that are consistent with other studies on foot-and-mouth disease . In contrast to previous work in real-time forecasting [7–10] , we used an individual-based model and included uncertainty regarding the location of infected and undetected farms [11 , 12] . As surveillance data become available , we improve our understanding , in a cumulative manner , of both how the outbreak is unfolding , via estimates of transmission parameters , and of where likely new infections may be located , via estimates of the spatial distribution of infected but undetected farms . Forward simulations are therefore conditional upon both the actual state of the outbreak ( i . e . the pattern of confirmed infected cases and the pattern of inferred , undetected infections ) and having enacted the actual outbreak controls until the time point in question . The outbreaks of FMD in the UK in 2001 and Miyazaki , Japan in 2010 contrast in a number of ways and thus provide a valuable comparison for investigating how policy recommendations are affected by additional information throughout an outbreak . Firstly , the UK outbreak had over 2000 infected premises over an area that included England , Wales , and Scotland ( roughly 230000km2 ) [13] whereas the outbreak in Miyazaki affected only 290 premises contained within the Miyazaki prefecture ( less than 8000km2 ) [14 , 15] . Secondly , at the time of confirmation of FMD , there were multiple foci of infection dispersed throughout the UK but only one focus of infection in the Miyazaki outbreak ( although additional foci occurred later ) . Finally , control interventions deployed during each outbreak were different; UK control interventions only included culling strategies whereas control of the outbreak in Miyazaki began with culling and shifted to vaccination after 5 weeks , necessitated by constraints on the disposal of accumulating carcasses [16] .
In the early stages of both the UK and Miyazaki outbreaks , estimates of the instantaneous risk of onward transmission are highly uncertain ( Figs 1 , S3 and S4 ) . Our analysis highlights that uncertainty reduces through time in several regards . Firstly , estimates of transmission parameters change through time ( Figs 1 , S3 and S4 ) . Secondly , our understanding of how transmission parameters relate to one another change through time ( S5 and S6 Figs ) . Finally , our confidence in the locations of premises we believe to be infected also changes through time ( Figs 1 , S7 and S8 ) . Given that the relationships between parameters change through time , marginal distributions of parameters ( S3 Fig ) do not tell the whole story , and we therefore summarize how the shape of the multidimensional posterior distribution evolves through time via a measure of instantaneous risk of onwards transmission . Note that the instantaneous risk of transmission indicates the overall relative risk of transmission , which does not have a direct epidemiological interpretation but provides a direct comparison across weeks . Projections of the final total number of animals culled that were made in the first three weeks of each outbreak were highly uncertain , grossly overestimating the size of the outbreak in the UK and , on average , tending to underestimate the size of the outbreak in Miyazaki ( Figs 2 and 3 ) . In the UK , the mean outbreak size for vaccination strategies was estimated to be ten times larger when using one week’s worth of data compared to when using all of the data ( Fig 2; accrued vs complete , week 1 ) . In Miyazaki , bimodal distributions of outbreak size were generated by forward simulations started in the initial weeks as the spatial extent of the outbreak was highly uncertain ( Fig 3 ) . This was particularly marked under strategies of non-ring control culling ( IP and IPDC ) . However , in both outbreaks , despite projections being highly variable , the relative performance of control interventions was resolved early on ( Figs 2B and 3B ) . The projected rankings of control interventions between those based on available data and those based on all outbreak data were identical at five weeks into the UK epidemic ( S9 and S10 Figs ) , and differences in the rankings of control interventions after week 5 were minor . Using only available data , the top three ranked control strategies ( vaccination at 3km , 10km , and IPDCCP culling , respectively ) do not change from weeks 5–8 , at which point vaccinating at 10km becomes optimal over vaccinating at 3km in week 9 . Vaccination interventions ( either at 3km or 10km ) are always ranked as optimal throughout the whole UK outbreak regardless of the week at which projections were made and regardless of the amount of data available . If vaccination was deemed politically unpalatable then culling of infected premises , dangerous contacts , and contiguous premises ( IPDCCP ) was consistently ranked as the next best intervention to minimize total culls from weeks 4–10 of the outbreak in the UK ( regardless of whether available or all data were used to estimate transmission parameters ) . Although the rankings of control interventions did change in the Miyazaki outbreak , the relative distribution of expected outbreak sizes across different control interventions remained consistent , with 3km and 10km vaccination strategies consistently ranked as optimal ( Figs 3 and S11 ) and consistently performing better than the other strategies when compared in repeated bootstrap simulations ( Figs 3C and S11C ) . Ring culling at 3km was ranked as optimal for weeks 1–5 if vaccination interventions were not considered , regardless of the amount of data used to estimate transmission parameters . IP culling was optimal in the final stages of both outbreaks when there were few or no more infected premises ( S1 , S2 and S10 Figs ) . The consistency in the relative rankings of control interventions lends support for trusting model comparisons of interventions at outbreak onset . For the UK outbreak on week 2 ( 5 March 2001 ) , when only using available data , projections of the magnitude of the four best performing control interventions occupied comparable overlapping ranges: V3 , V10 , IPDC and IPDCCP ( accrued information; Fig 2 ) . Such a set of projections may provide a strong case in support of only enacting IPDC , the minimum under EU and Japanese law , as this avoids the latent underlying costs and consequences of culling contiguous premises or emergency vaccination , for only a small increase in the expected number of total animals culled . However , had all the data for the outbreak been available at this date ( complete information; Fig 2 ) , then the expected performance of IPDC was revealed to be very poor compared to these other three control interventions , with the results supporting either vaccination ( V3 or V10 ) or culling in some proximity around infected farms ( RC3 or IPDCCP ) to quickly halt the outbreak . Here , despite variability in the absolute value of projections , our initial rankings and estimates of the relative performance of controls were again found to be robust . In the outbreak in Miyazaki the three best control interventions were correctly identified by week 2 using only the available data despite highly uncertain projections of total animals culled ( Figs 3 and S11 ) . Relative performance of control interventions was largely unchanged after this point , with the exception of the emergence of IP culling as optimal in the final weeks due to there only having been 1 infected case on 4th July , before which there had not been a case since 18 June 2010 [14] . Consistency in the performance of vaccination strategies is highlighted by noting that greater than 50% of the time vaccination actions are optimal when bootstrap samples are compared across the distributions of simulated total culls for weeks 2–5 ( Figs 3C and S11C ) . Bimodal distributions in outbreak size did reappear later in the outbreak between weeks 6–7 ( 1–8 June ) coincident with the occurrence of additional foci of infection ( S11 Fig ) , although these additional outbreaks were effectively controlled and policy recommendations were unchanged . This is in contrast to the UK 2001 outbreak where , at the time of the first case being reported , there were already infected farms in multiple foci spread across the country with no significant new emerging foci as the epidemic progressed that would give rise to the bimodal predictions that we observe for Miyazaki . Overall , outbreak size and optimal control choice for the Miyazaki outbreak were generally more straightforward to predict than for the UK outbreak , and policy recommendations for minimizing outbreak size are in line with what was actually implemented in 2010 ( vaccination at a 10km radius , V10 ) . Changes in policy recommendations and large uncertainty in model projections may be caused by uncertainty associated with the parameters governing the dynamics of the outbreak and/or uncertainty associated with the locations of infected farms . By looking at projections made using parameters generated with complete outbreak data , isolating the effect of changing the arrangement of infected farms ( since transmission parameters are fixed in these simulations ) , we see that policy recommendations are still changing . Between 5 and 12 March 2001 in the UK outbreak ( weeks 2–3 ) , there was a marked resolution in the performance of control interventions when parameters were estimated using only available data ( accrued information ) . The relative performance of control interventions at this time point comes to resemble analogous projections made using all outbreak data ( Figs 2 , S9 and S10 ) despite the absolute value of such projections still being highly uncertain . Here , this switch may have been caused by a constrained estimate of where the outbreak is–an improved estimate of the location of undetected infected premises ( S7 Fig ) . Our best estimate of the location of undetected infections on 5 March 2001 ( week 2 ) includes undetected infections in the north of Scotland . Additionally , the number of counties with a non-zero expected number of undetected infections is overestimated compared to analogous estimates if all the data were available ( S7 Fig ) . Our best estimate of undetected infections on 12 March 2001 ( week 3 ) had foci of infection in Cumbria , Midlands , Devon , North Wales , and Essex; these foci were in the same areas as estimates using all outbreak data . Furthermore , those counties that have a non-zero expected number of undetected infections are contiguous to counties that are estimated as a foci of infection ( S7 Fig ) . At this point , by excluding an allowance for additional culling in Scotland , ring culling at 3km became second only to vaccination actions . Ultimately , the change in our understanding of the locations of infected farms ( and having the resource capability to respond ) , drove this improved prediction of the relative performance of control interventions . Such a dramatic change in the relative performance of control interventions was not seen in the Miyazaki outbreak , despite starting with high variability in projections of the efficacy of control interventions . The Miyazaki outbreak started and largely stayed in one location , the Tsuno township . Though bimodal distributions in outbreak size did arise in simulations ( S11 Fig ) as additional foci of infection occurred later at the Ebino township , Miyakonojo , and Miyazaki city ( all over 40km away from the original point of infection [17] ) , the projections were largely seeded by an outbreak with a state characterized as a single point of infection , with compact and radial spread of infection ( Figs 3 , S8 and S11 ) . Accounting for the possibility of a larger outbreak size in Miyazaki did not stretch the simulated culling or vaccination resources , leaving policy recommendations for how we should respond unchanged . As with results from the UK outbreak , uncertainty associated with the locations of undetected infected premises , in the form of additional foci of infection , seemed to have a large effect on forward projections in the Miyazaki outbreak .
Our results have shown that policy recommendations using real-time parameter estimates and forward simulations can be correct from an early stage in an outbreak despite highly uncertain projections of epidemic severity , in line with previous research on one-time decisions for the control of Ebola [18] . Projections from mathematical and simulation models are useful for informing policy-making insofar as they help to identify what is the best course of action and we have included two summaries of how simulation results may be communicated , dependent upon whether policymakers have an objective of 1 ) optimizing an expectation in outbreak severity ( Figs 2B and 3B ) or 2 ) optimizing the number of times that an intervention was predicted to be optimal ( Figs 2C and 3C ) . In many situations , estimating the burden of infection and selection of the optimal control intervention are directly related . However , our results are a reminder that this may not always be the case; in many cases it is how projections influence change in the recommended control intervention , rather than the value of projections themselves , that matter and so , where appropriate , it is important to include projections of intervention efficacy when making real-time projections of disease burden . Our results showed that changes in the estimated relative performance of control interventions was strongly influenced by the spatial distribution of both observed and undetected infectious premises and culled premises ( the state of the outbreak ) . Although our analysis only looked at one change in control intervention ( from what was being applied historically ) this dependence highlights that , should one allow multiple changes of control through time , then the optimal policy is likely to be dynamic . In our results , even if accurate parameter estimates had been available from the first confirmed case ( such as those derived from the whole outbreak ) , it would still have been necessary to perform state-dependent control to find the best choice of control intervention as the outbreaks evolve . In practice , estimating the state of the outbreak ( i . e . prediction of locations of undetected infections ) depends on estimates of transmission parameters and , therefore , continual re-evaluation of control interventions alone makes little sense without also re-evaluating , and re-fitting , model parameters . Thus , control recommendations should adapt to the changing state of an outbreak . Our results included a marked stabilization in recommended control interventions in simulations for the outbreak in the UK . It remains an avenue for future research to investigate what are the drivers of such stabilization in recommendations in control intervention , and whether it is logistically feasible to measure ( and therefore act upon ) such drivers , beyond an in silico experiment . Such analyses would require a more general approach to seeding outbreaks , and generating landscapes , to more thoroughly explore the state-space of potential future outbreaks . Here , we only investigated two outbreaks , simulations for each of which were seeded from historical starting points . Often , the need for rapid response during an outbreak is discussed in terms of the initial response [13] , but not as often in terms of adapting as circumstances change throughout an outbreak . In both outbreaks we investigated , there was significant variability in epidemiological predictions during the early stages of disease outbreaks highlighting that real-time updating of parameters is vital in order to obtain accurate predictions of epidemic size and extent . However , additional foci of infection may be seen as independent outbreaks and , therefore , the rapid response that is called for at the start of an outbreak needs to be reapplied when such foci are discovered . For the outbreaks that we investigated , identifying the locations of undetected infected premises was of greater importance to determining the best course of action than identifying the underlying disease dynamics; this calls for increased vigilance in surveillance during an outbreak and highlights the importance of methodology for predicting undetected infections ( e . g . [11 , 12] ) , assimilating surveillance data into parameter estimates for individual-based models , and of confirming negative cases , especially when such premises are in locations that would be considered as additional foci of infection . Simply because an outbreak has progressed beyond its initial stages does not mean that the need to act swiftly according to changes in the outbreak are in any way diminished . At a coarse level , for one-time decisions , the idea of state-dependent control has already been adopted or discussed by several agencies , such as the Department for Environment , Food , and Rural Affairs , UK ( DEFRA ) and the United States Department of Agriculture ( USDA ) in flowcharts for determining when to perform emergency vaccination [19 , 20] , the dependence of different phases in smallpox eradication upon smallpox prevalence [21] , management of wildlife diseases [22] , and the use of adaptive surveillance of herds in the eradication of rinderpest [23] . Decision-making frameworks , such as adaptive management , have also highlighted the utility of modeling and optimization as tools for generating state-dependent policies [24] . We note that optimization methods such as dynamic programming or reinforcement learning would be required to generate state-dependent policies and that such methodologies , although a more complex optimization than what we have presented , would allow the possibility of changing control intervention repeatedly through time , or generating optimal interventions that are a combination of those defined here , such as county-specific interventions ( e . g . [25] ) . Our analysis shows there is potential to make large gains in the effectiveness of the response to an outbreak by adopting state-dependent control that is on a much more nuanced scale than one-time binary decisions , and that mathematical and simulation models can play a significant role in policy preparedness by investigating such strategies before an outbreak occurs .
Both outbreaks used a similar model structure . The infectious pressure , λj ( t ) , on a susceptible farm j at time t is λj ( t ) =ϵ ( t ) +∑i∈I ( t ) βijh ( Ij−Ii ) +∑i∈N ( t ) βij*h ( Ij−Ii ) , where I ( t ) and N ( t ) are the sets of infected and notified farms at time t respectively ( t = 0 , 1 , 2 , … T ) . Models of similar structure have been previously used for modeling FMD ( e . g . [1 , 4 , 11] ) . A summary of notation used to describe the model is given in Table 2 . Notification time is assumed to mean the time of laboratory confirmation of FMD , and removal time is assumed to mean the date when the farm is both culled and disposed of . We count t in days; day 0 is the first confirmed infected case in the data and T is the day of the final parameter estimate ( 24 December 2001 for the UK outbreak and 6 July 2010 for the outbreak in Miyazaki ) . By including both S ( t ) and R ( t ) , as the sets of susceptible and removed farms at time t respectively , we then have four sets essentially giving the state of the epidemic on the population at a given point in time , St=⟨S ( t ) , I ( t ) , N ( t ) , R ( t ) ⟩ . The infection time of the kth farm is denoted Ik , while Nk and Rk denote notification and removal times of farm k respectively . Note that in the next section we assume that for It , Nt , and Rt , the t subscript denotes the set of infection , notification , or removal times of all premises respectively; here the time subscript is dropped for succinctness . All times are listed in days unless otherwise specified . Infectious pressure can be decomposed into contributions from infectious ( but not yet notified ) farms , βij=γ1q ( i;ξ ) w ( j;ζ ) δ ( δ2+ρij2 ) ω , i∈I ( t ) , j∈S ( t ) , ( 1 ) and contributions from notified farms βij*=γ2βij ( i∈N ( t ) , j∈S ( t ) ) . We define q ( i;ξ ) =[ ( cic¯ ) ψ1+ξ2 ( pip¯ ) ψ2+ξ3 ( sis¯ ) ψ3] and w ( j;ζ ) =[ ( cjc¯ ) ϕ1+ζ2 ( pjp¯ ) ϕ2+ζ3 ( sjs¯ ) ϕ3] to be the ‘infectivity’ of farm i and ‘susceptibility’ of farm j respectively , where ck , pk , and sk are the numbers of cattle , pigs , and sheep on farm k , with mean numbers of cattle , pigs , and sheep , denoted by c¯ , p¯ , s¯ . The effect of distance is captured using a Cauchy-type kernel where ρij is the Euclidean distance between farms i and j , and δ the decay of transmission rate with distance . Baseline infectious pressure distinguishes between periods before and after the movement ban was implemented , ϵ ( t ) ={ϵ1t<movementbanϵ1ϵ2otherwise , and the latency of the disease was modeled using the function h ( t ) ={1t<40otherwise . An estimate at time t of the complete parameter set , of 16 parameters ( 12 parameters in Miyazaki ) , is denoted θt . Parameter ω was assumed fixed at 1 . 3 in both outbreaks , as were the initial priors . The priors for each parameter were elicited via expert opinion and assigned the following distributions: π0 ( δ ) ∼Gamma ( 4 , 8 ) , π0 ( ϵ1 ) , π0 ( ϵ2 ) ∼Gamma ( 1E−7 , 1 ) , π0 ( ψ1 ) , π0 ( ψ2 ) , π0 ( ψ3 ) , π0 ( ϕ1 ) , π0 ( ϕ2 ) , π0 ( ϕ3 ) , π0 ( ξ2 ) , π0 ( ξ3 ) , π0 ( ζ2 ) , π0 ( ζ3 ) ∼Gamma ( 0 . 5 , 1 ) , π0 ( γ1 ) , π0 ( γ2 ) ∼Gamma ( 1E−4 , 1 ) . Gamma ( a , b ) is the gamma distribution with shape parameter a and rate parameter b . Infection to notification time was assumed to be distributed according to a Gamma distribution such that Ni → Ii = di∼Gamma ( 4 , b ) with parameter b = 0 . 5 governing the scale of the probability distribution . Notification to removal time is an observed quantity since both events are recorded . Let the data ( demographics and event history ) observed up to time t be Xt− , with πt ( θt , It|Xt− ) denoting the joint posterior distribution of the model parameters and infection times ( including infection times of undetected infections ) immediately at time t . We assume that infections occurred according to a continuous-time non-homogeneous Poisson process , where infection rate is assumed to be λj ( t ) as above . Let I , N , and R be vectors of infection , notification , and removal times for individuals 1 , … , nI who were infected by observation time Tobs . Conditioning on this event time , the joint posterior distribution over parameters θ = {ϵ1 , ϵ2 , γ1 , γ2 , ξ , ζ , ψ , ϕ , δ , b} is π ( θ|I , N , R ) ∝∏j=1nI[λj ( Ij− ) ]exp[∫IκTobs∑j∈Pλj ( t ) dt] ×∏j=1nI ( Nj−Ij ) a−1e−b ( Nj−Ij ) ×∏p=1|θ|fθp ( θp ) , where the first line represents the infection process , the second line represents the detection ( infected to notified ) process , and the third line represents independent prior distributions for all components of θ . Ij− represents the time immediately before the infection time of the jth premises . Parameters in bold , represent the corresponding set of species-specific parameters , e . g . ζ = {ζ2 , ζ3} . κ denotes the initial infective , and P denotes the set representing all individuals in the population . The joint posterior distributions used in the forward simulations and figures of the instantaneous risk of onward spread represent the MCMC output sub-sampled to 2000 parameter coordinates . Multisite adaptive Metropolis-Hastings was used to draw from the conditional posterior distributions of {ϵ1 , ϵ2 , γ1 , , γ2 , δ} , ψ and ϕ . Furthermore , since the infection times are unobserved , we updated them component-wise using Metropolis-Hastings , with a reversible-jump update to explore the posterior over the presence of undetected infections . See [11 , 12] for further details . The introduction of the 3-species model ( in comparison to [11] and [1] ) results in a non-linearly correlated posterior distribution , particularly between γ1 and both of ξ and ζ . This presents particular difficulties for Metropolis-based MCMC algorithms , since the optimal proposal distribution scale changes with location in the posterior parameters space . To approximately orthogonalize the posterior distribution , and hence improve the convergence properties of our adaptive Metropolis algorithm , we employed the following non-centered update for ξ and ζ . To construct an efficient proposal distribution for ξ , we seek to exploit the shape of the posterior distribution with respect to γ1 . This can be thought of as a joint update of γ1 and ξ , respecting the contour of the joint posterior . First , we write the equation for q ( i;ξ ) ( see above ) as γ1q ( i;ξ ) =γ1Ci+γ1ξ2Pi+γ1ξ3Si =A{i , 1}+A{i , 2}+A{i , 3} where Ci= ( ci/c¯ ) ψ1 and similarly for Pi and Si . We then let R=A1+A2+A3=∑i=1nI ( A{i , 1}+A{i , 2}+A{i , 3} ) We then propose ( A2*+A3* ) T∼MVN2 ( ( A2+A3 ) T , 2 . 3822B ) and A1*=R−A2*−A3* where , with probability 0 . 05 , B = I2 the 2x2 identity matrix , and with probability 0 . 95 , B = Σk the empirical covariance matrix of the MCMC samples for ( A1 , A2 ) T up to iteration k ( see [30] ) . We then solve for γ1 and ξ . Similarly , we update γ1 and ζ . Uncertainty in the joint posterior distribution of transmission parameters was summarized using a measure of the instantaneous risk of onward spread , defined as the instantaneous force of infection from an average-sized infectious farm to an average-sized susceptible farm at time t ( gt ) . This risk measure was calculated at each time point as the integral over the joint posterior distribution of parameters at that point in time and over a distance of 20km ( from the infectious to the susceptible farm ) : gt=∫θ∫020γ1[ξ2 , t+ξ3 , t][ζ2 , t+ζ3 , t]δt ( δt2+ρ2 ) ωdρdθ . The equation for the instantaneous risk of onward spread is taken from the equation for the infectious pressure ( Eq 1 ) substituting the average number of cattle , pigs , and sheep for ci , pi , and si respectively . Each control intervention , a , is evaluated according to the expected total number of animals culled , U ( Yt|a ) , where the Bayesian predictive distribution of the ongoing epidemic , Yt , is given by the integral fYt ( Yt|Xt− , a ) =∫Θ∫IfYt ( Yt|Xt− , θt , It , a ) πt ( θt , It|Xt− ) dIdθ . This was estimated by simulating forward from fYt ( Yt|Xt− , θt , It , a ) using draws from πt ( θt , It|Xt− ) . The above formulation is using parameter estimates from the time point in question , so-called ‘accrued information’ . Forward simulations were also generated with parameters estimated using all data , fYt ( Yt|XT− , a ) , so-called ‘complete information’ , which were generated in an analogous fashion instead using draws from πT ( θt , It|XT− ) in the above integral . The optimal control strategy at⋆ is chosen as that which minimizes the mean expected total animals culled at⋆=argminamean[U ( Yt|a ) ] Vaccine efficacy was assumed to be 90% , whereby animal numbers on a vaccinated farm were reduced by 90% after a delay until conferment of immunity of 4 days . Delay from infection to notification time was simulated as 9 days , with a 4-day latency period . Notification to removal time was simulated as 1 day for infected premises and 2 days for culling of dangerous contacts or contiguous premises . The MCMC algorithm above was implemented in C++ ( GCC version 4 . 8 . 3 ) embedded within an R package . General-purpose graphics processing unit ( GPU ) using NVIDIA CUDA 7 . 5 was used to implement the calculation of the likelihood and speed up inference . The software is available under the GPLv3 license at http://fhm-chicas-code . lancs . ac . uk/InFER/InFER/tags/InFERfmd-v1 . 0 . | Mathematical and simulation models may be used to inform policy in the early stages of an infectious disease outbreak by evaluating which control strategies will minimize the impact of the epidemic . In these early stages , significant uncertainty can limit the ability of models to provide accurate predictions , and policymakers do not have the luxury of waiting for data to alleviate this state of uncertainty . For policymakers , however , what is most important is the selection of the optimal control intervention , rather than accuracy of model predictions . We fit an epidemic model to observed , spatially-explicit , infection data at weekly intervals throughout two historical outbreaks of foot-and-mouth disease , UK in 2001 and Miyazaki , Japan in 2010 , and compare forward simulations of the impact of alternative control interventions . These are compared to policy recommendations generated in hindsight using data from the entire outbreak . Our results show that the optimal control policy is identified accurately from an early stage in an outbreak , despite high levels of uncertainty in projections of epidemic size , and that the relative performance of control strategies is strongly mediated by our understanding of the locations of infected farms , rather than improved estimates of transmission parameters . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"livestock",
"medicine",
"and",
"health",
"sciences",
"animal",
"diseases",
"foot",
"and",
"mouth",
"disease",
"ruminants",
"immunology",
"vertebrates",
"animals",
"mammals",
"simulation",
"and",
"modeling",
"preventive",
"medicine",
"infectious",
"disease",
"control",
... | 2018 | Real-time decision-making during emergency disease outbreaks |
The World Health Organization has recently reemphasized the importance of providing preventive chemotherapy to women of reproductive age in countries endemic for soil-transmitted helminthiasis as they are at heightened risk of associated morbidity . The Demographic and Health Surveys ( DHS ) Program is responsible for collecting and disseminating accurate , nationally representative data on health and population in developing countries . Our study aims to estimate the number of pregnant women at risk of soil-transmitted helminthiasis that self-reported deworming by antenatal services in endemic countries that conducted Demographic and Health Surveys . The number of pregnant women living in endemic countries was extrapolated from the United Nations World Population Prospects 2015 . National deworming coverage among pregnant women were extracted from Demographic and Health Surveys and applied to total numbers of pregnant women in the country . Sub-national DHS with data on self-reported deworming were available from 49 of the 102 endemic countries . In some regions more than 73% of STH endemic countries had a DHS . The DHS report an average deworming coverage of 23% ( CI 19–28 ) , ranging from 2% ( CI 1–3 ) to 35% ( CI 29–40 ) in the different regions , meaning more than 16 million pregnant women were dewormed in countries surveyed by DHS . The deworming rates amongst the 43 million pregnant women in STH endemic countries not surveyed by DHS remains unknown . These estimates will serve to establish baseline numbers of deworming coverage among pregnant women , monitor progress , and urge endemic countries to continue working toward reducing the burden of soil-transmitted helminthiasis . The DHS program should be extended to STH-endemic countries currently not covering the topic of deworming during pregnancy .
Soil-transmitted helminthiasis ( STH ) are caused by the intestinal worms Ascaris lumbricoides ( roundworm , ) Trichuris trichiura ( whipworm ) , Necator americanus and Ancylostoma duodenale ( hookworm ) . STH are transmitted when individuals come in contact with environment contaminated by faeces containing nematode eggs . The parasites’ eggs or larvae enter the human body via ingestion of infested food , the larvae by skin penetration[1 , 2] . More than two billion people [1] in 102 endemic countries [3] were considered to be infected with STH in 2015 , causing a loss of 39 million of disability-adjusted life years . Most of the disease burden occurs in the tropical and sub-tropical areas of sub-Saharan Africa and South East Asia [4] . STH endemicity is a result of poor sanitary infrastructure and lack of understanding regarding the importance of safe disposal of faeces characteristic of the most impoverished communities of the world [5] . The World Health Organization ( WHO ) recommends administration of albendazole ( 400 mg ) or mebendazole ( 500 mg ) coupled with hygiene education to reduce STH-related morbidity among high risk populations . Annual PC is recommended in areas where STH prevalence is between 20% and 50% , and biannual PC in areas where STH prevalence exceeds 50%[1] . Completely eliminating STH requires long-term commitment and the allocation of extensive resources toward improving water and sanitation[6] . WHO recognizes three population groups at highest risk of STH-related morbidity: preschool-age children ( pre-SAC ) , school-age children ( SAC ) , and women of reproductive age ( WRA ) [1] . WRA are especially vulnerable as they are at elevated risk of certain comorbidities such as anaemia ( exacerbated particularly by hookworm and whipworm infections ) [7] . Drug donations are presently available for the regular PC of pre-SAC and SAC in STH-endemic countries . These countries consistently report the use of donated drugs among risk groups to WHO[8] . In 2015 , the implementation of PC averted over 44% of lost disability-adjusted life years due to STH in pre-SAC and SAC [9] . WHO has recently reiterated the importance of WRA as a target population for the scaling up of PC in the Bellagio Declaration[7] as STH among pregnant women can produce or seriously aggravate maternal and neonatal complications[4] . Yet , anthelminthic drug donations are currently not available for WRA and countries are not reporting coverage , making the estimation of PC provided to this group difficult to assess . In this study , we quantified the number of pregnant women being dewormed by health services in countries considered endemic for STH ( >20% prevalence ) [3 , 8] that conducted recent Demographic and Health Surveys ( DHS ) . The estimates offered by this study will provide useful information for the scaling up of PC programmes among WRA , however it is important that control programme managers collect further coverage and context specific epidemiological data to improve these estimates .
The United Nations World Population Prospects 2015 consists of population estimates and projections executed by the Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat . It includes results from national population censuses and demographic and health surveys , outlining key indicators by region , sub-region , country , and development group . The United Nations World Population Prospects 2015 was used to estimate the total number of pregnant women in STH-endemic countries[10] . DHS are standardized nationally-representative household surveys implemented over a span of 18–20 months about every five years . The surveys are implemented by ICF International and funded by the United States Agency for International Development ( USAID ) with contributions from other donors such as UNICEF , UNFPA , WHO , and UNAIDS . The surveys consist of four types of questionnaires ( Household , Woman’s , Man’s , and Biomarker ) that collect data on demographic , environmental , socio-economic , and health-related characteristics . One of the aim of DHS is to evaluate the performances of health services; in this context DHS asked to women in the sample with a live birth in the past five years , if the “took intestinal parasite drug” during the pregnancy of their last birth [11] . All DHS survey data were extracted and analysed using R statistical program and the rDHS and survey packages . Permission for sub-national DHS data between 2000 and 2017 in STH-endemic countries was first sought , downloaded through the DHS API , and national means and confidence intervals of coverages of deworming among pregnant women by antenatal services summarised using sample weights . The number of live births obtained from the United Nations World Population Prospects 2015 was assumed to be equal to the total number of pregnant women in each country . The 2015 report was used , as opposed to a more recent version , to maintain consistency with previously estimated deworming coverage among not pregnant WRA . Deworming coverage measured by DHS were assumed to be representative of the national situation . For DHS reports reporting deworming , the proportion of women dewormed during their last pregnancy was applied to number of pregnant women in each country . Total pregnant women and pregnant women dewormed in each of the endemic countries were aggregated by WHO region and globally .
Out of 102 STH-endemic countries , DHS reports for 49 countries were available , as depicted in Fig 1 . 53 countries did not have DHS reports or were excluded from the analysis as they were based on surveys conducted before 2000 . Overall , DHS was available for 48% of all endemic countries ( Table 1 ) . Out of the 1 . 24 million household interviews conducted , 85% were in the South East Asian ( 636 , 551 ) and African ( 425 , 650 ) regions . 45 out of 49 DHS Reports were based on surveys conducted after 2010 , with the majority of surveys being from 2011–2015 . Only 6 DHS reports were based on surveys conducted before 2011 . The proportion of DHS reports which include the percentage of pregnant WRA dewormed by antenatal services increased from 0% between 2000 and 2005 to 92% after 2016 ( see S1 Fig ) . Among the 49 countries investigated with DHS , it was estimated that out of 68 million pregnant women at risk of STH , over 16 million received deworming medication during their last pregnancy ( Table 2 ) . The kind of anthelminthic drug given was not specified in the DHS reports . The average coverage of deworming was estimated at 23% ( CI 19–28 ) among the countries surveyed , and varied between 0 . 6% ( CI 0 . 3–1 . 2 ) and 83% ( CI 75–89 ) . Overall , the 49 countries with DHS reports accounted for 62% of total pregnant women in all STH-endemic countries . DHS represented deworming coverage during pregnancy in the African region the best , with 31 out of 42 STH endemic countries surveyed . In contrast , the Western Pacific Region was least represented by DHS , with only 2 out 15 STH endemic countries surveyed for deworming coverage during pregnancy . There was no DHS report available for China , which accounts for 77 . 8% of pregnant women at risk of STH in the Western Pacific region and 14 . 6% of pregnant WRA in all endemic countries . Consequently , the Western Pacific region had the lowest percentage of pregnant women represented by available DHS ( Table 2 ) . Table 2 describes the regional distribution of total pregnant women at risk of STH as well as the estimated number of pregnant women dewormed in all STH-endemic countries .
In this study we estimate that in 49 STH endemic countries where DHS was recently conducted over 16 million have received deworming . More than 1 . 2 million households were interviewed and we consider the estimated deworming coverage representative of the performances of health services in those countries . At the moment countries are not requested to report coverage data on WRA because this risk group is not among the ones targeted by the NTD road map . Our data shows an overall positive trend in the number of STH-endemic countries reporting deworming of pregnant women in the absence of global monitoring , the initial spike in this trend coinciding with the first Global Partner’s Meeting on Neglected Tropical Diseases organized in 2007 [12] . Deworming initiatives for pregnant women are likely to continue growing as countries scale up their STH control programs . These estimates constitute a valuable step toward the coverage of this group at risk as they represent baseline figures of deworming coverage among pregnant women in multiple endemic countries . Although we do not have access to more recent DHS information on countries that conducted surveys between 2000 and 2005 , it is probable that those countries have since initiated deworming activity for pregnant women . Additionally , we recognize that countries that do not report deworming as a component of antenatal services may be providing anthelminthic drugs to pregnant women through other platforms , or women are procuring deworming tablets in local pharmacies or markets . The coverage of deworming among pregnant women in this study could therefore be underestimated . It is important to note that we found DHS for 73% of the STH-endemic countries in Africa , and for 50% of the STH-endemic countries in South East Asia , both accounting for the vast majority of pregnant women at risk of STH worldwide . As these two WHO regions comprise the highest concentration of STH , we consider that the data collected is particularly representative of those countries most in need of targeted PC . There was , however , no data on deworming during pregnancy from 53 STH endemic countries , meaning an evidence gap still remains in these countries and should be addressed , particularly in the American and Western Pacific regions , to better understand global deworming coverage in pregnant women . In addition to expanding the geographic scope of DHS , the questionnaire could be adapted in future surveys to elicit better quality data on deworming in WRA by , for example , by limiting the question on use of deworming to the latest pregnancy . A limitation of this study is the assumption that the number of pregnant women in a country is equal to the number of live births , as it omits all pregnancies that resulted in abortions , miscarriages , and stillbirths . On the other hand , multiple pregnancies ( i . e . those giving birth to twins ) may lead to overestimation of total number of pregnant women , granted an overall underestimation is more likely . In our opinion , antenatal services represent a relevant potential channel for systemic PC delivery considering that countries such as Congo have achieved 85 . 8% deworming coverage of pregnant women solely through antenatal services . Targeted PC can significantly reduce the burden of STH among WRA , improving maternal and child health outcomes and spurring productivity[13] . Considering the prospective impact of STH elimination and the relatively low implementation costs , PC among WRA should be at the forefront of public health operations in all endemic countries . We urge countries that have already integrated deworming into health programming to continue scaling up the provision of anthelminthic treatment among WRA . The estimates provided by this study will aid strategic efforts by informing the expansion of PC among WRA in endemic countries and promoting the definitive goal of eliminating STH globally . | Soil-transmitted helminths are intestinal worms that cause significant suffering among the poorest communities in the world . They are transmitted via contaminated water , food or soil , all of which result from poor sanitation . Children and women of reproductive age are at heightened risk of related morbidities such as malnutrition , cognitive impairment and anaemia . Pregnant women are particularly susceptible to severe maternal and neonatal complications . Deworming drugs are cheap , safe , and effective in reducing morbidity related to soil-transmitted helminthiasis . Large scale drug administration campaigns have distributed donated medicines to children in endemic countries , but women of reproductive age are currently not well covered . Yet , demographic surveys show that they are being treated for soil-transmitted helminthiasis through health care services . This study provides estimates for the number of pregnant women at risk of soil-transmitted helminthiasis being dewormed by antenatal services in endemic countries conducting Demographic Health Surveys . These estimates mark the preliminary reference point for deworming coverage among pregnant women in endemic countries , and will thus prove useful for tracking overall progress in the ongoing effort to eliminate neglected tropical diseases . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"cognitive",
"neurology",
"medicine",
"and",
"health",
"sciences",
"maternal",
"health",
"obstetrics",
"and",
"gynecology",
"united",
"nations",
"tropical",
"diseases",
"social",
"sciences",
"parasitic",
"diseases",
"neuroscience",
"political",
"science",
"health",
"car... | 2019 | Provision of deworming intervention to pregnant women by antenatal services in countries endemic for soil-transmitted helminthiasis |
Recognition of viruses by pattern recognition receptors ( PRRs ) causes interferon-β ( IFN-β ) induction , a key event in the anti-viral innate immune response , and also a target of viral immune evasion . Here the vaccinia virus ( VACV ) protein C6 is identified as an inhibitor of PRR-induced IFN-β expression by a functional screen of select VACV open reading frames expressed individually in mammalian cells . C6 is a member of a family of Bcl-2-like poxvirus proteins , many of which have been shown to inhibit innate immune signalling pathways . PRRs activate both NF-κB and IFN regulatory factors ( IRFs ) to activate the IFN-β promoter induction . Data presented here show that C6 inhibits IRF3 activation and translocation into the nucleus , but does not inhibit NF-κB activation . C6 inhibits IRF3 and IRF7 activation downstream of the kinases TANK binding kinase 1 ( TBK1 ) and IκB kinase-ε ( IKKε ) , which phosphorylate and activate these IRFs . However , C6 does not inhibit TBK1- and IKKε-independent IRF7 activation or the induction of promoters by constitutively active forms of IRF3 or IRF7 , indicating that C6 acts at the level of the TBK1/IKKε complex . Consistent with this notion , C6 immunoprecipitated with the TBK1 complex scaffold proteins TANK , SINTBAD and NAP1 . C6 is expressed early during infection and is present in both nucleus and cytoplasm . Mutant viruses in which the C6L gene is deleted , or mutated so that the C6 protein is not expressed , replicated normally in cell culture but were attenuated in two in vivo models of infection compared to wild type and revertant controls . Thus C6 contributes to VACV virulence and might do so via the inhibition of PRR-induced activation of IRF3 and IRF7 .
Mammalian cells respond to viral infection by producing pro-inflammatory cytokines and chemokines and also interferons ( IFNs ) , of which type I IFNs , consisting of IFN-β and several IFNα proteins , are particularly important . IFN-α and IFN-β then act in an autocrine and paracrine manner to switch on hundreds of target genes which contribute to anti-viral innate immunity by blocking virus replication and alerting neighbouring cells to the danger of infection ( reviewed in [1] ) . In addition to their role in innate immunity , type I IFNs also promote adaptive immune responses by priming T helper cells and cytotoxic T cells [2] . The initial production of type I IFNs is due to the activation of IFN regulatory factors ( IRFs ) , and in particular IRF3 , downstream of pattern recognition receptors ( PRRs ) , which recognize viral DNA , RNA and proteins . PRRs that detect the presence of foreign RNA include the RIG-I-like receptors ( RLRs ) melanoma differentiation-associated gene 5 ( MDA5 ) and retinoic acid induced gene I ( RIG-I ) , which sense intracellular double-stranded ( ds ) RNA and single-stranded ( ss ) RNA containing a 5′ triphosphate , respectively [3]–[6] . Other PRRs that aid the detection of viruses include the endosomal toll-like receptors ( TLRs ) , namely TLR3 which senses dsRNA , TLR7 and TLR8 which sense ssRNA and TLR9 which recognizes unmethylated DNA ( reviewed in [7] ) . Intracellular DNA sensors such as AIM2 , RNA polymerase III , DAI and IFI16 are also involved in sensing DNA viruses by recognizing the presence of dsDNA in the cytosol [8]–[15] . RNA polymerase III , DAI and IFI16 signal to cause type I IFN production , while AIM2 activates the inflammasome leading to processing of pro-interleukin ( IL ) -1β and release of IL-1β [9] , [11] , [12] , [15] . RNA polymerase III is unusual in that it does not signal directly in response to DNA , but instead transcribes AT-rich DNA into RNA species , which are then recognized by RIG-I [8] , [10] . The signalling pathways activated by the RLRs , the IFN-inducing intracellular DNA receptors and by TLR3 converge at the level of the kinases TNF receptor associated factor ( TRAF ) family member NF-κB activator ( TANK ) -binding kinase 1 ( TBK1 ) and IκB kinase-ε ( IKKε ) . These kinases exist in complexes with the scaffold proteins TANK , NAP1 ( NAK-associated protein 1 ) or SINTBAD ( similar to NAP1 TBK1 adaptor ) [16]–[18] . To activate these kinases , RLRs and consequently RNA polymerase III signal via the adaptor protein MAVS ( mitochondrial antiviral signalling ) [19]–[21] , other intracellular DNA sensors employ STING ( stimulator of IFN genes ) [22] , [23] , while TLR3 uses TRIF ( TIR-domain containing adaptor molecule inducing IFN-β ) [24] , [25] . PRRs also require TRAF3 for the activation of TBK1 and IKKε ( reviewed in [26] ) . Once activated by PRR signalling , TBK1 and IKKε phosphorylate IRF3 , causing its translocation to the nucleus and the transcriptional activation of promoters containing appropriate binding sites , such as the IFN-β and CCL5 promoters , and a subset of promoters containing IFN-stimulated response elements ( ISREs ) [27] , [28] . A related transcription factor of the IRF family , IRF7 , also plays an important role in anti-viral responses and can be activated in a similar manner to IRF3 during viral infection [29] . However , while IRF3 is expressed constitutively , IRF7 is present at low levels in most cells , but is induced by type I IFNs in a positive feedback loop . Thus , IRF7 is particularly important for the continued expression of IFN-β during viral infection , and also contributes to induction of IFN-β by co-operation with IRF3 [29] , [30] . In addition , IRF7 is essential for the induction of IFN-α genes that are not induced by IRF3 [29] . In plasmacytoid dendritic cells , an alternative TBK1- and IKKε-independent signalling pathway resulting in the activation of IRF7 is employed by TLR7 , TLR8 and TLR9 . These endosomal TLRs signal through the adaptor protein MyD88 ( myeloid differentiation factor 88 ) , leading to activation of the kinase IKKα which then phosphorylates IRF7 [29] , [31] , [32] . This is unusual , because in other PRR signalling pathways IKKα and IKKβ are involved in the phosphorylation of the inhibitor of NF-κB ( IκB ) , causing its degradation and the subsequent activation of NF-κB . NF-κB is another transcription factor activated by PRRs , and is critical for innate immunity . NF-κB and IRF3 ( or IRF7 ) co-operate with the activating protein 1 ( AP-1 ) transcription factor family to induce the transcription of the IFN-β promoter . A functional type I IFN response provides a potent means of controlling virus infections [33] and consequently viruses have evolved numerous counter-measures to stop the production or action of IFNs or IFN-induced anti-viral proteins ( for review see [34] ) . These strategies include blockage of antiviral PRR signalling pathways ( reviewed in [34] , [35] ) . Viruses with a large DNA genome , such as poxviruses , encode an extensive array of immunomodulatory proteins . Vaccinia virus ( VACV ) , an orthopoxvirus used as a vaccine to eradicate smallpox , has many immune evasion mechanisms , and these include intracellular proteins that block PRR signalling , secreted factors that sequester IFNs and proinflammatory cytokines , and proteins that inhibit the effector actions of an IFN response ( reviewed in [36] ) . However , the exact function of many of the approximately 200 virus proteins remains unclear . Here , a functional screen was used to identify VACV proteins that inhibit the induction of the type I IFN response after PRR activation . It is shown that protein C6 , the product of the C6L gene , is an inhibitor of IFN-β promoter activation . The C6 protein is a member of a family of VACV proteins that includes B14 , A52 and K7 [37] , [38] . The crystal structures of B14 , A52 and K7 were solved [39] , [40] and showed that they , and also VACV proteins N1 [41] , [42] and F1 [43] , adopt a Bcl-2-like fold . Functional characterisation showed that only F1 and N1 inhibit apoptosis [42] , and consistent with this these proteins have a surface groove for binding BH3 peptides from pro-apoptotic Bcl-2 proteins [42] , [43] . In contrast , proteins B14 , A52 and K7 lack this groove and inhibit innate immune signalling pathways instead [40] , [44] . Interestingly , the protein N1 is both anti-apoptotic and inhibits NF-κB activation induced by IL-1 [40] , [42] , [45] . In this paper we demonstrate that C6 inhibits the activation of IRF3 and IRF7 downstream of the kinases TBK1 and IKKε , while C6 does not inhibit signalling pathways using IKKα for IRF7 activation . Inhibition of IRF3 and IRF7 by C6 may be mediated by its interaction with the scaffold proteins TANK , NAP1 and SINTBAD . Consistent with the ability of C6 to inhibit IFN-β expression , recombinant viruses that do not express C6 are attenuated in vivo compared to the wild type and revertant viruses . C6 represents the first viral protein shown to target the TBK1 scaffold proteins .
To uncover novel VACV proteins that inhibit innate immune signalling pathways , a functional screen of proteins from VACV strain Western Reserve ( WR ) was used to identify those that inhibit type I IFN induction . For this , poorly characterized proteins encoded in the terminal regions of the VACV WR genome were selected , because these regions are rich in immunomodulatory proteins [46] . Proteins encoded within the highly conserved central region of the VACV genome were excluded from the screen , as were proteins with well-characterized functions , secreted proteins and those smaller than 8 kDa . This selection process identified 49 ORFs , and these were amplified from VACV WR strain genomic DNA , and cloned into mammalian expression vectors . Plasmids encoding these ORFs were transfected individually into HEK293 cells , and the effect on the IFN-β promoter following PRR stimulation was measured by reporter gene assays . ORF VACVWR022 ( gene C6L in VACV Copenhagen strain ) encoding the protein C6 emerged from this screen as an inhibitor of IFN-β promoter activation . Expression of C6 inhibited the activation of the IFN-β promoter by transfected poly ( dA-dT ) ( which acts via intracellular DNA sensors ) or poly ( I∶C ) ( which acts via RLRs ) in HEK293 cells ( Figure 1A , B ) , and by infection of cells with Sendai virus which activates RIG-I signalling [4] ( Figure 1C ) . C6 also inhibited poly ( I∶C ) -induced IFN-β promoter activation in mouse NIH3T3 cells ( Figure 1D ) . Furthermore , the presence of C6 inhibited the expression of endogenous IFN-β mRNA in Sendai virus-infected cells ( Figure 1E ) , as well as the secretion of the chemokine CCL5 from infected cells ( Figure 1F ) . Thus , the C6 protein significantly reduced the expression of IFN-β and CCL5 after stimulation of PRRs by ligands or viral infection . During viral infection , IRF3 and NF-κB co-operate to activate the IFN-β and the CCL5 promoters . IRF3 is phosphorylated by the kinases TBK1 and IKKε , leading to its dimerization and translocation to the nucleus . In a similar way , the kinases IKKα and IKKβ phosphorylate IκB , which then releases activated NF-κB and allows its nuclear accumulation . To investigate whether the signalling pathways leading to NF-κB or IRF3 activation are inhibited by C6 , the translocation of the NF-κB subunit p65 from the cytoplasm to the nucleus was measured by confocal microscopy . For this , HEK293T cells were transfected with a plasmid expressing GFP-tagged C6 , or a control plasmid expressing GFP for 16 h . The cells were then infected with Sendai virus for 6 h , or stimulated with IL-1 for 15 min , fixed and stained for endogenous p65 . During Sendai virus infection , approximately 20% of control cells expressing GFP displayed p65 accumulation in the nucleus ( Figure 2A , B ) , and the presence of GFP-tagged C6 did not affect the extent of p65 nuclear translocation . Furthermore , the expression of C6 did not affect the nuclear accumulation of p65 in cells stimulated with IL-1 , an activator of NF-κB , which , regardless of whether the cells expressed GFP or GFP-tagged C6 , caused p65 nuclear translocation in more than 80% of cells ( Figure 2A , B ) . The effect of C6 on the expression of a luciferase reporter gene under the control of an NF-κB-dependent promoter was examined next . Over-expression of C6 did not prevent activation of the NF-κB-dependent promoter stimulated by IL-1 or tumour necrosis factor ( TNF ) -α in HEK293 cells ( Figure 2C ) , or by poly ( I∶C ) in HEK293 cells expressing TLR3 ( Figure 2D ) . In contrast , the VACV protein B14 , another member of the poxviral Bcl-2-like protein family , inhibited NF-κB promoter activation under these conditions ( Figure 2C , D ) as shown previously [47] . The activation and nuclear translocation of IRF3 was investigated next . HEK293T cells transfected with V5-tagged C6 or V5-tagged GFP were infected with Sendai virus for 6 h , fixed and stained for IRF3 . In cells infected with Sendai virus , IRF3 translocated to the nucleus in approximately 30% of cells expressing V5-tagged GFP ( Figure 3A , B ) . However , in cells expressing V5-tagged C6 , the translocation of IRF3 was impaired , and only 5% of C6-expressing cells displayed nuclear accumulation of IRF3 ( Figure 3A , B ) . Similar results were obtained with GFP-tagged C6 ( data not shown ) . This indicates that the inhibition of promoter induction by C6 is due to the prevention of the activation and/or nuclear translocation of IRF3 , and not due to an effect on p65 activation . To measure phosphorylation-dependent transactivation activity of IRF3 , a luciferase-based IRF3 transactivation assay was utilized . This assay uses a fusion protein consisting of the DNA-binding domain of Gal4 and the transactivation domain of IRF3 . When the IRF3 transactivation domain is phosphorylated by upstream signalling events , it induces expression of a luciferase reporter gene under the control of a Gal4-dependent promoter . Using this assay , C6 inhibited poly ( dA-dT ) -stimulated IRF3 transactivation ( Figure 3C ) , providing further evidence that C6 inhibits the activation of IRF3 by PRRs . To determine which step of the signalling cascade that leads to IRF3 activation is targeted by C6 , the ectopic expression of signalling proteins that act upstream of IRF3 activation was used to drive the IRF3 transactivation assay . The RLR adaptor MAVS and the kinases TBK1 and IKKε all promoted IRF3 activation in this assay when overexpressed ( Figure 3D , E ) . Co-expression of C6 inhibited the activation of IRF3 in a dose-dependent manner in each case ( Figure 3D , E ) , indicating that C6 acts at the level of these signalling components or further ‘downstream’ to prevent IRF3 activation . To gain further mechanistic insight , the ability of C6 to inhibit the function of IRF3 once it is activated by phosphorylation was measured . To do this , a constitutively active form of IRF3 ( IRF3-5D ) was used in which serine to aspartate mutations mimic the phosphorylation of five key residues in the IRF3 sequence [48] . Over-expression of IRF3-5D induced the expression of a luciferase reporter driven by an ISRE element derived from the ISG15 promoter ( Figure 3F ) , which was shown previously to be transcriptionally activated by IRFs [27] . C6 was unable to prevent the activation of the ISRE by over-expression of constitutively active IRF3 ( Figure 3F ) , suggesting that C6 acts to prevent the activation of IRF3 , but is unable to interfere with IRF3 function once it is activated by phosphorylation . IRF7 is a transcription factor that is structurally and functionally related to IRF3 and also participates in the induction of the IFN-β promoter in response to PRR signalling ( reviewed in [49] ) . IRF7 can be phosphorylated and activated by two distinct pathways . TLR3 and cytosolic PRRs , such as sensors of poly ( dA-dT ) , act via TBK1 and IKKε , while the endosomal TLRs TLR8 and TLR9 activate IRF7 using a signalling pathway independent of TBK1 and IKKε , but involving MyD88 and IKKα [29] , [32] , [50] . The ability of C6 to inhibit the TBK1/IKKε-dependent and -independent signalling pathways to IRF activation was compared by employing a luciferase-based IRF7 transactivation assay . Activation of the TBK1/IKKε-dependent pathway following infection of HEK293 cells with Sendai virus ( outlined in Figure 4A ) was inhibited by C6 ( Figure 4B ) , as was the activation of IRF7 by the downstream signalling components MAVS , TBK1 and IKKε ( Figure 4C , D ) . To determine whether C6 could inhibit the TBK1/IKKε-independent pathway to IRF7 activation , HEK293 cells stably expressing TLR8 were used . Stimulation of TLR8 by the agonists CL075 or R848 activates IRF7 via MyD88 and IKKα ( outlined in Figure 4E ) , however , this was not inhibited by C6 ( Fig . 4F ) . Similarly , C6 was unable to inhibit IRF7 activation following over-expression of MyD88 or IKKα ( Figure 4G ) . As observed for IRF3 , C6 was unable to inhibit the activation of the ISRE element in response to constitutively active IRF7 ( IRF7-4D , Figure 4H ) . Taken together , these data indicate that C6 inhibits the activation of IRF3 and IRF7 by TBK1- and IKKε-dependent signalling pathways , implying that C6 acts on these kinase complexes , rather than acting on the transcription factors directly . Published sequence data show that C6 has orthologues in other orthopoxviruses ( OPVs ) , the capripoxvirus and deerpoxvirus ( www . poxvirus . org ) , and within the OPV genus the conservation is high ( 89–97% amino acid identity ) . To investigate if C6 function is also conserved , the ability of the C6 orthologue from monkeypox virus ( MPXV ) strain ZAI 1979-005 ( 92% amino acid identity ) to inhibit poly ( dA-dT ) -induced IRF3 transactivation was investigated ( Figure S1A ) . The MPXV ORF encoding the C6 orthologue was amplified from DNA extracted from MPXV-infected HeLa cells . When a MPXV C6 expression vector was transfected into cells , the MPXV C6 protein , like the VACV C6 protein , inhibited the pathway at the level of TBK1 and IKKε , because MPXV C6 inhibited the activation of IRF3 caused by the over-expression of either of the two kinases , or of the adaptor protein MAVS ( Figure S1B ) . Thus , the MPXV C6 orthologue behaved like VACV C6 in the assays tested . To investigate how C6 antagonises activation of the pathway at the level of the TBK1- and IKKε-containing complexes , interactions between C6 and components of the kinase-containing complexes were sought by immunoprecipitation . HEK293 cells were transfected with plasmids encoding FLAG-tagged proteins and then infected with a VACV expressing HA-tagged C6 . Immunoprecipitation with anti-FLAG antibody co-precipitated C6 with the scaffold proteins NAP1 , SINTBAD and TANK but not with a FLAG-tagged control protein , FLAG-GFP ( Figure 5A ) . In contrast , no interaction with TBK1 or IKKε was detected ( data not shown ) . A reciprocal immunoprecipitation using lysates from cells over-expressing Streptavidin-tagged C6 and FLAG-tagged scaffold proteins also showed an interaction between C6 and the adaptors since immunoprecipitation of C6 co-precipitated NAP1 , SINTBAD and TANK , but not GFP ( Figure 5B ) . It has been proposed that TBK1 and IKKε form distinct complexes , either as homodimers or heterodimers , which would contain a specific scaffold protein ( namely TANK , NAP1 or SINTBAD [17] , [51] , [52] ) . To investigate whether the interaction between C6 and NAP1 , TANK or SINTBAD affected the formation of signalling complexes , the scaffold-kinase interactions were investigated in the presence or absence of C6 . A LUMIER interaction assay was used , in which a FLAG-tagged scaffold protein and luciferase-tagged TBK1 were co-expressed in the presence or absence of C6 , and the amount of luciferase co-immunoprecipitated with the FLAG-tagged allele was quantified [17] , [53] . C6 did not prevent the association between TBK1 and NAP1 , SINTBAD or TANK ( Figure 5C , E , G ) . In contrast , expression of isolated TBK1-binding domains ( TBDs ) inhibited the formation of the scaffold-kinase complexes ( Figure 5D , F , H ) as described previously [17] . Similar results were obtained for the interactions between the scaffold proteins and IKKε , which were not disrupted by C6 ( Figure S2A–C ) . Thus , C6 appears to associate with the scaffold proteins TANK , NAP1 and SINTBAD , without disrupting the formation of the signalling complexes containing the kinases TBK1 or IKKε . The expression of C6 protein during infection was investigated by infecting BSC-1 cells with VACV WR in the presence or absence of cytosine arabinoside ( AraC ) , an inhibitor of viral DNA replication and hence intermediate and late protein expression . Using a polyclonal antiserum raised against C6 protein expressed in Escherichia coli , a 17-kDa C6 protein was detected starting from 2 h post infection , with continued expression at all time points thereafter ( Figure 6A ) . Also , C6 was detected in the presence of AraC confirming its expression prior to DNA replication and hence as an early protein during infection , consistent with previous data for C6L mRNA expression [54] . In contrast , the expression of the late protein D8 [55] was blocked by the presence of AraC . To characterize the contribution of the C6 protein to VACV replication , spread and virulence , recombinant viruses that did or did not express the C6 protein were generated . These viruses included a C6L deletion virus ( vΔC6 ) lacking the C6 ORF , a plaque purified wild type virus ( vC6WR ) that was isolated from the same intermediate virus as the deletion mutant during transient dominant selection ( see methods ) , a revertant virus where the C6L ORF was re-inserted into the deletion virus at its natural locus ( vC6Rev ) , and an additional recombinant virus ( vC6FS ) where the C6 translational initiation codon was disrupted by the insertion of an adenine nucleotide . As expected , this virus , as well as vΔC6 , did not express C6 protein whereas vC6WR and vC6Rev both did ( Figure 6B ) . For localisation and interaction studies a virus expressing HA-tagged C6 was constructed ( vC6HA ) . Reduced levels of C6 were detected from this recombinant virus using the anti-C6 serum , perhaps due to the lower expression of this protein compared to wild-type C6 or poorer detection by the antiserum . Nevertheless this protein was detected using an antibody against the HA epitope ( Figure 6B ) , and shown to be functional in that it was capable of inhibiting IFN-β promoter induction when expressed from a plasmid ( data not shown ) . The intracellular localisation of C6 was investigated by biochemical fractionation of cells into cytoplasmic and nuclear fractions , followed by immunoblotting using the anti-C6 serum ( Figure 6C ) . The integrity of nuclear and cytoplasmic fractions was confirmed by blotting for lamin A and C , and for tubulin , respectively . The expression of C6 was detected in both nuclear and cytoplasmic fractions of cells infected with the wild-type virus expressing C6 ( vC6WR ) , the revertant virus ( vC6Rev ) or the virus expressing HA-tagged C6 ( vC6HA ) ( Figure 6C ) . The anti-C6 serum was not suitable for the detection of wild-type C6 by immunofluorescence . However , both nuclear and cytoplasmic localisation of C6 was also observed by confocal microscopy when cells infected with vC6HA were stained with an antibody against the HA tag ( Figure 6D ) . The isolation of the deletion mutant and vC6FS virus indicated that C6 is not essential for VACV replication . To determine whether loss of C6 had an effect on virus replication kinetics or spread , the plaque size and virus growth in cell culture were analysed . The size of the plaques formed 72 h post infection with the various recombinant viruses was measured in three different cell lines: African green monkey BSC-1 cells , rabbit RK-13 cells and human TK-143 cells . The absence of C6 had no effect on the mean plaque size in any of the cell types studied ( Figure S3A ) . To assess viral replication , BSC-1 cells were infected at a high ( 10 ) or low ( 0 . 01 ) multiplicity of infection ( m . o . i . ) with the set of recombinant viruses , and virus in the intracellular and extracellular fractions at various time points post infection was titrated by plaque assay ( Figure S3B–E ) . No significant difference was observed between the titres of either the extracellular or intracellular forms of the recombinant viruses . The contribution of C6 to VACV virulence was tested in two murine models of infection . In the intranasal ( i . n . ) model , groups of 10 BALB/c mice were infected with the recombinant viruses at 5×103 plaque forming units ( p . f . u . ) per animal and weight loss and signs of illness were recorded and compared . A significant difference in weight loss was observed between the viruses that did not express C6 ( vΔC6 and vC6FS ) and those that did express C6 ( vC6WR and vC6Rev ) between days 6 and 12 post infection ( Figure 7A ) , with the mice infected with the C6 deletion viruses losing less weight overall and gaining weight more quickly during recovery . The viruses lacking C6 also caused significantly fewer signs of illness in infected mice between day 5 and day 12 after infection ( Figure 7B ) . In addition , the recombinant viruses were used to infect groups of 10 BL/6 mice intradermally with 104 p . f . u . virus per ear , in both ears , and the sizes of the resulting lesions were measured and compared . The lesions induced by vΔC6 and vC6FS were significantly smaller than those induced by vC6WR and vC6Rev between 6 and 26 days post infection ( Figure 7C ) . In addition the lesions induced by the viruses lacking C6 began to heal sooner ( 11 days post infection ) than those induced by the viruses that did express C6 ( 14 days post infection ) ( Figure 7C ) . Thus , these data show that a virus lacking C6 is attenuated in vivo and indicate C6 is a virulence factor in two different models of infection .
Here the VACV protein C6 is described as a novel modulator of the innate immune system . Data presented show that C6 inhibits IFN-β expression by preventing the activation of the transcription factors IRF3 and IRF7 , while not affecting NF-κB activation . C6 acts at the level of the kinases TBK1 and IKKε , and is able to associate with the kinase-associated scaffold proteins NAP1 , TANK and SINTBAD . The immunomodulatory function of C6 is likely to be important during infection , as a deletion virus lacking C6 is attenuated in mouse models in vivo . C6 was identified as an inhibitor of the initiation of the IFN-β response in a screen of poorly characterised VACV proteins . That C6 might be an immunomodulator had been suggested by the previous observations that it belongs to a family of poxvirus proteins [37] whose members ( A46 , A52 , B14 , N1 and K7 ) were shown subsequently to belong to the Bcl-2 protein family and to have immunomodulatory activity ( reviewed in [38] ) . While the family members share structural similarity , their degree of amino acid similarity is low indicating they diverged long ago and although they share an ability to manipulate innate immune signalling pathways , they differ in their targets and mechanisms of action . While A46 acts at the interface between TLRs and their adaptors [56] , A52 targets the more downstream signalling factors TRAF6 and IL-1 receptor associated kinase 2 ( IRAK2 ) [57] . B14 , K7 and now C6 all appear to target the kinase complexes located at the point of convergence between several different PRR signalling pathways . However , each of the viral proteins targets distinct components of these complexes . B14 binds IKKβ and thereby inhibits IκB phosphorylation and NF-κB activation [47] . In contrast , K7 inhibits IRF3 phosphorylation by binding to the helicase DDX3 , which is part of the complexes containing TBK1 and IKKε [58] . In this paper C6 is shown to inhibit the activation of IRF3 and IRF7 in a different way to K7 , namely by interacting with TANK , NAP1 and SINTBAD . These three proteins act as scaffold proteins that associate constitutively with TBK1 and IKKε [17] , [59] , [60] . They are essential for the innate immune response to several different viruses and PAMPs , and in particular for the activation of IRFs , but not NF-κB , upon stimulation with Sendai virus or poly ( I∶C ) [16]–[18] . The observation that the scaffold proteins are targeted by a viral immunomodulator provides additional evidence for the importance of these proteins in the antiviral response . The precise function of TANK , NAP1 and SINTBAD in the process of TBK1 and IKKε activation has not been defined fully , but there is evidence that the adaptor proteins link the kinases to the upstream signalling pathways , possibly by interaction with TRAF3 , which is a component of TLR and RLR signalling pathways [16] . The recruitment of the kinase complexes to TRAF3 would then require the scaffold proteins , and lead to the phosphorylation and activation of TBK1 and IKKε , ultimately leading to the phosphorylation of IRF3 and IRF7 . However , how exactly complex formation is linked to the activation of the kinases , and which functions of the adaptors may be redundant or unique , has yet to be elucidated . NAP1 , TANK and SINTBAD are related in domain structure , possessing N-terminal coiled-coil regions that are important for homodimerization , and a central TBD , which mediates the interaction with TBK1 and IKKε [17] . While C6 interacted with all three scaffold proteins , it did not affect their interaction with either TBK1 or IKKε . In contrast , truncated proteins containing only the TBD of either NAP1 , TANK or SINTBAD inhibited all the scaffold-kinase interactions in this assay , as described previously [17] . Thus the exact mechanism whereby C6 disrupts IRF activation remains to be determined . It is possible that C6 prevents the association of the scaffold-kinase complexes with TRAF3 , or else prevents the activation of TBK1 and IKKε once the complexes are formed . Also , it is possible that the interaction between C6 and the scaffold proteins is indirect and is mediated by additional proteins that may be part of the kinase signalling complexes . Further analysis of the effect of C6 on these protein complexes may shed some light on the mechanism by which TBK1 and IKKε are activated - and inhibited - during viral infection . VACV is not the only virus that inhibits the TRAF3-scaffold-kinase axis . Recently it was shown that the M protein of severe acute respiratory syndrome ( SARS ) coronavirus targets a complex containing TRAF3 , and prevents the association of TRAF3 with TBK1 , IKKε or TANK [61] . Like C6 , the M protein inhibits the induction of the IFN-β promoter by inhibiting IRF3 activation [61] . However to our knowledge C6 is the first viral protein shown to associate with all three scaffold proteins . VACV expresses several proteins that inhibit IRF activation , including the related Bcl-2-like protein K7 , which also targets a TBK1-containing protein complex [58] . However , despite this , the effects of K7 and C6 are evidently not duplicative , because when K7 is still expressed , loss of C6 caused a marked virus attenuation in two models of infection . Similarly , there are several VACV proteins that inhibit NF-κB activation , for instance A52 , A46 , N1 , B14 . Yet deletion of any single member of this group causes virus attenuation suggesting non-redundant functions . Possible explanations for this non-redundancy might be cross talk between different pathways , so the outcome of blocking a pathway is influenced by the point at which a virus inhibitor functions to block the pathway . Alternatively , the virus proteins might have multiple functions as has been demonstrated for VACV protein N1 . The need for the virus to express so many different non-redundant viral inhibitors of host signalling cascades may be due to the host innate immune system being able to partially compensate for the inhibition of an individual signalling component by using parallel pathways all ultimately leading to the induction of type I IFNs and pro-inflammatory cytokines . Furthermore , the importance of one particular signalling pathway or component may vary depending on the cell type infected or stage of infection , thus requiring the inhibition of several , seemingly redundant signalling proteins . Finally , it is plausible that the inhibition exerted by a single viral protein in not complete , particularly at early stages of infection , thus requiring the expression of several different factors targeting components of the same pathway to have an additive effect . The characterization of poxvirus proteins that inhibit the innate immune system is of interest , since elucidating the mechanism of viral inhibition often reveals new insights into how innate immunity operates . Furthermore , VACV strains are in development as vaccine delivery systems against smallpox and other pathogens ( reviewed in [62] ) , and as vectors for the targeting of tumours and for gene therapy ( reviewed in [63] ) . It is shown here that deletion of a single ORF can attenuate the virus , thus possibly making it safer for clinical use . Also , the deletion of ORFs that encode inhibitors of the innate immune system would be predicted to make the virus more immunogenic , and thus make a more effective vaccine . In this context , it is interesting to note that C6 is conserved in most OPVs and that the MPXV orthologue of C6 is functionally equivalent to the C6 protein from VACV strains WR .
For the screen of VACV candidate immunomodulators , 49 ORFs were selected from the VACV WR strain genome and amplified by PCR from VACV WR genomic DNA isolated by phenol-chloroform extraction from purified viral cores . Candidate ORFs including C6L and B14R were cloned into the expression vector pCMV-HA ( Clontech ) . C6 was also subcloned into pLENTI-Dest-V5 ( Invitrogen ) for immunoflourescence experiments . The ORF encoding MPXV C6 was amplified by PCR from DNA extracted from MPXV-infected HeLa cells ( a kind gift from K . Rubins , Whitehead Institute ) and cloned into pCMV-HA ( Clontech ) . IFN-β-promoter luciferase reporter was a gift from T . Taniguchi ( University of Tokyo , Japan ) and NF-κB luciferase was from R . Hofmeister ( University of Regensburg , Germany ) . ISRE-Luciferase and pFR-Luciferase were purchased from Promega . GL3-Renilla vector was made by replacing the firefly luciferase ORF from pGL3-control ( Promega ) with the renilla luciferase ORF from pRL-TK ( Promega ) . FLAG- and luciferase fusions with signalling proteins for the LUMIER assay were described in [17] . IKKα was from Tularik Inc . Vectors expressing IRF3-Gal4 , IRF7-Gal4 , TBK1 , IKKε and TRIF were a kind gift from K . A . Fitzgerald ( University of Massachusetts Medical School , USA ) , MAVS was from T . J . Chen ( University of Texas Southwestern Medical Centre , USA ) , MyD88 was from M Muzio ( Milan , Italy ) , TLR3 was from D . T . Golenbock ( University of Massachusetts Medical School , USA ) , and IRF3-5D and IRF7-4D were from J . Hiscott ( McGill University , Montreal , Canada ) . For construction of the C6 deletion virus 250-bp flanking regions of the C6L gene were amplified by PCR from VACV WR genomic DNA , ligated together and inserted into a plasmid containing the Escherichia coli guanylphosphoribosyl transferase ( Ecogpt ) gene fused in-frame with the enhanced green fluorescent protein ( EGFP ) gene ( Z11ΔC6 ) . For construction of Z11C6rev , Z11C6FS and Z11C6HA , C6L , C6L with an additional adenine nucleotide in the start codon or C6L with a C-terminal HA tag respectively , plus C6L flanking regions were amplified from VACV WR genomic DNA and inserted into the Z11 plasmid . C6 polyclonal antiserum was raised against C6 protein purified from Escherichia coli and injected into rabbits ( Eurogentec ) . Other antibodies were from the following sources: IRF3 ( IBL ) , V5 ( Cell Signaling ) , p65 ( Santa Cruz ) , IgG from rabbit serum ( Sigma ) , TBK1 ( Cell Signaling ) , IKKε ( Abcam ) , TANK ( Abcam ) , FLAG ( Sigma ) , Lamins A+C ( Abcam ) , tubulin ( Upstate Biotech ) . The mouse monoclonal antibody AB1 . 1 against D8 has been described previously [64] . Poly ( I∶C ) and poly ( dA-dT ) were from Sigma , TNF , IL-1 , CLO75 and R848 were from Invivogen . HEK293 cells were grown in Dulbecco's Modified Eagle's Medium ( DMEM , GIBCO ) supplemented with 10% fetal bovine serum ( FBS , Biosera ) and 10 µg/ml ciprofloxacin ( Sigma ) . BSC-1 cells were maintained in DMEM supplemented with 10% FBS and penicillin/streptomycin ( P/S ) ( 50 µg/ml ) . RK-13 and TK-143 cells were maintained in minimum essential medium ( MEM ) supplemented with 10% FBS and P/S ( 50 µg/ml ) . NIH 3T3 cells were maintained in DMEM supplemented with 10% newborn bovine serum ( NBS ) and P/S ( 50 µg/ml ) . HeLa cells were maintained in MEM supplemented with 10% FBS , 1∶100 non-essential amino acids ( NEAA ) ( Sigma ) and P/S ( 50 µg/ml ) . C6 recombinant viruses were constructed using the transient dominant selection method [65] . For construction of vΔC6 , RK-13 cells were infected with VACV strain WR at 0 . 01 p . f . u . per cell and then transfected with the Z11ΔC6 plasmid using polyethylenimine ( PEI ) ( 1 mg/ml ) according to the manufacturer's instructions . Progeny virus was harvested after 48 h and used to infect RK-13 cells in the presence of mycophenolic acid ( MPA , 25 µg/ml ) , hypoxanthine ( HX , 15 µg/ml ) and xanthine ( X , 250 µg/ml ) . EGFP-positive plaques were selected and purified by three rounds of infection using RK-13 cells in the presence of MPA , HX and X as above . Intermediate virus was resolved in BSC-1 cells by three rounds of infection in the absence of MPA , HX and X . The genotype of resolved viruses was analysed by PCR following proteinase K-treatment of infected RK-13 cells . Revertant viruses were constructed in a similar manner by transfection of plasmid Z11C6rev ( for vC6 rev ) , Z11C6FS ( for vC6FS ) or Z11C6HA ( for vC6HA ) into vΔC6 infected cells . Luciferase reporter gene assays were performed in HEK293 cells seeded in 96-well plates and transfected with 0 . 8 µl Genejuice ( Merck ) per well . Firefly reporter plasmid ( 60 ng ) , 20 ng GL3-Renilla control plasmid and 150 ng expression vector or empty vector control were used per well . For the IRF3 and IRF7 reporter gene assays , 60 ng pFR-Luciferase and 20 ng pGL3-Renilla were transfected together with 4 ng of IRF3-Gal4 or IRF7-Gal4 , and 150 ng expression vector or control plasmid . For luciferase assays in NIH3T3 cells , cells were seeded in 96-well plates and transfected with 60 ng IFN-β-luciferase , 10 ng pRL-TK , and 250 ng expression vector or empty vector control . Cells were lysed in Passive Lysis Buffer ( Promega ) , and firefly luciferase activity was normalized to renilla luciferase activity . Experiments were performed in triplicate and repeated at least 3 times . RNA from HEK293 cells grown in 12-well plates was extracted using the RNeasy kit ( QIAGEN ) , and converted to cDNA using the Quantitect RT kit ( QIAGEN ) . IFN-β mRNA was quantified by real-time PCR with the TaqMan gene expression assay Hs00277188_s1 and a β-actin endogenous control VIC-MGB probe ( 6-carboxyrhodamine–minor groove binder; Applied Biosystems ) . Experiments were performed in triplicate . Cell culture supernatants from HEK293 cells grown in 96-well plates were assayed for CCL5 protein using Duoset reagents ( R&D Biosystems ) . HEK293 cells were grown on glass coverslips and fixed with 4% paraformaldehyde in PBS . Cells were permeabilized in 0 . 5% Triton in PBS , pre-incubated for 1 h in blocking buffer ( 5% BSA , 0 . 05% Tween-20 in PBS ) , stained for 3 h with primary antibody ( 1∶300 in blocking buffer ) and for 1 h with Alexa488 or Alexa647-labelled secondary antibodies ( 1∶500 , Invitrogen ) . Coverslips were mounted in MOWIOL 4-88 ( Calbiochem ) containing DAPI ( 4 , 6-diamidino-2-phenylindole; 1 µg/ml ) . Images were taken on an Olympus FV1000 scanning confocal microscope . For co-immunoprecipitation , HEK293T cells were grown in 10-cm dishes and co-transfected with vectors expressing FLAG-tagged proteins and a vector expressing TAP-tagged ( consisting of FLAG and Streptavidin epitopes ) C6 , using Fugene-6 ( Roche ) , or transfected with the vectors expressing FLAG-tagged proteins alone and then infected 48 h later with vC6HA ( 2 p . f . u . per cell ) for 16 h . Cells were lysed in lysis buffer ( 0 . 1 ( v/v ) % Triton X-100 , 150 mM NaCl , 10% glycerol , 10 mM CaCl2 , 20 mM Tris-HCl pH 7 . 4 and protease inhibitors ) , pre-cleared by centrifugation and incubated with 30 µl of anti-FLAG M2 agarose beads ( Sigma ) , or strepavidin agarose beads ( Thermofisher ) for 3 h . Immunoprecipitates were washed 3 times in lysis buffer , and eluted from the beads by boiling in sample buffer containing SDS . Proteins were resolved by SDS-polyacrylamide gel electrophoresis ( PAGE ) and detected by immunoblotting . For the LUMIER assay , HEK293 cells were grown in 6-well plates and transfected with 0 . 5 µg FLAG-tagged plasmid , 0 . 5 µg luciferase-tagged plasmid and 3 µg C6 expression vector , TBD expression vector or empty vector control using GeneJuice ( Merck ) . Cells were harvested 24 h later in Passive Lysis Buffer ( Promega ) , and subjected to immunoprecipitation using 0 . 3 µl FLAG antibody pre-coupled to Protein A sepharose beads . Immunoprecipitates were washed 5 times in lysis buffer and eluted using 100 µM FLAG peptide ( Sigma ) in PBS , and renilla luciferase activity was measured . HeLa cells were infected with recombinant VACVs at 5 p . f . u . per cell for 16 h . The cells were washed twice in ice-cold LS buffer ( 20 mM Hepes pH 7 . 8 , 0 . 5 mM DTT , 0 . 5 mM MgCl2 in water ) and allowed to swell on ice for 20 min . The cells were gently scraped and disrupted by Dounce homogenisation on ice . The lysates were centrifuged at 600× g for 2 min at 4°C to pellet the nuclei . The supernatant ( cytoplasmic fraction ) was removed . The nuclei were washed five times in PBS , placed in nuclei resuspension buffer ( 50 mM Tris-HCl pH 8 , 0 . 5 mM MgCl2 , 20 mM iodoactetamide supplemented with protease inhibitor ( Roche ) ) and sonicated . Proteins were resolved by SDS-PAGE and detected by immunoblotting . For the single-step growth curves BSC-1 cells were infected with 10 p . f . u . per cell . At 0 h , 12 h and 24 h post infection the medium was removed and the cells were collected by centrifugation at 500× g for 10 min . The supernatant was removed and extracellular virus titres were determined by plaque assay on BSC-1 cells . For intracellular virus , cells were scraped , collected by centrifugation and subjected to three rounds of freeze-thawing before determining viral titre by plaque assay . For the multi-step growth curve BSC-1 cells were infected with 0 . 01 p . f . u . per cell and intracellular and extracellular virus was harvested at 0 , 12 , 48 and 72 h post infection as described above . RK-13 , BSC-1 and TK-143 cell monolayers were infected in duplicate with virus at 50 p . f . u . per well for 72 h to allow formation of well separated plaques . The cells were washed once with PBS and stained for 30 min with crystal violet ( 5% ( v/v ) crystal violet solution ( Sigma ) , 25% ( v/v ) ethanol ) . Wells were then washed with water and the sizes of six plaques per well were measured using Axiovision 4 . 6 software and a Zeiss Axiovert 200 M microscope . Female BALB/c mice ( n = 10 , 6–8 weeks old ) were infected intranasally ( i . n . ) with 5×103 p . f . u . and monitored as described previously [66] , [67] . Female C57BL/6 mice ( n = 10 , 6–8 weeks old ) were inoculated intradermally ( i . d . ) in the ear pinnae with 104 p . f . u . as described previously [68] , [69] . Data were analysed using Unpaired Student's T tests . Statistical significance is expressed as follows: * P<0 . 05 , ** P<0 . 01 , *** P<0 . 001 . VACV WR C6 , P17362 . 1; TBK1 , AF191838_1; IKKε , NP_054721; NAP1 , AAO05967; TANK , NP_001186064; SINTBAD , NP_055541; IKKα , NP_001269; MyD88 , AAC50954; MAVS , Q7Z434 . 2; IRF3 , AAH09395; IRF7 , NP_001563; NF-κB p65 , CAA80524; IFN-β , NC_000009 . 11; CCL5 , NC_000017 . 10 . | A key event in the innate immune response to virus infection is the detection of pathogen-associated molecular patterns ( PAMPs ) such as viral DNA and RNA by cellular pattern recognition receptors ( PRRs ) . This leads to expression of interferon-β ( IFN-β ) by an infected cell . Many viruses have evolved mechanisms to evade the induction of IFN-β . Here a screen of poorly characterized vaccinia virus ( VACV ) proteins identified protein C6 as an inhibitor of IFN-β induction by PRRs . Data presented show that C6 prevents the activation of the transcription factors IRF3 and IRF7 by the kinases TBK1 and IKKε , which are key components at the point of convergence of several PRR signalling pathways . C6 interacts with the scaffold proteins NAP1 , TANK and SINTBAD , which are components of the protein complexes containing TBK1 and IKKε , and this interaction might modulate the activity of these kinases . C6 is expressed early during infection and contributes to virulence because viruses that do not express C6 are attenuated in two in vivo models compared to wild type and revertant control viruses . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"signal",
"transduction",
"molecular",
"cell",
"biology",
"viral",
"immune",
"evasion",
"immunity",
"virology",
"innate",
"immunity",
"biology",
"microbiology",
"molecular",
"biology"
] | 2011 | Vaccinia Virus Protein C6 Is a Virulence Factor that Binds TBK-1 Adaptor Proteins and Inhibits Activation of IRF3 and IRF7 |
Mitochondrial dysfunction activates the mitochondrial retrograde signaling pathway , resulting in large scale changes in gene expression . Mitochondrial retrograde signaling in neurons is poorly understood and whether retrograde signaling contributes to cellular dysfunction or is protective is unknown . We show that inhibition of Ras-ERK-ETS signaling partially reverses the retrograde transcriptional response to alleviate neuronal mitochondrial dysfunction . We have developed a novel genetic screen to identify genes that modify mitochondrial dysfunction in Drosophila . Knock-down of one of the genes identified in this screen , the Ras-ERK-ETS pathway transcription factor Aop , alleviates the damaging effects of mitochondrial dysfunction in the nervous system . Inhibition of Ras-ERK-ETS signaling also restores function in Drosophila models of human diseases associated with mitochondrial dysfunction . Importantly , Ras-ERK-ETS pathway inhibition partially reverses the mitochondrial retrograde transcriptional response . Therefore , mitochondrial retrograde signaling likely contributes to neuronal dysfunction through mis-regulation of gene expression .
The use of ATP as a universal currency of energy transfer makes this molecule essential for life . ATP is generated either by glycolysis in the cytosol , or through the action of the tricarboxylic acid ( TCA ) cycle and β-oxidation of fatty acids coupled to oxidative phosphorylation ( OXPHOS ) in mitochondria . The mitochondrial electron transport chain ( ETC ) couples the transfer of electrons to the pumping of protons into the inter-membrane space [1] . This creates a membrane potential ( ΔΨ ) , which is used by the mitochondrial ATP synthase to convert ADP to ATP [2] . Under normal aerobic conditions mitochondria use OXPHOS to generate the majority of cellular ATP . Mitochondria also metabolize fatty acids , synthesize amino acids , buffer cellular calcium ions ( Ca2+ ) , produce the majority of cellular reactive oxygen species , synthesise iron-sulphur clusters and mediate programmed cell death . Mitochondria are abundant in almost every cell type and are particularly important in the nervous system . During nervous system development neural stem cell progenitors use mainly glycolytic metabolism , but upon differentiation into neurons switch to become dependent on OXPHOS for ATP [3 , 4] . Mature neurons must maintain their membrane potential through the action of ATP dependent sodium potassium pumps and so are vitally dependent on mitochondrial function . Failure of mitochondria and breakdown of the cellular power supply can lead to human disease [5 , 6] . Primary mitochondrial diseases are caused by mutations in mitochondrial genes , either in the nuclear or mitochondrial genome [7] . These diseases are rare , but can be severe and debilitating , often affecting the nervous system or muscle . The mitochondrial disease Leigh syndrome , caused by mutations in subunits of mitochondrial ETC complex I or IV , causes degeneration of the nervous system and death in infants [8] . Mitochondrial dysfunction is also strongly implicated in neurodegenerative diseases and may be a common pathway in Alzheimer’s disease , Parkinson’s disease , Amyotrophic lateral sclerosis and Huntington’s disease [9] . Reduced mitochondrial ETC complex I activity in the substantia nigra pars compacta is a hallmark of Parkinson’s disease and toxins that inhibit complex I cause dopaminergic neuron cell death and parkinsonian phenotypes in humans and rodent models [10–14] . Moreover , mutations in two genes involved in mitochondrial quality control , PINK1 and Parkin , cause familial Parkinson’s disease [15] . There is also evidence supporting a role for mitochondrial dysfunction in Alzheimer’s , as patients have been shown to have mitochondrial ETC complex IV and V deficits and mitochondrial DNA mutations have been associated with this disease [16–20] . Cells respond to changes in mitochondrial function by altering nuclear gene expression , a process known as mitochondrial retrograde signaling [21–23] . The mechanism of retrograde signaling varies depending on the organism and cell type . In yeast the retrograde response is triggered by decreased glutamate , whose synthesis is reduced due to failure of the TCA cycle to synthesise α-ketoglutarate , the precursor of glutamate [24] . This retrograde signaling pathway culminates in the partial dephosphorylation of the transcription factor Rtg3 which , together with its binding partner Rtg1 , relocalises to the nucleus to activate gene expression [25] . In other contexts , mitochondrial dysfunction can trigger a retrograde signaling pathway known as the mitochondrial unfolded protein response ( UPRmt ) . When activated , the UPRmt induces the expression of nuclear encoded mitochondrial chaperone proteins and proteases [26] . In C . elegans , activating transcription factor associated with stress-1 ( ATFS-1 ) is normally localised to mitochondria , but upon mitochondrial dysfunction a fraction of ATFS-1 localises to the nucleus where it regulates the expression of UPRmt genes [27 , 28] . ATFS-1 is not conserved in mammals and the RTG genes are not present in metazoans , but analogous retrograde signaling pathways have been identified that enable mitochondria to reprogram nuclear gene expression [29 , 30] . Transcriptional studies have shown that mitochondrial dysfunction elicits large scale changes in nuclear gene expression in diverse cell types in a range of model systems [22 , 23 , 31] . However , the molecular basis of mitochondrial signaling is still poorly understood , particularly in the nervous system [29 , 30] . Importantly , it is not known whether the retrograde transcriptional response contributes to mitochondrial dysfunction phenotypes or is protective . To address these questions , we performed a genetic screen and identified 30 genes that modify mitochondrial dysfunction in the Drosophila wing , implicating several new pathways in the mitochondrial retrograde response . Manipulation of one of the identified pathways , Ras-ERK-ETS signaling , also alleviates the effects of mitochondrial dysfunction in the Drosophila nervous system . Transcriptomic and functional analyses suggest that mitochondrial retrograde signaling is reversed and transcriptionally reprogrammed by Ras-ERK-ETS inhibition to restore neuronal function .
Mitochondrial retrograde signaling is activated in response to mitochondrial dysfunction . To induce mitochondrial dysfunction , we manipulated the expression levels of the mitochondrial DNA binding protein/transcription factor TFAM . TFAM expression is essential for mitochondrial DNA maintenance ( S1A Fig ) and gene expression , but overexpression of TFAM in mice and human cells also causes reduced mitochondrial gene transcription and mitochondrial dysfunction [32–35] . In Drosophila , both ubiquitous TFAM knock-down ( S1B–S1E Fig ) or TFAM overexpression [36] cause reduced mitochondrial gene expression and lethality at the larval stage . To develop a rapid , genetically modifiable assay for mitochondrial dysfunction in vivo we tested TFAM knock-down or TFAM overexpression in the wing using MS1096-Gal4 . Strong TFAM overexpression in the wing causes late pupal lethality , while weak TFAM overexpression or TFAM knock-down ( using several independent RNAi transgenes ) results in a curved adult wing phenotype ( Fig 1B , S1F , S1G , S1J–S1L Fig ) . Importantly , the TFAM knock-down curved wing phenotype is enhanced by heterozygosity for a loss-of-function mutation in TFAM ( TFAMc01716 ) ( Fig 1C , S1H Fig ) and almost completely rescued by co-expression of TFAM ( S1M–S1P Fig ) , proving that this RNAi phenotype is a result of reduced TFAM expression . Although neither TFAM knock-down or overexpression change ATP levels ( S2A–S2K Fig ) , or reactive oxygen species ( ROS ) ( S2L–S2S Fig ) in the developing wing , both cause reduced mitochondrial numbers and increased apoptosis ( Fig 1D–1G , S2T–S2X Fig ) . We previously showed that the gene Thor , encoding the eukaryotic initiation factor 4E binding protein ( 4E-BP ) , is a mitochondrial retrograde signaling response gene in neurons [36] . Both knock-down and overexpression of TFAM in the wing imaginal disc cause increased Thor expression ( Fig 1H–1K ) . TFAM overexpression causes a strong increase in Thor expression throughout and beyond the dorsal compartment , while TFAM knock-down causes increased Thor expression in discreet patches within the dorsal compartment ( Fig 1H–1K ) . As a result , using qRT-PCR from whole wing discs we could only detect increased Thor expression , or increased expression of the mitochondrial unfolded response pathway target gene Hsp22 [37] with TFAM overexpression ( S2Y and S2Z Fig ) . These data show that mitochondrial dysfunction caused by TFAM knock-down and TFAM overexpression activate mitochondrial retrograde signaling in the wing . In order to perform a genetic screen for modifiers of mitochondrial retrograde signaling , flies were generated that stably express the TFAM RNAi transgene , together with the TFAMc01716 mutation , in the wing . These flies , referred to as ‘MitoMod’ for ‘Mitochondrial Modifier’ , have a distinctive ~45° curve at the tip of the wing ( Fig 1C ) . To test the sensitivity of MitoMod flies to mitochondrial perturbation they were crossed to lines carrying RNAi or overexpression transgenes for genes associated with familial Parkinson’s disease that have mitochondrial associated functions . Transgenes that cause a wing phenotype when expressed on their own were excluded to avoid additive effects ( S1 Table ) . Overexpression or knock-down of Pink1 , overexpression of parkin , knock-down of Lrrk and overexpression and knock-down of DJ-1α and DJ-1β all enhance the MitoMod wing phenotype ( Fig 1L and 1M , S3 Fig and S1 Table ) . Therefore , the MitoMod wing phenotype provides a sensitised background for identifying mitochondrial retrograde signaling genes in vivo . MitoMod flies were used to perform a genetic modifier screen of 646 RNAi lines , targeting 579 genes ( Fig 2A ) . This RNAi collection was enriched for lines that target genes expressed in the nervous system and genes encoding chromatin remodelling factors ( S2 Table ) . RNAi lines that cause a phenotype when expressed alone in the wing were excluded ( S2 Table ) , to avoid additive effects with the MitoMod wing phenotype . Stringent criteria were used to identify interacting genes: only RNAi lines that caused a strong reproducible enhancement or suppression , which was also replicated by an independent RNAi line targeting the same gene were classed as hits . 25 genes were identified that enhance the MitoMod wing phenotype ( Fig 2F–2I , S3 Table ) . Gene ontology ( GO ) analysis shows that these genes are involved in a range of biological processes and functions ( Fig 2J , S3 Table ) . Five genes were identified that suppress the MitoMod phenotype ( Fig 2C–2E , S4 Table ) . The suppressor genes function in chromatin remodelling or transcriptional regulation ( Fig 2K , S4 Table ) . Overall , the variety in function of genes identified in the screen is consistent with the multifunctional cellular roles of mitochondria . To test whether suppression of the mitochondrial dysfunction phenotype in the adult wing reflects reduced apoptosis , cleaved Drosophila effector caspase ( Dcp-1 ) expression was analysed in the wing during development . Knock-down of TFAM in the wing causes increased Dcp-1 expression , but this is reduced to control levels by knock-down of the suppressors identified in the screen Aop , Ino80 , Chrac-14 , Ing3 and MTA1-like ( S4A–S4I Fig ) . Moreover , knock-down of Aop or Ino80 suppress the increase in apoptosis caused by TFAM overexpression ( S4J Fig ) . These data show that suppression of the adult wing phenotype in MitoMod flies reflects a reduction in apoptosis . The effects of mitochondrial dysfunction are particularly acute in the nervous system and manipulation of retrograde signaling may be a potential strategy to alleviate these effects . We aimed to use the wing screen to identify genes involved in mitochondrial retrograde signaling in the nervous system . Aop ( anterior open , also known as Yan ) , one of the suppressor genes identified ( Fig 2K ) , is an E-twenty six ( ETS ) transcription factor and a target of the highly conserved Ras-ERK ( mitogen-activated protein kinase ( MAPK ) ) pathway . Treatment of cultured neuronal cells with the mitochondrial uncoupler carbonyl cyanide p- ( trifluoromethoxy ) phenylhydrazone ( FCCP ) causes aberrant ERK activation [38] . We therefore hypothesised that Ras-ERK signaling is mis-regulated by mitochondrial dysfunction in the nervous system and that manipulation of Ras-ERK signaling would modify the effects of neuronal mitochondrial dysfunction . TFAM knock-down in the wing causes similar but weaker effects to TFAM overexpression ( S1J–S1L Fig ) . We have previously shown that TFAM overexpression in motor neurons causes a dramatic reduction in pre-synaptic tetramethylrhodamine methyl ester ( TMRM ) positive and mitochondrial-GFP labelled mitochondrial number and volume and a robust adult climbing phenotype [36] . Overexpression of TFAM in neurons with D42-Gal4 also causes a failure of wing inflation in approximately 50% of flies , due to dysfunction of the CCAP neurons that release the neuropeptide bursicon , which activates wing inflation ( S5L–S5N Fig ) [36] . TFAM knock-down in motor neurons also causes a reduction in mitochondrial volume and reduced adult climbing , but these phenotypes are much weaker than with TFAM overexpression ( S5A–S5G Fig ) and TFAM knock-down does not affect wing inflation . We therefore used TFAM overexpression as a model to test whether targeting Ras/ERK signaling modifies the effects of mitochondrial dysfunction in neurons . Consistent with the suppressive effect in the screen , knock-down of Aop in neurons ( using a validated RNAi transgene ( S5H–S5K Fig ) ) suppresses the TFAM overexpression climbing and wing inflation phenotypes ( Fig 3A and 3B ) . Furthermore , knock down of the Ras-ERK pathway components Downstream of raf1 ( Dsor; MAPK ) or the MAPK Rolled ( Rl; MAPK kinase ) both suppressed the TFAM overexpression climbing and wing inflation phenotypes ( S5O–S5V Fig ) , confirming that inhibition of Ras-ERK signalling alleviates the effects of neuronal mitochondrial dysfunction . To test whether increasing Aop affects neuronal function we overexpressed Aop alone in motor neurons , or together with TFAM . Overexpression of Aop in neurons causes a strong climbing deficit and complete failure of wing inflation ( Fig 3C and 3D ) . Moreover , overexpression of Aop and TFAM together causes pupal lethality ( Fig 3C and 3D ) , which is not affected by heterozygosity for the Ras85D loss-of-function allele Ras85DΔC40B , consistent with Aop acting downstream of Ras . The ETS domain transcription factor pointed ( Pnt ) is a second target of Ras-ERK signaling in Drosophila . To test whether reducing Pnt expression also modifies neuronal mitochondrial dysfunction we used both a validated RNAi targeting Pnt ( S6A–S6D Fig ) and heterozygosity for a loss-of-function mutation in pnt ( pntΔ88 ) . Knock-down of Pnt in motor neurons , or heterozygosity for pnt , suppresses the TFAM overexpression climbing and wing inflation phenotypes ( Fig 3E and 3F , S6E and S6F Fig ) . Combining Pnt knock-down with heterozygosity for Ras85D causes significantly increased suppression of TFAM overexpression phenotypes compared to either condition alone ( S6G and S6H Fig ) . However , this is not the case for Aop knock-down and Ras85DΔC40B ( S6I and S6J Fig ) . Simultaneous knock-down of Pnt and Aop with TFAM overexpression does not significantly improve the climbing phenotype , but significantly improves the wing inflation phenotype compared to TFAM overexpression with Aop knock-down ( S6K and S6L Fig ) . PntP2 overexpression in motor neurons causes a climbing deficit and lethality in combination with TFAM overexpression ( S6M Fig ) , which is not affected by heterozygosity for the Ras85D loss-of-function allele Ras85DΔC40B , consistent with Pnt acting downstream of Ras . Therefore , reduced expression of the Ras-ERK pathway components Aop and Pnt partially overcomes the damaging effects of mitochondrial dysfunction in motor neurons . Aop and Pnt are the main transcriptional effectors of the Ras-ERK pathway . To test whether directly activating Ras modulates neuronal activity in neurons with inhibited mitochondrial function we used a constitutively active form of Ras , RasV12 . Expression of Ras85DV12 in motor neurons causes lethality with D42-Gal4 and with OK371-Gal4 . Knock-down of either Aop or Pnt suppresses the Ras85DV12 lethality phenotype ( S6N and S6O Fig ) , while Ras85DV12 combined with Aop overexpression is lethal . Therefore , Ras-ERK pathway activation mimics the effects of mitochondrial dysfunction and Aop and Pnt both act as positive regulators of the pathway in motor neurons . We next looked for direct evidence of Ras-ERK-ETS pathway mis-regulation in neurons caused by mitochondrial dysfunction . Motor neurons overexpressing TFAM were stained with an antibody that recognises the activated ( di-phosphorylated ) form of ERK ( dpERK ) . dpERK expression is significantly increased in larval motor neurons overexpressing TFAM ( Fig 4 ) . Clonal analysis of TFAM overexpression shows that activation of dpERK is both cell autonomous and non-cell autonomous ( S6P and S6Q Fig ) . Therefore , Ras-ERK-ETS signaling is activated in response to mitochondrial dysfunction in Drosophila neurons . Similar to mitochondrial dysfunction in late larval sensory neurons , which does not affect ATP levels due to a compensatory increase in glycolysis [39] , overexpression of TFAM in motor neurons does not alter ATP levels ( S7A–S7C Fig ) . Active zones are the sites of pre-synaptic neurotransmitter release at chemical synapses and are enriched for the protein complexes that regulate synaptic vesicle release and recycling . TFAM overexpression in motor neurons causes altered mitochondrial morphology in the cell body , a dramatic loss of pre-synaptic mitochondria and a reduction in the number of active zones at the larval neuromuscular junction ( NMJ ) [36] . Therefore , to further investigate how modulation of Ras-ERK-ETS signaling affects neuronal mitochondrial dysfunction we focused on the synaptic compartment . Neither knock-down of Aop or Pnt affects the severe loss of pre-synaptic mitochondria caused by TFAM overexpression ( Fig 5A–5F . S7D and S7E Fig ) , suggesting that Ras-ERK-ETS pathway inhibition does not alter the primary mitochondrial defect . Knock-down of Aop does not rescue the active zone phenotype caused by TFAM overexpression , but Aop knock-down alone causes a reduction in active zone number ( S7F Fig ) , complicating the interpretation of this result . However , knock-down of Pnt fully rescues the active zone phenotype caused by TFAM overexpression ( Fig 5G and 5K ) . These data show that Ras-ERK-ETS pathway inhibition does not affect the primary mitochondrial defect , but modifies the active zone phenotype caused by neuronal mitochondrial dysfunction . Mitochondrial dysfunction in humans causes rare primary mitochondrial diseases and is also associated with more common neurodegenerative diseases , including Parkinson’s . We hypothesised that targeting the retrograde response , through inhibition of Ras-ERK-ETS signaling , would be beneficial in Drosophila models of human disease associated with mitochondrial dysfunction . To test this we used pan-neuronal ( nSyb-Gal4 ) knock-down of the OXPHOS complex IV subunit Surf1 , a model for the primary mitochondrial childhood encephalomyelopathy Leigh syndrome , and park25 homozygous mutant flies , a model for familial Parkinson’s disease [36 , 40 , 41] . dpERK expression is increased in the ventral nerve cord ( VNC ) by pan-neuronal knock-down of Surf1 , but not in park25 homozygous larvae ( S7G and S7H Fig ) , possibly because of the mild effect of loss of Parkin in neurons . Knock-down of Aop , knock-down or heterozygosity for pnt all suppress the severe climbing phenotype and rescue the wing inflation defect caused by pan-neuronal knock-down of Surf1 ( Fig 6A–6D , S8A and S8B Fig ) . Knock-down of Aop does not significantly improve climbing in the park25 mutant ( S8C Fig ) . However , heterozygosity for pnt does suppress the climbing phenotype in park25 mutant flies ( Fig 6G ) . To test whether the beneficial effects in these two models are specific to reduced Aop and Pnt expression , or a general property of Ras-ERK-ETS pathway inhibition , we used a loss-of-function allele of Ras85D ( Ras85DΔC40B ) . Heterozygosity for Ras85D suppresses the climbing and wing inflation defects caused by knock-down of Surf1 ( Fig 6E and 6F ) and the climbing deficit in park25 homozygous flies ( Fig 6H ) . Thus , targeting the retrograde response , through inhibition of Ras-ERK-ETS signaling , improves function in two independent models of human disease caused by mitochondrial dysfunction . Mitochondrial dysfunction transcriptionally reprograms cells by altering nuclear gene expression [22 , 29] . It is not known whether this transcriptional reprogramming is protective or damaging . Ras-ERK-ETS pathway inhibition could benefit neurons either by enhancing the retrograde transcriptional response , or by reversing the expression of genes that are mis-regulated in response to mitochondrial dysfunction . To understand the mechanism by which Ras-ERK-ETS signaling alleviates the effects of neuronal mitochondrial dysfunction , we performed transcriptomic analysis using central nervous system ( CNS ) tissue from larvae with pan-neuronal TFAM overexpression , Pnt knock-down , Aop knock-down , or TFAM overexpression combined with Pnt or Aop knock-down . The expression of 606 and 519 genes are significantly altered by knock-down of Pnt and Aop respectively ( Fig 7A , S7 Table ) . 189 genes were regulated by both Pnt and Aop knock-down and the expression of these genes is strongly positively correlated ( Fig 7B , S8 Table ) . These transcriptomic data support our epistasis analysis ( S6N and S6O Fig ) and are consistent with Pnt and Yan both acting as positive regulators of Ras-ERK-ETS signaling in neurons . The expression of 494 genes were significantly altered in control versus TFAM overexpression conditions , 671 genes in control versus TFAM overexpression combined with Pnt knock-down and 560 genes in control versus TFAM overexpression combined with Aop knock-down conditions ( Fig 7C , S7 Table ) . Around a third of these genes are commonly mis-regulated between these conditions and the expression of these common genes is very strongly positively correlated ( Fig 7D and 7E , S9 and S10 Tables ) . Therefore , the expression of around a third of genes mis-regulated by TFAM overexpression are unchanged by Pnt or Aop knock-down . These data show that Pnt or Aop knock-down does not enhance the transcriptional changes caused by mitochondrial retrograde signaling . To understand how Pnt and Aop knock-down modify the mitochondrial retrograde response we directly compared the TFAM overexpression transcriptome to TFAM overexpression combined with Pnt or Aop knock-down ( S7 Table ) . This comparison shows that in combination with TFAM overexpression , Pnt knock-down significantly alters the expression of 424 genes and Aop knock-down the expression of genes 314 , compared to TFAM overexpression alone ( S7 Table ) . Of these , 82 genes ( TFAM o/e vs TFAM o/e , Pnt RNAi ) and 78 genes ( TFAM o/e vs TFAM o/e , Aop RNAi ) are also mis-regulated in control versus TFAM overexpression conditions ( S11 and S12 Tables ) . Interestingly , the expression levels of these genes are strongly negatively correlated ( Fig 7F and 7G ) . Around half of these genes are mis-regulated by both Pnt and Aop knock-down ( S13 Table ) . Therefore , Pnt or Aop knock-down reverses the expression of a minor subset of genes that are mis-regulated in response to mitochondrial dysfunction . The enriched GO terms for these genes shows that Pnt or Aop knock-down reverses the expression of a number of functional classes , including transcriptional regulation and transmembrane helix , which may contribute to improved neuronal function ( S14 Table ) . Alternatively , Pnt and Aop knock-down may alleviate the effects of mitochondrial dysfunction through the action of the genes that are regulated independent of retrograde signalling . To test this directly we used RNAi to knock-down Hsc70-2 . Hsc70-2 encodes a chaperone of the heat shock 70 family and is the most strongly upregulated gene in TFAM overexpression conditions ( S7 Table ) . Knock-down of either Pnt or Aop dramatically reduces the retrograde induced upregulation of Hsc70-2 expression ( S11 and S12 Tables ) . Knock-down of Hsc70-2 in motor neurons suppresses the climbing and wing inflation phenotypes caused by TFAM overexpression ( Fig 7H and 7I ) . These results are consistent with a mechanism whereby inhibition of Ras-ERK-ETS signaling alleviates the damaging effects of neuronal mitochondrial dysfunction by reversing the expression of a specific subset of genes within the mitochondrial retrograde transcriptome .
Fluctuations in mitochondrial activity and function occur during the cell cycle , throughout development and in disease states . Changes in mitochondrial function affect the cell at multiple levels , but the cellular response to these homeostatic changes is very poorly understood . We have devised a method to identify genes potentially involved in mitochondrial retrograde signaling in vivo . The number and function of the genes identified suggests an extensive and orchestrated cellular response to mitochondrial dysfunction . Inhibition of one of the pathways identified in the screen , Ras-ERK-ETS signaling , also alleviates the effects of mitochondrial dysfunction in the Drosophila nervous system . Targeting Ras-ERK-ETS signaling also improves function in Drosophila models of Leigh syndrome and Parkinson’s . Inhibition of Ras-ERK-ETS signaling partially reverses the mitochondrial retrograde transcriptional response , evidence that retrograde signaling contributes to neuronal dysfunction . Proteomic and genetic methods have been highly successful in characterising the complement of proteins that make up the mitochondrion [42–44] . Human mitochondria consist of around 1158 proteins , only 13 of which are encoded by the mitochondrial genome [42] . However , mitochondria do not function in isolation and participate in a variety of cellular functions , acting within a homeostatic network that responds to changes in the cellular environment [22] . We have developed a sensitised phenotypic assay in the wing to identify genes involved in the cellular response to changes in mitochondrial activity . Using this for a genetic screen we identified 30 modifier genes , the majority of which enhance the wing phenotype . It is possible that screening in this way could identify components of the OXPHOS complexes , or other core mitochondrial proteins . However , knock-down of such genes on their own in the wing is likely to ( and in our experience does ) cause a wing phenotype , and so these genes would be excluded from the screen . We did identify several genes encoding cytosolic metabolic proteins ( Pgk , Adk1 , Pgi ) as enhancers of mitochondrial dysfunction , which did not cause a phenotype when knocked-down by themselves but enhanced the MitoMod phenotype ( S3 Table ) . However , a key point of this study is that many of the genes identified have roles in signal transduction and regulation of gene expression , strongly suggesting that mitochondrial dysfunction modulates the activity of a variety of cell signaling pathways . The ETS domain transcription factor Aop was identified as a suppressor in the modifier screen and also suppressed neuronal mitochondrial dysfunction phenotypes . Knock-down or heterozygosity for pnt also rescued neuronal mitochondrial dysfunction phenotypes . These results are surprising , as Aop and Pnt generally act antagonistically to each other , with Aop acting as an inhibitor of Ras-ERK signaling and Pnt as a positive effector . In the canonical model , activated ERK phosphorylates both Pnt and Aop , promoting cytosolic translocation and degradation of Aop , as well as enhancing the transcriptional activity of Pnt [45 , 46] . In the wing Pnt and Aop have opposite effects: knock-down of Pnt ( using JF02227 ) enhances , while knock-down of Aop suppresses the MitoMod phenotype . However , recent systems biology approaches have shown that the dynamic interplay between Aop and Pnt is more complex than previously thought and is context dependent [47–49] . Aop stability is in fact regulated differentially by Ras-ERK signaling depending on the neuronal differentiation state [49] . We find that PntP2 and Aop overexpression and expression of a constitutively active form of Ras all strongly inhibit neuronal function and act synergistically with TFAM overexpression . Knock-down of either Pnt or Aop suppress Ras-ERK pathway activation and the gene expression changes caused by Pnt or Aop knock-down in neurons strongly correlate . Moreover , reduced expression of Ras , Pnt or Aop are protective against the effects of neuronal mitochondrial dysfunction . Aop and Pnt therefore both act as positive effectors of Ras-ERK signaling in motor neurons and in the context of neuronal mitochondrial dysfunction . Mitochondrial activity plays a key role in healthy ageing . It was recently shown that inhibition of Ras , through ubiquitous expression of dominant negative Ras , or Ras knock-down in adult flies extends lifespan [50] . In contrast to our study , expression of an activated form of Aop in the gut and fat body of adult flies extended lifespan , while knockdown of Aop in these tissues had no effect on longevity [50] . The authors did not test whether inhibition of Pnt affected lifespan , but Pnt overexpression significantly reduced the lifespan of wild-type flies . Furthermore , administration of a pharmacological agent , Trametinib , which inhibits Ras activation of ERK kinase , increased Drosophila lifespan . Although there may be differences in the role of Aop in neuronal mitochondrial dysfunction versus healthy ageing , this previous study and our work together point to the exciting possibility that inhibition of Ras-ERK-ETS signaling may be beneficial to both healthy ageing and human diseases associated with mitochondrial dysfunction . Our data also suggest that Ras-ERK-ETS signaling acts both cell autonomously and non-cell autonomously in response to mitochondrial dysfunction . In the developing Drosophila eye , Ras-ERK signaling determines cell autonomous photoreceptor cell fate , acting downstream of the EGF receptor [51] . The secretion of Spitz , the ligand for the EGFR , from adjacent cells is also regulated by the Ras-ERK pathway [52] . Ras-ERK signaling thus acts both cell autonomously and non-cell autonomously to control photoreceptor differentiation . Mitochondrial signaling in Drosophila has previously been shown to act through non-autonomous systemic negative regulation of insulin signaling [53] . The mechanism of regulation of Ras-ERK-ETS in ageing is not known , nor do we know how mitochondrial dysfunction activates this pathway in neurons . A variety of mitochondrial retrograde signals have been identified , including ROS , Ca2+ , AMP , nicotinamide adenine dinucleotide ( NAD+ ) and acetyl coenzyme A [30] . Future studies will identify the factor ( s ) that mediate mitochondrial retrograde signaling in the nervous system . Large scale alterations in transcription in response to mitochondrial dysfunction have been observed in a wide range of cell types [23] . However , whether these altered transcriptomes have a functional consequence is not clear . Inhibition of Ras-ERK-ETS signaling restores neuronal function and pre-synaptic active zones in our models but does not appear to affect the primary mitochondrial defect . This is consistent with our transcriptomic data , which do not show large scale alterations in the expression of mitochondrial genes with Aop or Pnt Knock-down . We suggest that the beneficial effects of Ras-ERK-ETS pathway inhibition on mitochondrial dysfunction result from transcriptional reprogramming that leads to improvement in pre-synaptic structure and function . Reduced expression of Aop or Pnt , transcriptional targets of Ras-ERK signaling , alleviate the effects of neuronal mitochondrial dysfunction . We exploited this finding to test how the transcriptome is affected by Aop and Pnt knock-down in the background of mitochondrial dysfunction . Surprisingly , a significant number of the mitochondrial retrograde transcriptional changes are reversed by Pnt or Aop knock-down . This finding suggests that the transcriptional mis-regulation activated by mitochondrial retrograde signaling at least partly contributes to neuronal dysfunction . In support of this idea , we find that knock-down of Hsc70-2 , a retrograde response gene whose expression is reduced by Aop and Pnt knock-down , alleviates neuronal mitochondrial dysfunction phenotypes . Future analyses of these retrograde response genes will help elucidate the cellular mechanisms that contribute to neuronal dysfunction . Individual mitochondrial diseases are rare , but in total affect up to 1 in 4300 in the population [7] . The nervous system is frequently affected by mitochondrial mutations , resulting in a wide range of clinical outcomes including ataxia , epilepsy , neuropathy and deafness [54] . Treatments for mitochondrial diseases are limited and mostly symptomatic . Manipulation of the response to mitochondrial dysfunction in neurons may provide a new potentially curative strategy for mitochondrial diseases . Ras-ERK signaling has key roles in synaptic plasticity , learning and memory [55] . However , Ras-ERK-ETS signaling has not previously been identified as potential therapeutic target for mitochondrial disease . Determining how the transcriptional targets of Ras-ERK-ETS signaling contribute to neuronal dysfunction will provide important new insight into mitochondrial diseases such as Leigh syndrome .
Flies were maintained on standard yeast , glucose , cornmeal , agar food at 25°C in a 12 hour light/dark cycle unless stated otherwise . For imaging experiments , except AT[NL] and AT[RK] experiments , embryos were laid over a 24 hour period at 25°C , incubated for a further 24 hours at 25°C , then incubated at 29°C for three days prior to analysis . For AT[NL] and AT[RK] experiments embryos were laid at 25°C and larvae maintained at 25°C until dissection . Fly stocks were UAS-TFAM3M [36] , which was used for all TFAM overexpression experiments except where the weaker UAS-TFAM10M [36] is stated , park25 [40] , UAS-Surf123 . 4 RNAi [41] , Ras85DΔC40B [56] and UAS-mito-roGFP2-Grx1 [57] . UAS-ATeam1 . 03NL ( AT[NL] ) and UAS-ATeam1 . 03RK ( AT[RK] ) flies [39] were from the Kyoto Stock Center ( DGRC ) . The following fly stocks were from the Bloomington Stock Center: DJ-1α , DJ-1β , Lrrk , parkin and Pink1 RNAi and overexpression lines ( details in S1 Table ) , w1118 , FRT82B , TFAMc01716 , Da-Gal4 , nSyb-Gal4 , UAS-mitoGFP , MS1096-Gal4 , OK371-Gal4 , D42-Gal4 , UAS-CD8GFP , UAS-Aop , UAS-PntP2 , pntΔ88 , Thor-lacZ ( ThorK13517 ) , tub-Gal4 , TM6B , tub-Gal80 , GMR-Gal4 , UAS-RasV12 , TFAMJF02307 and TFAMHMC04965 RNAi lines , DsorHMS00037 and DsorJF01697 RNAi lines , RlJF1080 and RlHMS00173 RNAi lines . The TFAM RNAi line ( NIG , 4217R-1 ) , used in the MitoMod stock ( MS1096-Gal4; TFAM RNAi , TFAMc01716/ TM6B , tub-Gal80 ) , was from the NIG-Fly Stock Center , Japan . RNAi stocks used in the genetic screen were from the Bloomington Stock Center , the Vienna Drosophila Resource Center [58] and the NIG-Fly Stock Center , Japan and are listed in S2 , S5 and S6 Tables . park25 , Ras85DΔC40B and park25 , pntΔ88 lines used in Fig 6 were generated by recombination . Mosaic analysis with a repressible cell marker ( MARCM ) analysis of FRT82B , TFAMc01716 was performed using y , w , hs-flp;tub-Gal4 , UAS-mCD8GFP;FRT82B , tub-Gal80 flies as in Avet-Rochex et . al . [59] . Gene names are according to Flybase [60] . Virgin female flies carrying the MitoMod genotype balanced with TM6B , tub-Gal80 ( MS1096-Gal4; TFAM RNAi , TFAMc01716/TM6B , tub-Gal80 ) were crossed to males carrying RNAi transgenes . 1–2 days after eclosion of the progeny wings were observed and scored in males . RNAi lines were only classed as enhancers if most flies had a ≥90ᵒ wing curve . RNAi lines were classed as suppressors if most flies had a <45ᵒ wing curve ( S2 Table ) . Crosses from RNAi lines that enhanced or suppressed the MitoMod wing phenotype were repeated to confirm the result . To exclude RNAi lines that cause a phenotype by themselves , MS1096-Gal4 virgin females were crossed to all RNAi lines . If the progeny of this cross had a wing phenotype then the RNAi line was omitted from the MitoMod screen ( S2 Table ) . Genes for all interacting RNAi lines were tested with independent RNAi lines , where available ( S2 Table ) , and only classed as positive hits if the phenotype was replicated by the independent RNAi ( S5 and S6 Tables ) . RNAi lines were selected using gene expression data available on FlyAtlas , to select genes that are expressed more strongly in the brain than in the whole body [61] . RNAi lines for Drosophila chromatin remodelling genes were also used [62] . GO analysis of genes identified in the screen was performed using the Panther Classification System [63] . Climbing assays were performed as previously described [36] . Males were used for all climbing assays , apart from experiments involving knock-down of Surf1 , where females were used for all genotypes . To quantify wing inflation , flies were transferred into a new vial after eclosion and left for at least 24 hours to allow time for normal wing inflation to occur . Numbers of flies with fully inflated , semi-inflated and uninflated wings were then recorded . All flies that eclosed from the vial were counted . Statistical analysis was performed on raw data and data displayed as a percentage . Tissues were prepared , imaged and quantified as previously described [36] . Primary antibodies were Drosophila anti-TFAM ( Abcam , 1/500 ) , rabbit anti-Dcp1 ( Cell Signaling , 1/200 ) , mouse anti-Wingless ( DSHB , 1/200 ) , chicken anti-β galactosidase ( Abcam ab9361 , 1:1000 ) , mouse anti-Aop ( DSHB , 1/200 ) , mouse anti-brp ( DSHB , 1/200 ) , rabbit anti-dpErk1/2 ( Cell Signaling , 1/200 ) , HRP-Cy3 ( Stratech , 1/1000 ) , rat anti-PntP2 ( 1/500 ) [64] . Secondary antibodies were AlexaFlour 488 , AlexaFlour 594 and AlexaFlour 633 ( Invitrogen , 1/1000 ) . All images were taken using a Zeiss LSM710 confocal microscope with Zen software . Imaging of controls and experimental samples in each experiment was performed using identical confocal microscope settings . Mitochondrial number and volume were quantified using Volocity ( Perkin Elmer ) , 15μm x 15μm ( xy ) , 5 μm ( z ) . dpERK expression was quantified in ImageJ using the Point and Measure tools . Dcp1 expression was quantified using Volocity ( PerkinElmer ) using image projections . To quantify Dcp1 expression in the wing disc the dorsal compartment was selected as the area between the wingless expressing dorsoventral boundary and the third fold in the hinge area of the wing disc . MARCM analysis was performed as in Avet-Rochex et . al . [59] . For Förster resonance energy transfer ( FRET ) –based ATP biosensor imaging , AT[NL] ATP biosensor expressing and AT[RK] ATP insensitive expressing control wing imaginal discs or larval CNS tissues were dissected in Schneider’s medium ( Thermo Scientific ) and imaged immediately at 21°C using a 458nm excitation laser and detecting emitted light between 460-499nm ( CFP ) and 535-650nm ( FRET ) using a Zeiss LSM 710 confocal microscope . For control experiments wing imaginal discs were incubated in Schneider’s medium with 100μM oligomycin ( VWR ) /50 mM 2-deoxyglucose ( SLS ) for 40 minutes at 21°C , then imaged immediately in the same medium . CFP and FRET channel signal intensity at the same three randomly selected points in each wing disc , or three cell bodies in each VNC , was determined using the ImageJ Point and Measure tools and used to calculate the FRET/CFP ratio as a measure of ATP levels [39] . Imaging using Mito-roGFP2-Grx1 was performed as described previously [36] . For DHE staining , wing discs were incubated in 2μM DHE ( Cambridge Bioscience ) in Schneider’s medium for 10 minutes , rinsed twice in Schneider’s medium , then fixed for 5 minutes in 4% formaldehyde/PBS , washed briefly in PBS , dissected and imaged . 20 wing imaginal discs per genotype were dissected in PBS , homogenized in 100μl extraction buffer ( 6M guanidine chloride , 100mM TrisHCL , 4mM EDTA , pH 8 . 0 ) and incubated at 70°C for 5 mins . ATP levels were measured using the ATP Determination kit ( Molecular Probes ) according to the manufacturer’s instructions . Western blot analysis was performed as previously described [36] . Primary antibodies were diluted in TBS/0 . 1% Tween 20 ( TBS-T ) and incubated overnight at 4°C and were Drosophila anti-TFAM ( Abcam , 1/500 ) , mouse anti-ATP5A ( Abcam , 1/5000 ) , mouse anti-MTCO1 ( Abcam , 1/1000 ) and rabbit anti-Actin ( Cell Signaling , 1/4000 ) . After three ten minute washes in TBS-T , the membranes were incubated for 90 minutes with fluorescently labelled secondary antibodies ( anti-mouse IRdye 680 and anti-rabbit IRdye 800 , LI-COR , both at 1/5000 ) diluted in TBS-T , then washed three times for ten minutes in TBS-T . The membranes were then scanned and analysed using an Odyssey infrared scanner ( LI-COR ) . Odyssey infrared imaging systems application software version 3 . 0 . 25 was used to quantify the intensity of the bands on the blots . Normalised expression level was calculated by determining the band intensity relative to Actin . qRT-PCR and qPCR of mtDNA were performed as previously described [36] . Primers used for qRT-PCR were ThorFwd: CGAGGTGTACTCCTCGACGC and ThorRvs: GAGCCACGGAGATTCTTCATGA; Hsp22Fwd: AGCGTTGTCCTGGTGGAG and Hsp22Rvs: GAGCTATAGCCACCTTGTTCG; Rpl4Fwd: TCCACCTTGAAGAAGGGCTA and Rpl4Rvs: TTGCGGATCTCCTCAGACTT . CNS tissue from 20 late third instar larvae per genotype were dissected in cold PBS and transferred directly into 100μl lysis buffer containing β-mercaptoethanol ( Absolutely RNA Microprep kit , Agilent Technologies ) . The lysis buffer was kept on ice while all the brains were dissected . RNA was prepared following manufacturer’s instructions , including DNase treatment and stored at -80°C . Samples were prepared in triplicate . RNA was measured for quantity and integrity on an RNA Pico Chip ( Agilent Technologies ) . 10ng of RNA per genotype was converted into labelled cDNA with the Nugen Ovation System V2 ( NuGEN Technologies Inc . ) . 7mg of labelled cDNA was hybridised to Affymetrix Drosophila genome v2 GeneChips for 20 hours at 45°C . They were then washed , stained ( GeneChip Fluidics Station 450 ) and scanned ( GeneChip Scanner 3000 7G ) according to the manufacturer’s instructions ( Nugen Technologies Inc & Affymetrix ) . Microarray data was processed using the MAS5 . 0 algorithm using the Transcriptome Analysis Console ( ThermoFisher ) . Means were calculated using Tukey's Bi-weight average algorithm and differential expression between groups was calculated using un-paired one way analysis of variance ( ANOVA ) . A statistical cutoff of p<0 . 05 and a fold change cutoff of ±1 . 5 fold were used . Correlations between datasets were analysed using GraphPad Prism ( GraphPad Software Inc . ) . The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE114054 . GO analysis was performed using DAVID ( the database for annotation , visualization and integrated discovery ) bioinformatics resources [65] . GraphPad Prism ( GraphPad Software Inc . ) was used to create graphs and for statistical analysis . Data with a p-value less than or equal to 0 . 05 was considered significant . Comparisons of two samples of continuous data were analysed with an unpaired , two-tailed student’s t-test , where appropriate . Data were analysed for normality using the D’Agostino & Pearson omnibus normality test . Data that did not pass the normality test were analysed with the Mann Whitney test . Variance of the samples was assessed with an F test . If the variances of the two samples were significantly different then the Welch’s correction was applied to the t-test . In order to compare more than two samples of continuous data , one-way ANOVA was used with Tukey’s post hoc test . If data did not pass the D’Agostino & Pearson omnibus normality test , the Kruskal-Wallis , followed by Dunn’s post hoc test were utilised . Categorical data were analysed using chi-squared . | Loss of mitochondrial function activates the mitochondrial retrograde signaling pathway resulting in large scale changes in nuclear gene transcription . Very little is known about retrograde signaling in the nervous system and how the transcriptional changes affect neuronal function . Here we identify Ras-ERK-ETS signaling as a novel mitochondrial retrograde signaling pathway in the Drosophila nervous system . Inhibition of Ras-ERK-ETS signaling improves neuronal function in Drosophila models of mitochondrial disease . Targeting Ras-ERK-ETS signaling may therefore have therapeutic potential in mitochondrial disease patients . Using a transcriptomic approach , we find that inhibition of Ras-ERK-ETS signaling partially reverses the mitochondrial retrograde transcriptional response . Surprisingly therefore , the mitochondrial retrograde transcriptional response contributes to neuronal dysfunction . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"rna",
"interference",
"neuroscience",
"biological",
"locomotion",
"animals",
"motor",
"neurons",
"animal",
"models",
"drosophila",
"melanogaster",
"model",
"organisms",
"experimental",
"organism",
"systems",
"mito... | 2018 | Ras-ERK-ETS inhibition alleviates neuronal mitochondrial dysfunction by reprogramming mitochondrial retrograde signaling |
Prions adopt alternative , self-replicating protein conformations and thereby determine novel phenotypes that are often irreversible . Nevertheless , dominant-negative prion mutants can revert phenotypes associated with some conformations . These observations suggest that , while intervention is possible , distinct inhibitors must be developed to overcome the conformational plasticity of prions . To understand the basis of this specificity , we determined the impact of the G58D mutant of the Sup35 prion on three of its conformational variants , which form amyloids in S . cerevisiae . G58D had been previously proposed to have unique effects on these variants , but our studies suggest a common mechanism . All variants , including those reported to be resistant , are inhibited by G58D but at distinct doses . G58D lowers the kinetic stability of the associated amyloid , enhancing its fragmentation by molecular chaperones , promoting Sup35 resolubilization , and leading to amyloid clearance particularly in daughter cells . Reducing the availability or activity of the chaperone Hsp104 , even transiently , reverses curing . Thus , the specificity of inhibition is determined by the sensitivity of variants to the mutant dosage rather than mode of action , challenging the view that a unique inhibitor must be developed to combat each variant .
Alternative , self-replicating protein conformations have emerged as bona fide parallel protein-folding trajectories with significant biological consequences [1] . In most cases , these alternative conformations are β-sheet-rich and self-assembling , forming linear amyloid aggregates [2] . These amyloids replicate the conformation of their constituent monomers by acting as templates to direct the refolding of other conformers of the same protein as they are bound by and incorporated into the growing aggregate . In so doing , the majority of the protein is converted to the alternative conformation , changing protein activity and thereby inducing new phenotypes , such as neurodegenerative diseases ( i . e . , Transmissible Spongiform Encephalopathies or prion diseases , Alzheimer’s and Huntington’s diseases ) and organelle biogenesis in mammals and gene expression regulation in single-celled organisms [1 , 3] . The high efficiency of this process , when combined with the high kinetic stability of the aggregates [2] , contributes to the recalcitrance of amyloids to clearance by protein quality control pathways [4] . As a result , the associated phenotypes are frequently difficult—if not impossible—to reverse , especially in the clinic [5] . One notable exception to the persistence of amyloid-associated phenotypes is their reversal or “curing” by dominant-negative mutants of prion proteins . These sequence variants were first identified by their ability to confer resistance to scrapie in sheep ( Q171R or R154H in the mammalian prion protein PrP ) , sporadic Creutzfeldt-Jakob disease ( sCJD ) in humans ( E219K in PrP ) , and translation termination infidelity in yeast ( G58D in Sup35 ) [6–19] . Subsequently , these mutants were shown to interfere with the assembly of amyloid by the wildtype prion protein in vitro and to reduce or clear existing amyloid composed of the wildtype prion protein when delivered to tissue culture cells , mice , or yeast [15 , 19–31] . Given this unique curing ability , elucidating the mechanism ( s ) by which dominant-negative prion mutants act may reveal potential strategies for reversing amyloid persistence more generally . Despite the promise of this line of investigation , the inhibition achieved by dominant-negative mutants appears to be conformation-specific . For example , the resistance to sCJD conferred by the E219K PrP mutant in humans is not extended to the conformations , known as variants , responsible for genetic and iatrogenic forms of the disease [14 , 15 , 17 , 32–35] . Similarly , resistance to classical scrapie is not observed for bovine spongiform encephalopathy ( BSE ) or atypical scrapie variants in sheep with Q171R or R154H mutations in PrP [10 , 36–43] [44–52] . Finally , the G58D mutation of Sup35 cures the [PSI+]Strong and [PSI+]Sc4 variants ( n . b . [PSI+] denotes the transmissible amyloid state of Sup35 ) to different extents in different genetic backgrounds but is unable to cure the [PSI+]Sc37 and [PSI+]Weak variants in yeast [53 , 54] . What is the molecular basis of this differential inhibition ? One possibility is that the distinct recognition surfaces and/or rate-limiting steps in the self-replication process characteristic of the variants make them susceptible to only certain mechanisms of inhibition [55–61] . Alternatively , the conformational differences may confer distinct sensitivities to the same mechanism of inhibition . Given the conformational plasticity of amyloidogenic proteins [62 , 63] , understanding the forces limiting the efficacy of inhibitors can mean the difference between developing an infinite number of individual interventions for each variant or simply different dosing regimes for the same inhibitor . Here , we exploit the yeast prion Sup35 to gain this insight . We explored the sensitivity of three variants of Sup35 ( i . e . , [PSI+]Sc4 , [PSI+]Weak , and [PSI+]Sc37 ) to expression of G58D and the impact of this dominant-negative mutant on the self-replication of each variant . Our studies indicate that “resistance” to G58D can be partially overcome at higher dosage of the mutant , revealing differential sensitivity to the inhibition . G58D reduces the kinetic stabilities of the amyloids associated with the variants , which determines their efficiencies of fragmentation by chaperones [60] . Consistent with this correlation , G58D inhibition of the three variants was dependent on the chaperone Hsp104 , as was the case for the previously studied [PSI+]Strong variant [64] . In the presence of G58D , Sup35 amyloid was fragmented by Hsp104 with higher efficiency . This increase led to amyloid clearance in daughter cells , which could be reversed by transient inhibition of Hsp104 specifically in this population . Thus , G58D dominant-negative inhibition targets distinct conformational variants through the same mechanism with differing efficacy , suggesting that the observed “resistance” is relative rather than absolute .
To determine if the specificity of G58D on [PSI+] variants occurs through distinct mechanisms or through distinct sensitivities to the same mechanism of inhibition , we generated diploid [PSI+]Sc4 , [PSI+]Weak and [PSI+]Sc37 yeast strains expressing wildtype Sup35 at different ratios relative to G58D ( 2:1 , 1:1 , 1:2; S1 Fig ) . Inhibition of [PSI+] propagation can be monitored functionally because the formation of amyloid by Sup35 partially compromises its activity and leads to a defect in translation termination [65 , 66] . [PSI+] strains carrying the ade1-14 allele form white colonies on rich medium due to read-through of a premature stop codon in the ADE1 open reading frame . However , strains with defective prion propagation , or those that have lost the prion state ( known as [psi-] ) , form red colonies on rich medium as a result of the accumulation of active Sup35 [67] . Expression of G58D at any ratio in a [PSI+]Sc4 strain promoted the accumulation of red pigment on rich medium , indicating reversal of the prion phenotype ( Fig 1A ) . By colony color , the severity of this effect increased with G58D dosage , with a 1:2 ratio of wildtype to G58D leading to a colony phenotype for [PSI+]Sc4 that was indistinguishable from [psi-] ( Fig 1A ) . For the [PSI+]Sc37 and [PSI+]Weak variants , which were previously reported to be compatible with G58D expression [53 , 54] , efficient prion propagation was also dependent on the ratio of wildtype to G58D , but the critical threshold for phenotypic reversal was distinct in each case . The [PSI+]Sc37 variant formed colonies that were more pink on rich medium at a 1:1 ratio of wildtype to G58D relative to a wildtype strain and that were indistinguishable from [psi-] at a 1:2 ratio of wildtype to G58D ( Fig 1B ) , mirroring our observations for [PSI+]Sc4 ( Fig 1A ) . In contrast , the [PSI+]Weak variant phenotype was only partially reversed at the highest ratio of wildtype to G58D tested ( 1:2 ) , where the pinker colonies on rich medium relative to the wildtype [PSI+]Weak strain indicated a mild inhibition by G58D ( Fig 1C ) . Thus , the three [PSI+] variants are each dominantly inhibited by G58D expression in a dose-dependent manner , but the dose required for inhibition of [PSI+]Sc4 and [PSI+]Sc37 is lower than that of [PSI+]Weak . To assess whether reversal of the [PSI+] phenotype upon G58D expression reflected prion loss ( i . e . , curing ) , we determined the frequencies of [psi-] appearance during mitotic division for each strain . [PSI+] propagation was largely stable at the 2:1 ( ~0% curing ) and 1:1 ( ~1% curing ) ratios of wildtype to G58D for both [PSI+]Sc4 and [PSI+]Sc37 , where the colony phenotype was only mildly reversed ( Fig 1A , 1B , 1D and 1E ) . At a 1:2 ratio of wildtype to G58D , both [PSI+]Sc4 ( ~9% curing , Fig 1D ) and [PSI+]Sc37 ( ~8% curing , Fig 1E ) were more unstable , consistent with the stronger reversal of their prion phenotypes at this ratio ( Fig 1A and 1B ) . For the [PSI+]Weak variant , which is less sensitive to G58D inhibition ( Fig 1C ) , [PSI+] propagation was stable at all wildtype:G58D ratios tested ( Fig 1F ) . Thus , [PSI+] curing in diploids expressing G58D parallels the severity of the phenotypic reversal in all three variants and , for the most sensitive strains ( i . e . , [PSI+]Sc4 and [PSI+]Sc37 ) , arises in a dose-dependent manner . Together , these observations indicate that the previously described “resistance” of [PSI+]Sc37 and [PSI+]Weak to curing by G58D expression reflected their higher threshold for sensitivity rather than their absolute recalcitrance to inhibition by this mutant . Although the three [PSI+] variants studied here , in addition to the previously studied [PSI+]Strong variant [64] , differ in their sensitivities to G58D inhibition ( Fig 1 ) , the dose dependence of this inhibition suggests a common underlying mechanism [64 , 68] . We previously linked G58D inhibition to a reduction in the kinetic stability of Sup35 aggregates and a resulting increase in their fragmentation by the chaperone Hsp104 , which led to their disassembly [64] . In this model , the distinct effective inhibitory ratios of G58D on [PSI+] variants may reflect the impact that this mutant has on the kinetic stability of each . While it has been well-established that Sup35 aggregates in the [PSI+]Sc4 conformation are of lower stability than those in the [PSI+]Sc37 conformation , the relative stabilities of the four variants have not been previously reported [60 , 69 , 70] . To gain this insight , we first determined the kinetic stabilities of Sup35 aggregates , in the absence of G58D , by their sensitivity to disruption with 2% SDS at different temperatures as a baseline comparison [71] . Solubilized protein is then quantified by entry into a SDS-polyacrylamide gel and immunoblotting [64] . For wildtype strains , Sup35 was efficiently released from aggregates between 65°C and 75°C in lysates from strains propagating the [PSI+]Strong and [PSI+]Sc4 variants ( Fig 2A ) or between 70°C and 90°C in lysates from strains propagating the [PSI+]Weak and [PSI+]Sc37 variants ( Fig 2B ) . The higher kinetic stability of the latter variants is consistent with their lower efficiency of fragmentation , which leads to a larger steady-state size for their associated amyloids as assessed by semi-denaturing agarose gel electrophoresis ( SDD-AGE ) and immunoblotting for Sup35 ( S2 Fig ) [60 , 72] . To sensitize the assay in an attempt to reveal biochemical differences between the variants in each group , we deleted the NATA N-terminal acetyltransferase , which reduces the kinetic stability of Sup35 amyloid in [PSI+] strains [73 , 74] . In this genetic background , the fraction of soluble Sup35 released from amyloid of the [PSI+]Strong variant in the presence of SDS was significantly increased relative to that from the [PSI+]Sc4 variant over the same temperature range ( Fig 2C ) , indicating that the aggregates are less kinetically stable in the [PSI+]Strong than the [PSI+]Sc4 variant . Similarly , a significantly larger fraction of Sup35 was released from amyloid in the presence of SDS from the [PSI+]Sc37 variant than from the [PSI+]Weak variant ( Fig 2D ) , indicating that the aggregates are less kinetically stable in the [PSI+]Sc37 than the [PSI+]Weak variant . Thus , the kinetic stability of Sup35 aggregates in [PSI+] variants increases in the order [PSI+]Strong , [PSI+]Sc4 , [PSI+]Sc37 , [PSI+]Weak . If G58D inhibits these variants through a common mechanism , we would expect the kinetic stabilities of each of the variants to decrease in the presence of the mutant . To test this possibility , we assessed the sensitivity of Sup35 aggregates , isolated from diploid strains expressing a 1:1 ratio of wildtype to G58D , to disruption with 2% SDS at different temperatures . Soluble protein was then quantified by entry into an SDS-polyacrylamide gel and immunoblotting for Sup35 . For the [PSI+]Sc4 strain , G58D expression increased the amount of soluble Sup35 released from aggregates at all temperatures assayed ( 65°C , 70°C and 75°C ) in comparison with a wildtype strain ( Fig 3A ) . G58D similarly promoted Sup35 release from aggregates isolated from the [PSI+]Sc37 ( Fig 3B ) and [PSI+]Weak ( Fig 3C ) strains at 80°C and 85°C , but the magnitude of this effect was greater for the former . Thus , G58D incorporation destabilizes Sup35 aggregates from [PSI+] variants in a manner that correlates directly with the severity of their phenotypic inhibition ( Fig 1 ) . These observations are consistent with the idea that G58D acts through a similar mechanism to inhibit the [PSI+] variants . A decrease in the kinetic stability of amyloid should increase its efficiency of fragmentation and potentially lead to its clearance . To begin to determine the effects of G58D on the fragmentation of Sup35 amyloid associated with these [PSI+] variants , we first assessed the steady-state size distributions of these complexes by SDD-AGE and immunoblotting for Sup35 . As we have previously observed for [PSI+]Strong [64] , expression of G58D at any ratio relative to wildtype Sup35 in a [PSI+]Sc4 strain led to a decrease in the accumulation of slowly migrating aggregates in comparison to the same dose of wildtype protein alone ( Fig 3D ) . For [PSI+]Sc37 , similar decreases were observed ( Fig 3E ) , but for [PSI+]Weak , Sup35 aggregates were only shifted to smaller complexes at the lowest wildtype to G58D ratio tested ( 1:2 , Fig 3F ) . Together , these observations suggest that the kinetic destabilization of Sup35 aggregates by G58D results in a higher efficiency of fragmentation in vivo , and these effects correlate directly with the severity of their phenotypic inhibition ( Fig 1 ) . To determine how the kinetic destabilization of Sup35 aggregates by G58D impacts the number of heritable prion units ( propagons ) in [PSI+]Sc4 , [PSI+]Sc37 and [PSI+]Weak strains , we used a genetic assay [75] . Specifically , diploid strains expressing either two copies of wildtype SUP35 or one copy each of wildtype SUP35 and G58D were treated with guanidine HCl ( GdnHCl ) , a potent inhibitor of the fragmentation catalyst Hsp104 [67 , 76–81] , allowed to dilute existing aggregates through cell division , and plated in the absence of the inhibitor to quantify the number of cells inheriting an aggregate . As we have previously observed in a [PSI+]Strong strain [64] , G58D expression in either [PSI+]Sc4 and [PSI+]Sc37 diploids reduced propagon number by factors of ~2 and ~4 , respectively ( Fig 3G and 3H ) , consistent with the reversal of the [PSI+] phenotype and the loss of [PSI+] that we observed in these strains ( Fig 1A , 1B , 1D and 1E ) . In contrast , G58D expression in [PSI+]Weak increased propagon number by a factor of ~2 . 5 ( Fig 3I ) . Although we did not detect any changes in the severity or stability of the [PSI+]Weak phenotype at this ratio ( Fig 1C and 1F ) , this increase in propagon count provides an explanation for the previously reported strengthening of the [PSI+]Weak phenotype upon G58D expression to much higher levels [53] . Phenotypic strengthening is associated with a decrease in soluble Sup35 , which would result from an increase in amyloid templates , detected as propagons in this assay , through enhanced fragmentation [60] . Thus , the phenotypic consequences of G58D expression , both inhibition and enhancement , can be directly explained by changes in the steady-state accumulation of Sup35 propagons . Given the distinct kinetic stabilities of Sup35 amyloid in the [PSI+] variants studied here ( Fig 2 ) , the specificity of G58D inhibition and enhancement likely reflect thresholds for fragmentation activity that result in changes in the steady-state accumulation of Sup35 forms in vivo . If enhanced fragmentation is indeed the mechanism underlying G58D effects , these changes should be Hsp104-dependent . To determine if this is the case , we constructed heterozygous disruptions of HSP104 in diploid strains expressing G58D at different ratios ( S3 Fig ) . In strains expressing only wildtype Sup35 , heterozygous disruption of HSP104 significantly decreased the number of propagons in the [PSI+]Sc4 and [PSI+]Sc37 variants tested ( Fig 3G and 3H , compare lanes 1 and 3 ) , consistent with its catalytic role in fragmentation [78 , 79] and the size threshold for Sup35 aggregate transmission [72] . In contrast , heterozygous disruption of HSP104 in [PSI+] variant strains expressing both wildtype and G58D Sup35 increased the number of propagons ( Fig 3G and 3I , compare lanes 2 and 4 ) . Thus , the reduction in propagon number associated with G58D is suppressed by lowering the dosage of HSP104 and thereby fragmentation activity , suggesting that enhanced fragmentation is the underlying mechanism . Next , we determined if these changes in propagon number upon heterozygous disruption of HSP104 impacted the severity and stability of the [PSI+] phenotype . Heterozygous disruption of HSP104 restored the [PSI+] phenotype ( Fig 1A ) and efficiently suppressed [PSI+] loss ( Fig 1D ) in the [PSI+]Sc4 strains expressing any ratio of G58D . For the [PSI+]Sc37 and [PSI+]Weak variants , similar although attenuated trends were apparent . Heterozygous disruption of Hsp104 partially reversed the pinker colony color on rich medium for both [PSI+]Sc37 and [PSI+]Weak ( Fig 1B and 1C ) . For [PSI+]Sc37 , heterozygous disruption of Hsp104 increased [PSI+] loss in all strains , indicating that wildtype fragmentation levels must be close to the threshold required for efficient propagation of the amyloid state ( Fig 1E ) . Nonetheless , in the strain expressing the 1:2 ratio of wildtype to G58D , the frequency of [PSI+] loss was suppressed by heterozygous disruption of Hsp104 ( Fig 1E ) . Thus , reduction of Hsp104 reverses the G58D-induced inhibition of the [PSI+] phenotype . Together , these observations are consistent with the idea that the downstream effect of G58D is identical for all [PSI+] variants: an enhancement of the fragmentation efficiencies of their Sup35 amyloid . The enhanced efficiency of fragmentation of Sup35 aggregates in the presence of G58D ( Fig 3D and 3E ) and the reduction in propagon levels ( Fig 3G and 3H ) suggests that Sup35 aggregates are being destroyed in strains propagating the [PSI+]Sc4 and [PSI+]Sc37 variants . For [PSI+]Strong , we previously detected this disassembly by monitoring the soluble pool of Sup35 in response to cycloheximide treatment to follow the fate of existing protein [64] . However , [PSI+]Strong is more sensitive to G58D expression than [PSI+]Sc4 , [PSI+]Sc37 and [PSI+]Weak ( Fig 1 ) [64] , suggesting that release of soluble Sup35 from aggregates by enhanced fragmentation may be less readily detected in the latter variants . Specifically , the individual steps in prion propagation in vivo ( e . g . conversion , fragmentation , and transmission ) are variant-specific and difficult to monitor in isolation in a living system [60 , 78] . Moreover , the accumulation of soluble Sup35 is impacted not only by the inherent rate of conversion on fibers ends but also by the cumulative effect of each of the steps of prion propagation on the number of those ends [60 , 72] . Because the cumulative effects of each event on soluble Sup35 levels are not intuitive to qualitatively predict from those rates , we developed a deterministic model of Sup35 dynamics to deconstruct this complexity and gain additional mechanistic insight into the differential effects of G58D on the variants . This model uses a range of conversion and fragmentation rates that support [PSI+] maintenance to capture different variants ( see S1 Text ) . In addition , we have incorporated the concept of nucleation , which specifies a minimum size for a thermodynamically stable aggregate and has been previously established as a key event in Sup35 aggregation in vitro [82–84] . The steady-state size and number of Sup35 aggregates reflects a balance between conversion , which depends on continuous synthesis of Sup35 , and fragmentation [72]; when Sup35 synthesis is halted , aggregates are predicted to increase in number ( Fig 4A ) and decrease in size ( Fig 4B ) because fragmentation is proposed to exert a greater influence on the equilibrium state [72] . In line with this observation , our model predicts that cycloheximide treatment will decrease soluble Sup35 levels for prion variants that are stably propagating [PSI+] ( Fig 4C ) because additional templates have been created ( Fig 4A ) . Intriguingly , the extent of this decrease is predicted in our mathematical model to correspond inversely with the rate of fragmentation: that is , the slowest rate of fragmentation induces the largest decrease in soluble Sup35 ( Fig 4B , black ) , relative to the steady-state levels prior to the manipulation . If fragmentation produces more templates , which in turn promotes Sup35 conversion to the amyloid state , why would we predict a lower rate of fragmentation to have the most significant effect on soluble Sup35 levels ? The reason is , as we have previously demonstrated under heat shock conditions [85] , fragmentation resolubilizes Sup35 in addition to creating new templates . Thus , high rates of fragmentation will push the balance between conversion and fragmentation toward the latter , causing a shift from aggregated to soluble Sup35 . Consistent with this logic , our model predicts an increase in aggregate number that corresponds inversely with fragmentation rate ( i . e . the largest increase in aggregate number corresponds to the slowest fragmentation rate; Fig 4A , black ) . This correlation can be explained directly by changes in the rate of Sup35 resolubilization from aggregates: the slowest fragmentation rate leads to the slowest rate of resolubilization ( Fig 4D , black ) and thereby the largest increase in aggregate number ( Fig 4A , black ) . These predictions correlate with our observations of the [PSI+]Sc4 , [PSI+]Sc37 , and [PSI+]Weak variants upon treatment with cycloheximide . For strains where wildtype Sup35 was the only form present , the average size of Sup35 aggregates decreased ( Fig 5A–5C ) . In addition , the level of soluble Sup35 decreased upon cycloheximide treatment for the [PSI+]Weak and [PSI+]Sc37 variants , but no significant decrease was observed for [PSI+]Sc4 variant ( Fig 5D–5F , lane 1 ) . According to our model , these observations are consistent with a nucleation-dependent aggregation process , which permits resolubilization of aggregates that are fragmented below the minimum thermodynamically stable size ( Fig 4D , compare solid and dashed lines ) , and a higher rate of fragmentation for [PSI+]Sc4 , which would release more aggregated Sup35 into the soluble pool ( Fig 4D , red ) . In the presence of G58D , soluble Sup35 levels in [PSI+]Sc37 and [PSI+]Weak are no longer reduced ( Fig 5E and 5F , compare lanes 1 and 3 ) , suggesting that G58D expression promotes aggregate fragmentation and thereby resolubilization . Consistent with this idea , treatment of the variants with both cycloheximide and guanidine HCl led to an increase in aggregate size ( Fig 5A–5C ) and a decrease in soluble Sup35 levels in the presence of G58D ( Fig 5D–5F , compare lanes 3 and 4 ) , indicating that Hsp104-catalyzed fragmentation promotes Sup35 resolubilization . The ability of our mathematical model to capture the behavior of Sup35 in response to these manipulations strongly supports the idea that G58D destabilizes Sup35 aggregates to promote their increased fragmentation by Hsp104 and thereby their resolubilization . However , a more nuanced evaluation indicates that the threshold for inhibition cannot be explained by fragmentation efficiency alone . For example , [PSI+]Sc37 has a similar phenotypic sensitivity to G58D dosage as the [PSI+]Sc4 variant ( Fig 1 ) but a kinetic stability , size , and likely fragmentation efficiency closer to the [PSI+]Weak variant ( Fig 2 and S2 Fig ) . A bulk shift in Sup35 from aggregate to soluble requires that the resolubilized Sup35 does not efficiently reconvert to the aggregated state; thus , conversion efficiencies will also impact the outcome of the G58D effects on aggregate kinetic stability , fragmentation and resolubilization . Sup35 aggregates in the [PSI+]Sc37 conformation direct conversion at a higher rate than those in the [PSI+]Sc4 conformation [60] , but the relative rates of conversion for [PSI+]Sc37 and [PSI+]Weak have not been reported . To compare these variants , we transiently treated strains with GdnHCl in liquid culture to reduce propagon number and then monitored propagon recovery as a function of time after removal of GdnHCl by plating cells and assessing their colony-color phenotype . The [PSI+]Weak variant amplified its propagons at a faster rate than the [PSI+]Sc37 variant ( S4 Fig ) . This recovery rate is a function of the product of the conversion and fragmentation rates [60] . Because Sup35 aggregates in the [PSI+]Sc37 conformation are less kinetically stable than those in the [PSI+]Weak conformation ( Fig 2B and 2D ) and thereby likely fragmented at a higher rate , this observation suggests that the conversion rate of [PSI+]Sc37 is much lower than that of [PSI+]Weak . As a result , resolubilized Sup35 would be less likely to reconvert to the aggregated state in the [PSI+]Sc37 variant than in the [PSI+]Weak variant . Thus , the higher rate of resolution and the lower rate of conversion combine to increase the sensitivity of [PSI+]Sc37 to G58D inhibition relative to [PSI+]Weak . Together , our studies are consistent with the ideas that resolubilization of aggregated Sup35 is the mechanism of G58D inhibition and that the variant-specific rates of conversion and fragmentation dictate the threshold for phenotypic reversal . However , Weissman and colleagues previously reported that loss of [PSI+]Sc4 propagated by G58D alone was associated with propagon loss from daughter but not mother cells [54] . This observation was interpreted as a G58D-dependent defect in Sup35 aggregate transmission to daughter cells , but using a direct fluorescence-based microscopy assay for Sup35-GFP transmission , we were unable to detect a transmission defect in [PSI+]Strong strains expressing wildtype and G58D Sup35 [64] . The appearance of daughter cells without propagons could also arise if Sup35 aggregates were transmitted but subsequently disassembled by Hsp104 in this compartment . If this scenario is correct , inhibition of Hsp104 will lead to an increase in [PSI+] propagons in daughter cells . To test this hypothesis , we constructed [PSI+]Sc4 diploid strains expressing only G58D Sup35 and compared prion propagation in wildtype and HSP104 heterozygous disruption versions of this strain by plating on rich medium and observing colony-color phenotype . Consistent with previous observations [54] , [PSI+]Sc4 propagation is unstable in a wildtype strain ( ~50% prion loss ) , but we found that this instability is strongly suppressed by heterozygous disruption of HSP104 ( ~5% prion loss; Fig 6A ) . Propagons are normally distributed between mother and daughter cells in a 2:1 ratio [75] . However , analysis of propagon numbers in mother and daughter cells showed an even stronger bias in the distribution of propagons toward the mothers in the presence of G58D ( Fig 6B , black diamonds ) , including a population of pairs in which the mother but not the daughter retained a large number of propagons ( Fig 6B , red diamonds ) . By contrast , heterozygous disruption of HSP104 reduced the stronger mother bias associated with G58D expression , and more propagons were detected in daughter cells ( Fig 6B , white triangles ) . Notably , daughter cells lacking propagons were not isolated from the HSP104 heterozygous disruption strain , indicating that the suppression of prion loss ( Fig 6A ) correlated with an increase in propagons in daughter cells ( Fig 6B ) . Given the suppression of these phenotypes by heterozygous disruption of Hsp104 , we next directly determined if Hsp104 inhibition specifically in daughter cells is sufficient to suppress [PSI+] loss . To do so , we isolated daughter cells from [PSI+]Sc4 diploids expressing one copy each of wildtype and G58D SUP35 by FACS , based on the staining of bud scars with Alexa-647 WGA . The absence of bud scars in cells with the lowest fluorescence intensity indicates that this fraction contains the newborn population , in contrast to a mixed population before sorting ( Fig 6C and S5 Fig ) . The isolated daughters were then incubated on rich medium in the presence or absence of GdnHCl for three hours to transiently inhibit Hsp104 activity and then plated to determine the frequency of prion loss . Strikingly , GdnHCl treatment of daughter cells suppressed the frequency of prion loss ( Fig 6D ) . Because daughter cells were biochemically isolated before treatment , the GdnHCl suppression of prion loss cannot be explained by an increased transmission of Sup35 aggregates to daughter cells upon Hsp104 inhibition . Rather , Sup35 aggregates must have already been present , with the transient inhibition of Hsp104 blocking their resolubilization after transfer , consistent with the idea that G58D inhibits the propagon of all [PSI+] variants through the same mechanism .
Together , our studies indicate that a single inhibitor , the dominant-negative G58D mutant of Sup35 , can perturb the propagation of four different variants of the [PSI+] prion , [PSI+]Strong , [PSI+]Sc4 , [PSI+]Sc37 , and [PSI+]Weak , through the same mechanism . The effects of G58D expression are most easily detected at the protein level as kinetic destabilization of Sup35 amyloid ( Fig 3A–3C ) and related reductions in the size of their SDS-resistant core polymers ( Fig 3D–3F ) . These changes only become apparent at the phenotypic and inheritance levels when the impact on Sup35 amyloid rises above a threshold dictated by the rates of conversion and fragmentation for the variants , allowing disassembly to dominate over reassembly . The G58D mutation lies in the second oligopeptide repeat of Sup35 , a region of the protein that is essential for prion propagation [86–88] and that impacts the ability of the Hsp104 chaperone to thread monomers through its central pore during the fragmentation process [89] . Position 58 is located within the amyloid core of Sup35 in the [PSI+]Sc37 variant but is more accessible in the [PSI+]Sc4 variant [69] . Nonetheless , the kinetic destabilization of the four variants by G58D ( Fig 3A–3C ) [64] suggests this region contributes directly to associations within each of the aggregates . Structural studies on the isolated second repeat revealed that the G58D substitution introduced a turn into the otherwise extended conformation of the wildtype repeat , suggesting that packing and thereby amyloid kinetic stability could be altered by this conformational change [90] . Previous studies on the [PSI+]Strong and [PSI+]Sc4 conformational variants suggested two different mechanisms for G58D-induced curing . For [PSI+]Strong , curing depended not only on the dosage of G58D but also of HSP104 , suggesting that prion propagation was inhibited by amyloid disassembly . Indeed , in the presence of G58D , previously aggregated Sup35 transitioned to the soluble fraction [64] . For [PSI+]Sc4 , curing correlated with the loss of heritable aggregates in daughter cells , interpreted as a G58D-induced defect in amyloid transmission [54] . These distinct models for inhibition are consistent with the idea that different conformational variants must be cured through different molecular mechanisms [62 , 63] . However , our studies resolve this controversy: G58D inhibits both variants by promoting amyloid disassembly in daughter cells . This model is supported by both the Hsp104-dependence of the curing of both variants ( Fig 1D ) [64] and of the reduction in propagons ( Fig 3H ) [64] . In addition , overexpression of Hsp104 cures [PSI+]Sc4 propagated by G58D but not wildtype Sup35 , suggesting the former is more sensitive to higher fragmentation rates than the latter [54] . Consistent with this interpretation , overexpression of an N-terminally truncated Hsp104 mutant [54] , which is deficient in substrate processing [91] , is unable to cure [PSI+]Sc4 propagated by G58D . We have previously drawn parallels between the dominant-negative inhibition of [PSI+] propagation by Sup35 G58D and that of protease-resistant PrP by hamster Q219K ( corresponding to E219K in humans and Q218K in mouse ) . In both cases , the mutant is incorporated into wildtype aggregates but capable of destabilizing the amyloid state only when present in excess to wildtype protein , and the efficacy of dominant-negative inhibition is greater for less kinetically stable conformational variants [15 , 20 , 21 , 64 , 92] . Given the likelihood that the mechanisms of inhibition are similar between the yeast and mammalian dominant-negative mutants , the “resistance” of sCJD to E219K in humans and of 22L to Q219K in mice may be possible to overcome by increasing the dosage of the mutant , as we have demonstrated here for G58D and [PSI+]Sc37 ( Fig 1B and 1E ) . For G58D , inhibition occurs at a dosage far below that at which the prion state is induced to appear [53] , indicating that the threshold between curing and induction is wide enough to accommodate switches in one direction or the other specifically . A similar analysis in mammals would be prudent before pursuing increased dosage of dominant-negative mutants as a therapeutic strategy . How can the absence of heritable aggregates in some daughter cells be reconciled with amyloid disassembly as a common mechanism of inhibition for G58D ? Our previous studies have revealed that increasing chaperone levels by heat shock , leads to amyloid disassembly in a [PSI+]Weak strain [85] , suggesting that the ratio of chaperones:amyloid is a key contributor to the balance between amyloid assembly and disassembly . A similar skew in this ratio likely occurs during G58D curing but through a distinct mechanism . Our previous studies uncovered a size threshold for amyloid transmission during yeast cell division: larger aggregates were preferentially retained in mother cells [72] . This asymmetry created an age-dependent difference in aggregate load , with newborn daughters taking several generations to return to the steady-state level of propagons observed in mother cells [72] . This observation suggests that the chaperone:substrate ratio could be skewed toward the former in daughter cells . This altered ratio , when combined with the decrease in the kinetic stability of Sup35 amyloid induced by G58D ( Fig 3A–3C ) , likely creates a niche where amyloid disassembly dominates . Indeed , the normally resistant [PSI+]Strong variant is cured by transient heat shock when G58D is expressed [85] . Consistent with the idea that G58D cures [PSI+] by promoting amyloid disassembly , curing is reduced ( Fig 5A ) , and propagon numbers increase in daughters ( Fig 5B ) when Hsp104 levels are reduced . Most importantly , transiently blocking Hsp104 activity in daughter cells after division also greatly reduces prion loss ( Fig 5D ) . Thus , G58D–containing Sup35 amyloid is transmitted to daughter cells , but , once there , these aggregates are at greater risk of clearance by Hsp104-mediated disassembly . Beyond dominant-negative mutants , conformational variants of PrP and Sup35 also differ in their sensitivities to small molecule inhibitors [62 , 63] . Unfortunately , even sensitive conformational variants can develop resistance to these compounds , further complicating attempts to develop therapeutic interventions for these diseases . For example , treatment of prion-infected mice or tissue culture cells with quinacrine or swainsonine reduced the kinetic stability of protease-resistant PrP and altered its tropism in cell lines , but these properties were reversed when treatment was removed [93–95] . Although it remains unclear whether the emerging conformational variants were minor components that were selected or newly induced by the treatment , this conformational plasticity creates a moving target that is impossible to manage if a unique inhibitor must be developed in each case . Our studies suggest that as prion conformational variants evolve , adapt or mutate , changes in dosing regimes could be effective countermeasures , although the range of possible doses is likely to be restricted because overexpression of even a dominant-negative mutant can lead to prion appearance [53] . Nevertheless , quinacrine can eliminate the RML conformational variant of PrP from CAD5 cells at a 5-fold lower dosage than is required to eliminate an IND24-resistant variant [96] . Much research is focused on the appearance and self-replicating amplification of amyloid , yet these processes are clearly counteracted by disassembly pathways in vivo . This balance between assembly and disassembly contributes strongly to prion persistence , even in mammals . For example , inhibition of PrP expression can reverse accumulation of protease-resistant PrP , pathological changes and clinical progression of prion disease in mice , presumably by allowing clearance pathways to dominate , if initiated before extensive damage arises [97] . While mammals lack an Hsp104 homolog , a chaperone system , composed of mammalian Hsp70 , Hsp110 , and class A and B J-proteins , possesses strong disaggregase activity [98] , capable of directing amyloid disassembly , although this activity has yet to be tested against protease-resistant PrP [99] . Nevertheless , natural variations in the accumulation of prion and chaperone proteins may also serve as a new framework in which to consider phenotypic differences among variants . For example , tropism and clinical progression are likely to be impacted by the balance between assembly and disassembly pathways , as we have observed for mitotic stability and heat shock-induced prion curing in yeast [72 , 85] . Moreover , the steady-state ratio of chaperones:amyloid may be a key consideration in screening potential therapeutics and in their ultimate efficacy in vivo , particularly for small molecules proteostasis regulators that perturb the assembly/disassembly balance .
All plasmids used in this study are listed in S1 Table . pRS306-PADH contains PADH-Multiple Cloning Site-TCYC1 as a KpnI-SacI fragment from pSM556 ( a gift from F . U . Hartl ) in a similarly digested pRS306 . The SUP35 ( G58D ) ORF was then subcloned into pRS306-PADH as a BamHI-EcoRI fragment isolated from pRS306-SUP35 ( G58D ) to create pRS306-PADHSUP35 ( G58D ) ( SB468 ) . Oligonucleotides used in this study are listed in S2 Table . All strains are derivatives of 74-D694 and are listed in S3 Table . [PSI+]Sc4 ( SY2085 ) and [PSI+]Sc37 ( SY2086 ) haploid wildtype strains were gifts from J . Weissman . Yeast strains expressing ectopic copies of SUP35 or G58D from URA3 ( pRS306 ) or TRP1 ( pRS304 ) -marked plasmids were constructed by transforming yeast strains with plasmids that were linearized with BstBI or Bsu361 , respectively , and by selecting for transformants on the appropriate minimal medium . In all cases , expression was confirmed by quantitative immunoblotting for Sup35 . Disruptions of SUP35 ( FP35 , FP36 ) were generated by transformation of PCR-generated cassettes using pFA6aKanMX4 as a template with the indicated oligonucleotide primers ( S2 Table ) and selection on rich medium supplemented with G418 . HSP104 disruptions were generated by transformation with a PvuI-BamHI fragment of pYABL5 ( a gift from S . Lindquist ) and selection on minimal medium lacking leucine . Disruptions of NAT1 ( FP29 , FP30 ) were generated by transformation of PCR-generated cassettes using pFA6a-hphMX4 as a template with the indicated primers ( S2 Table ) and selection on complete medium supplemented with hygromycin . All the disruptions were verified by PCR and 2:2 segregation of the appropriate marker . Exponentially growing cultures of the indicated strain were plated on YPD for single colonies , and the frequency of [PSI+] loss was determined by the number of red colonies arising . Semidenaturing detergent agarose gel electrophoresis ( SDD-AGE ) , SDS-PAGE , quantitative immunoblotting and SDS-sensitivity experiments were performed as previously described [73] . To analyze the fate of aggregated Sup35 , cultures were grown to midlog phase and treated with cycloheximide ( CHX ) or both CHX and guanidine HCl ( GdnHCl ) for 1 . 7 hours . Yeast lysates were collected before and after treatment and incubated at 53°C and 100°C in the presence of 2% SDS before analysis by SDS-PAGE . Lysates were also prepared from the same cultures and analyzed by SDD-AGE . The number of propagons per cell was determined using a previously described in vivo dilution , colony-based method [75] . For propagon counting in mothers and daughters , a pair of mother and daughter cells was separated by micromanipulation onto minimal medium ( SD-complete with 2 . 5mM adenine and 4% dextrose ) with 3mM GdnHCl . After growing at 30°C for about 48 h , whole colonies were isolated using a cut pipette tip , resuspended in a small volume of water and plated onto YPD plates . The number of white colonies was then counted . Daughters were separated by FACS based on bud-scar labeling . Yeast cells were incubated for 1 h at room temperature in 1μg/ml Alexa-647 wheat germ agglutinin ( WGA ) in PBS . After washing twice in PBS , cells with the lowest fluorescence intensity ( 5% ) were sorted as newborn daughter cells , and a sample of this fraction was viewed by fluorescence microscopy to confirm bud scar number . This fraction was also moved to rich medium ( 1/4 YPD ) for color development . For Hsp104 inhibition , sorted fractions were first moved to a minimal medium with 3mM GdnHCl for three hours before being transferred to rich medium . In each case only completely red colonies were counted as [psi-] . Fluorescence microscopy was performed on a DeltaVision deconvolution microscope equipped with a 100x objective . WGA Alexa-647 fluorescence was collected using 650nm excitation and 668nm emission wavelengths and with an exposure of 50ms . Images were processed in ImageJ software . Cultures were grown in YPAD medium to an OD600 of 0 . 1 at 30°C . GdnHCl was added to 3mM , and the culture was returned to 30°C for 5 hours to decrease the propagon number . Cultures were then collected by centrifugation , washed and transferred to YPAD medium without GdnHCl for recovery . Samples were plated on YPD , and the number of propagons per cell was counted at the indicated timepoints . | Prion proteins adopt alternative conformations and assemble into amyloid fibers , which have been associated with human disease . These fibers are highly stable and self-replicate , leading to their persistence and resulting in a set of progressive and often fatal disorders . Inhibitors have been shown to interfere with some conformations but not others , suggesting that distinct strategies must be developed to target each . However , we show here that a single dominant-negative mutant can inhibit multiple conformations of the same prion protein through the same pathway but at distinct doses . Thus , the basis of this specificity is sensitivity rather than resistance to the mechanism of inhibition , suggesting that common strategies may be used to target a range of prion conformations . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"animal",
"diseases",
"medicine",
"and",
"health",
"sciences",
"molecular",
"probe",
"techniques",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"immunoblotting",
"light",
"microscopy",
"cell",
"disruption",
"fungi",
"animal",
"prion",
"diseases",
... | 2017 | A dominant-negative mutant inhibits multiple prion variants through a common mechanism |
In many organisms , dietary restriction appears to extend lifespan , at least in part , by down-regulating the nutrient-sensor TOR ( Target Of Rapamycin ) . TOR inhibition elicits autophagy , the large-scale recycling of cytoplasmic macromolecules and organelles . In this study , we asked whether autophagy might contribute to the lifespan extension induced by dietary restriction in C . elegans . We find that dietary restriction and TOR inhibition produce an autophagic phenotype and that inhibiting genes required for autophagy prevents dietary restriction and TOR inhibition from extending lifespan . The longevity response to dietary restriction in C . elegans requires the PHA-4 transcription factor . We find that the autophagic response to dietary restriction also requires PHA-4 activity , indicating that autophagy is a transcriptionally regulated response to food limitation . In spite of the rejuvenating effect that autophagy is predicted to have on cells , our findings suggest that autophagy is not sufficient to extend lifespan . Long-lived daf-2 insulin/IGF-1 receptor mutants require both autophagy and the transcription factor DAF-16/FOXO for their longevity , but we find that autophagy takes place in the absence of DAF-16 . Perhaps autophagy is not sufficient for lifespan extension because although it provides raw material for new macromolecular synthesis , DAF-16/FOXO must program the cells to recycle this raw material into cell-protective longevity proteins .
Dietary restriction , the reduced intake of food without malnutrition , increases the lifespan of many organisms , from yeast to mammals [1] . Dietary restriction increases lifespan , at least in part , by reducing the activities of pathways involved in growth and nutrient processing , including the TOR ( Target Of Rapamycin ) pathway . Inhibition of the TOR pathway extends lifespan in yeast , worms and flies [2–5] , and dietary restriction cannot further extend the lifespans of yeast , worms or flies in which the TOR pathway has been inhibited [3 , 4 , 6] . This suggests that down-regulation of the TOR pathway plays an important role in the longevity response to food limitation . TOR regulates several processes that could be involved in the longevity response to dietary restriction . For instance , TOR stimulates protein synthesis in yeast and in mammals by modulating key components of the translation machinery , including the ribosomal-protein S6 kinase ( S6K ) and the translation initiation factor 4E-binding protein ( 4E-BP ) . Inhibition of positive regulators of translation , including S6K , extends lifespan in both worms and flies [3 , 6–10] and inhibition of the negative regulator 4E-BP shortens lifespan in flies [11] . One could imagine that TOR inhibition extends lifespan solely by inhibiting protein synthesis . However , another process regulated by TOR , autophagy [12] , could also potentially influence the longevity of animals subjected to dietary restriction . Macroautophagy ( hereafter referred to as autophagy ) is a process in which portions of the cytoplasm , including mitochondria and other organelles , are degraded under conditions of nutrient limitation , allowing cellular macromolecules to be catabolized and recycled . During autophagy , large double-membrane vesicles , called autophagosomes , are generated and degraded in lysosomes , together with their contents . The breakdown products are subsequently recycled to the cytoplasm [13] . The regulation of autophagy has been studied extensively in yeast [14] . In this organism , autophagy is controlled by the ATG genes , many of which have functional homologs in other organisms [13 , 15] . In yeast , TOR inhibits the protein kinase Atg1 , which would otherwise mediate an early activation step in the autophagic process [16] . In response to Atg1 activity , the Vps34 complex , which contains the Class III phosphatidylinositol-3-kinase Vps34 as well as Atg6/Vps30 , the ortholog of the mammalian protein Beclin1 , stimulates and nucleates the formation of autophagosomes [14 , 17 , 18] . Autophagy is induced under conditions of stress , including nutrient limitation . For instance , dietary restriction stimulates autophagy in old rodents [19–21] , and in C . elegans larvae that enter a state of diapause , called dauer , in response to food limitation and crowding [22] . The process of autophagy has been linked to lifespan extension in long-lived insulin/IGF-1-pathway mutants . Mutations in components of the insulin/IGF-1 signaling pathway extend lifespan in many organisms [23 , 24] . In C . elegans , strong inhibition of the insulin/IGF-1 signaling pathway induces dauer formation , and weaker inhibition permits growth to adulthood and extends adult lifespan . Both daf-2-mutant dauers and adults exhibit increased levels of autophagy , and autophagy is required for their long adult lifespans [22] . RNAi inhibition of several autophagy genes , including ATG6/beclin1/bec-1 , prevents daf-2 mutations from extending lifespan , but has only minor effects on the lifespan of wild-type animals [22 , 25] . Despite the link between nutrient limitation and autophagy , it is not known whether autophagy plays a direct role in the longevity response to dietary restriction . In this study , we find that both dietary restriction and inhibition of the TOR pathway stimulate autophagy in C . elegans , and inactivation of genes required for autophagy specifically prevents these conditions from extending lifespan . We find that autophagy , like lifespan extension itself , is not a passive consequence of food limitation , but instead involves specific transcriptional control . Finally , our findings indicate that autophagy is neither necessary nor sufficient to extend lifespan in C . elegans , rather , autophagy appears to be an essential aspect of certain longevity pathways that are linked to nutrition .
To address the role of autophagy in the longevity response to dietary restriction , we made use of eat-2 ( ad1116 ) mutants , which are a genetic model for dietary restriction in C . elegans [26] . These mutants have defects in a pharyngeal nicotinic acetylcholine receptor subunit that is required for pharyngeal pumping ( feeding ) [27] . eat-2 mutants are long lived , and share many characteristics of animals that are directly food limited . These include a pale , thin morphology [26] , a lifespan extension that is dependent on the pha-4/FOXA transcription factor [28] , but independent of daf-16/FOXO ( a transcription factor required for the longevity of daf-2 insulin/IGF-1-receptor mutants ) [26 , 29] , reduced and prolonged progeny production [30] , and a characteristic spectrofluorimetric profile [31] . While these studies were in progress , Pilon's group reported increased levels of autophagy during the development of several feeding defective C . elegans mutants whose adult longevity phenotypes have not been well characterized [32] . To ask if autophagy occurs in eat-2 mutants and in animals subjected to direct dietary restriction , we visualized a GFP-tagged LGG-1 protein involved in autophagy ( Figure 1A ) [22] . LGG-1 is the worm ortholog of the vacuolar protein Atg8/MAP-LC3 , which is incorporated into pre-autophagosomal and autophagosomal membranes . In C . elegans , LGG-1::GFP is localized to puncta or foci in cells that are known to have increased numbers of autophagic vesicles [22 , 33] . The appearance of LGG-1::GFP-containing puncta has been used widely as an indicator of autophagy in C . elegans [22 , 32–35] . We found that the low number of autophagic events in wild-type L3 animals was increased ∼2 . 5-fold in eat-2 ( ad1116 ) mutants ( Figure 1B , p < 0 . 0001 , t-test ) . The longevity response to dietary restriction can be triggered in adults , and , consistent with this , we also observed increased levels of LGG-1::GFP-containing foci in the seam cells of adult eat-2 animals ( data not shown ) . In addition , we found that wild-type L3 animals subjected to direct dietary restriction by food limitation [30 , 36] exhibited a large increase in the number of autophagic puncta ( Figure 1C ) . Is autophagy required for the long lifespan induced by dietary restriction ? To investigate this , we inhibited the autophagic gene ATG6/beclin1/bec-1 in eat-2 mutants using RNAi . Because dietary restriction extends lifespan when initiated during adulthood [36] , we subjected the animals to RNAi on day-1 of adulthood by transferring them to culture dishes containing bacteria expressing bec-1 dsRNA . In this way , we were able to circumvent the requirement for bec-1 function during development [37] . We found that both of two different bec-1 RNAi clones shortened the mean lifespan of eat-2 ( ad1116 ) mutants by ∼15–30% ( Figure 2A; Table 1 ) , but did not shorten wild-type lifespan ( Figure 2B; Table 1 ) . In C . elegans , BEC-1 interacts with the class III PI3 kinase VPS-34 ( LET-512 ) [37] , an essential protein required for autophagy , membrane trafficking and endocytosis . We therefore asked whether vps-34 was also required for the long lifespan of eat-2 mutants . As with bec-1 RNAi , treating eat-2 mutants with vps-34 RNAi on day-1 of adulthood significantly shortened their long lifespan , but not that of wild type ( Figure 2C and 2D; Table 1 ) . Consistent with a role for bec-1 and vps-34 in autophagy , bec-1 and vps-34 RNAi disturbed the morphology and reduced the number of LGG-1 foci in the L3 progeny of eat-2 animals exposed to RNAi for their entire life ( see Methods ) ( Figure 3A and data not shown ) . We looked for a similar perturbation in LGG-1::GFP puncta under the RNAi conditions that we used to assay lifespan; that is , in eat-2 mutants treated with bec-1 or vps-34 RNAi from day-1 of adulthood . We did not observe a change in the LGG-1::GFP pattern within the first two days of adulthood ( in this or any other adult-only RNAi treatment we performed , including our experiments with daf-2 mutants [data not shown] ) . After day-2 of adulthood , the level of endogenous fluorescence , which increases with age , overwhelmed the LGG-1::GFP signal ( see Methods ) . Thus later disruption of the LGG-1::GFP pattern , which seems likely , could not be observed . BEC-1/Beclin1 is also known to interact with CED-9/Bcl-2 [37 , 38] , a protein that inhibits apoptosis . Therefore , we repeated the bec-1-RNAi experiment in animals in which cell death had been prevented using a caspase mutation , ced-3 ( n1289 ) . We found that bec-1 RNAi shortened the lifespan of ced-3 ( - ) ; eat-2 ( - ) mutants , as with eat-2 ( - ) single mutants ( data not shown ) , arguing against a longevity role for bec-1 in apoptosis . Taken together , these findings imply a requirement for autophagy in the longevity response to dietary restriction . Next , we asked whether RNAi treatments predicted to disrupt autophagy affected other phenotypes produced by dietary restriction . We found that eat-2 ( ad1116 ) mutants fed bec-1 or vps-34 RNAi-bacteria from hatching had the same low pumping rates as eat-2 ( ad1116 ) animals raised on control bacteria ( Figure S1 ) . In addition , feeding bec-1 RNAi-bacteria to eat-2 ( ad1116 ) animals did not have any effect on the brood size or the timing of the progeny production ( Figure S2 ) . We also asked whether inhibition of bec-1 affected the characteristic spectrophotometric spectrum of eat-2-mutant adults . Aging worms normally accumulate various fluorescent compounds that have a distinctive absorption maximum , and eat-2 mutants and wild-type animals subjected to dietary restriction exhibit a decrease in the absorption maximum of these age-related pigments [31] . We found that bec-1 RNAi fed to animals during adulthood did not significantly alter the fluorimetric profile of eat-2 mutants ( Figure S3 ) . Together these findings suggest that autophagy is specifically required for the longevity response to dietary restriction . How might dietary restriction induce autophagy ? As described above , dietary restriction appears to extend lifespan , at least in part , by down-regulating the TOR pathway , and inhibition of TOR is known to trigger autophagy in yeast and mammals [12] . To ask whether this was also the case in C . elegans , we assayed the levels of LGG-1 puncta in animals fed bacteria expressing TOR ( let-363 ) dsRNA . When we fed wild-type animals TOR RNAi-bacteria for their entire lives , we observed a significant increase in the number of autophagic vesicles in their L3 progeny ( Figure 4A ) , whose development , like that of let-363 ( h98 ) /TOR mutants , was arrested [2] . We also investigated the level of autophagy in mutants heterozygous for the TOR-binding partner daf-15/Raptor [39] . We found that daf-15 heterozygotes had increased levels of LGG-1::GFP-containing foci during development and as adults compared to wild-type animals ( Figure 4B and data not shown ) . Thus , the TOR pathway appears to regulate autophagy in C . elegans . To determine whether autophagy was likely to be required for the long lifespan of animals with reduced TOR activity , we asked whether the longevity of daf-15/Raptor mutants [39] ( Table 2 ) required bec-1 . We found that feeding bacteria expressing bec-1 dsRNA to adult daf-15 heterozygotes significantly shortened their lifespan in each of two independent experiments , but had no effect on wild-type animals ( Figure 4C and 4D; Table 2 ) . bec-1 RNAi slightly shortened the lifespan of wild-type animals when administered throughout life ( as reported earlier [22] ) , but we found that bec-1 RNAi shortened the lifespan of the long-lived daf-15 heterozygotes to a greater extent ( Table 2 ) . Taken together , these observations suggest that autophagy is required for the lifespan extension produced by the inhibition of TOR-pathway activity , and support the idea that dietary restriction induces autophagy via TOR inhibition in C . elegans . The small GTPase rab-10 appears to play a key role in the longevity response to dietary restriction in C . elegans [40] . As with TOR inhibition , rab-10 inhibition extends the lifespan of normal , well-fed animals , but does not further extend the lifespan of animals subjected to dietary restriction . rab-10 mRNA levels fall in response to dietary restriction , suggesting that the down-regulation of rab-10 activity plays a causal role in the longevity response to dietary restriction . Like animals subjected to dietary restriction , animals with reduced rab-10 activity also have delayed reproduction [40] and we found that they exhibited the dietary restriction-specific spectrofluorometric profile ( Figure S4 ) . RAB-10 is involved in vesicle transport in intestinal cells in C . elegans [41] and in mammalian epithelial cells [42 , 43] . In addition , RAB-10 was recently shown to regulate glutamate receptor recycling in neurons in C . elegans [44] . Because vesicle sorting is altered during autophagy , and because autophagy is increased in response to dietary restriction , we asked whether rab-10 inhibition might trigger autophagy . To do this , we subjected wild-type animals carrying the LGG-1::GFP reporter to rab-10 RNAi for their entire lives , and examined their L3 progeny . We found that this treatment , as well as the rab-10 ( ok1494 ) mutation , increased the number of LGG-1 foci in larvae ( Figure 5A and 5B ) and in adults ( data not shown ) . We also asked if autophagy might be required for rab-10 mutants to live long . To perform this experiment , we used a rab-10 ( ok1494 ) deletion mutant , which , as expected , was long-lived ( Figure 5C and 5D; Table 3 ) . We measured the lifespan of rab-10 ( ok1494 ) animals fed either bec-1 or vps-34 RNAi during adulthood , and we found that both RNAi clones significantly shortened lifespan ( Figure 5C ) . Taken together , these findings suggest that rab-10 inhibition is part of the mechanism by which dietary restriction stimulates autophagy . One could imagine that dietary restriction stimulates autophagy via purely post-translational mechanisms , such as changes in phosphorylation . However , recently the response to dietary restriction was shown to be subject to transcriptional regulation [28 , 45] . The FOXA transcription factor PHA-4 is required for the lifespan extension of animals subjected to dietary restriction [28] . Thus it was interesting to ask whether PHA-4 was required for dietary restriction to trigger autophagy . To do this , we fed pha-4 RNAi-bacteria to eat-2 animals expressing the LGG-1::GFP reporter for their entire lives and counted the number of GFP puncta in their progeny at the L3 stage . We found that the number of puncta was reduced significantly ( Figure 3A , control bacteria: 1 . 00 ± 0 . 05 SEM , pha-4 RNAi: 0 . 68 ± 0 . 04 SEM , p = 0 . 0001 , unpaired t-test ) . Thus , changes in transcription mediated by PHA-4 are likely to be required for dietary restriction to trigger autophagy . In principle , PHA-4 could trigger autophagy by reducing rab-10 transcription in response to dietary restriction . In this model , pha-4 would not be required to stimulate autophagy in animals already compromised for rab-10 function . However , we found that feeding pha-4 dsRNA significantly decreased the elevated number of LGG-1::GFP-containing foci in L3 stage rab-10 ( ok1494 ) mutants ( Figure 3B ) . This finding suggests that pha-4 acts either parallel to or downstream of rab-10 to regulate autophagy . Unlike animals subjected to dietary restriction , PHA-4/FOXA is not required for the increased longevity of daf-2/insulin/IGF-1-like mutants ( [28] and confirmed by us [data not shown] ) . Consistent with this , we found that the elevated levels of LGG-1::GFP foci in L3 stage daf-2 ( e1370 ) mutants [22] were not significantly affected by subjecting the animals to pha-4 RNAi ( Figure 6A ) . The long lifespan of daf-2 mutants is dependent on a different forkhead-family transcription factor , daf-16/FOXO [47] . Therefore , we asked whether daf-16 was required for daf-2 mutations to induce autophagy . daf-16 is known to act during adulthood to extend the lifespan of daf-2 mutants [48] . We found that this was also the case for bec-1 and vps-34: subjecting daf-2 ( mu150 ) animals to either bec-1 or vps-34 RNAi only during adulthood shortened lifespan ( Figure S5; Table S2 ) . This observation is consistent with earlier findings that treating daf-2 ( e1370 ) mutants with bec-1 RNAi throughout their lives significantly shorten their lifespan [22] . To ask whether daf-16 was required for autophagy in daf-2 mutants , we introduced a daf-16 ( null ) mutation into the daf-2 ( e1370 ) mutant and counted the number of LGG-1::GFP foci in the double mutant . We found that the daf-16 ( mu86 ) mutation had no effect on the level of foci in daf-2 ( e1370 ) larvae or adults ( Figure 6B and data not shown ) . This finding suggests that daf-16 is not required for the increased levels of autophagy in daf-2 mutants , and , conversely , that autophagy is not sufficient to extend lifespan . In addition , we found that bec-1 RNAi had no effect on the expression of any of the three transcriptional daf-16 target genes we investigated ( sod-3 , mtl-1 and dod-8 , data not shown ) . Together , these findings suggest that bec-1 and daf-16 act in parallel pathways to increase the lifespan of daf-2 mutants . IThe great majority of the phenotypes observed in insulin/IGF-1-pathway mutants require the daf-16/FOXO transcription factor . Therefore , it was striking to find that autophagy appears to be induced in daf-2 mutants independently of daf-16 . While these studies were in progress , the Jacobson group showed that protein turnover in daf-2 mutants is increased in a daf-16-independent fashion [49]; perhaps this turnover occurs , at least in part , via autophagy .
The process of autophagy allows an animal to recycle macromolecules during times of starvation and stress , presumably to deploy scarce resources in a more beneficial fashion . In this study , using a GFP reporter that indicates the presence of autophagic vesicles [22 , 32–35] , we have shown that autophagy is triggered in long-lived animals subjected to dietary restriction in C . elegans . To test whether autophagy is required for the longevity of animals subjected to dietary restriction , we inhibited the activities of two genes required for autophagy , bec-1 [ATG6/VPS30/Beclin1] and the PI 3-kinase vps-34 , and found that the treatment prevented food-limited eat-2 mutants from living long . Together , these findings suggest that autophagy is required for dietary restriction to extend lifespan . ( We note that , while this paper was under revision , Beth Levine's group independently reported that autophagy genes are required for the longevity of eat-2 mutants [34] . ) Disrupting genes required for autophagy did not perturb other phenotypes normally associated with dietary restriction , including morphological , spectrofluorimetric or reproductive changes . Thus , autophagy appears to be required specifically for lifespan extension . Perhaps autophagy allows an animal to clear away damaged proteins and other macromolecules that could accelerate the aging process and recycle their component amino acids into new cellular components . Is it possible that this interpretation is incorrect , and that bec-1 and vps-34 actually have different functions in the animal that are required for longevity ? In support of our interpretation , both bec-1 and vps-34 were required for the increased number of LGG-1::GFP-labeled autophagic vesicles we observed in eat-2 and daf-2 larvae ( see Figures 3A and 6A ) and LC3/LGG-1 is not known to have functions in processes other than autophagy . However , we did not observe changes in the adult LGG-1::GFP pattern when we produced changes in lifespan by inhibiting bec-1 or vps-1 function on day-1 of adulthood , though we were unable to assay LGG-1::GFP after day-2 , when the adults are still very young ( data not shown ) . This finding does not invalidate our interpretation , because it is possible that LGG-1::GFP recycling takes some time . Moreover , we observed the same phenomenon with two genes widely thought to influence autophagy: daf-2 and let-363/TOR . ( In our hands , daf-2 and let-363/TOR RNAi administered on day-1 of adulthood lengthened lifespan but did not induce an autophagic phenotype by day-2 of adulthood [data not shown] . ) What other functions could bec-1 and vps-34 have ? In addition to their roles in autophagy , Vps34 is also required for endocytosis [50 , 51] . Likewise , ATG6/VPS30/Beclin1 is involved in both autophagy and endocytosis in yeast , though Beclin1 is specifically involved in autophagy in mammals [52] . It is possible that bec-1 also regulates endocytosis in C . elegans , although C . elegans bec-1 ( + ) complements only the autophagy and not the vacuolar protein sorting function of yeast lacking VPS30 function [22] . Thus , knocking down bec-1 and vps-34 with RNAi could potentially shorten lifespan , at least in part , by blocking endocytosis . However , since bec-1 and vps-34 RNAi specifically affect the lifespans of long-lived mutants that have an elevated autophagic phenotype , we favor the interpretation that they shorten lifespan primarily by inhibiting autophagy . In addition to recycling cytoplasmic contents , autophagy is also involved in non-apoptotic , programmed cell death [53] . Physiological levels of autophagy are essential to C . elegans cell survival during starvation , whereas excessive or insufficient levels of autophagy promote organismal death [35] . While it has not been observed so far , it is possible that non-autophagic cell death contributes to the longevity induced by dietary restriction . We were prompted to investigate the role of autophagy in dietary restriction in part because autophagy is regulated by TOR , which in turn behaves as a downstream effector of the longevity response to dietary restriction in genetic tests [3 , 4 , 6] . In this study , we showed that TOR regulates autophagy in C . elegans and that genes required for autophagy are also required for the lifespan extension of TOR-pathway mutants . This finding suggests that autophagy is an integral part of the mechanism by which TOR inhibition increases lifespan , and supports the idea that dietary restriction extends lifespan via TOR inhibition . TOR inhibition also reduces the rate of protein synthesis , and inhibiting protein synthesis is sufficient to extend lifespan . Previously we suggested that the longevity of TOR mutants might be caused , in part , by reduced protein synthesis [6] . However , these and other new findings put a new twist into this line of reasoning . Recently , Kapahi's group showed that bec-1 RNAi does not prevent S6-kinase/rsks-1 or eIF-4G/ifg-1 mutations , which reduce protein synthesis , from extending lifespan in C . elegans [9] . We observed this , as well , for rsks-1 ( sv31 ) and ife-2 ( ok306 ) ( Figure S6A; Table S3; and data not shown ) . In addition , we looked for LGG-1::GFP foci in rsks-1 mutants in which protein synthesis had been inhibited and failed to see any increase in the number of LGG-1::GFP positive foci in seam cells ( Figure S6B ) . While it is possible that autophagy is taking place in other cells/tissues in the animal , the simplest interpretation of these findings is that the lifespan extension produced by the inhibition of protein synthesis does not involve autophagy . Thus , these findings raise an interesting question: If protein synthesis falls in response to dietary restriction , and the lifespan extension produced by inhibiting protein synthesis does not involve autophagy , why is the lifespan extension produced by dietary restriction dependent on autophagy genes ? One possibility is that disrupting protein synthesis in well-fed animals triggers a novel , lifespan-extending pathway that is not triggered by dietary restriction ( see model in Figure 7 ) . This seems plausible , since the reduction in protein synthesis caused by dietary restriction takes place in the context of a global physiological shift that down-regulates many other growth-related processes . Consistent with the idea that dietary restriction/TOR inhibition and direct protein synthesis activate distinct longevity pathways , the lifespan of eat-2 mutants is further extended by direct protein synthesis inhibition but not by TOR inhibition [6] . The idea that inhibiting protein synthesis in well-fed animals activates a novel longevity pathway does not rule out the possibility that the decrease in protein synthesis that occurs in response to TOR inhibition or dietary restriction , like autophagy , is required for increased longevity . It will be interesting to explore these pathways in more detail with biochemical and molecular experiments . Our studies have placed two new genes into the pathway by which dietary restriction triggers autophagy , rab-10 and pha-4 . rab-10 encodes a small GTPase whose mRNA levels fall in response to dietary restriction . rab-10 inhibition appears to be part of the mechanism by which dietary restriction extends lifespan . When rab-10 is inhibited with RNAi , a robust dietary-restriction phenotype ( lifespan extension , delayed reproduction , spectroflurorimetric shift ) is produced ( [40] and Figure S4 ) , and rab-10 RNAi does not further extend the lifespan of eat-2 mutants [40] . Our findings indicate that rab-10 inhibition stimulates autophagy . RAB-10 is involved in vesicle transport in intestinal cells in C . elegans [41] and in mammalian epithelial cells [42 , 43] . RAB-10 has also been shown to regulate glutamate receptor recycling in neurons in C . elegans [44] . Together these findings suggest the hypothesis that dietary restriction alters patterns of vesicle transport in a way that triggers autophagy and perhaps other events that promote lifespan extension . The transcription factor PHA-4 , which is required for the longevity of animals subjected to dietary restriction , is required for the elevated number of autophagic vesicles observed in eat-2 mutants [28] . Thus , the increase in autophagy that occurs in response to dietary restriction is not a passive consequence of food limitation but is likely to require changes in gene expression . It will be interesting to learn what genes act downstream of pha-4 to regulate autophagy . PHA-4 is also required for the inhibition of rab-10 to induce autophagy , so pha-4 may act downstream of rab-10 in the autophagy pathway . Perhaps changes in the pattern of vesicle transport are part of the signal that activates PHA-4 in response to dietary restriction . Alternatively , in food-limited animals , PHA-4 could regulate the expression of one or more genes that acts in the context of altered vesicle metabolism to induce autophagy . Autophagy seems like such a “purifying” process that it is tempting to think that it might be sufficient to extend lifespan . However , our findings suggest that is not the case . The longevity of daf-2 insulin/IGF-1 receptor mutants requires the FOXO-family transcription factor DAF-16 . Surprisingly , we found that daf-16 ( null ) ; daf-2 ( - ) double mutants had the same high level and distribution of autophagic LGG-1::GFP puncta as did daf-2 ( - ) single mutants . The fact that daf-16; daf-2 double mutants are not long-lived ( [47] and data not shown ) suggests that autophagy is not sufficient to increase lifespan . Why are transcription factors as well as autophagy required for lifespan extension in daf-2 mutants ? DAF-16/FOXO is known to stimulate the expression of a wide variety of antioxidant , chaperone , antimicrobial , metabolic and other genes that act in a cumulative fashion to extend lifespan [54–57] . Perhaps the role of autophagy in the longevity of daf-2 mutants is to provide new raw material for protein construction by recycling damaged cellular components , and the role of DAF-16 is to channel this raw material into proteins that protect and repair cells , and thereby extend lifespan . Not only is autophagy insufficient to extend lifespan , it is not necessary for lifespan extension . We found that subjecting the long-lived mitochondrial mutants clk-1 and isp-1 to bec-1 or vps-34 RNAi during adulthood has no effect on lifespan ( Figure S7; Table S3 ) ( though autophagy could conceivably play a longevity role in mitochondrial respiration during development; Table S4 ) . In addition , as discussed above , inhibiting protein synthesis in otherwise well-fed animals extends lifespan in the absence of autophagy . Taken together , these findings suggest that autophagy may be required specifically for longevity pathways that are fully integrated with , and regulated by , environmental signals that reflect the availability of food , such as the insulin/IGF-1 pathway and the response to dietary restriction .
All strains were maintained as previously described [58] . Single mutants: CF1037: daf-16 ( mu86 ) I , CF2846: rab-10 ( ok1494 ) I ( VC1026 outcrossed four times to Kenyon lab N2 wild-type strain ) , CF1908: eat-2 ( ad1116 ) II ( DA1116 outcrossed four times to Kenyon lab N2 wild-type strain ) , CF1041: daf-2 ( e1370 ) III , CF512: fer-15 ( b26 ) II; fem-1 ( hc17 ) III . CF1844: fer-15 ( b26 ) II; daf-2 ( mu150 ) III; fem-1 ( hc17 ) IV . VB633: rsks-1 ( sv31 ) III . Double mutants: CF1850: eat-2 ( ad1116 ) rrf-3 ( pk1426 ) II , CF2120: daf-2 ( mu150 ) III; ced-3 ( n1289 ) IV , CF2140: eat-2 ( ad1116 ) II; ced-3 ( n1289 ) IV [59] , DR412: daf-15 ( m81 ) /unc-24 ( e138 ) IV [39] . Transgenic strains: QU1: izEx1[Plgg-1::gfp::lgg-1 + rol-6] ( [22] , named in this study ) , QU2: daf-2 ( e1370 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] [22] , named in this study ) , CF2494: eat-2 ( ad1116 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] , CF2946: eat-2 ( ad1116 ) rrf-3 ( pk1426 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] , CF2544: daf-16 ( mu86 ) ; daf-2 ( e1370 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] , CF2821: daf-15 ( m81 ) /unc-24 ( e138 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] , CF2864: rab-10 ( ok1494 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] , CF2865: rsks-1 ( sv31 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] , CF2866: isp-1 ( qm150 ) ; izEx1[Plgg-1::gfp::lgg-1 + rol-6] . The identity of all RNAi clones was verified by sequencing the inserts using the M13-forward primer . The TOR RNAi clone was obtained from Dr . Xiaomeng Long , Massachusetts General Hospital . The daf-2 RNAi clone was published previously [48] . All other clones were from Julie Ahringer's RNAi library [60] or Marc Vidal's RNAi library [61] . The gene bec-1 is part of an operon that contains the stress-inducible transcription factor gene skn-1 , which is required for the lifespan extension induced by dietary restriction [45] . Non-specific inactivation of genes in operons by RNAi has been observed [62 , 63] . However , using quantitative RT-PCR , we found that RNAi of bec-1 did not affect the mRNA levels of skn-1 ( data not shown ) . Thus , the phenotypes observed in animals treated with bec-1 RNAi are likely to originate from reduced bec-1 mRNA levels . Lifespan analysis was conducted at 20 °C as described previously [40] unless stated otherwise . RNAi treatments were either performed as whole-life treatments or adult-only treatments . In the whole-life RNAi treatments , eggs were added to plates seeded with the RNAi-bacteria of interest . In the adult-only analysis , eggs were added to plates seeded with RNAi vector-only bacteria , and adult animals were transferred to gene-specific RNAi-bacterial plates . The chemical 2'fluoro-5′deoxyuridine ( FUDR , Sigma ) was sometimes added to adult worms ( 100 μM ) to prevent their progeny from developing . During this project , we experienced a time period in which the bec-1 RNAi clone failed to shorten the lifespan of eat-2 mutants in the presence of FUDR . These experiments were not included in this publication and we continued our experiments without FUDR . At least 80 worms were tested in each experiment . Strains were grown at 20 °C under optimal growth conditions for at least two generations before use in lifespan analysis . During the analysis of large numbers of RNAi clones , CF512 or N2 controls were performed either concurrently or in overlapping time frames . In all experiments , the pre-fertile period of adulthood was used as t = 0 for lifespan analysis . Censoring in the lifespan analysis included animals that ruptured , bagged ( i . e . , exhibited internal progeny hatching ) , or crawled off the plates . STATA software was used for statistical analysis and to determine means and percentiles . In all cases , p values were calculated using the Log-rank ( Mantel-Cox ) method . The level of autophagy in various mutants was assessed using an LGG-1::GFP translational reporter characterized previously [22] . Animals were raised at 20 °C . GFP-positive puncta were counted ( using 1000-fold magnification on a Zeiss Axioplan II microscope ) in the seam ( lateral epidermal ) cells of L3 transgenic animals , which were staged by gonad morphology and germline developmental phenotype . Counting puncta during adulthood was difficult due to the increased level of endogenous autofluorescence in the animal ( data not shown ) . In addition , examining puncta in adults was complicated by the difficulty in identifying seam cells . Between 3–10 seam cells were examined in each of 10–40 animals from at least two independent trials and averaged ( see Table S1 ) . Data analysis was done using unpaired , two-tailed t-test . When performing RNAi experiments to count LGG-1::GFP-positive foci , young adults were fed the RNAi bacteria , and the L3 progeny of their progeny ( “F2 generation” ) were examined . Analyzing the L3 animals in the first generation , even in daf-2 positive controls , was not sufficient to change the number of foci by the L3 stage ( data not shown ) . GFP-positive punctate areas were also counted in wild-type animals ( QU1 ) subjected to dietary restriction by direct food limitation . The bacterial culture was grown in a slightly modified , scaled up version of the protocol described in Gerstbrein et al . [31] , to yield cultures corresponding to ad libitum ( AL , or fully fed ) and dietary-restricted ( DR , or food-limited ) conditions . 500 μl of E . coli OP50 ( OD600 ∼1 ) was inoculated into 250 ml LB , grown for 5 hrs at 37 °C and resuspended in 25 ml complete S-basal medium . This culture corresponded to the stock as well as the ‘AL' culture . Cell density of the stock was determined by counting DAPI stained cells in a Petroff-Hausser counting chamber . The ‘AL' culture corresponded to a cell density of 1 . 9 × 1010 cells/ml and was diluted in complete S-basal medium to yield the ‘DR' culture ( 2 . 6 × 109 cells/ml ) . Worms grown in more dilute culture of cell density 5 . 2 × 108 cells/ml appeared to border on starvation while the worms grown in culture of cell density 2 . 6 × 108 cells/ml arrested . About 25 eggs were added to wells of a 24-well plate containing 600 μl of the bacteria-supplemented S-basal medium each and cultured at 20 °C with shaking . The media was changed every other day once the eggs developed into adults . Worms grown in the ‘DR' culture were considered to be dietary-restricted as they developed with a slight lag as compared to animals in ‘AL' culture and they had lower AGE pigments ( a biomarker of better healthspan and lifespan , data not shown ) . L3 animals were observed after ∼53 hours in ‘AL' culture and after ∼60 hours in ‘DR' culture . GFP-positive foci were counted in hypodermal seam cells of L3 transgenic animals . We note that although the genetic requirements for the longevity of eat-2 mutants and animals subjected to dietary restriction in liquid media are similar to one another [26 , 29 , 30] , initiating dietary restriction in a third way; namely , on plates during mid-adulthood , produces a lifespan increase with at least some different genetic requirements [64] . Therefore , it is possible that the role and regulation of autophagy in animals subjected to dietary restriction in different ways may not be the same . Eggs were incubated at 20 °C on control plates and 16 late-L4 stage worms were picked for each treatment and transferred to fresh RNAi or OP50 plates every 12 hours for 4–5 days . After this period , the worms were transferred every 24 hours . Worms that crawled off the plates , bagged or ruptured were censored . All progeny plates were incubated at 20 °C for about 2 days following transfer of the parental worms and then held at 4 °C . The number of worms that developed was determined at the end of the experiment . In vivo autofluorescence in C . elegans was measured using a spectrofluorimeter ( Fluorolog®-3 , Jobin Yvon Inc . , Edison NJ ) equipped with a plate reader ( MicroMax 384 ) . For each time point/scan , 50 animals per RNAi clone were cleaned on unseeded NGM plates , and then transferred to 50 μl of 10 mM NaN3 in a single well of a 96-well plate ( Cat #437842 , Nalge Nunc Internat'l ) . TRP and AGE fluorescence intensities and the excitation wavelength for maximal AGE fluorescence intensity were measured as described [31] . Each scan was done in triplicate . Data analysis was done using unpaired , one-tailed t-test . | Dietary restriction ( limited food intake ) increases lifespan in many organisms . However , the cellular processes underlying this fascinating phenomenon are still poorly understood . When an animal is starved , it degrades and recycles its organelles and other cellular components in a process called autophagy ( literally “self-eating” ) . Here , we have asked whether autophagy also occurs in response to dietary restriction , using the roundworm C . elegans for our studies . We find that autophagy does take place when food intake is limited . Moreover , inhibiting genes required for autophagy has little effect on well-fed animals but prevents food limitation from extending lifespan . This autophagy requires PHA-4/FOXA , a life-extension protein that regulates gene expression , suggesting that changes in gene expression are required for dietary restriction to stimulate autophagy . Because autophagy seems like such a rejuvenating process , it might seem to be sufficient to increase longevity . However , we find that , in long-lived hormone-pathway mutants , both autophagy and DAF-16/FOXO , another protein that controls gene expression , are required for longevity . We propose that autophagy frees up new resources for the cell , but that transcription factors like the DAF-16/FOXO protein must channel this raw material into new cell-protective proteins in order for lifespan to be increased . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"biochemistry",
"developmental",
"biology",
"caenorhabditis",
"molecular",
"biology",
"genetics",
"and",
"genomics"
] | 2008 | A Role for Autophagy in the Extension of Lifespan by Dietary Restriction in C. elegans |
DNA damage checkpoint activation can be subdivided in two steps: initial activation and signal amplification . The events distinguishing these two phases and their genetic determinants remain obscure . TopBP1 , a mediator protein containing multiple BRCT domains , binds to and activates the ATR/ATRIP complex through its ATR-Activation Domain ( AAD ) . We show that Schizosaccharomyces pombe Rad4TopBP1 AAD–defective strains are DNA damage sensitive during G1/S-phase , but not during G2 . Using lacO-LacI tethering , we developed a DNA damage–independent assay for checkpoint activation that is Rad4TopBP1 AAD–dependent . In this assay , checkpoint activation requires histone H2A phosphorylation , the interaction between TopBP1 and the 9-1-1 complex , and is mediated by the phospho-binding activity of Crb253BP1 . Consistent with a model where Rad4TopBP1 AAD–dependent checkpoint activation is ssDNA/RPA–independent and functions to amplify otherwise weak checkpoint signals , we demonstrate that the Rad4TopBP1 AAD is important for Chk1 phosphorylation when resection is limited in G2 by ablation of the resecting nuclease , Exo1 . We also show that the Rad4TopBP1 AAD acts additively with a Rad9 AAD in G1/S phase but not G2 . We propose that AAD–dependent Rad3ATR checkpoint amplification is particularly important when DNA resection is limiting . In S . pombe , this manifests in G1/S phase and relies on protein–chromatin interactions .
The DNA damage checkpoint is an elaborate signal transduction pathway that monitors the integrity of the DNA , prevents cell cycle progression and promotes appropriate DNA metabolism [1] reviewed in [2] . The DNA damage sensors associated with checkpoint activation define two separate DNA structure-dependent signal transduction cascades . Each pathway engages a phospho-inositol-3 kinase-like protein kinase ( PIKK ) ; either the Ataxia Telangiectasia Mutated ( ATM ) or the Ataxia Telangiectasia and Rad3 related ( ATR ) kinase [3] . ATM detects DNA double strand breaks ( DSBs ) by interaction with the Mre11-Rad50-Nbs1 repair complex , while ATR primarily senses single stranded-DNA ( ss-DNA ) through interactions with RPA . Both ATM and ATR are conserved in the model organisms S . pombe and S . cerevisiae . For ATR to recognise a DNA lesion , single-stranded DNA ( ssDNA ) needs to be formed - for example by DNA repair-dependent DNA processing [4] or following the replication machinery encountering the unrepaired lesion [5] . Once ssDNA is generated , it is immediately coated by replication protein A ( RPA ) ( Reviewed in: [6] ) . Multiple ATR molecules are initially recruited to ssDNA regions via ATRs obligate binding partner , ATRIP , which binds directly to RPA [7] , [8] . ATR-ATRIP recruitment to ssDNA-RPA is necessary for “basal” ATR activation , but is insufficient for full checkpoint activation: co-recruitment of a second complex consisting of three PCNA-like proteins , Rad9 , Hus1 and Rad1 ( known as the 9-1-1 clamp ) is also necessary . 9-1-1 is loaded in parallel to ATR recruitment at 5′ ssDNA/dsDNA junctions by the checkpoint clamp loader Rad17-RFC[2–5] [9] , [10] , [11] . ( Figure 1A ) . When ATR-ATRIP is first loaded at the site of ssDNA , its “basal” kinase activity promotes phosphorylation of its immediate neighbours , including ATRIP [12] , [13] , an in trans phosphorylation of a residue within ATR itself , T1989 [14] , and the subunits of the 9-1-1 clamp [15] , [16] . Dependent on the concomitant recruitment of 9-1-1 , a further protein , TopBP1 , is recruited [17] . TopBP1 is recruited via an interaction between its BRCT ( 1+2 ) domains and a constitutive phosphosphorylation on the C-terminus of Rad9 [18] , [19] . Similarly in both yeast systems , Saccharomyces cerevisiae and Schizosaccharomyces pombe , the TopBP1 homologs , Dpb11 and Rad4 respectively , are recruited by the phosphorylation of the C-terminus of Rad9Ddc1 creating a binding site for a pair of BRCT domains ( Figure 1A ) . In S . pombe , the C-terminal phosphorylations occur on Rad9 at residues T412 and S423 [20] . This subsequently recruits Rad4TopBP1 via interaction with BRCT pair ( 3+4 ) . However , unlike in mammalian cells , T412 and S423 in S . pombe are directly targeted by Rad3ATR in response to its ssDNA/RPA binding and concomitant 9-1-1 loading [16] , [20] , [21] . Despite these differences , in both S . pombe [20] and mammalian cells [14] , Rad4TopBP1 recruitment promotes the formation of a Rad3ATR/9-1-1/Rad4TopBP1 complex ( Figure 1A ) . However , the mode of interaction of Rad3ATR and Rad4TopBP1 within this complex has not been defined in S . pombe . Mammalian TopBP1 can directly activate ATR-ATRIP both in vitro in the absence of ssDNA/RPA and when over-expressed in cells . TopBP1-dependent ATR activation requires an ATR activation domain ( AAD ) situated between the 6th and 7th BRCT domains [17] and mutation of a conserved aromatic residue within this unstructured region , W1147 , prevents this mode of ATR activation . The AAD contacts a region within the C-terminus of ATR , between the kinase and FATC domains [22] , which has been termed the PIKK Regulatory Domain ( PRD ) . Mutation of a conserved PRD residue , K2598 , similarly abolishes TopBP1-dependent ATR activation . In both mammalian cells and Xenopus extracts the interaction between TopBP1 and ATR-ATRIP appears to be essential for checkpoint activation in response to replication stress [17] , [22] , although the initial in trans phosphorylation of ATR on T1989 , reportedly essential for full ATR activation , is TopBP1-independent [14] . In the budding yeast model system , the intrinsically disordered C-terminal extension of the TopBP1 homolog , Dpb11TopBP1 , contains an AAD which interacts with and activates Mec1ATR via a pair of aromatic residues , W700 and Y735 [22] , [23] , [24] , [25] . Interestingly , in S . cerevisiae , at least two distinct Mec1ATR activation domains have been identified: in addition to the Dbp11TopBP1 AAD , the C-terminal tail of Ddc1Rad9 ( S . cerevisiae homolog of the 9-1-1 subunit Rad9 ) contains an AAD that can directly stimulate Mec1ATR activity in vitro and contributes to checkpoint activation in vivo [11] , [26] . The key residues in Ddc1Rad9 required for Mec1ATR activation are W352 and W544 . W352 resides on the surface of the PCNA-like domain , while W544 lies within the intrinsically disordered C-terminus . In vivo the Ddc1Rad9 AAD is essential for Mec1ATR activation when S . cerevisiae cells are in G1 [24] , [26] , while the Ddc1Rad9 AAD acts redundantly with the C-terminal AAD of Dpb11TopBP1 during checkpoint activation in G2 . It is proposed that , at least in S . cerevisiae , a minimum of one other protein contains an equivalent AAD ( Reviewed in [27] ) . We have previously shown that S . pombe Rad4TopBP1 is not required for the Rad3ATR-dependent and DNA damage-dependent phosphorylation of Rad26ATRIP or the 9-1-1 clamp subunits [16] , [20] , demonstrating that Rad3ATR is active at sites of DNA damage in the absence of activation by the Rad4TopBP1 AAD . However , the presence of Rad4TopBP1 is clearly required to form a robust Rad3ATR/9-1-1/Rad4TopBP1 complex [20] , to recruit the Crb253BP1 mediator protein [28] , [29] and for Rad3ATR to phosphorylate downstream substrates such as Chk1-S345 [30] and promote robust checkpoint activation . To further explore the role of Rad4TopBP1 in checkpoint activation in S . pombe we identified and characterised the Rad4TopBP1 AAD . We show that Rad4TopBP1 can interact with Rad3ATR via its AAD and that the AAD contributes to Rad3ATR activation in vivo . We observe that the biological function of the Rad4TopBP1 AAD is most important in G1/S phase , when resection is limited , and that reducing DSB resection in G2 following ionising radiation results in compromised Chk1 phosphorylation in the absence of Rad4TopBP1 AAD function . In order to separate out and study Rad4TopBP1 AAD-dependent Rad3ATR activation we developed a Rad4TopBP1 AAD-dependent lacO-LacI checkpoint activation system for S . pombe and used this to show that Rad4TopBP1 AAD-dependent Rad3ATR activation is also dependent on histone H2A phosphorylation . Consistent with a role for this chromatin modification , mutations in Crb2 that interfere with phospho-binding by Crb2 also decrease Rad4TopBP1 AAD-dependent checkpoint activation . Thus , the Rad4TopBP1 AAD-dependent Rad3ATR activation pathway is chromatin dependent , implying a role in checkpoint amplification and maintenance .
Both rad4-Y599R and rad4-Δ[595–601] displayed normal cell cycle progression ( data not shown ) , indicating that , as is observed in S . cerevisiae [24] , the Rad4TopBP1 AAD is not required for unperturbed DNA replication . In response to UV , MMS and HU treatment , rad4-Y599R and rad4-Δ[595–601] cells showed intermediate sensitivity when compared to rad4+ and checkpoint defective rad3Δ ( Figure 1E and Figure S1A ) . To establish if the sensitivity to DNA damage correlated with a defective G2 DNA damage checkpoint , we monitored cell cycle progression after cells were synchronised in G2 and UV-irradiated . Following exposure to 50 Jm-2 ( Figure 2A ) , rad4-Y599R cells displayed premature release from cell cycle arrest ( ∼20 min earlier than rad4+ after 50 Jm-2 ) . We next monitored the sensitivity and checkpoint response to ionising radiation . rad4-Y599R mutant cells displayed only very mild sensitivity to IR ( Figure 2B ) and the checkpoint was mildly extended ( Figure 2C ) . We do not know the reason for the slight extension of the G2 delay: no obvious increase in numbers or duration of Rad22Rad52 foci were observed , indicating no significant delay to DSB repair ( Figure S1B ) . We next examined Chk1 phosphorylation status in rad4+ and rad4-Y599R cells as a surrogate for checkpoint activation ( Figure 2D ) . In response to 200 Jm-2 UV irradiation , asynchronously growing rad4-Y599R cells displayed reduced Chk1 phosphorylation when compared to rad4+ cells , consistent with the partial checkpoint defect observed . Conversely , in response to IR , no significant difference is seen between rad4-Y599R and rad4+ . An upstream target of Rad3ATR is the C-terminus of histone H2A [31] , [32] . To establish if the UV-specific defect in Rad3ATR-dependent phosphorylation is specific to Chk1 , we monitored γH2A formation following either UV or IR treatment ( Figure 2E ) . As was seen for Chk1 phosphorylation , a significant decrease in γH2A is observed in rad4-Y599R cells following UV but not IR treatment when compared to rad4+ cells . The pattern of DNA damage sensitivity seen for rad4-Y599R cells is consistent with a specific sensitivity within S phase . >70% of fission yeast cells in an asynchronous culture are in G2 and mitosis is followed rapidly by S phase: G1 is extremely short . In response to IR , the G2 DNA damage checkpoint is robustly activated and DNA repair completed before cells pass through mitosis and into S phase [33] . Thus , following IR , relatively few cells replicate damaged DNA . Conversely , the G2 checkpoint is not robustly activated following UV [34] and the majority of UV-irradiated cells pass through mitosis and enter S phase with damaged DNA . To monitor S phase-specific events , we thus examined γH2A induction in cells treated with either hydroxyurea ( HU ) , an inhibitor of ribonucleotide reductase , or Camptothecin ( CPT ) , an inhibitor of topisomerase I ( Figure 2F ) . Consistent with both agents manifesting cytotoxicity in S phase , γH2A levels were significantly reduced when comparing rad4-Y599R with rad4+ cells . Finally , since replication of UV damaged DNA induces Cds1Chk2 activity [35] , we monitored the kinase activity of immuno-precipitated Cds1Chk2 following UV irradiation of rad4-Y599R and rad4+ cells ( Figure 2G ) . Cds1Chk2 activation was reproducibly lower for rad4-Y599R , indicating an impaired S phase checkpoint activation ( n = 3 ) . If the rad4-Y599R mutant is deficient in activation of Rad3ATR in S phase , we would anticipate increased sensitivity to IR within S phase when compared to rad4+ cells . To test this possibility , rad4-Y599R mutant and rad4+ cells where either synchronised in G2 cells using cdc25-22 or in G1 using a cdc10-m17 . Following the block , cells were released by reducing the temperature and cell cycle progression was monitored by FACS analysis ( Figure 3A , 3B ) . Cells were irradiated with 50 Gy IR at the times indicated . rad4-Y599R cells showed significant increased sensitivity when compared to rad4+ when irradiated in S phase , but not when irradiated in G2 , when S phase is complete ( i . e . see Figure 3B ) . In an equivalent cdc25-22 block and release experiment , we monitored Chk1 phosphorylation and γH2A induction ( Figure 3C ) . Unlike when asynchronous rad4-Y599R cells are irradiated ( >70% of such cells are in G2 ) , when rad4-Y599R cells were irradiated in early-mid S phase , Chk1 phosphorylation was moderately reduced for the first 40 minutes after irradiation and γH2A levels were similarly decreased when compared to rad4+ . Interestingly , following progression through S phase and into G2 ( 150 minute time point ) , Chk1 phosphorylation levels increased significantly in rad4-Y599R cells , although the same was not seen for γH2A levels . To determine that the use of cdc25-22 synchronisation was not generating an artefact ( Cdc25 is an activator of Cdc2-Cyclin B , which itself is required for normal DNA damage responses in G2 [36] ) , we used the alternative method of synchronisation where cells were arrested in G1 using cdc10-M17 and released directly into S phase ( Figure 3D ) . Unlike IR treatment of asynchronous cultures where equivalent levels of γH2A were observed ( Async IR ) , treatment of rad4-Y599R cells at 30 , 60 or 90 minutes after release from arrest resulted in decreased γH2A levels and Chk1 phosphorylation when compared to rad4+ control cells . In S . cerevisiae , co-localisation of two or more checkpoint proteins to arrays of lacO repeats bypasses the requirement for DNA damage in Mec1-mediated checkpoint activation [37] . To establish the role of the Rad4TopBP1 AAD in a what has previously been characterised as an RPA-ssDNA independent system , rad4TopBP1 , rad9 and rad3ATR were each fused to a construct encoding GFP , the E . coli lac-repressor ( LacI ) and a nuclear localization signal ( NLS ) ; GFP/LN ( Figure 4A ) . The resulting plasmids express the fusion construct under the control of a thiamine-repressible ( nmt41 ) promoter . We established that each of the fusion constructs were functional by expressing them individually in the corresponding null mutants . Each was able to suppress the DNA damage sensitivity ( and for rad4TopBP1 , the thermosensitivity ) of the appropriate mutant , although for rad4-GFP/LN genotoxin resistance was not restored to wild-type levels ( Figure S1C–S1E ) . When expressed in cells harbouring 256 repeats of the lac operator sequence ( lacO ) integrated at the ura4+ locus , each fusion protein formed a single nuclear focus . No foci were detected in cells devoid of lacO arrays ( Figure 4B ) . We used Chk1 phosphorylation as a readout for DNA damage checkpoint activation ( Figure 4C ) . Following thiamine removal ( induction takes between 12 and 16 hours [38] ) , Chk1 became phosphorylated in lacO containing cells , but not in lacO-negative control cells , when either Rad3ATR , Rad4TopBP1 or Rad9 LacI fusion proteins were expressed . Similar results were obtained when each pair-wise combinations of two fusion proteins were expressed ( Figure 5B ) . In S . pombe , DNA damage checkpoint activation results in cell cycle arrest and cell elongation . Elongated cells were observed upon expression of single fusion proteins ( data not shown ) , confirming checkpoint activation . From these data we conclude that , in S . pombe , as in mammals [39] tethering of any of these single checkpoint proteins to a lacO array is sufficient to activate the DNA damage checkpoint and that , in contrast to the analogous experiments reported for S . cerevisiae , forced co-localisation of two checkpoint proteins is not required [37] . To establish if the Rad4TopBP1 AAD is involved in this damage-independent mode of checkpoint activation , we tested if Rad3ATR tethering could result in Chk1 phosphorylation in a rad4-Y599R mutant background ( Figure 5A ) . While Rad3-GFP/LN expression resulted in induced Chk1 phosphorylation in rad4+ cells , Rad3-GFP/LN expression did not increase Chk1 phosphorylation in rad4-Y599R cells , demonstrating a role for the Rad4TopBP1 AAD . Next we established if expression and tethering of the AAD-defective Rad4-Y599R protein to lacO arrays was able to activate the checkpoint ( Figure 5B ) . No induction of Chk1 phosphorylation was observed . Furthermore , while co-expression and tethering Rad3ATR and Rad9 , or of Rad3ATR and Rad4TopBP1 resulted in checkpoint activation ( Figure 5B ) , we observed that co-expression of Rad3ATR with Rad4TopBP1-Y599R mutant protein did not result in Chk1 phosphorylation . This data suggests that the AAD-defective mutant protein can act as a dominant negative , at least in this specific situation , preventing the endogenous wild-type Rad4TopBP1 from functioning with the tethered Rad3ATR to activate the checkpoint . It also supports the idea that the Rad4TopBP1 AAD domain is required for the activation of Rad3ATR and not simply recruiting it . While Rad3ATR kinase activity is essential for Chk1 phosphorylation in response to DNA damage [40] , it also depends on the 9-1-1 clamp , the Rad17 clamp loader and the Crb253BP1 mediator . To characterise the dependencies for lacO-dependent checkpoint activation we examined which checkpoint genes were required for Chk1 phosphorylation during Rad3ATR tethering ( Figure 5C ) . Rad3ATR-GFP/LN was expressed in lacO-positive strains deleted for rad1 , rad9 ( encoding 9-1-1 components ) , rad17 ( clamp loader ) and crb2 . Each was required for Chk1 phosphorylation . Thus , Rad3ATR tethering is not sufficient for checkpoint activation: the clamp loader , the 9-1-1 clamp complex and the Crb2 mediator are all required and this artificial checkpoint activation system does not bypass the usual requirements . However , Brc1 , the proposed MDC1/PTIP ortholog is not required for Chk1 phosphorylation in this system ( Figure S1F ) In both S . pombe and S . cerevisiae recruitment of the 53BP1 ortholog ( Crb2 and Rad9 respectively ) to chromatin in response to IR requires prior phosphorylation of histone H2A [32] , [41] , [42] . In addition to H2A phosphorylation , recruitment also requires the largely constitutive methylation of a further histone residue , H3K79 in S . cerevisiae or H4K20 in S . pombe . These modifications are effected by distinct methylransferases in the two yeasts: Dot1 methylates H3K79 in S . cerevisiae [43] while Set9 methylates H3K20 in S . pombe [44] , [45] . In S . pombe it has been demonstrated that the C-terminal BRCT domains of Crb253BP1 binds directly to γH2A [42] while the Tudor domain binds directly to di-methylated H3K20 [46] . Both interactions are required for Crb253BP1 chromatin association and show an epistatic relationship [45] . Using our Chk1 phosphorylation assay in response to Rad3ATR tethering , we tested two strains harbouring charge reversal mutations of residues within the phospho-acceptor site of the C-terminal Crb253BP1 BRCT domains , crb2-K617E and crb2-K619E ( Figure 5C ) that disrupt the interaction with γH2A [42] . Chk1 phosphorylation was reduced in both mutants . We next established if checkpoint activation by Rad3ATR tethering was affected in cells containing mutants in the two H2A genes that replace the phosphorylated residue with alanine , hta1-S129A hta2-S128A [32] . Chk1 phosphorylation was not observed in this background ( Figure 5D ) , indicating that the Rad3ATR tethering-dependent and Rad4TopBP1 AAD-dependent checkpoint activation acts in the context of chromatin modification . Upon activation of the DNA damage checkpoint in S . pombe , Rad3ATR phosphorylates the Rad9 C-terminus on T412 and this is required to recruit Rad4TopBP1 [20] . Recruitment of Rad4TopBP1 allows subsequent recruitment of Crb253BP1 and consequent Chk1 activation [29] . A similar requirement for TopBP1 recruitment via Rad9 C-terminal phosphorylation is also evident in S . cerevisiae and mammalian cells [19] , [47] , [48] . As expected , expression of Rad3ATR-LacI in cells harbouring a rad9-T412A mutation did not result in Chk1 phosphorylation ( Figure 5E ) implying that lacO-recruited Rad3ATR must phosphorylate endogenous Rad9 to promote Rad4TopBP1 recruitment to activate the checkpoint . We reasoned that the requirement for Rad9-T412 phosphorylation during activation by Rad3ATR tethering may solely be to bring the Rad4TopBP1 AAD into proximity of Rad3ATR . In this case , we should be able to bypass the requirement for Rad9-T412 phosphorylation specifically for checkpoint activation by Rad3ATR tethering by recruiting both Rad3ATR and Rad4TopBP1 at the same time . Indeed , Chk1 phosphorylation was restored when we co-expressed Rad3ATR-LacI and Rad4TopBP1-LacI in a rad9-T412A mutant background ( Figure 5E ) . Since checkpoint activation by co-expression of Rad3ATR-LacI and Rad4TopBP1-LacI remains lacO dependent ( Figure S1G ) , these data suggest that Rad4TopBP1 AAD can activate the Rad3ATR-dependent checkpoint cascade in the absence of the recruitment activity of the Rad9 C-terminal tail . We have shown that the Rad4TopBP1 AAD functions to protect cells from insult during S phase , but is not required for G2 checkpoint activation after IR . Further , we demonstrated that when rad4-Y599R ( AAD-defective ) mutant cells were synchronised in S phase , phosphorylation of both Chk1 and H2A in response to IR treatment was reduced when compared to rad4+ cells . The increase in Cdc2-Cdc13ClyclinB ( CDK ) activity as cells progress from S phase into G2 [49] is known to establish conditions conducive to HR by regulating factors required for DNA resection [36] , [50] , [51] . A consequence of this is that , in response to IR treatment but not in response to UV treatment , ssDNA RPA is predicted to be more prevalent in G2 cells when compared to G1/S phase cells . We thus predicted that reducing resection rates associated with IR treatment in G2 cells would create a dependency for full Chk1 phosphorylation on the Rad4TopBP1 AAD and thus that Chk1 phosphorylation would be reduced in rad4-Y599R strains compared to rad4+ strains in response to an equal dose of IR . To test this prediction we examined the induction of Chk1 phosphorylation in response to 100 Gy IR in exo1Δ rad4+ and exo1Δ rad4-Y599R cells ( Figure 6 ) . First we established that , when exo1 was deleted , Rad11RPA foci were reduced in number , consistent with the expectationt hat resection is decreased in this background ( Figure 6A ) . In contrast to rad4+ cells , where exo1 deletion did not reduce Chk1 phosphorylation levels , Chk1 phosphorylation was reduced to approximately 50% when exo1 was deleted in AAD-defective cells . Previous work in both budding and fission yeasts has indicated that , in the absence of resection , Chk1 phosphorylation can occur through an alternative double strand break end-dependent Tel1ATM pathway , as opposed to the canonical resection and ssDNA/RPA-dependent Rad3ATR pathway [52] . Such a response could potentially mask some aspect of defects seen in the exo1Δ background . Thus , we first confirmed that loss of tel1 alone does not influence Chk1 phosphorylation in our assay ( Figure S1H ) and then concomitantly deleted Tel1ATM in both rad4+ and rad4-Y599R strains ( Figure 6B , 6C ) . In the tel1Δ exo1Δ background , Chk1 phosphorylation was decreased by approximately 50% for both rad4+ and rad4-Y599R when compared to the exo1Δ alone background . These data are consistent with a general increase in Tel1-dependent checkpoint signalling when resection is reduced by exo1 deletion . In the background of the rad4-Y599R mutation , this is superimposed on a decrease in Rad3ATR-dependent signalling caused by the reduced resection . A second ATR activation domain has recently been identified in the S . cerevisiae Ddc1Rad9 C-terminal tail [26] . Mutations in this domain define a function in Mec1ATR activation during G1 , complementary to the function of the Dpb11TopBP1 AAD in promoting robust checkpoint activation in G2 in this organism [24] . Sequence alignments show that the two key Ddc1Rad9 AAD aromatic residues are conserved in S . pombe as Y271 ( equating to W352Sc within the PCNA-like domain ) and W348 ( equating to W544Sc in the intrinsically disordered C-terminal tail ) ( [26] and Figure 7A , 7B ) . We thus created an rad9-AAD mutant by mutating both aromatic residues to alanine . Analysis of the resulting rad9-AAD strain demonstrates no clear sensitivity to DNA damaging agents that create problems during S phase , including CPT , MMS and UV ( Figure 7C ) or in G2 to IR . However , some increased sensitivity is evident to CPT and MMS when the Rad4TopBP1 AAD mutant is present in the same strain . We next assayed the ability of rad9-AAD mutants to activate Chk1 in response to either IR or UV treatment . Consistent with the lack of sensitivity , the level of Chk1 phosphorylation after IR was not reduced ( Figure 7D ) , either in rad9-AAD mutant alone or in the rad9-AAD rad4-Y599R double mutant when compared to the rad4-Y599R single . In response to UV treatment , there was again no decrease observed for the single rad9-AAD mutant , but a further and reproducible decrease was seen for the rad9-AAD rad4-Y559R double mutant when compared to rad4-Y599R alone ( Figure 7E ) . Thus , the putative Rad9-AAD domain in S . pombe plays , at most , only a minor role in activating Rad3ATR in response to DNA damage and this is only revealed in the absence of the Rad4TopBP1 AAD .
Understanding the mechanism of ATR activation is an important facet of gaining insight into how cells respond to unwanted DNA structures , itself a key aspect in maintaining genomic integrity . TopBP1 was initially implicated in the ATR-dependent checkpoint in fission yeast and later this was extended to higher eukaryotes [53] . TopBP1 is a multi-BRCT-domain containing protein that acts to scaffold proteins during both the initiation of DNA replication and in response to DNA damage , a function dependent of the phospho-binding ability of the BRCT-domain pairs within TopBP1 [25] . In addition to scaffolding phospho-proteins , TopBP1 was shown to be able to directly activate ATR in Xenopus and human cells through a small domain of TopBP1 which is not part of any BRCT pair [17] . The ATR activating domain is sufficient , both in vitro and in vivo , to activate ATR - although it is not always necessary for ATR activation and the pathway in which this TopBP1 AAD domain functions is yet to be fully understood . It has recently been shown in S . cerevisiae that the AAD of the TopBP1 homolog ( Dpb11 ) is also able to activate the ATR homolg ( Mec1 ) . However , the Dpb11 AAD plays a relatively minor role in checkpoint activation which is specific to G2 phase [24] . In S . cerevisiae , a second ATR activation domain within the C-terminal tail of the 9-1-1 subunit , Ddc1Rad9 acts to help activate ATR in G1 and G2 and it is only when the function of this AAD is ablated a role for the Dpb11 AAD becomes apparent [26] . However , loss of both domains does not prevent checkpoint activation entirely , suggesting other AADs or modes of activation . Conversely , in the Xenopus system , ATR activation via the TopBP1 AAD is evident in S phase . Here we show that , in S . pombe , the activation of the ATR homolog ( Rad3 ) by a Rad4TopBP1 AAD is conserved . We demonstrate that the Rad4TopBP1 AAD makes a contribution to checkpoint activation and that this is specific to G1/S phase and is not evident in G2 . Note that log phase S . pombe spend little , if any , time in G1 and thus , while we can arrest cells before the onset of replication with cell cycle mutants , we cannot make a clear physiological distinction between G1 and S phase . We go on to demonstrate that , when DNA resection was limited in G2 by ablation of the Exo1 nuclease , checkpoint activation in response to DNA damage during G2 becomes partially dependent on the Rad4TopBP1 AAD , mimicking what we observed in G1/S cells . This leads us to propose that there is a threshold of ssDNA required for activation of the DNA damage checkpoint and that the Rad4TopBP1 AAD serves to amplify checkpoint signals when ssDNA is limiting . We next used a genetic system to separate Rad3ATR activation from the production of DNA damage and therefore ssDNA , thus allowing us to assess the pathway of Rad3ATR activation dependent on the Rad4TopBP1 AAD . In this system , specific checkpoint proteins are recruited to a defined chromatin locus through dsDNA:protein binding [37] , [39] . Interestingly , recruitment of any one of the three checkpoint proteins ( Rad3ATR , Rad4TopBP1 and Rad9 ) tested was sufficient to generate a checkpoint response and these responses followed the expected dependencies . This suggests that the recruitment of multiple copies of a single checkpoint protein results in the formation of active checkpoint complexes that utilise the endogenous proteins . Using this system , we observed that the ability of the Rad4TopBP1 AAD to activate Rad3ATR is fully dependent on phosphorylation of H2A ( γH2A ) and requires the ability of Crb2 to bind γH2A . This leads us to conclude that , in the absence of ssDNA , the ATR activation domain of Rad4TopBP1 is particularly important for Rad3ATR activation and acts in a chromatin:protein interaction dependent manner . Taking these data together with the requirement of the Rad4TopBP1 AAD to amplify checkpoint signals in either G1/S or G2 when resection was limited , we propose that Rad4TopBP1 acts to amplify the checkpoint in a chromatin-dependent manner when single-stranded DNA levels are limiting . We can therefore hypothesise that there is a threshold level for the amount of active Rad3ATR required for a full checkpoint response . When ss-DNA is limited , such as in S-phase , the chromatin-dependent Rad4TopBP1 AAD-dependent pathway for Rad3ATR activation becomes important to amplify the levels of activated Rad3ATR to obtain a full checkpoint response ( Figure 8 ) . In addition to analysing the Rad4TopBP1 AAD , we also created a mutant predicted to disable the Rad9 equivalent of the S . cerevisiae Ddc1Rad9 AAD and analysed the effect of this mutant in checkpoint activation . Unlike in S . cerevisiae , we observed no significant effect on DNA damage-induced checkpoint activation either in G1/S phase or G2 . Although when combined with a Rad4TopBP1 AAD mutant , an additive effect to S-phase but not G2 DNA damage can be seen . This suggests that the Rad9 AAD acts in a separate but redundant pathway for Rad3ATR activation in G1/S with the Rad4TopBP1 AAD . It appears that , during evolution , the mechanism of activating the ATR pathway has diverged significantly with the roles of different ATR activating domains being of more or less importance in different organisms . It will be interesting to establish if the ATR activating domain of TopBP1 in metazoan systems is particularly important in the context of low levels of ssDNA and whether its function is dependent on γH2AX , especially as a 53BP1 ( Crb2 ) and TopBP1 pathway for checkpoint activation in G1 has been previously reported in the mammalian system [54] . The differences in the dependencies of the specific ATR activators in different cell cycle phases between S . pombe and S . cerevisiae is not surprising as the checkpoint mechanism between these organisms has diverged . For example , in S . cerevisiae , the S phase checkpoint is activated independently of the 9-1-1 complex , whereas in S . pombe and mammalian cells ATR activation appears to be largely - if not entirely - dependent on 9-1-1 loading . Such distinctions are likely a result of evolutionary adaptation to the different cell cycle profiles of the two yeasts and it is interesting to note that significant evolutionary plasticity surrounds the interface between TopBP1 and the checkpoint apparatus . These distinctions will have to be considered when extrapolating mechanistic data from yeast to human systems . None the less , we believe that our findings shed light on the role of TopBP1 AAD in DNA damage responses and offer useful insights into metazoan mechanisms of DNA damage signalling .
Standard S . pombe protocols were carried out as previously described [55] . rad4 and rad9 mutant strains were created using PCR site directed mutagenesis and integrated at their endogenous locus using Cre recombinase-mediated cassette exchange [56] In brief , this system uses a “base strain” which is engineered so that the gene of interest is either replaced with the ura4 marker ( i . e . rad9 ) , or in the case of essential genes ( i . e . rad4 ) has the marker integrated immediately after the stop codon . In both cases the gene/marker and loci's promoter region are flanked by loxP and loxM sites . These two variant lox sites are incompatible with each other . The marker ( and , for essential genes , the actual gene also ) is then replaced by transforming in either the wild type ( as a control: rad+ ) or the various mutated copies on a plasmid . These are flanked by the equivalent loxP and loxM sites and the plasmid expresses Cre recombinase , which results in loxP:loxP and loxM:loxM recombination . For cdc10-M17 synchronisation cells were grown to log phase at the permissive temperature ( 25°C ) and shifted to the restrictive temperature of 36°C for 3 . 5 hours . Cells were then either irradiated with the indicated dose of gamma irradiation at 36°C and released at 25°C , or directly released at 25°C and irradiated at the given time points after release . cdc25-22 block and release [57] and lactose gradient synchronisation [12] were performed as described previously . For FACS analysis cells were resuspended in 50 mM tri-sodium citrate , 1 mg/ml final concentration RNAseA [Sigma] , stained with 5 µg/ml Propidium iodide [Sigma] and analysed on FacsCalibur [Becton Dickinson] . For live cell imaging concentrated culture was mounted onto a 2 . 5% agar patch in standard YE medim [Microworks] and imaged on a Deltavsion Microscope . Septation index was counted as previously described [12] . lacO::NAT chk1-HA strains were created by inserting the 10 Kb lacO repeats into the PUC19 plasmid containing the NAT marker and homology to ura4 . This was integrated into the genomic ura4 locus . The appropriate strains were transformed [58] with pRep41-GFP-LacI-NLS ( GFP/LN ) into which either rad3 or rad9 had been cloned in frame for N-terminal tagging or rad4 cloned in frame for C-terminal tagging . Transformants were grown and expression of the fusion protein induced by the removal of thiamine . All lacO repeats were checked by Southern hybridisation . Protein extracts for western were prepared by TCA ( trichloro-acetic acid ) extraction from 1×108 cells and resuspended in SDS sample buffer [59] . Crude extracts for affinity analysis were prepared by mechanical disruption in liquid nitrogen . Antibodies used: α-HA [Santa cruz] 1∶2500 , α-Myc [Santa cruz] 1∶2000 , α-GFP [Roche] 1∶2500 , α-H2ApS129 [Abcam] 1∶2500 or 1∶1000 , α-Tubulin [Sigma] 1∶5000 , α-Cdc2 Sc-53 [Santa cruz] 1∶2500 . α-Cdc13 [Jacky Hayles] 1∶500 . α-Cds1 1∶5000 [35] . The secondary antibodies used were Hrp rabbit α mouse [Dako] 1∶2500 or Hrp swine α Rabbit [Dako]1∶2500 . Chk1-HA phosphorylation was quantified as a percentage of total signal minus back ground on a ImageQuant LAS 4000 [GE Healthcare] . Cds1 kinase assay was carried out as described [35] . | DNA structure–dependent checkpoint activation and the amplification of checkpoint signals are carefully modulated to allow the checkpoint kinases to delay mitosis and regulate DNA metabolism . While much work has gone into understanding how this checkpoint functions , the mechanism by which the checkpoint signal is amplified is less clear . We have characterised a conserved domain in the Schizosaccharomyces pombe TopBP1 homolog , Rad4TopBP1 ( also known as Cut5 ) that is capable of activating the ATR homolog Rad3ATR . We demonstrate that this domain is not required for initial checkpoint activation , but functions to amplify the checkpoint signal , likely when the presence of single-stranded DNA is limiting . Our data suggest that the function of the Rad4TopBP1 ATR-Activation Domain ( AAD ) is mediated by interactions between checkpoint proteins and phosphorylated histone H2A , which is itself promoted by Rad3ATR . We propose that the resulting amplification of the checkpoint signal is particularly important in G1-S phase , when resection is limited . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biology"
] | 2012 | The Rad4TopBP1 ATR-Activation Domain Functions in G1/S Phase in a Chromatin-Dependent Manner |
Dimorphism or morphogenic conversion is exploited by several pathogenic fungi and is required for tissue invasion and/or survival in the host . We have identified a homolog of a master regulator of this morphological switch in the plant pathogenic fungus Fusarium oxysporum f . sp . lycopersici . This non-dimorphic fungus causes vascular wilt disease in tomato by penetrating the plant roots and colonizing the vascular tissue . Gene knock-out and complementation studies established that the gene for this putative regulator , SGE1 ( SIX Gene Expression 1 ) , is essential for pathogenicity . In addition , microscopic analysis using fluorescent proteins revealed that Sge1 is localized in the nucleus , is not required for root colonization and penetration , but is required for parasitic growth . Furthermore , Sge1 is required for expression of genes encoding effectors that are secreted during infection . We propose that Sge1 is required in F . oxysporum and other non-dimorphic ( plant ) pathogenic fungi for parasitic growth .
The fungus Fusarium oxysporum is found in both agricultural and non-cultivated soils throughout the world . The species consists of non-pathogenic and pathogenic isolates , both known as efficient colonizers of the root rhizosphere . The pathogenic isolates , grouped into formae specialis depending on their host range [1] , [2] , cause wilt or rot disease in important agricultural and ornamental plant species , such as tomato , banana , cotton and tulip bulbs , thereby causing serious problems in commercial crop production [3] , [4] . Recently , F . oxysporum has also been reported as an emerging human pathogen , causing opportunistic mycoses [5]–[7] . In the absence of plant roots F . oxysporum survives in the soil either as dormant propagules ( chlamydospores ) or by growing saprophytically on organic matter [1] , [8] . When growing on roots of a suitable host F . oxysporum appears to switch from a saprophyte into a pathogen . As a pathogen F . oxysporum needs to overcome host defence responses and sustain growth within the host in order to establish disease . To do so , F . oxysporum likely undergoes reprogramming of gene expression . In the last decade , genes have been identified that do not seem to be required for saprophytic growth , but are involved in or required for pathogenicity and/or are specifically expressed during in planta growth . Examples are SIX1 , encoding a small secreted protein , and FOW2 and FTF1 , both encoding Zn ( II ) 2Cys6-type transcriptional regulators [9]–[12] . In an insertional mutagenesis screen aimed at identification of pathogenicity factors of Fusarium oxysporum f . sp . lycopercisi ( Fol ) , a gene now called SGE1 ( SIX Gene Expression 1 ) was identified that shows homology to the transcriptional regulators Candida albicans WOR1 and Histoplasma capsulatum RYP1 [9] . These transcription factors have been identified as major regulators of morphological switching in these human pathogens: from a filamentous to a yeast form in H . capsulatum and from a white to opaque cell type in C . albicans [10]–[13] . In both fungi , these morphological transitions are correlated with the ability to cause disease . Targeted deletion of RYP1 in H . capsulatum or WOR1 in C . albicans locks the fungus in its filamentous form or white cell type , respectively . In this work we characterize SGE1 and show that it shares many characteristics with WOR1 and RYP1 . In addition , we show that expression of effector genes is lost in the SGE1 deletion mutant . We conclude that Sge1 plays a major role during parasitic growth , defined as extensive in planta growth leading to wilt symptoms , in F . oxysporum f . sp . lycopersici .
In an insertional mutagenesis screen aimed at identifying genes involved in pathogenicity a non-pathogenic mutant ( 5G2 ) and one severely reduced in pathogenicity ( 101E1 ) were identified that both carried a single T-DNA insertion into the ORF of FOXG_10510 [9] , hereafter called SGE1 ( SIX Gene Expression 1 ) . The SGE1 ORF contains no introns and encodes a protein of 330 amino acids ( http://www . broad . mit . edu/annotation/genome/fusarium_group/MultiHome . html ) . Sequence analysis revealed that the N-terminus ( amino acids 1–120 ) contains a TOS9 ( COG5037 ) and a Gti1_Pac2 family domain ( Pfam09729 ) and is conserved in the fungal kingdom; all fungi of which the genome sequence was examined , including ascomycetes , basidiomycetes and zygomycetes , contain related genes that divide in two groups based on sequence similarity of the predicted proteins . Most ascomycetes have one member in each group , except Neurospora crassa which lacks a member of the SGE1 group ( Figure 1A ) . The basidiomycete Coprinus cinereus and the zygomycete Rhizopus oryzae contain more than two members , still with at least one member in each group ( data not shown ) . The branching order within the two groups does not always follow species phylogeny , making orthology questionable . Examples are the placement of FoSge1 and FGSG_12164 basal to the Magnaporthe grisea homolog ( MGG_00850 ) , the placement of CAWG_04607 of C . albicans basal to Pac2 of Schizosaccharomyces pombe ( and closer to basidiomycete homologs ) and of NCU06864 of N . crassa basal to homologs of other filamentous fungi ( pezizomycotina ) ( Figure 1A ) . Sge1 is in the same group as Histoplasma capsulatum Ryp1 and Candida albicans Wor1 , both identified as regulators for morphological switching [10]–[13] , and Schizosaccharomyces pombe Gti1 , which plays a role in gluconate uptake upon glucose starvation [14] . A potential protein kinase A phosphorylation site ( KRWTDS/G ) is conserved between these proteins ( Figure 1B ) . In addition , a nuclear localization motif is present in Sge1 ( +93 to +100 ) that is shared with Ryp1 ( Figure 1B ) . The F . oxysporum protein related to Sge1 , encoded by FOXG_12728 , shows high similarity to S . pombe Pac2 , a protein controlling the onset of sexual development [15] . Interestingly , in the same insertional mutagenesis screen mentioned above , a Fol mutant with reduced pathogenicity ( 30C11 ) was identified in which one of two T-DNA insertions resides in the FOXG_12728 ORF [9] , hereafter called FoPAC2 . To assess the involvement of SGE1 and FoPAC2 in pathogenicity , gene knock-out mutants were generated by homologous recombination . Four independent SGE1 and eight independent FoPAC2 knock-out mutants were obtained , with the deletions confirmed by PCR and Southern analysis ( Figure S1 , S3 and S4 ) . The SGE1 deletion mutants were non-pathogenic on tomato in a root dip bioassay and corroborated the severely reduced to non-pathogenic phenotype of the insertional mutagenesis mutants ( Figure 2A ) . Re-introduction of the SGE1 gene in locus by homologous recombination in a knock-out mutant ( Figure S2 and S3 ) restored pathogenicity ( Figure 2B ) , confirming that the loss of pathogenicity was due to loss of SGE1 . Deletion of FoPAC2 only had a minor effect on pathogenicity . All knock-out mutants were significantly different from the wild type in disease causing ability , but seven out of the eight mutants were also significantly different from the original insertion mutant ( 30C11 ) ( Figure 3 ) . This indicates that FoPAC2 plays at most a minor role during infection and that most probably additional defects in the 30C11 mutant added to the reduced pathogenicity phenotype . Since the loss of pathogenicity was complete upon deletion of SGE1 , we focussed on this gene for further analysis . Previously , we reported that vegetative growth of the insertional mutants 5G2 and 101E1 are indistinguishable from that of the wild type on various carbon sources [9] . To more fully analyze potential metabolic defects of the sge1 mutant , we made use of BIOLOG FF MicroPlates , in which each well contains a different carbon source [16] . Also in this assay no reproducible differences were observed between growth of the wild type and the SGE1 deletion mutant on 95 different carbon sources ( Figure S5 ) . Microconidia and macroconidia generated in minimal or CMC liquid medium were phenotypically indistinguishable from wild type ( Figure S6 ) . However , the SGE1 deletion mutants produced about 6-fold less microconidia compared to the wild type in both media , and this phenotype was only partially restored in the SGE1 complementation mutants ( Figure 4A ) . The conidial germination rates were comparable to wild type ( Figure 4B ) , indicating that , although less microconidia are formed , they are fully viable . These observations indicate that SGE1 is quantitatively involved in conidiogenesis , but is not required for conidial fitness , overall ( colony ) morphology , vegetative growth or carbon source utilization . C . albicans WOR1 and H . capsulatum RYP1 are 45-fold and 4-fold upregulated upon transition from white to opaque cells in C . albicans and from filamentous growth to yeast cells in H . capsulatum , respectively [11] , [17] . To determine the relative expression levels of SGE1 during saprophytic and parasitic growth quantitative PCR was performed . Expression levels of SGE1 were determined both in axenic culture and during tomato root infection at different time points after inoculation and compared to the level of the constitutively expressed elongation factor 1 alpha gene ( EF-1α ) . We found that SGE1 expression is upregulated 2- to 5-fold during infection with maximal expression eight days after inoculation ( Figure 5 ) . To determine at which stage during infection the SGE1 deletion mutant is halted , tomato root colonization by the mutant was visualized using fluorescent binocular and confocal laser scanning microscopy . Tomato seedlings were infected with a GFP expressing wild type strain or SGE1 deletion strain and colonization was followed over time . After two days , patches of colonization on the roots were observed for both strains ( Figure S7 ) , indicating that the SGE1 deletion mutant is not impaired in root surface colonization . This observation was confirmed by confocal laser scanning microscopy . For both strains , germinated spores were observed on the root surface 24 hours after infection ( Figure 6A and 6B ) and mycelial mass increased in time leading to patches of colonization as described above ( Figure 6C and 6D ) . No difference in colonization was observed between the wild type and the sge1 mutant in the first 2 days after inoculation . However , after three days , plant cells filled with spores and mycelium were observed in roots infected with the wild type strain ( Figure 7A ) , indicating that penetration of the root surface had occurred ( Figure 7C ) . A mixture of germinated spores and mycelium was also observed on plant roots infected with the sge1 mutant , however , growth was dispersed over the entire root ( Figure 7B ) and was not confined to single plant cells as observed for the wild type . Occasionally , the sge1 mutant was observed within the root cortex ( Figure 7D ) , indicating that the sge1 mutant is capable of penetrating the surface . This is in line with our observation that the sge1 mutant is able to penetrate a cellophane sheet ( Figure S8 ) , an assay used to assess the ability to form penetration hyphae [18] . Four days after infection fungal growth within the roots was observed . In roots infected with the wild type , growth within the xylem vessel was observed once ( Figure 7E ) . Also in one case , extensive growth of the sge1 mutant was observed within the root ( Figure 7F ) ; however , this growth was mainly intercellular . In both cases the fungal growth extended from a point where the root was broken . Since these cases were too rare for a proper quantification , we can only conclude that both the wild type and the sge1 mutant are capable of in planta growth . To determine whether the sge1 mutant is able to colonize xylem vessels as extensively as the wild type , tomato seedlings were inoculated with the wild type , SGE1 knock-out or the SGE1 complementation strain and potted according to the bioassay procedure ( i . e . with damaged roots to allow relatively easy entry into vessels ) . One week after inoculation the hypocotyl was cut in slices of several millimeters and put on rich ( PDA ) medium . After two days F . oxysporum outgrowth was observed from hypocotyl pieces previously inoculated with the wild type and the SGE1 complementation strain , but not from the hypocotyl pieces previously inoculated with the SGE1 knock-out mutant ( Figure S9 ) . Based on the above described observations , it can be concluded that Sge1 is not required for early pathogenesis-related functions , such as root colonization and penetration , but appears to be required for extensive growth within plant cells and the xylem . The homologs of Sge1 in C . albicans and H . capsulatum are nuclear proteins [10]–[13] . To determine whether this is also the case for Sge1 its subcellular localization was investigated . For this purpose , Sge1 was fused C-terminally to the fluorescent proteins CFP or RFP . Constructs expressing these fusion proteins were introduced in the SGE1 deletion mutant by homologous recombination at the sge1 locus ( verified by Southern analysis ( Figure S3 ) ) . The functionality of the fusion proteins was determined in a bioassay . SGE1::CFP restored pathogenicity to almost wild type levels ( Figure S10 ) . Disease symptoms were also observed upon infection with strains expressing SGE1::RFP , albeit severely reduced compared to the wild type infection ( Figure S10 ) . Subcellular localization for both fusion proteins was similar in that they are localized in the nucleus in both spores and hyphae ( Figure 8 ) . Nuclear localization was verified by introduction of a construct encoding histone H2B::GFP [19] into the SGE1::RFP strain . Both fusion proteins localized to the same compartment ( Figure 8 ) . These observations support the potential role of Sge1 as a transcriptional regulator in F . oxysporum . As described above , the Sge1 protein contains a potential Pka phosphorylation site and this site is conserved in all Sge1 homologs ( Figure 1B ) . Replacement of the conserved threonine residue by an alanine impaired S . pombe Gti1 function [14] . In an attempt to identify Sge1 interacting partners using a yeast two-hybrid screen , the SGE1 gene , fused to the portion of GAL4 encoding the Gal4 DNA binding domain , was introduced in Saccharamyces cerevisiae . To our surprise , after numerous transformation attempts only one colony was obtained . After re-isolation of the plasmid from this colony followed by sequencing it turned out that a point mutation in the SGE1 sequence ( C196A ) had occurred resulting in an amino acid change from an arginine into a serine at position 66 , altering the Pka phosphorylation site ( Figure 1 ) . In contrast to wild type SGE1 , this mutant form of SGE1 could be easily re-transformed to yeast . We concluded that SGE1 is normally toxic in S . cerevisiae and that the R>S mutation leads to tolerance in this yeast . To determine the effect of this mutation on the activity of Sge1 in Fol , a construct expressing Sge1R66S was introduced in the SGE1 deletion mutant . Correct gene replacement was again confirmed by PCR ( Figure S11 ) . All strains encoding the Sge1R66S protein were non-pathogenic ( Figure 9 ) , suggesting that an intact Pka phosphorylation site is required for Sge1 to function properly . The transcriptional regulators Ryp1 and Wor1 regulate expression of phase specific genes [11] , [20] . Since Sge1 shares many features with Ryp1 and Wor1 , we hypothesized that expression of F . oxysporum genes specifically expressed during infection could be altered . Examples of such genes are those encoding effectors , small in planta secreted proteins , called Six ( Secreted in xylem ) proteins in Fol . Recently , it was shown that F . oxysporum secretes numerous Six proteins during infection [21] ( Houterman and Rep , unpublished ) . SIX1 , encoding Avr3 , is strongly induced upon penetration of the root cortex and plays a role during pathogenicity [22] , [23] . In addition , Six3 ( Avr2 ) and Six4 ( Avr1 ) have been shown to play a role in virulence as well as resistance protein recognition [24] , [25] . Since the SGE1 deletion mutant does not show extensive in planta growth , precluding assessment of SIX gene expression during root infection , we incubated the mutant together with tomato cells in culture , a situation known to induce SIX1 expression [22] . MSK8 tomato cells were inoculated with a conidial suspension of the wild type or the SGE1 deletion mutant . After 24 h the cultures were harvested and the presence of SIX gene transcripts was determined by RT-PCR and compared to transcript levels in axenic cultures . As expected , SIX1 gene expression was very low when the wild type strain was grown in axenic culture and expression was strongly upregulated upon incubation with MSK8 cells . A similar expression pattern was observed for SIX2 ( Figure 10 ) . In contrast , the SIX3 and SIX5 genes were expressed in the wild type strain in axenic culture as well as in the presence of MSK8 cells ( Figure 10 ) . Interestingly , in the sge1 mutant expression of all four SIX genes was lost both in axenic culture and in the presence of MSK8 cells . Expression was restored in the complemented strain ( Figure 10 ) . These results show that expression of at least four SIX genes tested is dependent on the presence of SGE1 . Biocontrol activity has been reported to be dependent on colonization of superficial cell layers , a capacity shared between pathogenic and protective strains and has been reported to be independent of competition for putative penetration sites or nutrients [26] . Fusarium oxysporum f . sp . lycopersici exerts biocontrol activity on flax when added in a 100-fold excess relative to a F . oxysporum strain pathogenic towards flax . To determine whether the sge1 mutant is still able to protect flax against wilt disease like its parental strain , flax cv . viking was treated with either a pathogenic isolate of F . oxysporum f . sp . lini ( Foln3 ) alone or in a 1∶100 ratio with either the sge1 mutant or the wild type strain 4287 . The first wilt symptoms were observed 22 days after inoculation in the treatment with Foln3 ( Figure 11 ) . In the Foln3/sge1 mutant and Foln3/wild type combination treatments the first disease symptoms were delayed by 3 days . Disease symptoms were always reduced in the Foln3/sge1 mutant and Foln3/wild type treatments compared to the Foln3 treatment: 48 days post inoculation disease symptoms were reduced by 27 and 33% , by the sge1 mutant and the wild type , respectively ( Figure 11 ) . The ANOVA performed on AUDPCs indicated that this difference was significant at the probability of 91% . We conclude that the sge1 mutant is able to protect flax against Fusarium wilt as well as the wild type strain , suggesting that it can colonize roots efficiently .
Fungi have various ways to adapt their morphology to the environment , one example being dimorphism . Dimorphism is a strategy frequently employed by fungal pathogens where this developmental transition is correlated with the ability to cause disease ( reviewed by [27] , [28] ) . In this study , we have characterized Sge1 , the F . oxysporum homolog of the master regulator of morphological switching Wor1 and Ryp1 . Although F . oxysporum does not display an immediately obvious morphological switch like C . albicans or H . capsulatum , Sge1 was found to be required for colonization of the xylem system and disease development . Sge1 , together with its homolog FoPac2 , described in this work , belong to a class of conserved fungal proteins . Both proteins have apparent orthologs across the fungal kingdom and the N-terminal domain of these proteins is always more conserved than the C-terminus , which is generally rich in glutamine residues . Despite their similarity , the SGE1 and FoPAC2 deletion mutants generated in this study display different phenotypes , indicating that the functions of these proteins are not redundant . The FoPac2 and Sge1 homologs in S . pombe , Pac2 and Gti1 , respectively , also have a different function although both proteins are involved in regulation of transition processes . Pac2 is a negative regulator of sexual development , which is induced under nitrogen starvation [15] and Gti1 is a positive regulator of metabolic reprogramming by inducing gluconate uptake under glucose starvation [14] . The best characterized members of the class to which Sge1 belongs are Wor1 in C . albicans and Ryp1 in H . capsulatum [10]–[13] . Ryp1 and Sge1 share a conserved and apparently functional nuclear localization signal . In addition , all three proteins contain a putative protein kinase A ( Pka ) phosphorylation site . Another common feature of these proteins is increased expression of their genes upon transition . WOR1 and RYP1 are both under tight transcriptional regulation in that hardly any expression of these genes can be detected in the white and the filamentous growth phase , respectively . Their expression is upregulated 45- and 4-fold during and after the transition to the opaque or the yeast growth phase [10] , [11] , [13] , [17] . Although SGE1 expression can easily be detected in the saprophytic growth phase , it was found to be upregulated 2- to 5-fold during in planta growth . WOR1 is preceded by an exceptionally large intergenic region of 10 . 3 kb ( average length in C . albicans ∼0 . 9 kb ( http://www . broad . mit . edu/annotation/genome/candida_albicans/GeneStatsSummary . html ) and it is able to bind to several positions in its own upstream region , although the protein does not contain an annotated DNA binding domain [13] . Also Ryp1 , which is preceded by an intergenic region of about 2 kb , is able to bind to several positions in its own upstream region [11] . The intergenic region preceding SGE1 is larger than the average in the F . oxysporum genome: 4 . 3 kb versus 2 . 0 kb ( http://www . broad . mit . edu/annotation/genome/fusarium_group/GeneStatsSummary . html ) , but whether Sge1 is able to activate its own transcription by binding to its promoter remains to be elucidated . Nevertheless , given the nuclear localization of Sge1 , the homology to the transcriptional regulators Wor1 and Ryp1 , and the effect of deletion of SGE1 on expression of SIX genes , we propose that also Sge1 also acts as a transcriptional regulator . The requirement of Sge1 for expression of effector genes may explain the non-pathogenic phenotype of the SGE1 deletion mutant . Based on the apparent absence of xylem colonization as observed with confocal microscopy and the absence of the sge1 mutant in hypocotyls from infected plants , we conclude that although the mutant is still able to colonize roots and penetrate the root surface , it is not able to grow extensively in living cells or the xylem system . The ability of the sge1 mutant to colonize roots superficially is supported by the biocontrol activity of the sge1 mutant . The only additional phenotype of the mutant that we found was a 6-fold lower microconidia production compared to the wild type strain . An aberrant conidiogenesis behaviour was also observed for the H . capsulatum ryp1 mutant [11] , indicating a conserved role for these proteins in conidiogenesis . In C . albicans , the regulatory network that drives white-opaque switching , to which Wor1 belongs , is beginning to be understood . A gene counteracting WOR1 is EFG1 , which is necessary for maintaining the white state [29] . Wor1 is able to bind to the upstream region of EFG1 , thereby regulating its expression [20] . Interestingly , the Fusarium homolog of Efg1 is FoStuA , a protein required for conidiogenesis in F . oxysporum [30] . However , STUA expression levels in the wild type and in the SGE1 deletion mutant were comparable ( data not shown ) . How Sge1 affects conidiogenesis remains , therefore , to be elucidated . As mentioned above , a common feature of Sge1 and its homologs is the presence of a potential protein kinase A phosphorylation site . The functionality of this phosphorylation site has been investigated in S . pombe . It was shown that gluconate import under repressing conditions ( high glucose concentration ) occurred in a protein kinase A deletion mutant . Therefore , it was speculated that Pka1 inhibits Gti1 protein function . However , alteration of the Pka phosphorylation site by replacement of the conserved threonine residue by an alanine did not result in activation of the Gti1 protein and the expected gluconate import was not observed , indicating that either Pka1 does not inhibit Gti1 activity by phosphorylation or that the mutation impairs Gti1 function or its stability [14] . In Sge1 , the same Pka site appears to be required for activity . A stable transformant of SGE1 in S . cerevisiae was obtained only after a spontaneous mutation of an arginine into a serine in the potential phosphorylation site , suggesting that expression of wild type SGE1 in yeast inhibits growth . S . cerevisiae contains a SGE1 homolog , YEL007W , and this gene has been implicated to play a role in regulation of smooth ER , cell adhesion and budding [31]–[33] . It could be that expression of both genes is lethal to yeast . Introduction of the mutated SGE1 gene in the SGE1 deletion mutant of Fol failed to restore the pathogenicity defect , confirming that the R66S mutation impairs Sge1 function , possibly due to an effect on phosphorylation . Since there is only a minor increase in SGE1 expression levels upon in planta growth , post-translational modifications such as phosphorylation could play a key role in activation of Sge1 . Unfortunately , initial attempts to demonstrate phosphorylation by Pka1 of purified Sge1 and Sge1R66S in vitro remained inconclusive ( data not shown ) and will be subject of further investigation . The loss of effector ( SIX ) gene expression in the sge1 mutant supports a role of Sge1 as a transcriptional activator controlling the onset of parasitic growth . In Fol , these effector genes encode small cysteine rich proteins secreted during colonization of xylem vessels , designated Secreted in xylem ( Six ) proteins [21] , [34] . SIX1 is only expressed during in planta growth and encodes Avr3 , as it is required for I-3 mediated resistance [22] , [34] . Six3 is Avr2 and is required for I-2 mediated resistance [25] . Both Six1/Avr3 and Six3/Avr2 are required for full virulence [23] , [25] . No function has yet been assigned to Six2 and Six5 . Here , we show that , like Six1 , Six2 is highly expressed during in planta-mimicking growth conditions ( co-cultivation with tomato cells ) , but SIX3 and SIX5 , divergently transcribed from the same promoter region , are expressed in synthetic medium . It could be that Sge1 activity under axenic growth conditions is too low for induction of SIX1 and SIX2 , but high enough to induce SIX3 and SIX5 expression . The increased SGE1 expression levels during in planta growth may then lead to expression of all SIX genes . Currently , a SGE1 over-expression mutant is being generated in order to determine whether increased SGE1 expression alone can lead to expression of all SIX genes under axenic growth conditions . At the moment , it is not known whether Sge1 influences the expression of the SIX genes directly or indirectly , for instance through involvement of other ( transcription ) factors , nor how SGE1 expression itself is regulated . Preliminary promoter analysis of the SIX genes revealed a potential common motif ( unpublished data ) . The role of this motif in SIX gene expression and the question whether Sge1 is able to bind DNA and the SIX gene promoter regions in particular remain to be elucidated . All the genes tested in the SGE1 deletion mutant ( SIX1 , SIX2 , SIX3 , and SIX5 ) were found to be dependent on Sge1 for their expression , even during growth in synthetic medium . Although deletion of individual SIX genes ( SIX1 or SIX3 ) only leads to a minor reduction in pathogenicity , the loss of expression of all SIX genes in the sge1 mutant may be the primary cause of the complete non-pathogenic phenotype , due to the inability to suppress host defence responses . Still , loss of pathogenicity upon deletion of SGE1 may not be due to loss of production of effector proteins only . The majority of pathogenicity genes found in F . oxysporum have pleiotropic functions [35] , [36] . It is therefore unlikely that expression of these genes is altered in the sge1 mutant , since this mutant did not display growth defects other than a minor effect in conidiogenesis . One described F . oxysporum mutant that is only disturbed in pathogenicity is the fow2 mutant [37] . Expression of FOW2 nor that of two other pathogenicity genes , the protein kinase gene SNF1 [38] and velvet homolog gene FOXG_00016 [39] , is altered in the SGE1 deletion mutant ( unpublished data ) . However , preliminary results suggest that the expression of the FTF1 transcription factor gene , implicated to be involved in pathogenicity [40] , seems to be altered in the sge1 mutant ( unpublished data ) . In addition , secondary metabolite profiling revealed that the sge1 mutant is reduced in the production of fusaric acid , beauvericin and a number of unknown metabolites compared to the wild type ( U . Thrane and C . B . Michielse , unpublished data ) . High concentrations of fusaric acid may contribute to pathogenicity by reducing host resistance [41] , [42] . Thus , processes implicated in pathogenesis other than SIX gene expression are also altered in the sge1 mutant , supporting the hypothesis that Sge1 plays a central role during parasitic growth . Deletion of the Sge1 homolog in Botrytis cinerea , BC1G_11680 , also leads to a severely reduced pathogenicity phenotype ( C . B . Michielse and P . Tudzynski , unpublished data ) , indicating that the role of Sge1-like proteins in ( plant ) pathogenic fungi might be conserved . The regulation of expression of phase specific genes ( SIX1 , SIX2 ) , and genes involved in virulence ( SIX1 , SIX3 ) by Sge1 resembles the functions of Wor1 in C . albicans and Ryp1 in H . capsulatum . Wor1 has been shown to bind directly to the upstream region of target genes [20] . Interestingly , these target genes all have a large upstream region , indicating that large upstream regions might be a prerequisite for Wor1 binding . Whether this is a common feature shared between Wor1 and Sge1 is still unknown . Future efforts include the identification of additional genes regulated by Sge1 and , eventually , to unravel the network of transcriptional regulators involved in activating the pathogenicity program in F . oxysporum . This will also reveal whether Sge1 indeed plays a master role in this network , like its homologs in dimorphic fungi . Given the striking conservation of Sge1 across fungi , we expect that many features of this network will turn out to be similar in pathogens of both plants and animals . Obviously , since Sge1 homologs are also present in non-pathogenic fungi , this regulatory network must serve a fundamental morphological or physiological switch function and pathogens would have adopted this network for adapting to the host environment .
Fusarium oxysporum f . sp . lycopersici strain 4287 ( race 2; FGSC9935 ) was used as the parent strain for fungal transformation . It was stored as a monoconidial culture at −80°C and revitalized on potato dextrose agar ( PDA , Difco ) at 25°C . Agrobacterium tumefaciens EHA105 [43] was used for Agrobacterium-mediated transformation of F . oxysporum and was grown in 2YT medium [44] containing 20 µg/ml rifampicin at 28°C . Introduction of the plasmids into the Agrobacterium strain was performed as described by Mattanovich et al [45] . Escherichia coli DH5 alpha ( Invitrogen ) was used for construction , propagation , and amplification of the plasmids and was grown in LB medium at 37°C containing either 100 µg/ml ampicillin or 50 µg/ml kanamycin depending on the resistance marker of the plasmid used . Plant line Moneymaker ss590 ( Gebr . Eveleens b . v . , The Netherlands ) was used to assess pathogenicity of F . oxysporum strains and transformants . Biocontrol assays were performed on flax ( Linum usitatissimum ) cv . viking using F . oxysporum f . sp . lini ( Foln3 , MYA-1201 ) isolated from a diseased flax plant in French Britany as pathogenic strain . To generate the SGE1 disruption construct , pSGE1KO , PCR was performed using a BAC clone containing the SGE1 gene as a template with primer combination FP878–FP879 and FP880–FP881 in order to amplify a 901 bp HindIII fragment corresponding to the upstream region and a 1032 bp KnpI fragment corresponding to the downstream region , respectively ( Table S1 ) . The fragments were sequentially cloned into pPK2hphgfp [9] and proper orientation of the inserts was checked by PCR . The SGE1 complementation construct , pSGE1com , was generated by cloning a 2234 bp HindIII fragment containing the SGE1 ORF including 901 bp upstream and 341 bp downstream sequences into pUC19 , resulting in pUC19SGE1 . The HindIII SGE1 fragment was subsequently transferred from pUC19SGE1 into pRW1p [24] , resulting in pSGE1com . The SGE1 complementation construct carrying a point mutation in the ORF of SGE1 , pSGE1comPM , was generated by replacing a 465 bp SacII/BglII SGE1 fragment in pSGE1com by a 465 bp SacII/BglII fragment isolated from the yeast-two-hybrid bait vector pASSGE1PM . This vector was re-isolated from Saccharomyces cerevisiae after previously being transformed with pASSGE1 . The latter vector was generated by cloning an NcoI-EcoRI 1011 bp PCR product obtained with the primers FP1484 and FP1485 corresponding to the SGE1 ORF in pAS2 . 1 ( Clontech ) . To generate the FoPAC2 gene disruption construct , a 1037 bp upstream fragment and a 749 bp downstream fragment was amplified from genomic DNA by PCR with the primer pairs FP1796–FP1797 and FP1798–FP1799 , respectively . The PCR products were cloned into pGEM-T Easy ( Promega ) and , subsequently , the KpnI/PacI upstream and the AscI/HindIII downstream fragment were sequentially cloned in pRW2h [24] . The Sge1-fluorescent protein fusion constructs pSGE1::CFP and pSGE1::RFP were generated in a multi-step approach . First , pUC19SGE1 was amplified by PCR with the primers FP1120 and FP1121 flanked at the 5′ end by an ApaI and a SpeI restriction site , respectively . The amplified product was digested with ApaI and religated , resulting in pUC19SGE1as , containing the SGE1 ORF with the ApaI and SpeI preceding the SGE1 stop codon . Secondly , a CFP [46] and a mRFP [47] fragment were generated by PCR using primer pair FP1122–FP1123 and FP1120–FP1121 , respectively . The CFP and RFP PCR products were digested with ApaI and SpeI and directionally cloned in pUC19SGE1as , resulting in pUCSGE1::CFP and pUCSGE1::RFP , respectively . Finally , a HindIII fragment corresponding to each fusion cassette was cloned into HindIII digested pRW1p and proper orientation of the fragments was checked by PCR . The gpdAH2B::GFP expression cassette was isolated as a XbaI/NheI fragment from pH2B::GFP ( kind gift from Dr . Ram , Leiden , The Netherlands ) and cloned into XbaI/NheI digested pRW2h [24] to generate pRWH2B::GFP . Agrobacterium-mediated transformation of F . oxysporum f . sp . lycopersici was performed as described by Mullins and Chen [48] with minor adjustments [49] . Depending on the selection marker used , transformants were selected on Czapek Dox agar ( CDA , Oxoid ) containing 100 µg/ml Hygromycin ( Duchefa ) or on CDA containing 0 . 1 M TrisHCl pH 8 and 100 µg/ml Zeocin ( InvivoGen ) . Plant infection was performed using 9 to 11 days old seedlings ( Moneymaker ss590 ) , following the root-dip inoculation method [50] . Disease index was scored and statistical analysis performed as described earlier [9] . For quantization of microconidia production , five independent experiments were performed , each with five replicates . Microconidia were harvested after five days of growth in 50 ml minimal medium ( 3% sucrose , 10 mM KNO3 and 0 . 17% yeast nitrogen base without amino acids and ammonia ) and 106 spores were used to inoculate 100 ml minimal medium . After five days the microconidia produced were harvested and counted in a Bürker-Türk haemocytometer . The isolated conidia were also used to determine germination rates . To this end 600 spores were added into a 6-wells plate with each well containing 250 µl PDA and incubated overnight at 4°C , then transferred to 25°C and germinated conidia were counted after six hours of incubation . Macroconidia development was analyzed in liquid carboxymethyl cellulose medium as described by Ohara and Tsuge [30] . Conidia were fixed in 0 . 4% p-formaldehyde and stained with Hoechst 33342 ( 250 µg/ml ) and calcofluor white ( 25 µg/ml ) to visualize nuclei and cell walls , respectively . Stained cells were observed with a BX50 fluorescence microscope and a U-MWU filter ( Olympus ) . Carbon-source utilization was tested in a BIOLOG FF MicroPlate ( BIOLOG ) . A conidial suspension ( 104 conidia in 150 µl ) of each strain was inoculated in each well of the plate and incubated at 25°C . The absorbance of each well at 600 nm was measured with a microtiter plate reader ( Packerd Spectra Count ) after 1 , 2 , 3 and 4 days of incubation . The cellophane penetration assay was performed as described [18] , with minor adjustments . CDA was used as basal medium in the assay and was inoculated with 105 conidia . After incubation of 5 days at 25°C , the cellophane was removed and after a subsequent incubation of 2 days at 25°C fungal growth was scored . Genomic DNA was isolated as described by Kolar et al [51] with minor adjustments [9] . For Southern analysis , 10 µg genomic DNA of each transformant was digested with Acc65I or BglII for SGE1 or FoPAC2 transformants , respectively , and incubated overnight at 37°C . The samples were loaded on a 1% 0 . 5× Tris-borate/EDTA gel and run for 18 hours at 45 V . The digested DNA was transferred to Hybond-N+ ( Amersham Pharmacia ) as described by Sambrook et al [44] . The probes used for Southern analysis are: ( 1 ) a 432 bp SGE1 upstream fragment obtained by PCR with primers FP842 and FP1174 ( Table S1 ) and ( 2 ) a 488 bp FoPAC2 downstream fragment obtained by PCR with primers FP2198 and FP 2199 . The DecaLabel™ DNA Labelling Kit ( Fermentas ) was used to label probes with [α-32P]dATP . Hybridization was done overnight at 65°C in 0 . 5 M sodium phosphate buffer , pH 7 . 2 , containing 7% SDS and 1 mM EDTA . Blots were washed with 0 . 2× SSC , 0 . 1% SDS . Hybridization signals were visualized by phosphorimaging ( Molecular Dynamics ) . Samples for expression analysis were obtained by adding 0 , 5 ml of 107 conidia/ml of wild type , SGE1 deletion or SGE1 complementation strain to 4 , 5 ml of a one week old MSK8 cell culture grown in BY-medium [52] . After 24 hours of incubation at 22°C , the material was harvested , washed two times with sterile water and freeze-dried . Total RNA was isolated using Trizol ( Invitrogen ) and prior to cDNA synthesis the RNA was treated with DnaseI ( Fermentas ) . First strand cDNA was synthesized with M-MuLV reverse transcriptase following the instruction of the manufacturer ( Fermentas ) . Supertaq ( SphaeroQ ) , 1 µl of the cDNA reaction and primers FP1999/FP2000 , FP998/FP1001 , FP962/FP963 , FP1993/FP1994 , FP2131/FP2132 and FP157/FP158 ( Table S1 ) to detect SIX1 ( CAE55879 ) , SIX2 ( CAE55868 ) , SIX3 ( CAJ83999 ) , SIX5 ( ACN87967 ) , SGE1 ( FOXG_10510 ) and FEM1 ( AAL47843 ) transcripts , respectively , were used in the RT-PCR . Quantitative PCR ( qPCR ) was performed with Platinum SYBR Green qPCR ( Invitrogen ) using a 7500 Realtime PCR System ( Applied Biosystems ) . To quantify mRNA levels of SGE1 and of the constitutively expressed EF-1α gene , we used primers FP2131/FP2132 and FP2029/FP2030 , respectively ( Table S1 ) . EF-1α was used to calculate the relative expression levels of SGE1 in axenic and MSK8 cultures infected with F . oxysporum wild type , SGE1 deletion or SGE1 complementation mutants . Real time PCR primer efficiencies were calculated using LinRegPCR [53] and relative expression levels were calculated according to the comparative C ( t ) method [54] . Sge1::mRFP , Sge1::CFP and H2B::GFP fusion proteins were visualized using a BX50 fluorescence microscope with the appropriate excitation and emission filters for CFP , mRFP and GFP ( Olympus ) . For this purpose , 10 µl of five days old minimal medium cultures was spotted onto glass slides . A Nikon A1 microscope was used to monitor tomato root infection by the wild type and SGE1 deletion mutant . Excitation was provided for the GFP signal with an Argon ( 488 nm ) laser ( emission: 405–455 nm ) and for the root auto-fluorescence with an UV diode ( 405 nm ) laser ( emission: 420–470 nm ) . Images were line-sequential scanned ( 2 µm slices , 1024×1024 pixels ) using water objective plan fluor 20× Imm , 0 . 75 NA . Pictures were analyzed with the Nikon NIS and ImageJ software . For preparation of the slides , 9 days old tomato seedlings were carefully removed from the potting soil , rinsed with water and intact roots were inoculated with 20 ml tap water containing 107 conidia/ml and incubated for 4 days at room temperature in a petridish . The roots were rinsed with water , cut from the hypocotyl , placed in a drop of water on a glass slide and covered with a cover glass . A bridge mounted on the glass slide prevented squashing of the root material . Inocula were prepared as described [55] , except that minimal liquid medium [56] in which sucrose was replaced by glucose ( 5 g/l ) and sodium nitrate by ammonium tartrate ( 1 g/l ) was used instead of Malt medium . A heat treated ( 100°C for 1 h ) silty-loam soil from Dijon ( 35 . 1% clay , 47% loam , 15 . 1% sand , and 1 . 22% organic C [pH = 6 . 9] ) was added to module trays containing 96 wells each of 50 ml . To prevent contamination between treatments , only every second row of wells was filled with soil . The soil in each well was inoculated with 5 mL of conidia suspension . The concentration of the conidia suspensions were adjusted to obtain the following inoculum densities: 1×104 conidia ml−1 of soil for the pathogenic strain Foln3 and 1×106 conidia ml−1 of soil for the two others strains . The soil surface was covered with a thin layer of calcinated clay granules ( Oil Dri Chem-Sorb , Brenntag Bourgogne , Montchanin , France ) and three seeds of flax , cv . viking , were sown in each pot . A thin layer of Chem-Sorb was used to cover the seeds . Plants were grown in a growth chamber in the first 2 weeks; the growing conditions were 8 h 15°C N/16 h 17°C D , with a light intensity of 33 µE m−2 s−1 . The plants were thinned to one plant per pot , and from week 3 the temperature was kept at 22°C N/25°C D . There were three replicates of 16 individual plants per treatment , randomly arranged , and the experiment was replicated . Plants were watered every day , and once a week water was replaced by a 500-fold dilution of a commercial nutrient stock solution ( “Hydrodrokani AO” , Hydro Agri , Nanterre , France ) . Plants showing characteristic symptoms of yellowing were recorded twice a week and removed . To compare the ability of strains to induce disease or , on the contrary , to protect the plant against wilt , ANOVA was performed on Area under the Disease Progress Curves followed by Newman and Keuls test at the probability of 95% . | Plant pathogenic fungi have evolved many ways to infect their hosts and can have devastating effects on commercial crop production . Dissecting their infection strategies and understanding the molecular pathways involved in pathogenesis have been and continue to be the subject of intensive research . New insights gained may help to develop disease controlling strategies . Fusarium oxysporum has become a model for root invading , pathogenic fungi . This fungus attacks a wide range of plant species worldwide and the only effective disease control strategies in the field are crop rotation and usage of resistant plant varieties , if at all available . In the last decade , many genes have been identified that play a role during pathogenesis , many of which are linked to general strain fitness . Only few genes have been identified that are required for pathogenesis but do not affect vegetative growth . This paper describes the characterization of such a gene . Its protein product , Sge1 , is conserved in the fungal kingdom and represents a new class of transcriptional regulators involved in morphological switching in dimorphic fungal pathogens . In F . oxysporum , the Sge1 protein is required for parasitic growth and is associated with expression of parasitic phase-specific genes . We suggest that the function of Sge1 is conserved in ( plant ) pathogenic fungi . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"microbiology/plant-biotic",
"interactions",
"pathology/molecular",
"pathology"
] | 2009 | The Nuclear Protein Sge1 of Fusarium oxysporum Is Required for Parasitic Growth |
Aurora kinases constitute a family of enzymes that play a key role during metazoan cells division , being involved in events like centrosome maturation and division , chromatin condensation , mitotic spindle assembly , control of kinetochore-microtubule attachments , and cytokinesis initiation . In this work , three Aurora kinase homologues were identified in Trypanosoma cruzi ( TcAUK1 , -2 and -3 ) , a protozoan parasite of the Kinetoplastida Class . The genomic organization of these enzymes was fully analyzed , demonstrating that TcAUK1 is a single-copy gene , TcAUK2 coding sequence is present in two different forms ( short and long ) and TcAUK3 is a multi-copy gene . The three TcAUK genes are actively expressed in the different life cycle forms of T . cruzi ( amastigotes , trypomastigotes and epimastigotes ) . TcAUK1 showed a changing localization along the cell cycle of the proliferating epimastigote form: at interphase it is located at the extremes of the kinetoplast while in mitosis it is detected at the cell nucleus , in close association with the mitotic spindle . Overexpression of TcAUK1 in epimastigotes leaded to a delay in the G2/M phases of the cell cycle due a retarded beginning of kinetoplast duplication . By immunofluorescence , we found that when it was overexpressed TcAUK1 lost its localization at the extremes of the kinetoplast during interphase , being observed inside the cell nucleus throughout the entire cell cycle . In summary , TcAUK1 appears to be a functional homologue of human Aurora B kinase , as it is related to mitotic spindle assembling and chromosome segregation . Moreover , TcAUK1 also seems to play a role during the initiation of kinetoplast duplication , a novel role described for this protein .
Cell cycle in eukaryotic cells involves the sequential transition between G1 , S , G2 and M phases and this progression is tightly regulated by protein kinases and phosphatases [1] . During the M phase , the cell divides its nucleus to originate two daughter cells with the same genetic content . In this event , Aurora kinase proteins play a crucial role . The Aurora kinase family of proteins presents a variable number of members among different organisms . While yeasts present a single Aurora kinase gene [2 , 3] , organisms like C . elegans and D . melanogaster have two genes [4–7] , whereas in vertebrates three Aurora kinase proteins are present [8] . In this last group , the three members of the Aurora kinase family are named Aurora-A , -B and -C and each protein plays specific functions during the cell cycle . In organisms with a single Aurora kinase gene , the encoded protein combines the function of both Aurora-A and -B , whereas in organisms with two Aurora proteins , one behaves as an Aurora-A while the other has functions similar to Aurora-B . In humans , Aurora-C is expressed in germinal cell lines and its function has not been elucidated yet . A distinctive feature of all Aurora proteins is that they change their cellular location during mitosis progression , according to their different roles . Aurora-A , the so-called Polar Aurora , is involved in centrosome maturation/migration and bipolar spindle formation/stabilization [9] and therefore , it is found in the neighborhood of the dividing centrosome in early mitosis , after which it moves with each of duplicated centrosome to opposite extremes of the cell , where the spindle poles locate during G2 phase [10] . Aurora-B is named Equatorial Aurora because it locates in the mid-plane of the cell during mitosis ( metaphase ) and , as nuclear division proceeds , it is tightly associated with segregating chromosomes . This Aurora forms the Chromosomal Passenger Complex ( CPC ) with other three proteins ( INCENP , Survivin and Borealin ) . As part of this complex , Aurora-B promotes chromatin condensation in prophase through phosphorylation of histone H3 at Ser10 [11] . In metaphase , it participates at the Spindle Checkpoint to ensure correct chromosome segregation during anaphase [12] . Finally , in cytokinesis the CPC settles in the cell midzone and participates in the formation of the contractile ring and the cleavage furrow [13] . Trypanosoma cruzi is a protozoan of the Kinetoplastida order and is the etiological agent of the American Trypanosomiasis , also known as Chagas disease . This disease is endemic in Latin America but in the last decades , due to the increasing migratory flux , a growing number of infections have been detected in non-endemic countries like the United States and Spain [14] . The complex life cycle of T . cruzi involves different forms: the epimastigotes and metacyclic trypomastigotes are present in the insect vector , and the amastigotes and bloodstream trypomastigotes are found in the vertebrate host . All these forms contain a single flagellum emerging from the basal body , a nucleus and a mitochondrion carrying the DNA complex known as kinetoplast . During cell division , all these organelles are replicated and segregated into two daughter cells in a synchronized manner . Moreover , both the existing and newly synthesized basal bodies physically interact with the duplicating kinetoplast and drive its division [15] , meaning that the synchronization is due in part to a physical contact between these organelles . During T . cruzi mitosis , contrary to what is observed in other organisms , the nuclear membrane remains intact and the chromosomes do not condensate . Despite this , the formation of a mitotic spindle inside the nucleus has been described [16] and it is well known that chromosome segregation is mediated by a microtubules-dependent mechanism [17] . While most eukaryotic cells contain many mitochondria with separate copies of circular DNA molecules , kinetoplastids have a single mitochondrion with a genome in which DNA molecules are physically interlocked forming a big network , the so-called kinetoplastid DNA ( kDNA ) . These DNA molecules consist of two type of circles: the larger maxicircles that are the equivalent of the mitochondrial genome in other organisms and are present as several dozen identical copies per cell , and the smaller minicircles that codified for guide RNAs ( gRNAs ) and are in thousands copies per cell . During cell division , this DNA does not only need to be duplicated but a well-orchestrated de-concatenation and segregation , driven by the recently duplicated flagella basal bodies , needs to take place [18] . The complexity of the mitochondrion genome adds for a particular kDNA S-phase besides from the classical G1 , S and G2/M cell cycle stages . Although contributions have been made by different authors to the knowledge about the kDNA replication mechanism [19] , many of the molecular players involved in the later steps of this process remain obscure , mainly the ones involved in kDNA division and segregation . Aurora kinase genes have been identified and described in different protozoan organisms . In Leishmania major , a single Aurora gene ( Lmairk ) has been reported but its function has not been studied yet [20] . In the apicomplexa Plasmodium falciparum , three Aurora genes have been described ( Pfark -1 , -2 and -3 ) Pfark-1 was defined based on its subcellular localization as the classic Aurora gene present in other organism , whereas the remaining two Pfark proteins seem to be involved in cellular processes exclusive of this organism [21] . Davids and coworkers found a single Aurora gene in the biflagellate Giardia lamblia [22] , showing that this protein adopts a cellular localization similar to mammalian Aurora-A and Aurora-B and is involved in microtubules dynamics reorganization at mitosis and interphase . In Trypanosoma brucei three Aurora genes were identified by Wang and coworkers but gene silencing experiments demonstrated that only TbAUK1 is a functional gene , at least in the procyclic form [23] . This protein is involved in mitotic spindle assembling and it was defined by the authors as a human Aurora-B ortholog . Later , these same authors showed that TbAUK1 silencing in the bloodstream form leads to failure to conclude cytokinesis and cell shape alteration , both effects associated to a microtubule filaments disruption [24] . TbAUK1 seems to be associated to others proteins conforming a complex like the CPC of mammals and , as with Aurora B , the proteins of this complex affect TbAUK1 localization and function [25] . A comparative analysis of the kinomes of three pathogenic kinetoplastids—including T . cruzi–by Parsons and co-workers reported the presence of several kinases normally associated to roles in cell division , the Aurora kinases being part of this group [26] . In this work we report the initial characterization of three Aurora kinase proteins in T . cruzi ( TcAUK1 , TcAUK2 and TcAUK3 ) . In addition , by a detailed analysis of TcAUK1 localization , we have found that this protein shows the canonical behavior of a chromosome passenger protein , being associated with the mitotic spindle during nuclear division . Furthermore , we detected that during interphase , TcAUK1 is located at both sides of the kinetoplast . Finally , we report that TcAUK1 overexpression in epimastigote forms causes a delayed G2-M transition , presumably by affecting the onset of kinetoplast duplication .
Radio chemicals were purchased from PerkinElmer Life Sciences , and restriction endonucleases were from New England Biolabs , Beverly , MA . Bacto-tryptose , yeast nitrogen base , and liver infusion were from Difco . All other reagents were purchased from Sigma . The gene sequences corresponding to TbAUK1 ( Tb927 . 11 . 8220 ) , TbAUK2 ( Tb927 . 3 . 3920 ) and TbAUK3 ( Tb927 . 9 . 1670 ) were used to screen T . cruzi sequences in TryTrip database using BlastN algorithm . Pairwise alignment and motif search were performed on high-scored targets by EMBL-EBI tools [27] and Pfam [28] , as well as manual inspection . Multiple sequence alignment was performed in MEGA 5 software [29] with ClustalW algorithm and visualized with BioEdit software [30] . T . cruzi epimastigote of CL Brener strain were grown at 28°C with 5% CO2 in LIT medium [5 g . l-1 liver infusion , 5 g . l-1 Bacto-tryptose , 68 mM NaCl , 5 . 3 mM KCl , 22 mM Na2PO4 , 0 . 2% ( w/v ) glucose , 0 . 002% ( w/v ) hemin] containing 10% v/v fetal bovine serum ( NATOCOR , Argentina ) , 100 units . ml-1 penicillin and 100 μg . l-1 streptomycin . Cell density was maintained between 1x106 and 1x108 cells . ml-1 sub-culturing parasites every 6–7 days . For growth curve determinations , a sample of culture supernatant was taken and swimming epimastigotes were fixed by incubation in 4% formaldehyde in PBS for 5 min at room temperature . Cell density was determined by counting at least three independent cultures in an hemocytometer . Specific growth rate ( μ , expressed as h-1 ) was estimated by the slope of the graphic “Ln of culture cell density” vs “culture time” ( h ) . Cells duplication time ( DT ) was calculated according to the formula: DT=ln2μ Cercopithecus aethiops ( green monkey ) Vero cells ( ATCC CCL-81 ) were cultured at 37°C and 5% CO2 supplied in Minimum Eagle Medium ( MEM , Gibco ) supplemented with 10% fetal bovine serum ( HyClone ) , 2 mM L-Glutamine ( Sigma ) , 100 units . ml-1 penicillin and 100 μg . l-1 streptomycin . For the obtention of T . cruzi trypomastigotes and amastigotes Vero cells were infected with trypomastigotes ( 1:50 ratio ) 24 h after being plated and maintained in MEM supplemented with 3% FBS . Trypomastigotes in culture supernatant were harvested by centrifugation and processed as needed . Amastigotes were collected from 10–11 day old cultures from the supernatant ( 90% or higher amastigotes/trypomastigotes ratio ) , centrifuged , and processed as needed . Based on the sequences of Aurora kinases homologs found in T . cruzi database , specific primers were designed: TcAUK1 ( TcAUK1-NcoI-fwd 5´-CCATGGTGAGTGCGGCGGAGGGCGGCCAA-3´ and TcAUK1-XhoI-rev 5´-CTCGAGGTTCTCCTTTCCGCCCGAGAAGT-3´ ) , TcAUK2 ( TcAUK2-BamHI-fwd 5´-GGATCCGCAGCACCACAACTTGAGTTCC-3´ and AUK2-XhoI-rev 5´- CTCGAGCTTCTTCTTCTTCTTCTCCCCATTT-3´ ) , TcAUK3 ( AUK3-NcoI-fwd 5´- CCATGGTGTGGTCGCTGGATGACTTTGAT-3´ and AUK3-XhoI-rev 5´-CTCGAGTAAATTCTCTGCCGCATCAACCGT-3´ ) . Polymerase Chain Reaction ( PCR ) was performed in a PTC-150 MiniCycler ( MJ Research ) . For this , genomic DNA of T . cruzi was isolated as described previously [31] and gene amplification was performed by using a high fidelity DNA polymerase ( Herculase II , Stratagene ) . The thermal cycling conditions were specific for each TcAUK gene . Amplification products were gel-purified , subcloned into pGEM-T Easy vector ( Promega ) , transformed into E . coli DH5α competent cells and both strands were sequenced ( Macrogen , Korea ) . Genomic DNA of epimastigote forms was digested ( approx . 30 μg ) overnight with 30 units of the indicated restriction enzyme ( New England Biolabs ) . After digestion , DNA was SpeedVac concentrated ( Jouan RCT 60 Refrigerated Cold Trap ) and electrophoresed for 8–12 h in 0 . 8% agarose gel ( 1 V/cm ) and was then denatured , neutralized and transferred onto nylon membrane ( GeneScreen , Perkin Elmer ) for Southern blot analysis [32] . For this , specific radiolabeled probes were generated by primer extension using full-length TcAUK genes and [α-32P]-dCTP ( NEBlot kit , New England Biolabs ) . dCTP radiolabeled probes were then purified with MicroSpin G-50 columns ( GE Healthcare ) and heat denaturalized before proceeding with hybridization . Probes were hybridized at 65°C ( overnight ) and washed at 65°C using 2x SSC , 1xSSC and 0 . 5x SSC with 0 . 1% SDS sequentially to remove excess of probe . Blot was developed by exposing membrane to Phosphoimager Storm system ( Pharmacia-Biotech ) . To perform Pulsed Field Gel Electrophoresis , CL Brener exponentially growing epimastigotes were washed with PBS-Glucose 2% , suspended in PBS and mixed with one volume of low melting point agarose 1 . 4% in PBS . After polymerization , the agarose plug was incubated with LIDS buffer ( 1% Lauryl Sulfate Lithium Salt , 10 mM Tris-HCl pH8 . 0 , 0 . 1 M EDTA ) for 48 h at 37°C . Afterwards the blocks were washed 6 times with NDS 0 . 2% buffer ( 0 . 2% N-Lauroylsarcosine-Sodium Salt , 0 . 1 M EDTA , 2 mM Tris base ) and , before running the agarose gel , they were equilibrated in TE buffer pH 8 , 0 . Gel electrophoresis was performed at 16°C in three steps: 1 ) 3 V/cm changing periods every 90–200 sec during 30 h; 2 ) 3 V/cm changing periods every 200–400 sec during 30 h; 3 ) 2 . 7 V/cm changing periods every 400–700 sec during 24 h . After electrophoresis , DNA was transferred to a nylon membrane and hybridized with TcAUKs probes as described above for Southern blot analysis . Total RNA was isolated from epimastigotes , trypomastigotes and amastigotes forms using Trizol according to the manufacturer’s protocol ( Invitrogen ) . cDNA was obtained from mRNA by Transcriptor First Strand cDNA Synthesis Kit ( Roche ) using oligo ( dT ) 18 primer , following the supplier´s instructions . These cDNA samples were used to amplify a fragment of TcAUKs genes and the housekeeping Actin gene ( TcCLB . 510945 . 30 ) . For all TcAUKs targets the same forward Tc-SL primer was used ( 5´-AACGCTATTATTGATACAGTTTC-3´ ) whereas a specific reverse primer for each one was designed: TcAUK1-RT-rev ( 5´-CCACCCAAAGTCTGCCAACTTA-3´ ) , TcAUK2-RT-rev ( 5´-AGCGTGCGGTGAACGTTGATCT-3´ ) and TcAUK3-RT-rev ( 5´-AATCCATCGTGCCGCAAAGCGT-3´ ) . In the case of the housekeeping actin gene the primers used were: TcActin-fwd ( 5´-ATGATCATCGTGGACTTTGGGT-3´ ) and TcActin-rev ( 5´-TTCCGCTTGGGTGTGAACAGC-3´ ) . The PCR reactions included an initial denaturalization step at 95°C for 2 min , followed by 35 cycles at 95°C for 1 min , annealing at 60°C for 45 sec , extension at 72°C for 45 sec and a final extension at 72°C for 5 min . Amplification products were visualized by electrophoresis on 1 . 5% agarose gel . Then , they were gel extracted , subcloned into pGEM-T Easy vector ( Promega ) and sequenced ( Macrogen , Korea ) . For western blot analysis cells were suspended in lysis buffer ( 50 mM Tris-HCl pH 8 . 0 , 1 mM EDTA , 1 mM DTT , 0 . 1% Triton X-100 , 1% NP-40 , 1 mM PMSF and 1 μg . ml-1 E-64 ) and lysed by freeze/thaw cycles in liquid nitrogen . The obtained lysate was centrifuged at 10 , 000 x g for 30 min and the pellet was discarded . Total protein concentration in the extracts ( supernatant ) was estimated by the Bradford quantification , and an aliquot containing 20–80 μg of proteins was loaded onto 12% ( w/v ) SDS-polyacrylamide gel , solved by electrophoresis as described by Laemmli [33] and electro-transferred to nitrocellulose membranes ( Hybond-C , Amersham Pharmacia Biotech ) . The membranes were then blocked with 5% ( w/v ) non-fat milk suspension in TBS-Tween 0 . 05% for 2 h and TcAUK1 was detected with a rabbit antiserum to TcAUK1 ( custom produced by GenScript Corporation against the peptide PRGKRMRGAADFSG , amino acids 292 to 305 of TcAUK1 ) and a goat antiserum to rabbit IgG HRP-labeled secondary antibody ( PerkinElmer ) . Specific TcAUK1 signal was developed with the ECL Plus Western blotting detection system ( PerkinElmer Life Sciences ) . The full-length coding sequence of TcAUK1 gene sub-cloned into pGEM-T Easy vector ( see above ) was isolated by digestion with EcoRI and XhoI nucleases and cloned into pTREX plasmid [34] digested with the same restriction enzymes . T . cruzi epimastigotes of CL Brener strain were electro-transfected with empty pTREX plasmid or the pTREX-TcAUK1 construct as described previously [34] . Stable transfectant pools were achieved after 60 days of treatment with 500 μg . ml-1 G418 ( Gibco BRL , Carlsbad , CA ) . Once selection has finished , single clone cell cultures were obtained for pTREX-TcAUK1 transfectants by the limit-dilution cloning method . Transgenic condition of several clones was confirmed by Southern and Western blot analyses . For the expression of the fusion protein TcAUK1-GFP , the coding sequence of TcAUK1 was amplified by PCR with primers TcAUK1-NcoI-fwd ( 5´-CCATGGTGAGTGCGGCGGAGGGCGGCCAA-3´ ) and TcAUK1-Rev-STOPLess-BamHI ( 5´-GGATCCGTTCTCCTTTCCGCCCGAGAAGTCC-3´ ) . The amplification product was sublconed into pGEM-T Easy vector , isolated by digestion with EcoRI and BamHI endonucelases and cloned into pTEX-eGFP-TEV-HA-EEF plasmid ( kindly donated by Dr . Leon A . Bouvier , Instituto de Investigaciones Biotecnológicas , IIB-INTECH ) digested with the same restriction enzymes . T . cruzi epimastigotes of CL Brener strain were electro-transfected with this construct and localization of fusion protein TcAUK1-GFP was evaluated after 48 h by fluorescence microscopy . T . cruzi epimastigote and trypomastigote forms were harvested by centrifugation , washed once with PBS and allowed to adhere to poly-L-lysine coated coverslips . Vero cells cultured in 24 wells plate with a sterile coverslip , were infected and at different days after infection , culture medium was removed and cells were washed once with PBS previous to further processing . Parasites and infected Vero cells were fixed in 4% paraformaldehyde and washed twice with PBS . After been permeabilized with 0 . 2% Triton-X100 in PBS ( PBT solution ) , cells were treated with blocking solution ( 1% BSA in PBS ) for 1 h at room temperature . For TcAUK1 and mitotic spindle double staining , epimastigote forms were first incubated with a mix of the rabbit antiserum to TcAUK1 ( 1:200 dilution , GenScript Corporation ) and the monoclonal mouse anti-β-tubulin KMX-1 antibody ( 1:400 dilution , Chemicon International ) and then incubated with a mix of goat anti-rabbit IgG Alexa Fluor 488-labeled ( 1:500 dilution , Invitrogen ) and goat anti-mouse IgG Alexa Fluor 594-labeled ( 1:500 dilution , Invitrogen ) . For TcAUK1 labeling in trypomastigote and amastigote , cells were first incubated with the rabbit antiserum to TcAUK1 and then with the secondary antibody goat anti-rabbit IgG Alexa Fluor 488-labeled . In all cases cells were incubated with the first antibody for 1 h at room temperature , then washed three times with PBT solution and finally incubated with the secondary antibody another hour at room temperature . After being washed three times with PBT solution , the slides were mounted in VectaShield mounting medium ( Vector Labs ) containing DAPI and examined with a fluorescence microscope ( model BX41 , Olympus ) . For actin filament staining , Vero cells were incubated for 10 min at room temperature with Rhodamine Phalloidin ( 1:1000 dilution , Invitrogen ) , washed and mounted as described previously for the secondary antibody . Images were processed with the ImageJ software [35] . Synchronization of epimastigote forms of T . cruzi in G1/S of the cell cycle was achieved using hydroxyurea ( HU ) . Cells in exponential growth phase were arrested by incubation with 15 mM of HU for 20–24 h and then released by washing twice with PBS and suspending the cells in culture medium . Cells continued to be cultured for 20 h; samples were taken at the indicated time points and processed as indicated . For flow cytometry analysis , 3x105 cells were harvested by centrifugation , washed with PBS-EDTA 2 mM and fixed in 70% ethanol at -20°C for 30 min . Then , they were washed once with PBS and suspended in staining solution ( 69 μM propidium iodide , 38 mM citrate buffer pH 7 . 40 , 0 . 2 mg . ml-1 RNase ) . The DNA content of propidium iodide-stained cells was analyzed with a fluorescence-activated cell sorting ( FACSAria II ) analytical flow cytometer ( BD Biosciences ) . Percentages of cells at different phases of the cell cycle were evaluated by Cyflogic software . For microscopic observation of cell cycle progression , synchronized cells were processed as described for the immunolocalization assay , using KMX-1 antibody .
The protein sequences of T . brucei Aurora kinase genes ( TbAUK1 , TbAUK2 and TbAUK3 ) were used as baits to search for orthologue sequences in the T . cruzi genome database ( http://tritrypdb . org/tritrypdb/ ) . As this database has been made from sequencing the genome of CL Brener strain , a hybrid [36 , 37] that arose from two different lineages ( Esmeraldo-like and Non-Esmeraldo-like haplotypes ) , we found several putative genes for each TbAUK . When TbAUK1 and TbAUK2 were used as query , two coding sequences ( CDS ) for each one were found , representing in both cases the same allele for the different haplotypes . In the case of TbAUK3 , three CDS were detected , two of them corresponding to the Esmeraldo-like and the other one from the non-Esmeraldo-like haplotype . After an exhaustive sequence analysis , we established that both CDS related to TbAUK1 codify for a single amino acid sequence , and the same result was found with the three CDS related to TbAUK3 . Nevertheless , the two CDS related to TbAUK2 showed conspicuous differences , including an insertion of 21 nucleotides . These identified sequences were named TcAUK1 , TcAUK2 and TcAUK3 , in accordance with the TbAUKs given names ( for details see Table 1 ) . In the case of TcAUK2 , the two protein sequences were named TcAUK2S ( short isoform ) and TcAUK2L ( long isoform with the 21 nucleotides insertion ) . Fig 1A shows the sequence alignment of TcAUK2S and TcAUK2L where the dissimilarities and the insertion of seven residues are highlighted . After establishing the CDS for each TcAUK , specific oligonucleotides were designed and used to amplify these genes , using genomic DNA of T . cruzi CL Brener strain as template . The obtained amplification products were then subcloned , sequenced and compared to nucleotide sequences found in the database . While sequenced TcAUK1 was identical to the gene found in the T . cruzi Genome Project database , TcAUK2 as well as TcAUK3 sequences showed some minor discrepancies . Particularly in the case of TcAUK2 , the sequencing reactions confirms the presence of the variants TcAUK2S and TcAUK2L in the genome of the parasite . Once the final sequence of each gene was determined , they were annotated in GenBank under the following accession numbers: TcAUK1 EU494590 . 1 , TcAUK2S EU494591 . 1 , TcAUK2L EU494592 . 1 , and TcAUK3 EU494593 . 1 . A series of multiple sequence alignments ( MSA ) were performed to analyze the detected Aurora kinase genes of T . cruzi . A MSA of TcAUK proteins with the catalytic domain of human Protein kinase A ( PKA ) showed that most amino acids with key roles in the kinase activity of PKA are conserved in TcAUKs ( S1 Fig ) . The most relevant is the presence of catalytic glutamic acid of PKA preceded by an Arginine , which is conserved in TcAUKs , allowing to classify them as members of the RD kinases group ( S1 Fig , indicated as ( 1 ) ) . In a second MSA , TcAUKs deduced protein sequences were aligned with Aurora kinase proteins from model metazoans . The four TcAUKs present the two most characteristic domains of Aurora proteins: the Activation loop ( DFGWSxxxxxxRxTxCGTxDYLPPE ) and the Destruction-box ( LLxxxPxxRxxLxxxxxHPW ) . The Threonine residue found in the highly conserved RxT motif from the Activation loop is phosphorylated in the active form of Aurora kinases enzymes ( Fig 1B ) . Described in detail in human Aurora A protein , the 3D active site of this family of proteins involves the Activation loop and the Glycine Rich Loop [38] . In between them is the Hinge region , which along with other residues , forms a hydrophobic pocket where the purine ring of the ATP’s adenosine substrate is located . TcAUKs proteins also show a strong conservation of the main residues implicated in the folding of this hydrophobic pocket . The 3D structure of human Aurora kinase A in complex with ATPγS reported by Nowakowsky and co-workers [39] reveals that in its active state , the αC helix in the protein adopts a position that allows a salt bridge formation between residues Glu181 and Lys162 . This salt bridge is essential for the catalytic activity of the enzyme , and it is very close to the β phosphate from de ATPγS ligand . The key residues are conserved in all the Aurora kinase proteins , including the TcAUKs . Finally , a phylogenetic analysis was carried out based on a third MSA of TcAUKs with Aurora kinases proteins of model metazoan and Aurora proteins described in other protozoans . The full-length amino acid sequences were aligned and a maximum parsimony tree was constructed ( Fig 1C ) . As expected , the three TcAUKs group with its orthologues in T . brucei . On the other hand , the longer C-terminal sequences for TcAUK3 and a presence of a large inserted region in the Activation Loop of TcAUK2S and TcAUK2L make these proteins to group separately from the Aurora kinases of other species . Similarly , when a MSA was performed considering only the catalytic domain ( S2 Fig ) of these proteins , again TcAUK3 and both TcAUK2 grouped apart from the proteins of others organisms . Thereby , is TcAUK1 the one closest related to Aurora proteins of other protozoan and metazoans . The complete analysis of the TcAUKs sequences show strong evidences that allow us to conclude that these genes most likely code for the T . cruzi aurora kinase functional orthologues . As described above , Aurora kinases from T . cruzi are represented in the database as more than one CDS . A detailed analysis of the information found in the database suggests that the CDS retrieved for TcAUK1 and TcAUK2 correspond to allelic variants of single copy genes . For TAUK3 the information in the database it is not conclusive about the number of CDSs and their corresponding genomic localization . A Southern blot analysis with specific probes for each TcAUK was performed on genomic DNA of T . cruzi CL Brener strain to experimentally determine the number of copies ( Fig 2A ) . When TcAUK1 or TcAUK2 probes were used , the results confirm that both are single copy genes . Notably , the pattern obtained with the TcAUKs probe on genomic DNA digested using PstI demonstrated the existence of the TcAUK2S and TcAUK2L variants . Endonuclease PstI has two recognition sites within the sequence of the TcAUK2S variant but none inside the TcAUK2L . Consistent with the presence of both variants of TcAUK2 , the total number of bands observed in the Southern blot were four: three corresponding to TcAUK2S and one for TcAUK2L ( Fig 2A , TcAUK2 panel , line Pst1 , asterisks ) . The Southern blot results obtained using a probe against TcAUK3 showed that when restriction enzymes without recognition sites within the protein coding sequence were used two bands were observed , indicating the presence of more than one copy of this gene . However , when restriction enzymes with one digestion site within the sequence were employed , there was a difference with the expected number of bands . The analysis of the CDS of TcAUK3 and their surrounding sequence obtained from TriTrypDB showed that TcAUK3 is present in three copies per haploid genome with two CDS in a tandem arrange on chromosome TcChr-33 and the third CDS in another contig without chromosome assignation . The tandem CDSs show high sequence identity on its flanking regions ( 98% identity ) , indicating the occurrence of a duplication event involving a large chromosomic region in which TcAUK3 is included . Given the difficult interpretation of this Southern blot results , Pulsed Field Gel Electrophoresis ( PFGE ) to separate intact chromosomes followed by hybridization with a TcAUK3 probe ( Fig 2B ) was carried out . Probes for TcAUK1 and TcAUK2 were included in this experiment to confirm the Southern blot results . While only one band was observed for TcAUK1 and TcAUK2 , supporting results that indicated these are single copy genes , two bands were detected for TcAUK3 , confirming the presence of more than one CDS located in different chromosomes . This result , together with the restriction profile observed in the Southern blot analysis , allows us to conclude that TcAUK3 gene is present in three copies per haploid genome of T . cruzi CL Brener strain . Additionally , in a recent publication [40] Dr . Robello and collaborators present the analysis of the genome sequences of two T . cruzi clones TCC ( TcVI ) and Dm28c ( TcI ) , determined by PacBio Single Molecular Real-Time technology . The assemblies obtained with this technology permitted accurately estimate gene copy numbers . Analyzing this improved genome sequence , it was possible to confirm the presence of three copies of TcAUK3 gene in Dm28c: two at scaffold 196 and one at scaffold 24 ( Dr . Robello , personal communication ) . After TcAUKs genes were identified , their expression through parasite life cycle was evaluated . The presence of the different TcAUK transcripts was evaluated by RT-PCR in epimastigote , trypomastigote and amastigote forms . Specific reverse primers for each TcAUK were used , while a primer corresponding to the Splice Leader ( SL ) region was used in all three cases . Amplification products were detected for each TcAUKs in all the three parasite forms ( Fig 3A ) . In order to confirm their identity all amplification products were subjected to sequencing reactions . Sequence data obtained from TcAUK2 RT-PCR products confirms that both variants of this gene–TcAUK2S and TcAUK2L–are transcribed . Furthermore , 5´UTR region shows 100% of identity between variants , supporting the hypothesis that these two forms correspond to alleles inherited from two different parental lineages . In the case of TcAUK3 , two specific amplification products of similar length ( Fig 3A , I and II ) were obtained . Sequencing showed that there a 100 bp stretch in the 5´UTR present in only one of the transcripts ( Fig 3B , TcAUK3 , II , underline ) , while the rest of the sequence had 100% identity with the shorter product ( Fig 3B , TcAUK3 ) . Thereby , the existence of two amplification products for TcAUK3 with differences in the 5´UTR agrees with what was observed in the Southern blot and the PFGE analysis , indicating the presence of more than one gene copy per haploid genome . The phylogenetic tree in Fig 1C suggests that TcAUK1 shows greater similarity to metazoan Aurora genes than the rest of the TcAUKs . Since TbAUK1 from T . brucei is regarded as the protozoan counterpart of human Aurora B [23] , we focused on unravelling the role of TcAUK1 in T . cruzi biology . Taking into account that in T . cruzi most of the regulation of protein expression occurs at the post-transcriptional level , the existence of TcAUK1 protein was evaluated by immunoblotting using a specific polyclonal antiserum against this protein . TcAUK1 protein was detected in whole cell extracts from epimastigotes , trypomastigotes and amastigotes ( Fig 3C ) , hence consistent with mRNA expression data . A main characteristic of Aurora kinase proteins is their dynamic redistribution during cell cycle . Human Aurora-B localization during mitosis has been well documented , at the beginning of the mitotic process it is dispersed in the nucleus , after which becomes concentrated in the nuclear midzone and migrates with chromatids to the cell poles , to finally locate in the constriction ring during cytokinesis . To address if TcAUK1 shares this dynamic localization during cell division , this protein was followed along cell cycle in epimastigote forms by immunostaining . DNA of the nucleus and kinetoplast were detected by DAPI staining while mitotic spindle microtubules were labeled using KMX-1 monoclonal antibody . This allowed to identify the different cell cycle stages in an asynchronous population based on the determination of the number of flagella ( F ) , nuclei ( N ) and kinetoplasts ( K ) , three organelles that duplicate in a tightly coordinate manner . In cells at interphase ( 1F1K1N ) , TcAUK1 was observed as a punctuated pattern localized in both extremes of the kinetoplast ( Fig 4A , upper panel ) . The same localization was observed at late G2 when cells have already duplicated their flagellum and kinetoplast ( 2F2K1N ) ( Fig 4A middle row panels ) . In these cells , the two kinetoplasts are arranged in an anteroposterior alignment indicating the beginning of kinetoplasts segregation . During mitosis ( 2F2K2N ) , when both kinetoplasts are moving in opposite directions , TcAUK1 was detected inside the nucleus . In cells where the nuclei have not segregated yet , TcAUK1 adopted a circular conformation ( Fig 4A lower row panels , yellow arrowhead ) , whereas in cells where both nuclei were segregating TcAUK1 was observed as an elongated structure ( Fig 4A lower row panels , white arrowheads ) . Furthermore , TcAUK1 adopted the same configuration as the mitotic spindle , as shown by microtubule staining , revealing a close association of TcAUK1 with this structure . To study TcAUK1 localization in the other forms of T . cruzi , trypomastigote and amastigote forms were collected from culture supernatant of infected Vero cells and subjected to immunostaining with TcAUK1 antiserum ( Fig 4B ) . In amastigotes , TcAUK1 appeared as a single focus close to the kinetoplast , whereas in trypomastigotes no TcAUK1 signal was detected . Given that amastigotes represent the intracellular replicative form of T . cruzi we prompted to study TcAUK1 localization in Trypanosoma cruzi during the progression of cell infections , using Vero as a host cell . In Fig 4C it can be observed that at day 2 after infection , both amastigotes and trypomastigotes were present inside the cytoplasm of Vero cells . Surprisingly , here TcAUK1 was detected not only in amastigotes but also in trypomastigotes and in both cases located inside the parasite nucleus ( Fig 4C , 2 days ) . At days 7 and 11 post-infection , only amastigotes could be observed inside Vero cells , and TcAUK1 was detected as two foci located one at each side of the kinetoplast , similar to what was observed in epimastigote forms during interphase ( Fig 4C , 7 days and 11 days ) . The above observations indicate that during the cell cycle TcAUK1 shows a changing localization . To accurate define each cell cycle phase , we synchronized epimastigote forms by hydroxyurea ( HU ) treatment . After release , samples were taken at different time points for flow cytometry and for determine TcAUK1 localization by immunofluorescence . Fig 5 shows that in G1 and S phases , when the cell has 1 nucleus and 1 kinetoplast , TcAUK1 was located at both extremes of the kinetoplast . At the moment that the kinetoplast has duplicated in G2 phase , TcAUK1 was still close to the kinetoplast but it moved out from the extremes and appeared as a single point . Finally , in M phase TcAUK1 was detected inside the dividing nucleus . According to what was observed in Fig 4A , and considering the role that Aurora proteins from other organisms play , during M phase TcAUK1 could be involved in mitotic spindle dynamics and chromatin segregation . Nevertheless , the localization of TcAUK1 in the proximity of the kinetoplast during interphase leads to ask about the role that it is playing in that location . The cell cycle stage dependent localization in epimastigotes of TcAUK1 is in accordance to the typical behavior of an Aurora B kinase , localizing inside the nucleus and appearing in association with the mitotic spindle during mitosis . Therefore , TcAUK1 could be fulfilling a possible role in spindle organization and chromosome segregation during cell division . The localization at each side of the kinetoplast during interphase is a novel observation that it has not been described in any other kinetoplastid . For this reason , to confirm this result with a different experimental approach , we generated epimastigote forms expressing a TcAUK1-GFP fusion protein by using the episomal , low-expression level vector pTEX . Fig 5B shows that TcAUK1-GFP presents a similar localization pattern to that observed by immunostaining . This novel localization could be reflecting roles of TcAUK1 in process other than nuclear division . In order to address the study of TcAUK1 possible roles , epimastigotes of T . cruzi overexpressing TcAUK1 were generated . Cells were transfected with the construction pTREX-TcAUK1 and stable overexpressing cells were selected under presence of G418 in the culture medium . After the selection period pTREX-TcAUK1 vector integration into the genomic DNA was confirmed by Southern blot analysis of the transfectant pool ( Fig 6A ) . Parasites transfected with an empty pTREX vector were used as a control . Individual clones were isolated by the limiting dilution method , and a single clone in which TcAUK1 overexpression was confirmed by western blot was used for subsequent studies ( Fig 6B ) . Considering that TcAUK1 seems to be involved in mitosis , we hypothesized its overexpression could alter cell cycle progression , modifying the culture growth rate . Three independent cultures of pTREX ( control ) and pTREX-TcAUK1 epimastigotes were initiated at 1x106 cells . ml-1 and counted for cell density every 24 h for six consecutive days . Fig 6C shows the growth curve for pTREX-TcAUK1 and pTREX epimastigotes cultures , where it can be observed that pTREX-TcAUK1 epimastigotes had slightly lower growth rate . When cell duplication time ( DT ) was calculated based on the specific growth rate ( μ ) between days two to five in each culture ( Fig 6C , inset ) , the mean of pTREX parasites was 20 . 6 ( SEM 0 . 2 ) hours whereas for TcAUK1 overexpressing epimastigotes was 25 . 4 ( SEM 0 . 895 ) hours ( statistical significance was determined by Student's Paired t-test , P<0 . 05 ) . In order to elucidate how cell cycle progression could be affected by TcAUK1 overexpression , hydroxyurea ( HU ) synchronized cultures were subjected to propidium iodide DNA staining followed by flow cytometry . For this , cultures of pTREX and pTREX-TcAUK1 cells were arrested at G1/S transition by incubating with hydroxyurea ( HU ) . After this treatment cells were released and samples were taken every two hours for cell cycle analysis . Results depicted as histograms in Fig 6D show that both cultures progressed similarly through G1 and S phases ( 0 to 8 h after HU removal ) , but when cells entered to G2/M ( 11 h after HU removal ) a delay was detected in TcAUK1 overexpressing cells when compared to pTREX epimastigotes . While at 13 h after-release from HU most of pTREX parasites completed mitosis ( 56% cells in G1 and 38% in G2/M ) , at this same time point pTREX-TcAUK1 parasites were still retained in G2/M phase ( 41% cells in G1 and 50% in G2/M ) . It was not still up to 14 hs post-release that most of pTREX-TcAUK1 cells complete mitosis ( 55% cells in G1 and 36% in G2/M ) , meaning that mitosis concretion in these cells was at least one hour retarded . Despite this observation , at 15 h after HU release , both cultures showed similar profiles in the cell cycle histogram . Thereby , the delay in the normal progression of the cell cycle in TcAUK1 overexpressing parasites supports the idea that TcAUK1 is involved in events occurring during cell division . A more detailed analysis of the progression through G2/M ( 11 h to 14 h after HU removal ) of synchronized control and TcAUK1 overexpressing cultures was performed ( Fig 7A ) . Considering the limited information available on T . cruzi mitosis and the wide differences between lab strains , we initially set to analyze the dynamic of organelle duplication in WT epimastigotes . Progression through G2/M is characterized by the organized and timed duplication of 1- the flagellum , 2- the kinetoplast and 3- the nucleus . According to our observations , once the new flagellum is synthesized and protrudes from the flagellar pocket , the kinetoplast starts to duplicate , first increasing its length ( Fig 7B , panel I DAPI ) and then acquiring a “V” shape with the concave side facing the base of the flagella ( Fig 7B , panel II DAPI ) . After that the kinetoplast “breaks” giving rise to the two daughter kinetoplasts ( Fig 7B , panel III DAPI ) , that localize one behind each other ( anteroposterior arrangement ) . Concurrently , the mitotic spindle starts to form in the nucleus , observed as a circular accumulation of β-tubulin ( Fig 7B , panel III KMX-1 ) . Next , while the new flagellum continues increasing its length , both kinetoplasts adopt a side by side configuration ( lateral arrangement ) and the mitotic spindle assembling begins ( polygonal structure ) in the nucleus ( Fig 7B , panel IV KMX-1 and DAPI ) . This last event marks the end of the G2 phase and the beginning of mitosis ( M ) . During this phase , the spindle elongates while the nucleus is divided in two ( Fig 7B , panel V KMX-1 and DAPI ) . At the end of the nuclear division , sister nuclei and kinetoplasts continue to migrate in opposite direction while cytokinesis begins with the cleavage furrow formation in the anterior edge of the cell ( Fig 7B , panel VI , VII and VIII KMX-1 and DAPI ) . As a result of this in-depth analysis of G2/M progression , two specific structural characteristics can be considered as landmarks: one is the length of the new flagellum in 2F1N1K cells and the other is the relative localization of the two kinetoplasts in 2F1N2K cells . The microscopic analysis of the above described events in pTREX and pTREX-TcAUK1 synchronized epimastigotes between 11 h and 14 h after HU removal showed that entry into mitosis was delayed in TcAUK1 overexpressing parasites when compared to control cultures ( Fig 7C ) . This is concluded from the observation that at 11 . 5 h after HU removal , control cultures showed a significant rise in the 1F1K1N population , indicating that most epimastigotes have undergone cell division . However , at this time point , most parasites in the overexpressing culture corresponded to the 2F1K1N subpopulation , with an augmentation of the 1F1K1N population only after the 12 . 5 h time point . The number of epimastigotes corresponding to the mitotic configurations 2F2K1N and 2F2K2N remains low throughout the analyzed period both in pTREX and pTREX-TcAUK1 cultures , indicating that this process occurs within a short time span . Moreover , this observation indicates that TcAUK1 overexpression does not alter mitotic events , in which case accumulation of the previously mentioned mitotic configurations would have been observed . With the aim to confirm the effect of TcAUK1 overexpression on kinetoplast replication , we defined a method that allowed us to time this event with considerable accuracy . As described above , the new flagellum protrusion from the flagellar pocket and the beginning of mitotic spindle assembling ( polygonal structure ) constitute two independent events that indicate the start and the end of kinetoplast replication , respectively . Considering this , 2F1N1K cells were analyzed for kinetoplast division and new flagellum length , while 2F1N2K cells were examined for mitotic spindle formation and the lateral arrangement of duplicated kinetoplasts . In pTREX epimastigotes , at every time point all cells that displayed the new flagellum protruding from the flagellar pocket also presented kinetoplasts in the process of duplication . However , in pTREX-TcAUK1 cultures , cells bearing a long new flagellum outside the flagellar pocket but with a kinetoplast that had not entered the duplication process could be observed ( Fig 7D , panel IV ) . Indeed at 12 . 5 h post-release , out of 53 2F1N1K cells , 47 had a protruding flagellum but only 22 showed a dividing kinetoplast . This structural configuration was detected in parasites from TcAUK1 overexpressing cultures obtained at every time point ( from 11 to 14 h post release ) . Cells with 2F1N2K configuration were evaluated for the kinetoplasts arrangement ( posterior or lateral ) as well as presence or absence of mitotic spindle polygonal structure . The total population of pTREX epimastigotes examined showed the presence of a polygonal structure of the spindle as well as a lateral arrange of the kinetoplasts , as expected according to what was described for WT cells . However , when 2F1N2K TcAUK1 overexpressing epimastigotes were analyzed , 50% of the cells that showed the polygonal structure of the mitotic spindle displayed kinetoplasts still placed in an anteroposterior arrangement at 12 . 5 h after HU was removed ( Fig 7D , panels III and V ) . This phenotype of 2F1N2K could observed at every time-point in pTREX-TcAUK1 epimastigotes . Moreover , an extreme phenotype of 2F1N1K cells with the polygonal structure of the mitotic spindle but a V-shaped kinetoplast was detected ( Fig 7D , panels I and II ) in epimastigotes overexpressing TcAUK1 . Taking into account that the function of Aurora kinases is closely related to their localization through the cell cycle , we next studied TcAUK1 dynamics in overexpressing epimastigotes during the different phases of cell cycle , considering kinetoplast and nucleus duplication and mitotic spindle assembly as hallmarks of this process . Interestingly , TcAUK1 in overexpressing cells was present almost exclusively in the nucleus during the entire cell cycle ( Fig 8 ) . This result differs from what was observed in WT epimastigotes , in which TcAUK1 localizes in both extremes of the kinetoplast during the interphase ( G1 and S phases ) and only is detected in the nucleus just before mitosis begins ( Fig 5 ) . This observation led us to hypothesize that , when overexpressed TcAUK1 loses its localization at the kinetoplast and this could be affecting replication process of this organelle .
In this work we identified three Aurora kinase genes present in T . cruzi genome: TcAUK1 , -2 and -3 . By analysis of genomic sequences , we found that TcAUK2 has two different forms differing only in its amino acid sequences by 3% , including a 7-residue insertion in the longer form of the gene ( TcAUK2L ) . By Southern blot and PFGE , we determined that TcAUK3 gene is represented by more than one copy per haploid genome . Moreover , cDNA sequencing shows that transcripts of these copies differ in their 5´-UTR . Considering all this data we postulate that TcAUKs family is not conformed only by 3 members , but 5 members form part of this group of genes . The presence of two different forms of TcAUK2 could indicate that these proteins have different and specific functions . In the case of TcAUK3 , considering that gene expression is mainly regulated at the post-transcriptional level in trypanosomatids the differences in the 5´ UTR of the different transcripts identified for these genes could denote differential regulation in their mRNA stability or translation efficiencies , being therefore subjected to selective protein expression regulation . These findings propose TcAUK2 and TcAUK3 as interesting proteins to be studied on the near future . TcAUK1 , which is present as a single copy gene , appears as the closest T . cruzi Aurora kinase to metazoans Aurora kinases , based on the phylogenetic analysis . Moreover , its orthologue in T . brucei ( TbAUK1 ) has been reported as the protozoan counterpart of mammalian Aurora B kinase [23] . These led us to focus our efforts on the in-depth characterization of TcAUK1 . After confirming the expression of TcAUK1 in the amastigote , trypomastigote and epimastigote stages ( Fig 3 ) , we next aimed to study its localization during epimastigotes cell cycle . By immunofluorescence , we detected TcAUK1 at two discrete cell cycle stage-dependent subcellular localizations: during interphase , it is localized at the extremes of the kinetoplast while in mitosis it resides inside the nucleus , associated with the mitotic spindle ( Fig 4A ) . Furthermore , we extended the localization analysis to trypomastigotes and amastigotes isolated from culture supernatant as well as in intracellular amastigotes . Although the result of the immunoblotting shows that this protein is expressed in trypomastigotes obtained from culture supernatants , TcAUK1 could not be detected in this form of the parasite by immunofluorescence . Nevertheless , when TcAUK1 was labeled in intracellular parasites we found trypomastigote forms that express this protein located in the nucleus ( Fig 4C , 2 days after infection ) . Therefore , it seems that TcAUK1 could be expressed in trypomastigotes at specific times during the infection cycle , probably indicating its participation in specific cellular processes . It is important to highlight that despite the trypomastigote , amastigote and epimastigote are the three forms that have been more deeply studied , during the differentiation that mediates the passage between these forms , the parasite adopts different intermediate stages . This involves a plethora of morphological events occurring in a dynamic way until the cell reaches its final form . According to what we have observed in the trypomastigote form , it is possible to postulate that TcAUK1 could be expressed for a short period of time and participates in events taking place during cell differentiation . Furthermore , the hypothesis of a short-lived TcAUK1 could explain why it can be detected when techniques that evaluate a wide population are used ( e . g . western blot ) but its detection remains elusive when observing individual events ( e . g . immunofluorescence ) . In amastigotes , similarly to what was observed for epimastigotes , TcAUK1 adopts two different locations . It is found inside the nucleus of intracellular amastigotes at two days after infection ( Fig 4C ) but it locates at the extremes of the kinetoplast in extracellular amastigotes and intracellular amastigotes at advanced days of infection ( Fig 4B and 4C ) . The fact that TcAUK1 adopts the same nuclear location in trypomastigotes than in amastigotes at early days of infections is another observation that suggests that this protein could be involved in the differentiation process . On the other hand , in actively replicating intracellular amastigotes ( advanced cellular infections ) and the epimastigote replicative form , the location of TcAUK1 at the extremes of the kinetoplast could indicate that this protein is involved in cell division processes . The role of TcAUK1 during cell division was studied in detail in the epimastigote form of T . cruzi . TcAUK1 localization alternates between kinetoplast extremes at interphase and the nucleus during mitosis . The arrangement that TcAUK1 adopts in the nucleus ( Fig 4A , 2F2K2N cells ) in close association with the mitotic spindle and migrating with segregating chromosomes strongly suggests that this protein is the orthologue of human Aurora B . The overexpression of TcAUK1 leads a delay in the duplication time of the epimastigote form , as observed in the growth curves of control and pTREX-TcAUK1 parasites ( Fig 6C ) . The results of cell cycle analysis by flow cytometry on synchronized pTREX and pTRX-TcAUK1 cultures showed that in the later , the time employed to conclude mitosis is longer than in control epimastigotes ( Figs 6D and 7A ) . This result confirms that TcAUK1 plays an important role in cell division . To further investigate this , we analyze organelle division dynamics during G2/M . The prevalence of 2F1K1N population at longer times after HU removal in TcAUK1 overexpressing cultures ( Fig 7C ) suggests that in these cells kinetoplast duplication is altered . By timing the sequential duplication of flagellum , kinetoplast and nucleus in WT cells ( Fig 7B ) , we noticed that overexpression of TcAUK1 leads to a delay in the start of kinetoplast duplication ( Fig 7D ) . Considering that the role of Aurora proteins is closely related to its localization , a possible explanation for the altered kinetoplast duplication is the fact that overexpressed TcAUK1 lost its location at the extremes of the kinetoplast during cell cycle interphase , as is detected in Fig 8 . The subcellular location of TcAUK1 in WT epimastigotes during interphase detected by immunofluorescence is very similar to the localization of the so-called Antipodal sites–a nucleation of proteins that are involved in different kinetoplast processes–in Kinetoplastids [18 , 41] . Hence , it is very likely that TcAUK1 interacts with different targets that participate in kinetoplast division at these sites . Further experiments should be performed to confirm this hypothesis . It has been widely documented that Aurora kinase B in metazoans is associated with other proteins to conform the so-called CPC . More recently it was found that T . brucei AUK1 ( TbAUK1 ) it is also associated with other proteins conforming a complex and that , like the CPC , this is crucial to guide TbAUK1 across different targets during cell division [25] . Nevertheless , the possible role of Aurora kinases in other cellular processes has not been explored yet . Here , we report for the first time that TcAUK1 is not only involved in mitosis but also in the duplication of the kinetoplast . It is well known that Aurora proteins cellular function is regulated by different post-translational modifications , such as protein phosphorylation and SUMOylation [42–45] . It is possible that TcAUK1 localization in the extremes of the kinetoplast is determined by its interaction with other proteins or by post-translational modifications , or a combination of these . The loss of its localization could be associated with the failure of one or both of these possibilities due to non-physiological expression levels . This hypothesis is supported by the observation that in western blot experiments , in whole cell extracts of pTREX-TcAUK1 epimastigotes only the lower molecular weight band increased its intensity , indicating that there are at least two different states for TcAUK1 ( probably corresponding to different post-translationally modified proteins ) but only one of them is augmented in overexpressing parasites ( Fig 6B ) . If TcAUK1 needs to be post-translationally modified to interact with specific proteins responsible in order to be recruited to the kinetoplast extremes , then it could be possible that unmodified overexpressed TcAUK1 interacts and sequesters specific proteins , therefore impeding the modified TcAUK1 to reach the kinetoplast . In summary , in this work we have identified the members of the Aurora kinase proteins in the protozoan T . cruzi . We have demonstrated that TcAUK1 protein behaves as the human Aurora kinase B regarding its role in nuclear division . Also , we have found that this protein is present in the three forms of the parasites life cycle and evidence shows that this protein could play a role in the differentiation between trypomastigote to amastigote forms . Interestingly , we have described a novel function for this protein in direct involvement in the initiation of kinetoplast duplication . This represents the first description of a role for Aurora kinases other than its widely studied function in mitosis . This stands as a valuable contribution in the attempt to understand in a molecular level the complex processes that take place during the life cycle of protozoan organisms . | The cell cycle is a complex and highly regulated cellular process in which different checkpoints are coordinated and a unique pattern of protein activities is present . In trypanosomatids , this process is even more complex because of the presence of the kinetoplast , a network of circular DNA inside a large mitochondrion that coordinates its division and segregation with the nuclear mitosis . Aurora kinases comprise a family of enzymes that regulate key steps during the cell cycle by varying their subcellular localization . In this work , we identified three Aurora kinase genes ( TcAUK1 , -2 and -3 ) in T . cruzi , the agent of Chagas disease , and established that TcAUK1 has the typical behavior of a mammalian Aurora B kinase , localizing inside the nucleus and associating with the mitotic spindle during mitosis . On the other hand , during interphase TcAUK1 , localizes at each side of the kinetoplast , showing a novel localization that has not been described before . Finally , here we present evidence for the relationship between the novel localization of TcAUK1 and its participation in kinetoplast division . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"parasitic",
"cell",
"cycles",
"enzymes",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"microbiology",
"enzymology",
"parasitic",
"protozoans",
"protozoan",
"life",
"cycles",
"parasitology",
"developmental",
"biology",
"trypomastigotes",
"protozoans",
... | 2019 | Aurora kinase protein family in Trypanosoma cruzi: Novel role of an AUK-B homologue in kinetoplast replication |
Regulatory networks often increase in complexity during evolution through gene duplication and divergence of component proteins . Two models that explain this increase in complexity are: 1 ) adaptive changes after gene duplication , such as resolution of adaptive conflicts , and 2 ) non-adaptive processes such as duplication , degeneration and complementation . Both of these models predict complementary changes in the retained duplicates , but they can be distinguished by direct fitness measurements in organisms with short generation times . Previously , it has been observed that repeated duplication of an essential protein in the spindle checkpoint pathway has occurred multiple times over the eukaryotic tree of life , leading to convergent protein domain organization in its duplicates . Here , we replace the paralog pair in S . cerevisiae with a single-copy protein from a species that did not undergo gene duplication . Surprisingly , using quantitative fitness measurements in laboratory conditions stressful for the spindle-checkpoint pathway , we find no evidence that reorganization of protein function after gene duplication is beneficial . We then reconstruct several evolutionary intermediates from the inferred ancestral network to the extant one , and find that , at the resolution of our assay , there exist stepwise mutational paths from the single protein to the divergent pair of extant proteins with no apparent fitness defects . Parallel evolution has been taken as strong evidence for natural selection , but our results suggest that even in these cases , reorganization of protein function after gene duplication may be explained by neutral processes .
Gene duplication and sub-/neo-functionalization are processes that both increase genomic complexity and are thought to be the major sources of genetic novelty in organisms [1] . Models seeking to explain the retention of paralog genes include relaxation of selection constraints leading to functional novelty through transcriptional changes , splicing divergence or functional repartitioning of protein domains . Particularly interesting are cases of repeated , parallel evolution of these paralog genes that has been taken to be strong evidence for natural selection ( several examples reviewed in [2] ) . In most model organisms , the spindle checkpoint pathway includes the paralogous Bub1 and Mad3 proteins . Surprisingly , although Bub1 and Mad3 are highly diverged , it was recently reported that this paralog pair has arisen from at least nine independent duplication events throughout the tree of life [3] . Perhaps even more striking is that in each event , the duplication almost always leads to a Bub1 homolog , containing a kinase domain and a Mad3 homolog containing a pseudo-substrate APC inhibitor motif [3] . This reorganization of protein function has therefore been thought to be adaptive , leading to a more complex spindle checkpoint pathway [3 , 4] . However , theoretical work suggests that a causal link between increased genetic complexity and adaptation may not be as clear as commonly assumed [5 , 6] . For example , if degeneration and complementation of the ancestral bi-functional protein must always lead to Bub1 and Mad3 division of function ( i . e . , kinase function separated from the pseudo-substrate inhibitor motif ) , then the repeated observation of Bub1 and Mad3 protein organization might be due to a rate of degeneration that is higher than the rate of reconstitution . Whether sub-functionalization of a bi-functional protein is adaptive or neutral cannot be easily distinguished by sequence analysis alone . Both models predict complementary conservation of function . However , precise quantitative fitness measurements in organisms with short generation time , such as yeasts , can be used to address questions about these two models of sequence evolution ( adaptive vs neutral ) . Using comparative approaches that delineate functional regions of proteins , we found that this gene duplication leads to sequence signatures that indicate repartitioning of the ancestral function in the extant paralogs . To test whether this increase in complexity ( from a pathway containing a bi-functional gene to a pathway containing two specialized genes ) in the spindle checkpoint is adaptive , we employ high-throughput quantitative fitness measurements in laboratory conditions stressful for the spindle checkpoint pathway and we systematically dissect the reorganization in protein architecture of an ‘ancestral’ bi-functional gene . To our surprise , we could not detect any increase in fitness for the stepwise functional reorganization ( duplication , degeneration and complementation ( DDC ) ) .
Preservation of duplicate genes as explained by the duplication-degeneration-complementation model is a neutral process by which repartitioning of the functions of the ancestral protein occurs within the two extant duplicates [7] . Although sub-functionalization is a neutral process , it can lead to adaptation in the case of adaptive conflicts: mutations that are precluded from occurring in the ‘ancestral’ gene but are adaptive when the functions of the ancestral gene are separated . This resolution of adaptive conflicts can sometimes explain the fitness benefit of mutations that ‘specialize’ multifunctional proteins [8] . Nevertheless , complementation by functional repartitioning or complete functional preservation in one of the two copies is still expected in the cases where mutations leading to neo-functionalization have occurred . It has previously been shown that Bub1 and Mad3 protein reorganization had occurred several times over the eukaryotic tree of life , leading to similar sequence profiles ( Fig 1A , and see [3] ) . To obtain an amino acid resolution view of the evolution of the paralogous proteins , we performed sequence analysis of the duplication that occurred in the whole-genome duplication of budding yeasts ( [9] , see Methods ) . Briefly , a phylogenetic hidden Markov model is used to identify short or large regions of conservation relative to the flanking amino acid sequence . This algorithm is applied to single copy proteins from species that diverged before the duplication to identify short motifs or domains that are under selective constraints . We then map these regions to the duplicates and test whether there is evidence of relaxation in constraints in the clade post-duplication using likelihood ratio tests . This analysis revealed several protein regions , in addition to the KEN boxes and kinase domain identified in previous studies [3] , that have repartitioned from the ancestral sequence to the paralogs ( Fig 1B ) . This more detailed view of the changes in constraints allowed us to propose why some of the changes were correlated ( S1A Fig for an example ) . For example , Mad3 is known to interact with Cdc20 via its N-terminal KEN box motif and its tetracopeptide repeat domain ( TPR ) [10] . According to 3D structural information of Mad3 ( S1B Fig ) , specific residues in the TPR domain ( shown in yellow in S1B Fig ) appear to contact the KEN box ( shown in red in S1B Fig ) and may stabilize the interaction of Mad3 with Cdc20 ( S1C Fig ) [11] . Bub1 , on the other hand , does not possess the N-terminal KEN box but still requires the TPR for proper function [12] . We can detect changes in constraints on these specific residues of the TPR of Bub1 , suggesting that the degeneration of residues in the TPR domain or the KEN box will disrupt the same function ( binding to Cdc20 ) and that loss of selection constraints on that function will lead to degeneration of both . Indeed , the same correlated changes in constraints exemplified here are also observed in the mammalian , drosophila , and fission yeast Bub1 and Mad3 , which occurred through independent gene duplication events ( S1D Fig ) . The changes in constraints at these specific residues are unlikely to disrupt the other functions of the TPR . On the other hand , the pattern of evolution on the ABBA motifs in the yeast paralogs is more complicated . The ABBA motif is known to be required for binding to Cdc20 [13] , and indeed , both Bub1 and Mad3 retain at least one copy of the ABBA motif ( Fig 1B ) . Nevertheless , we can clearly detect motif turnover in the first ABBA motif in Bub1 ( Fig 1B ) . This suggests two possible explanations: 1 ) the two ABBA motifs serve the same function and the loss of the first motif was compensated in the Bub1 protein by another motif , 2 ) the two ABBA motifs may bind Cdc20 for different functional reasons and the Bub1 protein underwent a single change in constraint on this motif . Since we have not identified a newly conserved motif in the Bub1 protein ( which could compensate for the loss of the first ABBA motif ) , we believe that the second possibility is more parsimonious . Consistent with important functions for the domains identified in the ancestral fungal protein , there were no conserved regions that were lost in both paralogs ( Fig 1B ) . The repartitioning of functional regions suggests that the duplication lead to sub-functionalization . We therefore sought to experimentally verify that sub-functionalization had occurred in the paralog pair . As a proxy for the ‘ancestral gene’ , we obtained the gene from Lachancea kluyveri which diverged prior to the whole-genome duplication event ( see Discussion ) . We refer to this gene as the ‘single-copy protein ( SCP ) ’ , and transformed it into S . cerevisiae . We first assessed the localization of the single-copy protein because it was known that Bub1 and Mad3 localize to different subcellular compartments: Bub1 is localized to the kinetochores in a Mps1 dependent manner during specific phases of the cell-cycle [14] and Mad3 is constitutively localized in the nucleus [15] . If Bub1 and Mad3 were products of a sub-functionalization event , we would also expect the L . kluyveri protein to localize to both subcellular compartments . To test this , we tagged the three proteins with GFP and visualized their localization by fluorescence microscopy . To account for possible differences in growth or imaging conditions that may influence the apparent localization of the proteins , we designed an assay where the tagged single-copy protein could be visualized alongside the S . cerevisiae tagged protein in the same field of view using the identical GFP tag . Briefly , we expressed additional cytoplasmic fluorescent proteins of other colors ( non-GFP ) that allow us to differentiate which cells carried the L . kluyveri protein fusion or the S . cerevisiae fusion and obtained fluorescent micrographs of mixed cultures ( see Methods ) . As expected , we observed a distinct localization pattern for the single-copy protein , which was quantitatively different from either Bub1 or Mad3 ( Fig 2A ) . Upon closer inspection , we found that the localization of the single-copy protein appears to be a mixture of the localization patterns of Bub1 and Mad3 . Consistent with this , when ordering by bud size as marker of cell stage , we observed that the L . kluyveri protein localized constitutively to the nucleus , and also showed a punctate localization within the nucleus in the early cell-cycle ( Fig 2B ) . Because Bub1 and Mad3 are required for the spindle-checkpoint pathway , we tested whether the single-copy protein could rescue defects of cells lacking Bub1 or Mad3 with respect to the functionality of the pathway . To do so , we took advantage of a strain carrying a non-essential chromosome containing the ochre suppressor SUP11 [16] . This strain forms red-pigmented colonies upon loss of this chromosome due to the presence of the ade2-101 allele . As has been reported previously [17] , we found that strains without a functional Bub1 gene have a very strong increase in chromosome loss rate ( 29/65 vs 3/131 sectored colonies , p-value < 0 . 0001 ) . However , we were unable to detect an increase in chromosome loss rate for strains lacking a functional Mad3 gene ( 5/177 vs 3/131 sectored colonies ) . Nevertheless , cells lacking Bub1 and Mad3 , but carrying the single-copy gene driven by the Bub1 promoter in this strain completely abolished the Bub1 chromosome segregation fidelity defect ( 3/207 vs 3/131 sectored colonies , Fig 3A ) . We next sought to verify if the single-copy protein could rescue the growth defects of impaired cell-cycle functions of cells lacking Bub1 and Mad3 . Cells lacking Bub1 or Mad3 are highly sensitive to benomyl , a microtubule destabilizing drug [18] , because cells fail to correctly detect mitotic spindle attachments . If the single-copy protein can perform the functions of both Bub1 and Mad3 , we expect that the single-copy protein would rescue the fitness defect of cells lacking Bub1 or Mad3 . If the phenotype is not fully rescued , it may indicate neo-functionalization and adaptation in the Bub1 or Mad3 protein ( or it may be due to an artifact of expressing a heterologous gene ) . To test this , we performed spot dilution assays and found that , if expressing the single-copy protein , cells lacking Bub1 and Mad3 can grow in the presence of benomyl with similar growth characteristics to wild-type S . cerevisiae cells ( Fig 3B ) . We could not simply test the function of the spindle checkpoint in other species as we observed that yeasts other than S . cerevisiae were highly resistant to benomyl ( S2 Fig ) . At the limits of the resolution of this plate assay ( we estimate that this assay can detect at the minimum only a 5% growth difference ) , and taken together with the localization data , these results suggest that the single copy protein can rescue Bub1 and Mad3 function in S . cerevisiae . Consistent with the DDC model , the sub-functionalization of the Bub1 and Mad3 proteins may confer no fitness advantage to S . cerevisiae . It is possible that the spot dilution assay did not have enough resolution to detect the fitness advantage of the reorganization of protein function in the paralogs . To address this , we designed a method to more precisely quantify relative selection coefficients ( s , see Methods ) . Briefly , our assay is a competitive fitness assay that takes advantage of flow-cytometry to provide relative counts of fluorescently labeled cells within a growing population [19] and this assay can be performed in a high-throughput fashion where the combination of alleles of interest are tested in cellular backgrounds expressing different fluorescent proteins as experimental replication . Replicate strains can be rapidly created by the SGA cloning strategy ( see Methods and [20] , S3 Fig ) . To determine the resolution limit of this fitness assay , we generated 16 spores from a cross between a strain carrying a wild-type Bub1 allele marked with CaURA3 ( the URA3 gene from Candida albicans , which complements the URA3 gene from S . cerevisiae ) and a query strain ( containing SGA mating type reporters ) expressing a green or a red fluorescent protein ( 8 spores for each cross ) . These spores were competed to form 64 fitness assays . We reasoned that if any additional single nucleotide polymorphisms ( SNPs ) between the query strains and parental strain of our mutant arrays had a detectable fitness effect ( but masked due to epistasis within their respective backgrounds ) , or if these SNPs showed non-transitive fitness effects , they would be uncovered within these 16 spores . We compared the relative proportions of cells expressing the green fluorescent protein and red fluorescent protein at the 20th generation to the 40th generation and we calculated the selection coefficient for each competition ( see Methods ) . Because our strains are supposedly genetically identical ( except for the possibility of non-shared SNPs ) , we expect an average selection coefficient of zero and the standard deviation obtained from this test can be used to estimate the resolution of our assay ( deviations due to growth conditions or to the non-shared SNPs ) . We measured the selection coefficients of the wild-type strains and observed a mean selection coefficient of 0 . 00069 , with a standard deviation of 0 . 0017 ( Fig 4A ) . This indicates that the resolution of our assay is in the order of s = 0 . 0033 ( 1 . 96 times the standard deviation ) and we believe this represents the difference in growth rate that we can detect . To account for other possible variations that may occur during the course of the study ( changes in media , etc ) we therefore chose to report as deleterious/beneficial any differences in fitness where both replicates of a competition exceeded a selection coefficient with an absolute value of 0 . 005 or greater while remaining consistent with all other competitions . We next compared strains carrying deletions of Bub1 or Mad3 , or single-copy protein rescues of these deletions to wild-type strains and found that cells lacking Bub1 have a strong fitness defect when growing on benomyl while cells lacking Mad3 have a more moderate defect ( s<-0 . 3 and s = -0 . 01317 respectively , Fig 4A ) . Remarkably , cells lacking both Bub1 and Mad3 , but expressing the single-copy protein at the Bub1 locus , retain wild-type growth rate when challenged with the same benomyl concentration ( s = 0 . 002 , Fig 4A ) . Thus , even at the much higher resolution of these quantitative competition experiments , we find no evidence that the duplication and reorganization of the Bub1 and Mad3 proteins ( Fig 1 ) confers a fitness benefit . The experiments above show that there is apparently no fitness advantage to the Bub1/Mad3 protein organization relative to the single-copy protein . However , the real evolutionary trajectory almost certainly did not replace the single-copy protein directly with the fully formed Bub1 and Mad3 . Instead , a series of evolutionary events ( likely separated by millions of years ) probably occurred from the single-copy protein to the extant paralogs . It is possible that certain steps along this path , particularly at the beginning after the duplication , were advantageous and drove the reorganization of the paralogs . Because there is insufficient phylogenetic resolution to infer the exact order and number of mutations , based on previous knowledge of the functional elements found in Bub1 and Mad3 ( Fig 1 ) , we created strains with genetic make-up of possible evolutionary intermediates that correspond to stepwise mutational events during protein function reorganization . The evolutionary paths assayed include mutations at multiple loci , and therefore possible paths were created using several rounds of the SGA cloning strategy ( see Methods ) . Briefly , we sought to test single and double mutations of the KEN boxes on the SCP at the BUB1 locus , loss of kinase of the SCP at the MAD3 locus , non-functionalization of the gene , or “evolution” to the extant protein . These mutations represent key evolutionary steps from the ancestral protein to the extant Bub1 and Mad3 ( Fig 5 ) . We introduced these mutations into the SCP and cloned them individually into different starting strains ( for example , we generated a strain containing the SCP with a mutated KEN box at the BUB1 locus ) . To combine them , query strains carrying the SGA markers and different fluorophores were crossed to the library in an ordered array and selected such that the final products of several rounds of mating were otherwise genetically identical haploid spores carrying different combinations of marked alleles . We then performed an all-by-all competitive fitness assay , and found that genotypes cluster in three distinct fitness classes , which correspond to cells lacking Mad3 function ( Δmad3-like , with an average selection coefficient of -0 . 015 relative to wild-type ) , cells lacking Bub1 function ( Δbub1-like , with an average selection of <-0 . 3 relative to wild-type ( fitness effects larger than -0 . 3 are not measurable in this assay and so we consider these genotypes to have fitnesses <-0 . 3 , see Methods ) , or cells with wild-type phenotype ( WT-like , with an average selection coefficient of 0 . 0026 relative to wild-type ) ( Fig 4B ) . Although we have not explored the fitness landscape exhaustively , this sample of genotypes suggests that it is made up of three distinct plateaus . To assay possible paths through the fitness landscape that lead to sub-functionalization , we focused on genotypes that differed from each other by one ‘evolutionary step’ . For example , we consider the initial duplication as one evolutionary step ( such that we compared the fitness difference of a strain containing one vs two SCP ) , as well as individual losses of functional motifs . Although in principle it is possible for a non-functional genotype to revert , we consider these events to be rare and therefore have not considered them for simplicity . The results of our analysis are displayed in Fig 5 , as an evolutionary landscape with multiple evolutionary paths that ‘travel’ through possible intermediate genotypes . Interestingly , we were able to find at least one path without a detectable fitness defect consisting of at least three degenerations ( Fig 5 , path through white nodes ) . If we only consider paths that do not go through nodes of fitness defects ( i . e . we do not allow crossing of fitness valleys ) , then the analysis suggests that the extant network in S . cerevisiae is an absorbing state ( see Discussion ) . Although we cannot be certain that evolution has taken any of the paths studied here , that we can find at least one seemingly neutral evolutionary path , at the resolution of our assay , strongly supports the DDC hypothesis that the reorganization of protein domains in the Bub1/Mad3 paralogs can be explained by neutral degenerative mutations ( Fig 6 and see Discussion ) . This model also predicts that the initial neutral step in the process ( the duplication ) is reversible , and that the mutations must be relatively common to explain the frequency at which the reorganization occurs during evolution ( Fig 1 ) . Consistent with this , we have identified at least one phylogenetic clade where the gene duplication in the ancestor of Saccharomyces reverted to the single-copy functional homolog: Vanderwaltozyma polysporus retains a single-copy protein with all the functional elements of Bub1 and Mad3 even though it diverged after the whole-genome duplication ( S4 Fig ) . It is estimated that the ancestor to the lineages leading to V . polysporus and S . cerevisiae had already non-functionalized ~20% of the duplicates [21] suggesting that neither sub-functionalization of Bub1/Mad3 and non-functionalization were rapid .
Neutral processes , such as described by the DDC model [7] , have been shown to be important in increasing genomic complexity ( see for example [22] for a study on non-adaptive increase in interactome complexity ) . At the limit of the resolution of our assay ( discussed below ) , we find no evidence that the Bub1/Mad3 protein reorganization after duplication in budding yeast provides a route for adaptive conflict resolution and therefore , further work is necessary to find a fitness advantage over a single-copy protein in the spindle checkpoint pathway . Consistent with the DDC model , none of the evolutionary intermediates that we consider functional through the comparative genomics analysis showed a fitness defect when tested under laboratory conditions requiring functional spindle checkpoint and when driven by the BUB1 promoter . Nevertheless , there are several caveats to our experiment . First , the resolution of our assay meant that we could only detect fitness effects in the range of s = 0 . 005 . Because of the population size of budding yeasts in nature [23] and our estimated effective population size during the experiment ( see Methods ) , selection could be efficient even on undetected differences in fitness ( given a high enough recombination rate ) . Therefore , it remains possible that the sub-functionalization to Bub1/Mad3 is truly adaptive . However , even if we assume a very small beneficial effect that was not detected , because we performed our assay in the presence of high concentration of benomyl ( which was used to characterize all the components of the spindle checkpoint pathway [24] ) , we believe that under reasonable growth conditions the adaptive effect of this sub-functionalization would be even smaller . Another possibility is that Bub1 and Mad3 participate in completely orthogonal molecular processes to the spindle checkpoint such that our assayed environment would not be able to detect the real functional differences between the duplicate genes and the ancestral single-copy protein . Yet a third possibility is that there exists one context where the effects of the duplication are under much stronger selection . However , if this context exists , it must be rare due to the finding that the phylogenetic clade leading to V . polysporus reverted to the single-copy gene . Ultimately , it is not possible to know the environmental context , nor the genetic context of the ancestral yeast and we cannot rule out that adaptation by escape of an adaptive conflict drove the sub-functionalization of the ancestral protein . Despite these caveats , we note that the functionally relevant motifs in Bub1 and Mad3 were identified in the context of the spindle checkpoint pathway , and this pathway is activated every cell division to ensure proper chromosomal attachment prior to anaphase . We therefore believe that large effects would have been captured even in this laboratory environment . Although in our study all functional evolutionary paths lead to the same genotype through the same number of degenerations , the probability that a path is taken is dependent on the rate of mutation , the selection coefficient of the intermediates and the effective population size [25] . We here discuss only the scenario of large population size because small population sizes would allow all genotypes , including any potential neo-functionalization , to be effectively neutral . When population size is large and mutation rate is low , then crossing a fitness valley ( as measured by our quantitative fitness assay ) requires the population to fix each intermediate genotype ( this is the deleterious sequential fixation regime discussed in [25] ) . In this large population size and low mutation rate regime , paths through deeper fitness valleys are essentially never taken because selection is very effective . In this regime , the relative frequencies of the effectively neutral states are entirely dependent on the mutational rate between these states [7] . Because the mutational rate to degenerate is likely to be higher than the mutational rate to re-create a functional element , the evolutionary outcome of Bub1/Mad3 homologs after sub-functionalization could be nearly deterministic . Therefore , the relative probability of observing Bub1/Mad3 functional homologs after duplication is equal to the rate of sub-functionalization divided by the rate of non-functionalization [7] . We propose that the rate of sub-functionalization can be high due to the very small number and mechanistically simple degenerations: we showed here that sub-functionalization and degeneration of the Bub1/Mad3 protein can occur within only two mutations following the gene duplication ( generation of stop codon to remove the kinase function in Mad3 , and a shift in start position to remove the first KEN box in Bub1 ) , both of which have been observed frequently over evolution in other genes [26 , 27] . If this sub-functionalization is truly neutral , we propose that the repeated reorganization of the Bub1/Mad3 homologs may be due to the fact that no other possible outcome of the duplication can be easily observed ( it is an absorbing state ) , such that the observed genotype is surrounded by fitness valleys or reversion to the single-copy protein . Fixation of the re-organization may have occurred through hitchhiking with other beneficial mutations , a scenario that often occurs with large population size and a high mutation rate [28] , or due to changes in gene expression ( discussed below ) . Previous studies have considered the functional implications of evolution of individual enzymes and protein complexes through gene duplication and divergence ( e . g . [29 , 30] ) . However , many proteins that function within regulatory networks contain complex multi-domain architectures and disordered regions [31] . In the case of Bub1 and Mad3 , numerous functional steps occurred after gene duplication , and it is not possible to reconstruct them at the resolution of single amino acid substitution . Nevertheless , we chose several key evolutionary steps during the functional reorganization and assessed the fitness of those intermediates . We believe that our study represents a practical way forward for studies of the evolution of complex eukaryotic proteins . Unlike other previous studies ( e . g . [29] ) , we have not performed ancestral gene resurrection via gene synthesis , but instead we have chosen a gene from another species which we believe is representative of the ancestral allele . Our approach has several advantages . First , it is more likely that the gene is functional in at least one genetic environment . Second , the reconstruction of the ancestral gene might not be accurate in proteins with highly diverged disordered regions . Finally , the gene has evolved for the same period of time as the duplicate genes , providing a direct test of whether adaptation from the ancestral allele is due to the resolution of an adaptive conflict . Similar experimental designs have been used to assay functional differences in whole transcriptional regulatory networks that happen on relatively shorter divergence times [32] . Transcriptional evolution has been shown to be an important aspect of functional divergence after gene duplication [33–35] . An important caveat of our study is that we have not tested the evolution at the promoter , due to the difficulty in finding the functional elements of the promoter region [36] . However , we do have evidence that the Mad3 promoter and the Bub1 promoter are not functionally equivalent as the single-copy protein does not fully rescue the spindle-checkpoint defects when placed at the MAD3 locus ( S5 Fig ) . Those promoter changes , however , are still consistent with a model of degeneration where the Bub1 promoter is similar to the ancestral SCP promoter . On the other hand , it is also possible that after duplication the Bub1 promoter acquires beneficial mutations relative to the ancestral promoter , and these changes drive fixation of the gene duplication . Thus , although there is clear evidence that the promoters of Mad3 and Bub1 have diverged , we did not test the effects of these changes . Under our tested conditions , our study shows that there is no evidence that the reorganization of protein function was to escape an ‘adaptive conflict’ for the spindle checkpoint pathway in mitotic cells of S . cerevisiae . Our data is consistent with the DDC model and our study suggests that parallel evolution through degenerative processes does not have to be rare or adaptive . Further work will be required to see if this is also the case in other organisms where this reorganization has been observed . This situation is reminiscent to the convergent evolution of holocentric chromosomes across the tree of life [37] , and it is still unclear whether it provides an advantage during growth , especially considering the more complex meiotic segregation of chromosomes [38] . Interestingly , although the core spindle checkpoint pathway is conserved in all eukaryotic life , several other differences exist in this pathway [39] . These differences include non-conserved proteins important for spindle checkpoint function ( such as p31 ) or different copy number of paralogous proteins ( such as Cdc20 in human ) . Our study provides experimental techniques to test the step-wise effects of evolutionary changes that have been detected through comparative genomics on multiple loci ( such as the ones in the spindle checkpoint ) . We anticipate that these sensitive quantitative fitness measurements will be useful in the computational modeling of sequence and protein evolution within the context of a complete regulatory network [40 , 41] , as has been performed on other important regulatory networks [42 , 43] .
All strains were derived from either BY4741 or BY4742 using standard yeast genetic techniques or synthetic gene arrays ( see next method subsection ) . The single-copy gene was PCR amplified from purified genomic DNA ( Fermentas , #K0512 ) of L . kluyveri ( NRRL Y-12651 ) . All integrations were verified by PCR , and key strains containing the single-copy gene were verified for absence of Bub1 or Mad3 when relevant . Strain construction for the library of alleles was performed using the same method as described in [44] . SGA query strains were created by transferring the Ste3pr_LEU2 marker from Y8205 into the CAN1 locus of BY4742 . Fluorophores with the Ste2pr_LkHIS3 were cloned into the pAN200a plasmid ( based on pFA6a [45] ) using standard cloning techniques and transformed into the CAN1pr locus using delitto perfetto [46] . Benomyl ( 10mg/mL DMSO stock ) is used at outlined concentrations and added to boiling-hot media until completely dissolved . 5-fluoroorotic acid ( 5-FOA , 100mg/mL DMSO stock ) was used to select against uracil biosynthesis prototrophs [47] and plates were poured at 1g/L 5-FOA final concentration ( supplemented with all amino acids , including 72ug/mL uracil ) . Geneticin ( G418 ) was used at 200ug/mL to select for geneticin resistance . Cells are grown according to slight modifications to the protocols outlined in [48] as shown as a schematic in S3 Fig . We modified our query strains to have the following cassette integrated at the CAN1 locus: RPL39pr_fluorophore_Ste2pr_LkHIS3_Ste3pr_LEU2 . Fluorophores used for our study were yeast-enhanced monomeric green fluorescent protein ( ymEGFP ) and yeast-mCherry ( ymCherry ) . For the general construction of our strains , a query strain is first crossed with all the desired alleles at a particular locus . Diploid selection is performed by selecting for complementary auxotrophies . Overnight diploid cells from plate patches are then scraped into liquid sporulation media ( 1% Potassium acetate , 0 . 005% Zinc acetate ) and supplemented with amino acid requirements for diploid strains at 25% of the normal usage and incubated on a roller wheel for three days at room temperature . Usually , about 30% sporulation is observed and 5ul of the mixture is spread or spotted on selection plates that select for MATα and other selection markers such as auxotrophies and drug markers . We found that modifications to the germination and outgrowth procedure were necessary in our hands to obtain colonies after the SGA procedure . In our hands , addition of lysine to the media greatly enhances the initial outgrowth of spores for all strains that were constructed using our query strains ( even the ones that did not express a fluorophore or were lysine prototrophs ) . Spore outgrowth was normal for standard SGA query strains , indicating that strain specific variation , or heterozygous lys2 deletion strains or homozygous LYP1 affected the outgrowth of our strains ( LYS2/Δlys2 LYP1/LYP1 compared with LYS2/LYS2 LYP1/Δlyp1 ) . Therefore , to select for lysine prototrophs , the colonies are replicated to media lacking lysine only after the initial growth . To select for lysine auxotrophy , replica plating was used to isolate colonies that did not grow on media lacking lysine , however alpha-aminoadipate could be used instead [49] . The final strains used in the competitive fitness assay all had the following genotypes: MATα , can1::RPL39pr_fluorophore_Ste2pr_LkHIS3_Ste3pr_LEU2 , bub1::allele::CaURA3MX , mad3::allele::KanMX , Δlys2 , Δhis3 , Δura3 , Δleu2 . Protein sequences used for the comparative analyses were from the Yeast Gene Order Browser [50] . These proteins were then aligned with MAFFT [51] . Genomic sequences for BY4741 and BY4742 were obtained from the Saccharomyces Genome Database [52] . To perform our comparative analyses , protein sequences were analyzed using methods described in [9] and by visual inspection . To test for changes in constraints in the duplicate protein , we first predict regions of conservation in the proteins pre-duplication using a phylogenetic hidden Markov model that detects regions that have significantly lower evolutionary rates relative to their flanking region [53] . Having predicted these conserved regions , we can then map them to the duplicate protein and ask whether two rates of evolution ( one for the pre-duplication clade , and one post-duplication ) better explain the evolution of the selected region as opposed to a single rate . Because the rate of evolution of predicted conserved regions in the ancestral protein is very low , when two rates of evolution better explain the data , it typically implies that the region under purifying selection prior to the whole-genome duplication is now under relaxed constraints after the whole-genome duplication . Partitioning of these losses in selection constraints in the two paralogs is an indication of sub-functionalization . To identify potential new motifs in the post-duplication proteins , the phylogenetic hidden Markov model can be used on the post-duplicate protein and the same analysis for changes in constraints can be performed . X-ray crystallography files ( 3ESL: Bub1 [54] and 4AEZ: Mad3 [11] ) were obtained from the Protein Data Bank [55] and analyzed using PyMOL [56] . Chromosome loss rate was measured in the classical strain carrying a linear non-essential chromosome in the W303 background [16] . Briefly , strains carrying the ochre allele ade2-101 mutation exhibit a visible red pigment when growing in media with low adenine supplementation . This phenotype is suppressed in strains with a single-copy functional SUP11 , which is present in the artificial chromosome . Therefore , chromosome loss rate can be measured by counting the number of red-sectored colonies in agar plates containing low adenine . This assay is performed by plating about 200 colonies , and we report the number of sectored colonies over the number of white colonies after two days of growth . Quantitative fitness assays were performed using the MACSQuant VYB ( Miltenyi Biotec Inc . ) . Briefly , strains are grown for 48 hours in 5mL of cultures on a rolling wheel . The competitive fitness experiment is started by mixing relatively equal proportion of green cells and red cells in deep 96-well blocks ( 100ul of a single ymCherry expressing strain and 100ul of a single ymEGFP expressing strain into 600ul distilled water ) at a 4-fold dilution . To obtain further dilutions , 20ul of the competition is diluted into 300ul distilled water to form a 16 fold dilution , then 20ul of this dilution is diluted into 300ul of defined media supplemented with amino acids and 100ug/mL ampicillin to form a final 16-fold dilution . The cells are therefore diluted 1024-fold and this operation is performed every 24 hours . Given a conservative estimate of 2*108 yeast cells per mL at saturation , we estimate an effective population size ( Ne ) of approximately 3 . 44 * 105 . Each screen competes 8 genotypes against the others ( 64 wells ) , with an additional 16 wells used as contamination control . We use the diagonal of the competitions as an additional negative control as these were competing genetically identical strains constructed independently with different fluorophores . Competitive fitness assays were performed in synthetic media with 10ug/mL benomyl , which is a condition at which wild-type cells still undergo ten divisions per day but seriously impairs growth of cells lacking spindle checkpoint function . We defined the relative selection coefficient ( s ) on the basis of the deterministic continuous time model of logistic growth of an allele against another: dR/dt = sRG , where R and G are the frequencies of red and green cells in the population , t the number of generations , and s the selection coefficient [57] . To simplify the parameter inference , we ignore drift ( because the timescale of the experiment is very short compared to Ne ) , mutational processes that happen during the fitness assay ( i . e . we ignore mutations occurring in red or green cells that may alter the lineage trajectories ) , recombination , and any non-transitive effects . Constraining on G = 1-R , this model describes the logistic equation and has solution [57]: R ( t ) =R ( 0 ) est ( 1−R ( 0 ) ) +R ( 0 ) est Because we do not observe the frequencies directly , we estimate the parameters s ( and R ( 0 ) if needed ) by counting each cell sampled by the flow-cytometer as a random variable , taking values of red or green ( as it is described for a binomial logistic regression ) . In total , 50 000 cells are analyzed at the 20th and the 40th generation for each competition experiment and well-defined proportions of green and red cells are gated to remove doublets and relative cell counts are obtained from each competition [58 , 59] . The likelihood function for this counting process is simply: L=∏t=1Tnt ! rt ! ( nt−rt ) ! R ( t ) rt ( 1−R ( t ) ) nt−rt Where t are the time points , r the number of red cells counted at that time point , n the total number of cells counted at that time point and R ( t ) the expected frequency of red cells at time point t . We estimate the initial frequency R ( 0 ) and s ( due to their relationship with R ( t ) ) by maximizing the likelihood , and the maximum likelihood estimate for the selection coefficient can be obtained by performing a log-linear regression on the ratio of red and green cells in the population: log ( R ( t ) G ( t ) ) =log ( R ( 0 ) G ( 0 ) ) +st For more than two time points , the parameters can be estimated by an iterative reweighted least-squares linear regression or with Newton’s method . For two time points , the solution is equivalent to the simple linear regression and we use the same formula as in [59]: log ( r ( t2 ) g ( t2 ) ) −log ( r ( t1 ) g ( t1 ) ) t2−t1=s Clearly , if s is positive , then the ratio of red cells to green cells increases at the next generation and the value of the selection coefficient is therefore the selective effect of a beneficial allele as in [60] . The upper limit of detection occurs when we expect fewer than 1 out of 50000 cells of the worst genotype at the 40th generation , which occurs at approximately s = -0 . 3 . In practice , some genotypes can no longer be detected at the 20th generation , and we simply report these as s < -0 . 3 . In some other cases , the number of red or green cells is fewer than 50 , which leads to highly inaccurate estimates of the selection coefficient and we also report these as s < -0 . 3 . When the total number of cells counted for both strains in a competition was fewer than 50 , we reported s = 0 to mean equally lethal . All genotypes were made in strains expressing red or green fluorescent proteins , and therefore assayed in two biological replicates . Key genotypes with undetectable fitness effects were further assayed with at least two technical replicates ( where the same strains were assayed more than once on different days ) . Cells were grown in low-fluorescence media with appropriate auxotrophic requirements to log-phase and imaged using a Leica SP8 confocal microscope . Proteins of interest were tagged with EGFP [61] , while the cytoplasm of cells were marked with cytoplasmic mCherry or mTagBFP2 [62] under a constitutive ribosomal promoter . We imaged the green fluorescence first , followed by the cytoplasmic marker to prevent bleaching . This setup enables highly controlled and quantitative analysis of localization because strains with differently tagged proteins can be imaged on the same field of view under identical conditions . Double-blind quantification of the localization pattern was performed by manually inspecting the green fluorescence localization pattern first and scoring for whole nucleus , kinetochore ( punctae ) , or a mixture of both . The scored cells were then assigned their proper genotype by looking at the red or blue fluorescence channels . | Parallel evolution of protein domain organization following gene duplication has been demonstrated in the spindle checkpoint pathway leading to the hypothesis that this organization is likely to be adaptive . We test this hypothesis by reconstructing budding yeast strains with a spindle checkpoint pathway containing a protein with ancestral domain organization , and systematically perform stepwise duplication , degeneration and complementation of the duplicated protein . We show that , under laboratory conditions where the spindle checkpoint pathway is necessary for growth , degeneration of the ancestral pathway organization to the extant sub-functionalized proteins is consistent with a neutral model of duplication-degeneration-complementation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"green",
"fluorescent",
"protein",
"fungi",
"luminescent",
"proteins",
"model",
"organisms",
"fungal",
"evolution",
"experimental",
"organism",
"systems",
"sequence",
"motif",
"analysis",
"saccharomyces",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
... | 2017 | Parallel reorganization of protein function in the spindle checkpoint pathway through evolutionary paths in the fitness landscape that appear neutral in laboratory experiments |
The formation of species in the absence of geographic barriers ( i . e . sympatric speciation ) remains one of the most controversial topics in evolutionary biology . While theoretical models have shown that this most extreme case of primary divergence-with-gene-flow is possible , only a handful of accepted empirical examples exist . And even for the most convincing examples uncertainties remain; complex histories of isolation and secondary contact can make species falsely appear to have originated by sympatric speciation . This alternative scenario is notoriously difficult to rule out . Midas cichlids inhabiting small and remote crater lakes in Nicaragua are traditionally considered to be one of the best examples of sympatric speciation and lend themselves to test the different evolutionary scenarios that could lead to apparent sympatric speciation since the system is relatively small and the source populations known . Here we reconstruct the evolutionary history of two small-scale radiations of Midas cichlids inhabiting crater lakes Apoyo and Xiloá through a comprehensive genomic data set . We find no signs of differential admixture of any of the sympatric species in the respective radiations . Together with coalescent simulations of different demographic models our results support a scenario of speciation that was initiated in sympatry and does not result from secondary contact of already partly diverged populations . Furthermore , several species seem to have diverged simultaneously , making Midas cichlids an empirical example of multispecies outcomes of sympatric speciation . Importantly , however , the demographic models strongly support an admixture event from the source population into both crater lakes shortly before the onset of the radiations within the lakes . This opens the possibility that the formation of reproductive barriers involved in sympatric speciation was facilitated by genetic variants that evolved in a period of isolation between the initial founding population and the secondary migrants that came from the same source population . Thus , the exact mechanisms by which these species arose might be different from what had been thought before .
Understanding how populations can diverge and become distinct species in the presence of gene flow is a central objective in evolutionary biology [1–3] . That gene flow poses a problem for speciation has for long been known [4–6] . Gene flow and recombination homogenize the genomes of diverging populations and break down associations of loci relevant for ecological adaptations and assortative mating; a condition usually required for speciation [7–9] . Yet , a growing body of research has shown that speciation can progress in the presence of gene flow [2 , 10–13] . Without a good understanding of the populations’ past it is , however , often difficult to distinguish between primary divergence-with-gene-flow and the sorting out of already partly diverged populations after secondary contact [3 , 14] . This distinction is important as the latter involves a period of geographic isolation in which the abovementioned problem of gene flow and recombination does not arise [2 , 15] . The evolution of reproductive incompatibilities in geographic isolation ( allopatry ) is well understood and not controversial , while primary divergence-with-gene-flow in the absence of strong geographic barriers demands other explanations [16] . From a population genetic perspective , the most extreme case of primary divergence-with-gene-flow is sympatric speciation [17] . In a biogeographic sense , sympatric speciation can be broadly defined as speciation in the complete absence of geographic ( external ) barriers [18] . The two definitions are not always in concordance [19–21] , but the ultimate question that relates both and motivates the study of sympatric speciation is whether and to what extent speciation requires the mediating effects of a period of geographic isolation . In other words , is geographic isolation necessary to reduce gene flow and initiate population divergence in the first place or can speciation commence in a panmictic population ? Thus , sympatric speciation has for long attracted theoreticians and empiricists alike , not because it is believed to occur frequently , but because—being the endpoint of the continuum of primary divergence-with-gene-flow—it may be particularly informative on the ecological conditions and evolutionary mechanisms that can lead to speciation in the presence of gene flow [19 , 22 , 23] . While theoretical models have shown that sympatric speciation is possible [8 , 22 , 24–26] , only few convincing empirical case studies have been published [reviewed in ref . 1 , 22] . And even in some of these cases critics remained doubtful [27 , 28] . This is partly due to the fact that speciation with geographic isolation is generally considered much more plausible , almost like a null hypothesis in speciation . Sympatric speciation appears thus not only to be rare , but also hard to demonstrate empirically . In their seminal book Coyne and Orr [16] proposed four criteria that have to be fulfilled to demonstrate that sympatric speciation is the most likely mode of speciation: ( i ) sympatric distribution of contemporary species , ( ii ) genetically-based reproductive isolation , ( iii ) phylogenetic sister relationship , and ( iv ) no historic phase of geographic isolation . Several cases are in concordance with some of these criteria , but almost none unambiguously fit all four [18 , 22] . Particularly the latter two criteria are inherently difficult to address and demonstrate . This is because a sister relationships between species ( criterion iii ) must reflect a true lineage bifurcation event and not simply result from a close genetic relationship due to secondary gene flow of evolutionarily more distantly related taxa . Especially inferences based on mitochondrial DNA alone are prone to error due to haplotype replacement [29–31] , but nuclear markers can lead to false inferences too , if gene flow and incomplete lineage sorting are not accounted for [32 , 33] . Further , demonstrating that a past allopatric phase of currently sympatrically occurring true sister species is unlikely ( criterion iv ) , can be difficult to do in practice . Following [34] the problem is that essentially three different scenarios can be imagined that would be consistent with the first three but differ in the fourth of Coyne and Orr’s criteria for sympatric speciation: ( 1 ) sympatric speciation after a single colonization , ( 2 ) sympatric speciation after several colonizations from the same ancestral lineage and the putative formation of a hybrid swarm , and ( 3 ) speciation after secondary contact and introgressive hybridization . The first scenario can be considered the ‘purest’ form of sympatric speciation in which reproductive barriers arise completely in sympatry . In the second scenario some of the genetic variation later involved in reproductive isolation could have evolved in the time of separation of the primary founder population and the secondary migrants . Importantly , these genetic variants would not immediately lead to divergence , but be absorbed into the gene pool—potentially leading to a hybrid swarm—and only later be recruited in the speciation process [35 , 36] . Speciation in this scenario could still be considered sympatric as population divergence happened in sympatry [34]; yet there is a role of geographic isolation if the admixture event was essential for speciation in sympatry . In the third scenario an initial level of ( incomplete ) divergence between the species evolved in geographic isolation , which would be strengthened by reinforcement [37] and/or ecological character displacement [38] upon secondary contact . The first two scenarios predict equal levels of shared ancestry with outgroups and the source population and no signs of differential admixture ( i . e . varying levels of admixture proportions ) among sympatric species within a radiation , whereas the latter scenario of secondary contact predicts varying levels of shared outgroups ancestry and signs of differential admixture [34] . Distinguishing between these three scenarios is especially difficult if the source population is not known or extinct; an issue that leads to lingering doubts in even the otherwise most convincing cases of sympatric speciation [39] . The attainability of big genomic data sets as well as theoretical and methodological advances in recent years have , however , markedly increased the power to investigate more complex demographic scenarios of secondary gene flow , admixture , and multiple colonization events [40–42] , thereby permitting to now infer if periods of geographical isolation were involved in putative cases of primary divergence-with-gene-flow and sympatric speciation . In this regard , recent evidence for a complex pattern of secondary gene flow and unequal shared outgroup ancestry of sympatric species of Cameroonian crater lake cichlids [34] , has shed some new light on this traditionally considered prime example of sympatric speciation [43] . Crater lake cichlids in Nicaragua , belonging to the Midas cichlid species complex ( Amphilophus sp . ) , represent a similar system in which fish from the two old and great lakes Managua and Nicaragua have repeatedly colonized small and isolated crater lakes [44] . The two great lakes are both inhabited by two species of Midas cichlids: A . citrinellus is a generalist species which presumably resembles the ancestral state and A . labiatus is adapted to feeding on invertebrates in rocky crevices with its characteristic hypertrophied lips and narrow head shape [45 , 46] . While most crater lakes harbor only one ( yet often polymorphic ) population of Midas cichlids , in two of the crater lakes , Lake Apoyo and L . Xiloá , several endemic species have been described [44] . According to the current taxonomy Crater Lake Apoyo harbors six [47] and L . Xiloá four species of Midas cichlids [48] . The species differ in their ecology and , notably , in both crater lakes a species with an elongated body shape inhabiting the open water niche ( from here on referred to as ‘limnetic’ as compared to the high-bodied and shore-associated ‘benthic’ species ) has evolved independently [49] . The small size of the crater lakes , the fact that they are surrounded by steep crater walls and no water connections exists , and the complete endemism of Midas cichlid species suggested sympatric speciation to be the most parsimonious scenario . And indeed , genetic data supported the monophyly of Midas cichlids in L . Apoyo [50] . Yet , this first study was criticized because the different benthic species inhabiting L . Apoyo were not considered separately and only one of the species , A . citrinellus , from the source L . Nicaragua was considered in certain analyses [27] . Furthermore , the different species in L . Apoyo were not equidistant to the source population in genetic space as might be expected after sympatric speciation . Thus , according to the critics , the null hypothesis of multiple colonizations and introgressive hybridization could not be ruled out completely [27] . Later studies taking several or all six described species into account and using different genetic markers concluded sometimes in favor of monophyly of the L . Apoyo flock and thus sympatric speciation [49 , 51 , 52] and sometimes not [53] . In addition the assignment of individuals to the proposed six-species taxonomy did not match in many cases [49 , 53] . Generally , L . Xiloá has been less in the focus of the debate around sympatric speciation , probably because its crater rim on the Eastern side is shallow and gene flow via intermittent direct water connections or vectors ( e . g . birds ) seems much more plausible than in the older , deeper and much more obviously isolated Crater Lake Apoyo . Nonetheless , also L . Xiloá’s species flock appears to be monophyletic [49 , 52] and appears to have resulted from a single founder event [54] . But , a comprehensive investigation of the plausibility of sympatric speciation in L . Xiloá has never been done . In addition to the questions of monophyly and sympatric speciation there have been discrepancies in the inferred order of speciation events based on different markers and types of analyses [49 , 52] . Most importantly , none of the abovementioned studies did explicitly take admixture between lakes , intralacustrine gene flow , and population size changes into account . Nonetheless , Midas cichlids still feature as one of the most prominent examples of sympatric speciation [18] . In this study we use genome-level analyses and demographic modeling in a coalescent framework to reconstruct the evolutionary history of the two parallel radiations of Midas cichlids in L . Apoyo and L . Xiloá using a comprehensive RADseq data set . More specifically we address all major points of previous criticism and more recent doubts concerning sympatric speciation in Midas cichlids [27 , 34] . To this end , we take all described species of Midas cichlids in the source and crater lakes into account and objectively assign individuals to genetic clusters to then ( i ) test for signs of unequal shared outgroup ancestry and differential admixture of sympatric species , ( ii ) establish the evolutionary relationships among species , and ( iii ) infer the demographic history of the two radiations to evaluate the evidence for primary divergence-with-gene-flow with or without secondary colonizations or secondary contact as outlined in the three scenarios of putative sympatric speciation above .
Previous studies of Midas cichlids had been partially hampered by difficulties concerning the taxonomic classifications . Thus as a first objective we investigated the population structure in our comprehensive data set . We were interested in both signs of genetic exchange and relationships among lake populations as well as population structure and individual ancestry within crater lakes . To this end , using Principal Component Analyses ( PCAs ) [55] and Admixture [56] , we first performed a ‘global’ analysis including all 446 individuals from the two great lakes and the crater lakes and then performed two ‘intralacustrine’ analyses focusing on each of the crater lakes separately . The first two principal components of the global PCA were highly significant ( p-value ~ 0 ) and clearly separated the four lake populations ( Fig 1 ) . In concordance with the geographic proximity and the assumed colonization history , the genetic cluster of L . Apoyo was closer to L . Nicaragua and L . Xiloá was closer to L . Managua , while the two great lake populations were in close proximity in the two-dimensional genetic space . Interestingly , two distinct genetic clusters could be identified for L . Xiloá , one being slightly closer to L . Managua than the other one . Individuals in this cluster corresponded exclusively to the two species A . amarillo and A . viridis . This presumably closer affiliation of these two species to the source population was also apparent in the global Admixture analysis , albeit , and importantly , only when assuming a priori the same number of clusters as lakes ( K = 4 ) ( S1A Fig ) . Considering all lakes , the highest support was found for nine ( K = 9 ) or twelve ( K = 12 ) clusters; the cross-validation error was almost equally low for the two runs ( S1B Fig ) . In the case of twelve clusters , four of the clusters corresponded to the two species A . citrinellus and A . labiatus in each L . Managua and L . Nicaragua while individuals from L . Xiloá and L . Apoyo were assigned to four different clusters each ( S1A Fig ) . Notably , there were no signs of admixture between the lake populations anymore . In the intralacustrine Admixture analysis of L . Apoyo the occurrence of four and five clusters had the highest support ( S2 Fig ) . Yet , 19 individuals , which are of strongly admixed ancestry in the case of four clusters ( S1 Fig ) , formed a distinct cluster in the case of five clusters ( Fig 2C ) . Five distinct clusters were also apparent in the PCA ( Fig 2A ) . Thus , our set of samples from L . Apoyo seemed to be best described by five genetic clusters . The main axis of variation ( PC1 ) clearly differentiated the limnetic A . zaliosus from the other four clusters . However , the delineation of the benthic individuals into the four different genetic clusters did in many cases not fit their species assignment based on morphology . Only in the case of A . astorquii were all individuals unambiguously assigned to one genetic cluster ( cluster 2 ) , albeit individuals from other species were included in this cluster as well . Since we think that the genetic clusters provide a more objective grouping of individuals than the sometimes difficult assignment based on morphology , we recoded benthic individuals as belonging to ‘clusters 2–5’ according to their genetic signature ( S1 Table ) . Note that from here on we will essentially adopt a genetic cluster species concept [57] and use the terms species and cluster interchangeably . Furthermore , a few individuals from all genetic clusters exhibited signs of admixed ancestry . In L . Xiloá three to four clusters had the highest support ( S2 Fig ) and four genetic clusters corresponded well to the four described species ( Fig 2B ) . Only in seven out of 123 cases ( three A . amarillo specimens assigned to A . viridis and four A . sagittae assigned to A . xiloaensis ) did the species assignment not match ( S1 Table ) and individuals were re-assigned . However , the admixture plot also revealed a substantial amount of hybridization; eighteen individuals exhibited varying degrees of admixed ancestry between A . sagittae and A . xiloaensis , two A . viridis showed signs of admixture with A . sagittae , and one A . amarillo with A . viridis . The same pattern was also apparent in a plot of the first three eigenvectors of a PCA ( Fig 2B ) . Putative hybrids are expected to occupy positions in genetic space along fictive lines connecting the species clusters [55] . Ten individuals in the center of this hybrid group , exhibiting more than 25% admixture proportions ( on average 43% ) , were re-labeled as belonging to a ‘hybrid’ group and considered separately or excluded from all subsequent analyses . Our rationale for this was that the inclusion of such a number of obviously admixed individuals ( also based on morphology , see below ) might have had a strong impact on the phylogenetic and demographic analyses and in many cases it would have been difficult to decide to which species they should be assigned to . To further investigate the occurrence of hybridization within crater lakes we performed morphological analyses . Indeed , individuals in the hybrid group exhibited an intermediate morphology ( S3 and S4 Figs ) . Thus , in both crater lake radiations we find evidence for distinct genetic clusters , yet also signs for ongoing gene flow . Pairwise levels of overall genetic differentiation among all species in the four lakes are provided in S2 Table . Patterns of genome-wide differentiation across the 24 linkage groups and among all sympatric species within the two crater lake radiations are visualized in S5 and S6 Figs . Analyses to detect loci putatively under divergent selection are described in S1 Text and detected outlier loci are given in S3 Table . The occurrence of two clusters in L . Xiloá in the global PCA , one being closer to L . Managua , would be consistent with two waves of colonization followed by introgressive hybridization . However , clustering methods do not explicitly take the demographic history into account and can thus sometimes falsely indicate admixture [58] . Thus we performed formal tests of admixture using f3-statistics [59] . f3-statistics are conceptually related to D-statistics ( ABBA-BABA tests ) and f4-statistics [60] , and readily interpreted: a test population is compared to two reference populations and a significant negative value provides evidence that the test population experienced some form of admixture from populations related ( or ancestral ) to both reference populations . If the two species A . amarillo and A . viridis , which appear closer to L . Managua in the PCA—or more accurately their ancestral population–resulted from secondary contact and subsequent introgressive hybridization with the already established crater lake population , tests including one of these two species as a test population and one of the other two species from L . Xiloá together with a species from the source lake as reference populations may be expected to yield significant negative f3-statistics . Yet , none of the tests with this constellation returned a significant negative value ( Table 1 ) . In fact , we performed the test among all 1 , 092 possible three-population combinations ( considering all populations and lakes in our data set ) and only three tests returned a negative score , and none of those turned out to be significant . Thus , the f3-statistics do not provide evidence for secondary contact followed by introgressive hybridization . We note , however , that a history of admixture will not always result in negative f3-statistics , especially if the test population has experienced a lot of population-specific drift [60 , 61] . We further note that tests based on the f3-statistics would not be able to detect an admixture event ( secondary colonization ) that occurred before the sympatric species diverged as the test and reference populations of the crater lakes would share equal proportions of admixed genotypes . Another way to investigate possible admixture events is by placing migration edges on a phylogenetic tree and evaluating whether they improve the fit of the model ( tree ) by reducing deviations in the residual covariance matrix: positive residuals indicate populations that exhibit observed covariances that are higher than accounted for by the model [61] . We used Treemix to build a tree and placed up to four migration edges ( m ) on it . The tree without migration ( m = 0 ) provided already a relatively good fit to our data: the fraction of variance explained in the observed covariance matrix by the tree ( “f” according to [61] ) was 99 . 7% . Importantly , no stark positive residual covariances between any of the source populations and any of the crater lake species was apparent ( S7 Fig ) . Adding four migration edges ( m = 4 ) improved the fit of the tree slightly ( f = 99 . 9% ) . The first three putative migration edges were placed between sympatric species within the two crater lakes ( S7C , S7E and S7G Fig ) and the fourth one between the ancestor of all L . Xiloá species and A . viridis from L . Xiloá itself ( S7I Fig ) . The latter migration edge is difficult to interpret as we would expect secondary gene flow from the source population or a related species into a crater lake species to be reflected by a migration edge coming from the lineage leading to the two species in the respective source lake , but not from its own ancestral lineage . We note that we are not aware of any closely related species that could have hybridized with a Midas cichlid species in the last few thousand years . Furthermore , the small increase in fit provided by the fourth migration edge does not come from a decrease of positive residual covariances between A . viridis ( or any other species in L . Xiloá ) and the source populations—the fit is already good without any migration ( S7B Fig ) . Instead , it seems to improve the fit of the relationships among species within L . Xiloá ( S7H and S7J Fig ) . Thus , rather than indicating secondary gene flow from the source ( or a related ) population into A . viridis we think this migration edge rather reflects the difficulty of fitting the evolutionary relationships of the species within L . Xiloá in a bifurcating tree ( even with migration ) : the topology within L . Xiloá is not robust and when fitting three migration edges A . viridis , and not A . amarillo , is the first species to split ( S7G Fig ) . That some of the divergence events do not adhere to a strict bifurcating manner was also supported by other phylogenetic analyses that we performed ( see below ) . In any case , the f3-statistics did not provide evidence for differential admixture of A . viridis and this putative migration edge is thus not significant: the statistical support of migration edges in Treemix have to be considered with caution and three- or four-population tests are recommended as formal tests of admixture [61] . We stress that we used Treemix in an explorative approach , but refer readers to the f3-statistics for formal tests of differential admixture . Given the high fit of the model with four migration edges and the fact that the highest scaled residual covariance between any two populations was very low with less than 1 . 5 SE , we did not attempt to fit more than four migration edges . Monophyly of the two radiations was strongly supported ( 100% bootstrap support ) . Also the phylogenetic sister relationship of L . Managua and L . Xiloá , providing evidence for the former being the source of the latter , was found in a 100% of bootstrap replicates . Interestingly , apart from the node grouping A . sagittae , A . xiloaensis , and the hybrid group in L . Xiloá ( 100% bootstrap support ) , the branching order within the radiations was not well supported ( bootstrap support ranged from 59 . 8%–86 . 2% ) . The low bootstrap support of nodes within the crater lake radiations in our Treemix tree led us to further investigate the evolutionary relationships among the sympatric species . To this end , we first built phylogenetic trees using SNAPP , which is explicitly designed to handle biallelic markers such as SNPs and employs the multispecies coalescent [62] . SNAPP returns a sample of species trees , which can be visualized in a “cloudogram” . Due to the computational burden and since we were only interested in the topology as well as relative branching times within the two radiations we built two separate trees , as their respective monophyly was strongly supported . In both trees it was evident that the two species from the source lakes are sister species and are equally distantly related to the crater lakes radiations ( Fig 3 ) . Within L . Apoyo the cloudogram indicated an almost starlike topology with extremely short internal nodes ( Fig 3A ) . The overall consensus ( root canal ) suggested that A . zaliosus diverged first followed by a split of cluster 2–3 from cluster 4–5 . Yet , every possible topology within the radiation was represented by some trees . In total 196 different consensus trees were found ( differing in topology and divergence time ) . Also for the species that are endemic to L . Xiloá the cloudogram indicated a simultaneous split of three species ( Fig 3B ) . A . amarillo , A . viridis and the ancestral population of A . sagittae and A . xiloaensis seemed to have split at the same time followed by the split of the latter two . Interestingly the hybrid group did not take an intermediate position between A . sagittae and A . xiloaensis , but formed the sister group to A . xiloaensis in all trees . The first three consensus trees ( nine in total ) covered 35% , 30% , and 27% of the individual trees and supported either a sister relationship of A . amarillo and A . viridis , an earlier split of A . amarillo , or an earlier split of A . viridis , respectively . Due to the computational demand of this method the phylogenetic trees were limited to only four individuals per species and a subset of loci [63 , 64] . To evaluate whether the phylogenetic results might be influenced by using only few individuals and excluding missing data [65] , we built individual-based phylogenetic split networks including all samples and more markers allowing for missing data ( see Methods for details ) . For both radiations they revealed essentially an identical pattern ( S8 Fig ) . In L . Apoyo all species seemed to diverge simultaneously , whereas in L . Xiloá there was one split between A . amarillo , A . viridis , and the ancestor of A . sagittae and A . xiloaensis . The hybrid group occupied an intermediate position between the latter two species , which is expected considering that the networks were based on genetic distance . In both analyses the two species in the source lakes were almost not distinguishable and were equally distantly related to the crater lake radiations . The fact that the great lake species were almost not distinguishable is probably due to the fact that the networks were based on genetic distance only . Overall genetic differentiation between the great lake species was very low ( S2 Table ) —presumably due to their relatively large effective population sizes—leading to a low resolution in the networks . In the SNAPP analyses differences in effective population sizes were taken into account and the two species appeared probably therefore clearly diverged in the SNAPP trees in contrast to the networks . A limitation of the described phylogenetic methods is that they do not take gene flow and changing population sizes into account . Moreover , the f3-statistics may not detect admixture events that happened before the split of the sympatric species . To overcome these limitations and furthermore infer the demographic history of the radiations we used fastsimcoal2 to perform coalescent simulations in pre-defined models and evaluated their fit against our empirical data summarized in the multi-dimensional site frequency spectrum ( SFS ) [66 , 67] . To better account for the complexity of multi-population models , we started with one-population models for both species in both great lakes ( the source populations ) . For each of the four populations six different models were tested ( S9A Fig ) . A model incorporating a sudden reduction in population size in the past followed by exponential growth until the present ( ‘bottlegrowth’ ) had the highest support in all four populations ( S4 Table ) . Since a signal of recent population expansion could be driven by rare alleles resulting from sequencing and genotyping error we repeated the analyses for A . citrinellus from Nicaragua using only genotype calls that were based on at least 15x coverage . Importantly , the ‘bottelgrowth’ model was again the most supported one ( S4 Table ) . Next , we tested each crater lakes species together with A . citrinellus from the respective great lake as a source population in two-population models ( S9B Fig ) . We used A . citrinellus because it presumably resembles the ancestral state of fish in this species complex and because fish with hypertrophied lips , resembling A . labiatus , are not present in L . Apoyo and are extremely rare in L . Xiloá [44] . Moreover , our phylogenetic analyses suggest that both species in the source lakes are equally distantly related to the crater lake radiations ( Fig 3 ) . We tested between nine and eleven models for each species and the same class of model ( differing only in migration ) was supported for all species ( S5 Table ) . This model included ( i ) exponential growth after a population bottleneck in the source population ( ‘bottlegrowth’ ) , ( ii ) divergence followed by ( iii ) exponential growth in the crater lake species , and ( iv ) an admixture event from the source into the crater lake . In the case of L . Apoyo gene flow between the lakes was not supported , whereas in L . Xiloá migration from the source improved the fit of the model . However , the relative statistical support for the different models with or without migration is not very different ( S5 Table ) . Thus , our data strongly support the population size changes and the admixture event , but we have rather low power to distinguish the different migration scenarios . For all species a model in which the colonization event happened after ( in forward time ) the bottleneck in the source populations was superior to a model in which we forced the colonization to happen before the bottleneck ( S5 Table ) . This could indicate a limitation of the inference method , as a bottleneck in the source could lead to a loss of information and bias lineages to coalesce before ( backwards in time ) the bottleneck [68] . In order to test this , we simulated data using the maximum likelihood parameter estimates and data structure of cluster 2 in L . Apoyo ( i . e . using the same number and length of loci ) but added 10 , 000 generations to the divergence time . Importantly , we were able to infer the correct ( i . e . simulated ) divergence time in this case ( S6 Table ) . This suggests that theoretically we have enough power to correctly infer divergence times that happened before the bottleneck in the source populations . Finally we analyzed the demographic history in five-population models . Even though the f3-statistics did not provide evidence for secondary contact we wanted to make use of the likelihood framework to explicitly evaluate the evidence for the two main competing hypotheses: sympatric speciation ( after admixture from the source population ) and secondary contact followed by introgressive hybridization ( Fig 4 ) . In addition , we aimed to evaluate different topologies within the radiations to further investigate the support for simultaneous divergence events . Building up on the results of the two-population models , for all models in both radiations we included a ‘bottlegrowth’ event in the source population and exponential growth in the crater lake species . Furthermore , in L . Apoyo we did not include gene flow between the lakes , whereas in models of L . Xiloá we allowed for migration from the source population ( L . Managua ) into the crater lake species . Migration was assumed to be identical , that is only one migration parameter was used . In both radiations we added gene flow between the sympatric species , assuming it again to be identical and symmetrical . While this assumption may be overly simplistic , including different migration parameters for all twelve possible migration routes would have likely over-parameterized our models . For L . Apoyo we tested six different models . Five models of sympatric speciation and one model of secondary contact were evaluated . Within the sympatric speciation models our aim was to evaluate the support for three different topologies ( see below ) , intralacustrine gene flow , and an admixture event prior to sympatric speciation ( we refer to the model of sympatric speciation without prior admixture as “single colonization” ) . Incorporating an initial split of A . zaliosus from the benthic species was strongly supported over a simultaneous split of all species ( Table 2 ) . However , including another parameter to model an additional split of cluster 5 from the other two benthic species , as weakly indicated in our phylogenetic analysis , did not increase the likelihood . Removing gene flow among the sympatric species or the admixture event into the crater lake population before sympatric speciation strongly decreased the likelihood of the model . With four species a multitude of two-colonization scenarios is conceivable , yet it is computationally unfeasible and biologically not sensible to test all possible models [69] . Hence , based on the firmly established finding that A . zaliosus is genetically the most distinct species within the radiation of L . Apoyo we formalized the main competing hypothesis of secondary contact as: an initial colonization by A . zaliosus followed by a secondary colonization by the ancestral population of the benthic species and admixture . This model was 2 . 5 times less likely than the best model of admixture prior to sympatric speciation . Similar to L . Apoyo , for L . Xiloá we tested seven models of sympatric speciation and two models of secondary contact . Modeling two intralacustrine divergence events , one between A . amarillo , A . viridis , and a third population which later split into A . sagittae and A . xiloaensis , was strongly supported over a model in which one ancestral population split simultaneously into all species . However , a sister relationships of A . amarillo and A . viridis , or an initial split by either of the two species did not further improve the model . Gene flow between the sympatric species and an admixture event before the onset of the radiation was again strongly supported ( Table 2 ) . Given the seemingly closer genetic affiliation of A . amarillo and A . viridis to the source population in our global PCA , we framed the main alternative hypothesis of secondary contact ( in contrast to sympatric speciation ) for L . Xiloá to be: an initial colonization by the ancestral population of A . sagittae and A . xiloaensis followed by a secondary colonization by the ancestral population of A . amarillo and A . viridis and subsequent admixture . For this type of model we tried two different topologies , one in which A . amarillo and A . viridis split before A . sagittae and A . xiloaensis , and a simultaneous split of the two lineages . The latter model was more strongly supported , yet it was about 70 times less likely than the model of sympatric speciation .
Any argument for or against sympatric speciation has to rest on a valid taxonomic assignment [27] and while we agree that it is important to take all species in a respective radiation into account the current taxonomy in L . Apoyo has been in conflict with genetic data . For example , only a single species formed a monophyletic group in [53] and [49] found the highest support for only two genetic clusters . Thus we decided to use a more objective approach and assign individuals to genetic clusters—essentially applying a genetic cluster species concept [57]–and adhere to these clusters for all subsequent analyses . But we note that the assignment based on morphology fits the genetic signature in all cases of A . zaliosus in L . Apoyo and almost all cases in L . Xiloá . And even in the case of the genetic clusters in L . Apoyo there are clear trends . For example , all individuals of A . astorquii are assigned to cluster 2 , and eight of the nine individuals in cluster 4 are A . globosus . Consequently , our results do not imply that there are no morphologically distinct species of Midas cichlids , but rather that the assignment based on morphological criteria alone is often difficult in these young radiations with ongoing hybridization . Nonetheless , our data do not support the current six-species taxonomy in L . Apoyo since we only find strong support for five instead of six genetic clusters . It is worth mentioning though that our results only apply to our data set and that it cannot be ruled out that more genetic clusters exist . In L . Xiloá our results agree with previous studies that have reported four genetic clusters corresponding to the currently described four species [48 , 49] . Interestingly , a high number of admixed individuals between A . sagittae and A . xiloaensis is already apparent in these studies , although this was not the focus of their discussion . The occurrence of some gene flow between the sympatric species is not unexpected , as reproductive barriers are thought to be incomplete and mainly based on mate choice ( pre-mating ) and divergent selection against hybrids ( extrinsic postzygotic ) . No intrinsic incompatibilities are known to occur in Midas cichlids [72] and species can be easily crossed in the laboratory [73 , 74] . That we find so much hybridization between the limnetic A . sagittae and the benthic A . xiloaensis is puzzling . If these two species arose by ecological speciation one would expect hybrids to have a reduced fitness and be thus less frequent [75–78] . We can think of two main possible explanations for the existence of the hybrid group: the two species A . sagittae and A . xiloaensis , which seem to have diverged only ca . 750 generations ago , might still be in an early stage of the speciation continuum [23] and reproductive barriers are weaker than between other species . Alternatively , the hybrid group itself could be in the process of becoming a stable and distinct population . In other words we could be witnessing the early stages of another speciation process due to ecological niche partitioning . Our genetic data provide conflicting evidence for the two alternative scenarios . The hybrid group did not form a distinct cluster in our Admixture analysis , yet it formed the sister group of A . xiloaensis in our phylogenetic analysis and did not exhibit an intermediate position between the two species . We acknowledge , however , that the latter result was probably affected by the admixture proportions of the randomly selected individuals used to infer the phylogeny: a post hoc examination showed that the average admixture proportions of the four used hybrid individuals were slightly in favor of A . xiloaensis ( 51 . 7% of their ancestry ) over A . sagittae ( 48 . 3% ancestry ) . This could explain why the hybrid group was resolved as the sister group of A . xiloaensis and did not result in a trifurcation . On the other hand , overall genetic differentiation between the two species A . sagittae and A . xiloaensis was relatively high ( S2 Table ) , which might not be expected if there was a lot of ongoing gene flow between them . More detailed genomic analyses as well as ecological experiments beyond the scope of this study are necessary to determine the fitness and ecological niche of hybrids compared to both parental species . The two large and old source lakes contain two species of Midas cichlids and sympatric speciation requires that none of the species in the respective crater lakes are more closely related to either of the species in the source lakes than the other sympatric species are [27] . Our phylogenetic analyses suggest that the species in the great lakes are sister species and thus equally distantly related to each of the species in the crater lake radiations . They are also equally distant in genetic space in our global PCA ( Fig 1 ) . This pattern either suggests that the two species in the great lakes only diverged after the crater lakes had been colonized from their shared ancestral population , or that we do not have enough power to resolve the exact relationships with our current data set; the two species in the great lakes are genetically almost not distinguishable ( S2 Table ) [52] . Only the latter case would make an interpretation in regard to sympatric speciation more difficult . If the crater lakes had been colonized by a set of two species , two ( but not all ) of the species in the respective radiations could theoretically simply be the descendants of the two founding species . Efforts to characterize the genomic differences between the two great lake species are currently underway and diagnostic haplotypes might help to finally resolve whether these crater lakes were colonized by either one or both of the species . Yet , the fact that no fish resembling A . labiatus with its characteristic hypertrophied lips exists in these crater lakes ( they have only anecdotally been reported to occur in L . Xiloá ) speaks in favor of a colonization by A . citrinellus alone . We note that fish resembling A . labiatus do occur at considerable frequencies in two other crater lakes , L . Masaya and L . Apoyeque [79] . Hence , the ecological niche ( foraging in rocky crevices ) that A . labiatus is adapted to [46] is probably present in crater lakes Apoyo and Xiloá as well and if A . labiatus colonized these crater lakes it is difficult to conceive of why their phenotype would have changed completely . We further note that A . labiatus occurs much less frequently in the great lakes than A . citrinellus ( ca . 5% ) and it is therefore not unlikely that A . labiatus never colonized these crater lakes while A . citrinellus did . In conclusion , while we cannot rule out at the moment that two of the sympatric species in the crater lake radiations are the result of a double-colonization by the two species from the great lakes ( if they diverged before the colonization of the crater lakes ) , we think the fact that no clear traces of an A . labiatus-like phenotype are present in these two crater lakes makes a colonization by only one species more parsimonious . Besides establishing the relationship between the two species in the great lakes and the crater lake species we were further interested in the branching pattern within the radiations . In L . Apoyo the cloudogram resembles a starlike phylogeny with an almost simultaneous split of all five species ( Fig 3A ) . Yet , A . zaliosus is the first species to branch off the stem lineage , albeit only slightly before the other species . This split is also supported in our demographic models and is consistent with previous studies [49 , 52] . However , neither our phylogenetic tree , network , nor demographic models can unambiguously resolve the relationships among the other four endemic species from L . Apoyo . Similarly , the tree in L . Xiloá remains only partially resolved with a simultaneous split of A . amarillo , A . viridis , and the ancestor of A . sagittae and A . xiloaensis ( Fig 3B ) . The sister relationship of the latter two is again consistent with our earlier work [49] , supporting the interesting conclusion that the limnetic-benthic divergence happened via non-parallel routes in the two parallel adaptive radiations of the crater lakes . Yet , in our previous phylogenetic analysis [49] A . amarillo splits off first in all bootstrap replicates . The discrepancy with these current results could be due to the fact that the former phylogeny was based on a concatenated SNP matrix , which may be problematic in this young species complex , where shared ancestral variation and incomplete lineage sorting prevail [80] . Our analyses in this study explicitly take incomplete lineage sorting and in the case of the demographic models also gene flow and changing population sizes into account . Thus , we are led to support the hypothesis that some of the speciation events happened simultaneously and represent hard polytomies , as has been recently suggested to occur in birds [81] . And even if the splits did not happen strictly simultaneously they seem to have occurred in extremely rapid succession , which suggests that ecological interactions among the incipient species may have played a role . The possibility of a multispecies outcome of sympatric speciation was proposed based on a theoretical model ten years ago [82] , strikingly invoking crater lake cichlids as an example where it might have occurred . Indeed , our phylogeny of L . Apoyo resembles the outcome of a simulation , in which one panmictic population diverges into six species after only ca . 400 generations [82] . Thus , we propose that Midas cichlids represent , to our knowledge , the first empirical case of a multispecies outcome of sympatric speciation . The model was , however , one of ‘pure’ sympatric speciation and if the admixture event prior to radiation that we detected is real and did facilitate speciation in sympatry the theoretical model may not directly correspond to the situation in Midas cichlids . Unequal levels of shared ancestry with an outgroup or the source population can be indicative of a past period of allopatry of sympatric species followed by introgression upon secondary contact . This pattern would be expected to be reflected in intermediate positions of certain species along the major axes of genetic variation in a PCA [27 , 34 , 83] . In our global PCA ( Fig 1 ) all individuals from L . Apoyo are equidistant to the source population , consistent with sympatric speciation . In L . Xiloá , however , two species are closer to the source population than the other two . While this pattern might suggest secondary contact and hybridization , the f3-statistics do not support this explanation . This signal of admixed ancestry did also disappear in the Admixture analyses when assuming more than four clusters ( S1 Fig ) . Instead , we propose that this pattern in the global PCA might rather reflect a difference in population sizes; our demographic models suggest that the population sizes of A . amarillo and A . viridis have been larger than the other two species and they may have thus retained more of the ancestral variation . Thus , altogether our phylogenetic and genetic clustering results are consistent with sympatric speciation and provide no evidence for an initial divergence of the sympatric crater lake species in geographic isolation followed by introgressive hybridization ( a scenario of secondary contact ) . Yet , f3-statistics may not have enough power to detect admixture in species that have subsequently experienced a considerable amount of genetic drift [60 , 61 , S1 Text therein] , which might be the case in Midas cichlids . Furthermore , neither the clustering or phylogenetic methods , nor f3-statistics will detect multiple colonizations from the same source population ( admixture ) that happened prior to the onset of the radiation , as all species within the radiation would share the same amount of shared ancestry and drift paths compared to the source population . Thus , we formulated the most plausible hypotheses for the different evolutionary scenarios in demographic models and evaluated their evidence using information-theory-based criteria [69] . The main three models we aimed to compare were: sympatric speciation after a single colonization , sympatric speciation after a secondary colonization ( admixture prior to sympatric speciation ) , and two waves of colonization followed by admixture ( secondary contact and introgressive hybridization ) . The first two models are scenarios of primary divergence-with-gene-flow , whereas the latter one models secondary gene flow after an initial period of allopatry . For both radiations the respective models of sympatric speciation after a single colonization had essentially no support and an admixture event from the source population into the crater lakes prior to sympatric speciation is strongly supported . Yet , consistent with our clustering and phylogenetic analyses as well as f3-statistics , the evolutionary history of both radiations is better modelled by a scenario of primary divergence-with-gene-flow than by an initial period of allopatry of the crater lake species themselves: in L . Apoyo sympatric speciation after an admixture event is 2 . 5 times more likely than a scenario of secondary contact . Furthermore , even though the likelihood for secondary contact is not negligible the scenario seems biologically less plausible . According to the parameter estimates the population of secondary colonizers ( i . e . the ancestral population of the four benthic clusters ) would have received 90% of its gene pool from the established population ( i . e . the lineage of A . zaliosus ) , but only about 250 generations after they arrived in the crater lake . For the radiation in L . Xiloá the evidence is clearly in favor of sympatric speciation after admixture , which is 70 times more likely than secondary contact . The fact that the likelihood ratio between the model of sympatric speciation and secondary contact is so much higher in L . Xiloá than in L . Apoyo could be due to the fact that A . zaliosus in L . Apoyo is much more distinct from the other sympatric species than any of the species in L . Xiloá are compared to each other . Furthermore , we had slightly more data in the site frequency spectrum ( SFS ) of L . Xiloá and thus potentially more power to distinguish between the models . Overall , for both radiations our data provide more support for a model of primary divergence in sympatry than one in which already partly diverged populations diverged further and speciated upon secondary contact . However , an admixture event from the source population into the stem lineage of the crater lake flocks ( before the species diverged ) is strongly supported in both cases compared to the respective models of sympatric speciation after a single colonization , thus opening the possibility of sympatric speciation after formation of a hybrid swarm [34 , 36] . The fact that the admixture event happened in both radiations shortly before the first speciation event makes it tempting to assume a causal relationship . We think that this is certainly possible , but we warrant caution at this point . It is an interesting hypothesis that the 4% admixture into L . Apoyo from the same ancestral lineage after only about 800 generations of separation provided the genetic substrate to initiate sympatric speciation . But only when the traits involved in reproductive isolation and their genetic basis is identified can a causal relationship be investigated . Furthermore , distinguishing between the causes of shared polymorphisms remains inherently difficult [84–87] . Once a trinucleotide substitution matrix [88 , 89] is available for cichlids , future studies making use of information about ancestral and derived allelic states , that is , using the more powerful derived SFS , should be used to evaluate our results . Moreover , whole-genome data will likely increase the power to test the different hypotheses due to a higher number of segregating sites and information about the size of linkage blocks [90] . The 29% admixture into L . Xiloá seem more likely , at least probabilistically , to have had an impact . In any case , the admixture events would primarily explain the first speciation events in the two radiations and further speciation might have been sympatric in the ‘pure’ sense . Yet , we acknowledge that further speciation events could also have been driven by bouts of ecological interactions and complex sorting of partial reproductive incompatibilities , once two or more ( incipient ) species had evolved [36] . The support for a bottleneck in the great lake populations around 1 , 500 and 2 , 000 generations ago was unexpected , yet it is not inconceivable that major geological events in this tectonically active area of Nicaragua have strongly affected the fauna in the lakes . Indeed , there is geological evidence for an underwater eruption of a volcano in L . Managua that caused a tsunami only about 3 , 000–6 , 000 years ago [91] and other possible tsunamis in L . Nicaragua triggered by debris avalanches [92] . Assuming a generation time of two years we propose that the inferred bottleneck coincides with such an event . The signal of exponential growth in the great lakes after such an event is not unexpected and also the inferred exponential growth of the crater lake populations after colonization by a small founder population seems biologically sensible . Nonetheless , a signal of population growth can be falsely inferred for several reasons . Technical reasons , such as sequencing or PCR-based errors leading to an excess of singletons seem unlikely , as growth was also supported with a more stringent threshold of 15x read depth and since we used a low number of amplification cycles , performed ten PCR replicates , and used a high-fidelity polymerase for genomic library preparation . Multiple-merger coalescent events [93] and background selection [94] , however , cannot be ruled out to have affected the analyses . Yet , the site frequency spectrum ( SFS ) contains often enough information to distinguish between multiple-mergers and exponential growth [95] and methods to jointly infer the demographic history and the effects of selection are an active and promising area of research that may help to sort out the relative effects of selection and demography [94 , 96] . The inferred colonization and divergence times for these endemic crater lake cichlid species are much lower than we anticipated . Considering that L . Apoyo is ca . 24 , 000 and L . Xiloá ca . 6 , 100 years old [97] , our results would imply that especially L . Apoyo has been devoid of a stable population of Midas cichlids for much of its history . Previous studies have reported divergence times that are closer to the age of the lakes , yet these studies were based solely on a single mtDNA marker for which calibration times and molecular clock rates are uncertain or debated [50 , 54] . Moreover , using a local substitution rate that was calibrated by equating the geological age of L . Apoyo with a signal of population expansion ( mismatch distribution of mtDNA ) as a proxy for divergence time [98] might have been a too strong assumption . Uncertainty about the substitution rate might also be a source of error in this study . Similarly , the ratio of monomorphic to polymorphic sites is important for obtaining absolute estimates and is to some extent affected by the way the data has been processed . Too strict filtering can lead to an underestimate of the number of polymorphisms and bias the absolute estimates . However , while our absolute estimates may change depending on the substitution rate and data filtering criteria , the relative values and the model selection procedure should not be affected by this . Assuming that the substitution rate is approximately correct and the effect of data filtering unbiased and negligible , with ca . 1 , 690 and 1 , 320 generations for L . Apoyo and L . Xiloá , the ages of these two radiations are much younger than previously thought . Alternatively , we cannot rule out the possibility that older populations of cichlids in these crater lakes might have been almost or completely exterminated—for example by volcanic activity ( although there is no geological evidence for this ) —and these earlier populations or species were replaced only recently by the extant radiations that are less than 2 , 000 generations old . Recurrent mass extinctions due to volcanic activity have possibly occurred in other crater lakes such as Lake Apoyeque [79] and such events would lead to a loss of information stored in the SFS [68] and could thus bias our estimates downwards . The same argument applies to the bottleneck in the source populations , yet our simulations suggest that we can correctly infer divergence times that happened before the bottleneck . Ultimately , fossils from the beds of the crater lakes might further inform on this issue . Cichlid fishes in general exhibit one of the fastest known speciation rates [99] and , acknowledging the caveats described above , our data suggest that speciation rates in Midas cichlids might even be the fastest reported yet . According to our estimates it took only around 1 , 000 and 600 generations from the time of colonization to the last splits leading to the five species in L . Apoyo and four species in L . Xiloá , respectively . Such a rate of speciation is unprecedented , even though it might be considered that speciation is not fully completed yet as there is still some level of ongoing gene flow—at least between a subset of these species . Ecological speciation in general can commence very rapidly [100] and a theoretical model of Midas cichlids showed that sympatric speciation can happen in less than 20 , 000 generations [101] . Moreover , in the model of [82] five species evolved in as little as ca . 400 generations–quite similar in both number and timing as we have inferred in the case of these two crater lake adaptive radiations . Both of these models are based on several assumptions and investigating whether these are met in Midas cichlids will require future behavioral and ecological research . In this study we reconstructed the demographic history of two endemic radiations of Midas cichlids inhabiting the small and isolated crater lakes Apoyo and Xiloá to infer whether complex periods of geographic isolation may have facilitated speciation within the two radiations . Apart from this main objective our large genome-wide data set suggests that most of the species evolved in a burst of speciation , thus making these two radiations of cichlids , to our knowledge , the first empirical examples of multispecies outcomes of sympatric speciation [82] . Moreover , these radiations of nine species took place within only about a thousand generations , making them some of the fastest speciation rates reported so far . Unlike recent evidence presented for Cameroonian crater lake cichlids [34] , our population clustering and phylogenetic analyses are consistent with a scenario of primary divergence-with-gene-flow ( sympatric speciation ) of the crater lake species . Interestingly though , our models do provide evidence for a secondary colonization from the source population that happened shortly before the species radiated in both crater lakes . Whether Midas cichlids represent therefore a good case of sympatric speciation may ultimately depend on the definition of sympatric speciation [19–21] . The species flocks of Midas cichlids in the Nicaraguan crater lakes arose via sympatric speciation in the sense that their divergence happened most likely in a geographic setting that does not offer geographic barriers to gene flow . Yet , if the admixture event from the source population was instrumental for seeding speciation in sympatry by providing some of the genetic variation involved in reproductive isolation , then a short period of geographic isolation would have been involved in speciation . This would make the radiations of Midas cichlids no longer a case of ‘pure’ sympatric speciation , similarly to the sympatric divergence of apple maggot flies in North America for example [35] . Whether the admixture event was essential for speciation remains to be elucidated . If confirmed , this could partly explain how Midas cichlids have speciated so rapidly in sympatry . It would also exemplify that the term ‘primary divergence’ may have a different meaning when applied at the level of populations and incipient species or at the level of individual genetic variants that distinguish them while the rest of the genome can be exchanged freely [102] . Overall , rather than adding to the debate over whether speciation conforms to a single category of speciation or not we think the results presented here open up a new and exciting hypothesis of how speciation may have happened in the extremely young and repeated radiations of Midas cichlids .
Sampling was approved and conducted in accordance with the regulations of the local authorities , the Ministerio de Ambiente y Recursos Naturales , Nicaragua ( MARENA ) . Fish were collected in the field in 2001 , 2003 , 2005 , 2007 , 2010 and 2012 with gill nets or by harpooning . Specimens were photographed in a standardized way and tissue samples from fin or muscle were taken and preserved in pure ethanol . Genomic data were generated using double-digest RAD sequencing [103] following an in-house protocol [104] with minor modifications . Briefly , for each individual 600 ng DNA template were digested with the restriction enzymes PstI-HF ( NEB ) and MspI ( NEB ) . The success of every single digestion reaction was visually inspected on a 2% agarose gel and samples showing a heterogeneous fragment distribution ( e . g . due to incomplete digestion or degradation ) were replaced . After ligation of individually-barcoded adaptors [provided in ref . 104] individuals were combined into pools of 50 samples . Fragments in a range of 320–500 bp were selected using Pippin Prep technology ( Sage Science , Beverly , MA ) and amplified in replicates of ten PCRs per pool , running for ten cycles , using a Phusion high-fidelity polymerase ( NEB ) . Oligonucleotide dimers were removed by gel electrophoresis and fragment size distribution was inspected with an Agilent 2100 Bioanalyzer machine . Finally , genomic libraries ( pools ) were single-end sequenced for 101 cycles using Illumina HiSeq 2000 technology at the genomics core facility of TUFTS University ( Boston , MA ) . Sequence quality was inspected with FastQC and no systematic bias or quality drop-off at the end of the reads was observed . Thus , no trimming was performed . Individually-barcoded reads were de-multiplexed using the process_radtags script included in the Stacks v . 1 . 29 software pipeline [105 , 106] . Reads containing uncalled bases and/or showing an average quality score of less than 25 in a sliding window of 10% total read length were discarded . The remaining reads were mapped to an anchored in-house genome assembly of an individual of A . citrinellus from Lake Nicaragua [49] with bwa v . 0 . 7 . 12 [107] . Reads mapping to several positions in the genome , containing soft-clipped positions , or showing a mapping quality of less than 25 were discarded using custom bash scripts . Genotyping was conducted with Stacks using a minimum of five reads to form a locus . Based on population level information , the rxstacks correction module of Stacks was used to remove loci being confounded in more than 25% of individuals , or showing an excess of haplotypes within populations . This module furthermore corrects individual genotype calls based on population information . Genotypes were called setting an upper bound of 0 . 05 for the error rate and using a 5% significance level cut-off ( non-significant likelihood ratios of genotype models resulted in uncalled genotypes ) . At each locus and individual , log-likelihood values for each genotype call ( every nucleotide position ) were summed up and individual genotype calls at loci with an overall log-likelihood of less than -10 were filtered out and did not contribute to any subsequent analyses . If an individual is , for example , unambiguously homozygous across the whole length of a locus the respective log-likelihood value will be zero . Similarly , if a heterozygous position is supported by an equal representation of alleles , the log-likelihood for the call will be close to zero . Loci with many poorly supported genotype calls ( e . g . due to sequencing errors ) will exhibit more negative log-likelihood values . The cut-off value of -10 was chosen based on the empirical distribution of log-likelihoods in our data set . On average 70 , 538 ± 17 , 191 ( sd ) loci were obtained per individual with a mean coverage of 13 . 8 ± 4 . 9 ( sd ) reads per locus and individual . Individual- and population-level information on number of loci , average coverage per locus , proportion of missing data included in the matrix used in the global PCA ( Fig 1 ) and Admixture analysis ( S1 Fig ) , as well as genomic library IDs are provided in S7 Table . Our data exhibited an excess number of polymorphisms in the last two positions of the reads and loci with polymorphisms in these positions were thus excluded ( i . e . blacklisted ) from all subsequent analyses . In an attempt to account for hidden paralogy , loci deviating from Hardy-Weinberg-Equilibrium ( HWE ) ( 5% significance level ) or containing more than three SNPs within a population were excluded from further analyses . HWE exact tests [108] were performed in Plink v . 1 . 19beta [109] . Hybrid individuals were treated together as a separate group and no HWE tests were conducted in this group . Furthermore , unless noted otherwise , only loci that were genotyped in at least six individuals per population were used in subsequent analyses . Population structure in our data set was explored with the model-based approach of Admixture v . 1 . 23 [56] and with model-free principal component analyses ( PCA ) as implemented in the Eigensoft v . 5 . 0 . 2 package [55] . Both methods were applied in a hierarchical design . First , all samples were included in one analysis ( ‘global’ ) . In a second step , samples from the two crater lakes were analyzed separately to investigate population structure within lakes ( ‘intralacustrine’ ) in more detail . Admixture was run from 1–18 predefined clusters ( K ) in the global analysis . In the intralacustrine analyses Admixture was run for 1–8 clusters . Statistical support for the different number of clusters was evaluated based on ten rounds of the implemented cross validation technique . Missing data in the PCAs were accounted for by solving least squares equations ( applying the lsqproject function ) . Statistical significance of principal components was determined by means of the implemented Tracy-Widom statistics . PCAs were visualized in R v . 3 . 1 . 2 [110] using the scatterplot3d library [111] . For both approaches only one SNP per locus was used to reduce the effect of non-independence ( linkage ) among markers . Overall pairwise genetic differentiation was calculated in Arlequin v . 3 . 5 . 1 . 3 [112 , 113] and statistical significance was assessed by means of 10 , 000 permutations . The same data set that was used for the global PCA and Admixture analyses went into this analysis . We examined body shape differentiation among populations of the two crater lakes using the body height index ( BHI ) and geometric morphometrics . Data for both measures were obtained from standardized pictures . The BHI is defined as the ratio of body height divided by standard length and is a simple measure to capture the main morphological differentiation between the elongated limnetic and high-bodied benthic species . For geometric morphometrics seven homologous body landmarks were digitized in TPSDIG 2 . 17 [114] for all individuals of A . sagittae , A . xiloaensis , and the hybrid group in L . Xiloá . Landmarks are a subset ( labels 1 , 6 , 9 , 10 , 12 , 14 , 15 ) of previously defined positions [44 , Fig 2] that capture the main differentiation in body shape between the focal species . Shape analyses were performed in MorphoJ 1 . 03d [115] . Landmarks were first aligned using a full Procrustes superimposition , which involves scaling all shapes to unit centroid size , translation to a common position , and rotation to minimize the Procrustes distance between landmark configurations [116 , 117] . Allometry is common in fish and thus morphology and total body size are typically related [117] . Therefore , a multivariate regression of body shape ( Procrustes coordinates ) on size ( centroid size ) was used to correct for allometric effects . Regression residuals were then used for all downstream geometric morphometric analyses . Individual variation in body shape across and within species was visualized using a PCA on the regression residuals . First , a maximum likelihood phylogenetic tree was built from allele frequency data using Treemix v . 1 . 12 [61] . Support for the tree topology was assessed by means of 1 , 000 bootstrap replicates using a block size of 20 adjacent SNPs . Trees were rooted with A . citrinellus from Lake Nicaragua . Up to four migration events were fitted on the tree . Admixture between populations was formally tested with f3-statistics [59] implemented in the threepop software of Treemix . Standard errors were calculated in blocks of 20 adjacent SNPs . Only SNPs assigned to the 24 linkage groups of our reference genome were used in Treemix and threepop analyses . Phylogenetic trees were built using the Bayesian method implemented in SNAPP v . 1 . 10 [62]; an add-on package of BEAST v . 2 . 2 . 1 [118] . Due to the computational demand of SNAPP only four randomly selected individuals per species were used and trees for the two crater lakes and their respective source populations were built separately . Backward and forward mutation rates ( u+v ) were estimated from the stationary allele frequencies in the data ( u = 0 . 6196; v = 2 . 5907 for L . Apoyo and u = 0 . 6596; v = 2 . 0661 for L . Xiloá ) . Only one SNP per locus and only SNPs genotyped in all individuals were used . Each analysis was run for more than five million generations , discarding the first 10% as burn-in . Trace files were inspected with Tracer v . 1 . 6 [119] and effective sample sizes were higher than 200 for all parameters . Trees were visualized with DensiTree [120] . In a second approach we built individual-based phylogenetic networks with SplitsTree v . 4 . 13 . 1 [121] . Similar to the phylogenetic trees , the networks were built separately for the two crater lake radiations and their respective sources . Individual genotype calls were transformed from VCF to Nexus format using custom scripts and networks were built using the NeighborNet method based on uncorrected P distances . The demographic history was inferred using the information contained in the multidimensional site frequency spectrum ( SFS ) and fastsimcoal v . 2 . 5 . 2 . 3 [66] . Briefly , fastsimcoal2 performs coalescent simulations under an arbitrarily complex predefined demographic model and then uses a conditional maximization ( ECM ) algorithm to optimize each parameter in turn to maximize the likelihood given the data . Demographic models are not restricted to a certain number of populations and can include a variety of demographic events such as migration , population size changes , population splits and admixture events . In an attempt to reduce the effect of selection , loci presumably located in coding regions were excluded; these loci were identified via a blastn search against a compilation of transcriptomic data from various species and tissues of Midas cichlids [45 , 122 , 123] . Furthermore , only one SNP per locus was used to reduce the effect of non-independence of markers [124] . The SFS was created in the following way: data were parsed from variant call format ( VCF ) files using a custom python script and transformed into the MSFS using a modified script available from δaδi [67] . Since no trinucleotide substitution matrix is available for cichlids to correct for ancestral misidentification [88 , 89] we used the minor ( folded ) site frequency spectrum . Initially , simple one-population models were run for both source lake populations . Subsequently , each crater lake species was analyzed together with its respective source population in two-population models . Finally , for both crater lakes each four species were analyzed jointly with the source population in five-population models . Including all five sympatric species in one analysis in the case of Apoyo was not possible as fastsimcoal2 is currently limited to handling SFS files of up to one million entries and including another population would have meant reducing the sample size to only four samples per population . The number of entries in the multidimensional SFS is the product of the number of alleles plus one ( for the state of zero ) per population . In the case of L . Apoyo we excluded cluster 4 as we only had nine individuals in our data set . In L . Xiloá all four sympatric species could be included in one analysis; the hybrid group was not considered in these analyses . To alleviate the problem of missing data in building the SFS from RADseq data , sample sizes were projected downwards to a certain size using δaδi’s projection function [67] . In one- and two-population models the source populations were projected down to 50 alleles ( 25 individuals ) , except for A . labiatus from L . Managua ( projected to 30 alleles ) , and the crater lake species each to 30 alleles , except for clusters 3 , 4 , and 5 in Apoyo , which were projected down to 20 , 14 , and 20 alleles , due to their small sample sizes , respectively . In the five-population models sample sizes had to be projected down to 18 alleles for the source populations and 14 alleles for each of the crater lake species due to the limitation of a maximum of one million entries in the SFS . Note that the projected samples sizes were used as a minimum threshold to create the VCF files , that is , a locus for which fewer samples were genotyped than the targeted sample size for projection in any one population was excluded . This filter was applied to both polymorphic and monomorphic loci , which is crucial to obtain the correct ratio of monomorphic to polymorphic sites , by which all demographic parameters are scaled . In more detail , the number of monomorphic sites was added manually to the SFS and theoretically equals the respective number of loci times the 89 potentially variable sites ( obtained by subtracting the 5 bp of restriction site and the 2 blacklisted sites from the 96 bp reads ) minus the number of segregating sites . Yet , using only one SNP per locus decreases the ratio of polymorphisms and thus biases the estimates . This bias was corrected for by first calculating the ratio of monomorphic to polymorphic sites using all SNPs . The resulting number of monomorphic sites is then the number of SNPs ( one per locus ) multiplied by this ratio . To convert the inferred parameters into demographic units , a substitution rate of 7 . 5 x 10−9 per site and generation similar to a recent estimate from nine- and three-spine sticklebacks was assumed [125] . For each demographic model at least 25 independent fastsimcoal2 runs with relatively broad prior search ranges for the parameters were conducted . If several competing models returned similar likelihoods or we were interested in the maximum likelihood parameter estimates the number of runs was increased . Likelihoods are approximated and increasing the number of runs thus increases the chance of minimizing the error [66] . In this case the prior search range was adjusted to better accommodate more likely values of the parameters . Especially in the case of more complex models and time parameters reducing the prior search ranges often enhanced convergence . But note that prior bounds only define initial search ranges and are not to be understood like priors in a Bayesian approach . Only the lower bound is an absolute boundary in fastsimcoal2 . The upper bound can increase each round . Each run consisted of 20–50 rounds of parameter estimation via the ECM algorithm with a length of 100 , 000–250 , 000 coalescent simulations each ( increasing by 5 , 000 steps each round ) . The relative fit of the different demographic models to the data was evaluated by means of the Akaike Information Criterion ( AIC ) after transforming the log10-likelihood values to ln-likelihoods . Following [66] 95% confidence intervals were calculated by parametric bootstrapping . Bootstrap replicates ( n = 25 ) were obtained by simulating minor site frequency spectra using the same overall corrected sequence length as the empirical data ( unlinked regions of 89 bp ) and according to the maximum likelihood parameter point estimates followed by re-estimating the parameters . | Speciation is the main driver of biological diversity and how species arise is a central question in evolutionary biology . For speciation to occur in sexually reproducing organisms the exchange of genetic material ( gene flow ) between populations has to be reduced . Ultimately this has to be due to genetically determined reproductive incompatibilities between species . Yet , whether ( an initial period of ) geographic isolation is necessary for these incompatibilities to evolve has been subject to one of the most persistent debates in evolutionary biology . Sympatric speciation is the most extreme case of primary divergence-with-gene-flow and lies at the heart of this question . However , only few empirical examples of sympatric speciation are generally accepted and in most of these cases some ambiguities and doubts remain . This study provides evidence that the Nicaraguan crater lake cichlids can indeed be considered a valid example of sympatric speciation in the sense that the species themselves probably started to diverge in the absence of geographic barriers . However , the data also suggests that this divergence in sympatry may have been facilitated by genetic variants that evolved during a time of isolation between an initial founding population and a secondary wave of colonizers stemming from the same source population . This highlights the limitations in the definitions of sympatric speciation when the mosaic nature of genomes is taken into account: some of the genetic regions driving divergence may have evolved in allopatry while the populations themselves diverged in sympatry . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"species",
"colonization",
"ecology",
"and",
"environmental",
"sciences",
"demography",
"population",
"genetics",
"aquatic",
"environments",
"phylogenetic",
"analysis",
"speciation",
"bodies",
"of",
"water",
"molecular",
"biology",
"techniques",
"population",
"biology",
"... | 2016 | Multispecies Outcomes of Sympatric Speciation after Admixture with the Source Population in Two Radiations of Nicaraguan Crater Lake Cichlids |
Single-molecule techniques for protein sequencing are making headway towards single-cell proteomics and are projected to propel our understanding of cellular biology and disease . Yet , single cell proteomics presents a substantial unmet challenge due to the unavailability of protein amplification techniques , and the vast dynamic-range of protein expression in cells . Here , we describe and computationally investigate the feasibility of a novel approach for single-protein identification using tri-color fluorescence and plasmonic-nanopore devices . Comprehensive computer simulations of denatured protein translocation processes through the nanopores show that the tri-color fluorescence time-traces retain sufficient information to permit pattern-recognition algorithms to correctly identify the vast majority of proteins in the human proteome . Importantly , even when taking into account realistic experimental conditions , which restrict the spatial and temporal resolutions as well as the labeling efficiency , and add substantial noise , a deep-learning protein classifier achieves 97% whole-proteome accuracies . Applying our approach for protein datasets of clinical relevancy , such as the plasma proteome or cytokine panels , we obtain ~98% correct protein identification . This study suggests the feasibility of a method for accurate and high-throughput protein identification , which is highly versatile and applicable .
Modern DNA sequencing techniques have revolutionized genomics [1] , but extending these methods to routine proteome analysis , and specifically to single-cell proteomics , remains a global unmet challenge . This is attributed to the fundamental complexity of the proteome: protein expression level spans several orders of magnitude , from a single copy to tens of thousands of copies per cell; and the total number of proteins in each cell is staggering [2] . Given the lack of in-vitro protein amplification assays the ability to accurately quantify both abundant and rare proteins hinges on the development of single-protein identification methods that also feature extraordinary-high sensing throughput . To date , however , protein sequencing techniques , such as mass-spectrometry , have not reached single-molecule resolution , and rely on bulk averaging from hundreds of cells or more [3] . Affinity-based method can reach single protein sensitivity [4] , but depend on limited repertoires of antibodies , thus severely hindering their applicability for proteome-wide analyses . Consequently , in the past few years single-molecule approaches for proteome analysis based on Edman degradation [5] or FRET [6] have been proposed . To date , however , profiling of the entire proteome of individual cells remains the ultimate challenge in proteomics [7] . Nanopores are single-molecule biosensors adapted for DNA sequencing , as well as other biosensing applications [8 , 9] . Recent nanopore studies extended nucleic-acid detection to proteins , demonstrating that ion current traces contain information about protein size , charge and structure [10–17] . However , to date , the challenge of deconvolving the electrical ion-current trace to determine the protein’s amino-acid sequence from the time-dependent electrical signal has remained elusive . In an analogy to the field of transcriptomics , in many practical cases it is sufficient to identify and quantify each protein among the repertoire of known proteins , instead of re-sequencing it . Yao and co-workers showed theoretically that most proteins in the human proteome database can be uniquely identified by the order of appearance of just two amino-acids , lysine and cysteine ( K and C , respectively ) [18] . But taking into account experimental errors , for example due to false calling of an amino-acid , or an unlabeled amino-acid , sharply reduces the ID accuracy . Motivated by recent experiments suggesting the ability to translocate SDS-denatured proteins through either small nanopores ( ~0 . 5 nm ) [19] , or large nanopores [20] ( ~10 nm ) , and the possibility to differentiate among polypeptides based on optical sensing in nanopore [21] , we here introduce a protein ID method that according to simulation remains robust against the expected experimental errors . We show that relatively low-resolution , tri-color , optical fingerprints produced during the passage of proteins through a nanopore , preserve sufficient information to allow a deep-learning classification algorithm to accurately identify the entire human proteome with >95% accuracy . Even in cases where the apparent spatial and temporal resolutions of the optical system appear to be prohibitively low , and the amino-acids labelling efficiency is incomplete , whole proteome ID efficiency remains high and robust . Particularly , the expected protein ID efficiency is of an extremely high clinical relevancy . We illustrate the broad applicability of the method by analyzing the human plasma proteome , as well as commercially-available cytokine identification panel based on antibodies , showing that our antibody-free method can readily surpass current techniques in a number of key parameters , while displaying a near perfect accuracy .
In our method , proteins extracted from any source ( serum , tissue or cells ) , are denatured using urea and SDS ( Fig 1A ) . Three amino-acids lysine ( K ) , cysteine ( C ) and methionine ( M ) are labeled with three different fluorophores using three orthogonal chemistries: the primary-amines in lysines are targeted with NHS esters; thiols in cysteines are targeted with maleimide groups , and methionines are labeled using the two-step redox-activated chemical tagging [22] . The negatively charged SDS-denatured polypeptides are electrophoretically threaded , one at a time , through a sub-5 nanometer pore fabricated in a thin insulating membrane to ensure single file threading of the SDS-coated polypeptide . The voltage , nanopore diameter and other factors , such as solution viscosity are used to regulate the protein translocations speed . The nanopore is illuminated using laser beams for multi-color excitation [23] . The excitation volume ( Fig 1A , yellow highlighted region ) is centered with the nanopore , and importantly , its axial depth is confined by plasmonic focusing of the incident electromagnetic field [24] . Consequently , depending on the excitation depth , either a single or multiple labeled amino-acids will be simultaneously illuminated , during the passage of the protein . Three-color fluorescence time traces ( “fingerprints” ) are recorded for each protein passage and are classified using deep-learning ( Fig 1B ) . The theoretical likelihood of protein ID can be tested by calculating the percentages of unique matches of all proteins in the human Swiss-Prot database [25] based on the number and the order of appearance of three amino-acids only . Simply counting the number of K , C and M residues in each protein identifies 72% of the total proteins uniquely , and another 14% identified as either one of two proteins in which one of them is the correct match ( online methods ) . Moreover , the percentage of uniquely identified proteins is close to 99% with the determination of the KCM order of appearance along all proteins in the human proteome database ( Fig 1C ) . Thus , in principle , the boundaries for the expected ID accuracies fundamentally permit whole-proteome , single-protein , identification . The theoretical analysis shown in Fig 1C may be considered as an upper limit for the accuracy of a protein ID method based on a three amino-acid labelling , which neglects inter-dye distances . However , it ignores experimental limitations , such as the sensing spatial and temporal constraints , the labelling efficiency and the photophysical properties of fluorophores . These factors are likely to impact the accuracy of the protein ID method , and hence must be considered . To this end we developed a detailed photophysical model to numerically calculate the time-dependent photon emission during the passage of each SDS-denatured protein through a solid-state nanopore . Our model consists of three layers: first , we used Finite Difference Time Domain ( FDTD ) computations to evaluate the expected electromagnetic field distribution for a simple plasmonic structure fabricated on top of the nanopore ( Materials and Methods ) . Second , an amino-acid labelling simulation was applied to each protein , in order to generate partial labelling of each of the three target amino-acids . Finally , SDS-denatured proteins were allowed to slide through the plasmonic nanopore complex while illuminated at three distinct wavelengths . The expected detected photon emissions were calculated at each step of the protein translocation taking into account the photophysical properties of the fluorophores , as well as energy transfer ( FRET ) , bleaching kinetics and collection efficiencies . This allowed us to generate detailed photon emission time traces for each and every protein translocation . To illustrate our method , we schematically show in Fig 2A snap-shots of the system at two time points during the passage of the PSD protein . This figure is plotted in scale to illustrate the relative dimensions of the plasmonic field , the nanopore and the SDS-coated polypeptide chain ( marked as orange layer around the chain ) . Specifically , the axial FWHM of the plasmonic field is 20 nm calculated from the FDTD field distribution , and the nanopore diameter is 3 nm . Each protein was modeled as a fully-denatured , SDS-coated , wormlike polymer [26] , translocating across the nanopore at an instantaneous velocity ui = 〈u〉+δui where 〈u〉 is its average velocity , and the random term δui accounts for thermal fluctuations in its motion . Since the SDS-coated biopolymers have a Kuhn length of approximately 7 nm [26] , they can be assumed to be partially-stretched ( unfolded ) wormlike polymers during translocation through a sub ~5 nm pore . Moreover , when threaded through a 3 nm pore , the roughly 2 nm wide SDS-coated proteins are confined laterally in a small volume in the nanopore proximity where the electromagnetic field remains nearly constant . Hence , in this study the protein translocations can be treated as one dimensional [27] . The excitation profile calculated from the FDTD simulations was approximated by a one-dimensional Gaussian function as shown in S1 Fig . The fluorescence emission rate of each labeled amino-acid while passing through the excitation zone was modeled as a two-state system ( Fig 2C ) , as described in the Materials and Methods section . Triplet state transition rates , which may result in microsecond-long dark-states were also investigated ( equations not shown ) based on literature values of three specific fluorophores [28–30] . We explicitly took into consideration energy transfer rates ( Fig 2B and 2C ) , which directly depend on the amino-acid sequence , as well as photo-bleaching rates ( indicated by dotted yellow lines and solid grey arrows in Fig 2 , respectively ) . At each time step of the simulation the emitted light from all fluorophores residing in the excitation zone were split to three spectrally-resolved , photon-counter channels as shown in Fig 2D . In addition to the collection and detection efficiency of each channel , we also considered photon statistics by incorporating shot-noise . The labeling efficiency was modeled by randomly positioning fluorophores at the K , C and M amino-acid , such that in each protein only a fraction Γj of them ( j represents K , C or M ) was actually labelled ( indicated by purple arrows in Fig 2A ) . In all the following computational results presented the three amino-acids , K , C and M were labelled by Atto488 , Atto565 and Atto647N fluorophores , and the fluorophores properties were taken into account when simulating the photon emission rates . Additionally , we introduced cross-labelling efficiency ( green arrows in Fig 2A ) , although this is known to be negligible [31] . In order to estimate the translocation velocity of SDS-denatured polypeptides we performed electrical translocation measurements using SDS-denatured albumin ( 585 amino-acids ) proteins using ~4 nm-wide solid-state nanopores , as described in the Materials and Methods section . Representative translocation events measured at a bias voltage of V = 300 mV , in which a single blockage current level is observed , are shown in Fig 3A . Examining a statistical set of >900 translocation events showed a single blockade current level ( IB = 0 . 7 ) indicative of single-file polypeptide translocations . This experiment supports the assumption that proteins are likely to be fully denatured as they thread through the narrow nanopore , in agreement with a previous publication [20] . Fig 3B displays an overlay of the scatter plot of the fractional blockade current IB versus the translocation dwell-time tD , with its corresponding density map . The area delimited by the dashed red lines approximate the typical full-width-half-maximum of a Gaussian centered on the characteristic dwell-time ( 94 . 3±7 . 2 μs as determined by the histogram shown in the inlet panel ) . Accordingly , we estimate the mean translocation velocity by 0 . 2 cm/s . Notably , this velocity is slower than the previous report , presumably due to the fact that in our experiments a much smaller nanopore was used . We first focus on the simulated optical signals calculated for two proteins having nearly the same length: the EGF precursor , and its receptor EGFR ( 1208 and 1210 amino-acids , respectively ) . Under near-ideal experimental conditions ( 100% labelling , 0 . 5 nm resolution , and velocity of 0 . 035 cm/s ) their tri-color fingerprints were readily distinguishable from each other , despite similar K , C and M compositions , and followed the actual K , C , M amino-acid order in each protein ( Fig 4A ) . We then extended our protein translocation simulations under much lower spatial resolutions , lower labelling efficiencies and higher translocation velocities . As expected , in the more realistic conditions we no longer can resolve individual fluorophore photon bursts , associated to single K , C or M residues . Instead , the resulting signals appear as continuous tri-color fingerprints of each protein translocation . Importantly , however , the fingerprints , even at the poorest resolution of 50 nm maintain an overall pattern characteristic of each protein ( Fig 4B ) . Analyzing >5·107 single protein translocations events , under different conditions suggest that even at 100 nm resolution some characteristic features of each protein are preserved ( S2 Fig ) . Moreover , we expect that small variations in the nanopore size would result in different translocation velocities . To evaluate this effect , we repeated the translocation simulation experiments at mean values of 0 . 035 , 0 . 2 and 2 cm/s and increasing the translocation velocity fluctuations ( 20% , 30% and 40% of the mean velocity ) . Our result presented in ( S3 , S4 & S8 Figs ) suggest that as long as the velocity is in the order of ~ 0 . 2 cm/s ( or below ) in accordance with our experimental result ( Fig 3 ) , the identification accuracy remains sufficiently high . We tested the similarity among repeated translocations of the same proteins , which were subject to different labeling and random velocity fluctuations , by evaluating the Pearson correlation coefficients between all pairs of 50 translocation repeats of the same protein . The results , showed in all cases high values ( 0 . 85–0 . 97 ) when considering auto-correlation ( Fig 5 , diagonal values ) . In contrast , attempting to cross-correlate among 5 different , randomly-chosen , proteins produced in most cases much lower Pearson coefficient values ( 0 . 03–0 . 35 ) . Obviously , this is just a small fraction of all possible cross-correlations . However , even as is , this sample of data suggests that the protein translocation simulator generates highly-reproducible signals . Next we vastly scaled-up our simulations to include thousands of different proteins , each one repeated hundreds of times under different labeling efficiencies , translocation velocities and spatial resolutions . The accurate classification of noisy , low-resolution , time-dependent signals is often encountered in areas such as image and speech recognition and is effectively handled by Convolutional Neural Networks ( CNN ) approaches [33 , 34] . We postulated that provided sufficient training , the CNN would be able to identify most proteins based on the tri-color fingerprints . To check this hypothesis , we set up deep-learning whole-proteome analyses . First , we trained the CNN network using a large data set containing at least 80 individual nanopore passages of each protein in the Swiss-Prot database . Then the CNN was presented with new protein translocation events and queried as to the protein identity . This procedure was repeated at least 5 times for whole-proteome analysis allowing us to establish the mean ID accuracy and its standard deviation , for 16 different experimental conditions ( Fig 6A ) . Starting with the highest labelling efficiency ( 90% , right-hand set ) we observed that 96%-97% of all protein translocations were correctly identified , as long as the spatial resolution was ≤50nm . The correctly identified protein fraction dropped down to 92% using a 100 nm resolution . A similar pattern can be observed for the other labelling efficiencies with somewhat lower numbers . In the worst-case scenario considered here ( 100 nm resolution and only 60% labeling efficiency ) the CNN nevertheless was able to correctly classify 68% of all translocation events , similar to the ideal case considered in Fig 1C , ( C , K , M counts only ) . In other words , despite the fact that 40% of the target amino-acids were not labeled , and the resolution of the probing was about a third of the optical diffraction limit , the pattern recognition algorithm identified correctly nearly 70% of all protein translocation events . When the labelling efficiency was improved to the expected standards ( between 70%-90% ) [22 , 35] , and the sensing resolution assumed to be in the 20–30 nm , the correct identification of all translocation was roughly 95% . Increasing the translocation speed of proteins by nearly two orders of magnitude to 2 cm/s ( an order of magnitude higher than the mean measured velocity in Fig 3 ) , reduced the ID accuracy ( S8 Fig ) . However , for high labeling efficiencies ( 80% and 90% ) the ID accuracy was high ( 72% and 81% , respectively ) . In addition to the mean accuracies , the CNN algorithm produces a “confusion matrix” , which presents the number of times each and every protein x was identified as protein y ( where x and y could be any of the proteins in the set ) . We used this information to calculate the probability density function ( pdf ) of correct ID for each and every classification set , namely the likelihood that a given protein is correctly identified with probability p . The pdf of correct ID calculated for the case of 30 nm resolution and 80% labelling efficiency ( Fig 6A right panel ) indicates that 51% , 71% and 89 . 2% of proteins were correctly identified with probability of 1 . 0 , 0 . 98–1 . 0 and 0 . 9–1 . 0 , respectively . The probability distributions for all other conditions are shown in SI S5 and S6 Figs . We also analyzed the results for misclassified proteins . Specifically , we were interested to know whether a misclassified protein is likely to be deterministically or randomly misclassified . To investigate the degree of randomness in misclassification , we first selected proteins that had at least 10% misclassified events . Then , we determined the fraction of identical mismatch ri = maxjnij/Ni for each protein i , where nij is the number of translocation events misidentified to protein j and Ni the total number of misclassified translocation events . With this a high ri was characteristic of a deterministic misidentification , i . e . protein i is consistently mistaken with another specific protein j , and conversely a low ri was indicative of a rather random misidentification . As shown in the right panel of Fig 6A , proteins were often confused with several others , suggesting a relatively high degree of randomness in misclassification , while only 10% were consistently mis-identified , that is with the same partner . The distributions for all other conditions are shown in SI S5 and S7 Figs . We further evaluated the performance of our approach for clinically-relevant applications including whole human plasma proteome and a cytokine panel . In both studies , we kept the CNN training at the whole human proteome , rather than restricting it to the clinical sub-set . Then we presented nanopore translocation traces of the plasma/cytokines proteins and evaluated the classification accuracy as before . Interestingly for the high-spatial resolutions ( 20 nm and 30 nm ) the correct ID of the 3852 plasma proteins was only slightly larger than the whole proteome accuracy at the different labelling efficiencies , reflecting the fact that there is a small set of proteins that are hard to be classified in both cases ( Fig 6A and 6B right panel ) . However , at the lower resolutions , especially for the 100 nm case in which we observed a significant drop in the ID accuracy for the whole proteome results , we still obtained very high scores for the plasma proteome . Even at the lowest labelling efficiency of 60% at 100 nm resolution the CNN classified correctly 93% of all translocations ( Fig 6B ) . In addition , the fraction of proteins correctly identified with probability between 0 . 9–1 . 0 improved over that of the whole-proteome classification , reaching 96 . 8% for the case of 30nm resolution and 80% labeling efficiency . Finally , close to 30% of mis-identified proteins were consistently mistaken with another specific partner , suggesting that the accuracy of classification could be further significantly improved by relaxing the requirements of correct ID for selected proteins . These results indicate that single-molecule plasma proteome application , which holds great clinical value , does not require extremely-stringent experimental resolutions or super-efficient labelling chemistries ( S9–S11 Figs ) . The cytokine panel ( CytokineMAP [36] ) contains 16 proteins involved in inflammation , immune response and repair . We evaluated the CNN classification under 16 different experimental conditions ( Fig 6C ) . At the lowest labelling efficiency of 60% the ID accuracy drops between 43% - 85% , and at the realistic 80% labelling we obtain correct ID in the range of 73% - 97% . However , despite the functional similarity between the candidate cytokines , and the wide range of conditions tested , each was distinguishable from all other cytokines within the commercial test panel . This indicates that our approach has the potential to meet the requirements of a broad range of clinically relevant applications–that are less demanding than whole-proteome identification–with extremely high accuracies and yet very poor experimental conditions ( S12 Fig ) .
Single-molecule protein ID and quantification techniques are on the verge of revolutionizing the field of proteomics by enabling researches to achieve single-cell proteomics and to identify low abundance proteins that are essential biomarkers in biomedical and clinical research [7] . Specifically , nanopore discrimination among poly-peptides based solely on two color labeling of C and K residues has recently been demonstrated [21] . Here , we have proposed and simulated the feasibility and limits of a novel method for single-molecule protein ID and quantification using tri-color amino-acid tags and a plasmonic nanopore device . Specifically , we designed a simulator that incorporates a range of physical phenomena to predict and model the behavior of our proposed device and performed a computational analysis taking into account a broad range of experimental conditions to characterize its performance . Importantly , we developed a whole-proteome single-molecule identification algorithm based on convolutional neural networks providing high accuracies ( >90% overall ) , reaching up to 95–97% in challenging but attainable experimental conditions . To facilitate the computational efforts , in this study we approximated each protein translocation dwell-time using a Gaussian distribution function . Notably , past studies [37] successfully utilized CNN to identify signals from exponentially-distributed time-dependent signals , which may better reflect the experimental dwell-time distribution ( Fig 3 ) . However , further studies will be required to evaluate the full impact of the temporal distributions of proteins translocation dwell-time on the CNN identification accuracy . In clinical samples lysine residues may be post-translationally modified hence reducing their labelling efficiency . To account for this effect and for the limitations in the chemical labelling yield , we evaluated the protein identification accuracy under partial labelling conditions . Our results ( Fig 6 ) show that our tri-color protein identification method nevertheless largely circumvents this potential issue , yielding very high accuracies for up to 40% of unlabeled residues . This is attributed to a redundancy in the tri-color labelling scheme that provides a higher degree of robustness against partial labelling . Solid-state nanopores can process tens of individual proteins per second , and importantly because our method does not rely exclusively on measurements of the ion-current through the pore , it lends itself for parallel readout of high-density nanopore arrays fabricated on a sub mm2 membranes , using multi-pixel single-photon sensors [38] . The versatility and robustness of convolutional neural networks tremendously simplify any calibration procedures and even potentially allow protein ID based on partial reads [39] . This ensures that the whole-proteome ID is reliable and compatible with a wide variety of systems , able to overcome real experimental challenges . Furthermore , in many cases ( notably for the plasma proteome ) misidentified proteins were consistently confused with another specific protein , which in a broad range of applications such as identifying disease-specific biomarkers , may not pose a significant issue as only small-subsets of the proteome are considered , or since the quantification of proteins can be cross-examined with expected counts ( e . g . low , medium or high abundance ) . Finally , we evaluated the expected efficacy of our approach with commercially available applications , even resolving functionally similar proteins in rather poor experimental conditions .
The theoretical identification values were calculated using the human proteome Swiss-Prot database , which contains 20 , 328 entries . For each entry we extracted the number of the target amino-acids ( C , K and M ) , as well as their order of appearance . For example , the p53 protein would either be characterized by its C , K , M counts ( 10 , 20 , 12 , respectively ) or by the sequence below: MKMMMKKCKMCKCMKMCCCMCCMMCCKKKKKKMKKKKKKKMK , in which all intervening amino-acids were deleted . Proteins having identical characteristic sequences ( or C , K and M counts ) are grouped together . A protein is identified when it is the sole member of a group . In the case of p53 , both the C , K and M counts and the characteristic sequence gave a unique identification . The pie charts ( Fig 1C ) distribute the proteins according to the size of the group in which they belong to . Each protein primary sequence was transformed into a string ( B ( i ) ) to which we assigned a value of 1 , 2 or 3 corresponding to each of the three aa tags ( K , C , and M ) , respectively; and 0 for all other aa in the protein sequence . To account for partial or nonspecific labelling a set of randomly selected labeled positions in the string were omitted according to a given labeling efficiency ( ηL ) , and a set of artificial labeled positions were inserted according to a given nonspecific labeling efficiency ( ηNS ) . It is important to note that nonspecific labeling did not affect all aa equally . For instance , in generating a barcode for lysine ( K ) positions , nonspecific labeling could only be inserted at positions of either threonine , serine and tyrosine ( amino-acids which have been shown to compete with NHS-ester-based labeling ) with a probability of typically 1% [31] . The strings were generated for the entire Swiss-Prot data base , and were re-generated each time to simulate an uneven labelling of the same protein data sets , as well as whenever we used different values of ηL and ηNS . The three-dimensional near field enhancement of the plasmonic structure ( 2D vertical cross-section shown in Fig 2A ) was determined using a finite difference time domain ( FDTD ) [40] method solving for Maxwell’s time-dependent electromagnetic equations . The architecture over which the FDTD computations were performed comprised a 10 nm-thick silicon ( Si ) membrane–exhibiting a 3 nm-wide nanopore–on top of which a gold ( Au ) plasmonic structure was deposited ( Fig 2D ) . An additional 2 nm-thick titanium oxide ( TiO2 ) adhesive layer was inserted in between the Au structures and underlying Si membrane . The plasmonic structure consisted of a gold ring ( inner and outer diameter of 12 and 32 nm , respectively , and a height of 40 nm ) centered at the nanopore and embedded inside a gold nanowell ( diameter of 120 nm and a height of 100 nm ) . Water was used as the immersion media . The excitation field was modeled as a total-field scattering-field source ( TFSFS ) [41] and the spatial sampling frequency was set to 5 nm-1 ( taking 60 frequency points over the 500–800 nm wavelength range ) . The FDTD boundary conditions consisted of 8-layer PMLs ( perfectly matched layers ) symmetric in the x axis and antisymmetric in the y axis thus minimizing the reflections and the computational cost , respectively . Frequency domain power monitors only were incorporated in the simulation to determine the near field enhancement in the vicinity of the nanopore . All numerical simulations were performed using Lumerical FDTD Solutions ( Lumerical , Inc ) . To simulate the translocation of the linearized protein through the nanopore , we assumed a unidirectional motion with steps of a single aa length ( Δ≈0 . 35 nm ) and an average velocity u ( cm/s ) . To account for thermal fluctuations in this process we added a random noise term δu at each step ( δu can be positive or negative ) . Hence the simulation step time of the i-th aa was defined as τi = Δ / ( u + δu ) . The average protein velocity value was typically ~0 . 2 cm/s , based on experiments using SDS denatured proteins in solid-state nanopores as shown in Fig 3 . Additionally , we tested faster translocations ( 2 cm/s ) . The fluorescence emission rate of each fluorophore n in our system Kfl , j , n ( t ) was modeled as a two-state system: Kfl , j , n ( t ) =kfl , jPj , n ( t ) Eq 1 where j = 1 . . 3 correspond to each of the three excitation/emission channels , kfl the fluorescence transition rate and Pn ( t ) the occupation probability of the excited molecular state S1 . The fluorophores are excited by up to three laser lines corresponding to the three channels , that form sub-wavelength excitation volumes by means of a plasmonic nanostructure or total internal reflection . The axial full width at half maximum of our Gaussian excitation volume Iex is defined as ξ and is allowed to vary from 5 nm to 200 nm in order to account for broad possible experimental conditions . The emitted light from the three-color channels is assumed to be acquired with given efficiencies ηj , which include both the optical transmission efficiencies and the photodetector efficiencies . The photon counts Iij at each channel j during each step i of the protein translocation is then determined by summing the emissions of all the fluorophores n that resides within the excitation volume . Namely: Iij=ηj∑nKfl , j , n ( ti ) +kbgτi=ηj∑nkfl , jPj , n ( ti ) +kbgτi Eq 2 {Pj , n ( ti ) =Pj , n ( ti−1 ) + ( kex , j ( n ) kj ( n ) −Pj , n ( ti−1 ) ) ( 1−e−kj ( n ) ti ) kj ( n ) =kex , j ( n ) +kS1 , j=σex , jIex , j ( n ) λex , jhc0+τS1 , j−1 Eq 3 , Eq 4 where kbg is the background emission rate , ti the time at which step the translocation occurred such that ti−ti-1 = τi , kex , j ( n ) is the excitation rate of the fluorophore n of channel j , σex , j is its absorption coefficient , λex , j is the excitation wavelength and τS1 , j is its excited state lifetime . The number of cycles ( S0→S1→S0 ) undergone by each fluorophore was capped to account for photobleaching according to a decaying exponential distribution . Specifically , the maximum number of cycles performed by each fluorophore before photobleaching was given by a random number drawn from a decaying exponential distribution with a characteristic decay of ~106 . Finally , we applied a Poisson distribution to the photon counts Iij to simulate shot noise . To include energy transfer ( such as Förster Energy Transfer and homo-transfer ) in our system we calculated a 2D distance matrix for each fluorophore in our system . The distances between the labelled aa’s ( or fluorophores ) in each linearized protein were subsequently used to calculate the Förster energy transfers of each fluorophore from and to each of its neighboring emitters . As a proxy for the exact energy transfer , two additional transition rates accounting for energy gain and loss were incorporated in the fluorophore two-state model: {kFRET+ , j ( n ) =1hc0∑i∑m≠nσex , iIex , i ( m ) En←mλex , ikFRET− , j ( n ) =σex , jIex , j ( n ) λex , jhc0∑i∑m≠nEm←n Eq 5 , Eq 6 where Em←n = ( 1+ ( |xm−xn|/R0 , m←n ) 6 ) −1 is the FRET energy transfer efficiency from fluorophore n to m , xn is the position of fluorophore n along the denatured protein and R0 , m←n is the Förster-radius of the ( n , m ) dye pair when considering an energy transfer from fluorophore n to m . The transition rates kex , j ( n ) and kj ( n ) in Eq 4 were corrected to account for FRET accordingly: {kex , j ( n ) →kex , j ( n ) +kFRET+ , j ( n ) kj ( n ) →kj ( n ) +kFRET+ , j ( n ) +kFRET− , j ( n ) The code was implemented using MATLAB , and the optical readouts of the three channels were determined by running this procedure for each labeling string . For the purpose of a multi-class ( the human proteome comprises more than twenty thousand proteins ) classification of time-series that exhibit specific patterns , we used convolutional neural networks ( CNN ) that have shown great promise in the field of pattern recognition , including image classification , which similarly requires tens of thousands of classes [42 , 43] . Specifically , we used the python deep learning package Keras on a four GPU architecture ( NVIDIA Tesla K40 ) , which leads to a CNN whole-proteome training time of ~2 h only . The CNN model relied on four sequential layers–a convolutional layer , a normalization layer in which dropout was applied and a pooling layer–followed by a multi-layer perceptron . In brief , the convolutional layer filters ( at a given step or stride size ) the translocation time-series with a large set of kernels of a specific size . The resulting activation or feature map it provides is further transformed by the normalization layer such that the mean and standard deviation of the activation map approach zero and one , respectively . Next , the dropout circumvents overfitting of the CNN to the training dataset by setting a random subset of activations to zero . The last pooling layer performs a down-sampling operation on the activation map to further prevent overfitting of the training dataset and the computational load . The multi-layer perceptron consists of a single densely-connected neural network layer , each neuron outputting the probability of belonging to the class it represents ( ‘softmax’ activation function ) . The hyper-parameters were optimized according to standard procedures , that is maximizing the accuracy of the CNN trained over five to ten epochs per hyper-parameter set . Once finely adjusted , the CNN was trained using twenty epochs to yield the greatest accuracy . The protein identification accuracy as determined by the CNN was calculated as the fraction of correctly classified translocation events from the test dataset . We partitioned randomly the dataset into five pairs of training and testing sub-sets , and for which we determined the identification accuracy . The final accuracy was calculated as the average between them where a typical test set included ~400 , 000 translocation events . Solid-state nanopores were fabricated using a laser drilling method in 17 nm-thick SiNx membranes as described previously [44] . Human serum albumin ( Biological Industries Inc . 30-O595-A ) was first treated by TCEP ( 5 mM ) at room temperature for 30 min to break disulfide bonds and subsequently denatured at 90°C for 5 min in PBS with 2% sodium-dodecyl sulfate ( SDS ) . The resulting albumin concentration was further diluted ( 100:1 ) to <1 nM in buffer ( PBS/0 . 4M NaCl/ 0 . 1% SDS/ 1mM EDTA ) for nanopore translocation experiments performed under a 300 mW bias . A custom-made LabVIEW interface was used to acquire and analyze each event . Scatter plots and dwell-time distributions were generated using Igor Pro ( Wavemetrics ) . | Macromolecules identification methods are central for most biological and biomedical studies , and while the field of genomics advanced to single-molecule resolution , the proteomic field still relies on bulk and costly techniques . We describe a solution for single protein identification , based on the analysis of optical traces obtained from fluorescently-labeled proteins threaded through a nanopore and processed by a pattern recognition algorithm . To evaluate the feasibility of our method we constructed computer simulations of the system , producing and analyzing nearly 108 individual protein translocations from the human Swiss-Prot database . Our results suggest protein identification of >95% for the whole human proteome , even under non-ideal conditions . These results constitute the basis for a novel whole proteome identification method , with single molecule resolution . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"protein",
"transport",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"cytokines",
"chemical",
"compounds",
"particle",
"physics",
"energy",
"transfer",
"cell",
"processes",
"immunology",
"organic",
"compounds",
"developm... | 2019 | Simulation of single-protein nanopore sensing shows feasibility for whole-proteome identification |
Zic3 regulates early embryonic patterning in vertebrates . Loss of Zic3 function is known to disrupt gastrulation , left-right patterning , and neurogenesis . However , molecular events downstream of this transcription factor are poorly characterized . Here we use the zebrafish as a model to study the developmental role of Zic3 in vivo , by applying a combination of two powerful genomics approaches – ChIP-seq and microarray . Besides confirming direct regulation of previously implicated Zic3 targets of the Nodal and canonical Wnt pathways , analysis of gastrula stage embryos uncovered a number of novel candidate target genes , among which were members of the non-canonical Wnt pathway and the neural pre-pattern genes . A similar analysis in zic3-expressing cells obtained by FACS at segmentation stage revealed a dramatic shift in Zic3 binding site locations and identified an entirely distinct set of target genes associated with later developmental functions such as neural development . We demonstrate cis-regulation of several of these target genes by Zic3 using in vivo enhancer assay . Analysis of Zic3 binding sites revealed a distribution biased towards distal intergenic regions , indicative of a long distance regulatory mechanism; some of these binding sites are highly conserved during evolution and act as functional enhancers . This demonstrated that Zic3 regulation of developmental genes is achieved predominantly through long distance regulatory mechanism and revealed that developmental transitions could be accompanied by dramatic changes in regulatory landscape .
Early embryonic patterning is achieved through a process involving the determination of body axes and defining which cell types develop at each coordinate . The Zic family of transcription factors ( TFs ) is involved in such process [1]–[4] . Zic genes are the vertebrate homologues of the odd-paired gene , which is involved in the generation of segmental body plan in the Drosophila embryo [5] , [6] . Although functions of Zic proteins partially overlap , their loss-of-function cause distinct phenotypes , suggesting unique roles in development [7] , [8] . Of particular interest is ZIC3 , which is linked to the heritable defects of the left-right internal organs placement ( situs inversus ) in humans [9] . Studies in animal models reveal the involvement of Zic3 the establishment of left-right ( L-R ) asymmetry [1] , [10]–[12] . In Xenopus , Zic3 established left-sided expression of Xnr1 and Pitx2 [12] , two determinants of internal organs asymmetry [13]–[15] . However , zic3 is expressed symmetrically along the L-R axis in the Xenopus embryo and its loss-of-function ( LOF ) affects structures in which its expression was not detected [1] , [12] . Results from several studies provided clues to the mechanism of L-R patterning by Zic3 . First , Zic3 acts in organizer formation by inhibiting the canonical Wnt signaling pathway [16] . Second , Zic3 regulates gastrulation in mouse [1] , [17] . Furthermore , studies in zebrafish revealed a correlation between convergence-extension ( C-E ) and L-R patterning defects in Zic3 LOF [10] . These suggest that Zic3 may regulate L-R patterning through its role in an earlier developmental event such as C-E . Zic3 is one of the earliest TFs expressed in the neuroectoderm [3] , [18] . Its expression is regulated by determinants of the early neural fate specification and dorsal-ventral ( D-V ) axis formation , including BMP , FGF , and Nodal signaling [3] , [17] , [19] , [20] . The role of Zic3 in establishing neural cell fate was demonstrated through experiments in Xenopus , where its overexpression resulted in the expansion of the neuroectoderm and induction of neural and neural crest markers [18] . This led to the assumption that Zic3 activates the expression of proneural genes such as Achaete-scute homologs , Neurogenin , and NeuroD [2] . However , Zic3 lacks the ability to induce ectopic neuronal differentiation in the epidermis [18] , which suggested the complex interaction between Zic3 and the proneural genes . Increasing evidence has established the presence of long-distance interactions between TFs and their target genes [21]–[24] . This feature is especially true for TFs regulating specific functions outside of the core transcription machinery [25]–[27] . Therefore , an unbiased evaluation of binding sites throughout the whole genome would be a more comprehensive and biologically relevant method in the context of a developing organism . However , genomic approaches to study TFs in vivo are often limited by the quantity of available tissue sample . Furthermore , in mammalian systems , this problem is exacerbated by the short supply of embryos at early developmental stages . The zebrafish , with its unlimited supply of embryos and external development , substitutes for the inconveniences of a mammalian system . Its genome annotation is also the most complete among non-mammalian vertebrates and the expression of many genes are well-defined . This makes the zebrafish a robust model system for functional studies of vertebrate development . To understand the developmental role of Zic3 , we applied a genomic approach to identify genes directly regulated by Zic3 . To capture genome-wide binding sites of Zic3 , chromatin fragments bound by Zic3 were immunoprecipitated from gastrulating embryos at 8 hpf and zic3 expressing cells were sorted from transgenics [21] , [28] at 24 hpf and sequenced in-depth using ChIP-seq methodology . This provided unbiased coverage of Zic3 binding events during the period of gastrulation and segmentation . We used microarray expression profiling to characterize changes at the transcription level as a result of Zic3 LOF during gastrulation . In addition , we compared gene expression profiles of zic3-positive and -negative cells at 24 hpf to identify genes co-expressed with zic3 . Combining binding site analysis and expression data , we demonstrated that Nodal and Wnt pathways are the main downstream targets of Zic3 during gastrulation , and show distinct pathways regulated by Zic3 in the dorsal neural tube at the end of segmentation . Finally , in vivo enhancer assay validated selected binding sites as developmental enhancers . Our results provide novel insights into the molecular mechanism underlying Zic3 regulation of developmental events during gastrulation and neural development , which ultimately results in the L-R patterning and neural fate specification and patterning .
The earliest zic3 transcript was detected at 3 hpf ( Fig . 1A , B ) , coinciding with the initiation of zygotic transcription during mid-blastula transition [29] . At 4 hpf zic3 expression is restricted to dorsal blastoderm ( Fig . 1C , C′ ) , and is subsequently found in the dorsal neuroectoderm and marginal blastomeres ( Fig . 1D , D′ ) . To capture genome-wide Zic3 binding profile during zebrafish gastrulation , we performed ChIP-seq analysis at 8 hpf , a time coinciding with the beginning of neurogenesis [30] . At this time zic3 is expressed largely in the dorsal neuroectoderm ( prospective neural plate ) and blastoderm margin ( presumptive mesendoderm; Fig . 1E , E′; [3] ) . Hence , the interaction of Zic3 with its targets could be considered within a context of neural induction and mesendodermal development . Although neuroectoderm does not show any obvious morphological organization at this time , its anteroposterior patterning at the molecular level was shown by fate mapping studies [31] and in vitro explant assays [32] , [33] . At 24 hpf zic3 is expressed in the brain and dorsal spinal cord ( Fig . 1F , F′ ) . To identify Zic3 binding sites specifically in zic3-expressing cells , we performed ChIP-seq using sorted cells from transgenic line SqET33 [28] , [34] at this stage . Since gfp expression in this line faithfully recapitulates zic3 expression ( Fig . 1G–H″ ) , we considered GFP-positive cells as zic3-expressing cells and GFP-negative cells as non- zic3-expressing cells . However , it is worth to note that in SqET33 line at least one zic3-positive domain ( presomitic mesoderm ) does not express GFP . This suggests that a small fraction of non-neuronal zic3-expressing cells may be present in the GFP-negative pool of cells . Sequencing of the 8 hpf ChIP sample generated 23 , 945 , 552 reads ( 11 , 037 , 221 or 46% were mapped to the zebrafish genome ) ; the 24 hpf ChIP sample generated 23 , 083 , 504 reads ( 11 , 797 , 011 or 51% were mapped ) . We identified 3209 and 2088 Zic3 binding sites ( hereafter referred to as peaks ) with high significance value at 8 hpf ( Table S13 ) and 24 hpf ( Table S14 ) , respectively . Interestingly , both datasets showed that only a small fraction ( 8 . 6% at 8 hpf and 4% at 24 hpf ) of the peaks mapped to promoter regions ( within 5 kb of transcription start site , TSS ) , while the rest were aligned to intragenic ( 26 . 8% at 8 hpf and 29% at 24 hpf ) and intergenic ( 64 . 6% at 8 hpf and 67% at 24 hpf ) regions ( Fig . 2A ) . This suggested that Zic3 mainly acts via distal regulatory elements . To validate the ChIP-seq performance , we carried out quantitative PCR ( qPCR ) on randomly selected peaks from the 8 hpf dataset , five within promoter region and sixteen at regions outside of gene promoters . Taking a fold-change of 2 as a cutoff for positive enrichment , the qPCR analysis validated all but one peak tested ( Table S1 ) . To determine the biological relevance of our data , we used the gene association rule ‘basal plus 100 kb extension’ according to GREAT algorithm [35] ( Fig . 2B ) . Using this criterion , the number of peaks associated with either none , one , or two genes were evenly distributed in both 8 hpf and 24 hpf datasets ( Fig . 2C ) . Distribution of the peaks relative to the TSS of genes associated with them showed strong bias towards regions beyond 5 kb of the TSS ( Fig . 2D ) . In agreement with known Zic3 functions at 8 hpf [10] , [16] , [18] , [36] functional categories enriched were embryonic morphogenesis , gastrulation , and dorsal/ventral pattern formation ( 2835 genes , Fig . 2F; Table S2 ) . Enrichment was also observed for neural tissue-specific genes , predominantly expressed in the neuroectoderm at 8 hpf ( Fig . 2G ) . In contrast , at 24 hpf , different categories were enriched ( neural crest development and migration , nervous system development; Fig . 2H , I ) in agreement with these events of neurodevelopment taking place at this stage [18] , [37] . To identify the common regions bound by Zic3 as well as those unique to either developmental stage , we overlapped the 8 hpf and 24 hpf peaks ( Fig . 2E ) . Taking the combined list of peaks from 8 hpf and 24 hpf , we performed clustering using ChIP-seq signals around the peaks . We found 937 regions bound by Zic3 at both stages ( class I ) , 2729 regions bound only at 8 hpf ( class II ) , and 1630 regions only at 24 hpf ( class III ) . A clear distinction of functional categories was observed among genes associated with each individual class ( Fig . S2 ) , which reflect the shift of Zic3 function from regulating gastrulation at 8 hpf , to directing neurodevelopment at 24 hpf . To identify the consensus motif in Zic3-binding sites , we performed de novo motif search using sequences within 50 bp ( total length 100 bp ) of the top 1000 peaks summit . The highest scoring motif in both datasets consisted of a CAGCAG core ( Fig . 3A ) and was similar to that previously identified in mouse ES cells using ChIP-chip [38] ( Fig . S3A ) and Zic3 motif in UniPROBE database [39] . This motif occurred in 48 . 5% ( 1556/3209 ) of 8 hpf peaks and 54 . 3% ( 1134/2088 ) of 24 hpf peaks ( Fig . 3B ) . This consensus motif was bound in a dose-dependent manner by a recombinant protein encompassing the Zic3 DNA binding domain ( Zic3_ZF2-5; Fig . 3C ) . This binding was reduced upon introducing three-point mutations to the motif , confirming binding specificity . The mouse Zic3 recombinant protein mZic3-DBD-HisMBP [38] also recognized the consensus motif derived from the zebrafish genome ( Fig . S3B ) , demonstrating cross-species conservation of Zic3 consensus motif . On the other hand , two other motifs enriched in the dataset to a lesser extent were not specifically recognized by Zic3_ZF2-5 recombinant protein ( Fig . S3C ) . Enrichment of these motifs among the identified peaks might signify an indirect binding of Zic3 to these sequences through interaction with other TFs . Interestingly , Gli motif was found in both 8 hpf and 24 hpf datasets ( 273 peaks , 8 . 5% in 8 hpf; 203 peaks , 9 . 7% in 24 hpf; Fig . 3B ) . More than half of peaks containing Gli motifs also had an adjacent consensus Zic3 motif at both developmental stages , in support of interactions between Gli and Zic3 [40] , [41] . To identify Zic3 target genes during gastrulation and early neural development , we profiled the transcriptome of 8 hpf embryos after Zic3 morpholino ( MO ) -mediated knockdown . Embryos injected with the same MO dosage as in Cast et al . [10] exhibited similar gastrulation and convergent extension ( C-E ) defects ( data not shown ) . However , to minimize the detection of non-direct targets in microarray , we injected the embryos with a lower dose of MO ( 1 . 7 ng in our experiments versus 7 . 5 ng in [10] ) which did not cause visible morphological defects during gastrulation ( refer to Methods section ) , but affected heart laterality and caused curvature of the A-P axis at later stages ( Fig . 4A ) . These phenotypes were rescued by co-injection with Zic3 mRNA which , when injected alone , had little effect ( Fig . 4B ) . This confirmed the specificity of the phenotypes caused by Zic3 MO injection . We identified 1316 genes differentially expressed in MO injected embryos ( morphants , fold change >1 . 2; p≤0 . 05; Table S3 ) . GO analysis revealed prominent enrichment in functions related to embryonic morphogenesis ( Table S4 ) . When the same or higher dose of MO ( 3 . 4 ng ) was injected , the expression of several representative genes showed similar trend when measured by qPCR . This validated a possibility of their regulation by Zic3 ( Fig . 4C; Table S7 ) . We then determined the presence of Zic3 binding peaks within 100 kb of the TSS of these differentially expressed genes , which we defined as a selection criterion for Zic3 target gene . Based on this selection , 454 genes out of the total 1316 were identified as putative targets of Zic3 ( Table S5 and Table S6 ) . This set contains genes of the Nodal signaling pathway such as oep , lft1 and pitx2 ( Fig . 5 ) . While the presence of Zic3 binding in association with oep suggests direct regulation of Nodal pathway , the association of Zic3 peaks with lft1 and pitx2 suggests that Zic3 could also regulate the pathway through its modulators [42] , [43] . These three genes , along with other members of this pathway not associated with Zic3 peaks ( foxh1 , bon , and gsc ) , were concurrently upregulated in Zic3 morphants ( Fig . 4C; Table S3 ) suggesting negative regulation of the Nodal pathway by Zic3 . Inhibition of Nodal signaling indicates suppression of endodermal fate [15] , [44]–[46] . This correlated with broader expression of endodermal marker sox17a in 8 hpf Zic3 morphants ( Fig . S4A ) . The inhibition of endodermal development by Zic3 is in line with previous observation in murine ES cells [38] . Similarly , peaks were associated with three genes of the canonical Wnt signaling pathway: axin1 , jun , and vent ( Table S5 ) . In support of this association , microarray analysis revealed that the negative regulator of canonical Wnt pathway axin1 was downregulated in Zic3 morphants , while the downstream components jun and vent were upregulated ( Fig . 5; Table S3 ) . The expression of some other members of this pathway ( axin2 and nlk1 ) without association with peaks has changed in Zic3 morphants based on microarray data . This implied that such genes could be the indirect targets of Zic3 . Such observation provided further support for Zic3 regulation of the canonical Wnt pathway . The inhibition of canonical Wnt signaling by Zic3 was previously reported in frogs as a mechanism for organizer development [16] . Interestingly , Zic3 LOF only affected downstream components of these signaling pathways , and not the ligands , suggesting that at 8 hpf Zic3 is more likely to modulate the response to Wnt signaling in the target cells rather than initiation of signaling . Apart from genes previously implicated as targets of Zic3 , the combined ChIP-seq and microarray screen also identified novel candidates . Zic3 peaks were found in association with genes known to regulate cell proliferation in the neural plate , dlx4b and msxe [47] , [48] . These genes perform a function [49] , [50] similar to that of msxc , irx1a , and irx7 , which do not have associated peaks but were nevertheless downregulated in Zic3 morphant ( Table S3; S7 ) . This observation suggests the role of Zic3 in promoting proliferation of neural progenitors at 8 hpf . Since these genes are known to inhibit neural differentiation , we assayed the expression of proneural gene neurog1 [51] in Zic3 morphants at 10 hpf . As expected , neurog1 was upregulated , in concert with the downregulation of her9 ( Fig . 4C; Table S7 ) , which provided further support for Zic3 role as a promoter of proliferation of neural progenitors and repressor of neural differentiation . More interestingly , the novel candidate targets include members of the non-canonical Wnt signaling pathway ( dvl2 , rock2b and invs ) . These genes were co-expressed with zic3 during gastrulation ( Fig . S5A , B ) and were downregulated in the microarray ( Table S3; Fig . 4C ) . One of the non-canonical Wnt pathways , the planar cell polarity ( PCP ) , regulates convergence-extension ( C-E ) [52] and controls the positioning of the motile cilia [53] . The changes in expression of sox17 , ntl , pax3a and sox19a mark correspondingly , endoderm , mesoderm , neural crest and neural plate . The broadening of their expression domains suggested that in Zic3 morphants C-E is affected ( Fig . S4B–D , [10] ) . On the other hand , the disorganized expression of foxj1a and sox17a in the dorsal forerunner cells at an earlier stage indicated abnormalities of their migration in Zic3 morphants ( Fig . S6 ) , which may lead to abnormalities in L-R patterning . A correlation between C-E defects and L-R defects in Zic3 morphant was reported [14] , suggesting Zic3 regulation of these events through the non-canonical Wnt pathway . Several genes implicated in cell migration and polarity were among the targets . These include npy [54] , ptenb [55] , sepn1 , srsf1a [56] , and sparc [57] , [58] , all of which were downregulated in microarray and associated with peaks . WISH analysis showed that their expression overlap that of zic3 ( Fig . S5C; ZFIN; University of Oregon , Eugene , OR 97403-5274; URL: http://zfin . org/; 21 June 2013 ) . In addition , other genes with similar function , such as ccdc88a ( probe generated from BC057440 which correspond to the annotated ccdc88a sequence ) [59] , [60] and tsg101 [58] , were also downregulated in the microarray despite not having associated peaks . Hence the direct and indirect regulation of these genes by Zic3 could be the mechanism behind cell movements during gastrulation . To identify potential zic3 targets during late neurogenesis , we performed microarray expression analysis on 24 hpf GFP-positive zic3 expressing cells that were FACS-sorted ( Table S8 ) . Comparing expression levels to a control dataset derived from GFP-negative cells ( cells negative for zic3 expression ) , we identified genes enriched in GFP-positive cells ( zic3-expressing cells ) . A total of 689 genes ( p-value<0 . 05; fold change ≥1 . 5 ) were enriched in zic3-expressing cells ( zic3-coexpressed genes ) . Among these genes were six members of the Zic family and other genes expressed in the dorsal neural tube . This confirmed the identity of the sorted cells as dorsal neural cells . Among the zic3-coexpressed genes , 167 had at least one peak within 100 kb of their TSS , rendering them putative Zic3 targets ( Table S10 ) . Similar to the 8 hpf stage , members of the Wnt pathway were also among the targets . However , Zic3 seems to regulate a different set of Wnt components , including wnt11r and lef1 ( Fig . 6 , Table S8 ) . qRT-PCR revealed that wnt11r , were down-regulated in Zic3 morphants at 24 hpf ( Fig . 4C; Table S7 ) , confirming their positive regulation by Zic3 . Two other genes encoding Wnt ligands , wnt10a and wnt10b , were co-expressed with zic3 , and regulated upon Zic3 knockdown ( Table S7; Fig . 4C ) although they were not associated with peaks in ChIP-seq , suggesting that they may be indirect targets of Zic3 . A striking difference between 8 hpf and 24 hpf regulatory landscape is apparent from the distinct functions associated with Zic3 target genes at each stage . For example , many genes regulating cell migration and polarity were identified as Zic3 targets at 8 hpf , whereas at 24 hpf neural crest determinants were found . The latter included foxd3 , and pax3a which were further confirmed to be responsive to Zic3 knockdown ( Fig . 4C , Table S7 , S11 ) . On the other hand , in zic3-negative cells , 835 genes were enriched by at least 2-fold ( non zic3-coexpressed genes enriched for endoderm and mesoderm-specific expression terms , Table S9 ) . Among these , 195 had peaks within 100 kb of their TSS , suggesting repression of these genes in cells expressing zic3 ( Table S10 ) . Several proneural genes ( neurod , neurod4 , ascl1a ) were found under this category , which may reflect that the zic3-expressing cells in the dorsal neural tube are not differentiating . Interestingly , the presence of a Zic3 peak in association with oep suggests that a similar inhibition of Nodal by Zic3 occurs at both 8 hpf and 24 hpf ( Fig . 6 ) . Taken together , an entirely different set of candidate Zic3 target genes were found at 24 hpf compared to 8 hpf ( Fig . 6 ) . Although similar signaling pathways , such as the Wnt and Nodal pathways , were regulated by Zic3 at both developmental stages , different members of these pathways were targeted by this regulation at each stage . Furthermore , the global shift in Zic3 binding sites from 8 hpf to 24 hpf suggested the presence of complex regulatory changes accompanying developmental transitions . The large number of Zic3 binding sites in the distant intergenic regions suggested that Zic3 may direct the expression of target genes by binding to the distal regulatory elements . In support of this idea , relevant biological categories could be observed among genes associated with peaks located outside of their basal regions of −5 kb to +1 kb of TSS ( 2716 genes; Table S2; Fig . S7A ) or at a distance more than 50 kb ( 989 genes; Table S2; Fig . S7B ) . In contrast , no particular enrichment of GO categories could be observed for 119 genes associated with peaks in their basal region ( Table S2 ) . Of these , 77 had expression data in ZFIN ( University of Oregon , Eugene , OR 97403-5274; URL: http://zfin . org/; 21 June 2013 ) , but none of these were co-expressed with zic3 at 8 hpf , while only 6 ( lppr3a , p2rx3b , lingo1b , myo15aa , robo4 , gng3 ) had expression overlapping with zic3 at 24 hpf ( not shown ) . To test whether peaks associated with distal genes function as regulatory elements , we used the enhancer activity reporter assay [61] . We chose five distal peaks associated with genes from Nodal and Wnt signaling pathways , including oep ( fragment 10-02 , 94 . 7 kb downstream from TSS ) , axin1 ( fragment 3-43 , 71 . 53 kb downstream ) , lft1 ( fragment 20-35 , 29 . 77 kb downstream ) , dvl2 ( fragment 7-214 , 55 . 92 kb downstream ) , and invs ( fragment 16-297 , 78 . 08 kb downstream ) . A canonical Zic3 motif was present within 100 bp of each peak summit except for fragment 10-02 . Only fragment 16-297 , associated with invs , showed enhancer activity ( Fig . 7B , C , G; Table S12 ) . When the association region was extended to 500 kb , we found more peaks associated with dvl2 ( fragment 7-211 , 236 . 6 kb upstream ) , axin2 ( fragment 3-56 , 147 . 9 kb upstream ) , and pitx2 ( fragment 14-37 , 180 . 32 kb upstream ) . These peaks had at least one canonical Zic3 motif and exhibited positive enhancer activity ( Fig . 7 , Table S5 , S12 ) . Intriguingly , some of the expression patterns driven by the tested enhancers only partially matched that of the associated genes ( fragments 14-37 and 3-56; Fig . 7D , E ) , which could be due to functional dependence on interaction of multiple regulatory elements [62] , [63] . Nevertheless , the presence of Zic3-binding sites with an enhancer activity near genes responding to Zic3 LOF suggested that these genes were direct targets of Zic3 . To validate the activation of the enhancer fragments by Zic3 , we co-injected fragment 7-211 , which drove the strongest reporter gene expression at 8 hpf and 24 hpf ( Fig . 7C ) , and Zic3 MO into the zebrafish embryo . When assayed by qRT-PCR at 8 hpf , a significant decrease in reporter expression in a MO dose-dependent matter was observed ( Fig . S8 ) . No reduction in reporter expression was observed when control MO was used . A similar result was obtained when two other fragments , 4-16 and 17-24 which coincided with CNEs ( Tables 1 , S12 ) , were tested ( Fig . S8 ) , demonstrating Zic3-dependent induction of reporter expression through these fragments . To study whether Zic3 binding sites were evolutionarily conserved , we overlapped the 8 hpf dataset with a list of known conserved non-coding elements ( CNEs; ANCORA database ) [64] . We identified 228 peaks as CNEs conserved between zebrafish and Tetraodon , and 56 as CNEs conserved between zebrafish and humans ( Fig . 8A ) , with 31 in common between the two groups . Similar to the distribution profile of the full set of peaks , these CNE peaks were mostly located outside of the basal promoter region ( Fig . 8B ) . Genes associated with these CNEs were enriched for developmental functions and neural tissue-specific expression ( Fig . 8C , D; Table S2 ) . Of 15 CNE peaks tested for enhancer activity , 11 ( 73% ) drove gfp expression at either 8 hpf or 24 hpf , or both ( Table 1 ) . Of these eleven , eight drove higher gfp expression compared to the reporter vector alone at 8 hpf ( fold change at least 1 . 5 compared to enhancer-less vector ) . Of these eight , four continuously drove reproducible tissue-specific gfp expression in various regions of the CNS up to 24 hpf ( Fig . 8E–H ) , which overlapped with known expression domains of zic3 ( Fig . 1F ) . Another three CNE peaks drove reporter expression only at 24 hpf . The CNE peaks with enhancer activity included the fragments 4-16 and 20-4 , which drove expression in the brain , eye and trunk . In the hindbrain , both drove similar expression in neuroepithelial cells with radial morphology . In the trunk , activity of 4-16 was detected in muscle cells , whereas that of 20-4 was largely confined to the neural tube ( Fig . 8E , F ) . The gfp expression pattern driven by 4-16 partially recapitulated that of a nearby gene , sox5 . On the other hand , 20-4 was located in a gene desert region , suggesting long distance regulation . Fragment 15-26 drove gfp expression largely in cells along the neural tube ( Fig . 8G ) , which partially recapitulated the expression of tbx2b nearby . Fragment 1-22 drove gfp expression mainly in the hindbrain region ( Fig . 8H ) , which partially recapitulated that of the nearby mab21l2 . On the other hand , out of 12 non-CNE peaks tested only two ( 17% ) drove higher gfp expression than the reporter vector alone at 8 hpf ( Table 1 ) . Together with the fragments corresponding to peaks associated with microarray-identified genes , out of 35 fragments tested for activity as enhancers , 17 ( 49% ) were positive . Two thirds of the active peaks were previously identified as CNEs . Whereas this indicated somewhat better chance of finding enhancers amongst CNEs , it also suggested that a significant number of enhancers are not conserved in evolution .
The majority of Zic3 binding sites were found outside promoter regions . While this could be partially attributed to the incomplete annotation of promoter regions in the zebrafish genome , the predominantly distal distribution of Zic3-binding sites revealed that Zic3 regulates transcription largely via distal regulatory elements . Such distribution of binding sites was previously observed in other genome-wide analyses of several TFs in cell culture or mammalian tissues [21] , [22] , [25] , [65] . Our findings therefore establish that a similar distal regulatory mechanism is in effect within the context of Zic3 function during development in vivo . Some of the Zic3 binding sites overlapped with CNEs , most of which drove expression in neural tissues . CNEs are known to regulate developmental genes [66]–[69] . However , in our dataset CNEs represented only 5% of the total Zic3 binding sites identified , while the majority was under weak evolutionary constraint . Tissue-specific enhancers have been shown to differ in the extent of evolutionary conservation of their sequence [70] , [71] . Having only 5% overlap with CNEs , the set of Zic3-binding sites showed a similar trend . The lack of sequence conservation could be explained by the relaxation of selection pressure towards regulatory elements [72] owing to the genome duplication event in teleosts [73]–[75] . Given that at least for now the data available in zebrafish and mammals suggest that only a minority of sites are conserved in both classes of animals , other explanations should be considered . Detailed characterizations of other TFs in the zebrafish would provide a better understanding of the extent of conservation in regulatory regions in teleosts . Cell culture studies have demonstrated interactions between multiple enhancer elements in regulating the transcription of a target gene [24] , [62] , [63] , [76] , [77] , as well as interactions between a TF and different binding partners which can result in alternative transcriptional outputs [26] , [78] , [79] . Our results provide an insight of such complexity of transcriptional regulation by Zic3 in developmental context in vivo . For instance , the concurrent upregulation and downregulation of different subsets of direct target genes by Zic3 suggest that Zic3 binding can result in either activation or repression of target genes , and implies that additional mechanisms determine these two outcomes . Another facet of the data revealed distinct Zic3 binding profiles at 8 hpf and 24 hpf . The genes associated with binding events at these two stages showed relevant functional enrichments . This shift in binding was not dictated by a change in DNA recognition motif as almost identical dominant motifs were identified in both stages . The combinatorial analysis of ChIP-seq and microarray datasets revealed an entirely distinct set of candidate Zic3 target genes at 8 hpf and 24 hpf . Whereas not totally unexpected , this analysis revealed some surprises . First , a developmental switch towards regulation of different members within the same signaling pathway was detected . In the context of Wnt signaling this shifted Zic3 impact from the intracellular part of Wnt signaling towards extracellular ligands in this pathway . Second , that cells expressing Zic3 show a reduced level of transcription of proneural genes placed an impact of Zic3 on cells that are in a state either before or after neural differentiation . Zic3 has been linked with pluripotency of stem cells in mammals [80] . Whereas it is less likely that Zic3 positively regulates the proneural genes at 24 hpf , at the same time this does not exclude a possibility that it could be involved in this process ( as suggested [2] ) during earlier stages . Taken together , these observations suggest that functional relationship between Zic3 and its target gene could not be deduced from a simple one-to-one interaction model . Factors , such as the presence of different subsets of interacting partners or accessibility of certain binding sites as dictated by chromatin states , in different spatiotemporal contexts may affect transcriptional output . One implication of an interactive regulatory landscape is that genes targeted by a particular TF may not be determined by simply observing binding of the TF near its genomic locus . Additional proof , such as responsiveness of the particular target gene to LOF of the TF , would be necessary . In our data , there is a surplus of Zic3 binding events compared to those associated with responsive target genes . Widespread binding of TFs exceeding their known target genes have been reported in cell culture and in Drosophila [81]–[87] and is suggestive of non-functional binding . This may happen due to interaction of TFs with randomly occurring target sequences in the genome [78] , [88] . The availability of expression data helps to identify candidate target genes within the vicinity of a TF binding event by providing additional functional cues . Nevertheless , given that TF-target genes interactions could occur over long distances [22] , [89] , [90] , it is still possible that seemingly isolated Zic3 binding events with no responsive genes within a set distance criteria might actually be regulating a target located further away . Until a more detailed understanding of the architecture of genome-wide interactions have been achieved , this possibility could not be ruled out . The highly interconnected TF regulatory network also necessitates a careful interpretation of enhancer function by reporter assays: while such assays can be useful to identify independently acting regulatory elements , evidence exists for regulatory elements acting in tandem , resulting in higher transcriptional output [24] , [62] , [63] , [76] , [77] . While other possibilities such as non-functional occupancy and repressive interactions could not be ruled out , the TF interaction model could account for the inactivity of several of the tested enhancers inferred from the reporter assay . The occurrence of Zic3 consensus motifs in close proximity to 50% of peaks containing Gli consensus motif supports this idea . Interestingly , the presence of Gli motifs does not seem to be specific to a particular developmental stage , as both 8 hpf and 24 hpf data show similar proportions of Zic3 peaks containing Gli motifs nearby . As in vitro data have demonstrated physical and functional interactions between Zic and Gli proteins [40] , [41] , such interaction , as well as interactions with other binding partners , may also occur in vivo in regulating transcription of target genes . Our identification of novel target genes of Zic3 has improved an understanding of the mechanism by which Zic3 regulates development . These results demonstrated that Zic3 inhibits Nodal signaling ( either directly or indirectly ) which is implicated in mesendodermal specification [15] , [44]–[46] . Similarly , Lim and colleagues [38] observed that murine ES cells acquired endodermal fate upon Zic3 knockdown , which supported an idea that Zic3 acts as an inhibitor of endodermal fate . Coincidentally , Nodal and Wnt signaling is known to regulate gastrulation [91]–[94] . Their regulation by Zic3 therefore may account for the gastrulation defect observed in Zic3 morphants . On the other hand , proper midline development during gastrulation is essential for proper L-R patterning [15] , [95] , [96] . Therefore , an involvement of Zic3 in regulating gastrulation through Nodal and canonical Wnt per se could have been sufficient to ensure a proper L-R asymmetry . However , our results suggested that Zic3 may also regulate the non-canonical Wnt ( PCP ) signaling pathway which is implicated in ciliogenesis . Interaction of these signaling pathways culminates in the establishment of a proper embryonic L-R axis [97]–[102] . Therefore , we could not rule out the possibility of direct involvement of Zic3 in later events specific to L-R patterning . In this context , it is noteworthy that mkks was also found as one of the Zic3 targets ( Table S5 ) which is implicated in both L-R patterning and C-E movements during gastrulation through interaction with vangl2 [103]–[106] . Therefore , the regulation of non-canonical Wnt signaling by Zic3 could be at a core of developmental events linking C-E movement and L-R patterning [10] . Our finding that Zic3 regulates genes implicated in proliferation of neural progenitors agrees with the idea that Zic3 has properties of a stem cell factor [38] , [80] . A mode of Zic3 regulation of genes responsible for the proliferation of neural progenitors reconciles the role of Zic3 in both early neuroectodermal specification and later events of neurogenesis . In essence , it establishes a particular role of Zic3 ( and possibly other Zic family members ) as an important regulator of proliferation of neural progenitors [7] . This model challenges previous assumptions that Zic3 induces the expression of proneural genes shown in overexpression studies [18] , and suggests that an activation of proneural genes could be a downstream consequence of Zic3 regulation of proliferation of neural progenitor at an earlier stage of neurodevelopment . Given that neurog1 expression was upregulated upon Zic3 knockdown , and Zic3 binding sites were found near neurog1 , as well as other proneural genes such as neurod4 and ncam1a , Zic3 may have an additional direct role in neural differentiation as its inhibitor . This possibility is also supported by the downregulation of her9 . This places Zic3 within a regulatory landscape of Notch signaling in support of an early hypothesis based on functional analysis of Zic1 [107] .
Zebrafish of wild type ( AB strain ) and transgenic line SqET33 [28] , [34] were maintained according to established protocols [108] following all the ethical practice recommended for fish maintenance . Embryos were staged according to standard morphological criteria [109] . Dechorionated 24 hpf transgenic embryos were deyolked in PBS by pipetting through the 1 ml pipette tip . Cells were dissociated with trypsin solution ( 0 . 05% trypsin and 0 . 2 mM EDTA ) in PBS for 15 min at room temperature . To facilitate dissociation of cells , embryos were pipetted through the 200 µl pipette tip . Trypsin was inhibited with complete protease inhibitor cocktail ( Roche ) and cell suspension was filtered through a nylon mesh ( 40 µm Cell Strainer , BD Falcon ) . Immediately , an equal volume of 4% paraformaldehyde ( PFA ) in PBS was added to cell suspension and cells were fixed for 10 min at room temperature . Reaction was stopped by an equal volume of ice-cold 0 . 25 M glycine in PBS , cells were washed three times with 0 . 125 M glycine-PBS and resuspended in the same buffer . Cell sorting was carried out with FACSAriaII Cell Sorter ( BD Bioscience ) . To set autofluorescence level , cell sorter was calibrated with PFA-fixed GFP-negative cells before cell separation . GFP-positive and GFP-negative cells were collected in 0 . 125 M glycine-PBS , frozen in liquid nitrogen and kept at −80°C until use . For microarray analysis , PFA fixation step was omitted and cells were sorted into complete L-15 Leibovitz medium ( Gibco ) containing 20% fetal bovine serum . Chromatin Immunoprecipitation ( ChIP ) was performed according to an established protocol ( Wardle et al . , 2006 ) with an addition of a deyolking step according to Link and colleagues ( 2006 ) , with modifications ( see Text S1 ) . ChIP DNA was sequenced on the Illumina Genome Analyzer II ( Illumina , USA ) . Detailed ChIP-seq methods are described in Supplementary information . Sequencing reads were mapped to the zebrafish Refseq genome assembly ( Zv9 ) , following which peak finding was performed using the QuEST algorithm [110] using the following parameters: bandwidth = 30 bp , region size = 600 bp , and FDR q-value<0 . 01 . Peaks mapped to unassembled chromosomal contigs , centromeric regions , telomeric regions , segmental duplications and peaks consisting of >70% repeat sequence were removed . The ChIP-seq data have been deposited in the Gene Expression Omnibus database under the accession number GSE41458 . To validate the ChIP-seq performance , we carried out quantitative PCR ( qPCR ) on randomly selected peaks , 5 within promoter region and 16 at regions outside of gene promoters . Taking a fold-change of 2 as a cutoff for positive enrichment , the qPCR analysis validated all but one peak tested ( Table S1 ) . The Database for Annotation , Visualization , and Integrated Discovery ( DAVID ) [111] , [112] and Genomic Regions Enrichment of Annotations Tool ( GREAT ) [35] was used to find gene ontology-enriched terms . Overlapping of 8 hpf and 24 hpf ChIP-seq signals around peaks was performed within a region of +/−2 kb from each peak summit . Notice that some peak regions in 8 hpf dataset were not detected as peaks in 24 hpf dataset but they could be having sufficient amount of ChIP-seq tags at 24 hpf because of true binding by Zic3 . Similarly there were regions detected as peaks in 24hpf samples and not detected in 8hpf but they may be bound by Zic3 in both samples and be having enriched ChIP-seq tag count in both . Hence ChIP-seq signal based clustering further clarified the status of detected peaks . Motif search was performed with MEME de novo motif finder [113] . From the top 1000 peaks by statistical significance , we extracted sequences comprising +/−50 bp from the summit of each peak . After finding the similarity of de novo motif from MEME with other published Zic3 motifs [39] , [80] , the quantification of occurrence of these motifs was done on all ChIP-seq peaks . For this the sequences within 400 bp from the peak summit were matched with PWM of motifs and the best matching score were calculated . After having the best matching score a threshold was used to determine the presence of motif . The PWM-matching threshold value for each motif was calculated using simulation such that when 10000 sequences were randomly designed to have probability similar to corresponding nucleotides in its PWM then 85% of those sequences could be detected . CNE peaks were identified by comparing the 8 hpf ChIP-seq dataset against a list of known CNEs in ANCORA database [64] . We performed the comparison to both human and Tetraodon CNE database to take into consideration the genome duplication event during teleosts evolution , which relaxed selection pressure on the conservation of important developmental enhancers [68] , [72] . The genomic coordinates of each peak summit were extended by 500 bp on each side and compared against the genomic coordinates of CNEs identified through comparison with either human hg19 or Tetraodon tetNig2 assemblies . A threshold of at least 70% sequence conservation within every 50 bp was used to define CNEs in each species . Two recombinant constructs of the zebrafish Zic3 protein were produced , the full-length protein ( Zic3_ORF ) and the DNA-binding domain encompassing Zn-fingers 2 to 5 ( Zic3_ZF2-5 , amino acid residues 273–391 ) . DNA sequences corresponding to each domains were PCR-amplified using the following primers: Zic3_ORF: 5′-GGG GAC AAG TTT GTA CAA AAA AGC AGG CTT CGA AAA CCT GTA TTT TCA GGG CAG CTT ACG TGA AAT TGC G CTC-3′ and 5′-GGG GAC CAC TTT GTA CAA GAA AGC TGG GTT TAC TCC ACC TGA AAA CGG ACT TG-3′; Zic3_ZF2-5: 5′-GGG GAC AAG TTT GTA CAA AAA AGC AGG CTT CGA AAA CCT GTA TTT TCA GGG CGC CTT CTT CAG ATA CAT GCG-3′ and 5′-GGG GAC CAC TTT GTA CAA GAA AGC TGG GTT TAT GAT TCG TGT ACC TTC ATA TG-3′ . Each forward and reverse primer contained an attB recombination site overhang , with an additional Tobacco Etch Virus ( TEV ) protease cleavage site in the forward primer preceding the N-terminal Zic3 coding sequence . Protein expression and purification was performed as previously described ( Lim et al . , 2010 ) . Electrophoretic mobility shift assay ( EMSA ) was performed as previously described [38] . Briefly , Cy5-labeled oligonucleotide pairs ( 1st BASE , Singapore ) were annealed by heating to 95°C for 5 minutes in annealing buffer ( 500 mM MgCl2; 500 mM KCl; 200 mM Tris-HCl , pH 8 . 0 ) and left in room temperature to cool down overnight . These were subsequently incubated with the recombinant Zic3 in EMSA buffer ( 10 mM Tris , pH 8 . 0; 0 . 1 mg/ml BSA; 50 µM ZnCl2; 100 mM KCL; 0 . 5 mM MgCl2; 10% glycerol , 0 . 1% SDS; 2 mM β-mercaptoethanol ) for 1 hour at 4°C . The reaction was subsequently run on 5% native Tris-Glycine polyacrylamide gel electrophoresis . Gel was scanned in Typhoon Scanner ( GE Healthcare , USA ) . The affinity of protein to DNA was determined by titrating 0–250 nM of protein against 1 nM of annealed probes . Zic3 knockdown was performed using a translation-blocking antisense morpholino oligonucleotide ( MO ) purchased from Gene Tools , LLC ( USA ) . The MO sequence was 5′-AGG TTA GTG GAG TGA ACG GGT ACC G-3′ . A standard control antisense MO was also obtained from Gene Tools , LLC with the following sequence 5′-CCT CTT ACC TCA GTT ACA ATT TAT A-3′ . For microarray , 1 . 7 ng Zic3 MO was injected into 1-cell stage embryos . Rescue was performed using 20 pg of zic3 mRNA without morpholino-binding site . Capped zic3 mRNA was synthesized using mMessage mMachine Kit ( Ambion , USA ) . Results were obtained from at least three different experiments on embryos from random pairs . For gene expression profiling , custom made zebrafish oligonucleotide microarray ( Agilent Technologies; GIS V2 with some modifications ) containing 44 , 000 oligonucleotide probes ( 60 mer long; including positive and negative controls designed by Agilent and beta-actin controls ) was used . The microarray was performed according to Agilent's One-Color Microarray Based Gene Expression Analysis ( Quick Amp Labeling ) protocol ( Version 5 . 7 , March 2008 ) and RNA Spike-In-One Color . Arrays were probed using cDNAs reverse transcribed in the presence of Cy3-dUTP using 600 ng of total RNA from either wild-type control or Zic3 knockdown embryos ( 8 hpf ) , or from either non zic3-expressing cells or zic3-expressing cells ( 24 hpf ) . Labeled cDNA was denatured and hybridized at 42°C for 16 h in a hybridization oven ( Agilent Technologies , USA ) . After hybridization , the slides were washed and scanned for fluorescence detection on Agilent DNA Microarray Scanner . Scanned images were analyzed using Agilent Feature Extraction Software ( v10 . 5 . 1 . 1 ) . Feature extracted data were analyzed in Genespring software ( Agilent Technologies , USA ) . Statistically significant gene expression was identified using Significance Analysis of Microarrays ( SAM 3 . 05 ) for each successive time point [114] . Threshold values were set as follows: q-value<0 . 8 , predicted false discovery rate ( FDR ) <0 . 05% . Genes were annotated using the “Unigene & Gene Ontology Annotation Tool” available at GIS site ( http://123 . 136 . 65 . 67/ ) . Genes were subjected to pathway assembly using Ingenuity Pathway Analysis ( IPA; http://www . ingenuity . com ) . Selected genes ( Fig . 4C; Table S7 ) were validated using real time RT-PCR ( qRT-PCR ) by assessing their expression level changes in embryos injected with higher dose of morpholino ( 3 . 4 ng ) to show similar trend with microarray regulation . Tested genomic regions encompassing the peaks with ∼200 bp flanking sequence at each side were amplified using PCR ( primer list in Additional file 5 ) and cloned into SalI and BamHI sites of the pTol2-GFP reporter vector containing a minimal promoter from the mouse cFos gene [115] . Transposase mRNA was synthesized using mMESSAGE mMACHINE T3 Kit ( Ambion , USA ) and purified using RNeasy Mini Kit ( QIAGEN , Germany ) . A total of 20 pg of the circular reporter plasmid and 50 pg of transposase mRNA were co-injected into 1–2-cell stage embryos . For each construct , two batches of at least 100 embryos were injected and assayed for egfp expression at 24 hpf . A consistent egfp expression pattern observed in at least 20% of injected embryos was considered as positive . The reporter vector alone showed expression in muscles and blood cells in G0 embryos ( data not shown ) . Embryos positive for egfp expression were subsequently processed for whole mount immunohistochemistry ( IHC ) with anti-GFP antibody . qPCR was used to determine egfp expression level at 8 hpf since morphological identification of tissue specificity at this stage was difficult . | The Zic3 transcription factor regulates early embryonic patterning , and the loss of its function leads to defects in left-right body asymmetry . Previous studies have only identified a small number of Zic3 targets , which renders the molecular mechanism underlying its activity insufficiently understood . Utilizing two genomics technologies , next generation sequencing and microarray , we profile the genome-wide binding sites of Zic3 and identified its target genes in the developing zebrafish embryo . Our results show that Zic3 regulates its target genes predominantly through regulatory elements located far from promoters . Among the targets of Zic3 are the Nodal and Wnt pathways known to regulate gastrulation and left-right body asymmetry , as well as neural pre-pattern genes regulating proliferation of neural progenitors . Using enhancer activity assay , we further show that genomic regions bound by Zic3 function as enhancers . Our study provides a genome-wide view of the regulatory landscape of Zic3 and its changes during vertebrate development . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Genome Wide Analysis Reveals Zic3 Interaction with Distal Regulatory Elements of Stage Specific Developmental Genes in Zebrafish |
Horizontal gene transfer ( HGT ) in bacteria generates variation and drives evolution , and conjugation is considered a major contributor as it can mediate transfer of large segments of DNA between strains and species . We previously described a novel form of chromosomal conjugation in mycobacteria that does not conform to classic oriT-based conjugation models , and whose potential evolutionary significance has not been evaluated . Here , we determined the genome sequences of 22 F1-generation transconjugants , providing the first genome-wide view of conjugal HGT in bacteria at the nucleotide level . Remarkably , mycobacterial recipients acquired multiple , large , unlinked segments of donor DNA , far exceeding expectations for any bacterial HGT event . Consequently , conjugal DNA transfer created extensive genome-wide mosaicism within individual transconjugants , which generated large-scale sibling diversity approaching that seen in meiotic recombination . We exploited these attributes to perform genome-wide mapping and introgression analyses to map a locus that determines conjugal mating identity in M . smegmatis . Distributive conjugal transfer offers a plausible mechanism for the predicted HGT events that created the genome mosaicism observed among extant Mycobacterium tuberculosis and Mycobacterium canettii species . Mycobacterial distributive conjugal transfer permits innovative genetic approaches to map phenotypic traits and confers the evolutionary benefits of sexual reproduction in an asexual organism .
Sexual reproduction in eukaryotes promotes genetic diversity by increasing gene flow through a population , permitting both the loss of mutant genes and the acquisition of functionally distinct gene alleles . The diversifying potential is further enhanced by crossover events that create new mosaic recombinant meiotic products , which in turn may impart new functionalities not present in either parent . In contrast , bacterial fission provides rapid clonal expansion to fill an environmental niche , but lacks the evolutionary advantages of sexual reproduction . Horizontal gene transfer ( HGT ) mitigates the diversification constraints of asexual reproduction by mediating limited gene flow through the population . The fundamental forms of HGT include transformation , transduction , and conjugation . Conjugation is considered a major contributor to HGT , as it can transfer more extensive segments of DNA between different species and even kingdoms [1]–[4] . Conjugation describes the unidirectional transfer of DNA from a donor to a recipient , and requires cell–cell contact . Conjugal processes are traditionally plasmid encoded , or encoded by a discrete genetic element integrated into the chromosome . Transfer proteins are generally classified into those that establish and maintain mating-pair formation or those responsible for DNA transfer [5] , [6] . These latter proteins recognize and nick the unique origin of transfer ( oriT ) on the plasmid and guide the DNA into the recipient cell . oriT is cis-acting , and thus , when recombined into the chromosome , it can mediate transfer of chromosomal DNA , as first described for E . coli Hfr strains [7] . DNA transfer in M . smegmatis displays all of the hallmarks of conjugation: it requires stable and extended contact between a donor and a recipient strain , it is DNase resistant , and the transferred DNA segments are incorporated into the recipient chromosome by homologous recombination [8] . While the process clearly meets the traditional definition of conjugation , the similarities with the classical E . coli Hfr system end there [9]–[13] . Mycobacterial conjugation is chromosome—not plasmid—based , and bioinformatic and genetic studies have yet to identify a genetic element that might mediate transfer [14] , [15] . In E . coli , Hfr transfer always initiates at the sole plasmid-encoded oriT site , and the DNA is transferred in a 5′ to 3′ direction , such that only genes proximal and 3′ to oriT are inherited at high frequencies [10] , [16] . By contrast , in M . smegmatis , all regions of the chromosome are transferred with comparable efficiencies as demonstrated by equivalent transfer of a kanamycin-resistance marker regardless of its chromosomal location [11] . This position independence is consistent with the presence of multiple , but ill-defined , initiation sites [17] . Transposon mutagenesis screens provided initial insights into the genetic requirements of transfer [14] , [15] . These studies established a prominent role for the Type VII secretion apparatus , ESX-1 , in both donor and recipient activity . ESX-1 clearly plays different roles in each cell type . ESX-1 donor mutants are hyperconjugative , suggesting secretion plays a role in negatively regulating transfer activity [15] . By contrast , recipient strain ESX-1 mutants do not receive donor DNA [14] . Although these studies provided novel insights into the functional roles of ESX-1 , they did not provide insights on the transfer mechanism , or define what determines the mating type of a cell ( either donor or recipient ) . Here , as an alternative approach , we examined the products of DNA transfer to better understand this process and its contributions to mycobacterial evolution . We used next-generation sequencing to determine the parental inheritance profiles in transconjugant M . smegmatis progeny . The genomic sequence of each of the M . smegmatis parental strains has been determined , and the abundant single nucleotide polymorphisms between the two strains indicated that the transferred segments comprising the transconjugant genomes could be mapped with precision . We found that the parental contributions to the transconjugants were much more complex than expected , indicating a surprisingly major role for conjugal DNA transfer in generating genomic diversity . The blending of the parental genomes is reminiscent of that seen in the meiotic products of sexual reproduction . This comparison is validated by our use here of genomic approaches previously developed and applied in sexual reproduction systems to define candidate genes for conjugal mating identity .
To provide a selectable marker for chromosomal DNA transfer , a kanamycin resistance gene ( Kmr ) was integrated in the chromosome of mc2155 , the standard laboratory and conjugal donor strain of M . smegmatis . Donor mc2155 derivatives that differed in their Kmr insertion site were mated to an apramycin-resistant ( Apr ) recipient strain , mc2874 ( Figure 1A ) . mc2874 is an independent isolate of M . smegmatis that we have used as a standard recipient strain [8] , [18] . Apramycin resistance was episomally encoded to avoid inheritance biases caused by selecting for this gene on the recipient chromosome . From matings between these strains , 12 independent KmrApr F1 progeny were isolated , and the DNA sequences of their genomes were determined ( sequence data deposited in the EBI/ENA database at http://www . ebi . ac . uk/ena/data/view/ERP002619 ) . Our comparative sequence analyses of the parental strains had shown that the circular mc2155 and mc2874 genomes are collinear , and that they contained abundant single nucleotide polymorphisms ( SNPs; averaging one per 56 bp ) providing a clear distinction between parental DNA origins ( Figures 1A and S1 ) . Individual sequence reads from each transconjugant were aligned with the donor strain genome to identify all transferred donor segments . When evaluating transconjugant sequences , we conservatively required the presence or absence of two consecutive recipient SNPs to define a boundary between recipient and donor sequence tracts , respectively ( Figure S2 ) . Donor segments replaced the corresponding recipient sequences , as evidenced by a concomitant localized loss of recipient-specific SNPs in transconjugants . Unique segments of transferred donor DNA , predicted by alignment analyses in transconjugants , were confirmed by conventional PCR and Sanger sequencing ( Table S1 ) . Two transconjugants had 11 regions that were merodiploid ( approximately equal contributions of donor and recipient SNPs ) . As this was a resequencing and not a de novo sequencing strategy , we cannot determine the precise architecture and location of these regions . These regions did not contain repetitive elements , though it is possible that integration occurred at nonsynonymous sites via microhomology or through mechanisms not requiring homology . The most striking observation from an alignment of our initial set of 12 transconjugant genomes with the parental genomes was that the transconjugant genomes were broadly mosaic , containing at least two , and as many as 21 , separate tracts of cotransferred mc2155 DNA embedded in an mc2874 background ( Figure 1B and Table S2 ) . These separate segments of DNA were acquired in a single cell–cell transfer event , as determined in earlier studies [11] . To our knowledge , this degree of genome-wide diversity is unprecedented in genetic transfer events between bacteria . This contrasts directly with the iconic plasmid-transfer systems in which a single segment of donor DNA linked to oriT is inherited [10] , [19] . Therefore , we refer to mycobacterial conjugation as distributive conjugal transfer to distinguish it from oriT-mediated transfer . As expected , all transconjugant progeny acquired the selected Kmr gene , along with variable amounts of flanking mc2155 DNA ( Figure 1B , Kmr , green segments embedded in yellow recipient DNA ) . Surprisingly , 5-fold more mc2155 DNA was co-inherited in segments that were not selected , and these segments were distributed around the genome with no obvious regional biases ( Figure 1B , alternating blue and magenta improve visual discrimination between adjacent tracts; Table S2 ) . The 12 transconjugant genomes analyzed contained from 57 kb to 679 kb ( of 6 . 9 Mb ) of mc2155-derived sequence . The sizes of the donor segments varied >1 , 000-fold , ranging from 59 bp to 226 kb ( Figure S3 and Table S2 ) , with an average size of 33 . 8 kb , and a mean of 10 tracts per genome ( Table 1 ) . Some regions showed intricate microcomplexity of multiple inherited segments separated by short intervals of recipient DNA ( Figure 1C and highlighted in Table S2 ) . Note that the single-nucleotide discrepancies ( colored SNPs ) derive from parental inheritance , not de novo mutation ( see reciprocal parental reference sequence alignments in Figure 1C ) . These likely resulted from a combination of repair and recombination events occurring between the recipient chromosome and a single molecule of introduced donor DNA , as some segments are separated by only a few base pairs . Regardless of the mechanism , the net effect was to create a localized composite blend of parental contributions at the nucleotide level . The image in Figure 1B shows the extent of mc2155 DNA transferred to recipients when selecting for a single event: acquisition of the gene encoding Kmr . Based on the distributive nature of transfer , we reasoned that we could employ secondary screens of the transconjugants to map any additional genetic trait regardless of its linkage to the Kmr gene . Tracking parental SNPs within a group of individual transconjugants exhibiting a given phenotype should identify those shared SNPs ( and parental genes ) associated with that phenotype . We have previously observed that a subset of transconjugants become donors , suggesting that these progeny acquired a donor-conferring locus [11] . We hypothesized that an unbiased genome-wide mapping approach would identify a shared segment of mc2155 DNA among those progeny encoding this trait . Transconjugants derived from crosses of the differentially marked donor strains were screened for donor ability , and 10 independent donor-proficient transconjugants were identified . We note that mating identity is a mutually exclusive phenotype , and transconjugants exhibit transfer efficiencies comparable to parental strains ( [11] and Table S3 ) . Genomic DNA from each donor-proficient transconjugant was prepared and its sequence determined . Comparative sequence analysis showed that all donor-proficient transconjugants , regardless of the location of the Kmr gene in the parent , shared only one segment of mc2155 DNA ( Figure 2A and Table S4 ) , with the smallest region of overlap encompassing coordinates 74 , 522 to 119 , 788 bp ( Figure 2B ) . This result is consistent with transfer of a single 45 kb locus ( mid ) that is sufficient to switch mating identity from recipient to donor in these transconjugants . This region is not simply a hot spot for integration of acquired DNA , since the 12 recipient-proficient ( i . e . , did not become donors ) transconjugants in Figure 1B were not similarly enriched for this segment of mc2155 DNA ( compare Figures 1B and 2A , and see below ) . Closer examination of the region acquired by donor-proficient transconjugants established that they all had inherited a minimal segment of DNA encompassing the mc2155 esx1 locus ( Figure 2B , 74 , 600–107 , 334 bp , esx1D , where the subscript differentiates donor or recipient origin ) . The esx1 locus encodes a Type VII secretion system [20] , [21] . The encoded ESX-1 apparatus assembles in the cell membrane and secretes a specific set of proteins , which , in M . tuberculosis , are essential for pathogenicity [22]–[24] . Proteins secreted by ESX-1 lack a signal peptide that would aid in their identification , and the most notable substrate is a heterodimer of two small proteins , EsxB and EsxA . Other proteins encoded within the esx1 locus and elsewhere in the genome are also secreted through ESX-1 , some of which are co-dependent on EsxBA secretion . The functions of most of the proteins encoded by esx1 genes are unknown , but the overall composition of the esx1 loci between the parental mc2155 and mc2874 strains are similar ( see below ) . Although our previous transposon mutagenesis studies have shown that ESX-1 plays an important role in the process of DNA transfer in both donor and recipient strains , mating-type identity is not reversed in ESX-1 mutants [14] , [15] . Therefore , the role of ESX-1 in determining mating identity was quite unexpected , and underscores the utility of a “change-of-function” mapping approach . While all of the donor-proficient transconjugants inherited an intact esx1D locus , none of the recipient-proficient F1 strains did . Notably , four of the F1 recipient-proficient strains were derived from the Km0 . 1 parent , in which only 15 kb separate esx1D and the selected Kmr gene . Despite this tight linkage , distributive conjugal transfer readily segregated the Kmr gene and intact esx1D locus when appropriately screened , thereby augmenting the mapping resolution ( Figure 1B , Table S2 , and below ) . Helpfully , one of these recipient-proficient transconjugants ( Km0 . 1c ) inherited parts of esx1D , excluding these esx1 genes from mid candidacy ( 0064–0068 and 0077–0083 , Table S2 ) . These negative correlations affirm the functional dependence of the donor trait on the mid genes of esx1D and demonstrate the robust nature of distributive conjugal transfer in generating the level of genetic diversity necessary for our mapping analyses . In classical genetic studies , fine mapping of a genetic determinant can be achieved by performing successive backcross introgression analyses to genetically purify a locus in a recipient background . We reasoned a similar strategy would achieve two goals: ( 1 ) discard mc2155 parental genes not required for the donor transfer trait and ( 2 ) further narrow the key conjugal mid gene region . Six F1 donor recombinants were backcrossed with mc2874 recipient derivatives that were marked with a different episomally encoded antibiotic resistance gene ( Hygr or Apyr ) in successive generations . Introgression entailed co-selection for Kmr transfer and the recipient marker to identify transconjugants at each generation ( Nx ) , and then screening progeny for donor proficiency ( Figure 3 ) . Comparative analyses of genomes of three donor-proficient strains showed a purifying selection of the donor-conferring locus and Kmr genes in an otherwise recipient genome ( Figure 4 , Table S4 ) . In each case , the majority of the F1 mc2155 DNA was lost . For example , the F1 parent of Km0 . 1BCb contained 19 mc2155 segments totaling over 869 kb , yet following six backcross generations this DNA was trimmed to three segments totaling 110 kb , most of which encompassed the selected mid and Kmr genes ( 79 kb , Table S4 ) . As expected , backcross matings also resulted in recipient-proficient progeny , several of which were also sequenced ( Figure 3 ) . Coincident with a reversal of mating identity , the esx1D locus failed to transfer . One recipient strain , Km0 . 8BC , retained only 75 kb of mc2155 DNA of the 920 kb originally present in the F1 parent ( Figure 5 , Table S4 ) . Analyses of two recipient-proficient strains derived from independent F1 Km6 . 9 parents further refined the region of interest . Km6 . 9BCa included donor genes 0055D–0067D and 0079D–0083D and Km6 . 9BCb contained genes 0072–0075D ( Figures 5 and 6 , Table S4 ) . Thus , these esx1D genes are insufficient to confer a donor phenotype . Taken together , the mapping data identify esx1 genes in 0068D–0071D and/or 0076D–0078D as being critical for determining mating identity . Ongoing studies requiring multiple , precise , targeted gene swaps will identify the key gene ( s ) . While most esx1 gene products are highly conserved among mycobacterial species , M . smegmatis proteins 0069 , 0070 , and the N-terminal two-thirds of 0071 have notably low amino acid identity between donor and recipient orthologs ( Figure 6 and Figure S4 ) [14] and are therefore good candidates for switching mating identity . The proteins encoded from this region are not predicted to contain an obvious motif or domain that would provide mechanistic insight into their role in conjugation . However , the location of the mid genes within esx1 suggests that the encoded proteins modify ESX-1 structure or function , to perhaps affect cell–cell communication or physically mediate DNA transfer .
We used next-generation sequencing to examine transconjugant genomes and found that mycobacterial conjugation generates highly mosaic genomes created by a robust distributive conjugal transfer process . Transconjugants acquired large amounts of donor DNA ( some exceeding one-fourth of the transconjugant genome; Table S4 , Km4 . 5a ) , in varied segment sizes ( spanning four orders of magnitude ) that were distributed around the genome . We exploited these characteristics of distributive conjugal transfer ( DCT ) to map mating identity genes of M . smegmatis . Hfr transfer in E . coli is initiated from the unique oriT and results in transfer of a single segment of the donor chromosome [9] , [19] , [25] . Thus , while the recipient acquires new genetic information , that new information is limited to DNA immediately adjacent and 3′ to oriT ( Figure 7 , left ) . Genetic analyses and an understanding of the RecBCD recombination machinery suggest that a single segment is integrated into the recipient chromosome via a recombination event occurring at each end of the transferred DNA molecule [16] . To our knowledge , whole genome sequencing has not been reported for Hfr– transconjugants , preventing a detailed comparison of the two conjugation systems . Thus , our study provides the first genome-wide analysis of bacterial conjugal transfer . In contrast to oriT-mediated transfer , the complex inheritance profiles exhibited by mycobacterial transconjugants suggest stochastic co-transfer from multiple origins , as previously predicted [17] . Based on our genome sequence data , we speculate that random chromosomal DNA fragments are generated in the donor , some of which are co-transferred into the recipient strain where they replace recipient sequences through homologous recombination . An alternative scenario is that a single large DNA molecule is transferred , which is processed into smaller segments before their integration into the recipient chromosome by homologous recombination . This scenario seems less likely as we would have expected to identify some transconjugant progeny containing exceedingly large chunks of donor DNA ( 3–4 Mb ) integrated into the chromosome . These would have resulted from recombination close to the ends of the transferred molecule , before creation of small segments . This latter scenario is also less consistent with our previous observations , which indicated that the donor chromosome contained multiple initiation sites and that the efficiency of gene transfer was location-independent . We have considered examining boundary sequences to determine whether they provide insight on the mechanism of conjugation . However , there are multiple factors influencing boundary regions , which together prevent a unifying mechanistic insight . For example , the actual breakpoints generated by conjugation are almost certainly lost as the boundaries are driven by the requirement for homology and by different recombination mechanisms mediating integration , as evidenced by inheritance of both regions of microheterogeneity and single large integration events . Mycobacteria encode multiple nonredundant recombination pathways ( RecBCD , AdnAB , and nonhomologous end-joining ) , but are not known to encode a mismatch repair system [26] , [27] . We postulate that homologous recombination mediated by AdnAB is likely responsible for the simple crossover events , which is consistent with the absolute requirement for RecA in DCT [17] . However , this form of homologous recombination alone seems insufficient to explain regions of microcomplexity . The clustered proximity of recombinant tracks indicates that an imported donor segment initially encompassed the entire region , but the mechanism underlying the internal mosaicism is unclear . Characterization of the mechanism and the enzymes behind this process will require careful directed approaches using defined recombination mutants . Every facet of the transfer process contributes to the genetic complexity of the transconjugants ( Figure 7 ) . The large number and distributive character of the transferred segments , combined with the microcomplexity in some tracts , makes each transconjugant uniquely different from the others , as well as from the parental strains . The widely varied sizes of the transferred segments allows transconjugants to acquire both major changes , potentially bringing in entire operons encoding biological pathways , and minor nucleotide substitutions that provide subtle diversity , which could , for example , modify the activity or interaction specificity of an enzyme . Multiple pan-genomic changes that typically accompany evolution of bacteria are assumed to be a serial accrual of HGT and spontaneous mutation events ( Figure 7 ) . By contrast , a single step DCT event between two single cells generates a transconjugant strain that is a mosaic blend of the parental genomes , and not merely an incrementally altered derivative . Thus , distributive conjugal transfer provides an unparalleled mechanism for quickly generating tremendous genetic diversity , which rivals that seen in sexual reproduction [28] . Recent genome-wide studies of naturally competent strains provide an interesting contrast between the progeny of transformation and conjugation [29]–[32] . In these studies , nonselected segments of DNA were also observed around the recipient chromosome and thus contribute to variation . Microcomplexity in these segments suggested that , as for DCT , integration of transformed DNA was mediated by both recombination and/or repair machinery . However , the nonselected segments were significantly smaller ( 1–4 kb , depending on the species ) than those described here , which average 49 kb and can be as large as 249 kb ( Table S4 , Km4 . 5b: 6 , 942 , 375–202 , 798 ) . The limitation on recombination sizes in pneumococci correlated with an underrepresentation of large insertions , which together argued that transformation led to genome reduction and was unlikely to act as a mechanism for uptake of accessory loci [29] . The large DNA segments acquired via DCT , in contrast , facilitates inheritance of novel operons and genes . For example , one large recombination tract introduced a contiguous stretch of ∼55 kb of nonhomologous donor-derived DNA into the transconjugant chromosome ( Km6 . 9b ) . Perhaps an example more functionally pertinent to our work was an insertion–deletion exchange observed in the divergent mid candidate region of esx1 in transconjugants switched to donors ( Figure S5 ) . We have demonstrated conjugal DNA transfer in additional naturally derived M . smegmatis strains [8] , indicating a broader presence for mycobacterial distributive conjugal transfer . The rough-colony morphology members of the Mycobacterium tuberculosis complex ( MTBC ) exhibit extremely low genetic variation , suggesting that they do not undergo HGT , are evolutionary young , and resulted from a recent clonal expansion [33] . However , there is now convincing evidence for HGT among M . canettii , and other smooth-colony MTBC strains , which display genome-wide mosaicism , although the precise mechanism ( s ) of HGT are unknown [34] , [35] . Based on sequence comparisons , it was proposed that M . canettii strains are extant members of a genetically diverse MTBC progenitor species , M . prototuberculosis , whose members underwent frequent HGT [34] , [36] , [37] . The unspecified HGT process underlying that mosaicism is presumed to result from a series of sequential transfer events . However , based on our studies , distributive conjugal transfer involving the ancestral M . prototuberculosis offers a plausible and parsimonious explanation for the remarkably similar mosaicism observed among the extant M . canettii . We could envision that distributive conjugal transfer in M . prototuberculosis rapidly incorporated the necessary blend of parental genotypes that drove the emergence of the pathogenic , rough-colony morphology species , like M . tuberculosis , allowing their subsequent clonal expansion . Moreover , if DCT drove these postulated HGT events , the evolutionary clock for M . tuberculosis is likely much shorter because of the capacity of DCT to generate genome-wide mosaicism in a single step . Given the widespread nature of conjugation , we speculate that distributive conjugal transfer also occurs in other bacteria , conferring similar evolutionary benefits . The characteristics of mycobacterial distributive conjugation suggested to us that tools developed for mammalian genetics could be applied here . Using a eukaryotic-style genome-wide association mapping approach , we mapped the mating identity locus ( mid ) for mycobacterial conjugation ( Figure 7 ) . Similarly , we applied a backcross introgression strategy to refine the mapping and to purge extraneous mc2155 sequence ( Figure 7 ) . The purifying selection of successive backcross generations effectively introgressed the mc2155 mid locus into the mc2874 background; this created a strain that was nearly isogenic to the mc2874 parent strain , but which now functioned as a conjugal donor . We note that the hybrid esx1 loci produced by distributive conjugal transfer have not been disabled ( as in transposon mutagenesis screens ) , and still encode functional ESX-1 secretory apparatuses that secrete the major ESX-1 substrates ( Figure S6 ) . The un-annotated theoretical proteins encoded by the mid candidate genes bear no overt resemblance to those known to be involved in conjugation in other bacteria . Their association with the esx1 locus suggests that Mid proteins modify the ESX-1 secretion system , are secreted by ESX-1 , or interact with other ESX-1–secreted substrates . The next step in their functional assessment will likely result from an extension of this work to identify which protein ( s ) or protein motifs are necessary and sufficient to impart conjugal sex identity . Interestingly , orthologs for the mid candidate genes are found in the sequenced genomes of other environmental mycobacteria , suggesting a possible ongoing role for distributive conjugal transfer in gene flow between mycobacteria . Orthologs of these mid candidates are not apparent in the esx1 locus of M . tuberculosis , consistent with our speculative model that the MTBC represents a clonally expanded product of distributive conjugal transfer , not necessarily an active participant in this process . Nevertheless , recent evidence from genome sequencing comparisons indicates that some form of genetic exchange has occurred between M . tuberculosis and M . canettii [35] . While we applied DCT to map mid genes , in principle any genetic trait that differs between the parental strains can be mapped using this genome-wide mapping strategy . For example , mc2155 and mc2874 grossly differ in colony morphology , biofilm formation , and phage susceptibility , any of which could have been scored as a change of function in the recipient and mapped by DCT . Similarly , biochemical differences between these strains could be discerned through simple , high-throughput assays . We recognize that more traditional approaches for mutagenic loss-of-function mapping [38] , [39] will remain important in mycobacterial studies , but this new application of conjugation now allows any phenotype that differs between a mating pair to be unambiguously mapped . Our analysis of distributive conjugal transfer ( DCT ) in M . smegmatis has practical and conceptual ramifications . It brings new tools to mycobacteriology , including those traditionally used exclusively in eukaryotic genetics . It also shows how bacterial evolutionary time scales can be compressed by generating incredible genetic diversity in a single step . Identifying the necessary components , such as esx1 and mid , will help to elucidate the mechanism , to allow modification of the system , and to computationally identify bacteria that actively participate in DCT—or engineer them to do so . Our previous finding of DCT in a mixed biofilm [40] underscores the importance of predicting how prevalent DCT may be in nature , for a more accurate interpretation of metagenomic datasets and to model gene flow through bacterial populations . Regardless of these secondary ramifications , our primary finding of the tremendous genomic variation generated by DCT takes a significant step toward bringing the evolutionary benefits of sexual reproduction to bacteria .
M . smegmatis donor strains were derivatives of the laboratory strain , mc2155 [41] . Each derivative has a KmR gene inserted at a unique location in the chromosome [11] , which was mapped by DNA sequencing the flanking DNA and alignment to the mc2155 genome sequence ( http://cmr . jcvi . org/tigr-scripts/CMR/GenomePage . cgi ? org=gms ) , or the draft genome of the recipient ( GenBank CM001762 ) . The recipient strain mc2874 [18] , [42] was transformed with a plasmid encoding either apramycin or hygromycin resistance to allow counterselection against the donor . M . smegmatis strains were cultured at 37°C in Trypticase Soy Broth with 0 . 05% Tween80 , or on Trypticase Soy Agar ( TSA ) plates . Antibiotics were added at 100 µg/ml ( apramycin ) , 100 µg/ml ( hygromycin ) , and 10 µg/ml ( kanamycin ) . DNA transfer experiments were carried out as described previously selecting for dual-resistant transconjugants [8] . To allow selection in the reiterative backcrosses , the recipient strain was alternated between that encoding either apramycin or hygromycin resistance . Each independent transconjugant was assayed in subsequent mating experiments to determine whether they were donor or recipient , in parallel with positive controls . As we have observed previously [8] , this phenotype was mutually exclusive . Donor transfer frequencies were determined based on the average of three , independent mating experiments as described previously [8] . Zero transconjugants were obtained with recipient strains , below the sensitivity threshold of one event per 108 cells [8] . Transconjugants were colony purified , and genomic DNA was prepared and then subjected to whole-genome DNA sequence analysis at the Institute for Genome Sciences ( IGS ) , U . Maryland , using paired-end Illumina technology . The sequence coverage for each genome was between 50-fold for F1 progeny and ∼1 , 000-fold for backcross strains . Sequence reads were mapped to the mc2155 reference sequence by IGS . Single nucleotide polymorphisms ( SNPs ) or sequence gaps were identified using the Integrative Genomics Viewer ( IGV ) sequence viewer [43] to define genomic regions of different parental origins . Boundaries of recipient- and donor-derived segments were recorded as the last recipient SNP observed with a minimum of two consecutive SNPs defining parental identity ( Figure S2 ) . A donor segment unique to each transconjugant was identified to confirm accuracy of the aligned sequence reads . Primers were designed to specifically amplify these segments , and the amplified products were cloned and sequenced ( Table S1 ) to confirm that donor SNPs had been inherited by the recipient . A compilation of the donor and recipient segments from each transconjugant was projected onto the circular mycobacterial donor chromosome reference sequence , arranged as concentric circles of a Circos plot [44] , with color optimization guided by ColorBrewer ( Cynthia Brewer , The Pennsylvania State University ) . Collinearity of the donor and recipient genome was determined using Mauve , a program that was also used to identify SNPs and in/dels [45] , [46] . All sequence data have been deposited at the European Nucleotide Archive at http://www . ebi . ac . uk/ena/data/view/ERP002619 . | Bacteria reproduce by binary fission , generating two clones of the original; this restricts the genomic diversity of the population , which brings with it inherent evolutionary drawbacks . This problem can be eased by conjugation , which transfers DNA from a donor to a recipient bacterium . Understanding the potential of conjugal DNA transfer for generating genetic diversity is necessary for estimating gene flow through populations and for predicting rates of bacterial evolution . The influence of chromosomal conjugal DNA transfer on mycobacterial diversity has not been previously addressed . Here , we determine and compare the complete genome sequences of independent progeny from bacterial matings between defined donor and recipient strains of Mycobacterium smegmatis . We find the resulting hybrid bacteria to be extremely diverse blends of the parental strains , reminiscent of the genetic mixing that occurs through meiotic recombination in sexual organisms . This novel mechanism of conjugation can create genome-wide mosaicism in a single event , generating segments of donor DNA that range from small ( ∼0 . 05 kb ) to large ( ∼250 kb ) , widely distributed around the recipient chromosome . We exploit this mixing by using genetic tools originally developed for finding mammalian disease genes to locate the genes that confer a donor phenotype in M . smegmatis . We speculate that similar genomic mosaicism observed in pathogenic mycobacteria arose from conjugation between ancestral progenitor strains . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genome-wide",
"association",
"studies",
"genome",
"evolution",
"microbiology",
"genome",
"sequencing",
"prokaryotic",
"models",
"model",
"organisms",
"molecular",
"cell",
"biology",
"microbial",
"evolution",
"molecular",
"genetics",
"microbial",
"pathogens",
"comparative",... | 2013 | Distributive Conjugal Transfer in Mycobacteria Generates Progeny with Meiotic-Like Genome-Wide Mosaicism, Allowing Mapping of a Mating Identity Locus |
The human immunity-related GTPase M ( IRGM ) has been shown to be critically involved in regulating autophagy as a means of disposing cytosolic cellular structures and of reducing the growth of intracellular pathogens in vitro . This includes Mycobacterium tuberculosis , which is in agreement with findings indicating that M . tuberculosis translocates from the phagolysosome into the cytosol of infected cells , where it becomes exposed to autophagy . To test whether IRGM plays a role in human infection , we studied IRGM gene variants in 2010 patients with pulmonary tuberculosis ( TB ) and 2346 unaffected controls . Mycobacterial clades were classified by spoligotyping , IS6110 fingerprinting and genotyping of the pks1/15 deletion . The IRGM genotype −261TT was negatively associated with TB caused by M . tuberculosis ( OR 0 . 66 , CI 0 . 52–0 . 84 , Pnominal 0 . 0009 , Pcorrected 0 . 0045 ) and not with TB caused by M . africanum or M . bovis ( OR 0 . 95 , CI 0 . 70–1 . 30 . P 0 . 8 ) . Further stratification for mycobacterial clades revealed that the protective effect applied only to M . tuberculosis strains with a damaged pks1/15 gene which is characteristic for the Euro-American ( EUAM ) subgroup of M . tuberculosis ( OR 0 . 63 , CI 0 . 49–0 . 81 , Pnominal 0 . 0004 , Pcorrected 0 . 0019 ) . Our results , including those of luciferase reporter gene assays with the IRGM variants −261C and −261T , suggest a role for IRGM and autophagy in protection of humans against natural infection with M . tuberculosis EUAM clades . Moreover , they support in vitro findings indicating that TB lineages capable of producing a distinct mycobacterial phenolic glycolipid that occurs exclusively in strains with an intact pks1/15 gene inhibit innate immune responses in which IRGM contributes to the control of autophagy . Finally , they raise the possibility that the increased frequency of the IRGM −261TT genotype may have contributed to the establishment of M . africanum as a pathogen in the West African population .
Autophagy is induced by the formation of intracellular double-membrane structures which form autophagosomes to sequester and , after further maturation to autolysosomes , degrade cytosolic protein aggregates and corrupted cellular organelles . Thereby , it allows a re-cycling of amino acids , which is of particular importance during periods of cell starvation . Autophagy is also an efficient innate defense mechanism in that it may lead to deposition of intracellular pathogens into an autophagosome for subsequent autolysosomal acidification and peptidase-mediated degradation . The net effect of well functioning autophagy is protein degradation , optimized loading of Human Leukocyte Antigen ( HLA ) class II molecules and intensified antigen presentation [1] . The activation pathways of autophagy and phagocytosis are closely related in that they share phosphatidylinositol-3-phosphate ( PI3P ) and other factors in facilitating the fusion step and closure of the phagophore at the final stage of phagosome and autophagosome formation [2] . Among other factors , immunity-related GTPases exert significant effects on both autophagy and phagosome maturation [3] . It was recently shown in human U937 cells that inhibition of the immunity-related GTPase IRGM by siRNA caused impaired conversion of light chain 3 ( LC3 ) , an exclusive marker of the autophagy cascade , into its active form which is required for the elongation of the double membranes and eventual completion of the autophagosome [4] . In these experiments , impaired LC3 conversion resulted also in an extended survival of BCG in phagosomes . These experiments were based on earlier experiments in mice where Irgm1 ( syn . : LRG-47; encoded by Irgm1 ) , the murine homologue of IRGM , was critically involved through induction of autophagy in the control of several intracellular pathogens , including Mycobacterium bovis BCG and M . tuberculosis H37rv [1] , [3] , [5] , [6] . Notably , phagosome development into phagolysosomes may alternatively be induced by the autophagic component LC3 even when the conventional PI3P dependent pathway of phagosome maturation is blocked by experimental infection with the M . tuberculosis H37rv strain [3] . IRGM ( syn: LRG47 , IFI1 ) is encoded by the immunity-related GTPase protein family , M gene ( IRGM; 5q33 . 1; OMIM *608212 ) . IRGM consists of a long first exon encoding 181 amino acids , and four shorter putative exons that span more than 50 kilobases ( kb ) downstream of the first exon [7] . Most recent evidence indicates that the IRGM gene was deactivated during evolution but has regained its function after the insertion of a retroviral ERV9 segment and an Alu repeat sequence ( Figure 1; [8] ) . Genetic variants of IRGM , including a 20 . 1 kilo-base ( kb ) upstream deletion polymorphism , have been found to confer in Caucasians an increased risk of developing Crohn's disease , whereby coding-sequence variation has been excluded to be the liable source of the association [9]–[11] . The phylogenetic tree of the M . tuberculosis complex contains two independent clades . Clade 1 represents all M . tuberculosis sensu strictu lineages that may be exclusively pathogenic for humans , while clade 2 comprises lineages that are pathogenic for both humans and animals ( M . africanum , M . bovis ) [12] . A major branch of M . tuberculosis sensu strictu is the M . tuberculosis Euro-American ( EUAM ) lineage that recently was shown to be prevalent and cause significant numbers of infections in West Africa as well [13] . While M . tuberculosis lineages occur throughout the world , M . africanum strains are almost exclusively restricted to West Africa [12] , [14] , suggesting the existence of factors favoring spread and preservation of M . africanum in West Africa . Based on the in vitro evidence on the role of IRGM and autophagy in experimental M . tuberculosis infections of murine and human cells [3] , [4] we hypothesized that naturally occurring IRGM variants , including the large upstream 20 . 1 kb deletion , might also be relevant to the phenotype of in vivo human pulmonary tuberculosis ( TB ) . We have , in a case-control design , re-sequenced fragments of the IRGM gene and genotyped distinct genetic variants . An influence exerted by these variants on susceptibility or resistance to TB should be reflected in a large sample of active sputum-positive cases with pulmonary TB that we collected in Ghana , West Africa , and compared it to a control group of notable size .
A power of >90% of detection was achieved for multiplicative and additive models , assuming an approximative TB prevalence of 0 . 004 in West Africa , a frequency of 0 . 1 for high risk alleles , and a genotype relative risk of 1 . 3 ( α = 0 . 05 ) with our sample size ( case-control ratio = 1 . 18 ) . All variants tested were in HWE in cases and controls except IRGM −261 and IRGM −71 , where the distribution of alleles was in HWE among cases , but deviated in controls . The deviation could be traced to the subgroup of controls of Ewe ethnic background . As IRGM −261 and IRGM −71 are in strong LD but were determined by independent genotyping assays , a genotyping failure appears to be highly unlikely . The most plausible explanation is that the deviation results from the low number of Ewe controls ( 5% ) . Calculational exclusion of this subgroup from statistical analyses did not affect the significance of our results . As data on the frequency of the IRGM −261T variant in other ethnic groups are not available so far , we have typed this variant also in a small panel of 47 healthy Caucasian subjects . The genotypes CC , CT and TT were observed at frequencies of 0 . 81 , 0 . 15 and 0 . 04 , respectively , versus frequencies of 0 . 44 , 0 . 43 and 0 . 13 in the Ghanaian study population . Although not representative due to the low number of Caucasian individuals and to the bias imposed by a non-randomly selected African study group , the data suggests that the IRGM −261T allele occurs by far more frequently among West Africans . Re-sequencing revealed , in addition to previously recognized polymorphisms , ten yet unidentified IRGM variants which were submitted to the NCBI dbSNP database ( Table 1 ) . The positions and preliminary NCBI ss numbers of the novel variants are as follows: IRGM −908A>C ( ss105106760 ) , −797C>T ( ss105106761 ) , −420C>T ( ss105106763 ) , −284G>A ( ss105106764 ) , 281C>A ( ss105106765 ) , 370A>G ( ss105106766 ) , IRGM intron +2T>C ( ss105106767 ) , intron +106T>C ( ss105106768 ) and a deletion of two bases at positions −386/−387delAG ( ss105106769 ) . A tetranucleotide polymorphism ( rs60800371; repeats starting at nucleotide position −308 ) was newly recognized to appear as triple repeat . This variant is located at the 5′-end of the Alu segment ( Figure 1; [15] ) . Potential accumulation of transcription factor binding sites ( FOXP3 , GR , PR A/PR B ) may occur , depending on the number of tetranucleotide repeats . All novel variants were confirmed by forward and reverse DNA sequencing and were observed in at least two different DNA samples . The IRGM genotypes identified by re-sequencing of 69 DNA samples are given in Table 1 . Out of the total of 1567 isolates that were obtained and characterized , 1029 ( 65 . 7% ) were M . tuberculosis Euro-American ( EUAM ) strains ( pks1/15 7 base-pair [bp] deletion ) , 472 ( 30 . 1% ) were M . africanum , and 10 ( 0 . 6% ) were M . bovis ( the latter two lineages exhibiting the RD9 deletion ) . Fifty-six ( 3 . 6% ) isolates belonged to the M . tuberculosis East-African-Indian ( EAI ) , Beijing or Delhi lineages . Genotyping of eight IRGM variants was performed in 2010 HIV-negative TB cases and 2346 controls . The genotyping detection rate was >94% for all variants . In the entire study group , no allelic or genotypic association with disease or resistance was observed . A trend for an association with resistance to TB was , however , seen for the distribution of the IRGM −261TT genotype ( OR 0 . 79 , CI 0 . 65–0 . 96 , uncorrected nominal P value [Pnom] 0 . 017; Table 2 ) . Stratification for the two major phylogenetic mycobacterial clades , M . tuberculosis sensu strictu and M . africanum/M . bovis , revealed a significant difference in the distribution of the IRGM −261 genotype among cases and controls . IRGM −261TT was significantly associated with protection from TB caused by M . tuberculosis , but not by M . africanum/M . bovis ( OR 0 . 66 , CI 0 . 52–0 . 84 , Pnom 0 . 0009 , Pcorr 0 . 0045 vs . OR 0 . 95 , CI 0 . 70–1 . 30 , Pnom 0 . 8 ) . The association was confirmed in additive and recessive statistical models ( ORadd 0 . 86 , CI 0 . 77–0 . 95 , Pcorr 0 . 025 and ORrec 0 . 68 , CI 0 . 53–0 . 85 , Pcorr 0 . 005 , respectively; Table 3 ) . Further stratification with respect to mycobacterial genotypes revealed that the association of the −261TT genotype applied exclusively to carriers of the M . tuberculosis Euro-American ( EUAM ) genotype , but not to individuals infected with M . tuberculosis East-African-Indian ( EAI ) , Beijing or Delhi genotypes ( OR 0 . 63 , CI 0 . 49–0 . 81 , Pnom 0 . 0004 , Pcorr 0 . 0019 vs . OR 1 . 20 , CI 0 . 57–2 . 52 , Pnom 0 . 6 ) . Again , the association was substantiated in the additive and recessive models ( ORadd 0 . 85 , CI 0 . 76–0 . 95 , Pcorr 0 . 016 and ORrec 0 . 64 , CI 0 . 50–0 . 82 , Pcorr 0 . 0017 , respectively; Table 4 ) . No association was observed for carriers of the heterozygous IRGM −261CT genotype ( OR 0 . 96 , CI 0 . 82–1 . 13 , P 0 . 6 ) . For the low numbers of patients infected with mycobacteria of the EAI/Bejing/Delhi group , the statistical power might not allow to obtain significant results and a possible association could be easily overlooked . To rule out this possibility statistically , a supplementary test of interaction was performed [16] . This test permitted to verify complete independence of ORs and validation of the exclusive liability of the M . tuberculosis EUAM genotype for the observed association . Cluster analyses of cases were done as previously described [17] . The distribution of the two variants at IRGM position −261 among and within clusters of cases did not differ significantly . This applied also when stratifications for ethnicity were performed . Haplotypes and linkage disequilibria ( LD ) were reconstructed with the UNPHASED and Haploview softwares , respectively ( Table 5 , Figure 1 ) . The corrected global P value of 0 . 017 for haplotypes after 1000 permutations comprising all variants that were tested indicated that the observed association might apply also to a distinct haplotypic combination . Detailed analyses including 1000 permutations for each combination showed that the IRGM haplotype 20kbdel_DEL/−4299C/−566G/−420C/rep4/−261T/−71A/313T was associated with resistance to TB caused by EUAM strains ( OR 0 . 83 , CI 0 . 73–0 . 93 , P 0 . 001; Table 5 ) , albeit the haplotype was not found to exert a stronger effect than the −261T variant alone when occurring homozygously ( Tables 4 and 5 ) . This underscores the significance of the IRGM −261TT genotype in protection from human pulmonary tuberculosis caused by the species M . tuberculosis . A two-tailed student's t-test revealed a significant difference in the normally distributed ratio of firefly∶Renilla luciferase activity of five independent transfections per IRGM variant in the luciferase reporter gene assay , indicating increased gene expression in cells transfected with pGL3-Control Vector carrying the mutant variant IRGM-261T than in cells transfected with the pGL3-Control Vector with the IRGM wild-type variant ( −261C ) ( P = 0 . 013 ) ( Figure 2 ) .
We found in a large Ghanaian study group of HIV-negative patients with pulmonary TB and healthy control individuals that a distinct IRGM genotype , IRGM −261TT , was as a trend associated with decreased susceptibility to TB . After stratification for the two major mycobacterial clades and molecular subtypes the trend observed in the entire study group could be traced back to an association of IRGM −261TT with tuberculosis exclusively when caused by the M . tuberculosis EUAM lineage . No association was observed in comparisons of cases caused by M . tuberculosis EAI , Beijing , Delhi and M . africanum/M . bovis with controls . The frequencies of the synonymous variant 313T>C , the variant at position −4299 and of the 20 . 1 kb deletion , all found to be associated with Crohn's disease in other studies [9]–[11] , did not differ significantly between TB patients and controls . Upon infection of macrophages , M . tuberculosis initially resides in phagosomes . During their normal maturation , phagosomes fuse with lysosomes and establish a hostile environment for the pathogen , characterized by lysosomal enzymes , acid pH , reactive oxygen and nitrogen intermediates , and toxic peptides . Most macrophage mediated killing of intracellular pathogens occurs within the phagolysosome and pathogens have developed several mechanisms to avoid this vacuolar attack by escape and multiplication in the cytoplasm , inhibition of phagosome-lysosome fusion or circumventing the common endocytic pathway ( reviewed in [18] ) . With regard to M . tuberculosis H37Rv , translocation of the bacteria from phagosomes to cytosolic compartments has been observed to occur in nonapoptotic cells [19] . Notably , escape into the cytosol has also been observed of the M . tuberculosis strain CDC1551 in laboratory infections of the amoeba Dictyostelium discoideum [20] and loss of phagosomal membranes with release of mycobacteria has been observed in an in vitro system five days post infection [21] . Autophagy induced by IRGM can efficiently interfere with cytosolic replication of pathogens by trapping and recapture and subsequent degradation of bacteria [22] , [23] . In the LC3 dependent activation pathway , induction of autophagy contributes not only to the degradation of cytosolic components , but also to the maturation of mycobacterial phagosomes [3] , [4] . While M . tuberculosis can impair phagosome maturation by blocking the normal PI3P dependent pathway , this effect is restored and outbalanced by the alternate activation mode triggered by LC3 . This is consistent with the finding in mice and in human U937 cells that murine Irgm1 and human IRGM , respectively , play an important role in the containment of M . tuberculosis through efficient induction of autophagy [4] . The polymorphism that is associated with relative resistance to TB , IRGM −261T , in particular when occurring homozygously as TT genotype , might enhance expression of the mature IRGM protein which triggers autophagic degradation of translocated bacteria . The transcription factor AHR is expressed in the monocytic cell line THP1 that we used in our transfection experiments and inhibits differentiation of monocytes in vitro [24] , [25] . This suggests that an unaffected AHR/ARNT transcription factor complex that is likely to occur in the IRGM wild-type gene promoter , may decrease innate immune responses mediated by macrophages . Inversely , and based on our luciferase reporter gene assays , the loss of the potential PAX5 , AHR and ARNT transcription factor binding sites predicted for the IRGM variant −261T upstream of the coding region is likely to contribute to increased IRGM gene expression and , thus , enhanced innate responses . The fact that IRGM is not inducible by IFN-γ as are Irg genes in mice [4] , [7] , [26] but initiates successfully autophagy as does IFN-γ , argues for additional resistance mechanisms in individuals carrying the IRGM −261T variant homozygously , independent of other Th1-mediated immune responses . The association that we observed was restricted to infections caused by the M . tuberculosis EUAM lineage . The major characteristic of that lineage is the pks1/15 7 bp deletion which does not occur in M . africanum , M . bovis and in the other M . tuberculosis lineages identified in our study , EAI , Beijing , and Dehli . Why relative protection by IRGM −261TT is exclusively provided against disease caused by the lineage that exhibits the pks1/15 deletion remains to be explained further . One may hypothesize that absence or presence of the pks1/15 deletion may contribute to a modulation of the pathogenic potential of infecting strains in different ethnicities , leading to an adaption of mycobacterial lineages to their sympatric host population . Inhibition of innate immune responses and a tendency of increased spread of bacteria by phenolic glycolipid TB ( PGL-tb ) , the product of undamaged pks1/15 gene , has been demonstrated [27] , [28] . It is , therefore , reasonable to assume that lineages which do not produce PGL-tb as a suppressor of the innate immune response are more susceptible to IRGM-triggered innate immune mechanisms , namely phagosome maturation and autophagy . However , a more recent report indicates that PGL-tb itself does not confer hypervirulence , but rather differentially may modulate early cytokine response of the host [29] . As such , variable PGL-tb levels in conjunction with other yet inadequately defined mycobacterial virulence factors might result in a lineage-specific immunogenicity and/or degree of virulence that is related to the occurrence of human genetic variants in a particular population ( 14 , 29 ) . Since translocation of bacteria from the phagosome has only been observed for M . tuberculosis H37rv and M . leprae , but not for M . bovis BCG [19] , and was not tested for other lineages it is also conceivable that distinct pathogenic strains are subjected to mechanisms preventing translocation to cytosolic compartments and allowing escape from autophagy , a hypothesis that awaits further substantiation . This would support the view of increased virulence of lineages carrying intact pks1/15 genes such as M . tuberculosis Beijing and other lineages , but also underline the equivalent pathogenic potential of M . africanum compared to M . tuberculosis as has been observed in our study group [17] . It is intriguing to speculate that the high prevalence of IRGM −261TT in our Ghanaian study population and the relative protection that it confers from TB caused by M . tuberculosis EUAM might reflect one of multiple selection factors which promote the preferential and virtually exclusive propagation of M . africanum in West Africa , an assumption to be verified in other data sets of mycobacterial and human genotypes which are not yet available . The phylogeography of mycobacteria implies that lineages have become differentially adapted to different ethnicities with allelic variations conferring traits associated with certain infection phenotypes [14] . While the association of the IRGM polymorphism that we found cannot unquestionably confirm the role of IRGM in tuberculosis , it adds evidence to the in vivo experiments in mice and human cells [3] , [4] and supports the relevance of autophagy in the control of tuberculosis . Unfortunately , genetic replication data on the role of human IRGM polymorphism in infections caused by M . tuberculosis subtypes that have unambiguously been determined by mycobacterial genotyping are not available so far , neither for Caucasian nor for African or Asian study groups . Data of functional experiments on the effect of IRGM in tuberculosis caused by strains that produce PGL-tb are not available . The inhibition of the innate immune mediators TNF-alpha and other interleukins by PGL-tb has been described earlier [26] . A corresponding inhibition of IRGM would explain our findings and contribute to understanding the role of PGL-tb as a factor of virulence and modulator of innate immune responses .
The study protocol was approved by the Committee on Human Research , Publications and Ethics , School of Medical Sciences , Kwame Nkrumah University , Kumasi , and the Ethics Committee of the Ghana Health Service , Accra . Patients were treated according to the “Directly Observed Treatment , Short-course” ( DOTS ) strategy organized by the Ghanaian National Tuberculosis Programme . Blood samples for genetic analyses and HIV testing were taken only after a detailed explanation of the study aims and written or thumb-printed consent for participation provided , including HIV testing . Study participants were recruited in Ghana , West Africa , between September 2001 and July 2004 . The recruitment area and the enrollment procedure have previously been described [17] , [30] . Patients were enrolled at the two Teaching Hospitals in Accra and Kumasi and at additional hospitals or policlinics in Accra , Tema , Kumasi , Obuasi , Agona , Mampong , Agogo , Konongo and Nkawie ( Ashanti Region ) , Nkawkaw and Atibie ( Eastern Region ) , and Assin Fosu and Dunkwa ( Central Region ) . Characterization of patients included i ) the documentation of the medical history on standardized structured questionnaires , including self-reported duration of cough and symptoms of TB ( dyspnea , chest pain , night sweats , fever , hemoptysis , weight loss ) , ii ) two independent examinations of non-induced sputum specimens for acid-fast bacilli , iii ) serological determination of the HIV status and confirmation of positive results by an alternative test system , iv ) culturing and molecular differentiation of phylogenetic mycobacterial lineages , and v ) a posterior-anterior chest radiography . Inclusion criteria were two sputum smears positive for acid-fast bacilli , no history of previous TB or anti-mycobacterial treatment and an age between 6 and 60 years . Two sputum smears were examined in order to corroborate and confirm the phenotype . Patients were also included if only one smear and the culture for mycobacterial growth was positive . Exclusion criteria were incomplete information provided on the questionnaire , HIV positivity , evidence of alcoholism , drug addiction and other apparent generalized disease . A total of 2010 patients fulfilling the criteria for participation were enrolled . Unrelated personal contacts of cases and community members from neighboring houses of cases and public assemblies were recruited as controls . The leading criterion for enrollment as a control was no history of TB or previous anti-mycobacterial treatment . Characterization of controls included a medical history , posterior-anterior chest X-ray radiography and a tuberculin skin test ( Tuberculin Test PPD Mérieux , bioMérieux , Nürtingen , Germany ) . 1211 personal contacts and 1135 community members fulfilled the criteria for participation and were available as controls . Study participants belonged to the following ethnic groups ( cases/controls ) : Akan including Ashanti , Fante , Akuapem ( 63 . 6%/59 . 1% ) , Ga-Adangbe ( 14 . 5%/19 . 8% ) , Ewe ( 7 . 1%/9 . 3% ) and ethnic groups of northern Ghana including Dagomba , Sissala , Gonja , and Kusasi ( 12 . 9%/10 . 4% ) . The proportions of ethnicities among patients and controls did not differ significantly . Disclosure of HIV test results was dependent on the documented willingness of participants to be informed and included for HIV-positive patients their prompt referral to counseling and treatment provided by the Ghanaian AIDS Control Programme . The firm diagnosis of pulmonary TB was made as described previously [13] , [17] , [30] . M . tuberculosis complex isolates were cultured on Löwenstein-Jensen ( LJ ) media and shipped to the German National Reference Centre for Mycobacteria ( Borstel , Germany ) for minute analyses of biochemical , growth and molecular characteristics . Molecular differentiation of 1567 mycobacterial isolates included spoligotyping , IS6110 fingerprinting and typing of the pks1/15 deletion as described previously [31]–[33] . Mycobacterial strains were for further stratification grouped according to the major phylogenetic lineages [12] . The stepwise procedure of typing of mycobacteria included an initial cluster analysis of IS6110 fingerprinting data and lineage identification according to specific spoligotype signatures . Assignment of lineages was based on the MIRU-VNTRplus webpage ( www . miru-vntrplus . org; [34] ) and a reference strain collection using the Bionumerics 5 . 1 software ( Applied Maths , Sint-Martens-Latem , Belgium ) . Classification was confirmed by random selection of 20 strains of each group and testing for the presence of lineage-specific deletions as follows: pks1/15 for M . tuberculosis Euro-American , region of difference ( RD ) 239 for M . tuberculosis East African Indian ( EAI ) , RD9 and RD711 for M . africanum West African 1 , RD9 and RD702 for M . africanum West African 2 , and RD9 and RD4 for M . bovis . All M . tuberculosis strains with ambiguous lineage identification were confirmed as belonging to the Euro-American lineage by identifying the presence of the pks1/15 7 bp deletion . Deletion typing was performed using protocols available at the MIRU-VNTRplus webpage [34] . Clusters were defined as groups of affected individuals infected with mycobacterial strains exhibiting the same IS6110 fingerprinting patterns . If less than 5 bands were identified , analyses were supplemented by spoligotyping . Identical strains isolated from 2 or more patients were regarded as a cluster and strains found in 1 patient only were considered unique . For HIV-1/-2 testing of TB cases , a capillary test system ( Capillus , Trinity Biotech , Bray , Co Wicklow , Ireland ) was applied . HIV positivity was confirmed by the Organon Teknika Vironostika HIV-1/-2 EIA system ( Organon Teknika , Turnhout , Belgium ) . The rate of confirmation was 100% . HIV-positive TB patients were excluded from further genetic analyses . DNA was isolated from peripheral blood samples of participants ( AGOWA® mag Maxi DNA Isolation Kit , Macherey & Nagel , Germany ) following the instructions of the manufacturer . The reference sequence was derived from the chromosome 5 contig , GenBank NW_922784 . 1 , region 23935681…23938058 . In order to reliably identify the IRGM variants occurring in our study population and to select variants for genotyping , a segment containing 1053 bp of the 5′UTR region upstream of the ATG start codon including the intron and the Alu segment , the open reading frame ( ORF ) of 546 coding bp and 250 bp of the 3′ region distal of the ORF of the IRGM gene were sequenced . Sequence analysis was performed in 23 TB patients , 23 PPD-positive and in 23 PPD-negative controls . Sequencing reactions were run on an automated ABI 3100 DNA sequencer ( Applied Biosystems , Foster City , USA ) . Forward ( F ) and reverse ( R ) primers for re-sequencing were IRGM_pro1000F ccttgaaaaagagcagagcatt and IRGM_pro1000R tagcatccccagccctca ( 598 bp ) , IRGM_pro500F ttgctccctgaagaaatgtg and IRGM_pro500R ctcaacattcatggcttcca ( 599 bp ) , IRGM_p1F aatatctgcgtccagggttc and IRGM_p1R tgaactgcatttccatcagg ( 572 bp ) and IRGM_p2F tgtgcctcctatttctcttcc and IRGM_p2 tgatataatcttgcatccattttaag ( 600 bp ) . Based on the results of re-sequencing and evidence of association of with other conditions , eight IRGM variants were selected for genotyping ( Table 6 ) . The variant at position −566G/C ( rs17111379 ) was chosen for genotyping , as this variant was by re-sequencing identified in the groups of TB cases and PPD-positive controls only . Notably , IRGM −566 allele frequencies are unevenly distributed between Caucasians and the West African ethnicity of Yoruba ( Caucasians: G 1 . 0; Yoruba , originating from Nigeria: G 0 . 74 , C 0 . 26; www . hapmap . org/cgi-perl/snp_details ? name=rs17111379&source=hapmap_B35 ) . The variant at position −420C/T ( ss105106763 ) was included as it was identified as a novel variant in our study population and was not in strong linkage with other polymorphisms . The IRGM variant −261 ( rs9637876 ) was selected for genotyping , because , according to an in silico prediction of transcription factor binding properties , the allele causes loss of several binding sites ( PAX5 , ARNT , AHR; http://alggen . lsi . upc . es/recerca/menu_recerca . html ) . IRGM −71 ( rs9637870 ) and the microsatellite repeat ( rs60800371; repeats starting at position −308 ) are in LD with IRGM −261 and were included in genotyping to more reliably identify a potentially causative variant . The synonymous exonic variant 313C/T ( L105L; rs10065172 ) , the 20 . 1 kb deletion and the variant located 124 bp downstream of the 20 . 1 kb deletion ( rs13361189; position −4299 ) were included in the genotyping panel due to their proven association with Crohn's disease [9]–[11] . Further variants that were identified by DNA re-sequencing were not subjected to genotyping , as they were either in perfect LD with +313 ( −964A/C , −787C/T , +87A/G ) , or too low in their frequencies to allow for reasonable statistical calculations ( −908A/C , −844C/T , −797C/T , −386/7delAG , −284G/A , 281C/A , 370A/G , +2T/C , +106C/T ) . IRGM variants were analyzed by dynamic allele-specific hybridization with fluorescence resonance energy transfer ( FRET ) in a LightTyper device ( Roche Diagnostics , Mannheim , Germany ) . Primer pairs , sensor- and anchor-oligonucleotides utilized are listed in Table 6 . As no frequency information was available for the occurrence of the IRGM −261 polymorphism among Caucasians , this variant was also genotyped in 47 healthy voluntary German individuals ( employees of the Bernhard Nocht Institute for Tropical Medicine , Hamburg; informed consent obtained ) . For the microsatellite repeat , a FRET-based LightTyper method was developed using a sensor nucleotide that covered the entire tetranucleotide repeat with adjacent 5′- and 3′-nucleotides [15] . Depending on the actual number of the repeats , distinct melting temperatures allow explicit specification of the number of repeats . For the 20 . 1 kb deletion upstream of the IRGM promoter , the LightTyper design was combined with a gap-PCR comprising two forward primers matching upstream and within the region of deletion . The assay enabled the determination of gene fragments with and without the 20 . 1 kb deletion . All plasmid constructions were based on the pGL3-Control Vector ( Promega , Mannheim , Germany ) which contains the SV40 promoter driving the firefly luciferase gene . Two fragments , each of 0 . 4 kb , comprising the Alu-repeat sequence and , thus , including the IRGM variants of interest , −261C or −261T , were PCR-amplified with oligonucleotide primers 5′-cgaagctttcacactctattagctgcatccttaac-3′ and 5′-cgccatggctttctcaacattcatggcttccat-3′ from 10 ng of genomic human DNA ( Biomers . net GmbH , Ulm Germany ) . PCR conditions were: Initial denaturation ( 98°C , 30′ ) , 30 amplification cycles ( 98°C , 7′; 64°C 20′; 72°C , 35′ ) and final elongation ( 72°C , 7′ ) . HindIII and NcoI restriction sites at the 5′ and 3′ end , respectively , were engineered on each PCR product . Fragments were then ligated into the HindIII-NcoI digested pGL3-Control Vector to generate plasmids pGL3 −261C and pGL3 −261T . Enzymatic digestions and ligations were performed according to the instructions of the manufacturer . Small-scale preparations were done applying the NucleoSpin Plasmid Kit ( Macherey and Nagel , Düren , Germany ) . Plasmid DNAs were propagated in E . coli XL1- Blue cells ( Stratagene , La Jolla , CA , USA ) and prepared using the EndoFree Plasmid Maxi Kit ( Quiagen , Hilden , Germany ) . DNA sequences of the final constructs were confirmed by sequencing . 1×106 cells of the human monocytic cell line THP1 ( German Resource Centre for Biological Material , DMSZ , Braunschweig , Germany ) were transfected with either 0 . 5 µg of the two plasmid constructs , with the addition of 0 . 5 µg of the plasmid phRL-CMV which contains the gene encoding Renilla luciferase ( Promega , Mannheim , Germany ) in order to normalize results . Transfectiosn with 0 . 5 µg of the unmodified pGL3-Control Vector and 0 . 5 µg phRL-CMV , as well as a mock transfection , were performed for control and to minimize effects of reagents on the cells . All transfections were performed with the Amaxa Cell Line Nucleofector Kit V ( Lonza , Cologne , Germany ) . Four hours after transfection , cells were harvested and luciferase activities were measured ( Dual-Luciferase Reporter Assay System; Promega , Mannheim , Germany ) . Firefly luminescence was measured using a single tube Junior LB9509 luminometer ( Berthold Technologies , Bad Wildbad , Germany ) . After a 10 second measurement period , 100 µl 1× Stop & Glo Reagent were added for the detection of Renilla luminescence . Measurements of luminescence are expressed as relative light units ( RLU ) . For each variant , five independent transfections were performed . Demographic data , self-reported signs and symptoms as documented on structured questionnaires as well as laboratory results were double-entered into a Fourth Dimension database ( San Jose , CA , USA ) . Bacteriological data were provided as Excel datasheets . Data were locked before using them in a pseudonymized form for statistical analyses . Power calculations were performed with the public CATS software ( http://www . sph . umich . edu/csg/abecasis/CaTS/ ) . Multivariate logistic regression analyses were calculated for different models to determine odds ratios ( OR ) for allele and genotype distributions ( STATA 10 . 0MP software; Stata Corporation , College Station , TX , USA ) . As age , sex and ethnicity were significant confounders , they were appropriately adjusted for . Analyses of allele distributions and Hardy-Weinberg equilibria ( HWE ) were calculated with a public STATA module ( www-gene . cimr . cam . ac . uk/clayton/software/stata/genassoc; David Clayton , Cambridge , UK ) . Haplotypes were estimated with the UNPHASED software ( version 3 . 0 . 13; Frank Dudbridge , http://www . mrc-bsu . cam . ac . uk/personal/frank/software/unphased/ ) , whereby incomplete haplotypes were subjected to simulations comprising all possible haplotypic completions of available data . In order to verify the independence of ORs that were obtained for associations of different mycobacterial clades/genotypes with phenotypes , ORs were subjected to tests of interaction according to the method described in [16] . Corrections of nominal P values ( Pnom ) for multiple testing applied to the number of variants tested and when stratifications by mycobacterial lineages were made . Bonferroni corrected P values ( Pcorr ) <0 . 05 were considered significant and correction values are indicated where applicable . The tetranucleotide repeat rs60800371 and SNPs at positions −261 ( rs9637876 ) and −71 ( rs9637870 ) are in strong LD ( r2∼0 . 8 ) and were , therefore , combined as a single correction entity . Global P values of haplotype associations and of distinct haplotypes were subjected to 1000 permutations ( UNPHASED software ) . Shapiro-Wilk tests and a two-tailed student's t-test were calculated to confirm a normal distribution and differences in the ratio of firefly∶Renilla luciferase activity in the reporter gene assay . | Autophagy is a process in which cell components are degraded by the lysosomal machinery . It has recently been described that activation of autophagy reduces the viability of M . tuberculosis in phagosomes due to an intimate autophagy-phagocytosis interaction . M . tuberculosis may also be directly accessible to autophagy , as M . tuberculosis was found to translocate into the cytoplasm . The immunity-related GTPase IRGM is a mediator of innate immune responses and induces autophagy . We have studied genetic variants of the human IRGM gene in a Ghanaian tuberculosis case-control group and found that the IRGM variant −261T provides relative protection against disease when the infection is caused by the Euro-American lineage of M . tuberculosis . This lineage is characterized by the pks1/15 seven base-pair ( bp ) deletion . The product of an intact pks1/15 gene , phenolic glycolipid-tb , might contribute to mycobacterial virulence by suppressing innate immune responses . It is , therefore , conceivable that only the Euro-American lineage is exposed to IRGM-triggered innate defence mechanisms . Our observations suggest that the increased frequency of the IRGM −261TT genotype may have allowed the establishment of M . africanum as a pathogen in West Africa . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases",
"genetics",
"and",
"genomics",
"microbiology/microbial",
"evolution",
"and",
"genomics"
] | 2009 | Autophagy Gene Variant IRGM −261T Contributes to Protection from Tuberculosis Caused by Mycobacterium tuberculosis but Not by M. africanum Strains |
The standard approach for identifying gene networks is based on experimental perturbations of gene regulatory systems such as gene knock-out experiments , followed by a genome-wide profiling of differential gene expressions . However , this approach is significantly limited in that it is not possible to perturb more than one or two genes simultaneously to discover complex gene interactions or to distinguish between direct and indirect downstream regulations of the differentially-expressed genes . As an alternative , genetical genomics study has been proposed to treat naturally-occurring genetic variants as potential perturbants of gene regulatory system and to recover gene networks via analysis of population gene-expression and genotype data . Despite many advantages of genetical genomics data analysis , the computational challenge that the effects of multifactorial genetic perturbations should be decoded simultaneously from data has prevented a widespread application of genetical genomics analysis . In this article , we propose a statistical framework for learning gene networks that overcomes the limitations of experimental perturbation methods and addresses the challenges of genetical genomics analysis . We introduce a new statistical model , called a sparse conditional Gaussian graphical model , and describe an efficient learning algorithm that simultaneously decodes the perturbations of gene regulatory system by a large number of SNPs to identify a gene network along with expression quantitative trait loci ( eQTLs ) that perturb this network . While our statistical model captures direct genetic perturbations of gene network , by performing inference on the probabilistic graphical model , we obtain detailed characterizations of how the direct SNP perturbation effects propagate through the gene network to perturb other genes indirectly . We demonstrate our statistical method using HapMap-simulated and yeast eQTL datasets . In particular , the yeast gene network identified computationally by our method under SNP perturbations is well supported by the results from experimental perturbation studies related to DNA replication stress response .
Recent advances in the next-generation sequencing and other high-throughput technology has allowed researchers to collect various types of genome-scale datasets , providing unprecedented opportunities to discover detailed gene regulation processes in cells via analysis of the massive data . The standard approaches for identifying the causal regulatory relationship among genes have been based on examining gene-expression data collected after an experimental perturbation of one or two genes such as in gene knockout studies [1] , [2] or over-expression studies [3] . In a typical experimental design , genome-wide gene-expression levels are measured using microarrays under different experimental conditions such as strains with one or two genes knocked out [1] , [2] . Then , the differential gene-expression patterns between control and experimental conditions are examined to obtain clues as to the organization of genes into functional modules and key regulators of those modules . However , this standard approach for identifying the wiring of gene networks based on experimental perturbations of gene regulation system comes with many limitations . Experimentally perturbing the activity of a gene can be very costly , time-consuming , and laborious and it is even more so for repeating such perturbation for every single gene in an organism to obtain a comprehensive picture of gene network wiring . Furthermore , the experimental methods are usually limited to a perturbation of one or two genes at a time due to experimental infeasibility and combinatorial explosion in the number of experiments to perform . Thus , they cannot be used to perturb more than two genes at the same time to obtain information on multifactorial gene interactions . More importantly , it is often not possible to apply such experimental perturbation studies to humans for ethical reasons . Finally , given the set of differentially expressed genes under each perturbation , it is difficult to distinguish between those genes that are directly regulated by the perturbed gene and those genes in the downstream of the pathway whose expressions are influenced as secondary/indirect effects . Genetical genomics approach has been proposed as a less expensive but more powerful alternative to the approach with experimental perturbations [4] , [5] . Genetical genomics treats genetic variation as naturally-occurring perturbation of gene regulatory networks and tries to learn gene networks by examining the effects of genetic variation on gene expression measurements within a large population of individuals . The key advantage of genetical genomics approach is that unlike experimental perturbations that can be performed only on one or two genes at a time , there are more than millions of genetic variants across genomes in the case of a human population , enabling the effects of multifactorial perturbations to be observed directly in gene expression data . Another advantage is that while experimental perturbation studies involve artificial perturbations in lab with often large perturbation effects , the perturbations of gene network by genetic variants occur in nature and usually induce more subtle changes in gene expressions . Thus , understanding the consequences of network perturbations by genetic variants is likely to lead to more direct understanding of the gene networks that exist in nature . In addition , genetical genomics approach can be easily applied to humans as well as other organisms , since genotype and gene-expression data are routinely collected for the purpose of expression quantitative trait locus ( eQTL ) mapping to understand the genetic architecture of complex phenotypes and diseases [6] , [7] . However , the genetical genomics approach poses a significant computational challenge , because it is not obvious how to decode the effects of multifactorial genetic perturbations from genotype and gene expression data . In an experimental approach with a perturbation of one or two genes , the genes that are differentially expressed under each perturbation experiment can be easily identified with a simple computation . However , in genetical genomics approach , the gene expression variability is the result of aggregated effects of multifactorial perturbations by a large number of genetic variants and it is not obvious how to decouple the aggregated perturbation effects to identify the set of genes that are differentially expressed with respect to each perturbation by each individual genetic variants . Furthermore , as many of the genetic variants do not have any functional consequences or perturbation effects , genetical genomics approach has an additional computational challenge of identifying eQTLs or the genetic variants that affect gene expressions , while identifying the gene network perturbed by these eQTLs at the same time . Because of these computational challenges , given gene-expression and genotype data , researchers tended to limit their analysis to eQTL mapping , where eQTL mapping can be viewed as a special case of genetical genomics analysis that assumes an isolated effect of genetic perturbation on a single gene with no downstream effects in gene network . While statistical methods such as graph-guided fused lasso ( GFlasso ) [8] have been developed to detect genetic effects on multiple correlated gene-expression traits , they focused only on eQTL mapping assuming a known gene network , instead of performing a more powerful genetical genomics analysis . Those few existing computational methods for genetical genomics analysis have been limited in terms of computational efficiency and statistical power [9]–[12] . For example , discovering eQTLs and reconstructing the gene network perturbed by those eQTLs were performed in two separate steps , leading to reduced statistical power [9] , or average gene-expression levels within each gene module were used as a trait to identify eQTLs rather than using the original gene-expression data , leading to the loss of information on individual gene activities [13] . In this paper , we propose a statistical framework that directly addresses all of the above computational challenges of genetical genomics analysis within a single statistical analysis to achieve the maximum statistical power for identifying gene networks via single-nucleotide-polymorphism ( SNP ) perturbations . Given SNP genotype and gene-expression data collected for a large number of individuals , our statistical method simultaneously identifies 1 ) the gene network structure by decoding the effects of multifactorial perturbations of gene regulation system by a large number of SNPs , 2 ) eQTLs that perturb this gene network , 3 ) the genes whose expressions are directly perturbed by eQTLs and the genes whose expressions are indirectly perturbed as secondary downstream effects of the direct perturbations in the network , and 4 ) detailed characterizations of the SNP perturbation effects on the gene network by decoupling the complex multifactorial SNP effects on the gene network with respect to each individual perturbation . Our proposed statistical framework is based on probabilistic graphical models , and in particular , we introduce a new statistical model , called a sparse conditional Gaussian graphical model ( CGGM ) , that models a gene network under SNP perturbations as an undirected graphical model . In our statistical model , the unknown gene network is represented as a graph over gene-expression traits and this graph is associated with an unknown probability distribution that models the strengths of gene-gene interactions in the gene network and the strengths of perturbation effects of SNPs ( Figure 1A ) . Then , both the gene network structure perturbed by SNPs and probability distribution associated with the network structure are learned jointly from data . We show that the learning problem is convex , leading to increased statistical power and a guarantee in the quality of the estimated model , and develop an efficient learning algorithm that scales to a large dataset . Given the estimated graphical model , we describe inference methods to characterize the detailed mechanisms of how the effects of SNP perturbations propagate through the network ( Figures 1B–D ) . From the computational point of view , addressing the challenges of genetical genomics analysis requires handling the computational challenges of both gene-network analysis given gene-expression data and eQTL mapping given gene-expression and genotype data , namely gene-network structure learning and SNP feature selection , at the same time . We show that in fact our sparse CGGM subsumes as special cases both a sparse Gaussian graphical model [14]–[17] , which is popular as a model for gene network , and a sparse linear regression model [8] , [18] , [19] , which is widely used for eQTL mapping , thus providing a natural unifying representation for a gene network and eQTLs perturbing this network . Moreover , by embedding the standard regression model for eQTL mapping within a probabilistic graphical model and leveraging the representational power of a graphical model , our approach allows to extract a significantly more detailed characterization of the functional roles of eQTLs than any of the existing methods for eQTL mapping . In our experiments , we apply our statistical framework to HapMap-simulated and yeast eQTL datasets [20] , [21] . Using HapMap-simulated data , we demonstrate our approach can recover the true underlying gene network under SNP perturbations , and at the same time , can recover true eQTLs with greater statistical power than other existing methods that have been developed for eQTL mapping . In addition , we applied our method to yeast eQTL dataset collected for 112 segregants of two yeast parent strains , BY4716 and RM11-1a [21] . Nearly all of the previous analyses of this dataset focused either on eQTL mapping or on gene network analysis ignoring the genetic information , with an exception of the genetical genomics analysis of this dataset performed by multivariate regression with covariance estimation ( MRCE ) [22] that has also been mistakenly called sparse CGGM [23] as we further discuss in the next section . However , the analysis by MRCE [23] was performed for only a small subset of the full dataset because of its expensive computational cost . In our experiment , MRCE took more than weeks of computation to analyze the full yeast eQTL dataset , whereas our method ran within a day . We provide an in-depth analysis of SNP perturbations of a subnetwork over genes involved in stress response in yeast , and show that this subnetwork obtained from SNP perturbations by our approach is well supported by the network inferred from experimental perturbations in knock-out studies in the literature .
A Gaussian graphical model defines a probability distribution over an undirected graph that models a gene network . The nodes of the graph correspond to continuous-valued random variables for gene-expression traits and the edges represent probabilistic conditional dependence relationships between pairs of nodes [14]–[17] . Given microarray gene-expression measurements for genes and individuals , a Gaussian graphical model assumes that the gene-expression measurement for the th individual is an independently and identically distributed sample from a Gaussian distribution , where is a vector of 0's and is a covariance matrix . Then , it is well-known that the inverse covariance represents a Gaussian graphical model , where the non-zero ( or zero ) value for in the th entry of represents the presence ( or absence ) of edges between the th and th gene-expression traits in the gene network . While each non-zero element in implies conditional dependency between the th and th gene-expression traits given all the other gene-expression traits , computing the inverse of to obtain the covariance amounts to performing an inference in this graphical model to obtain marginal dependencies , or equivalently dependencies between the two nodes without consideration of any other nodes . Graphical lasso [14] , [16] , [24] has been widely used to learn a sparse Gaussian graphical model , where only statistically significant gene-gene interactions have edges with non-zero entries in . Graphical lasso minimizes the negative log-likelihood with sparsity-inducing penalty as follows: ( 1 ) where is the trace of matrix , is the sample covariance matrix , is the norm of , and is the regularization parameter that determines the amount of sparsity . A large value of leads to a sparser estimate with a greater number of zero elements in . The optimal value for can be determined using cross-validation . The problem in Eq . ( 1 ) is convex and can be solved efficiently [16] . In eQTL mapping , the problem of identifying SNPs influencing gene-expression levels from eQTL data is often formulated as that of learning a multivariate linear regression model [8] , [18] , [19] . Given genotype data for SNPs , where is a vector of length with each element taking values from for the number of minor alleles at the given locus , and the expression measurements for genes for samples , a linear regression model for the functional mapping from SNPs to gene-expression traits is given as: ( 2 ) where is the regression coefficient matrix representing the unknown association strengths , and is the matrix of noise terms whose rows are Gaussian-distributed with mean zeros and covariance . Typically , a model without an intercept is considered , assuming that the genotype data are standardized to have mean and unit variance . Since each gene-expression trait typically has only a small number of eQTLs , lasso [19] , [25] has been widely used to obtain a sparse estimate of . Lasso minimizes the squared-error criterion with penalty as follows: ( 3 ) where is a regularization parameter that controls the amount of sparsity in . Eq . ( 3 ) is convex with a single globally optimal solution , and efficient algorithms are available for solving it [26] . Lasso essentially performs separate regression analyses , treating the gene-expression traits as independent of each other . In order to combine the statistical power across multiple correlated gene-expression traits , GFlasso [8] assumed a known gene network and extended the standard lasso by including an additional penalty , called graph-guided fusion penalty , that encourages multiple related genes in the gene network to be influenced pleiotropically by a common SNP . Given a gene network with a set of edges and edge weights 's for each edge between the th and th genes , the graph-guided fusion penalty takes the form of , where each term in the penalty encourages the amount of influence of the th SNP on the expression levels of the th and th genes to be similar if the two genes are connected with an edge in the network . GFlasso was capable of identifying SNPs with pleiotropic effects , but it was restrictive in that the gene network should be known a priori . Within the same statistical framework of linear regression method , MRCE [22] attempted to shift the focus from eQTL mapping towards genetical genomics analysis by identifying the gene network and eQTLs jointly . Towards this goal , MRCE relaxed the assumption of uncorrelated noise ( i . e . , being a diagonal matrix ) in lasso and estimated the full noise covariance matrix . Then , the inverse of the noise covariance corresponds to a gene network . MRCE minimizes the negative log-likelihood of data with an penalty for both and : ( 4 ) where and are the regularization parameters . We notice that unlike GFlasso , MRCE does not have any mechanisms to leverage the estimated gene network to model pleiotropic effects of SNPs on multiple correlated gene-expression traits in . The optimization problem in Eq . ( 4 ) is not convex , but bi-convex , since fixing either or and solving for the other is a convex optimization problem . Thus , Rothman et al . [22] proposed to optimize for each of and alternately given the other over iterations . However , they noted that this strategy often does not converge , and instead suggested to use an approximate method that prematurely terminates the iterative optimization procedure after two iterations . As we discuss in the Results section , we found that even this approximate method was too slow to be applicable to a dataset of even moderate size . The same statistical method for MRCE has been proposed independently in the literature under the name of sparse conditional Gaussian graphical models ( CGGMs ) [23] . However , we emphasize that the statistical model that is learned in MRCE is not a graphical model , because as we further discuss in detail in the next sections , the parameters do not model conditional dependencies as in graphical models but only models marginal dependencies . We believe that MRCE was mistakenly called a sparse CGGM due to the resemblance between the inverse noise covariance matrix in MRCE and the inverse covariance matrix in graphical lasso as well as the aspect of the standard regression model as a conditional model for given . The sparse CGGM that we propose in this paper is set up as a proper probabilistic graphical model and as we show in the next sections , is significantly more powerful than MRCE in terms of representational power and computational efficiency . Under the same name of sparse CGGMs , Li et al . [27] introduced a related but different statistical method that also models a gene network corresponding to in MRCE . However , unlike our approach and MRCE , the estimation procedure for the sparse CGGM in [27] amounted to a two-stage process , where the gene-expression data are pre-processed to remove SNP effects in the first stage and then these pre-processed gene-expression data are used to learn a gene network in the second stage . Thus , their graphical model was defined only on gene-expression traits and did not directly model the relationship between SNPs and gene expressions to identify eQTLs . In contrast , our sparse CGGM is set up as a graphical model on both gene expressions and SNPs , performs a joint estimation of gene network and eQTLs , and infers various perturbation effects of SNPs on gene expressions via inference . In this section , we introduce a statistical model for CGGM as a model for a gene network under SNP perturbations . Then , in the next sections , we describe a learning algorithm for estimating a sparse model for CGGM from data and discuss inference schemes for the estimated sparse CGGM . A sparse CGGM estimated from data captures the gene network structure and direct perturbations of the gene-expression levels by eQTLs via conditional dependency structure in the estimated graph structure . By performing inference on this estimated sparse CGGM , we can obtain a detailed characterization of how the direct SNP perturbation effects propagate through the gene network to perturb the expression levels of other genes indirectly . The key idea behind our proposed approach is to model a gene network under SNP perturbations as a Gaussian graphical model for a gene network conditional on SNPs . We derive a CGGM as a conditional distribution from the Gaussian graphical model for a joint probability distribution for SNPs and gene-expression traits for the th individual . Let us assume a Gaussian graphical model with covariance and inverse covariance , where zero mean is assumed after mean-centering each gene-expression trait and SNP . Then , the conditional distribution of given can be obtained as . We further re-write this conditional distribution , using the inverse covariance matrix and the partitioned inverse formula [28] to obtain a CGGM: ( 5 ) CGGM parameters and represent a gene network and SNP perturbation effects on this gene network , respectively . A non-zero value for the th element of indicates that the th SNP is an eQTL for the th gene-expression trait . This SNP perturbation captures the direct influence of the th SNP on the th gene expression , since the graphical model captures conditional dependencies . While in experimental perturbation studies the expressions of only one or two genes can be directly perturbed ( e . g . , by knocking out the genes ) , our CGGM for genetical genomics study allows multiple gene-expression traits to be perturbed by multiple SNPs at the same time . Then , this multifactorial genetic perturbations of gene-expression levels are decoded to learn a gene network by a learning algorithm that estimates and simultaneously . In order to show the direct correspondence between a CGGM and a general undirected graphical model , we re-write Eq . ( 5 ) by expanding the quadratic term in the Gaussian distribution in Eq . ( 5 ) to obtain: ( 6 ) where is a constant , also known as a partition function in the literature of probabilistic graphical models [29] , which ensures that forms a proper probability distribution integrating to . As Eq . ( 6 ) is equivalent to the Gaussian-distribution form in Eq . ( 5 ) , this constant can be obtained in a closed-form by directly comparing Eq . ( 6 ) with Eq . ( 5 ) . If is positive definite , the integral in the partition function is finite and the probability distribution is well-defined . The representation in Eq . ( 6 ) explicitly shows that a CGGM is an undirected graphical model [29] defined over a graph with two sets of edges , namely the set of edges connecting each pair of gene-expression traits in gene network and another set of edges connecting each SNP to gene expressions that the SNP is influencing ( Figure 1A ) . Then , following the definition of an undirected graphical model [29] , the numerator in Eq . ( 6 ) is a weighted sum of features over the graph edges , where 's and 's are features and and define edge weights . The gene network is modeled an undirected graph , but the directions from SNPs to gene-expression traits are implicit , since the model is a conditional probability model for gene-expression traits conditional on SNPs . Since both gene-gene interactions and SNP perturbations of gene-expression traits are highly modular and localized , we are interested in learning a sparse model for CGGM . In other words , only statistically significant gene-gene interactions should be represented as edges with non-zero entries in and only a small number of statistically significant direct SNP perturbations should be estimated as having non-zero effect sizes in . In order to impose a sparsity constraint , we learn a sparse CGGM by minimizing the negative log-likelihood of data with an penalty as follows: ( 7 ) where is the negative log-likelihood of data based on Eq . ( 6 ) or equivalently Eq . ( 5 ) , and and are the regularization parameters that control the amount of sparsity . It is not necessary to explicitly consider the positive-definite constraint for within the optimization problem in Eq . ( 7 ) , because the partition function in the data log-likelihood contains term that acts as a log-barrier function for the positive-definite constraint [30] . Within the penalty for , we do not penalize the diagonal elements of , since we found that this leads to a slightly better performance in our experiments , consistent with what has been reported for graphical lasso . It is straightforward to prove that the problem in Eq . ( 7 ) is convex ( Text S1 ) [30] . Thus , the learning algorithm is guaranteed to find the globally optimal solution that achieves the maximum statistical power . The main challenge for solving Eq . ( 7 ) arises from the non-smoothness of the penalty function . We adopt a variant of accelerated proximal gradient algorithms , called a Nesterov's second method [31] , that has been developed as a general-purpose algorithm for handling a non-smooth component of the parameter estimation problem while improving the convergence ( and thus , computation time ) of the standard gradient descent algorithm [31]–[33] . We provide details of the learning algorithm in Text S2 . Lasso [25] and graphical lasso [14] , [16] , [24] can be viewed as special cases of the sparse CGGM estimation problem in Eq . ( 7 ) . When , the sparse CGGM learning problem in Eq . ( 7 ) essentially reduces to applying graphical lasso to gene-expression data , ignoring genotype data , since the large encourages all or nearly all of the elements of to be set to zeros . On the other hand , if , the sparse CGGM learning problem becomes equivalent to lasso that fits a regression model for each gene-expression trait separately , ignoring gene network , since the large tends to set all or almost all of the off-diagonal elements of to zeros . The optimal values for and that strike the right balance between these two extreme cases can be found by cross-validation . So far , we showed that by learning a sparse CGGM , it is possible to decode the underlying gene network and its direct multifactorial perturbations by SNPs from data . Now , we show that given a sparse CGGM estimated from data , we can perform inference on this graphical model to characterize the mechanisms of SNP perturbations of gene network in detail . Below , we discuss how inference schemes can be used on our estimated model to learn about indirect/secondary downstream effects of the direct SNP perturbations , a decomposition of the overall multifactorial SNP perturbation effects with respect to each individual direct perturbation , and a decomposition of observed covariance in gene expressions into genetic and non-genetic components . We note that all of these inference schemes involve only few simple matrix operations and are highly efficient .
In order to simulate eQTL datasets , we used the SNP genotype data from HapMap phase III release 2 [20] as SNP data and simulated gene-expression traits , given and known model parameters . We used the SNP data for chromosome 21 of the 343 individuals of African origin , including ASW , LWK , MKK , and YRI population groups . After removing SNPs with minor allele frequency and highly correlated SNPs with squared correlation coefficient , we obtained 4 , 901 SNPs . In each simulated dataset , we randomly selected a region of 500 SNPs and simulated the values of 30 gene-expression traits for each individual , based on the CGGM in Eq . ( 5 ) . As almost all statistical methods for eQTL mapping assumes the standard linear regression model in Eq . ( 2 ) , we performed experiments on gene-expression traits simulated from this model as well . In order to set the model parameters , we first set the sparsity pattern and then assigned values to the non-zero elements of the parameters as follows . While in our simulation studies we primarily focused on the relatively small datasets of 500 SNPs and 30 gene-expression traits as described above , in order to demonstrate the performance and scalability of our method , we also applied our method to larger-scale simulated datasets of 1 , 000 SNPs and 500 gene-expression traits . Because MRCE , the main competing method to our approach , required substantially more computation time than our approach even on the smaller datasets , we were unable to compare the performance of MRCE on these larger simulated datasets . Instead , we compared our method with other computationally efficient methods , including GFlasso and a base-line approach of applying graphical lasso [16] and lasso [25] sequentially to learn gene networks and eQTLs . We use precision-recall curves and prediction errors as quantitative measures of the performance of different statistical methods . Precision-recall curves summarize how accurately each method recovers the true eQTLs and gene network structure by plotting precisions and recalls on - and -axes . In order to compute precisions and recalls , for each simulated dataset , we ranked all the elements of the estimated parameter matrices in a descending order according to their absolute values and compared this ranked list with the set of non-zero elements in the true parameters . On the other hand , prediction errors evaluate the performance of different methods on how accurately each method can predict gene-expression levels , given SNPs and estimated model parameters . Once the parameters are estimated using training data , prediction errors are obtained as , where is the prediction of gene expressions given by the model for the th individual in an independent test dataset and is the number of samples in the test set . Given the 343 samples in the full dataset , we used 300 samples as a training dataset and the remaining 43 samples as a test dataset . In order to determine the optimal regularization parameters in sparse CGGM , MRCE , and GFlasso during the training phase , we created a grid for different choices of regularization parameters , performed a five-fold cross-validation for each point on the grid , and selected the values that give the smallest cross-validation error as the optimal regularization parameters . In order to illustrate the behavior of a sparse CGGM , we present the results from applying our method to a single dataset simulated from a CGGM parameterization and compare them with what we obtained from GFlasso and MRCE ( Figure 2 ) . The non-zero elements of the true model parameters were drawn from for and for . Since GFlasso requires the gene network to be known , we used the correlation coefficient matrix of gene-expression trait data thresholded at as a gene network in GFlasso estimation . The true parameters for and along with are shown in the left , middle , and right columns , respectively , in Figure 2A . The estimated parameters for sparse CGGMs , MRCE , and GFlasso are shown in Figures 2B–D , respectively . In the plots for and , the rows and columns correspond to gene-expression traits and SNPs , respectively , and the results are shown only for the first 150 SNPs . In each panel , the white pixels correspond to the zero elements of the parameters and the darker pixels to non-zero elements . We note that while sparse CGGM provides the estimates of both and , MRCE and GFlasso provide a single estimate of eQTL effect sizes in and do not distinguish between direct and indirect effects of eQTLs on gene-expression traits . As shown in Figure 2B , our method successfully recovers the three gene modules in the true gene network as the block-diagonal structure in along with the sparse direct perturbation of this network by eQTLs in . When we perform inference in the estimated sparse CGGM by computing to learn indirect perturbation of the network by eQTLs , the direct perturbations of eQTLs in propagate primarily within each gene module in , leading to vertical stripes in . Although MRCE learns a gene network from data , unlike sparse CGGM , it does not have any mechanism to leverage this gene network to learn pleiotropic or indirect effects of eQTLs on gene modules and the estimated in Figure 2C shows isolated eQTLs for individual gene-expression traits rather than vertical stripes . As can be seen in Figure 2D , the GFlasso estimate of shows vertical stripes for eQTLs common within each gene module . However , it is immediately clear that GFlasso results have significantly more false positives for eQTLs than sparse CGGM and MRCE . This demonstrates that genetical genomics approach has the potential to improve the accuracy for detecting eQTLs than the conventional approach that focuses solely on eQTL mapping . We performed a quantitative comparison of the performance of sparse CGGM , MRCE , and GFlasso , by obtaining precision-recall curves and prediction errors averaged over 50 simulated datasets . Since MRCE and GFlasso are based on the standard linear regression model and sparse CGGM is based on a graphical model , we evaluate the different methods on datasets simulated from both models . Since MRCE and GFlasso use the standard linear regression model , we also compared the performance of the different methods , using datasets simulated from the model in Eq . ( 2 ) with known parameters for and ( Figure 6 ) . We present the precision-recall curves for and averaged over 50 simulated datasets in Figures 6A and B , respectively , and show the prediction errors in Figure 6C . In our simulation , we set the true parameter values to random draws from for and for . As can be seen in Figure 6 , even if the datasets were simulated from the standard linear regression model as used in MRCE and GFlasso , our method still outperforms MRCE and GFlasso . The simulation results so far demonstrated that our method has greater power for identifying gene networks and eQTLs than other methods . However , in these experiments , we were constrained to use relatively small datasets of only 30 gene-expression traits with 500 SNPs , because MRCE could not handle larger datasets effectively in a systematic simulation study . In this section , we demonstrate the effectiveness and scalability of our method , using substantially larger simulated datasets of 500 gene-expression traits and 1 , 000 SNPs . Given a region of 1 , 000 SNPs from chromosome 21 of the African individuals in HapMap phase III SNP data as described above , we simulated the values for 500 gene-expression traits , assuming sparse CGGMs with the true parameters determined as follows . We assumed that the true gene networks are scale-free networks , and set the network using the following strategy . First , we determined the number of neighbors of each node by making a random draw from a power-law distribution . Then , we applied the algorithm for generating a scale-free network [36] that repeatedly connects two nodes until we achieve the desired node degrees initially determined according to the power-law distribution . Given this network structure , we set the edge weights to random draws from a uniform distribution , and set to the graph Laplacian of the edge-weight matrix with small positive values added to the diagonal elements . We set the true eQTLs in by choosing each SNP as an eQTL for each gene-expression trait with probability and selecting one additional SNP as an eQTL for hub nodes with more than 20 neighbors in the network . For each eQTL in , we set the eQTL effect sizes to random draws from a uniform distribution with the signs of the values determined randomly . In Figure 7 , we compare the performance of the different methods averaged over 30 datasets simulated according to the above strategy . The results are shown for the precision-recall curves for the accuracy of detecting gene-network structures ( Figure 7A ) and eQTLs ( Figure 7B ) as well as prediction errors ( Figure 7C ) . As MRCE could not run on a single dataset of the given size within a few days , instead of using MRCE in our experiment , we compared our method with GFlasso and also with a two-stage method of applying graphical lasso and lasso to learn gene networks and eQTLs separately . We observe from Figure 7 that sparse CGGMs outperform all the other methods on these large-scale datasets . In order to examine the scalability of sparse CGGM , MRCE , and GFlasso , we compared the computation time for a single run of the different methods on varying sizes of datasets in Figure 8 . Figure 8A shows the computation time for varying the number of gene-expression traits with the number of SNPs fixed at , whereas Figure 8B shows the results from varying the number of SNPs with the number of gene-expression traits fixed at . Even though we used the approximate method for MRCE to reduce the computational cost of the exact method , our sparse CGGM optimization is more efficient by orders of magnitude than MRCE for problems of all sizes . Although GFlasso is more efficient than both sparse CGGM and MRCE , it is significantly more limited in that it focuses only on the problem of eQTL mapping . We applied our method to an eQTL dataset collected for two yeast parent strains , BY4716 ( BY ) and RM11-1a ( RM ) , and their 112 segregants [21] . We obtained SNP genotypes for 1 , 260 loci after removing the redundant SNPs with the same genotypes in neighboring regions of the genome and obtained expression measurements for 3 , 684 genes after removing the genes whose expression measurements were missing for more than 5% of the 114 samples . In order to select the optimal regularization parameters and , we performed a cross-validation with three random splits of data into 100 samples for estimating model parameters and 14 samples for computing cross-validation errors . Then , a final estimate of parameters was obtained by training a model on the entire dataset using the optimal regularization parameters . Below , we examine the yeast gene network and eQTLs with direct and indirect perturbations estimated by our method . In addition , we provide an in-depth analysis of a subnetwork with a strong evidence of being involved in DNA replication stress response based on the literature . We also provide a quantitative comparison of sparse CGGMs and other methods in terms of prediction errors . Since many previous works showed that gene networks tend to have a scale-free topology with few hub genes having many neighbors , we examined the gene network parameters in the estimated sparse CGGM for a scale-free property [37] . Given the network edge weights in , we defined the degree of each node as the sum of the absolute values of all incoming edge weights for the node . Then , the best ordinary least square fit of linear model for the empirical cumulative degree distribution was , showing that the estimated network has a strongly scale-free topology . Overall , in our estimate of network , 14 genes were connected to more than 100 other genes , 156 genes had more than 10 neighbors , and 1 , 593 genes had at least one neighbor . We hypothesized that each hub gene and its immediate neighbors form a hub-gene module and are involved in a common biological process . In order to test this hypothesis , we performed a gene ontology ( GO ) enrichment analysis for the 25 largest hub-gene modules , using Fisher's exact tests ( Table 1 ) . We found that each hub-gene module was significantly enriched with genes in a common GO category , showing that the genes in each hub-gene module are likely to participate in a common pathway . Next , we examined the eQTLs identified by our method as perturbing the above gene network . The eQTLs with direct perturbations of the gene network as captured in tended to concentrate on a small number of genetic loci , forming eQTL hotspots . Although the sparsity pattern of showed that 1 , 248 out of 1 , 260 SNP loci regulate directly at least one gene-expression trait , the top 10 SNPs that affect the largest number of gene-expression traits accounted for 15 . 8% of all SNP/gene-expression-trait pairs with direct influence of SNPs on gene-expression traits , and the top 20 SNPs accounted for 28 . 5% . We defined these top 20 SNPs as eQTL hotspots , and the genes directly regulated by each eQTL hotspot as a hotspot-regulated gene module ( Table 2 ) . In order to avoid redundancy , in the case of multiple eQTL hotspots within a 20 kb region with largely overlapping hotspot-regulated gene modules , we examined only one of those hotspots with the largest hotspot-regulated gene module . Out of the 13 eQTL hotspots that have been previously reported in analysis of the same dataset in [38] , 9 hotspots overlapped with the results from our method . In order to investigate whether the genes in each hotspot-regulated gene module are involved in a common biological function , we performed GO enrichment analysis ( Table 2 ) . We performed Fisher's exact tests , using GO slim categories downloaded from http://www . geneontology . org/GO . slims . shtml , after removing the GO categories with more than 500 genes . Our results show that all of the hotspot-regulated gene modules are significantly enriched for some GO categories , providing evidence that each eQTL hotspot regulates a functionally coherent set of genes . Since indirect SNP perturbations result from direct SNP perturbations propagating through the network , eQTLs with direct perturbations are likely to have stronger effect sizes than indirect perturbations . In Figure 9A , we compared the overall distribution of effect sizes of direct and indirect SNP perturbations as captured in and of the sparse CGGM by plotting histograms of the absolute values of non-zero elements in and . For indirect SNP perturbations , only those SNP/gene-expression-trait pairs that were estimated to be zero in but non-zero valued in were included . As can be seen in Figure 9A , the direct perturbations are generally stronger than indirect perturbations , confirming our hypothesis . Then , we examined whether the direct SNP perturbations estimated by our method are more likely to be cis eQTLs than the indirect perturbations . We declared the direct and indirect SNP perturbations in and as cis eQTLs , if for a given pair of SNP/gene-expression-trait , the gene sequence overlaps with the linkage region represented by the given SNP . The histogram in Figure 9B shows the distribution of the effect sizes of the estimated direct and indirect SNP perturbations for cis eQTLs . As can be seen in Figure 9B , direct SNP perturbations are significantly more frequent in cis eQTLs than indirect SNP perturbations , and explain nearly all of the cis eQTLs with strong effect sizes . When we examined the cis eQTLs with direct perturbations in our estimated model , we found that our approach was able to identify some of the well-known direct genetic perturbations in the literature . The genotypes for LEU2 , URA3 , HO , and LYS2 are known to differ in the parent strains , BY and RM , where these genetic differences have a large impact on the expressions of the corresponding genes as well as other genes [21] . While LYS2 was not included in our analysis , the LEU2 , URA3 , and HO expressions were found to have cis eQTLs with direct perturbations in our analysis , and at the same time , in our estimate of gene network , LEU2 and URA3 appeared as hub genes with more than 50 neighbors . In particular , the cis eQTLs with direct perturbations of LEU2 and URA3 were found to have the strongest effects among all cis eQTLs shown in Figure 9B , whereas HO had a cis eQTL with moderately strong direct perturbation . These results provide evidence that our method can recover the true direct SNP perturbations of a gene network , by decoupling direct SNP perturbations of gene expressions from their secondary/indirect effects on other gene expressions . We performed an in-depth analysis of the subnetwork around the TFS1 gene and its perturbation by eQTLs . This subnetwork is shown in Figure 10A , where the edge thicknesses correspond to the absolute values of edge weights in , representing the strength of dependency between two gene-expression traits . To avoid clutter , we only show the edges with the absolute values of edge weights . This subnetwork in Figure 10A contains many genes involved in DNA replication stress response and other types of stimulus response . In particular , TFS1 has been identified as a high-copy suppressor of guanine nucleotide-exchange factor CDC25 , which activates the Ras/cyclic AMP pathway regulating growth and metabolism in response to nutrients [39] . Many of the genes in this subnetwork , including TFS1 , its immediate neighbors ( PGM2 , SOL4 , RTN2 , GDB1 , RME1 , PRB1 , SDS24 , IGD1 ) , and 22 other genes , have been previously observed with changed abundance or localization under DNA damage [40] . In addition , the DDR2 gene in the subnetwork that codes for the DNA damage responsive protein has been found to have multi-stress response function [41] . Also , several hub genes in the subnetwork have been annotated as responding to other stress conditions such as heat shock ( HSP26 ) and oxidative stress ( CTT1 and GAD1 ) [42]–[44] .
In this paper , we presented a new statistical framework for genetical genomics analysis to learn a gene network by treating SNPs as naturally occurring perturbants of a gene network . Within this framework , we introduced a statistical model , called a sparse CGGM , for modeling a gene network under SNP perturbations and discussed an efficient learning algorithm and inference methods . While genetical genomics approach has been recognized as a more effective and less costly method for learning a gene network than experimental methods , this approach has not been widely used mainly because of the computational challenge that the effects of perturbations by often millions genetic variants at a time need to be decoded from data . Our approach directly addresses this challenge and identifies a gene network by decoupling the effects of multifactorial perturbations in eQTL data . At the same time , our approach addresses many of the weaknesses of the experimental methods and is able to identify which genes are directly perturbed by each SNP or are indirectly perturbed as downstream effects in the pathway . As eQTL data collection is being routinely performed for model organisms , and is more amenable for human tissues than experimental perturbations , our approach opens up doors to the possibility of leveraging these datasets for gene network learning rather than focusing on finding eQTLs from such data . Although the primary goal of our work and more generally genetical genomics analysis is to identify a gene network , our statistical approach has additional advantages of enhancing the current statistical tools of eQTL mapping and extracting significantly more detailed information on the functional role of eQTLs in the context of gene network . Our approach provides a flexible statistical framework for learning a gene network along with eQTLs that can be easily extended in several different ways . For example , although in this paper , our gene network was defined over the expression levels of mRNAs , it is straightforward to include microRNA expression data to construct a network over both mRNA and microRNA expressions , both of which can be perturbed by genetic variants . Another possible extension is to model epistatic interactions among SNPs within sparse CGGM by introducing additional features for SNP interactions in the probabilistic graphical model . However , there are certain limitations to our approach . Although our approach can handle thousands of gene-expression traits and SNPs efficiently , it is still not efficient enough to be directly applied to genome-wide analysis of eQTL datasets of higher-level organisms with tens of thousands of gene-expression traits and millions of SNPs . For such large-scale datasets , we suggest to split the full dataset into smaller sets of gene-expression traits by applying a clustering algorithm to obtain coarse-grained gene modules . Then , our approach can be applied to each subset of gene-expression traits for coarse-grained gene modules to extract fine-grained information on gene-network connectivities . In order to perform a full joint analysis of all data , in future work , we will consider improving the computational bottleneck of matrix inversion for the gene network parameters in the learning algorithm by replacing it with an approximate but computationally less expensive inversion . Another future direction is to relax the assumption in our model that the gene-expression traits under SNP perturbations follow a Gaussian distribution . Although Gaussian graphical models have been widely used to infer a gene network from gene-expression data due to many of the properties of Gaussian distributions that lead to easy computations , this assumption can be potentially restrictive for modeling realistic biological processes . The software for sparse CGGMs is available at http://www . cs . cmu . edu/~sssykim/softwares/softwares . html#scggm . | A complete understanding of how gene regulatory networks are wired in a biological system is important in many areas of biology and medicine . The most popular method for investigating a gene network has been based on experimental perturbation studies , where the expression of a gene is experimentally manipulated to observe how this perturbation affects the expressions of other genes . Such experimental methods are costly , laborious , and do not scale to a perturbation of more than two genes at a time . As an alternative , genetical genomics approach uses genetic variants as naturally-occurring perturbations of gene regulatory system and learns gene networks by decoding the perturbation effects by genetic variants , given population gene-expression and genotype data . However , since there exist millions of genetic variants in genomes that simultaneously perturb a gene network , it is not obvious how to decode the effects of such multifactorial perturbations from data . Our statistical approach overcomes this computational challenge and recovers gene networks under SNP perturbations using probabilistic graphical models . As population gene-expression and genotype datasets are routinely collected to study genetic architectures of complex diseases and phenotypes , our approach can directly leverage these existing datasets to provide a more effective way of identifying gene networks . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"systems",
"biology",
"regulatory",
"networks",
"population",
"genetics",
"biology",
"computational",
"biology"
] | 2014 | Learning Gene Networks under SNP Perturbations Using eQTL Datasets |
How cells communicate to initiate a regenerative response after damage has captivated scientists during the last few decades . It is known that one of the main signals emanating from injured cells is the Reactive Oxygen Species ( ROS ) , which propagate to the surrounding tissue to trigger the replacement of the missing cells . However , the link between ROS production and the activation of regenerative signaling pathways is not yet fully understood . We describe here the non-autonomous ROS sensing mechanism by which living cells launch their regenerative program . To this aim , we used Drosophila imaginal discs as a model system due to its well-characterized regenerative ability after injury or cell death . We genetically-induced cell death and found that the Apoptosis signal-regulating kinase 1 ( Ask1 ) is essential for regenerative growth . Ask1 senses ROS both in dying and living cells , but its activation is selectively attenuated in living cells by Akt1 , the core kinase component of the insulin/insulin-like growth factor pathway . Akt1 phosphorylates Ask1 in a secondary site outside the kinase domain , which attenuates its activity . This modulation of Ask1 activity results in moderate levels of JNK signaling in the living tissue , as well as in activation of p38 signaling , both pathways required to turn on the regenerative response . Our findings demonstrate a non-autonomous activation of a ROS sensing mechanism by Ask1 and Akt1 to replace the missing tissue after damage . Collectively , these results provide the basis for understanding the molecular mechanism of communication between dying and living cells that triggers regeneration .
Organisms are continuously exposed to a wide variety of environmental stressors that can cause deterioration and cell death . Tissues overcome the effect of those stressors by replacing damaged cells to restore homeostasis . Therefore , understanding the early signals that initiate the response to damage is an essential issue in regenerative medicine . Regeneration can be monitored in Drosophila imaginal discs , which are well-characterized epithelial sacs capable to regenerate after genetically-induced apoptosis or when parts are physically removed ( reviewed in [1] ) . Compiling evidence supports that reactive oxygen species ( ROS ) fuel wound healing and oxygen-dependent redox-sensitive signaling processes involved in damage response [2–4] . Actually , genetically-induced apoptosis in the imaginal discs , using the Gal4/UAS system , leads to the production of ROS which propagate to the surrounding neighbors [5–7] . Although oxidative stress has been associated with several pathologies , it has been described that low levels of ROS can be beneficial for signal transduction [8] . The Jun-N Terminal kinase ( JNK ) and p38 signaling pathways are MAP kinases that respond to many stressors , including ROS , and foster regeneration and cytokine production in Drosophila [5 , 6 , 16–25 , 7 , 26 , 27 , 9–15] . Both pathways control numerous cellular processes as diverse as cell proliferation and cell death . For example , ectopic activation of JNK induces apoptosis [28 , 29] , but its inhibition results in lethality [30] . It is though that these disparities could be due to either different levels of activity or different mechanisms of activation . After genetic ablation in a specific zone of the wing imaginal disc , dying cells produce high levels of ROS , and show high JNK activity [5 , 7] . High levels of JNK have been associated to dying cells and are involved in a feedback amplification loop to enhance the apoptotic response to stress [31] . However , the living cells nearby the dying zone show low levels of ROS , which are beneficial for the cell as they turn on the activation of p38 and low levels of JNK [5] . Both MAP kinases in the living cells are required for a cytokine-dependent regenerative growth . A key question is how the balance between the beneficial or detrimental effects of ROS is controlled and , in particular , how ROS control JNK and p38 activity . A candidate molecule to perform this function is the MAPKKK Apoptosis signal-regulating kinase 1 ( Ask1 ) , which responds to various stresses by phosphorylation of the kinases upstream JNK and p38 [32–34] . Hence , in a reduced environment , thioredoxin ( Trx ) inhibits Ask1 kinase activity by directly binding to the N-terminal region of Ask1 . Upon oxidative stress , the redox-sensitive cysteines of Trx become oxidized , resulting in the dissociation of Trx from Ask1 . Consequently , Ask1 is oligomerized and its threonine-rich kinase domain is phosphorylated , inducing Ask1 activation [35 , 36] . Mammalian Ask1 is highly sensitive to oxidative stress and contributes substantially to JNK-dependent apoptosis . Nevertheless , recent studies have also revealed other functions of this kinase , including cell differentiation and survival [37] . Ask1-interacting proteins promote conformational changes that lead to the modulation of Ask1 activity and result in various cellular responses . For example , Ask1 is a substrate for phosphorylation by Akt1 , a serine/threonine kinase activated by lipid kinase phosphatidylinositide 3-kinase ( Pi3K ) pathway in response to insulin receptor activation . This phosphorylation is associated to a decrease of Ask1 activity in vitro [38] . The Pi3K/Akt pathway , which is conserved between mammals and Drosophila , is one of the main effector signals for the regulation of tissue growth [39–41] . In Drosophila , loss-of–function mutants of various components of the pathway result in reduced body size or lethality [42] . Conversely , mutants of the phosphatase PTEN , an antagonist of Pi3K , result in high activity of Akt and tissue overgrowth [43–45] . Thus , it is conceivable that Pi3K/Akt signaling is involved in regenerative growth , but whether Pi3K/Akt and Ask1 interact for controlling regeneration is unknown and deserves attention . In this work , we genetically induced cell death in wing imaginal discs to explore the link between ROS production and regeneration . We found that Ask1 acts as a sensor of ROS upstream the JNK and p38 pathways . Moreover , we describe that Akt1 is necessary for modulating Ask1 activity in living cells to trigger regeneration . In addition , our results indicate that oxidative stress generated in the damaged cells signals the neighboring living cells to promote tissue repair .
The Gal4/UAS/Gal80TS transactivation system has been extensively used to temporarily and spatially activate pro-apoptotic genes such as reaper ( rpr ) [10 , 11 , 46] . To study the capacity to regenerate , we induced cell death in the wing disc using the wing-specific salE/Pv-Gal4 strain to activate UAS-rpr , in a temporally controlled manner thanks to the tub-Gal80TS thermo sensitive Gal4-repressor ( henceforth salE/Pv>rpr ) . We first kept the embryos at 17°C until the 8th day ( 192 hours after egg laying ) to activate the tub-Gal80TS and therefore to prevent rpr expression . Subsequently , we moved those larvae to 29°C for 11 hours , to inhibit tub-Gal80TS , and activate apoptosis in the salE/Pv-Gal4 zone . Then , larvae were returned to 17°C to avoid further apoptosis and allow tissue to regenerate . After adults emerged , wings were dissected and regeneration was analyzed . With these experimental conditions , 100% of the wings of salE/Pv>rpr flies contained the full set of veins and interveins , which demonstrates that the missing parts were regenerated ( Fig 1A ) . The same experiment was carried out in Ask1 mutant backgrounds . The Ask1MB06487 mutant flies are viable in homozygosis but lethal over the deficiency Df ( 3R ) BSC636 , which suggests that is a hypomorphic allele . The Ask1MI02915 mutant is lethal in homozygosis as well as over the Df ( 3R ) BSC636 , which suggests that is an amorphic allele . We found that after genetically-induced cell death in Ask1MB06487 or Ask1MI02915 heterozygous background , full regeneration is only achieved in 15% and 23% of the wings , respectively ( Fig 1A ) . The phenotypes of the abnormally regenerating wings vary and consisted in lack of some veins or parts of a vein or anomalous vein segments ( S1A Fig ) . To further address the role of Ask1 in regeneration , we used a double transactivation system to simultaneously induce cell death in the salE/Pv domain and inhibit Ask1 with UAS-RNAi ( UAS-Ask1RNAi ) in an adjacent compartment ( Fig 1B ) . RNAi knockdown of Ask1 ( UAS-Ask1RNAi ) reduces Ask1 mRNA total levels to approximately 25–30 percent of that observed in controls ( S1B Fig ) . The UAS-Ask1RNAi transgene was activated in the anterior compartment using the Gal4/UAS system ( ci-Gal4>UAS-Ask1RNAi ) and cell death was induced in the salE/Pv domain using the Gal80-repressible transactivator system LHG ( LexA-Hinge-Gal4 activation domain ) , a modified form of the lexA lexO system ( salE/Pv-LHG>lexO-rpr ) ( Fig 1B ) [5 , 47] . Both transgenes were activated at the same time , following the same protocol as in the previous experiment . The resulting adult wings ( ci>Ask1RNAi salE/Pv>rpr ) lacked some veins or interveins and their size was reduced in all cases observed . However , the inhibition of Ask1 in the anterior compartment for the same time but without cell death , did not affect vein pattern nor wing size ( ci>Ask1RNAi ) ( Fig 1B ) . Similar results were obtained after driving UAS-Ask1RNAi to a different compartment ( dorsal ) and inducing cell death in salE/Pv ( ap>Ask1RNAi salE/Pv>rpr ) ; in this case , we found that 82% of wings could not regenerate properly . As for the anterior compartment , Ask1RNAi in the dorsal compartment without cell death did not affect vein pattern nor wing size ( ap>Ask1RNAi ) ( Fig 1B ) . To discard any toxicity due to the insertion of the transgene , all experiments were carried out in parallel but constantly at 17°C to maintain tub-Gal80TS activity and block transgene ( UAS- or lexO- ) expression . In these conditions no defects in the wings were detected ( salE/Pv> OFF in Fig 1A and 1B ) . To evaluate whether these anomalies were due to impairment of proliferation , we analyzed the mitotic index , calculated as the number of cells positive for the phosphorylated form of Histone 3 ( P-H3 ) in the anterior compartment ( ci> ) , where the Ask1RNAi transgene was expressed . It is well known that apoptosis in discs induces compensatory proliferation of nearby cells ( reviewed in [1] ) . Accordingly , genetic ablation ( salE/Pv>rpr ci>GFP ) resulted in an increase of mitosis compared to the neutral ( salE/Pv>GFP ci>RFP ) or to the Ask1 knockdown ( salE/Pv>GFP ci>Ask1RNAi ) conditions . However , combining genetic ablation in the salE/Pv zone with Ask1 knockdown in the nearby anterior compartment , the number of mitosis associated to damage did not increase ( salE/Pv>rpr ci>Ask1RNAi ) ( Fig 1C ) . This observation confirms that Ask1 reduction impairs regenerative growth . The Drosophila Ask1 locus encodes two protein isoforms of different lengths , Ask1-RB and Ask1-RC , with predicted molecular weights of 136 . 5 kDa and 155 . 5 kDa , respectively . Both have a protein kinase-like domain that contains a highly conserved core of threonines conferring functionality to the protein ( reviewed in [32] ) . It has been reported that after Trx release , the Ask1 oligomer undergoes conformational change leading to trans-autophosphorylation of the threonine residue corresponding to Thr838 , Thr845 and Thr747 in human , mouse and Drosophila Ask1 , respectively [32] . In human cells , phosphorylation of the N-terminus Ser83 by Akt results in attenuation of Ask1 activity in vitro [38] . This is noteworthy , because attenuation of Ask1 could result in moderate levels of activity necessary for turning on JNK and p38 in living cells . Drosophila Ask1 lacks this residue and the Akt consensus motif in the N-terminus ( S2 Fig , S1 Appendix , S2 Appendix ) . However , we found that the longer Ask1-RC isoform contained a conserved domain of unknown function DUF4071 with a highly conserved Ser ( human Ser174 , Drosophila Ser83 ) , which we hypothesized that could be key in Ask1 regulation ( Fig 2A , S2 Fig ) . Active Ask1 can be traced with antibodies against the phosphorylated threonine residues of the highly conserved kinase domain ( henceforth P-Thr ) [48] and the attenuated form with specific antibodies against phosphorylated Ser83 ( P-Ser83 ) . We first tested both antibodies in wild-type wing imaginal discs . We detected low levels of P-Thr all over the disc including some transient high activity during mitosis ( Fig 2B and S3A Fig ) , as occurs in dividing mammalian cells [49] . In Ask1 mutant backgrounds , both the general and the mitosis-associated P-Thr levels were reduced but not abolished ( S3B Fig ) , and Ask1RNAi discs did not show any effect on P-Thr levels in mitosis ( Fig 2E ) . This discrepancy may be due to the hypomorphic condition of the mutant combinations , but also to that the antibody may recognize some unspecific epitope in mitotic cells . We also found basal levels of P-Ser83 in wild-type discs ( Fig 2C ) . Ectopic expression of Ask1RNAi suppressed the low endogenous P-Ser83 Ask1 levels ( S3C Fig ) . Remarkably , upon apoptosis , high levels of P-Thr were localized in dying cells ( Fig 2D ) and absent or , in some cases , increased weakly above the basal levels in nearby living cells . This increase of P-Thr in apoptotic cells was inhibited after Ask1RNAi expression ( Fig 2E ) . In contrast , P-Ser83 accumulated in living cells adjoining the apoptotic zone and was completely absent in apoptotic cells . The increase in P-Ser83 varied from discs with strong accumulation near the dying domain to those with an extended increase in the whole wing pouch ( 2D and S4A Figs ) . Similar results were obtained after killing cells with a different pro-apoptotic gene ( salE/Pv>hid ) ( S4B Fig ) . In the presence of apoptosis and blocking Ask1 , the P-Ser83 increase in living cells was inhibited ( salE/Pv>rpr ci>Ask1RNAi , Fig 2F ) . Moreover , P-Ser83 was also found to be elevated at the wound edges after physical injury ( S4C Fig ) . Together , these observations indicate that living cells respond to damage by phosphorylation of Ask1 Ser83 . In contrast to the high activity of Ask1 in dying cells ( high P-Thr ) , the presence of P-Ser83 in living cells could be indicative of tolerable levels of Ask1 achieved by attenuation of P-Thr activity . To test this hypothesis , we mutated Ask1 at serine 83 to alanine and cloned it into a UAS vector ( UAS-Ask1S83A ) . In parallel , we also cloned a wild-type form of Ask1 ( UAS-Ask1WT ) . We ectopically expressed Ask1WT in the posterior compartment of the wing disc ( hh-Gal4 UAS-Ask1WT ) , which resulted in an increase of P-Ser83 , in addition to low levels of P-Thr ( Fig 2G ) . Interestingly , the ectopic expression of Ask1S83A in the posterior compartment ( hh-Gal4 UAS-Ask1S83A ) , did not show a rise in P-Ser83 , but strong activation of P-Thr ( Fig 2H ) . The high levels of P-Thr were not homogeneously distributed , varied from disc to disc , and were always found within the posterior compartment ( hh> ) , where the transgene was activated . This observation concurs with the P-Ser83 residue as responsible for the attenuation of P-Thr . We next analyzed whether the Ser83 residue was key for the damage response . Using the double transactivation system , we found that the expression of UAS-Ask1WT or UAS-Ask1S83A transgenes in the anterior compartment without cell death did not cause any defects in wing morphology , size or vein and intervein patterning ( salE/Pv>GFP ci>Ask1WT or Ask1S83A in Fig 2I ) . After genetic ablation , expression of UAS-Ask1WT resulted in full regeneration in 89% of wings ( salE/Pv>rpr ci>Ask1WT ) . In contrast , expression of UAS-Ask1S83A led to a fall to only 29% of individuals being capable of regenerate and the rest showed strong effects on wing morphology , with altered pattern of veins and interveins , as well as the appearance of notches , which together are indicative of disrupted regeneration ( salE/Pv>rpr ci >Ask1S83A in Fig 2I ) . The suppression of the ability to regenerate by UAS-Ask1S83A , which may act as a dominant negative allele , is likely due to the lack of non-autonomous P-Ser83 increase in the living cells near the damaged zone . Next , we decided to identify the upstream signal responsible for Ser83 phosphorylation . In mammalian cells , attenuating phosphorylation of Ask1 is driven by the serine-threonine Akt kinase , the core kinase of the insulin pathway [38] . We wondered whether the Drosophila Akt1 as well as its upstream Pi3K92E kinase ( the Drosophila Pi3K kinase also known as dp110 ) , were required for Ser83 phosphorylation . We first found that the active phosphorylated form of Akt ( P-Akt ) was increased in the living tissue and decreased or was absent in the dying zone ( Fig 3A–3D ) . To test whether the Akt1 phosphorylated the endogenous Ask1 Ser83 in vivo , we specifically inhibited this kinase with a UAS-AktRNAi in a stripe of cells at the center of the disc , the ptc domain ( ptc>AktRNAi ) , and found that P-Ser83 was reduced ( Fig 3E and 3F ) . In addition , ectopic activation of Akt1 ( UAS-myr-Akt1 . S ) , a constitutively activated membrane-anchored form of Akt1 [50] , resulted in an increase in the levels of P-Ser83 ( Fig 3G and 3H ) . We next studied the role of P-Akt on Ask1 in the context of regeneration . In the presence of cell death and blocking Pi3K92E , with the dominant negative form dp110DN in the adjacent compartment , we observed a reduction in the accumulation of P-Akt and P-Ser83 induced after genetic ablation ( salE/Pv>rpr hh> dp110DN in Fig 3I–3K ) . Moreover , in the overlapping zone of dying cells and dp110DN , the accumulation of P-Thr was unaffected ( Fig 3L and S5 Fig ) . We also scored the effects on wing regeneration after blocking Akt and Pi3K92E kinases . Regeneration was impaired when the dominant negative form of Pi3K92E was expressed in the anterior compartment ( salE/Pv>rpr ci>dp110DN ) as well as in the Akt11 heterozygous background ( salE/Pv>rpr Akt1/+ ) , an allele that encodes a catalytically inactive protein [51] ( Fig 3M ) . Moreover , regeneration was severely affected in double heterozygous flies containing Akt11 and Ask1MB06487 or Ask1MI02915 alleles ( Fig 3M ) . Neither the dominant negative form of Pi3K92E nor the allelic combination Akt11 and Ask1MB06487 or Ask1MI02915 affected wing development in the absence of cell death ( salE/Pv>OFF wings in Fig 3M ) . These results further support the notion that Pi3K92E/Akt1 and Ask1 genetically interact and that their epistatic interaction is key to drive regeneration . Next , we modulated the ROS levels in order to determine if Ask1 can sense oxidative stress after damage . We fed salE/Pv>rpr larvae with food supplemented with N-acetyl cysteine ( NAC ) , a potent non-enzymatic scavenger that decreases ROS production , and examined Ask1 phosphorylation ( Fig 4A ) . After induction of cell death , we found a significant decrease in both P-Thr and P-Ser83 in discs from NAC-fed larvae ( Fig 4B and 4C; S3 Appendix ) . In addition , we fed larvae with H2O2-supplemented food and observed a significant increase in both P-Thr and P-Ser83 levels ( Fig 4D–4F ) . This oxidative stress-induced increase was blocked in the hypomorphic Ask1MB06487 homozygous mutant ( Fig 4E and 4F; S3 Appendix ) . It is known that in mammalian cells Ask1 can also be activated by endoplasmic reticulum ( ER ) stress [52] . Feeding larvae with tunicamycin , an inhibitor of N-glycosylation in the ER that induces ER stress , led to an increase in Ask1 activation , which was also blocked in Ask1MB06487 mutant discs ( Fig 4E and 4F; S3 Appendix ) . Furthermore , we examined whether Akt1 activation in the living cells is targeted non-autonomously by ROS produced by the dying cells . To examine this hypothesis , we enzymatically blocked ROS production using ectopic expression of the ROS scavengers Superoxide dismutase 1 and Catalase ( Sod1:Cat ) in the rpr ablated region ( salE/Pv>rpr , Sod1:Cat ) and monitored the pixel intensities of two adjacent zones in each disc , the anterior ( A ) and posterior ( P ) to the salE/Pv ( Fig 4G ) . The activation of the Sod1:Cat transgene concomitantly with salE/Pv>rpr did not inhibit apoptosis and P-Thr was detected in the dying cells ( S6 Fig ) . However , in these conditions , we found that P-Akt leveis in neighboring cells were reduced in both A and P zones when Sod1:Cat was co-expressed with rpr . Accordingly , we found that the increase of P-Ser83 in neighboring cells was blocked when Sod1:Cat was co-expressed with rpr ( Fig 4G; S3 Appendix ) . Together , these results demonstrate that the oxidative stress generated from the dying cells targets Akt1 and Ask1 in the neighboring undamaged tissue . We have shown that the Ask1 P-Ser83 attenuating signal is essential for regeneration , and that this signal depends on ROS and Akt . We next hypothesized that this P-Ser83-mediated attenuation does not silence Ask1 , but instead maintains the low levels of Ask1 activity necessary for regeneration . Therefore , we analyzed the expression of its known targets , the stress-activated MAP kinases JNK and p38 [48 , 53] , both required for ROS-dependent regeneration [5 , 6] . We induced cell death in the salE/Pv domain and simultaneously blocked Ask1 in the anterior compartment , and used the posterior as an internal control for the same disc ( ci>Ask1RNAi salE/Pv>rpr ) . We found most phosphorylated p38 ( P-p38 ) , as an indicator of p38 activity , in the posterior compartment ( Fig 5A and 5B ) . The effects of Ask1 on JNK activity were monitored by matrix metalloproteinase 1 ( Mmp1 ) expression as it is one of the bona fide transcriptionally regulated read-outs of the JNK pathway [54] . After salE/Pv>rpr cell death , Mmp1 was found in both compartments , but in ci>Ask1RNAi salE/Pv>rpr discs Mmp1 mostly accumulated in the posterior ( Fig 5C and 5D ) . The effects of Ask1 interference on JNK pathway were strengthened after staining with P-JNK antibody . P-JNK was found accumulated around the dead zone in salE/Pv>rpr discs . However , when Ask1 was knockdown in the anterior compartment ( ci>Ask1RNAi salE/Pv>rpr discs ) , P-JNK was reduced in this compartment and accumulated in the posterior ( S7A and S7B Fig ) . To confirm that Ask1 is responsible for JNK and p38 activation , we tested whether Ask1 was required for their activation after physical damage . We made two incisions in hh>GFP , Ask1RNAi discs , one in the UAS-Ask1RNAi compartment and another into the control compartment . We found that P-p38 localized at the wound edges in the control compartment , whereas the levels of P-p38 were reduced at the wound edges of the Ask1RNAi compartment ( Fig 5E and 5G ) . Likewise , cut Ask1RNAi discs resulted in less Mmp1 than in the control compartment ( Fig 5F and 5H ) . In addition , we tested the puckered ( puc ) reporter of the JNK pathway , which normally accumulates in wounds , and found to drop in heterozygous Ask1 discs after a physical injury ( S7C and S7D Fig ) . Together , these results indicate that Ask1 is necessary for the activation of p38 and JNK during regeneration .
We have shown here that Ask1 acts as a sensor of ROS after damage , and that synergizes with Pi3K/Akt1 to phosphorylate the Ask1 Ser83 , a key phosphorylation event to initiate regeneration . In addition , we have demonstrated that the activation of the Pi3K/Akt1 and Ask1 in undamaged cells is originated non-autonomously by ROS produced by the damaged tissue . An essential question to understand regeneration is how damaged cells communicate to the nearby living tissue to initiate regenerative growth . One of the most classical and determining studies on epithelial regeneration was the discovery that massive cell death caused by ionizing radiation in Drosophila larvae resulted in proliferation of the unaffected neighbors to compensate the lost [55] . Since then , compensatory proliferation , a cellular response linked to regeneration , has been considered as a result of signals released from apoptotic cells ( reviewed in [56] ) . We propose here that the oxidative stress generated by damaged or apoptotic cells signals undamaged tissue to initiate the Ask1/Akt1 machinery that will culminate with repair and compensatory proliferation . ROS acting in signaling after wounding has been subject of extensive research in various organisms and tissues [5 , 6 , 64–67 , 26 , 57–63] . Here , we have uncovered two scenarios of Ask1 activity: high activity in apoptotic cells , with high levels of P-Thr; and low activity in undamaged cells , where P-Ser83 prevails over P-Thr . This fits with the high levels of ROS and JNK in dying cells and the low levels of ROS , low JNK and P-p38 in living cells reported previously [5] . Together , our observations address the question of how JNK signaling selectively fosters apoptosis or proliferation , and how this signal is related to high or low ROS levels . The same applies to p38 , which is only activated in unharmed cells when neighbors enter in apoptosis or are damaged [5 , 53] . Our study demonstrates that Ask1 operates in an Akt1-dependent manner in living cells and in an Akt1-independent manner in dying cells . It is conceivable that Akt1-mediated attenuation results in either low or transient levels of Ask1 activity necessary to stimulate P-p38 and low JNK levels for regeneration . In contrast , apoptotic cells induce high or sustained levels of JNK because high levels of ROS result in high Ask1 P-Thr in the absence of Akt1 attenuation . Based on the results presented here and previous observations , we propose a model in which oxidative stress and the absence or presence of Pi3K/Akt1 survival signals are integrated in Ask1 to control the selective activity of JNK and p38 , and in turn regenerative growth ( Fig 5I ) . In this model , only dying cells produce high ROS levels and high JNK activity [5 , 7] . In addition , other mechanisms can exacerbate apoptosis through the lethal levels of JNK , and therefore increase ROS production . For example JNK , as well as p53 , functions in a positive feedback after apoptosis as it transcriptionally activates the pro-apoptotic genes hid and rpr , thus amplifying apoptosis [31 , 68] . It is therefore likely that the high ROS that result in high Ask1 activation , inabilities the dying cells to respond to Pi3K/Akt1 , preventing the Ask1 attenuation and resulting in deleterious JNK activity . We also have demonstrated that P-Akt1 is absent , or low , in dying cells , which supports a lack of Pi3K/Akt1-dependent Ask1 attenuated activity in dying cells . These cells will enter into a committed-to-die status with no return to survival . In contrast , the living neighbor cells , albeit adjacent to dying cells , show tolerable levels of ROS [5 , 7] , and activate moderate levels of Ask1 through the Akt1-dependent phosphorylation of Ask1 P-Ser83 . In these cells , Ask1 will turn on the activation of tolerable levels of JNK and p38 , leading to a committed-to-regenerate status . Eventually , the JNK and p38 pathways will promote the transcription of the unpaired cytokines necessary for survival and regeneration [5 , 15 , 69–72] . In addition to cytokines , Wnt/Wg signaling pathway acts as a mitogenic signal involved in wing disc regeneration that responds to JNK [11 , 18 , 69 , 73] as well as the inhibition of the Hippo pathway , which results in increased Yki activity [74–76] . Several phosphatases are sensitive to ROS , and could be also integrated into the model presented here . This is the case of the JNK phosphatases , which are critical molecular targets of ROS . Oxidative inhibition of JNK phosphatases results in sustained JNK activation in order to promote caspase 3 cleavage and eventually cell death [77] . Therefore , the early burst of ROS in dying cells could activate lethal levels of JNK activity through inhibition of the Drosophila JNK phosphatase puckered concomitantly with the high levels of Ask1 , and together boost the committed-to-die status . Additionally , the PTEN phosphatase , which regulates many cellular processes through direct antagonism of Pi3K signaling , is also ROS sensitive . Increasing ROS can enhance insulin signaling attributable to the oxidative inhibition of PTEN [78] . PTEN inactivation enhances the activation of AKT signaling , which in turn promotes the expression of cell survival genes [78] . Thus , PTEN could be involved in the living cells to release the Akt1 activity that is necessary to reach the tolerable levels of Ask1/JNK/p38 . Therefore , the low oxidative stress that reaches the living cells can trigger not only Trx2 dissociation from Ask1 , but also the PTEN-oxidation dependent Pi3K/Akt1 activity required for attenuated Ask1 necessary for survival and proliferation . Our model concerns to an inherent response in the imaginal disc epithelium . However , other signals could enhance or stabilize the ROS-dependent regenerative response . For example , ROS produced after wounding are necessary for recruitment and activation of the Drosophila macrophage-like immune cells , which will secrete the TNF ortholog Eiger . Eiger activates JNK signaling in the epithelial cells to contribute to the regenerative response [6] . Because the moderate activity of Ask1 for regeneration is achieved in an Akt1-dependent manner , and because Akt1 functions downstream of the insulin signaling , it seems conceivable an implication of nutrition in imaginal disc regeneration . The production and secretion of insulin and AKH , the fly analog of mammalian glucagon , vary depending of the food content . Thus , a challenge for the future will be to analyze how these hormones can modulate the regenerative response to damage . For example , the Drosophila TNF Eiger is produced by the fat body when larvae are exposed to low protein diet [79] . As Eiger activates JNK non-autonomously in many tissues , it could have a central role in imaginal disc regeneration , depending on food availability . The relation between nutrition and metabolism , as well as their involvement in regeneration deserves further analysis in order to achieve an integrative view of regeneration .
The Drosophila melanogaster strains , salE/Pv-LHG and lexO-rpr , are previously described [5]; lexO-GFP ( gift from K . Basler ) . UAS-H2B-RFP ( from J . Knoblich ) , Akt11 ( gift from H . Stocker ) , sal-Gal4 , salE/Pv-Gal4 ( gift from J . F . de Celis ) , ci-Gal4 ( from R . Holmgren ) and en-Gal4 ( gift from G . Morata ) . The following strains were provided by the Bloomington Drosophila Stock Center: tubGal80TS ( RRID:BDSC_7017 ) , ptc-Gal4 ( RRID:BDSC_2017 ) , Ask1MB06487 ( RRID:BDSC_26048 ) Ask1MI02915 ( RRID:BDSC_36163 ) , UAS-GFP ( RRID:BDSC_4776 ) , UAS-dp110DN ( RRID:BDSC_25918 ) , Df ( 3R ) BSC636 ( RRID:BDSC_25726 ) , UAS-Ask1RNAi ( RRID:BDSC_35331 ) , UAS-Cat . A ( RRID:BDSC_24621 ) , UAS-Sod . A ( Sod1 ) ( RRID:BDSC_24754 ) , UAS-rpr on the X chromosome ( RRID:BDSC_5823 ) , UAS-rpr on the third chromosome ( RRID:BDSC_50791 ) . These strains are described in FlyBase: hh-Gal4 , ap-Gal4 , UAS-hid . The UAS-AktRNAi ( 2902 ) strain was obtained from the Vienna Drosophila Resource Center ( VDRC ) . We used the UAS-myr-Akt1 . S [50] . Canton S and w118 were used as controls . A full list of genotypes is provided at the end of this section . Cell death was genetically induced as previously described [5] . We used the salE/Pv-Gal4 as a driver , which consists of the spalt wing enhancer with expression confined to the wing to score adult wing parameters . The UAS lines used to promote cell death were UAS-rpr or UAS-hid , two pro-apoptotic genes , controlled by the thermo-sensitive Gal4 repressor tubGal80TS . We also used the salE/Pv-LHG and LexO-rpr strains [5] for genetic ablation , utilizing the same design as for Gal4/UAS . Embryos were kept at 17°C until the 8th day/192 hours after egg laying to prevent rpr expression . They were subsequently moved to 29°C for 11 hours and then back to 17°C to allow tissue to regenerate . Two types of controls were always treated in parallel; individuals without rpr expression ( UAS-GFP , moved to 29°C for 11 hours ) and individuals kept continuously at 17°C to avoid any transgene activation . In most of the experiments shown here , we induced cell death for 11h . However the induction of cell death for testing the effects of signals emerging from the dying zone was longer . This is the case of the P-Akt experiment in Fig 3B and 3C and for the genetic scavenging of ROS ( Sod1:Cat ) ( Fig 4G and S6 Fig ) , in which sal>rpr was activated for 24h . In dual transactivation experiments , we used salE/Pv-LHG LexO-rpr to ablate the salE/Pv domain ( abridged as salE/Pv>rpr ) , whereas Gal4 was used to express different transgenes under the control of Gal4 drivers ( ci-Gal4 for anterior compartment; ap-Gal4 for dorsal compartment and hh-Gal4 for posterior compartment ) . To test the capacity to regenerate in different genetic backgrounds , we used adult wings emerged from salE/Pv>rpr individuals in which patterning and size defects can be scored easily . Flies were fixed in glycerol:ethanol ( 1:2 ) for 24 hours . Wings were dissected in water and then washed with ethanol . Subsequently , they were mounted in lactic acid:ethanol ( 6:5 ) and analyzed and imaged under a microscope . The percentage of regenerated wings refers to fully regenerated ( for genetic ablation genotypes ) or normally developed wings ( for testing transgenes ) and was calculated according to the number of wings with a complete set of veins and interveins , as markers of normal patterning . For each sample , we scored the percentage of individuals belonging to the “regenerated wings” class . We calculated the standard error of the sample proportion based on a binomial distribution ( regenerated complete wing or not ) SE = √p ( 1-p ) /n , where p is the proportion of successes in the population . Serine/threonine putative phosphorylation sites were determined by scanning the orthologous human/fly sequences with the RxRxx[ST] domain motif for the Akt kinase ( purple block on S2 Fig ) and the degenerated pattern xxRxx[ST] ( green blocks on S2 Fig ) . Strikingly the human Ser83 site was not found on Drosophila melanogaster; which was confirmed by using GeneWise ( version wise2 . 4 . 1 ) [80] to map the M3K5_HUMAN protein sequence ( Q99683 ) , downloaded from UniProt [81] , over a window of 10kbp centered at the Ask1 transcription start site; this sequence segment was retrieved from FlyBase FB2017_06 genome version ( [82] , chr3R:19 , 875 , 000–19 , 885 , 000[+] ) . The GeneWise output is provided as S1 Appendix , the software tool was not able to find a match on the fly genome for the initial 126 residues from the human amino term . In order to assess the relevance of the fly Ser83 two homology-based approaches were implemented . First , a search on PFAM database [83] returned the domain of unknown function DUF4071 ( PF13281 ) . The first amino acid positions for this domain are highly conserved across a set of diverse homologs , including fly Ask1-PC and human MAP3K5 but also paralogs sharing the domain . The consensus sequence for DUF4071 has been included on the right alignment block from S2 Fig for illustrative purpose . The second approach was considering a precomputed set of homolog sequences downloaded from EggNOG database ( version 4 . 5 ) [84] . A cluster of homologous sequences already included fly Ask1 ( KOG4279 ) , providing a total of 338 protein sequences for 164 species; a total of 49 sequences were selected from that set , fly Ask1-PC isoform among them , that were also reliably annotated as orthologs of human MAP3K5 . Fly Ask1-PB isoform protein sequence was retrieved from FlyBase and appended to the selected MAP3K5 orthologs . Then , the 50 protein sequences were aligned by MAFFT ( version 7 . 271 ) [85] using the following parameters: maxiterate = 1000 , localpair , op = 10 . The complete alignment is available from S2 Appendix , from which two conserved blocks centered on the human and fly Ser83 domains were trimmed and processed with TeXshade [86] to produce the S2 Fig . Note that on S2 Fig the domain highlighted in cyan , which includes the fly equivalent column highlighted in red , is highly conserved in a wide range of taxa , from sponges to humans . Other domains are conserved only within chordata or even tetrapoda species . Such conservation signature makes Drosophila Ser83 a good candidate for functional and mutational studies . pUASt-attb_Ask1WT was constructed by cutting Ask1 cDNA EcoRI /BamHI from DGRC clone FI02066 and cloning it into pUASt-attb . pUASt-attb_Ask1S83A . Two serines in the 75 and 83 residues in the DUF4071 domain showed putative non-canonical AKT phosphorylation sites ( ILTQQRPLSYHYGVRESF ) . Both residues are spatially exposed similarly to human Ser83 of ASK1 , which makes them accessible to kinases . pUASt-attb_Ask1 S83A was constructed by mutating serine 83 to alanine by PCR using oligos Ask1Mut-Fwd and Ask1S83A-Rev for partial PCR1 and Ask1S83A-Fwd and Ask1Mut-Rev for partial PCR2 . Complete PCR was performed using the two partial PCRs as templates with oligos Ask1Mut-Fwd and Ask1Mut-Rev . The complete PCR was then cut with EcoRI /PflMI and cloned into FI02066-cut EcoRI/PflMI . The mutation introduced a new StuI site that was used to check the mutated clones . Mutated Ask1S83A cDNA was then cut from EcoRI/BamHI and cloned into pUASt-attb . Both clones were injected by standard procedures in line zh-86Fb-attP and transgenic lines were selected . Ask1Mut-Fwd: AAT ACA AGA AGA GAA CTC TGA ATA CGG AAT Ask1Mut-rev: CGG CGG TGT GGT TTT GTG CAC AAA CCG ATC Ask1S83A-Fwd: CGT TAG GGA GGC CTT CGG GAT GAA GGA GA Ask1S83A-Rev: CGG CGG TGT GGT TTT GTG CAC AAA CCG ATC Immunostaining was performed using standard protocols . The primary antibodies used in this study were the polyclonal Ask1 P-Ser83 ( Santa Cruz Biotechnology sc-101633 1:100 , which recognizes the conserved region surrounding human P-Ser83 ) and Ask1 P-Thr ( Santa Cruz Biotechnology sc-109911 1:100 , which labels the preserved region nearby mouse P-Thr845 ) . We selected the anti-Ask1 P-Ser83 because of the following in vivo analysis: ( 1 ) its localization dropped under a UAS-Ask1RNAi ( S3C Fig ) ; ( 2 ) Its localization increased after ectopic activation of Ask1WT ( Fig 2G ) ; ( 3 ) Mutation in the Drosophila Ser83 residue ( UAS-Ask1S83A ) , inhibited its localization ( Fig 2H ) ; ( 4 ) Akt activation or repression , resulted in increased or decreased localization of P-Ser83 ( Fig 3E–3H ) . Other antibodies used were: Ptc ( DSHB , 1:100 ) , P-p38 ( Cell Signalling 1:50 ) , Mmp1 ( cocktail of three antibodies: DSHB 3A6B4 , 5H7B11 , 3B8D12 1:100 ) , P-Akt ( S473 Cell Signalling 1:100 ) and P-Histone-H3 ( Millipore , 1:1000 ) . The anti-P-JNK used was the Anti-ACTIVE® JNK pAb , Rabbit , ( V7931 , Promega , 1:100 ) . The images were taken using the Thermal LUT from the 3D Surface Plot of FIJI , with maximum values of 50% and minimum 24% to minimize the background . Calculation of pixel intensities was obtained from raw images . Fluorescently labeled secondary antibodies were from ThermoFisher Scientific . Discs were mounted in SlowFade or ProLong ( ThermoFisher Scientific ) supplemented with 1 μM TO-PRO-3 ( TP-3 ) or YO-PRO-1 ( YP1 ) ( Life Technologies ) to label nuclei . For apoptotic cell detection , we used the TUNEL assay . We employed the fluorescently labeled Alexa Fluor® 647-aha-dUTP ( ThermoFisher Scientific ) , incorporated using terminal deoxynucleotidyl transferase ( Roche ) . The number of mitotic cells per mm2 ( Fig 1C ) was calculated after counting the number of P-Histone-H3 ( P-H3 ) positive cells in the anterior compartment of the wing pouch and hinge , excluding the notum , for all the genotypes shown . Note that two genotypes in Fig 1C show the anterior compartment labeled with the transgene ci-Gal4 UAS-GFP ( ci>GFP ) , whereas in the two ci>Ask1RNAi experiments the anterior compartment was labeled with anti-Ci . In these experiments shown in Fig 1C , the induction of cell death or activation of transgenes was done for 16h at 29°C . The number of mitosis after analyzing the stacks of confocal images was calculated using Fiji software . Wing discs were dissected from third instar larvae in Schneider’s insect medium ( Sigma-Aldrich ) , and a small fragment was removed with tungsten needles . To visualize Mmp1 staining , the discs were cultured in Schneider’s insect medium supplemented with 2% heat-activated fetal calf serum , 2 . 5% fly extract and 5 μg/ml insulin , for 5 hours at 25°C . For P-p38 , discs were injured in Schneider’s insect medium and immediately fixed and stained . Ex vivo images were taken using a Leica SPE confocal microscope and processed with Fiji software . Ex vivo discs were also monitored live after a cut , using puc-Gal4 UAS-GFP as a JNK reporter , using the same medium . For nuclei visualization in ex vivo culture , we used NucRed Live647 ( Life Technologies; 1 drop per slide ) added 20 minutes before imaging . To prevent ROS production in salE/Pv>rpr discs we pursued two different protocols . First , we inhibited ROS chemically ( Fig 4A–4F ) . To do this , standard fly food was supplemented with the antioxidant N-acetyl cysteine ( NAC 100 μg/ml; Sigma-Aldrich ) . NAC treatment was dispensed on the 7th day of development at 17°C . On the 8th day , experimental larvae were moved to 29°C for 11 hours to promote cell death , whereas controls were transferred to a vial with standard food and moved to 29°C for the same time period . Afterward , the larvae were move back to 17°C to allow tissue recovery . Second , we decreased ROS production genetically by the ectopic expression of the Sod1 and Catalase enzymes using a recombinant fly UAS-Sod1:UAS-Cat ( Sod1:Cat ) in the salE/Pv>rpr domain ( Fig 4G and S6 Fig ) . For those experiments , Sod1:Cat and rpr were activated for 24 hours at 29ºC . Third instar larvae were transferred to vials with 5mL of special medium containing 1 . 3% UltraPureTM LMP agarose ( Invitrogen ) , 5% sucrose ( Fluka ) and the desired concentration of 0 . 1% H2O2 ( Merck ) or 1ng/μl tunicamycin ( Sigma-Aldrich ) . To avoid loss of oxidative capacity , these substances were added to the media at a temperature below 45°C . The larvae were fed for 2 hours prior to dissection and fixation of the discs . Controls without H2O2 or tunicamycin were always handled in parallel . Control ( hh>RFP ) and experimental ( hh>Ask1RNAi ) conditions were induced for 16h at 29°C ( 192h after egg laying ) using the Gal4/tub-Gal80TS system . Wing imaginal discs ( n = 40 ) were dissected after induction in three biological replicates for each condition . RNA was extracted with the Zymo Research ZR RNA MicroPerp ( R1060/R1061 ) and RNA Clean and Concentrator ( R1015/R1016 ) kits following standard protocols . RT-PCR was performed using SYBR green Master Mix ( Roche ) . Specific primers for Ask1 and Rps18 are detailed in S1B Fig . ΔΔCt method was used to normalize the data . Ask1RNAi and control samples were normalized against the housekeeping gene Rps18 . Average standard error of the mean ( SEM ) of the three biological replicates was computed for each one based on three technical replicates by ΔΔCt method . Fig 1 . ( A ) +/+ → wUAS-rpr/+; salE/Pv-Gal4/+; tubGal80TS/+ Ask1MI02915/+ → wUAS-rpr/+; salE/Pv-Gal4/+; tubGal80TS/Ask1MI02915 Ask1MB06487/+ → wUAS-rpr/+; salE/Pv-Gal4/+; tubGal80TS/Ask1MB06487 ( B ) salE/Pv>rpr → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS-GFP ci>Ask1RNAi → w; ci-Gal4/LexO-GFP; salE/Pv-LHG:tubGal80TS/UAS-Ask1RNAi salE/Pv>rpr ci> Ask1RNAi → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS- Ask1RNAi ap> Ask1RNAi → w; ap-Gal4/LexO-GFP; salE/Pv-LHG:tubGal80TS/UAS- Ask1RNAi salE/Pv>rpr ap> Ask1RNAi → w; ap-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS- Ask1RNAi ( C ) sal>GFP ci>RFP → w; ci-Gal4/LexO-GFP; salE/Pv-LHG:tubGal80TS/UAS-RFP sal>GFP ci>Ask1RNAi → w; ci-Gal4/LexO-GFP; salE/Pv-LHG:tubGal80TS/UAS-Ask1RNAi salE/Pv>rpr ci>GFP → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS-GFP salE/Pv>rpr ci> Ask1RNAi → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS- Ask1RNAi Fig 2 . ( B , C ) WT → Canton S ( D ) salE/Pv>rpr → wUAS-rpr/+; salE/Pv-Gal4/+; tubGal80TS/+ ( E , F ) salE/Pv>rpr ci> Ask1RNAi → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS- Ask1RNAi ( G ) hh>Ask1WT , GFP → w; UAS-GFP/+; hh-Gal4/UAS- Ask1WT ( H ) hh>Ask1S83A , GFP → w; UAS-GFP/+; hh-Gal4/UAS- Ask1 S83A ( I ) sal>GFP ci>Ask1WT → w; ci-Gal4/LexO-GFP; salE/Pv-LHG:tubGal80TS/UAS- Ask1WT sal>GFP ci>Ask1 S83A → w; ci-Gal4/LexO-GFP; salE/Pv-LHG:tubGal80TS/UAS- Ask1 S83A salE/Pv>rpr → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS-GFP salE/Pv>rpr ci> Ask1WT → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS- Ask1WT salE/Pv>rpr ci> Ask1 S83A → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS- Ask1S83A Fig 3 . A ) Canton S ( B ) ptc>rpr → wUAS-rpr/+; ptc-Gal4: tubGal80TS /+ ( C ) sal>rpr → wUAS-rpr/+; sal-Gal4/+; tubGal80TS/+ ( D , E ) ptc>AktRNAi → w; ptc-Gal4:tubGal80TS/+;UAS-AktRNAi/+ ( F , G ) ptc>myrAkt → w; ptc-Gal4:tubGal80TS/+; UAS-myrAkt/+ ( H-K ) salE/Pv>rpr hh>dp110DN → w; LexO-rpr /UAS-dp110DN; salE/Pv-LHG:tubGal80TS/hh-Gal4 ( L ) ci>dp110DN → w; ci-Gal4/UAS-dp110DN; salE/Pv-LHG:tubGal80TS/+ +/+ → w; LexO-rpr/ +; salE/Pv-LHG:tubGal80TS/+ ( control for LexO-rpr in the second chromosome ) and wUASrpr/+; salE/Pv-Gal4; tubGal80TS ( control for mutant backgrounds ) salE/Pv>rpr ci>dp110DN → w; ci-Gal4/UAS-dp110DN; salE/Pv-LHG:tubGal80TS/ LexO-rpr salE/Pv>rpr Akt1/+ → wUAS-rpr/+; salE/Pv-Gal4; tubGal80TS; Akt1/+ salE/Pv>rpr Ask1MB06487/+ Akt1/+ → wUAS-rpr/+; salE/Pv-Gal4; tubGal80TS; Akt1/Ask1MB06487 salE/Pv>rpr Ask1MI02915/+ Akt1/+ → wUAS-rpr/+; salE/Pv-Gal4; tubGal80TS; Akt1/Ask1MI02915 Fig 4 . ( B , C ) Std food and NAC → wUAS-rpr/+; salE/Pv-Gal4/+; tubGal80TS/+ ( E , F ) CTRL → w118; +; + Ask1-/- → w118; +; Ask1MB06487/Ask1MB06487 ( G ) sal>rpr , GFP → w; salE/Pv-Gal4/UAS-GFP; tubGal80TS/UAS-rpr sal>rpr , Sod1:Cat → w; salE/Pv-Gal4/UAS-Sod1:UAS-Cat; tubGal80TS/UAS-rpr Fig 5 . ( A , C ) salE/Pv>rpr → w; LexO-rpr/ +; salE/Pv-LHG:tubGal80TS/+ ( B , D ) salE/Pv>rpr ci>Ask1RNAi → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS-Ask1RNAi ( E-H ) en>Ask1RNAi , GFP → w; en-Gal4/UAS-GFP; UAS-Ask1RNAi /+ S1 Fig . ( A ) salE/Pv>rpr Ask1MI02915/+ → wUAS-rpr/+; salE/Pv-Gal4/+; tubGal80TS/Ask1MI02915 salE/Pv>rpr Ask1MB06487/+ → wUAS-rpr/+; salE/Pv-Gal4/+; tubGal80TS/Ask1MB06487 salE/Pv>rpr ci> Ask1RNAi → w; ci-Gal4/LexO-rpr; salE/Pv-LHG:tubGal80TS/UAS- Ask1RNAi ( B ) Rt-q-PCR Controls: w; tubGal80ts/+; hh-Gal4/UAS-RFP UAS-Ask1RNAi: w; tubGal80ts/+; hh-Gal4/UAS-Ask1RNAi S3 Fig . ( A ) w118; +; + ( B ) Ask1+/Ask1+ → w118; +; + Ask1MB06487/Ask1MB0647 → w118; +; Ask1MB06487/Ask1MB06487 Ask1MB06487/Def ( 3R ) BSC636 → w118; +; Ask1MB06487/Def ( 3R ) BSC636 ( C ) hh>Ask1RNAi , GFP → w; UAS-GFP/+; hh-Gal4/UAS- Ask1RNAi S4 Fig . ( A ) wt and ptc>rpr → wUAS-rpr/+; ptc-Gal4: tubGal80TS /+ ( B ) salE/Pv>hid → w; salE/Pv-Gal4/+; tubGal80TS/UAS-hi ( C ) wt and physically injured → Canton S S5 Fig . ( A-D ) salE/Pv>rpr ci>dp110DN → w; ci-Gal4/UAS-dp110DN; salE/Pv-LHG: tubGal80TS/ LexO-rpr S6 Fig . w; salE/Pv-Gal4/UAS-Sod1:UAS-Cat; tubGal80TS/UAS-rpr S7 Fig . ( A ) wt→Canton S salE/Pv>rpr → w; LexO-rpr/+; salE/Pv-LHG: tubGal80ts salE/Pv>rpr , ci>Ask1RNAi→ w; ci-Gal4/LexO-rpr; SalE/Pv-LHG: tubGal80ts/UAS-Ask1RNAi ( C ) w; UAS-GFP/+; puc-Gal4/+ ( D ) w; UAS-GFP/+; puc-Gal4/Ask1MB06487 | One of the early events that occur after tissue damage consists on the production of Reactive Oxygen Species ( ROS ) , that signal to the surrounding tissue to initiate wound healing and regeneration . Many signaling pathways , such as JNK and p38 , respond to oxidative stress and are necessary for regenerative growth . As the link between ROS and regenerative signaling is not well understood , we decided to explore the mechanism that underlies this process . To do that , we genetically induced cell death in specific areas of Drosophila wing imaginal discs and then studied the mechanism that drives living cells to replace the damage zone until it is completely regenerated . We found that the Drosophila Apoptosis signal-regulating kinase 1 ( Ask1 ) , a protein that is sensitive to oxidative stress , is a key player in this scenario . This protein acts as an intracellular sensor that upon damage activates the JNK and p38 regenerative signaling pathways . However , high activity of Ask1 can be toxic for the cell . This is controlled by Akt , a kinase downstream the insulin pathway , which attenuates the activity of Ask1 in the living cells that will participate in the regeneration process . In consequence , Ask1 and Akt act synergistically to respond to the stress generated after tissue damage and drive regeneration . Our results provide a first overview within the framework of how insulin signaling inputs could modulate the capacity to overcome tissue damage . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"cell",
"death",
"invertebrates",
"redox",
"signaling",
"cell",
"processes",
"animals",
"animal",
"models",
"developmental",
"biology",
"drosophila",
"melanogaster",
"model",
"organisms",
"organism",
"development",
"experimental",
"organism",
"systems",
"morphogenesis",
"... | 2019 | Ask1 and Akt act synergistically to promote ROS-dependent regeneration in Drosophila |
Although social behaviour can bring many benefits to an individual , there are also costs that may be incurred whenever the members of a social group interact . The formation of dominance hierarchies could offer a means of reducing some of the costs of social interaction , but individuals within the hierarchy may end up paying differing costs dependent upon their position within the hierarchy . These differing interaction costs may therefore influence the behaviour of the group , as subordinate individuals may experience very different benefits and costs to dominants when the group is conducting a given behaviour . Here , a state-dependent dynamic game is described which considers a pair of social foragers where there is a set dominance relationship within the pair . The model considers the case where the subordinate member of the pair pays an interference cost when it and the dominant individual conduct specific pairs of behaviours together . The model demonstrates that if the subordinate individual pays these energetic costs when it interacts with the dominant individual , this has effects upon the behaviour of both subordinate and the dominant individuals . Including interaction costs increases the amount of foraging behaviour both individuals conduct , with the behaviour of the pair being driven by the subordinate individual . The subordinate will tend to be the lighter individual for longer periods of time when interaction costs are imposed . This supports earlier suggestions that lighter individuals should act as the decision-maker within the pair , giving leadership-like behaviours that are based upon energetic state . Pre-existing properties of individuals such as their dominance will be less important for determining which individual makes the decisions for the pair . This suggests that , even with strict behavioural hierarchies , identifying which individual is the dominant one is not sufficient for identifying which one is the leader .
Animals can gain many benefits from associating in groups [1] , but there are disadvantages that need to be considered as well . Competition for resources that cannot easily be shared equally ( such as food , security , or access to mates ) can lead to conflict between the members of the group [2] . Many species have social mechanisms for quickly or automatically resolving these conflict situations , such as the quick and definitive formation and maintenance of dominance relationships [3] , [4] . Because group-living often means that individuals within a hierarchy are constantly interacting , these relationships can have long-term consequences upon the behaviours shown by individuals of different social standing , and therefore the effects of dominance need to be considered when we are interested in understanding the individual and collective behaviours of the group’s members . Ignoring the effects of these interactions could lead to us misunderstanding behavioural dynamics at the levels of the individual and of the group [5] , [6] . Differing effects of dominance upon the behaviour of interacting individuals have been integrated into a number of theoretical studies . Some have concentrated upon how dominance affects the order of access to resources , where dominants may have priority or exclusive access to a foraging resource [7] , [8] . Alternatively , dominant individuals may benefit during social foraging by reducing the degree of predation risk they experience during foraging by forcing subordinates into riskier positions [9]–[11] or more dangerous foraging periods during the day [12] . Other models have considered these imbalances in access to resources [13] , as well as cases where dominance interactions lead to the spatial displacement of lower ranked individuals away from a patch [14] , [15] . As well as effects upon foraging ability and group composition , living in social hierarchies can impose differing costs on individuals [16]–[20] . Rands et al . [6] considered an individual-based model consisting of a population made up of dominant and subordinate individuals , where all the individuals followed simple foraging rules when foraging , and where all individuals experienced the same energetic costs and gains during foraging , regardless of whether they were dominant or subordinate . However , subordinate individuals also experienced an additional cost when they foraged in close proximity to dominant individuals . This single additional cost had effects upon the movement behaviour and energetic stores of the subordinate individuals within the population . This single cost was considered to give a simple representation of a ‘socially mediated interference’ cost [21] , [22] , and considered the situation where a subordinate suffered if it foraged at the same time as a dominant individual ( which could then be compared to a similar model where dominance costs weren’t considered [23] ) . The way in which this socially mediated interference cost was implemented within the spatially-explicit individual-based framework of [6] involved making some broad assumptions about the rules that individuals use . As Rands [24] , [25] discusses , making assumptions about rules within these models is useful , but we can greatly enhance the value of these rules if we are able to derive them from an even simpler set of assumptions . Therefore , in this paper I describe a state-dependent dynamic game model that explicitly considers the effects of several different socially-mediated costs upon the behaviour of a pair of socially foraging animals . I aim to show that imposing socially-mediated costs has effects upon the optimal behaviours of both the individual that is paying the costs ( the subordinate ) , and in addition , upon the individual who is not directly paying these costs ( the dominant ) . The dominant is instead being affected indirectly by these costs due to their effects upon the behaviour of the subordinate . In addition , I will also consider whether imposing costs of dominance mean that a dominant individual is the individual driving the behaviour of the pair .
The model followed here builds on the dynamic foraging game described by Rands et al . [26] , [27] . In the simpler model described in [26] , the decisions made by a pair of individuals are considered . The model uses dynamic programming techniques [28]–[32] to identify the optimal behaviours of a pair of animals , who are both characterised by possessing energy reserves ( which defines their ‘state’ ) which change stochastically as a result of the actions that both individuals take over a series of consecutive decisions . Both members of the pair are able to accurately assess each other’s energetic reserves as well as their own , and their actions are informed by this information . During a period , each individual can choose to conduct one of two actions: either to rest or to forage for the entire period . Both actions incur an energetic cost which depletes the reserves of the individual , but foraging can also lead to the forager finding food ( within a stochastic environment ) , meaning that , on average , it should see a net gain in its energetic reserves if it forages . Energetic reserves are important within the model: it is assumed that if they fall too low , an individual starves to death . It is also assumed that there is an upper limit to the capacity of the reserves , beyond which they cannot be increased further . As well as the risk of starvation if an individual doesn’t forage , there is also the risk of predation , which depends on the actions of both individuals in the pair . If an individual choses to rest during a period , it incurs a low risk of being predated . If it forages at the same time as its colleague , it incurs a moderate risk of being predated ( which could be through increased protection against predation from being in a small group , or enhanced detection , or simply a dilution of risk ) . If it forages on its own ( whilst its colleague rests ) , it incurs the greatest risk of being predated . Therefore , there is a trade-off within this framework between being predated when foraging , and starving whilst resting . Rands et al . [26] demonstrate that these assumptions can be modelled using a stochastic dynamic game , and show that optimal policies can be calculated , which describe the optimal actions of an individual within a pair: the policies allow an individual to identify the suitable action to conduct given that it knows its own energetic reserves and those of its colleague at a given moment in time . Rands et al . [27] extend these models by considering what occurs when individuals are not identical in the costs they incur for conducting actions , or the amount of energy they gain during a period , or in the risks they face when conducting specific actions . The model I describe here builds further on this framework . Although Rands et al . [27] considered possible differences between individuals in various parameters , they did not consider what could happen in a dominance interaction , where specific behavioural interactions between the two members of a pair incurred additional costs to one of the members ( which I refer to as the ‘subordinate’ ) , similar to the socially-mediated interference costs proposed in [6] . Note that I assume both that the dominance hierarchy has been decided by the pair members prior to the start of the period modelled , and that this hierarchy is adhered to throughout , with no further requirements to maintain it ( see [33] for work considering the formation and maintenance of hierarchies ) . Here , I consider there to be four possible situations where an additional cost can be incurred by the subordinate: In ( iii ) and ( iv ) , I assume the dominant individual is conducting the opposite behaviour to the subordinate . The model considers a dynamic game between pair of players consisting of a dominant and a subordinate individual . This dynamic game builds on solution procedures outlined in [28] and [34] , following a state-dependent framework as described in [28]–[31] . General computation methods build on the dynamic game framework for pairs of foragers , outlined in [26] and described in detail in [27] , and the reader is referred to the latter for full details of the assumptions and the computational solution process , which are not repeated here . Unless described here , details are identical to those given in [27] , and consequently I do not repeat any analysis for the effects of variables other than the socially-mediated costs that are introduced in this paper . To summarise the procedure described in [27] in a brief , using the assumptions about the effects of pair members’ actions as detailed in the overview above , an initial candidate strategy is assumed . This defines all possible actions that each individual should take , given that it knows its own energetic reserves and those of its colleague at a moment in time . Assuming one of the pair members is using the current candidate strategy , a best response can be calculated for its colleague using dynamic programming . To do this , I make an additional initial assumption about how energetic state relates to fitness ( where fitness is used as a common currency to compare all possible actions [35] ) , but this initial assumption about how fitness relates to state is rendered unimportant through strong backwards convergence [28] . Once an optimal response strategy to the current population strategy has been identified , the best response to that strategy could be calculated , and then the best response to that , and so forth , with the aim of identifying an evolutionarily stable strategy ( ESS ) . However , it is difficult to iterate to an ESS using this direct route ( e . g . [36] , [37] ) , and I instead used a error-making approach [34] , where the candidate strategy is updated at each iterative step by combining the previous candidate strategy with the newly identified best response strategy ( weighting the new candidate strategy strongly towards the previous candidate strategy ) . Using this technique , an ESS is identified through an iterative computational process . For finer detail of the assumptions , please see [26] and [27] . Where the current model differs from that presented in [27] is in the detail of the function denoted Hi ( xi , xj , t; ui , uj , π ) , which defines the probability that an individual of type i who is alive at the start of time step t , in state xi ( >0 ) , paired with a living colleague of corresponding type j in state xj ( >0 ) who follows a strategy defined by the candidate strategy π , will survive until the start of the final time step T , if it adopts action ui and its colleague adopts action uj in the current time step ( assuming that the focal individual i thereafter behaves so as to maximise its chances of surviving until time step T , taking into account errors in decision making ) . Note that , as with the model described in [27] , the candidate strategy π encompasses the candidate responses of both subordinate and dominant individuals within the population ( which means that it defines the current ‘best’ action that an individual should take given that it knows its own energy reserves , and those of its colleague: the modelling process considers all possible state combinations of energy reserves for both dominant and subordinate individual , and the candidate strategy therefore includes current ‘best’ actions for all of these ) . In order to consider the effects of dominance on behaviour , I replace the Hi ( xi , xj , t; ui , uj , π ) function described in [27] with the following set of definitions , which are dependent upon whether the focal individual is dominant or subordinate , and assume that both individuals in a pair are alive at the moment the decision is made ( note that the associated functions that describe what occurs when only the focal individual is alive at the decision point , or that describe what happens if no individuals are alive at the decision point , are identical to those presented in [27] , and are therefore not described here ) . Throughout , terms relevant to subordinates are denoted with a subscript s , and terms relevant to dominants are denoted with a subscript d . If the individual is dominant , I assume If the focal individual is subordinate , I instead assume Apart from the novel interference cost terms ( described below ) , terminology here follows [27] , and is only briefly summarised here . An individual of type a has a probability maR of being predated if it is resting , maA if it is foraging alone , and maT if it is foraging with its colleague . The function Wa ( xa , xb , t;π ) is the fitness at time t of an individual of type a with energy reserves xa , and whose colleague has energy reserves xb , assuming that both individuals follow policy π from that point forward in time . Λa ( x ) defines a ‘chop’ function for an individual of type a as defined in [30] , where Λa ( x ) = min ( Sa , max ( x , 0 ) ) , and Sa is the maximum state value possible for an individual of type a . κa ( ca;u ) denotes the probability that an individual of type a spends ca state units of energy during a period if it conducts action u , and γa ( ga ) denotes the probability that it gains ga state units of energy during the period if it forages . Both energetic costs and gains were represented within the current model using the same functions described in [27] , where the probabilities defined followed a discretised distribution based on normal distributions with defined means and standard deviations . For simplicity , both dominant and subordinate individual were assumed to have identical probabilities of incurring given gains and costs , and share identical predation risks when conducting particular activities ( so mdT = msT , etc . ) . The only difference considered between them was in the extra cost paid by the subordinate individual when it was conducting a paired behaviour that incurred extra energetic costs . These energetic costs are denoted DdRsR , DdFsR , DdRsF and DdFsF , which represent the extra energetic cost ( in state units ) incurred for the four possible pairs of behaviours ( where the general form DdUsV represents the cost paid by the subordinate when the dominant conducted action U and the subordinate conducted action V ) . The dominant is not expected to pay any extra costs for its actions dependent upon the behaviour of the subordinate . Once stable behavioural policies for subordinate and dominant individuals had been identified , forward iterations using Markov chain processes [29] were then used to calculate the distribution of a pair’s states within a stable population . Again , full details of the process and assumptions made follow those described in [27] . Having identified a stable distribution of paired states , I then calculated the following summary statistics: I explored the effects of the costs by randomly generating 1 , 000 sets of other model parameters , and then calculating optimal policies and population distributions for all possible combinations of the four socially-mediated interference costs . Table 1 describes the parameters used within this model , including the use of randomisation to generate differences between parameter sets . For each set of parameters , I considered the sixteen possible scenarios where DdRsR , DdFsR , DdRsF and DdFsF could each take a value of either 0 or 1 state units , representing a full spectrum of cases where there was a potential cost to be paid by a subordinate individual dependent upon the actions of the dominant member of the pair . After calculating policies and stable population distributions , summary statistics were calculated for each of these sixteen possible scenarios , and exploratory analyses were conducted as detailed below . The model considers four possible interference costs to a subordinate individual , all of which could potentially have a separate effect upon its energetic turnover during a period dependent upon the behaviour of the pair . Therefore , I was interested in the interactions of the costs , as well as each of the costs themselves . To explore this , standard analysis of variance was used to generate F values for the four costs and the eleven possible interactions involving two , three or all four of these . The distribution of the results generated would not fit the standard assumptions necessary for ANOVA , and so I used resampling methods to identify critical F values , following recommendations in [39] . Because I was potentially interested in the effects of interactions , I would have been unable to generate resampled critical F values by the random assortment of results such that the untested costs or interactions were kept correctly assorted . However , my sample population of results was large , and I therefore generated resampled critical F values by freely permuting my entire dataset without restriction , following recommendations in [39] . For each , I used R 2 . 12 . 1 [40] to permute the entire dataset without replacement 50 , 000 times , harvesting the fifteen F values for an ANOVA conducted on each permuted set , and used the quantile function within R to identify the value of the 95% quantile values .
All individual increases in dominance cost led to an increase in the amount of foraging behaviour shown by the subordinate ( Table 2 ) – there were also significant interactions between paired costs ( and most three- or four-way interactions ) , although in all cases adding a cost led to an increase in foraging behaviour . Increasing the costs experienced by the subordinate led to increases in most of the individual foraging behaviour shown by the dominant , except for the case where the subordinate only experienced costs when it was resting and the dominant was foraging , which is likely to be a situation when the dominant is not going to be affected by the actions of the subordinate too much . The increases in individual foraging behaviour were also echoed in the paired behaviours ( Table 2 ) . Considering all the single costs within the statistical model , increasing any of the costs experienced by the subordinate individual led to an increase in its foraging behaviour , and a decrease in its resting behaviour . The fact that the direction of change is dictated solely by the subordinate individual suggests that the action of the dominant individual is being driven primarily by its foraging partner . Paired costs led to increases in both individuals foraging together , and ( apart from the case where the subordinate always paid a cost when the dominant was foraging ) , decreases in resting together . The paired interactions when the members of the pair were conducting differing behaviours were mostly non-significant , although there were increases in cases where the dominant rested and the subordinate foraged when the costs experienced by the subordinate occurred when the pair differed in their behaviour . As would be expected , pairs become more synchronised when there are costs involved with not being paired , and become less synchronised when there are costs to conducting the same action as each other ( Table 2 ) . Considering this alongside the paired behaviour results suggests that although the subordinate is driving the behaviours of the pair , its own actions are therefore partially dictated by the costs that it pays . Note also that this measure ( with similar reasoning for the following S statistic ) does not discriminate between resting together and foraging together . Therefore , an increase in only one of these paired behaviours may not lead to a corresponding increase in the general level of synchronisation within the pair . The S statistic decreased in response to a dominance cost when foraging together , and decreased when resting together ( Table 2 ) . Resting alone doesn’t incur any more risk of predation than when resting together , and it is therefore feasible that as resting together becomes more costly to the subordinate , it should therefore become more dependent upon the state of its colleague dictating its actions ( in this case , avoiding resting together ) , leading to the decrease in synchrony shown by the synchrony coefficient . The decrease in dependence with an increasing cost of foraging together is echoed in the observation that the subordinate individual should be increasing its foraging regardless of the actions of the dominant . Both the dominant and subordinate individuals tended to increase their repetition of behaviour when there was an extra cost to the subordinate of foraging at the same time as the dominant ( Table 3 ) . This is likely to be an effect of foraging being highly synchronised: the subordinate has to forage at the same time as the dominant , and consequently needs the pair to spend more time foraging than the dominant in order to fund the energy it spends ( especially , but counterintuitively , whilst foraging ) . Both individuals tended to reduce their repetition of behaviour when there was an extra cost to the subordinate of resting together . This is likely to be due to the reduction in the amount of time that the subordinate rests overall – an increased likelihood of foraging suggests that a pair of individuals will be swapping between different pairs of behaviours , and is demonstrated in a similar reduction in the mean length of time that pairs of individuals repeated a paired behaviour ( Table 3 ) . The subordinate individual also tended to repeat its own behaviour more often when there was a dominance cost associated with conducting the opposite behaviour to the dominant individual ( note here that this means an overall increase in the subordinate repeating a behaviour irrespective of what the dominant is doing , rather than a statement that the subordinate is increasing conducting the opposite behaviour to the dominant ) . Most of the interactions shown for the subordinate individual also indicate a positive trend . Paired behaviours were also repeated more often when these costs were incurred . These increases are likely to be due to the increase in synchronisation behaviour seen when there is an extra cost to being non-synchronised . As would be expected , incurring an extra cost of dominance to the subordinate meant that its energetic reserves tended to be reduced ( Table 3 ) . This was not the case where there was a dominance cost to the subordinate when it rested and the dominant foraged , which may be due to an increase in the subordinate tending to forage in order to avoid this cost . Regardless of which sort of cost was imposed on the subordinate , the dominant tended to gain energetic reserves when there was a cost , which ties in with the increase in subordinate foraging behaviour and corresponding synchronisation by the dominant individual . The length of time that the subordinate remained heaviest ( when it managed to reach that state of being ) was in most cases reduced by imposing a cost of dominance ( Table 3 ) . The exception to this followed a similar pattern to the energetic reserves , where imposing a cost when the subordinate rested and the dominant foraged tended to lead to an increase in the length of time that the subordinate remained heaviest . Again , this is likely to be due to the subordinate increasing the amount of time it forages in response to this cost , therefore leading to an increase in its reserves . Increasing the length of time the subordinate individual remained heaviest should logically lead to a decrease in the length of time that the dominant individual remained heaviest , and vice versa . This trend was seen , but was only significant for the situations where the subordinate’s costs were paid for conducting the opposite behaviour to the dominant individual .
This model demonstrates that if a subordinate pays energetic costs when it interacts with a dominant individual , this has distinct effects upon the behaviour that it shows , and subsequently it affects the behaviour of the interacting dominant individual . Considered independently , both individuals tended to increase the amount of foraging behaviour they conducted when there were interaction costs . Considered together , the behaviour of the pair was driven by the subordinate individual . Costs imposed when the subordinate forages tend to increase paired foraging behaviour . In the model presented here , the subordinate individual tended to be the lighter individual for longer periods of time when interaction costs were imposed . This lends support to the suggestion that the lighter individual acts as the decision-making ‘pace-maker’ of the group [26] , [27] , giving leadership-like behaviours that are based upon state [24] , rather than specific pre-existing properties of individuals such as their dominance level [41] or tendency towards leadership [42] . As Rands et al . [27] discuss , consistent leadership behaviour can be a property of individuals with a higher metabolic requirement ( such as in lactating female zebras [43] ) . Therefore , although a dominance relationship exists in the pairs modelled , the behaviour of the pair is determined by an individual whose identity emerges from the interaction between the pair , rather than being strictly set by which individual is dominant to which . This ‘leadership’ status should also be transient within the pair , with both the dominant and the subordinate individual taking it in turns to be lightest and thus determine the actions of the pair . Within the model , imposing most sorts of interaction cost leads to a reduction in the reserves of the subordinate individual , leading in turn to it remaining heaviest for less consecutive periods of time . This means that imposing a cost of interaction should lead to the subordinate individual becoming the decision-maker more often . Of course , it should be noted that although there are examples where subordinates tend to be the ones gaining the most energy reserves ( e . g . [10] , [44] , [45] ) , there are also many empirical examples where dominant individuals tend to be both the decision-makers and the ones gaining the most food ( e . g . [46]–[49] ) . Therefore , imposing direct energetic costs of dominance should lead to effects upon the paired behaviour of foragers , lending support to the rules proposed for larger groups by Rands et al . [6] . Many of the costs of dominance are not directly energetic [9] , [50]–[55] . Although these rules are a relatively simple representation of a possible cost , the idea of behaviourally-mediated interaction costs has biological merit . Differences may exist in energetic expenditure between individuals of different social ranks , as has been demonstrated in fish [18] and birds [19] , [20] ( but see [56] ) . Mass gain may differ between individuals of different ranks , even if they appear to show equal feeding rates [57] , [58] . These differences in mass gain by individuals of differing social status may be due to differences in digestive ability [59] , [60] , or simply a behavioural difference in the amount of time spent foraging [52] , [61] – both of which could be represented by the cost modelled here . Individuals could also be behaviourally mediating the costs that they pay in interactions [54] , such as subordinates taking a lower share of resources when social dominance exists as a means of mediating the behavioural interaction . Lindström et al . [54] discuss whether a larger body mass could mean that individuals are better at mediating these costs . The model presented here only considers a difference in costs spent during activities , but could be extended to consider individuals with very different metabolic requirements , in a similar manner to the model presented by Rands et al . [27] . However , it is likely that a larger scale dominance model including differences between individuals would yield complex relationships that would not be simple to describe from a qualitative perspective , and should maybe be reserved for systems where some amount of parameterisation is possible . Furthermore , the current model makes a simplifying assumption by assuming that the subordinate paid additional costs ( and indeed , being subordinate is solely defined by paying these costs within the model ) . We could conceivably see a situation where the dominant individual also pays additional costs for being dominant . If these costs are less than those paid by the subordinate , these could simply be subsumed into the general metabolic costs paid by individuals , giving us a similar model structure to that described . However , if the dominant and the subordinate individual paid different levels of cost for different behaviours such that both paid more than the other for at least one of the four behavioural pairs , then this would be a situation not covered within the current model . For example , we could imagine a situation where the dominant paid the higher metabolic cost when foraging at the same time as the subordinate ( such as through having to be aware of the subordinate’s foraging actions , and through expending energy in forcing the subordinate away from resources ) , whilst the subordinate could show a higher metabolic cost than the dominant when it was foraging on its own ( such as through raising vigilance levels to spot both predators and in anticipation of the currently absent dominant individual ) . In this hypothetical example , the current model is not sufficient , and an extended version would need to be considered where costs to the dominant individual are also modelled . I would suggest that this exercise might be useful if exact predictions are needed for a well-defined system ( such as tying the model in with an empirical system ) , but investigating a more general model would be unlikely to yield more tangible results than described in the simpler model I present here . As well as behavioural interactions leading to subordinates gaining less energy during an interaction , physiological processes may also mean that they spend more energy , and could be mediated hormonally , such as through stress responses by individuals . Studies on many species have demonstrated that social stress and dominance interactions have effects upon body mass and composition [9] , . Hierarchy rank and measures of stress typically depend on the social conditions experienced by animals , and whether there have been recent changes in how the social structure is organised [68] . Stress , as measured by levels of hormones such as glucocorticoids , shows no obvious relationship with dominance rank [69] , although there are some correlations with species social system [70] . Stress has subtle short- and long-term effects upon an individual’s physiology , and care would be needed to catch these effects within a dynamic game , although it has been demonstrated that stress can be successfully captured using a state-dependent approach [71] . Again , careful parameterisation is necessary , but could be very useful for extending the rules suggested here to larger models considering complex social interactions ( such as [6] , [23] , [72]–[74] ) , enhancing predictions about social behaviour and interactions . | Dominance hierarchies could offer interacting animals a quick way to settle disputes without having to use too much effort . However , individuals may pay a price for acknowledging their position within the hierarchy , which could influence how they choose to behave within the group . Consequently , the actions of the group may be shaped by the effects of the hierarchy on each of the group’s members . I consider the behaviour of a pair that consists of a dominant and a subordinate individual , where the subordinate pays an energetic cost when it interacts with the dominant . I show that having to pay this cost affects the behaviour of the pair . I also demonstrate that , although a social hierarchy is imposed , the behaviour of the pair is not determined by the dominance relationship , but is instead influenced by the energetic reserves of the pair , where the decision-maker may just be whoever is the hungriest . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"game",
"theory",
"computer",
"science",
"evolutionary",
"ecology",
"computer",
"modeling",
"animal",
"behavior",
"mathematics",
"theoretical",
"biology",
"ecology",
"evolutionary",
"biology",
"applied",
"mathematics",
"theoretical",
"ecology",
"biology",
"behavioral",
"e... | 2011 | The Effects of Dominance on Leadership and Energetic Gain: A Dynamic Game between Pairs of Social Foragers |
Cystic echinococcosis ( CE ) is a neglected parasitic zoonosis with considerable socioeconomic impact on affected pastoral communities . CE is endemic throughout the Mediterranean , including Morocco , where the Mid Atlas is the most prevalent area for both human and animal infection . The highest hospital annual incidence of human CE is recorded in the provinces of Ifrane and El Hajeb . However , hospital-based statistics likely underestimate the real prevalence of infection , as a proportion of cases never reach medical attention or official records . In 2012 , a project on clinical management of CE in Morocco was launched with the aims of estimating the prevalence of human abdominal CE in selected rural communes of the above mentioned provinces using ultrasound ( US ) screening and training local physicians to implement US-based focused assessment and rational clinical management of CE according to the WHO-IWGE Expert Consensus . A total of 5367 people received abdominal US during four campaigns in April-May 2014 . During the campaigns , 24 local general practitioners received >24 hours of hands-on training and 143 health education sessions were organized for local communities . We found an overall CE prevalence of 1 . 9% , with significantly higher values in the rural communes of Ifrane than El Hajeb ( 2 . 6% vs 1 . 3%; p<0 . 001 ) . CE cysts were predominantly in inactive stage , especially in older age groups . However , active cysts were present also in adults , indicating acquisition of infection at all ages . Province of residence was the only risk factor consistently associated with CE infection . Our results show a high prevalence and on-going , likely environmental transmission of CE in the investigated provinces of Morocco , supporting the implementation of control activities in the area by national health authorities and encouraging the acceptance and divulgation of diagnosis and treatment algorithms based on imaging for CE at both national and local level .
Cystic Echinococcosis ( CE ) is a globally distributed parasitic zoonosis , caused by the larval stage of the dog tapeworm Echinococcus granulosus sensu lato . Its life cycle develops between the dog ( and other canids ) , which is the definitive host harbouring the adult parasites in the intestine , and the sheep ( and other ungulates ) , which is the intermediate host where the larval form may develop in organs and tissues . Humans , which are aberrant “dead-end” intermediate hosts , as well as natural intermediate hosts , become infected through ingestion of eggs released with the faeces of parasitized dogs . Eggs can remain viable for months in the environment . In the intermediate host , the parasite larvae develop in organs and tissues , mainly the liver and the lungs , forming fluid-filled cysts ( commonly referred to as hydatid cysts ) that expand centrifugally . The cycle completes when the definitive host ingests viscera infected with hydatid cysts . CE affects mainly pastoral communities where close contact with the dog-sheep cycle and a contaminated environment occur . However , clear-cut association with risk factors is still lacking [1] . The global burden of CE has been estimated in 1 . 2 million people infected worldwide with over 3 . 5 million DALYs ( Disability-Adjusted Life Years ) lost globally every year [2] . However , these figures are likely underestimated . The geographical dispersal of the infection on vast rural areas with a patchy distribution , the absence of specific symptoms , and the lack of an effective disease record system , hamper a more precise assessment of infection prevalence and disease burden . Furthermore , CE mostly affects poor communities , is a disease with a low case-fatality rate with difficult and expensive diagnosis and treatment , and is an infection with a transmission cycle that is difficult to interrupt in the absence of sustained , expensive and well-coordinated programs involving both human and veterinary health services [2] . As a result , this disease is neglected [2 , 3] . Morocco is highly endemic for CE . The prevalence of infection in animals has been reported to be up to 58 . 8% in dogs , 19 . 3% in sheep and 48 . 7% in cattle , with consistent variations between regions ( Guide des activites de lute contre l’hydatidose/echinococcose . Comité Interministeriel de lute contre l’hydatidose/echinococcose , Royaume du Maroc , 2007 ) . The Mid Atlas region was found to be the most prevalent area for animal CE in surveys conducted between 2001 and 2004 by Azlaf and Dakkak [4] . A total of 23 , 512 operated human cases were recorded by the Ministry of Health of Morocco during the periods 1980–1992 and 2003–2008 [5] . An increasing average annual incidence of surgical cases was shown , from 3 . 6 to 5 . 2 per 100 , 000 inhabitants in 1980 and 2008 , respectively , with the region of Meknes-Tafilalet , in the Mid Atlas , recording the highest figures ( 11 . 9 per 100 , 000 in 2008 ) [5] . More recent data from the Ministry of Health of Morocco for 2014 report an annual incidence of human CE in the Meknes-Tafilalet region of 7 . 04 per 100 , 000 , with the highest figures recorded in the provinces of Ifrane ( 16 . 33 per 100 , 000 ) and El Hajeb ( 12 . 90 per 100 , 000 ) . These figures , however , are likely not representative of the real prevalence of infection , as a proportion of cases remain clinically silent often for many years and , even when symptomatic , may never reach medical attention or official disease records . More accurate data may come from screening campaigns , which allow to detect also asymptomatic cases and to evaluate the distribution of CE stages among age groups . In 2000 and 2001 , Macpherson and colleagues conducted an ultrasound ( US ) screening of 11 , 612 people in the provinces of Ifrane and Khenifra in the Meknes-Tafilalet region , finding a prevalence of abdominal CE of 1 . 1% ( 1 . 3% in Ifrane and 0 . 9% in Khenifra ) [6] . Most of the diagnosed CE cases were asymptomatic , even in patients with large active cysts , and symptoms , when present , were non-specific . The diagnosis and clinical management of CE are complex and require a multidisciplinary approach , often available only in referral centres . In 2003 , the World Health Organization Informal Working Group on Echinococcosis ( WHO-IWGE ) implemented a consensus US-based classification of CE cysts stages , which allows classifying unequivocally all morphological stages of cysts [7] . This classification also groups cyst stages into clinical categories to guide the rational stage-specific allocation of CE patients to different management options , including surgery , medical therapy , percutaneous treatment and the “watch and wait” approach [8 , 9] . This approach not only allows choosing the most appropriate treatment ( or the need for no treatment ) depending on cyst characteristics , patient-related factors , and therapeutic resources available , but also helps in rationalizing the expenses for CE management . Unfortunately , the use of this consensus approach , and of CE classifications of any kind , are still appallingly scarce , and the management of the disease is often inappropriate , exposing patients ( and health systems ) to unnecessary treatments , risks , and costs [10 , 11] . A survey on clinical practices in CE carried out in Morocco in 2008 found that 73 . 6% of interviewed physicians ( n = 148 ) would treat inactive CE cysts with surgery , only 4 . 1% would use the ( correct ) watch and wait approach , while 20 . 3% of the interviewed did not know what approach to take with inactive cysts [12] . In Morocco , the direct cost of surgery for abdominal CE has been estimated at 1500–3000 US$ per patient [4 , 5] . Further costs include the reduction or loss of income due to hospitalization and complications and , not less important , the impact on patient’s quality of life due to the risks , complications , and long hospital stay associated with surgery . Nonetheless , the treatment of CE in Morocco is still almost uniquely surgical , while other options such as percutaneous treatment are implemented in only a minority of cases [5] . According to data from the Ministry of Health of Morocco , in 2014 , percutaneous treatment was used in 1 . 8% , and medical treatment in 0 . 9% of recorded CE cases . In 2004 , an inter-ministerial committee for the control of CE was established in Morocco , involving the Ministry of Agriculture , the Ministry of Health and the Ministry of Interior . The committee started its activities in 2009 . However , the implementation of integrated control measures proved extremely difficult due to problems of inter-ministerial collaboration at the organizational and funding level , and so far only improvement of human CE case report system has been implemented . A re-evaluation of the prevalence of human CE was therefore considered necessary . In 2012 , an EUR 300 , 000 project “Clinical Management of Cystic Echinococcosis in Morocco” funded by the Italian Ministry of Health and coordinated by the WHO was launched , with the aims of: i ) estimating the prevalence of human abdominal CE in target endemic areas of Morocco using US screening; ii ) encouraging the use of US-based focused assessment of CE in peripheral endemic areas [13 , 14]; and iii ) encouraging a rational clinical management of CE by local physicians according to the WHO-IWGE Expert Consensus on diagnosis and management of human echinococcosis [8] . Here we present the results of the community-based US screening and related activities carried out between April and May 2014 in the target endemic provinces of Ifrane and El Hajeb , Meknes-Tafilalet region ( designated as such until December 2014 and indicated as such in the manuscript ) , Mid Atlas , an area among the most endemic in Morocco for both human and animal infection ( Guide des activites de lute contre l’hydatidose/echinococcose . Comité Interministeriel de lute contre l’hydatidose/echinococcose , Royaume du Maroc , 2007; and [4 , 5] ) .
Approval was granted by the Ethics Committees of the University of Pavia , Italy , and of the University Hospital Centre Hassan II of Fès , Morocco . The project developed as a 2-year collaboration between the University of Pavia , San Matteo Hospital Foundation , WHO Collaborating Centre for the Clinical Management of Cystic Echinococcosis , Pavia , Italy , and the Ministry of Health of Morocco , Service of Parasitic Diseases , Rabat , Morocco . The project was coordinated by the WHO , Department of Neglected Infectious Diseases and the WHO Regional Office for the Eastern Mediterranean , Cairo , Egypt . Moroccan study centres were represented by the Service of Medicine C and the Service of Surgery B , Avicenne ( Ibn Sina ) hospital , Rabat; the Service of Parasitology , Mohammed V Military Teaching Hospital and Faculty of Medicine and Pharmacy , Rabat; the Prince Moulay Hassan Hospital , El Hajeb; and the 20 August Hospital , Ifrane . Local support and coordination was provided by the WHO Office in Morocco , Rabat , the Regional Health Directorate of Meknes-Tafilalet region , and the Provincial Delegations of Ifrane and El Hajeb . The main objective of this cross-sectional study was to estimate the prevalence of abdominal CE in the target endemic provinces of Ifrane and El Hajeb , Meknes-Tafilalet region , Mid Atlas . Secondary objectives were i ) to assess risk factors associated with CE infection , ii ) to train local health professionals on the diagnosis and clinical management of abdominal CE according to the WHO-IWGE Expert Consensus on diagnosis and management of human echinococcosis [8] and , iii ) to provide educational inputs on the disease , its transmission , treatment and prevention , to the inhabitants of the communities involved in the screening . The evaluation of the effectiveness of the stage-specific clinical management of patients with abdominal CE according to the WHO-IWGE Expert Consensus is on going and will be the object of a further publication . A sample size of 5 , 000 subjects ( 2 , 500 per province ) was calculated to provide an estimate of CE prevalence with a 95% confidence level and 0 . 5% precision , based on an expected prevalence of 1 . 5% . The Shapiro-Wilk test was used to assess the normal distribution of quantitative variables . These were expressed as the mean and standard deviation , as they were normally distributed , and were compared by t-test . Qualitative variables were described as number and percentage . Differences between groups were evaluated using χ2 or Fisher’s exact test , as appropriate . Association between CE infection and potential risk factors , with the exception of age and sex , were assessed using univariate logistic regression models . To take into account potential confounders we calculated Odds Ratios ( ORs , with their 95% Confidence Interval ) adjusted for age , sex , and province by multivariable logistic models . Each type of “dog role” was analyzed as “yes/no” as dogs may have several roles in the same households . At cyst level , risk of being in active or inactive stage was assessed by logistic regression with robust standard errors , clustered at patient level . As missing data were below 5% , no statistical method for missing data was necessary . A p-value < 0 . 05 was considered statistically significant and a p-value < 0 . 10 was considered borderline significant . All tests were two sided . Data analysis was performed with the STATA statistical package ( version 14 . 1; Stata Corporation , College Station , TX , USA ) . Consent to conduct the screening was obtained from local authorities and community leaders during the preparatory phase of the project . The surveys comprised four 2-day campaigns in April and May 2014 , in the rural communes of Ain Louh and Timahdit in the province of Ifrane , and in the rural communes of Bouderbala and Sebt Jehjouh in the province of El Hajeb , Meknes-Tafilalet region ( Fig 1 ) . The main population of these areas of central Morocco are Amazighs . The rural communes of Ain Louh and Timahdit in Ifrane province are located at an altitude of 1300 and 1900 m a . s . l . and are 24 km and 34 km distant from the closest city , respectively . Their population is 9669 and 10945 inhabitants , respectively . The rural communes of Bouderbala and Sebt Jehjouh in the province of El Hajeb are located at a lower altitude , between 760 and 1000 m a . s . l . They are located 16 km and 15 km , respectively , from the closest city , with a population of 7907 and 7485 inhabitants , respectively ( data from the Haut Commissariat au Plan , Morocco , 2014 ) . These areas were chosen after evaluation of available human and animal data on the presence of CE , and agreement from local authorities . According to official data of the Ministry of Agriculture and Fishery of Morocco , in 2014 the province of Ifrane counted 900 , 000 ovines , with a prevalence of CE infection in this species of 12 . 2% , while the province of El Hajeb counted 330 , 000 ovines , with a prevalence of CE infection of 3 . 5% . These figures are considerably lower than those reported by the Comité Interministeriel de lute contre l’hydatidose/echinococcose ( 2007 ) , an underestimation possibly in part deriving from the fact that abattoirs survey data reflect the age of the animals slaughtered . Indeed , young animals are less infected than older ones , and , in addition , data from abattoirs are cumulative , not differentiating between age categories , therefore inducing bias and underestimation . On average 40 staff members were involved in the implementation of each campaign . Pre-screening activities included two preliminary meetings with all staff involved , inspections to the screening sites , and regular announcement of the screening dates to the population during social and religious occasions ( weekly markets , Mosques , etc ) . The screening activities were carried out in primary school buildings , in each of which eight rooms were used: one for the registration and waiting of participants; one for the health education sessions , one for the administration of the Participant Information Sheet and the risk factors questionnaire , and the signature of the Informed Consent Form; four for clinical examination using eight portable ultrasound machines , and one for blood sampling , medical interview and counselling of patients diagnosed with CE . All residents of the target study areas aged between 10 and 80 years were invited to take part into the screening . However , younger children brought by their parents and older persons responding to the invitation were also examined . The Informed Consent Form was signed by the parent/legal representative for subjects <18 years of age . Males and females were examined by US in separate rooms . All participants were asked to answer a risk factors questionnaire before US examination . The questionnaire was written in French and verbally translated into the local language Amazigh by the survey staff . Questions included recognition of CE cysts upon show of a picture of an infected sheep liver , information on main water source for human use , livestock breeding , home slaughter , disposal of offal , ownership and management of owned dogs ( purpose of dog ownership , dog confinement inside and outside the owner’s premises , feeding habits , deworming with praziquantel , use of same source of water by dogs and livestock ) , and access of unowned dogs to the premises All patients diagnosed with abdominal CE or suspected CE lesions/surgical scars were asked about previous CE diagnosis and treatment , underwent blood sampling for the examination of laboratory parameters and CE serology , and received a chest X ray for the detection of possible pulmonary CE lesions . All women of childbearing age were also assessed for pregnancy using hCG urine rapid test . After receiving detailed information and counselling about the most appropriate clinical management according to the WHO-IWGE Expert Consensus indications , patients were invited to sign the Informed Consent Form to treatment . Subjects <18 years of age were addressed to a paediatric hospital for treatment . Treatment was offered free of charge . Patients were classified as positive for abdominal CE if: i ) they had abdominal lesions with pathognomonic features of CE at US irrespective of their serology results; or if ii ) they had abdominal lesions compatible with CE and positive serology; or if iii ) they had post-treatment lesions from previous CE treatment . In the latter case , where a residual cavity was visualized on US with suspect features for relapse , a diagnostic puncture was proposed and the cyst fluid analysed microscopically and by PCR ( see below ) to define the nature of the lesion . Suspect lesions were investigated with diagnostic puncture , Magnetic Resonance Imaging , and US re-evaluation , as appropriate . All subjects diagnosed with medically relevant conditions other than CE were referred to the reference provincial hospital or regional hospital , as appropriate , for free of charge treatment according to the agreement with the Ministry of Health of Morocco . Serology for echinococcosis was performed using ELISA ( RIDASCREEN Echinococcus IgG , R-Biopharm , Darmstadt , Germany ) and Western Blot ( WB ) ( Echinococcus IgG , LD-BIO Diagnostics , Lyon , France ) according to manufacturers’ instructions . For the analysis , patients with both hepatic and extra-hepatic cysts were classified into 3 groups according to the stage of the hepatic cyst . When more than one hepatic cyst was present , patients were grouped according to the stage of the cyst known to have the most influence on a positive serology result , i . e . CE2-CE3a-CE3b>CE1>CE4-CE5 [15] . To assess the E . granulosus genotypes infecting CE patients , hydatid fluids available after percutaneous or surgical interventions of patients with CE or suspected CE were analysed by PCR according to the method described by Boubaker et al [16] . Before the survey , a 1-day meeting was organized , directed to the medical personnel of the target areas involved in the study . The workshop included 6 hours of frontal lectures on the current techniques of diagnosis and recommendations on the clinical management of CE . Educational material was also provided to the participants . Twenty-four general practitioners practicing in the two target provinces ( 12 per province ) received >18 hours of hands-on training on general abdominal US and focused assessment of CE with US during the screening campaigns , by flanking expert sonographers conducting the screening . None of the physicians had ever received training on CE and ultrasonography before the project . Physicians from El Hajeb specialized in gastroenterology and surgery ( one per specialty ) agreed to receive practical training on surgical and percutaneous CE treatment techniques at the Avicenne ( Ibn Sina ) hospital , Rabat , during the treatment of diagnosed patients . Unfortunately physicians from Ifrane were not available for training during the envisaged period . Educational posters and handouts in Arabic were made available in the places of the US screening to inform the local population on the infection , its transmission , and its prevention . Participants were organized in groups of 30–35 people and received a 15-minute presentation by the Provincial Training Facilitator on Information , Education , and Communication with the aid of simple Power Point slides written in Arabic and explained in the local language Amazigh . Educational material was also provided to the participants .
A total of 5 , 367 people , aged 3–94 years , participated voluntarily in the screening and were evaluated by abdominal US during the 4 campaigns , of which 2 , 705 ( 50 . 4% ) in Ifrane and 2 , 662 ( 49 . 6% ) in El Hajeb . This constituted 1 . 7% of the whole population of Ifrane in 2014 and 13 . 1% of that of the two investigated rural communes in this province; and 1 . 1% of the whole population of El Hajeb and 17 . 3% of the two investigated rural communes in this province ( data from the Haut Commissariat au Plan , Morocco , 2014 ) . During the screening campaigns , 143 health education sessions were performed , for a total of 5249 local people . Of the 5367 people screened , data were available for analysis from 5221 people ( 97 . 3% ) , with results seen in Fig 2 , of which 2633 from Ifrane and 2588 from El Hajeb . The majority ( 70 . 7% ) of screened people were females . The demographic and social distribution of the general resident population and the screened population is detailed in Table 1 . The number of examined people and CE cases found during the screening campaigns are summarized in Fig 2 . Of the 5 , 221 people who received abdominal US , 132 subjects had at least one abdominal CE or suspect abdominal lesion . Of these , 102 subjects ( 1 . 9% [95% CI 1 . 6%-2 . 4%] ) had at least one abdominal CE lesion , either a CE cyst ( 92 . 2% ) , or a residual lesion from previous surgery for abdominal CE ( 7 . 8% ) . CE was excluded in 32 subjects while for 7 people the aetiology of the lesion is still not determined at the time of writing . As shown in Table 2 , the prevalence in the province of Ifrane ( 2 . 6% [95% CI 2 . 0%-3 . 3%] ) was significantly higher than that of El Hajeb ( 1 . 3% [95% CI 0 . 9%-1 . 8%] ) ( p<0 . 001 ) . Using logistic regression analysis , the risk of having CE was significantly higher in the two investigated rural communes of Ifrane province ( OR 2 . 8 [95% CI 1 . 3–5 . 7] in Ain Louh and OR 3 . 3 [95% CI 1 . 6–6 . 7] in Timahdit ) compared to the rural commune of Sebt Jahjouh in the province of El Hajeb ( p = 0 . 005 and 0 . 001 , respectively ) , showing the lowest prevalence , while there was no significant increased risk compared to this area in the second investigated rural commune of Bouderbala in the El Hajeb province ( OR 1 . 8 [95% CI 0 . 9–3 . 9]; p = 0 . 114 ) . The association between CE infection and each risk factor ( Table 3 ) was investigated taking into account province and age and gender , which were not significantly different between CE positive and CE negative subjects ( age p = 0 . 734 and p = 0 . 856 , and gender p = 0 . 896 and p = 0 . 166 in Ifrane and El Hajeb , respectively ) . Province of residence was constantly found associated with high statistical significance with CE infection ( p<0 . 001 in univariate logistic regression model; p<0 . 01 in all multivariable logistic models ) . Dog ownership was found to be associated with borderline statistical significance ( p = 0 . 063 ) with CE infection only in Ifrane province , where the proportion of infected subjects owning a dog was 68% vs 57% of non infected subjects , and statistically associated ( p = 0 . 035 ) with infection when adjusting for age , sex , and province . Other variables associated with borderline significance to infection were livestock breeding in the household ( p = 0 . 098 ) , owned dogs allowed to roam ( p = 0 . 099 , only for Ifrane province ) , and raw viscera given to dogs ( p = 0 . 077 , only in El Hajeb province ) . The proportion of people recognizing CE lesions in pictures of animal infected organs was very high ( 93 . 1% in Ifrane and 57 . 8% in El Hajeb ) . Although the questionnaire was administered after the health education session , and therefore answers may have been influenced by the recently heard information , it is possible they were true recognition of the parasitic lesion , as the absence of abattoirs in these rural communes makes inhabitants exposed to infected organs after home slaughter . The characteristics of CE cases found during the screening campaigns are summarized in Fig 2 . Of the 102 subjects classified as having abdominal CE , 94 ( 92 . 2% ) had CE cyst in abdominal organs , while 8 ( 7 . 8% ) patients had only residual lesions from previous surgery for abdominal CE . None of the patients with abdominal CE had lung infection , as assessed by X ray of the chest . Twenty-three patients with suspect CE lesions were excluded from having the parasitic infection after re-examination , as appropriate . The alternative diagnoses were biliary cysts ( n = 15 ) , haemangioma ( n = 1 ) , hepatocellular carcinoma ( n = 1 ) , other kidney diseases ( n = 4 ) , and absence of lesions at re-evaluation ( n = 2 ) . Of the 94 patients with abdominal CE cysts , 68 ( 72 . 3% ) did not know they were infected , and were therefore newly diagnosed . The remaining 26 ( 27 . 7% ) patients already knew about their condition . Infected patients were symptomatic in 47 . 9% of cases; the most frequent reported symptom was abdominal pain ( 91 . 7% ) . Of the 32 patients with a previous history of treatment for CE , 27 ( 84 . 4% ) were treated surgically and 5 ( 15 . 6% ) received medical treatment with albendazole . Nobody reported previous percutaneous treatment . Of these previously treated CE patients , 11 ( 34 . 4% ) had active cysts on US examination , but unfortunately it was not possible to assess whether these were new infections or relapses after treatment , due to the lack of medical documentation . The prevalence and distribution of CE lesions by age group and gender is shown in Fig 3A . The 94 patients with abdominal CE cysts had a total of 131 CE cysts ( mean 1 . 4 CE cysts per patient; range 1–8 ) . Of these , 84 ( 89 . 4% ) patients had CE cysts only in the liver; 3 ( 3 . 2% ) in the liver and another localization , notably the peritoneum in 2 patients and the spleen in 1 patient; and 7 patients ( 7 . 4% ) had CE cysts only in extra-hepatic locations , notably the peritoneum in 3 patients , the spleen in 3 patients , and the kidney in 1 patient . The distribution of hepatic CE cysts by stage is shown in Fig 3B . Liver CE cysts were most frequently inactive ( 55 . 8% were CE4-CE5 stages ) , followed by stages CE3b ( 19 . 2% ) , CE1 ( 13 . 3% ) , CE2 ( 8 . 3% ) and CE3a ( 3 . 3% ) . As CE has a slow progressive evolution and the exposure may vary due to gender and age-related activities , we investigated the distribution of cyst stages according to these variables . As it was not possible to discriminate between relapse and reinfection in previously treated patients , the distribution of CE stages by sex and age was analysed here only in the 70 previously untreated patients ( 68 newly diagnosed and 2 known infected but never treated patients ) . We found that the relative frequency of inactive cysts increased with age , while that of active cysts decreased with age . The risk of having an active cyst significantly decreased with age ( p = 0 . 003 OR 0 . 33 CI 0 . 16–0 . 69 ) . Results are shown in Fig 3C and 3D . Of the 29 patients who had been previously treated for abdominal CE , 17 ( 58 . 6% ) had only residual cavities/scars or inactive CE4-CE5 cysts on US examination ( 7 [24 . 1%] only residual cavities/scars; 10 [34 . 5%] inactive cysts ) , while 12 ( 41 . 4% ) had cysts in active stage . However , as stated above , it was not possible to discriminate between relapse and reinfection . Sera from patients with CE were tested by ELISA and WB and the results analysed by cyst stage group as detailed in the methods . Of the 32 patients previously treated for CE , 78 . 2% had positive serology on both ELISA and WB . Of note , among these previously treated patients , only 1 of the 8 patients having just residual cavities/scars from previous surgery for CE was seronegative , and one seronegative patient had a CE3b cyst in the spleen . Of the 70 previously untreated patients , 21 ( 30% ) were seronegative on both ELISA and WB , 6 ( 8 . 6% ) had only positive ELISA serology , and 41 ( 58 . 6% ) were seropositive with both tests . For 2 patients data regarding serology were not obtained . The percentage of ELISA positive results according to cyst stage in untreated and previously treated patients is shown in Fig 4 . Of note , only 1 patient was receiving albendazole treatment at the time of blood sampling . One ( 4 . 3% ) of people that were excluded from having CE had ELISA and WB positive results . Cyst fluid was obtained from 13 lesions and evaluated by microscopy and PCR . Three suspect lesions were excluded from being CE upon the result of negative microscopy and PCR; of note , one of these patients had a suspect liquid cyst together with a CE5 cyst and positive serology , therefore the non-parasitic nature of the liquid cyst could have not been determined by the serology result only . The negativity of microscopy and PCR of aspirated liquid from 4 lesions of patients with a previous history of surgery and images suspect for relapse allowed to exclude relapse and classify the lesions as residual cavities . All the remaining 6 cyst fluids had protoscoleces on microscopy . Of these , 5 were also positive by PCR and identified as genotype G1 ( E . granulosus sensu stricto; sheep strain ) .
CE is endemic in Mediterranean countries , including Morocco , however its real prevalence , incidence and burden are difficult to estimate . This is due to the uneven distribution of transmission areas in endemic countries , the high proportion of asymptomatic infected individuals and symptomatic patients living in resource-poor areas with logistical and/or economic constraints , who never reach medical attention , and the underreporting of diagnosed cases . Furthermore , commonly used measures such as surgical case incidence are not appropriate to evaluate the dynamics of a chronic and clinically complex infection such as CE . A comprehensive evaluation of infection and disease burden , and of transmission risk factors , is at the basis of the decision , by public health authorities , upon the implementation of control programmes and rationalization of diagnosis and treatment recommendations . Population US surveys using portable and relatively inexpensive scanners allow obtaining more comprehensive , accurate and detailed information on infection prevalence and stage distribution of CE . The finding of early active cyst stages arguably reflects transmission pressure . Furthermore , US is non-invasive and repeatable , thus it can be used to monitor the effectiveness of control interventions [17] . Training of local physicians on focused assessment with US of CE and the rational allocation of infected patients to clinical management options allow for the reduction of costs associated with the need of travel to tertiary care facilities and provide an efficacious and less expensive tool for patients and health care systems [13] . Finally , community-based US surveys may constitute a useful educational activity , raising awareness of the importance of the infection in the population living in endemic areas [14] . CE is endemic throughout the Mediterranean , which is an area of intense migration , and infected patients may be diagnosed long after infection and in a different country from where it was acquired . Recent data from the WHO Collaborating Centre on Clinical Management of CE in Pavia , Italy , showed that 38 . 2% of the 203 patients with CE followed by the Centre between January 2012 and February 2014 were foreign-born , and Morocco was the country of birth of the majority ( 27 . 2% ) of these patients [18] . In this work , carried out through an Italian-Moroccan partnership , we estimated the prevalence and the characteristics of human abdominal CE by means of a community-based US screening in the provinces of Ifrane and El Hajeb , Meknes-Tafilalet region , Mid Atlas , an area among the most endemic in Morocco for both human and animal infection . In four 2-day campaigns using 8 portable US machines operated by experienced clinicians , we screened 13 . 1% of the population of the two investigated rural communes of Timahdit and Ain Louh in Ifrane and 17 . 3% of the two investigated rural communes of Bouderbala and Sebt Jahjouh in El Hajeb . This study had several limitations . First , this sample population included a lower percentage of young ( <20 years of age ) people and a higher percentage of middle-aged ( 40–60 years ) people compared to the general registered population of the investigated provinces; also , more females than males volunteered to participate to the survey . Second , the screened sample was self-selected due to the voluntary nature of the participation , which could have biased the estimate in either direction . A random sampling was not considered suitable here due to concerns about the acceptability of such sampling method by the population . Third , children younger than 10 years of age were excluded from the target population due to practical constraints regarding treatment of CE in this age range population . We found an overall CE prevalence of 1 . 9% ( CI 1 . 6%-2 . 4% ) , a figure almost double compared to what reported by Macpherson and colleagues [6] , who carried out a population US survey in the provinces of Ifrane and Khenifra , Mid Atlas , in 2000–2001 , observing a prevalence of 1 . 1% ( CI 0 . 9%-1 . 3% ) . This discrepancy could be due to a real increase in the infection pressure in the absence of control measures , or derive from differences in the investigated areas and target populations . Indeed , CE infection burden is difficult to estimate also due to the uneven distribution of transmission even within relatively small areas . Although a direct comparison of the two surveys is therefore not possible , these results demonstrate , as expected , an on-going transmission of the infection in the region . In our survey , CE prevalence in Ifrane was twice that in El Hajeb , with no differences between rural communes within provinces . The analysis of the prevalence of infection by age and gender showed the highest values in males aged 21–40 years , with comparable values in males and females within age groups . These results are in line with what found by Macpherson and colleagues [6] . However , we observed a decrease in infection prevalence in the population aged > 40 years . This was not explained by an increase in treatment rate with age , as prevalence was calculated including both subjects untreated and previously treated for CE . Possible explanations may be related to the structure of the resident population ( e . g . more people aged 40–60 years may be emigrated from rural areas ) and/or to the spontaneous resolution of the infection with time , with disappearance of the lesions possibly combined with the acquisition over time of some protective immunity to new infections [19] . When we investigated risk factors using multivariable analysis , we found that only province of residence ( Ifrane ) and dog ownership were significantly associated with CE infection , while age and gender were not . Of note , Ifrane is the province with more extensive livestock breeding activity and meat production ( i . e . slaughter activity ) . These results are different from previous work , where dog ownership was generally not associated with CE infection in studies investigating adults or communities , while gender and source of water were consistently associated with CE infection [20–24] . These results , however , are consistent with those reported in the recent systematic review and meta-analysis of potential risk factors associated with CE infection [1] . In the investigated area , “dog ownership” may be not intended in the same strict manner as in Europe . Indeed , from the analysis of the answers to the questionnaire , we noticed that at-risk practices such as home slaughter , unsafe disposal of livestock viscera in public places ( e . g . garbage or open fields accessible to dogs ) , and feeding of dogs with raw viscera , were carried out independently of the strict ownership of livestock and dogs . Possibly , a more precise investigation of the type of contact with dogs rather than the general terminology of “dog ownership” could be more informative on the role of direct contact with dogs in the transmission of CE . Our result that province of residence was the only risk factor consistently associated with CE infection suggest that environmental contamination is likely the main factor responsible for CE transmission in this area . A more detailed assessment of habits and analysis of materials ( water , soil , food ) would be necessary to try individuating the actual route of infection , as also suggested by Possenti et al [1] . However , the acquisition of infection a long time before diagnosis makes the evaluation of such causality very difficult . When considering CE infection in previously untreated patients , we observed that CE cysts were predominantly in an inactive stage , supporting the findings of previous longitudinal and observational studies showing that cysts evolve spontaneously to inactivation over time [25–30] . However , active cyst stages , including CE1 cysts ( i . e . cysts likely acquired in recent times ) , were present also in adult age groups , as also found by Macpherson and colleagues in the same area [6] , indicating acquisition of infection even in adulthood , although at lower rate . This may be due to the acquisition of some degree of immunity or to a decrease in exposure to infection at older age . In any case , these results are of particular importance . First , the predominance of inactive cysts in the sample population and the increase of the presence of inactive cysts with increasing age support the need for a stage-specific approach . Indeed , most cysts evidently evolve spontaneously to inactivation . As a consequence , aggressive invasive therapy , when not needed because of symptoms or complications , is not appropriate in most cases , and spontaneously inactivated cysts ( the majority of those found in our population ) do not have to be treated at all [8 , 9] . Second , the evidence of acquisition of infection even in adulthood implies that a benefit from control programmes may be observed in all age groups , as already observed by Beard [31] . Indeed , the assumption that most cysts are acquired at early age but only evident after years hindered control measures , as this implied that investments for decades were needed before visible results may be obtained [31] . On the contrary , the practical consequence of the finding that adults are also susceptible and that latency between infection and diagnosis may be shorter than believed , is that expenditures for control may be encouraged by the expectation of early measurable benefits to the whole community . To conclude , our results show a high prevalence and on-going transmission of CE in the investigated provinces of Ifrane and El Hajeb , Mid Atlas of Morocco . A plan for a control program in Morocco was envisaged after the study on the incidence of hospitalized CE cases at the national level ( 1980–1992 ) carried out by the Ministry of Health . In 2003 a national register of hospital CE cases was implemented . In 2004 , an inter-ministerial committee for the control of CE was established in Morocco . The activities envisaged in the control program included health education , improvement of general hygiene , strengthening of the case registration system , control of the stray dog population , and screening and early treatment of patients . The intersectoral strategy was developed in 2007 and the activities started in 2009 , but so far only improvement of human CE case report system has been implemented . Our data confirm the need for control activities in the area by national health authorities , through the full implementation of envisaged activities , and possibly the inclusion of other measures [32] after revision of the current plan , in the light of the difficulties and constraints to the implementation of such control programs . Also , our results encourage the acceptance and use of diagnostic-based algorithms using imaging rather than serology , and a stage-specific management approach for CE . Indeed , with presently available tools cyst activity cannot be assessed by serology , as more than 50% of subjects with spontaneously inactivated cysts ( which remain stably inactive in >97% of cases [9] ) and more than 85% of subjects with post-surgical residual cavities/scars had a positive serology in our population . These patients should be correctly diagnosed as not having an active infection , and should be solely monitored over time , avoiding expensive and risky treatments such as surgery [8 , 9] . Focused training of local physicians on US in CE diagnosis , management options , and follow-up would be a valuable tool . Indeed , in Morocco , surgery is still almost the sole treatment option offered to patients [5] , as also shown by our results of treatment type reported by previously treated patients , and compliance to current recommendations on the clinical management of CE is still very low [10 , 12] . Our experience also shows that population-based screening campaigns are useful to assess the prevalence , dynamics , and risk factors of CE in endemic areas , and may provide the favourable occasion to implement focused training for local health care workers and health education for general population . In this study , children younger than 10 years of age were excluded from the target population due to practical constraints regarding treatment of CE in this age range population . However , the inclusion of a younger population would be important in studies preliminary to the implementation of control activities , to provide baseline data , and for the monitoring of infection incidence once the program is in ongoing . A highly coordinated multidisciplinary team is pivotal and a systematic pre-survey planning is absolutely required to address research questions and implement operational activities on an infection such as CE whose management is the exemplification of the "One Health" approach . However , the building and coordination of such teams is difficult , and requires time and continuous work over time with the same group . Also , the individuation and agreement on the relative share of costs and coordination between human and veterinary health services is an issue [33] . In this regard , it would be desirable to include all or part of the same staff in future surveys , to take advantage of the experience gained and lessons learned from previous activities through the identification of critical points after completion of field work ( e . g . here the administration of the education session before the questionnaire for logistical convenience posed then problems in the interpretation of answers to some questions of the questionnaire ) . A pilot testing of the risk factors questionnaire and the inclusion of a social scientist in the team are also advisable to better adapt epidemiological tools to the habits and structure of the target communities , improve attendance from all age groups and genders and advise on the social acceptability of different sampling methods [34] . | Cystic Echinococcosis ( CE ) is a parasitic infection whose natural domestic cycle develops between dogs and sheep ( and other livestock ) . Human infection is endemic in pastoral communities , where close contact with the dog-sheep cycle occurs . In humans , as well as in livestock , the parasite develops as fluid-filled cyst mainly in the liver . CE is a neglected disease , as it is a disabler without high mortality , it affects mainly poor communities , and requires complex and expensive clinical management and long-term integrated public health control strategies . The prevalence of infection in an area is often unknown or largely underestimated , therefore the problem is perceived as not important . In Morocco , the Mid Atlas is the most prevalent area for human and animal infection . We performed an ultrasound survey on 5 , 367 people in Ifrane and El Hajeb provinces , and found an overall prevalence of 1 . 9% . CE cysts were predominantly inactive , however , active cysts were present also in adults . Our results show a high prevalence and on-going transmission of CE , encouraging the prompt strengthening and complete implementation of control activities envisaged by Moroccan health authorities in the area and the adoption of diagnosis and treatment algorithms based on imaging at both national and local level , to avoid a risk-associated and expensive treatment of inactive cysts . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"animal",
"types",
"livestock",
"morocco",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"tropical",
"diseases",
"geographical",
"locations",
"vertebrates",
"parasitic",
"diseases",
"dogs",
"mammals",
"animals",
"pets",
"and",
... | 2017 | Human cystic echinococcosis in Morocco: Ultrasound screening in the Mid Atlas through an Italian-Moroccan partnership |
Parasitic diseases caused by kinetoplastid parasites of the genera Trypanosoma and Leishmania are an urgent public health crisis in the developing world . These closely related species possess a number of multimeric enzymes in highly conserved pathways involved in vital functions , such as redox homeostasis and nucleotide synthesis . Computational alanine scanning of these protein-protein interfaces has revealed a host of potentially ligandable sites on several established and emerging anti-parasitic drug targets . Analysis of interfaces with multiple clustered hotspots has suggested several potentially inhibitable protein-protein interactions that may have been overlooked by previous large-scale analyses focusing solely on secondary structure . These protein-protein interactions provide a promising lead for the development of new peptide and macrocycle inhibitors of these enzymes .
Infections caused by the kinetoplastid parasites Leishmania spp . , Trypanosoma brucei , and Trypanosoma cruzi are estimated collectively to put at risk one billion people , resulting in tens of millions of infections and upwards of ten thousand deaths per year [1] . Neglected tropical diseases ( NTDs ) caused by these parasites primarily occur in the developing world and are infrequently the target of commercial drug-development efforts [2] . A number of highly conserved enzymes are present across these pathogenic species , despite substantial genomic diversity [3] . Furthermore , the proliferation of high-resolution crystallographic data affords the opportunity to identify new mechanisms for inhibiting both established and emerging drug targets in these organisms . Recent drug-repurposing efforts have allowed for the development of promising new leads based on previous work on homologous targets , such as kinases and heat-shock proteins , in human diseases [4 , 5] . Just as neglected tropical diseases have received comparatively little attention from the drug discovery community , so too have protein-protein interactions ( PPIs ) , which are characterized by larger surface area and lower binding affinity than is typical for drug-like molecules [6 , 7] . A substantial fraction of the protein-protein interaction energy is localized in a few amino acid residues , known as “hot spots , ” which are often surface-exposed hydrophobic amino acid residues [8] . Computational alanine scanning can generally predict these interface hot spots with a 79% success rate [9] . This has led to the successful development of several inhibitors of PPIs [10–12] . Of greatest relevance to NTDs , this approach has been applied to inhibition of the cysteine protease cruzain , based on the interaction with its native inhibitor chagasin [13] . Targeting PPIs of multimeric enzymes [14 , 15] in these pathogens , by avoiding the highly conserved substrate-binding domains , should allow for fine-tuning selectivity to avoid inhibition of the homologous host enzymes [15] . This approach has been successful in PPI-based inhibition of the homodimeric enzyme , triosephosphate isomerase ( TIM ) , in P . falciparum [14] and T . cruzi [16] . Thus , a systematic analysis of these overlooked targets for neglected diseases may reveal both new drug targets and new approaches to inhibit well-established targets .
Structures of multi-protein complexes from the family Trypanosomatidae were obtained using the advanced search functionality of the Protein Data Bank [17] . Structures with >4 Å resolution or >90% similarity were excluded . The PDB files were cleaned to remove headers , retaining only ATOM line entries , using a shell script . Computational alanine scanning [9] was performed using Rosetta 3 . 6 and PyRosetta [18] , with a modified version of the alanine-scanning script originally developed by the Gray lab [19] . The updated Talaris2013 scorefunction [20] was parameterized to match an established general protocol [9 , 21] without environment-dependent hydrogen bonding terms . Default score function weights were retained , but line 129 of the script was replaced as follows to implement these changes: Interfaces that were determined to have at least three hot spots ( ΔΔG ≥ 1 . 0 Rosetta Energy Units ( REU ) , average of 20 scans , 8 . 0 Å interface cutoff ) by this method were further examined for proximity of the hot spots in both primary [22] sequence and secondary/tertiary structure . Complexes with at least two hot spots in close proximity were cross-checked for presence in existing databases of helix [23 , 24] and loop [25 , 26] interaction motifs , then with existing literature for experimentally verified interface hot spots , and finally for identity as an established or emerging drug target [27–31] . Amino acid residues falling just below the threshold ( ΔΔG between 0 . 8 and 1 . 0 REU ) were also considered when proximal to multiple interface hot spots . During the preparation of this manuscript , the authors became aware of the Peptiderive server [32] , which allows for the rapid examination of single PPI interfaces for hot-spot rich segments of a defined length . The PPIs identified in this study were subsequently re-examined using Peptiderive to locate decameric “hot segments” for comparison .
Of the 1 , 076 kinetoplastid protein structures deposited in the PDB , 207 are multi-chain biological assemblies . Computational alanine scanning identifies 56 structures containing at least three putative interface hot spots ( 27% ) . Hot spots are defined as any amino acid residue that , when mutated to alanine , increased the ΔΔGcomplex by at least 1 . 0 REU [33] , a threshold that generally has a 79% correspondence with experimentally observed hot spots [9] . Among these 56 structures , 46 contain multiple hot spots on the same helix ( 27 ) or loop ( 19 ) . Despite the 90% sequence identity cutoff , several homologous proteins from Leishmania spp . , Trypanosoma spp . , and the non-pathogenic model organism Crithidia fasciculata appear multiple times ( vide infra ) , reducing the number of unique interfaces to 34 . Analysis of these 34 complexes reveals 12 unique PPIs that are either established drug targets in T . cruzi , T . brucei , or Leishmania spp . or essential enzymes and structurally or functionally obligate multimers . Drug targets with inhibitable PPIs , their potentially inhibitory peptide sequences , and a comparison to HippDB , Loopfinder , Peptiderive , and experimental results are listed in Table 1 . Five targets are involved in the redox metabolism of trypanothione , an essential pathway for the parasites’ antioxidant defense that has been the target of numerous drug development efforts [36] . Central to this pathway is Trypanothione reductase ( TryR ) , an essential enzyme which maintains trypanothione , T ( SH ) 2 , in the reduced state . T ( SH ) 2 is produced from glutathione ( GSH ) , which is both synthesized de novo by glutathione synthetase ( GS ) and scavenged extracellularly from the host . TryR utilizes NADPH as a reductant , produced primarily from the pentose phosphate pathway ( PPP ) by glucose 6-phosphate dehydrogenase ( G6PDH ) , which itself is induced by the presence of hydrogen peroxide , and 6-phosphogluconate dehydrogenase ( 6PGDH ) [37] . Several enzymes use T ( SH ) 2 to detoxify specific reactive oxygen species , including tryparedoxin peroxidase ( TXNPx ) , which reduces hydroperoxides produced by the host’s immune response . Trypanothione reductase is a well-established drug target , with almost all known inhibitors targeting the active site through covalent inactivation of the catalytic cysteine residues or binding of polycationic species in the active site [2 , 30 , 38 , 39] . Recent computational and experimental studies have identified a hot-spot-containing helix in L . infantum TryR that inhibits TS2 reduction by disrupting dimerization of the enzyme , as demonstrated by kinetics and ELISA [34 , 40] . This helix overlaps , but shares little sequence homology with , a helix that disrupts dimerization of human glutathione reductase ( hGR ) , although it presents a strikingly similar helical face ( S1 Fig ) . The hGR peptide prevents refolding of denatured hGR , but does not inhibit the activity of the native enzyme [41] , minimizing the possibility of hGR inhibition from an isosteric TryR inhibitor . Beyond this known mode of inhibition , this study identified a short helix-breaking loop in C . fasciculata ( Fig 1A ) , T . brucei , T . cruzi , and L . infantum TryR containing three hotspots , Ile-72 , Phe-78 , and Leu-82 , which matches a loop in LiTryR identified by Loopfinder ( heat score: 3 ) , overlaps a segment predicted by Peptiderive ( 22% interface energy ) , and contains one hotspot ( Trp-80 ) that has been verified experimentally ( Fig 1 ) [34] . This proposed inhibitory peptide is predicted to contribute more to the interface energy than the established helix-based inhibitor . Glutathione synthetase is the second step in de novo synthesis of TS2 , producing glutathione from γ-glutamylcysteine , glycine , and ATP . Knockout of GS in T . brucei results in a growth-restriction phenotype that is not rescued by addition of exogenous glutathione , suggesting that GS may be druggable [43] . Initial structural characterization of T . brucei GS had suggested that the high homology in the regions involved in substrate and cofactor binding and catalysis would make GS a suboptimal drug target [44] . However , the helix identified in this study differs greatly from the human homolog in both primary sequence and secondary structure . While the proposed inhibitory sequence , 2-VLKLLLEL , contains only two putative hot spots , Leu-3 and Leu-7 , recombinant TbGS with an N-terminal His6 tag has drastically reduced catalytic turnover [45] , suggesting the importance of the N-terminus in dimerization and providing a plausible route to selective PPI-based inhibition of TbGS in this region . Both HippDB ( 17% ) and Peptiderive ( 22% ) identified segments near the N-terminus predicted to contribute substantially to the stability of the PPI . Tryparedoxin peroxidase catalyzes the TS2-dependant detoxification of peroxides , is an essential enzyme in T . brucei and L . major , and has been proposed as a potential drug target due to the constitutively high levels of peroxide , especially in T . brucei [46–50] . TXNPx is an obligate homodimer , with catalytic cysteine residues for a single active site located on both subunits [48] . This study identifies an unstructured loop containing three hot spots , Asn-145 , Val-149 , and Arg-151 , in L . major , T . brucei , and C . fasciculata ( Ile-149 ) in a hydrophobic cleft on the surface . This peptide matches a loop identified by Loopfinder ( heat score: 9 ) and a segment identified by Peptiderive ( 27% interface energy ) . HippDB contains one short , two-turn helix , predicted to contribute only 9% interface energy ( Fig 1 ) . Glucose-6-phosphate dehydrogenase catalyzes the first reaction in the pentose phosphate pathway , oxidizing glucose-6-phosphate to 6-phosphogluconolactone and producing NADPH required for TS2 reduction [37] . G6PDH is both an essential enzyme and a validated drug target [51] , in addition to being catalytically active in both dimeric and tetrameric forms . Inhibition by peptide-based PPI disruption has been successfully applied to human G6PDH [35] . This study identifies a homologous loop from T . cruzi , 441-AMYLKLTAKTPGLLNDTHQTEL , containing three hot spots , Met-442 , Leu-446 , and Leu-462 , which are tightly clustered on adjacent strands of a beta sheet in a hydrophobic cleft ( Fig 1 ) . This suggests that this approach may find similar success in trypanosomatids with a carefully designed , smaller macrocyclic peptide . Moreover , Peptiderive identifies a decameric segment predicted to contribute 38% of interface energy , which , when extended to include an adjacent two-turn helix , is expected to contribute >50% of interface energy by Loopfinder ( heat score: 4 ) . Considering the successful inhibition of hG6PDH and a second predicted inhibitory peptide , G6PDH presents a logical opportunity to explore PPI-based inhibition . 6-phosphogluconate dehydrogenase catalyzes the third step in the pentose phosphate pathway , converting 6-phosphogluconate to ribulose 5-phosphate and CO2 . G6PDH and 6PGDH are the primary source of NADPH for the reduction of TS2 [37 , 52 , 53] . This essential enzyme has a highly conserved sequence identity between T . cruzi , T . brucei , and L . major , yet differs substantially from the human 6PGDH homolog , making it an ideal drug target [54] . Substrate analogs have shown potent inhibition of 6PGDH and trypanocidal activity in the low micromolar range [55] . Substrate binding involves residues from both protomers , suggesting PPI disruption may also be a viable inhibition strategy [56] . This study identifies a loop in T . brucei 6PGDH , 251-LTEHVMDRI , containing three hot spots , Asp-253 , Val-255 , and Ile-259 . HippDB , Loopfinder , and Peptiderive all identified peptides immediately surrounding a helix , 445-YGQLVSLQRDVFG , predicted to contribute 10–13% of the interface energy . The remaining seven targets represent a variety of essential metabolic and biosynthetic processes . Two targets emerged in sugar metabolism: ribose-5-phosphate isomerase B ( RpiB ) in the non-oxidative branch of the PPP and UDP-glucose-4’-epimerase ( GalE ) in galactose catabolism . Three other targets are involved in varied essential biosynthetic processes: Farnesyl pyrophosphate synthase ( FPPS ) in the isoprenoid biosynthetic pathway , tyrosine aminotransferase ( TAT ) in tyrosine catabolism , and pteridine reductase ( PTR1 ) in cofactor biosynthesis . Finally , two essential targets are found in nucleotide synthesis: deoxyuridine triphosphate nucleotidohydrolase ( dUTPase ) and dihydroorotate dehydrogenase ( DHODH ) . Ribose 5-phosphate isomerase B catalyzes the interconversion of D-ribose-5-phosphate and D-ribulose-5-phosphate in the non-oxidative branch of the pentose phosphate pathway . RpiB is essential for viability of the bloodstream form of T . brucei and is a subtype with no mammalian homologue [57–59] . RpiBs are functionally obligate dimers , with catalytic residues spanning both subunits , suggesting that targeting the RpiB interface may a viable inhibition strategy . The peptide sequence from T . cruzi RpiB identified in this study overlaps a helix predicted by HippDB to contribute 49% of the interface energy . This helix , 140-RRIEKIRAIEASH , contains two predicted hot spots , Ile-145 and Ile-148 , on adjacent turns of the helix and the indispensable residue Glu-149 . Experimental mutation of this residue disrupts both structure and function in LdRpiB [60] . This helix is also immediately C-terminal to His-138 , another deactivating mutant ( Fig 2 ) . All four amino acid residues are conserved between T . cruzi and L . donovani , suggesting the generality of a peptide helix-based inhibitor . UDP-galactose-4’-epimerase is an essential enzyme for the growth and survival of trypanosomatid parasites [61] . Unable to acquire galactose from the host , they rely on GalE to synthesize galactose from glucose [61 , 62] . T . brucei GalE has 33% homology to the human enzyme [63] , and thus has received substantial attention as a target for trypanocidal drugs . Several small-molecule inhibitors have been identified , mainly targeting the active site of the enzyme [62 , 64 , 65] . Additionally , GalE is only fully functional as a dimer [61 , 62] , suggesting that the interface of this enzyme is potentially druggable . This study identifies a helix , 111-PLKYYDNNVVGILRLL , with two hotspots , Val-119 and Ile-123 , on adjacent turns of the same buried helical face of T . brucei GalE , overlapping sequences identified by both HippDB ( 33% interface energy ) and Loopfinder ( heat score: 9 ) ( Fig 2 ) . Farnesyl diphosphate synthase , a key enzyme in sterol biosynthesis , catalyzes the sequential condensation of isopentenyl diphosphate and dimethylallyl diphosphate to form geranyl diphosphate and ultimately farnesyl diphosphate , which is the obliged precursor for the biosynthesis of sterols , ubiquinones , dolichols , heme A , and prenylated proteins [66] . Recently , FPPS has been validated as a drug target [67] and the sterol biosynthesis pathway has been targeted at numerous other steps [68] . Most established inhibitors are bisphosphonate substrate mimics; however , they are commonly associated with poor drug-like characteristics [69–72] . FPPS is a functionally obligate homodimer , with the active site located at the protein-protein interface [66 , 73] . This study identifies a turn between two helices , 25-FDMDPNRVRYL containing three hotspots , Phe-25 , Tyr-34 , and Leu-35 , in T . brucei FPPS [66] . This same segment is identified by both Peptiderive ( 15% interface energy ) and Loopfinder ( heat score: 3 ) . Tyrosine aminotransferase , which is involved in the first step of amino acid catabolism , catalyzes transamination for both dicarboxylic and aromatic amino-acid substrates [74] . Structural studies suggest that TAT is only fully functional in the dimeric state [75] . TAT is overexpressed in T . cruzi from patients with acute Chagas [76] and associated with resistance to oxidative damage . This study identifies a three-turn interface helix , 54-AQIKKLKEAIDS , in T . cruzi and L . infantum TAT with two proximal hotspots , Leu-59 and Ile-63 , on adjacent turns , presenting a hydrophobic face buried in the opposite protomer . This is in contrast to the only helix found in HippDB , 275-PSFLEGLKRVGMLV ( 15% interface energy ) , which interacts primarily with the domain-swapped , N-terminal 15 amino acids . Pteridine reductase , a short-chain reductase , participates in the salvage of pterins , for which trypanosomatids are auxotrophic [77] . PTR1 catalyzes the NADPH-dependent two-stage reduction of oxidized pterins to the active tetrahydro-forms and provides an alternate pathway for folate reduction , allowing de novo thymidylate synthesis to occur even in the presence of methotrexate [77 , 78] . PTR1 is essential in T . brucei and has been targeted in numerous small-molecule efforts [79–82] . The enzyme is a functional tetramer with substantial surface contacts between the A chain and B and C chains [79 , 83] , suggesting the viability of targeting the PPI . This study identifies six hotspots on helix 5 of L . major PTR1 , with hotspots clustered in hydrophobic pockets at the N-terminal ( Thr-192 and Met-196 ) and C-terminal ( Leu-210 , Glu-211 , Leu-212 , and Leu-215 ) ends of an otherwise convex surface at the A-B interface ( Fig 2 ) . The C-terminal portion of this helix is predicted by Peptiderive to contribute 24% of the A-B interface energy and is positioned to mediate the A-C interaction as well . This same helix was identified by HippDB ( 67% A-B interface energy ) and contains two key catalytic residues , Tyr-194 and Lys-198 , mutation of which inactivates PTR1 [84] . Loopfinder identified a complementary loop ( heat score: 9 ) that appears to contribute substantially to the A-C interaction . Deoxyuridine triphosphate nucleotidohydrolase is necessary for both DNA repair and de novo synthesis of dTTP . It converts dTUP to dUMP and pyrophosphate . dUTPase maintains a high ratio of dTTP:dUTP , preventing accidental incorporation of uracil into DNA [85 , 86] . This enzyme was shown to be essential in L . major and T . brucei with decreased proliferation in both the procyclic and bloodstream forms of the organism [85 , 87] . T . cruzi dUTPase , an obligate dimer , shows little homology to the human counterpart , which is a functional monomer , contributing to its potential as a drug target [86 , 87] . However , among trypanosomatids , the interface residues are highly conserved [85] . This study identifies an unstructured loop on the interface of L . major dUTPase , 51-ELLDSYPWKWWK , with two hotspots , Leu-53 , Trp-58 , in close proximity . An overlapping segment was identified by Peptiderive ( 35% interface energy ) . Trp-58 , Trp-60 , and Trp-61 are buried in a deep hydrophobic cavity on the opposite protomer , although only the former was identified by this computational alanine scan . Dihydroorotate dehydrogenase catalyzes the oxidation of dihydroorotate to orotate in the de novo pyrimidine biosynthetic pathway [88] . The highly conserved DHODHs found in trypanosomatids bear less than 20% sequence homology to the analogous human enzyme [89 , 90] . DHODH knockout studies demonstrated that the protein is essential in T . cruzi and an obligate dimer [89 , 90] , suggesting that DHODH would be an ideal drug target in trypanosomatids . This study identifies a long , unstructured loop in T . brucei DHODH , 202-VIDAETESVVIKPKQGFG , containing three hotspots , Ile-203 , Val-210 , and Phe-218 , tightly clustered in a hydrophobic groove . Both Loopfinder ( heat score: 3 ) and Peptiderive ( 23% interface energy ) identify overlapping portions of this peptide , suggesting an ideal starting point for the development of macrocyclic inhibitors . In the past decade , inhibition of PPIs has evolved from the short , primary epitopes exemplified by RGD-peptide-like integrin antagonists and AVPI-peptide-like Smac mimetics to include clinically relevant molecules that recapitulate increasingly complex secondary and tertiary structures like those presented by the BCL family and IL-2 , respectively [12 , 91] . The diversity of topologies , interaction motifs , and binding affinities at these interfaces presents an intriguing challenge for the development of new PPI inhibitors [7] . PPI-based inhibition of NTD targets has achieved some pre-clinical successes , including non-peptide inhibitors of the cysteine protease cruzain [92 , 93] , and interface-peptide-derived inhibitors of triosephosphate isomerase [14 , 16] . Ultimately , this analysis identified solely homomultimeric enzymes . This is likely due to the bias of existing structural data towards these types of targets , which have received substantial attention as targets for structure-based design of small-molecule inhibitors [28 , 31 , 94] . Of the 207 multi-chain trypanosomatid crystal structures in the PDB , 148 ( 71 . 5% ) are for enzymes ( Table 2 ) . Nevertheless , two of the 12 targets identified in this study have been successfully inhibited by interface-derived peptides . An interface-derived peptide helix has been demonstrated to inhibit LiTryR through a mechanism that disrupts the PPI [34 , 40] . This helix had also been identified by HippDB as potentially contributing 15% of the interface energy , while the loop region identified in this analysis is predicted to contribute 22% . Similarly , the G6PDH interface peptide matches a homologous region in the human enzyme , which has been successfully developed into an inhibitory peptide [35] . Given the proximity of the hot spots in space rather than sequence , it appears amenable to inhibition by a macrocycle or peptidomimetic . Overall , PPI-based inhibition of multimeric enzymes [14 , 15] represents a complementary , but underutilized , approach to these targets . The interface peptides identified in this study predominantly contain hot-spot amino acids with aliphatic side chains ( Leu , 29%; Ile , 24%; Val , 12% ) and Phe ( 9% ) . Surprisingly [95–97] , other aromatic amino acids ( Tyr , 3%; Trp , 3% ) appear to be underrepresented in this analysis . These percentages do not differ substantially from the hot spots found over the entire interface ( Leu , 30%; Ile , 16%; Val , 16%; Phe , 13%; Tyr , 3%; Trp , 2% ) . Bogan and Thorn observed a general enrichment of Trp , Tyr , and Arg at interface hot spots [98]; the Loopfinder dataset observed enrichment of Trp , Phe , His , Asp , Tyr , Leu , Glu , and Ile in hot loops [25] . This contrast is most apparent when examining specific PPIs in this study . In the case of TryR ( Fig 1 ) , an experimentally verified hot spot ( Trp-80 ) [34] was not identified by the computational alanine scan , despite being buried in the opposite chain of the protein . TryR Trp-80 is conserved across kinetoplastids ( S5 Table ) , as are two of the three calculated hot spots , Ile/Leu-71 and Phe-78 . Similarly , the hydrophobic face presented by the interface helix identified for GalE contains a third residue , Tyr-115 , not identified in this analysis , but contained in the single alpha-turn found by HippDB as contributing 33% of interface energy ( Fig 2 ) . As in the case of TryR , the GalE interface peptide contains three highly conserved hotspots , Pro-111 , Val-119 , Leu/Ile-123 . Sequence conservation of both interface peptides and hot spots was highly variable from protein to protein , with large differences in enzymes such as G6PDH and RpiB , but high homology in TXNPx and PTR1 ( S5 Table ) . Overall , since the interface peptides were identified manually rather than algorithmically , and this is a relatively small data set , it is impractical to extrapolate broader conclusions about the nature of potentially inhibitory interface peptides . Computational alanine scanning has revealed 12 drug targets in kinetoplastid parasites that are likely amenable to PPI-based inhibition . While all 12 targets are covered by previous PDB-wide analyses focusing on particular structural motifs , manual inspection of this subset has revealed a number of unique sequences that provide a logical starting point for the development of new inhibitors . Nine of the identified targets have sequences overlapping those identified in previous databases , and two have been experimentally verified , suggesting the potential generality of PPI-based inhibition for these homomultimeric enzymes . Moreover , this approach leverages the power of freely available databases and computational tools , allowing for the rapid analysis of newly disclosed structures for novel modes of inhibition . While a generally predictive model of PPI inhibition has yet to be established , the targets identified in this work present particularly attractive opportunities for the exploration of new modes of inhibition for these targets . | Neglected tropical diseases caused by parasites of the genera Trypanosoma and Leishmania affect millions of people , primarily in the developing world . Due to a historical lack of incentive or interest , few new drugs have been developed to treat these conditions . Numerous efforts have targeted the metabolism of trypanothione , an essential molecule for maintaining the redox homeostasis of these parasites . This study uses freely available structural databases and computational tools to identify new druggable sites on several essential proteins in these organisms by disrupting the protein-protein interactions that allow multimeric enzymes to function . Five of the targets identified in this study are involved in redox homeostasis , while the remainder are involved in other essential metabolic or biosynthetic processes . Nine have been identified in other computational databases , and two have already been experimentally verified , which suggests that protein-protein interaction inhibition of multimeric enzymes may be a general and viable route for the development of new trypanocidal agents . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"and",
"discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"enzymology",
"microbiology",
"parasitic",
"protozoans",
"protozoans",
"sequence",
"motif",
"analysis",
"pharmacology",
"enzyme",
"inhibitors",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"bioinformatics",
"bi... | 2017 | Essential multimeric enzymes in kinetoplastid parasites: A host of potentially druggable protein-protein interactions |
Invasive Non-typhoidal Salmonella ( iNTS ) are an important cause of bacteraemia in children and HIV-infected adults in sub-Saharan Africa . Previous research has shown that iNTS strains exhibit a pattern of gene loss that resembles that of host adapted serovars such as Salmonella Typhi and Paratyphi A . Salmonella enterica serovar Bovismorbificans was a common serovar in Malawi between 1997 and 2004 . We sequenced the genomes of 14 Malawian bacteraemia and four veterinary isolates from the UK , to identify genomic variations and signs of host adaptation in the Malawian strains . Whole genome phylogeny of invasive and veterinary S . Bovismorbificans isolates showed that the isolates are highly related , belonging to the most common international S . Bovismorbificans Sequence Type , ST142 , in contrast to the findings for S . Typhimurium , where a distinct Sequence Type , ST313 , is associated with invasive disease in sub-Saharan Africa . Although genome degradation through pseudogene formation was observed in ST142 isolates , there were no clear overlaps with the patterns of gene loss seen in iNTS ST313 isolates previously described from Malawi , and no clear distinction between S . Bovismorbificans isolates from Malawi and the UK . The only defining differences between S . Bovismorbificans bacteraemia and veterinary isolates were prophage-related regions and the carriage of a S . Bovismorbificans virulence plasmid ( pVIRBov ) . iNTS S . Bovismorbificans isolates , unlike iNTS S . Typhiumrium isolates , are only distinguished from those circulating elsewhere by differences in the mobile genome . It is likely that these strains have entered a susceptible population and are able to take advantage of this niche . There are tentative signs of convergent evolution to a more human adapted iNTS variant . Considering its importance in causing disease in this region , S . Bovismorbificans may be at the beginning of this process , providing a reference against which to compare changes that may become fixed in future lineages in sub-Saharan Africa .
Invasive Non-typhoidal Salmonella ( iNTS ) are a major cause of morbidity and mortality in sub-Saharan Africa . Especially in young children , iNTS are either the first or second most common cause of bacteraemia [1] , [2] , meningitis and septic arthritis [3] , [4] with high morbidity . HIV infection is the primary risk factor for iNTS bacteraemia in adults , and it has been suggested that iNTS emerged together with the HIV pandemic in sub-Saharan Africa [5] . The most important clinical risk factors for iNTS disease in children are malnutrition , malaria and anaemia , with one in five cases of NTS bacteraemia in children also associated with HIV infection [2] , [6] , [7] . There is considerable interest in identifying any underlying bacterial genetic basis for the apparent increase in invasiveness and transmission of African NTS strains . Strain collections of iNTS isolates from Africa are dominated by the serovars Typhimurium and Enteritidis [8]–[12] . However a seven year study of iNTS isolates associated with bacteraemia in Malawi showed that S . Bovismorbificans was the third most common serovar , with 46 cases , which accounts for 1% of the total number of NTS isolates [11] . In contrast to this , a study of Salmonella bacteraemia in developed countries ( Finland , Denmark , Canada and Australia ) showed that among 490 bacteraemia NTS isolates , isolated between 2000 and 2007 , only one was S . Bovismorbificans ( 0 . 2% ) [13] . However , S . Bovismorbificans has been found responsible for gastroenteritis outbreaks: in Malaysia between 1973 and 1996 S . Bovismorbificans accounted for 2–11% of salmonellosis cases [14] , [15] as well as isolated outbreaks all over Europe and around the world [16] , [17][18] , S . Bovismorbificans phage type 32 ( PT32 ) [19] , [20] . It has been shown that bacteraemia isolates of S . Typhimurium from Kenya and Malawi belong to a distinct Multi Locus Sequence Typing ( MLST ) group , ST313 , that harbours a specific repertoire of prophages and shows evidence of specific patterns of genome degradation with many parallels to human-specific Salmonella serovars such as S . Typhi and Paratyphi A , which cause acute invasive disease [21] . ST313 is significantly distant from the common gastroenteritis-associated S . Typhimurium ST19 [22] . These studies have raised the possibility that bacteremia-associated iNTS serovars that were previously able to infect a broad host range are becoming human-host adapted [23] , [24] . Multidrug resistance is also a significant factor in the emergence of iNTS strains in Africa , resulting in a reliance on fluoroquinolones [11] , [25] . While S . Typhimurium and Enteritidis isolates from Malawi exhibited resistance to commonly-used antimicrobials ( including ampicillin , co-trimoxazole and chloramphenicol ) , iNTS S . Bovismorbificans isolates have remained comparatively susceptible to these commonly-used antimicrobials . Here , we report the genome sequence of S . Bovismorbificans 3114 ( ST142 ) , a paediatric bacteraemia isolate from Malawi , and describe a detailed analysis of iNTS strains of S . Bovismorbificans causing disease in humans in Malawi and compare them to other isolates from the same region and from veterinary isolates from the UK . We investigated whether there are markers of adaptation to the human host , similar to those described in other iNTS serovars from this region .
Malawian S . Bovismorbificans isolates used in this study were taken from a previous study by Gordon et al [11] and date from 1997 to 2004 . S . Bovismorbificans serovar designations for Malawian strains were confirmed by serotyping at the National Salmonella Reference Laboratory , Galway , Republic of Ireland . Veterinary strains of S . Bovismorbificans were obtained from Professor Paul Barrow ( University of Nottingham ) , and were taken from a collection dating from the 1970s and 1980s . All Salmonella strains were stored in 10% ( v/v ) glycerol broth at −80°C . The antimicrobial susceptibility profiles of the four veterinary strains were determined by the disc diffusion method , in accordance with BSAC guidelines ( http://bsac . org . uk/wp-content/uploads/2012/02/Version-11 . 1-2012-Final- . pdf ) , using a total of 11 antimicrobials: ( AML10 ( amoxicillin 10 µg ) , AMC30 ( amoxicillin/clavulanic acid 30 µg ) , CTX30 ( cefotaxime 30 µg ) , CN 10 ( gentamicin 10 µg ) , CIP 1 ( ciprofloxacin 1 µg ) , W 2 . 5 ( trimethoprim 2 . 5 µg ) , NA 30 ( nalidixic acid 30 µg ) , RL25 ( sulphamethoxazole 25 µg ) , C10 ( chloramphenicol 10 µg ) , TET30 ( tetracycline 30 µg ) , CXM 5 ( cefuroxime sodium 5 µg ) , RD 2 ( rifampicin 2 µg ) , CAZ30 ( ceftazidime 30 µg ) , S 25 ( streptomycin 25 µg ) . The susceptibility profiles of the human S . Bovismorbificans isolates from Malawi have been determined previously , in accordance with BSAC guidelines [11] . Salmonella strains were cultured in Luria broth overnight at 37°C shaking at 200 rpm . Genomic DNA extractions were performed using the Wizard Genomic DNA Purification Kit ( A1120 , Promega , Madison , USA ) as described in the manufacturer's instructions . The genome of S . Bovismorbificans strain 3114 was sequenced using the Roche 454 Genome Sequencer FLX ( GS-FLX ) following the manufacturer's instructions ( Roche 454 Life Science , Branford , CT , USA ) . In brief , each sample was made into both a paired-end and fragment library using the standard FLX chemistry for 454 . Fragment libraries were prepared by fragmentation , attachment of adapter sequences , refinement of the ends and selection of adapted molecules . Paired-end libraries were produced by hydroshear shearing , circularisation , addition of adapters and selection , as for the fragment library . Both libraries were amplified by emPCR and fragment-containing beads were recovered and enriched . Sequencing primers were added and each library was deposited onto a quarter of a PicoTitrePlate plate and sequenced . Multiplexed Illumina standard libraries were prepared for S . Bovismorbificans 3114 and 17 additional strains following standard protocols with 200 bp inserts and sequenced on the Illumina Genome Analyzer II . Paired end sequence runs were performed with 54 bp read length . Raw sequence data is submitted to the public data repository , ENA , under accession ERP000181 . For S . Bovismorbificans 3114 454 data , reads from the fragment and paired-end libraries were de-novo assembled into contigs using the Roche 454 Newbler assembler ( version 2 . 0 . 01 . 12 ) with default settings . Illumina data was then used to extend and order the 454-assembled contigs using the PAGIT package [26] as follows: 454 contigs were extended using ICORN [26] , and the resulting contigs were ordered and orientated with respect to the genome of S . Typhimurium LT2 using ABACAS [27] . Finally , gap closure was attempted where possible using IMAGE [26] . For each scaffold-contig in turn , putative coding sequences ( CDSs ) were predicted using Glimmer version 3 . 02 ( http://www . cbcb . umd . edu/software/glimmer/ ) . A further in-house Perl script was then used to identify and correct those CDSs likely to have been split due to sequencing errors when handling homopolymer repeats . This involved BLASTP alignment of CDS protein translations against a database of translations generated from previously annotated Salmonella genomes , the identifications of likely indels within homopolymer regions , the modification of coding sequence feature positions to correct errors , and the merging of relevant CDSs . Where such modification occurred , this was recorded as metadata ( in the form of the eventual GenBank feature note field ) . Such CDSs were also marked with the exception flag set to ‘low-quality sequence region’ for the final GenBank submission to signify poor quality sequencing . A putative function was then assigned to each gene by BLASTn ( NCBI Blast 2 . 2 . 17 ) comparison with a database of sequences generated from the previously annotated genome of S . Typhimurium LT2 . Putative tRNA genes were detected using tRNAscan-SE 1 . 23 ( ftp://selab . janelia . org/pub/software/tRNAscan-SE/ ) . A pseudochromosome , consisting of a concatenation of contigs arranged into scaffolds , with 100 Ns separating adjoining scaffolds , was prepared for comparison purposes; the final version of the 3114 genome was manually curated using Artemis [28] , [29] . The S . Bovismorbificans 3114 chromosome has been submitted to EMBL under accession number HF969015 , and its 93 . 8 kb virulence plasmid under accession number HF969016 . Illumina sequence data for the 17 additional genomes was assembled as follows: for each strain , Velvet [30] was used to create multiple assemblies by varying the kmer size between 66% and 90% of the read length . From these assemblies , the one with the best N50 was chosen and contigs which were shorter than the insert size length were removed . An assembly improvement step was then run on the chosen assembly . The contigs of the assembly were scaffolded by iteratively running SSPACE [31] . Then gaps identified as 1 or more N's , were targeted for closure by running 120 iterations of GapFiller [32] . Finally , the reads were aligned back to the improved assembly using SMALT ( http://www . sanger . ac . uk/resources/software/smalt/ ) and a set of statistics was produced for assessing the QC of the assembly . All of the software developed is freely available for download from GitHub ( https://github . com/sanger-pathogens ) under an open source license , GNU GPL 3 . The improvement step of the pipeline is also available as a standalone Perl module from CPAN ( http://search . cpan . org/~ajpage/ ) ( see Table S1 for assembly data and statistics ) . Pseudogenes were identified , using ACT comparisons [33] , by comparing the genome of strain 3114 , first to S . Typhiumurium LT2 and then to the genomes of S . Typhimurium D23580 , SL1344 and DT104 . Pseudogenes were identified according to whether CDS showed frameshift mutations , missing N- or C-terminals or carried nonsense mutations . Once pseudogenes were identified in the 3114 genome , their orthologous sequences were checked in the assemblies of a further six S . Bovismorbificans isolates , from both Malawi ( 3180 , D1253 , D993 , A1668 , ) and the UK ( 653308 , 276608 ) . These isolates were chosen as representatives of distinct subclades in the tree . In some cases it was not possible to identify mutations due to gaps in the sequence; these are highlighted in Table S5 . MLST sequences were obtained from Illumina reads and sequence types were assigned through the MLST website ( http://www . mlst . net/ ) . Ambiguous results for some individual loci were subsequently confirmed by PCR amplification of the locus and sequencing of the PCR product ( Beckman Coulter ) . In order to build the MLST-based phylogenetic tree of Figure S1 , the concatenated sequences of the seven MLST loci of S . Bovismorbificans 3114 and the most common S . enterica STs from published databases were loaded into SeaView v3 . 2 [34] . The phylogeny was reconstructed using PhyML [35] within the Seaview package , and FigTree v1 . 3 . 1 [36] was used to edit and label the final figure . For the purpose of mapping and visualization of the genomic content of all S . Bovismorbificans samples , a pseudomolecule was constructed comprising the reference 3114 genome ( chromosome and virulence plasmid ) and all non-redundant accessory regions found by tblastx in each sample assembly with respect to the others in an iterative manner . Briefly we performed pairwise comparisons , firstly of one isolate against the reference sequence ( genome and its plasmid ) . Regions in the comparator that were not present in the reference were identified and added to the end of the reference sequence to form a pan genome pseudomolecule . This was repeated in an iterative process involving manual curation of the sequences to be included in the growing pseudo molecule . Mapping of illumina reads per sample was carried out against this resulting pseudomolecule using SMALT ( http://www . sanger . ac . uk/resources/software/smalt/ ) without mapping to repeats . SNP calling was performed as previously described [37] . In order to construct a robust phylogenetic tree a Bayesian approach [38] , which identifies high density SNPs and recombinant regions and ignores them when constructing the phylogeny , was used . A detailed list of sites removed from the chromosome , 48 , 423 bases in total , is summarised in Table S2 and the alignment of variant sites used to construct the final phylogenetic tree is presented in Table S3 . In order to construct the phylogeentic tree shown in Figure 1 an initial tree using S . Heidelberg str SL476 ( acc no . CP001120 ) ( see Figure S1 ) , as an outgroup was constructed . This identified the root position in the ingroup and final tree . Using this root position the final tree shown in Figure 1 was constructed using all 954 variant sites ( Table S3 ) within the chomosome: A maximum likelyhood approach ( RAxML ) was used to construct the initial bipartitions tree followed by a reconstruction of the SNPs onto the tree branches using delayed transformation ( DELTRAN ) parsimony [39] . Mapped illumina read data was saved in Bam format [40] and converted to coverage files ( number of reads mapped to each base coordinate of the reference ) with an in-house script . To aid visualisation the read depth per base position of each isolate against the pan-pseudomolecule reference sequence was constructed ( Figure 1 ) . Base positions with 0 or 1–14 reads mapped were coloured white or grey , respectively . Base positions with 15× coverage or greater are coloured black . The cutoff of 15 or more times coverage per base position was selected because it was just below the minimum median coverage obtained across all of the isolates we sequenced ( 16 to 31× coverage [data not shown]; see Figure 1 ) . Using the observed coverage regions >4000 bps showing a significant deviation from the median coverage were identified by manual curation and checked against the genome assembly . The raw sequence data is available under the accession number ERP000181 at the European Nucleotide Archive , ENA . The sequence and annotation data for S . Bovismorbificans strain 3114 chromosome and virulence plasmid , pVIRBov are available from ENA under accession numbers HF969015-HF969016 .
To establish a phylogenetic framework for the S . Bovismorbificans samples we sequenced the genomes of 18 isolates , 14 of which were derived from Malawian adults and children isolated between 1997 and 2004 at the Queen Elizabeth Hospital Blantyre , Malawi . Suspecting these could be clonal , we brought into the analysis the sequences of 4 further isolates of different origin ( pigs and alpaca ) , geographical location ( UK ) and temporal isolation ( 1970s/80s ) to provide context to investigate wider serovar variations . Genomic DNA from the S . Bovismorbificans isolates were assayed using either 454 or multiplex Illumina sequencing ( see methods ) . The genome data of these sequences are summarised in Table 1 . Only chromosomal SNPs were used to construct the maximum likelihood phylogenetic tree ( see methods ) shown in Figure 1 ( left ) , for which the root branch ( leading to sample 51892776 ) was previously identified by including an outgroup ( see methods ) . Recombinant regions and regions that were unlikely to reflect the core phylogeny , such as prophage , were removed from this analysis ( 48 , 423 total sites; Table S2 ) . Amongst our samples , we found a total of 954 variable sites randomly distributed around the S . Bovismorbificans chromosome , which is approximately one SNP per 4 , 742 bp or just under 0 . 001% nucleotide divergence , within the core regions . It is evident that the human and animal isolates are intermixed . To give an idea of the level of nucleotide divergence within the core genome the animal isolate 51892776 closest to the root , was separated by 328 or 341 SNPs from the reference human isolate 3114 or most the divergent isolate shown in the tree , respectively . Consistent with this , we extracted the sequences of the MLST loci from the whole genome sequence data . All strains belonged to the major S . Bovismorbificans sequence type ST142 except two Malawian strains D993 and D4891 which were single locus variants ( SLV ) of ST142 ( see Table 1 ) but were not found to be located on long branches on the tree . It is evident from Figure 1 ( left ) that the human S . Bovismorbificans isolates taken in Malawi are phylogenetically extremely closely related , when compared to each other . This level of sequence divergence is comparable to the evolutionary distance between the S . Typhimurium lineages causing invasive disease in Africa [41] which form two distinct lineages ( differentiated from each other by 455 SNPs ) , that are separated from the nearest gastroenteritis lineages by >700 SNPs [41] . However , when including the animal isolates from the UK it is apparent that this limited variation is a feature of S . Bovismorbificans as a serovar despite temporal , geographic and host differences . Also , in contrast to the S . Typhimurium lineages causing invasive disease , there was no clustering within the human samples with age or year of isolation . Together these data suggested that the S . Bovismorbificans isolates causing invasive disease in Malawi , unlike S . Typhimurium , were not a specialised clade , at least according to the core phylogeny , therefore we looked at the accessory genome for clues to the observed differences in disease outcome for these isolates causing invasive disease in Malawi . In order to visualise the variation across S . Bovismorbificans isolates we constructed a pan-genome . To do this we concatenated the whole genome sequence of strain 3114 ( chromosome and virulence plasmid ) as well as the regions found to be variable present in one or more isolate , in a non-redundant manner ( see methods ) . Figure 1 ( right ) shows the read coverage of all isolates included in this study mapped against the pan-genome pseudomolecule ( see methods ) . There are three main regions of difference ( RODs ) in chromosome of strain 3114 , denoted ROD_13 , ROD_14 and ROD_34 . These are the only significant regions of difference in the core genome being either absent or partially present in the four veterinary strains and variably distributed in the human isolates ( Table 1 and Figure 1 ) . They are predicted to encode prophage; ROD_34 and ROD_13 are highly similar to the S . Typhimurium prophage elements Gifsy-1 and Gifsy-2 , respectively , and have therefore been termed Gifsy-like ( See Table S4 ) . ROD14 represents a novel 46 . 4 kb prophage inserted into a spermidine/putrescine operon on the S . Bovismorbificans 3114 genome ( See below; Table S4 ) . A distinctive region of variation between human ( carried in all ) and animal samples ( variably present ) was a ∼93 kb virulence plasmid , here named pVIRBov ( Figure 2 ( B ) ) . pVIRBov is highly simlar to the S . Typhimurium LT2 virulence plasmid , pSLT ( See Figure 2 ( B ) and S2 ) and , like pSLT , carries the defining spv virulence gene cassette and the pef ( plasmid-encoded fimbriae ) operon mediating adhesion to murine intestinal epithelial cells [42] . Located downstream of the pef operon is the rcK ( resistance to complement killing ) gene . RcK is required by S . Typhimurium for survival in macrophages and for virulence in mice [43] . pVIRBov also shows a 7 . 463 kb deletion , and a 6 . 705 kb insertion , compared to pSLT . The deleted region contains a putative gene for single strand binding protein B ( ssbB ) , as well as a number of putative genes encoding membrane-associated proteins . The insertion relative to pSLT includes a pilA-like gene ( SBOV4711 ) , repC ( SBOV47701 ) , a gene encoding a putative outer membrane protein ( SBOV47871 ) , traJ ( SBOV48411 ) , the primary activator of tra ( polycistronic transfer ) operon expression [44] and a number of hypothetical proteins . pVIRBov does not carry rsk ( resistance to serum killing ) , thought to be associated with the control of serum resistance [45] . In contrast to a previous PCR based screen of S . Bovismorbificans isolates , these data show that 100% of the human bacteraeamia isolates and only one in four of the UK veterinary isolates carry pVIRBov [46] . In pairwise comparisons with respect to the S . Bovismorbificans str 3114 genome , eight human isolates ( D1253 , D4891 , D4451 , A1104 , A1608 , A1668 , A16892 and A8737 ) were almost identical in genomic content to the reference , with accessory regions of only 4–8 kb ( summarised in Table 1 ) . The veterinary isolate 499208 and the paediatric bacteraemia isolate 3476 both carry the largest accessory genomic regions , of ∼290 kb and 132 kb , respectively ( Table 1 ) . The accessory genome of 3476 contains a ROD inserted at a t-RNA ( Phe ) -downstream of the CDS homologue to SBOV31771 in the 3114 strain- , showing high sequence homology to the SPI-7 of the human restricted pathogen S . Typhi CT18 ( Figure 1 , Figure S3 ) . SPI-7 is a large mosaic pathogenicity island carrying a collection of virulence-related genes; versions of SPI-7 have been identified in S . enterica serovars Typhi , Paratyphi C , and Dublin [47]–[49] . This particular version of sample 3476 also encodes a putative Vi capsule although lacks the sopE phage of S . Typhi CT18 . It also contains an operon related to carbohydrate metabolism and extracellular structure modifications ( Figure S3 ) . We did not find any genomic scar or any other evidence to show that this island might have been contained in any of the other human isolates or most recent ancestors . Despite the existence of substantial accessory regions in half of the samples sequenced , they were mostly found to be unique to single isolates without correlation to the phylogenetic inference or the disease outcome . In addition to gene gain , functional gene loss plays an important role in the adaptation of the Salmonella to different lifestyles , with host restricted Salmonella carrying over 200 pseudogenes [21] , [22] , [47] , [50]–[52] compared to their broad host range counterparts . It is also evident that the patterns of gene loss are nonrandom with nonsense mutations and frame-shifts being over-represented in genes that are associated with aspects of virulence or host interaction . The parallels in the patterns of pseudogene accumulation is also a feature of the iNTS S . Typhimurium isolate D23580 , causing invasive disease in Malawi . By comparing the genome of strain 3114 to those of S . Typhimurium D23580 , SL1344 and DT104 , a total of 43 pseudogenes were identified in S . Bovismorbificans 3114 . Six further S . Bovismorbificans isolates , which were chosen as representatives of distinct groups on the phylogenetic tree were also analysed for pseudogene content ( summarised results in Table S5 ) Of the 43 pseudogenes identified , 15 ( 34 . 9% ) were conserved in all of the six S . Bovismorbificans isolates tested , but intact in the three S . Typhimurium strains . There was no clear distinction in pseudogene carriage between human bacteraemia ( Malawi ) and veterinary ( UK ) S . Bovismorbificans isolates , with only a single pseudogene specific to Malawian isolates . Five ( 11 . 5% ) pseudogenes have been identified as pseudogenes in all seven S . Bovismorbificans and all three of the S . Typhimurium isolates ( Table S5 ) . Genes that are pseudogenes in all seven S . Bovismorbificans isolates tested , but intact in all three S . Typhimurium isolates ( D23580 , SL1344 , DT104 ) , include genes involved in sugar metabolism , a putative autotransporter of the haemagglutinin family ( SBOV37821 ) and putative surface-exposed virulence protein BigA ( SBOV35501 ) . We observed no clear link between pseudogene carriage and source ( human and veterinary isolates ) . Although the extent of pseudogene formation in S . Bovismorbificans does not compare to the host adapted Salmonella serovars or to the iNTS Typhimurium D23580 , there are tantalising glimpses that suggest S . Bovismorbificans as a species may carry pseudogenes , such as BigA that are consistent with host specialisation . What is clear , though , is that this has not been driven by the emergence of a new dominant lineage adapted to humans in Africa as we have seen before with S . Typhimurium ST313 . Since this is the first time a S . Bovismorbificans genome has been described and to identify functions that may be involved in the apparent plastiticity in pathogenicity we performed whole-genome comparison between S . Bovismorbificans and S . Typhimurium LT2 . These comparisons revealed a high level of synteny and colinearity , with no inversions ( Figure 3 ) . A comparison of S . Bovismorbificans genome statistics with other serovars is summarized in Table 2 . Figure 2 ( A ) shows orthologous genes , identified by reciprocal fasta searches , in 10 S . enterica subspecies 1 strains from eight Salmonella serovars as well as wider members of the Enterobacteriaceae . This analysis showed that S . Bovismorbificans gene content broadly resembles those common to S . enterica subspecies 1 NTS serovars . Genome comparisons to S . Typhimurium LT2 show that , while SPI-1 , -2 , -4 , -5 , -9 and -11 are largely synonymous , SPI-3 , -6 , -10 and -12 show deletions compared to the genome of S . Typhimurium LT2 . Of note is SPI-6 , formerly known as SCI ( Salmonella enterica centisome 7 island ) , which is approximately 10 . 6 kb in size , compared to the 47 kb SPI-6 in the genome of S . Typhimurium LT2 , and simply retains part of the fimbrial saf operon while lacking the Type VI secretion system ( T6SS ) encoded by this island . The T6SS is thought to play a role in adaptation to different lifestyles and environments , particularly animal hosts . SPI-6 T6SS was found to be absent from serovars Enteritidis , Gallinarum , Agona , Javiana , Virchow and IIIb 61∶1 , v∶1 , 5 , ( 7 ) [53] . Like S . Typhimurium LT2 , SPI-13 and -14 , are largely absent from S . Bovismorbificans [54] as are SPI-15 and -17 ( Figure 3 see Table S6 for details on SPI repertoires ) S . Bovismorbificans 3114 and S . Typhimurium LT2 share the same repertoire of 13 fimbrial operons ( stf , saf , stb , fim , stc , std , lpf , stj , sth , bcf , sti , csg and pef ) although safA of the saf operon is absent from the genome of strain 3114 . The positions of the fimbrial operons within the S . Bovismorbificans 3114 genome are summarized in Figure 3 . Regions of difference ( RODs ) were defined as insertions ( or replacements ) in the genome of S . Bovismorbificans 3114 when compared to published S . Typhimurium genomes ( see methods; summarized in Table S4 ) . A total of 27 RODs were identified many of which were predicted to encode proteins of unknown function . The most significant class of RODs were those related to prophage elements ( ROD7 , -12 , -13 , -14 , -17 , -21 , -30 , -31 , 34; ROD 13 , -14 and -34 are described above ) . Prophages are important sources of genomic variation in Salmonella , with most serovars being polylysogenic [55] , [56] . Cryptic prophages have been shown to contribute to bacterial survival in adverse environments . They have been shown to help bacteria overcome acid , osmotic and oxidative stresses , influence growth and biofilm formation and contribute significantly to resistance to ß-lactams and quinolones [57] . In comparison to the genome of S . Typhimurium LT2 , the genome of S . Bovismorbificans 3114 has a number of deletions or variations in regions related to common Salmonella prophages . There are no putative genes matching the common Salmonella prophage Fels-1 and Fels-2 , with the exception of the Fels-1 ybjP gene . Also absent , compared to S . Typhimurium LT2 , are inducible prophages Gifsy-1 and Gifsy-2 , which have been replaced by prophage-like RODs 34 and 13 respectively ( See above; Table S4 ) . Although , ROD13 presents a partial match to Gifsy-2 , unlike Gifsy-2 , it does not carry the same genetic cargo sodC1 which is associated with intracellular survival [58] or gtgA , which together with sodCI and gtgB is also absent from Gifsy-1 of S . Typhi [59] . ROD34 is 45 . 8 kb in size and carries Gifsy-1 like regions in both terminal regions , as well as one Fels-1 like region . ROD14 represents a novel prophage 46 . 4 kb in size with some similarities to a predicted E . coli ( UMKN88 ) phage and to Bacteriophage P27 . The cargo of ROD14 largely constitutes hypothetical proteins with the exception of the SPI-2 effector sifA gene ( SBOV11471 ) which is essential for Sif formation , a process linked to SCV ( Salmonella Containing Vacuole ) integrity [60] , [61] . The remaining prophage regions show no similarity to other Salmonella prophage but are present in all of the strains we have sequenced , regardless of source ( summarised in Table S4 ) Multidrug resistance is a serious problem in sub-Saharan Africa . S . Typhimurium and S . Enteritidis isolates from Malawi exhibit extensive resistance profiles . The empirical treatment for adults with sepsis in Malawi was chloramphenicol and benzyl penicillin . With the emergence of chloramphenicol resistance in S . Typhimurium isolates in 2002 , treatment was switched to ciprofloxacin and parenteral gentamicin was added . While S . Typhimurium isolates have been resistant to ampicillin and trimethoprim-sulphamethoxazole for a long time , a dramatic increase of resistance to ampicillin , trimethoprim-sulphamethoxazole and chloramphenicol was observed in S . Enteritidis isolates in 1999 [62] . Phenotypic antibiotic resistance profiles ( Table 1 and Figure 1 ) were obtained for all S . Bovismorbificans samples in this study . Of the 14 Malawian S . Bovismorbificans isolates , 11 showed resistance against sulphamethoxazole , cefuroxime and rifampicin , while three isolates were resistant to cefuroxime and rifampicin only . Contrary to the findings for S . Typhimurium and S . Enteritidis , S . Bovismorbificans isolates remain susceptible to chloramphenicol and ampicillin , and resistance to sulphamethoxazole and cefuroxime follows no dicernible temporal distribution . Table 3 and Table S7 summarize putative resistance related genes identified in both the core and accessory genomic regions of all S . Bovismorbificans isolates . Consistent with other studies in enteric bacteria , linking antibiotic resistance phenotype and genotype was problematic [63] . However , we were able to identify a number of putative β-lactamase genes in S . Bovismorbificans that may explain the resistance to cephalosporins , such as Cefuroxime . Moreover all the S . Bovismorbificans isolates ( including the veterinary ones ) carry the mutation in the rpoB gene associated with resistance to rifampicin [64]–[66] ( Figure S4 shows the rpoB gene alignment of strain 3114 together with those of S . Typhimurium LT2 and DT104 ) . Despite phenotypic resistance to sulphamethoxazole no sul genes could be identified in any of the 14 Malawian S . Bovismorbificans isolates to explain this . Prevalence of tuberculosis ( TB ) is high in HIV-infected patients [67] , and co-infection with invasive NTS and TB is common [68] . Rifampicin is the standard treatment for TB . Ours and previous data show that all of the NTS isolates , regardless of serovar , were resistant to rifampicin [11] . It is not possible to comment on whether co-infection and treatment for TB might be a causal mechanism for the emergence of resistance among S . Bovimorbificans or simply allowed an already resistant S . Bovimorbificans to exploit this niche . There is , however , strong evidence antimicrobial resistance , particularly to chloramphenicol in the case of S . Typhimurium have been key drivers in the spread of Salmonella pathovars across Africa [41] . A number of additional putative resistance genes were identified in the S . Bovismorbificans core genome , including those directed against aminoglycosides ( aadA ) , streptomycin/spectinomycin ( aphA ) and trimethoprim ( dhfr1 ) ( Table S7 ) . For completeness the four veterinary UK isolates compared to the human isolates showed an even more diverse antimicrobial resistance profiles ( summarised in Table 1 ) , some of them showing an extensive array of resistance related genes , perhaps associated with the type of antibiotics and dosages used in the UK at the time they were isolated . In conclusion all S . Bovismorbificans isolates included in this study showed extremely close phylogenetic relationships regardless of source , place of isolation , host or disease outcome , even though morbidity and mortality caused by NTS is much more severe in sub-Saharan Africa and the developing world [22] , [68] . Genome comparisons between the Malawian bacteraemia and UK veterinary isolates showed few clear differences . In our study , all of the bacteraemia isolates from Malawi were of the most prevalent S . Bovismorbificans sequence type , ST142 . Unlike iNTS S . Typhimurium isolates causing invasive disease in Malawi there is no evidence that functional gene loss was a significant feature of the evolution and adaptation to a more invasive lifestyle for African S . Bovismorbificans isolates . The only differences from those strains circulating elsewhere were in the mobile genome , largely prophage , and the presence of the virulence plasmid ( only in one of four of the UK veterinary samples ) . However comparing the accessory genomic variations of the African S . Bovismorbificans isolates , such as the apparently random presence or absence of SPI-7 , it strongly suggested that those causing disease originate from a mixed population of bacteria circulating within the region and that invasive disease by this serovar was caused by multiple sporadic independent bacteraemia infections . All isolates , regardless of source , appear to display multiple phenotypic and genotypic drug resistance markers . In Malawi this is likely to have been essential to colonise a susceptible population , which tend to take regular antibiotic therapy . Although there is no obvious sign of convergent evolution to a more human adapted iNTS variant of S . Bovismorbificans , these strains -considering their importance in causing disease in this region- , may be at the very beginning of this process and so this study provides the reference point against which to compare changes that may become fixed in future lineages in sub-Saharan Africa . This study also highlights the likely importance of the patterns of evolutionary change we have previously highlighted in S . Typhimurium and show how , given the opportunity , multiple Salmonella serovars are able to cause more acute disease in susceptible populations . | Bacteraemia and meningitis caused by non-typhoidal Salmonella ( including serovars Typhimurium , Enteritidis and Bovismorbificans ) are a serious health issue in sub-Saharan Africa , particularly in young children and HIV-infected adults . Previous work has indicated that a distinct S . Typhimurium sequence type , ST313 , has evolved and spread in these countries , and may be more human-adapted than isolates found in the developed world . We therefore investigated the genomes of Salmonella enterica serovar Bovismorbificans bacteraemia isolates from Malawi and compared them to genomes of veterinary S . Bovismorbificans isolates from the UK using Next Generation Sequencing Technology and subsequent genomic comparisons to establish if there is a genetic basis for this increase in invasive disease observed among African NTS . Contrary to the previous findings for S . Typhimurium , where a distinct ST is found only in sub-Saharan Africa , we discovered that the S . Bovismorbificans isolates from Malawi belong to the most common ST of the serovar and the genome is highly conserved across all sequenced isolates . The major differences between UK veterinary and African human isolates were due to prophage regions inserted into the genomes of African isolates , coupled with a higher prevalence of a virulence plasmid compared to the UK isolates . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"and",
"Discussion"
] | [] | 2013 | Genomic Characterisation of Invasive Non-Typhoidal Salmonella enterica Subspecies enterica Serovar Bovismorbificans Isolates from Malawi |
Understanding the evolution of a protein , including both close and distant relationships , often reveals insight into its structure and function . Fast and easy access to such up-to-date information facilitates research . We have developed a hierarchical evolutionary classification of all proteins with experimentally determined spatial structures , and presented it as an interactive and updatable online database . ECOD ( Evolutionary Classification of protein Domains ) is distinct from other structural classifications in that it groups domains primarily by evolutionary relationships ( homology ) , rather than topology ( or “fold” ) . This distinction highlights cases of homology between domains of differing topology to aid in understanding of protein structure evolution . ECOD uniquely emphasizes distantly related homologs that are difficult to detect , and thus catalogs the largest number of evolutionary links among structural domain classifications . Placing distant homologs together underscores the ancestral similarities of these proteins and draws attention to the most important regions of sequence and structure , as well as conserved functional sites . ECOD also recognizes closer sequence-based relationships between protein domains . Currently , approximately 100 , 000 protein structures are classified in ECOD into 9 , 000 sequence families clustered into close to 2 , 000 evolutionary groups . The classification is assisted by an automated pipeline that quickly and consistently classifies weekly releases of PDB structures and allows for continual updates . This synchronization with PDB uniquely distinguishes ECOD among all protein classifications . Finally , we present several case studies of homologous proteins not recorded in other classifications , illustrating the potential of how ECOD can be used to further biological and evolutionary studies .
The billions of proteins in extant species constitute a bewilderingly diverse protein world . To understand this world , systematic classifications are needed to reduce its complexity and to bring order to its relationships . As proteins are the products of evolution , their phylogeny provides a natural foundation for a meaningful hierarchical classification . As in the classification of species , a phylogenetic classification of proteins identifies evolutionary relationships between proteins and groups homologs ( proteins that are descendants of a common ancestor ) together . Because homologs generally share similar three-dimensional ( 3D ) structures and functional properties , such a classification provides a valuable platform for studying the laws of protein evolution by comparative analysis as well as for predicting structure and function by homology-based inference . Many protein classifications are currently available . Comprehensive sequence-based classifications such as Pfam [1] and CDD [2] are among the most popular protein annotation tools . When sequence-only methods fail to reveal more distant evolutionary links , 3D structures allow us to see further back in time , as protein structure is generally better preserved than sequence in evolution [3] . Currently , the two leading structure classifications are SCOP ( Structural Classification of Proteins ) [4] and CATH ( Class , Architecture , Topology , Homology ) [5] , both of which are widely used in analyzing protein sequence , structure , function , and evolution and in developing various bioinformatics tools . CATH ( http://www . cathdb . info ) is largely automatic with added manual curation and emphasizes more on geometry , while SCOP is mainly manual and focuses on function and evolution . In the SCOP [4] ( http://scop . mrc-lmb . cam . ac . uk/scop/index . html ) hierarchical classification , closely related domains are grouped into families; families with structural and/or functional similarities supporting common ancestry are grouped into superfamilies; superfamilies with similar 3D architectures and topologies are grouped into folds; and folds with similar secondary structure compositions are grouped into classes . Cataloging remote homologies identified by a combination of visual inspection , sequence and structure similarity search , and expert knowledge , the SCOP superfamily is the broadest level indicating homology and offers invaluable insights in protein evolution . However , SCOP tends to be conservative in assessing evolutionary relationships , and many homologous links reported in literature are not currently reflected [6] , [7] , [8] , [9] , [10] , [11] . Also , the recent dramatic increase of available structures in the PDB [12] ( http://www . pdb . org ) hinders careful manual curation in SCOP . Recently , a new version of SCOP ( SCOP2 ) [13] was introduced that eschews hierarchical classification in place of a network of relationships ( homologous and structural ) , although this database has not been made current with PDB . To partially alleviate this problem , ASTRAL now offers SCOPe , a sequence-based extension of the original SCOP hierarchy [14] . Nevertheless , not a single protein classification database has kept current with the PDB database . We maintain that the most recently determined structures , especially those evolutionarily distant from classified proteins , attract the most interest and hence are the most important to classify quickly and accurately . However , automatic updates , such as those in ASTRAL , are only able to deal with easily classifiable proteins . Here we introduce the ECOD ( Evolutionary Classification Of protein Domains ) database . Our goal is threefold: ( 1 ) to construct a comprehensive domain classification based on evolutionary connections , ( 2 ) to extend the realm of connections to include remote homology , and ( 3 ) to maintain concurrent updates with the PDB . Because experimental data is very sparse compared to sequence data , establishing an evolutionary-based classification scheme of structures allows for biological insight into related proteins that otherwise lack functional information . In such a scheme , close homologs admittedly represent the most relevant source of functional inference . However for most proteins , only distant homologs have been studied in detail . Fortunately , many examples have shown that analysis of proteins in the context of their distant homologs provides functional clues that advance biological research [15] , [16] , [17] , [18] . In addition , remote homology offers deeper insights in protein evolution . In order to extend distant evolutionary relationships beyond the SCOP superfamily level in ECOD , we apply state of the art homology-inference algorithms both developed in our group [19] , [20] as well as by others [21] , [22] , manually analyze and verify the suggested homologous links , and incorporate findings from literature . For weekly updates , we rely on a computational pipeline that automatically and confidently classifies the majority of newly released structures and flags incompletely classified and unclassifiable structures , as well as a web interface that presents those difficult to deal with structures and pre-computed data in a convenient way for rapid manual inspection and classification . ECOD is a publicly available database ( http://prodata . swmed . edu/ecod/ ) . By focusing on remote homology and weekly updates , ECOD strives to provide a more simplified and up-to-date view of the protein world than is currently available in existing classifications . As such , ECOD is unique in combining the following features: 1 ) the aforementioned weekly updates , following new releases from the PDB; 2 ) a hierarchy that specifically incorporates sequence-based relationships in a family level of close homology; 3 ) a classification that reflects more distant evolutionary connections; 4 ) a hierarchy that lacks a SCOP-like fold level , as the definition of “fold” is often subjective [23]; 5 ) domain partitions for all former members of the SCOPmulti-domain protein class; and 6 ) combination of membrane proteins with their soluble homologs where an evolutionary relationship can be hypothesized . Theoretically , ECOD catalogs rich and up-to-date information about protein structure for the studies on protein origins and evolution; and practically , it helps homology-based structure and function prediction and protein annotation by providing a pre-compiled search database .
We first developed a pilot version of ECOD based on SCOP 1 . 75 [4] . To detect remote homologies beyond the SCOP superfamily level , 40% identity domain representatives in the first 7 classes in SCOP 1 . 75 were retrieved from ASTRAL [24] and compared in an all-versus-all fashion . Four scores were computed for each pair: HHsearch probability [21] , DALI Z-score [22] , HorA combined score [20] , and HorA SVM score [19] . Domain pairs with high scores were manually inspected and analyzed . The decision on whether any given pair is homologous was based on considerations of the aforementioned scores , literature , functional similarity ( such as common cofactor-binding residues ) , shared unusual structural features [25] , domain organization , oligomerization states , and disulfide bond positions . Since the SCOP superfamily level is reliable and conservative , we typically only merged SCOP superfamilies into homologous ( H- ) groups . In addition to merging SCOP superfamilies , we split SCOP entries with multiple domains or with duplications , and corrected rare inconsistencies in the SCOP classification . Cytoscape [26] clustering was used to aid manual analysis by displaying domains and high-scoring links . After 40% representatives were classified , other SCOP 1 . 75 domains were automatically mapped into the ECOD hierarchy using MUSCLE alignments [27] . Many hierarchical groups in the ECOD pilot version retained the names of their original SCOP counterparts . Those structures not classified in SCOP 1 . 75 were partitioned and assigned to ECOD using a combination of sequence and structural homology detection methods . We used an iterative pipeline of three sequence homology detection methods of increasing sensitivity and decreasing specificity to partition input proteins into domains ( Fig . 1 ) . First , the input protein sequence is queried against a library of known ECOD full-length chains ( containing both single-domain and multi-domain architectures ) using BLAST [28] , [29] . Where significant sequence similarity ( E-value<2e-3 ) is detected to a known domain architecture with high coverage ( <10 residues uncovered ) , the entire series of domains in the input chain was partitioned in one pass . Second , the protein sequence is queried using BLAST against a library of domain sequences . Here single-domain proteins and components of multi-domain proteins were assigned individually by sequence similarity ( E-value<2e-3 ) and hit coverage ( >80% ) . Finally , for detection of more distant homology , a query sequence profile was generated using HHblits [21] . This profile was used to query a database of ECOD representative domain profiles using HHsearch . Domains from the input chains could be classified by any combination of the three sequence-based methods ( chain BLAST , domain BLAST , or domain HHsearch ) . Following partition , a boundary optimization procedure based on the structural domain parser , PDP , was run to eliminate small interstitial gaps between assigned domains and at termini [30] . Input protein chains with a set of detected domains with full residue coverage from the sequence pipeline were considered to be complete . Domains from these chains were then assigned to the ECOD hierarchy broadly using the classification of their hit domain . Following this assignment a combination of HMMER/Pfam and HHsearch-based clustering was used to finely tune family assignments [1] , [31] . Domains were clustered into F-groups by Pfam where confident HMMER3-based assignments could be found . Where domains had no confident Pfam assignment , all-versus-all HHsearch-based complete linkage clustering was used to generate an F-group [32] where all domains shared 90% HHsearch probability . We specifically designate provisional representatives in F-groups where no member shares close homology with a representative ECOD domain for manual examination . Input protein chains that could not be fully assigned by the sequence pipeline were passed to the structural pipeline . If a protein chain could not be assigned by the sequence pipeline , it was queried against a library of representative ECOD domain structures using DaliLite [33] . Domains were assigned where significant structural similarity existed to a known ECOD domain and where the aligned region passed a simple BLOSUM-based alignment score [34] . As in the sequence pipeline , the boundaries of structurally assigned domains were optimized , and those chains that could be completely assigned ( 100% residue coverage ) were added to the classification . Where a chain could not be completely assigned , it was passed to the manual curators for boundary refinement or assignment . As we neared completion of the PDB , the need for structural search decreased as the number of remaining structures was small enough to manually curate . Difficult structures that could not be completely and confidently classified by the pipeline required manual curation . We first inspected the mapping suggested by the pipeline . Oftentimes , the suggested mapping was correct for most or part of the query structure , and we typically accepted this mapping but modified the domain boundaries . For other queries where the suggested mapping was wrong or absent , we used HorA server [20] to search for remote homologs . In evaluating HorA results , we applied the same considerations used in developing the ECOD pilot version to determine homology between a query and a hit . When a homologous hit with similar topology could be found , the query was classified into the same T-group as the hit; when a homologous hit with different topology could be found , the query was classified in a new T-group but the same H-group as the hit; when only a possibly homologous hit with similar overall structure could be found , the query was classified in a new H-group but the same X-group as the hit; when no possible homologs can be identified , the query is classified in a new X-group by itself ( see Results and Discussion for a description of the ECOD hierarchy ) . To facilitate manual analysis , we developed a web interface that presented relevant information in a clear format as well as recorded and incorporated feedback and annotations from manual curators .
ECOD is a hierarchical classification with five main levels ( Fig . 2 , from top to bottom ) : architecture ( A ) , possible homology ( X ) , homology ( H ) , topology ( T ) , and family ( F ) . The architecture level ( A ) groups domains with similar secondary structure compositions and geometric shapes . The possible homology level ( X ) groups domains where some evidence exists to demonstrate homology ( but where further evidence is needed ) . The homology level ( H ) groups together domains with common ancestry as suggested by high sequence-structure scores , functional similarity , shared unusual features [25] , and literature . The topology level ( T ) groups domains with similar topological connections . The family level ( F ) groups domains with significant sequence similarity ( primarily according to Pfam , secondarily by HHsearch-based clustering ) . ECOD has 20 architectures that were developed both by consulting SCOP fold descriptions and inspecting numerous structures . We note that clear-cut boundaries between architectures do not always exist and that domain assignment to an architecture is sometimes subjective . This level is introduced largely for convenience of users and does not directly correspond to evolutionary grouping . A-level lies in between SCOP class and fold and groups proteins by simple visual features such as bundles , barrels , meanders , and sandwiches . Coiled-coils , peptides , fragments , largely disordered structures , and low resolution structures were put in special architectures with no X- , H- , T- , or F-levels , as confident evolutionary classification of these structures is challenging at the moment . Nucleic acids , in addition to proteins , are kept within a special architecture and are not currently classified . Within architectures , X-groups are ordered by structural similarity between them . The ECOD X-level groups domains that may be homologous as is frequently suggested by similarity of their spatial structures . A domain's overall structure is traditionally referred to as its ‘fold’ . Fold similarity usually refers to general resemblance in both architecture and topology and can result from either common ancestry ( homology ) or physical/chemical restrictions ( analogy ) [35] , [36] , [37] . Both SCOP and CATH have a fold level in the hierarchy: “SCOP fold” and “CATH topology” . However , the definition of fold can be subjective [23] , and fold is a geometrical concept without explicit evolutionary meaning . Therefore , ECOD generally avoids the fold concept . However , domains that share strong overall architectural and topological similarity and are possibly homologous , but which lack further evidence to exclude analogy , are attributed to the same X-group but different H-groups . The conceptual difference between ECOD X-group and SCOP fold can be shown , for example , in the classification of domains with a ferredoxin-like topology . In SCOP , the ‘Ferredoxin-like’ fold is a large assembly of various superfamilies that share the ( βαβ ) ×2 topology . Among all these superfamilies , 4Fe-4S ferredoxins seem unique for their small size and cysteine-rich nature ( cysteines are used to coordinate the Fe-S clusters ) . Thus we suspect 4Fe-4S ferredoxins have an independent evolutionary origin and keep 4Fe-4S ferredoxins and other superfamilies in separate X-groups . On the other hand , although domains in the SCOP fold ‘Ribosomal proteins S24e , L23 and L15e’ do not have the ferredoxin-like ( βαβ ) ×2 topology , their structures can easily be transformed into that topology by a circular permutation . Their structural similarity and functional similarity with the ‘RNA-binding domain , RBD’ superfamily in SCOP ‘Ferredoxin-like’ fold may imply homology . Therefore , ECOD classifies ‘Ribosomal proteins S24e , L23 and L15e’ and ‘RNA-binding domain , RBD’ as two H-groups in the same X-group as possible homologs . When further evidence coming either from additional sequences or 3D structures accumulates , classification decisions are adjusted to agree best with all available data . We examined the distribution of domains mapped to SCOP folds and CATH topologies among ECOD X-groups . Of 1 , 799 ECOD X-groups , 598 include domains from only one SCOP fold and 564 include domains from only one CATH topology , reflecting agreement between classifications for these groups . 89 ECOD X-groups contain domains from multiple SCOP folds and 315 X-groups include domains from multiple CATH topologies . For example , the SCOP folds c . 1-TIM beta/alpha-barrel and c . 6-7-stranded beta/alpha barrel both contain domains mapped to the ECOD TIM beta/alpha barrel X-group . ECOD unifies such groups due to their shared structural similarity ( 7- versus 8- stranded ) and similar locations of functional sites , but with insufficient evidence of homology to belong to the same H-group . 935 ECOD X-groups are not mapped to any SCOP fold , whereas 1 , 014 ECOD X-groups are not mapped to any CATH topology . The majority of these unmapped X-groups are simply due to proteins that are not classified by SCOP or CATH ( 722 and 872 X-groups , respectively ) ; the remainder are shared proteins that are partitioned differently . Taken together , these results suggest that ECOD tends to merge both SCOP folds and CATH topologies into X-groups . An ECOD H-group can contain more distant homologous links than the equivalent SCOP superfamily or CATH homologous superfamily . Although the majority of ECOD H-groups contain only a single SCOP superfamily ( 88% ) or CATH homologous superfamily ( 81% ) , some H-groups contain many more ( Fig . 3 ) . For example , the Immunoglobulin-related and the Rossmann-related H-groups contain the most SCOP superfamiles ( 47 and 28 , respectively ) and CATH homologous superfamilies ( 81 and 40 , respectively ) . Superfamilies were merged based on multiple high-scoring homologous links between domains . These merges reflect the homology between domain members of these previously split groups . In total , 53 ECOD H-groups contain domains from two or more SCOP folds , and these H-groups contain domains from 151 unique SCOP folds , indicating that fold change in evolution of protein structures is not a very uncommon phenomenon . Similarly , 169 ECOD H-groups contain domains from two or more CATH topologies , and these H-groups contain domains from 357 unique CATH topologies . Additionally , 36 H-groups contain domains mapped to more than one CATH class , indicating homologous domains that nonetheless contain fairly different topologies . To readily incorporate the observation that homologs can adopt different folds , ECOD has a topology ( T- ) level below the homology ( H- ) level . As a result , homologs with different topologies that SCOP necessarily separates into different folds ( and thus different superfamilies ) are unified in the same H-group but different T-groups in ECOD . For example , β-propellers are comprised of differing numbers of repeated β-meanders , all of which are evolutionarily related . The five different beta-propeller folds outlined in SCOP are organized in ECOD into a single H-group , with child T-groups for domains with differing number of blades [38] . Also , the domain contents of 11 SCOP folds are organized into multiple T-groups under the Rift-related H-group in the cradle-loop barrel X-group [39] . If we find sufficient evidence for homology between these proteins this consideration results in merging not only SCOP superfamilies , but also SCOP folds . Within T-groups , ECOD organizes domains into families based on sequence similarity . We employ Pfam as the standard for family definition . ECOD domains were attributed to Pfam families by HMMER3 [31] . Therefore , the majority of ECOD F-groups are simply Pfam families . However , not all protein domains with known structure can be attributed to the current version of Pfam by sequence similarity . Those domains are grouped into families by HHsearch as outlined in Materials and Methods . As a result , ECOD contains 8 , 947 F-groups , 7 , 156 of which can be mapped to Pfam families , and 1 , 622 composed of homologous domains not mapped to any Pfam family . Summary statistics for the ECOD database as of July 31stth , 2013 ( version 22b ) are presented in Table 1 . The majority of the 317 , 021 domains in ECOD were assigned automatically to a smaller set of 15 , 969 manually curated domain representatives . Domains in ECOD were derived from five sources: 1 ) domains originally in SCOP ASTRAL40 , inherited and reclassified manually in ECOD ( 11 , 462 ) , 2 ) domains originally in SCOP , but not in the ASTRAL40 set , mapped by MUSCLE alignment with their ASTRAL representative ( 98 , 702 ) , 3 ) novel domains not contained in SCOP , usually from chains deposited to the PDB in the intervening period between the release of SCOP v1 . 75 and ECOD , manually curated and added to the representative set ( 4 , 373 ) , 4 ) domains automatically added to ECOD by detection of homology by pairwise sequence or structure search ( 153 , 381 ) , and 5 ) domains added to ECOD by MUSCLE alignment of non-representative sequences to closely related ECOD representatives ( 48 , 817 ) . The vast majority of domains classified in ECOD have been added by automatic methods . ECOD provides for domains which are assembled from multiple PDB chains , either due to photolytic cleavage ( i . e . order-dependent assembly ) or obligate multimers ( i . e . order-independent assemblies ) . For order-independent assemblies , we distinguish between those domains where the assembly is primarily relevant for display , or appears to be biologically necessary . These are fairly rare in the database; only 132 representative order-independent assemblies have been defined . At the time of writing , 100% of PDB depositions could be accounted for in the ECOD classification ( including those members of the special architectures ) . We also compare ECOD to the most recent releases of SCOP and CATH . ECOD , SCOP , and CATH differ in domain partition strategy , classification hierarchy , and simply in the number of structures considered . At the time of writing , ECOD classifies 93 , 663 PDB depositions containing 239 , 303 protein chains , SCOP 1 . 75 contains 38 , 221 PDBs and 85 , 141 chains , and CATH v3 . 5 contains 51 , 334 PDBs and 118 , 792 chains . Of those chains classified in ECOD that are not in SCOP ( and not in a special architecture ) , 137 , 794 were automatically classified and 2 , 484 were classified manually . Of those chains classified in ECOD , but not in CATH ( and not in a special architecture ) , 106 , 474 were automatically classified and 2 , 521 were classified manually . The growth of the PDB over time is compared to the number of structures classified in ECOD , CATH , and SCOP ( Fig . 4 ( a ) ) . The difference between the number of structures in the PDB and those in the main architectures of ECOD can be primarily accounted for by the number of structures contained in ECOD special architectures ( i . e . coiled-coil , peptide , non-peptide polymers , and low-resolution structures that could not be classified by sequence ) . The growth of the hierarchical levels from 2000–2013 indicates that although evolutionary distinct groups ( i . e . X- and H- groups ) are being discovered at a steady pace , the predominant source of new domains in ECOD is from sequence families ( F-groups ) being associated with existing homologous groups ( Fig . 4 ( b ) ) . Since the July 2013 version , whose statistics are presented here , the subsequent 25 weekly releases by the PDB have been automatically classified ( Fig . 5 ) . Each week , protein chains are clustered at 95% redundancy , representatives for those non-redundant chains are classified; those remaining chains are classified when the initial automatic and manual classification pass are completed . For each weekly update , the majority ( ∼89% ) of non-redundant ( <95% ) chains can be partitioned and assigned automatically ( 134 . 1±40 . 4 ) . Those chains that cannot be resolved automatically are manually curated . On average , 11 . 7±4 . 9 chains per week were classified as manual representatives in ECOD , whereas 5 . 1±3 . 2 were chains not containing domains ( i . e . peptides , coiled-coils , or fragments ) that were resolved by assignment to special categories or other methods that did not modify the hierarchy . Overall , the majority of protein chains in weekly PDB releases can be classified automatically into ECOD . We analyzed the distribution of domains in hierarchical levels in ECOD . The most populated homologous groups ( H-groups ) are placed in context with their architecture in ECOD ( Fig . 6 ( a ) ) and are also ranked by population ( Fig . 6 ( b ) ) . The Ig-related and Rossmann-related H-groups , in addition to containing the most merged SCOP and CATH homologous groups , are the most populated H-groups in ECOD . The merging of many previously distinct helix-turn-helix ( HTH ) SCOP superfamilies in ECOD boosts the population of this H-group considerably compared to its original SCOP population . The inset ( Fig . 6 ( b ) ) shows those most populated H-groups by number of F-groups . Where many sequence families have been merged by distant homology , such as the RIFT-related or Immunoglobulin-related domains , H-groups will contain many F-groups . In ECOD , as opposed to SCOP or CATH , there exist fewer distinct homologous groups with related topologies , as many of these groups have been linked by homology . For example , in ECOD , there is a single Rossmann-related H-group among the most populated ( top 15 ) groups , whereas in the most populated SCOP superfamilies or CATH homologous superfamilies , there are two ( NAD ( P ) -binding Rossmann fold domains and SAM methyltransferases ) and four ( 3 . 40 . 50 . 720 , 3 . 40 . 50 . 1820 , 3 . 40 . 50 . 150 , and 3 . 40 . 50 . 2300 ) , respectively . We compared our H-groups to SCOP superfamilies and folds by considering sequence and structure similarity of domain pairs within each level . ECOD manual representatives and ASTRAL40 domains were evaluated by HHsearch to reflect sequence similarity and TMalign to reflect structure similarity [21] , [40] . SCOP superfamilies tend to contain more close homologs that can be detected by sequence homology search methods than ECOD H-groups ( Fig . 7 ( c ) ) . Domains classified in SCOP folds ( excluding pairs from the same superfamily ) emphasize structural similarity , as the distribution is mostly populated in the low sequence similarity region and the peak shifts right compared with others ( Fig . 7 ( a , b ) ) . On the other hand , as ECOD H-group readily incorporates homologous links from SCOP superfamilies and also many remotely homologous relationships that were previously overlooked , its peak sizes lie between SCOP fold and superfamily in high and low sequence similarity regions . Also it is worth noting that the peak of ECOD H-group does not have the right shoulder in the intermediate sequence similarity group but has a relatively evident left shoulder in the high sequence similarity group ( Fig . 7 ( b , c ) ) , which potentially supports the idea that ECOD classification is homology-centric . We compared the domain partition observed in ECOD , SCOP , and CATH . Domain partition strategy can differ markedly between classifications , depending generally on whether the presence of compact structural units or overall sequence similarity is emphasized . The number of domains per chain observed in the domain classifications is presented in Figure 8 ( a ) . ECOD splits more protein chains ( 29% ) into multiple domains than SCOP ( 23% ) , but splits slightly less than CATH ( 35% ) . The size distribution of domains in ECOD , SCOP , and CATH was compared ( Fig . 8 ( b ) ) . ECOD favors slightly shorter domains than SCOP , and favors slightly longer domains over CATH , but the size distributions are very similar . These results are consistent with the differences in domain definition strategy employed by different classifications . CATH emphasizes on structural integrity of the domain and its structural separation from other domains , SCOP focuses on the occurrence of an individual domain in different domain combinations , and ECOD attempts to find a compromise between these two strategies . The difference in homologous links among equivalent domains was analyzed in ECOD , SCOP , and CATH . We define equivalent domains as those that share 80% residue coverage in all classifications . This subset of domains contains those domains whose partition is similar among classifications , but whose classification and homologous cluster size differ . We then analyze whether those domains that share a homologous link within one classification also share that link in other classifications . For the purposes of this analysis , only SCOP domains from canonical SCOP classes [a–d] are considered . Of the total domains in ECOD , 67 , 559 are defined equivalently ( by 80% residue coverage ) in SCOP and CATH . As many of these domains are identical or near identical in sequence , only domains with less than 95% sequence identity are used . There are 9 , 523 equivalent , non-redundant domains , shared among SCOP , CATH , and ECOD . Any pair of those equivalent domains belonging to the same H-group is considered to be homologous , 1 , 030 , 085 of these homologous domain pairs were observed in ECOD . Similar analysis was performed on SCOP superfamilies and CATH homologous superfamilies , where 711 , 894 and 680 , 726 homologous domain pairs were observed respectively . On average , 49 . 5% of domain pairs were shared between classifications , 36 . 6% of domain pairs were only observed in ECOD , 11 . 4% of domain pairs were observed only between ECOD and CATH ( Fig . 9 ) . Negligible numbers of domain pairs were observed in SCOP only , CATH only , or SCOP/CATH only . These results reflect a set in which most known homologous relationships among similarly partitioned domains are similar in ECOD as in SCOP and CATH . Additionally , ECOD catalogs many homologous relationships ( among these similarly partitioned domains ) that are not observed elsewhere . SCOP recently diverged into two separate projects: SCOPe [14] , which continues to update the original SCOP hierarchy using conservative automated methods , and SCOP2 [13] , which is a dramatic reimagining of protein classification away from a hierarchal tree to a network model . We compared both of these more recent SCOP databases to ECOD . SCOP2 ( February prototype version ) eschews the traditional classification model; individual residues can be classified at multiple nodes in the network . We considered all SCOP2 domains , regardless of level , in comparison to ECOD . Of 995 PDBs and 1010 chains classified in SCOP2 , equivalent domains were found in 725 PDBs and 732 chains . 70% of SCOP2 domains defined at the sequence family level were ECOD-equivalent . Conversely , only 56% of SCOP domains defined at the structural fold level ( 340/605 ) and sequence superfamily ( 272/482 ) level were equivalent to an ECOD domain . Only 61 of 257 domains defined at the hyperfamily ( HF ) level , are equivalent any domain in ECOD . Only 121 of 2 , 973 ECOD H-groups in this comparison were mapped to any domain in SCOP2 . In general , the incomplete coverage of SCOP2 makes general statements about differences from ECOD premature . SCOPe ( v2 . 03-stable ) uses a conservative automated method to add domains to the SCOP v1 . 75 hierarchy . Since both ECOD and SCOPe were derived from SCOP v1 . 75 , we were particularly interested in classification of recent chains . ECOD v49 and SCOPe v2 . 03 ( stable ) contain 261 , 704 and 163 , 351 domains from shared protein chains , respectively . Of those SCOP-mapped ECOD domains , 94 , 292 were derived from SCOP v1 . 75 domains and 57 , 929 were independently classified . 27 , 142 ECOD domains derived from SCOPe shared chains do not map to any SCOPe domain , reflecting direct differences in domain partition strategy between SCOPe and ECOD . 1 , 493 SCOPe domains arise from structures classified only by SCOPe , but these structures are dominated by peptides and coiled-coils , regions that are not classified as domains by ECOD . 9 , 164 ECOD domains were derived from SCOP v1 . 75 domains , but are not mapped to SCOPe . These domains were generally the result of subdivision of a larger SCOP domain . There is a core set of domains that are shared by SCOPe and ECOD , both arising due to their shared origin and also due to independent classification of more recent domains . The differences in domain partition likely arise from differences in treatment of domain duplication and subdomains and are a potential target for further study . We consider the growth-over-time analysis of ECOD in the context of the domain mapping between ECOD , SCOP and CATH ( Fig . 10 ) . Where an ECOD level ( X- , H- , T- , or F-group ) contains one or more domains with a mapping to a SCOP or CATH domain , we remove that level from consideration . We then re-plot the growth over time of ECOD using only those groups that contain no mapping to domains from other classifications . There is marked increase in novel ECOD classifications beginning in January 2005 . The most recent deposition dates contained in SCOP 1 . 75 and CATH 3 . 5 are October 2008 and August 2011 , respectively . However , the increase in novel classifications begins when the total PDBs and the PDBs classified in SCOP and CATH begin to diverge . The novel H-groups in ECOD ( 997 ) account for nearly 45% of total H-groups in ECOD . Those F-groups with no manual representative ( where all domains were assigned automatically ) are assigned a provisional manual representative . The majority of these automatically generated F-groups with no manual representative are derived from known Pfam families ( Fig . 11 ) . The increase in novel hierarchical levels in ECOD clearly demonstrates the value of an updated and comprehensive domain classification . In addition to comparison of broad statistics of ECOD , we also present three examples of homologous relationships recorded in ECOD but not observed in other classifications . We consider any two homologous domains to have a “homologous link . ” Firstly , we demonstrate the homologous link between SAM-dependent methyltransferases and NAD ( P ) -binding Rossmann-fold domains . These domains share topological connections , but a strand invasion causes them to bear distinct topologies , nonetheless , they share strong sequence similarity . Secondly , we show how members of the cysteine-rich domains of Frizzled share homology with other domain families that can primarily be detected by conserved patterns of cysteines . Finally , we describe a novel homologous link between Duf371 and the GutA-like PTSIIA component domain families within the topologically diverse cradle-loop barrel X-group . Each of these distinct examples demonstrates how the particular focus of distant homology in ECOD can reveal previously unknown relationships . ECOD contains many homologous links that are not recorded in other classification databases . One example is the relationship between S-adenosyl-L-methionine-dependent methyltransferases ( SAM MTases ) and NAD ( P ) -binding Rossmann-fold domains ( Rossmann domains ) . SAM MTases methylate a wide range of substrates using the methyl group donated by the cofactor SAM , which is comprised of an adenosine nucleoside and a methionine amino acid joined together . Rossmann domains are found in many oxidoreductases that transfer electrons between substrates and the cofactor NAD ( P ) , which is comprised of a nicotinamide nucleotide and an adenine nucleotide joined together . Thus , SAM and NAD ( P ) share the adenosine part but differ in the other half , and the two enzyme superfamilies exploit the dissimilar parts of the cofactors to catalyze different reactions [41] , [42] , [43] . SAM MTases have a consensus structure of a 7-stranded β-sheet sandwiched between connecting α-helices ( strand order 3214576 with strand 7 antiparallel to the other six strands , Fig . 12 ( a ) ) [44] . Rossmann domains have a consensus structure of a parallel 6-stranded β-sheet sandwiched between connecting α-helices ( strand order 321456 , Fig . 12 ( b ) ) [45] . Thus , the SAM MTase structure can be viewed as Rossmann domain structure with a strand invasion: the additional strand 7 is inserted into the β-sheet between strands 5 and 6 . In SCOP , SAM MTases and Rossmann domains are classified in different folds ( and therefore different superfamilies , SAM MTases: c . 66 . 1; Rossmann domains: c . 2 . 1 ) , while in CATH , they are in the same topology group but different homology groups ( SAM MTases: 3 . 40 . 50 . 150; Rossmann domains: 3 . 40 . 50 . 720 ) . Although both SCOP and CATH indicate by their classification that SAM MTases and Rossmann domains are not homologous , literature suggests that they are actually related [46] , [47] . As noted in reference [41] , the overall structural similarity between SAM MTases and Rossmann domains is reflected in the observation that they are reciprocally the closest DALI hits to each other . In addition , SAM MTases and Rossmann domains bind their respective cofactors in a very similar fashion: the common adenosine part of the cofactors resides on top of a glycine-rich loop between the first strand and the first helix , and the adenosine ribose hydroxyls usually form hydrogen bonds with a conserved aspartate or glutamate residue at the end of the second strand ( Fig . 12 ( a , b ) ) [41] , [45] , [46] . Indeed , the sequence-based homology detection algorithm HHsearch [21] and server HHpred [48] also provide statistical evidence that SAM MTases and Rossmann domains are related . In Cytoscape [26] display of SCOP domains and high-scoring links between them , numerous links with HHsearch probability above 90% exist between SAM MTases and Rossmann domains . In HHpred runs , for instance , when the Rossmann-domain in formaldehyde dehydrogenase ( SCOP domain d1kola2 , classified in c . 2 . 1 , Fig . 12 ( b ) ) is submitted as query to search against scop95_v1 . 75B database with secondary structure scoring turned off , the top hits within the same c . 2 . 1 superfamily are followed by a region of mixed hits from both Rossmann domains superfamily ( c . 2 . 1 ) and SAM MTases superfamily ( c . 66 . 1 ) . The highest-scoring hit from SAM MTases superfamily is hypothetical protein TM0748 ( SCOP domain d1o54a_ ) with a 97 . 89% probability , E-value 9 . 4e-09 , and identities 17% out of 110 aligned residues . Another SAM-MTase , ribosomal protein L11 methyltransferase ( SCOP domain d2nxca1 , Fig . 13 ( a ) shows a same domain d2nxea1 with SAM bound ) , is detected with a 97 . 33% probability , E-value 3 . 4e-07 , and identities 23% out of 102 aligned residues . Based on overall structural similarity , cofactor-binding resemblance , the number of confident homologous links observed between domains in each group , and statistically significant sequence similarity , ECOD classifies SAM MTases and Rossmann domains in the same homology ( H- ) group but different topology ( T- ) groups . Frizzled receptors possess an extracellular cysteine-rich domain ( FZ-CRD ) for binding the Wnt ligands . FZ-CRD , as a mobile evolutionary module , has been found in other proteins such as the Smoothened receptor in Hedgehog signaling , secreted Frizzled-related proteins ( SFRPs ) , and receptor tyrosine kinases MuSK and ROR . Sequence similarity searches and structural comparisons revealed distant similarities among FZ-CRD , Niemann-Pick type C1 protein ( NPC1 ) that functions in cholesterol transportation , folate receptors and riboflavin-binding proteins ( FRBPs ) [17] . Recently , the core structures of two glypicans , proteoglycan molecules that regulate the signaling of a number of morphogens , were solved [49] , [50] . Interestingly , comparative structural analyses suggested that glypicans also contain a cysteine-rich domain homologous to FZ-CRD and NPC1 [51] . Domains homologous to FZ-CRD and NPC1 have a wide distribution in eukaryotes , as they were also found in a number of other protein families currently without available structures , such as Hedgehog interacting proteins ( HHIPs ) , RECK ( reversion-inducing-cysteine-rich protein with Kazal motifs ) proteins , the calcium channel component Mid1 in fungi , and the uncharacterized FAM155 proteins in metazoans [51] . The ECOD database unifies available structures of FZ-CRD , NPC1 , folate receptor , and glypicans in one homologous group based on compelling sequence and structural similarities among them [17] , [51] . These domains share similar disulfide bond patterns and adopt a similar overall structure fold with four core α-helices . Structural studies of three FZ-CRDs , in mouse Frizzled8 ( Fig . 13 ( a ) ) [52] , mouse SFRP3 [52] , and rat MuSK [53] ( Fig . 13 ( b ) ) , revealed a common fold mainly consisting of four core α-helices ( H1–H4 in Fig . 13 ) . These FZ-CRD domains exhibit conservation of ten cysteines with a general pattern of ‘C*C*CX8CX6C*CX3CX6 , 7C*C*C’ ( C: conserved cysteine; *: a variable number of residues , Xn: n residues , and Xm , n: m to n residues ) ( Fig . 13 ( g ) ) . The disulfide connectivity patterns among the ten conserved cysteines are C1–C5 ( between the first and fifth conserved cysteines ) , C2–C4 , C3–C8 , C6–C10 , and C7–C9 ( marked by black * , # , + , = , and & signs , respectively in Fig . 13 ( a , b and g ) ) . The homologous cysteine-rich domain in glypicans possesses 12 conserved cysteines with a similar pattern of ‘C*C*CC*CX8CX2 , 3C*CX3CX6C*C*C’ ( Fig . 13 ( c , g ) ) similar to that of the FZ-CRD . Such a pattern and disulfide bond connectivity ( C1–C3 , C2–C5 , C4–C7 , C5–C10 , C8–C12 , and C9–C11 ) ( Fig . 13 ( g ) ) in glypicans are also seen in the structures of FRBPs including a folate receptor [54] ( Fig . 13 ( d ) ) and a riboflavin-binding protein [55] . The structure of the cholesterol-binding domain of NPC1 [56] possesses eight of these 12 conserved cysteines , while lacking two disulfide bonds formed by C8–C12 and C9–C11 in glypicans , FRBPs ( Fig . 13 ( c ) ) . Together , the homologous cysteine-rich domains in Frizzled , NPC1 , FRBP , and glypicans define a diverse superfamily of extracellular protein domains with an ancient eukaryotic origin and potential ligand-binding activities . Duplication and divergence of such a domain have resulted in a number of families with various functions in eukaryotic membrane transport and signaling . Despite overall similarity in fold and disulfide connectivity patterns , high structural divergence , reflected by low Dali Z-scores ( Fig . 13 ( f ) ) , was observed between some of these structures . The ECOD classification of this homologous group of proteins includes recently solved structures such as glypicans [49] , [50] and the folate receptor [54] . In contrast , both the SCOP and CATH databases only have structures of FZ-CRDs from Frizzled receptors and SFRPs and do not include the structures of FZ-CRD of MuSK [53] , NPC1 [56] , glypicans , and folate receptor ( although the sequence of MuSK is classified in the related CATH FunFam database ) . ECOD establishes a previously unrecognized homologous link between a domain of unknown function ( Duf371 , PDB:3cbn ) and the bacterial GutA-like PTS system glucitol/sorbitol-specific IIA component ( PTSIIA , PDB:2f9h ) . While Duf371 is absent in SCOP , CATH classifies its fold ( 2 . 60 . 120 . 630 ) separately from that of PTSIIA ( 2 . 40 . 33 . 40 ) . Duf371 forms an 8-stranded β-barrel from the intertwined β-strands of a tandem duplication ( Fig . 14 ( a ) ) . The duplicated structure elements can be superimposed ( RMSD 1 . 3 Å ) , with a conserved His-containing motif from the N-terminal repeat overlapping a somewhat less conserved His-containing motif from the C-terminal repeat ( Fig . 14 ( b ) ) . Accordingly , PSI-BLAST [57] provides sequence evidence for this duplication , with both halves of the Duf371 query ( PBD:3cbn , gi|169404770 ) confidently detecting the Methanocaldococcus fervens sequence Mefer0473 ( 3cbn[A:6-141] hits Mefer0473 with E-value 1e-30 in the first iteration , and 3cbn C-terminal range [A:77-142] hits with E-value 0 . 003 in second iteration ) . PTSIIA adopts a similar β-barrel topology as Duf371 and is noted in SCOP as consisting of two intertwined structural repeats ( Fig . 14 ( c ) ) . The overside connections between adjacent β-strands of the duplicated structure motifs in Duf371 and PTSIIA do not frequently appear in barrel architectures and distinguish the two folds . A similar overside connection occupies the N-terminal half of pyruvate kinase ( PK ) β-barrel domain-like folds ( ECOD/SCOP domain e1pklA1/d1pkla1 ) . The PK barrel adopts a duplicated topology like PTSIIA and Duf371 , although it lacks the C-terminal overside connection . The absence of this structural element in PK results in a 7-stranded β-barrel ( Fig . 14 ( d ) ) . The PK barrel half lacking the overside connection forms a ββxβ unit characteristic of the cradle-loop barrel metafold , which encompasses homologous folds of different topologies [39] . Based on the presence of a GD-box motif , the PK barrel was described as related to ancient RIFT-related folds ( i . e . translation protein EF-Tu PDB: 1d2e ) by a strand invasion of the N-terminal ββxβ unit that creates the overside connection [39] . Interestingly , the GD-box was also identified in both halves of PTSIIA ( N-terminal GD and C-terminal GT ) [58] , but is not present in Duf371 . Structural similarity between Duf371 and PTSIIA is evidenced by their being reciprocal top Dali hits of each other ( Dali Z-score 6 ) , with the next best hits being to various RIFT-related homologs such as the PH barrel . The resulting structural alignment of the PTSIIA C-terminal sequence with both Duf371 sequence repeats is shown in Figure 14 ( e ) . A conserved C-terminal PTSIIA His residue ( highlighted in black ) marks the potential active site ( Fig . 14 ( f ) ) . Although the corresponding site in the Duf371 C-terminal repeat sequence is less conserved , an almost invariant threonine in the Duf371 N-terminal repeat aligns to the proposed PTSIIA functional site . Accordingly , the two folds may be related by a circular permutation of the structural repeats , maintaining a similar conserved active site position within the symmetry-related fold of Duf371 ( Figure 14 ( g ) ) . Considering the distinct topology of the duplicated structural motifs containing unusual overside connections , the unique way the two motifs tangle together to form an 8-stranded barrel , and the maintenance of similar active site positions , ECOD classifies PTSIIA and Duf371 as homologs in the same T-group within the RIFT-related H-group . The ECOD database summarizes our views about partitioning of protein structures into domains and this evolutionary classification is a comprehensive resource for the research community . Data about a protein can be retrieved by PDB ID , keyword ( s ) , or protein sequence search . Protein domains of interest are placed close to their close homologs , facilitating analysis of closely related protein structures . Information about more distant homologs is available by browsing representatives of this protein's homology group . ECOD database emphasizes distant evolutionary relationships that otherwise cannot be found . Finally , it is the only classification of protein domain structures that is kept current with the PDB , and every structure is classified with a week delay from its release by the PDB . This feature is significant because other classifications lag behind in updates and researchers are frequently interested in the newest protein structures . Future developments of ECOD will include incorporation of protein sequences without experimentally determined structures to cover as much of the protein world as possible . | Protein structural domain databases offer a vital resource for structural bioinformatics . These databases provide functional inference for homologous structures , supply templates for structural prediction experiments , and differentiate between homologs and analogs . The rate of structure determination and deposition has increased dramatically over recent years , overwhelming the ability of current classifications to incorporate all new structures . We have developed a fast and reliable methodology for updating domain databases automatically , and created a revised hierarchy for domain classification that emphasizes evolutionary relationships . By classifying all known structures in our database with continuing automatic updates , we provide an up-to-date alternative to current resources . We illustrate several concepts that guided our classification scheme with examples of homology between domains in ECOD that are not observed in other resources . | [
"Abstract",
"Introduction",
"Methods",
"Results/Discussion"
] | [
"protein",
"structure",
"comparison",
"biochemistry",
"proteins",
"protein",
"structure",
"protein",
"structure",
"databases",
"biology",
"and",
"life",
"sciences",
"protein",
"domains"
] | 2014 | ECOD: An Evolutionary Classification of Protein Domains |
The brain constantly infers the causes of the inputs it receives and uses these inferences to generate statistical expectations about future observations . Experimental evidence for these expectations and their violations include explicit reports , sequential effects on reaction times , and mismatch or surprise signals recorded in electrophysiology and functional MRI . Here , we explore the hypothesis that the brain acts as a near-optimal inference device that constantly attempts to infer the time-varying matrix of transition probabilities between the stimuli it receives , even when those stimuli are in fact fully unpredictable . This parsimonious Bayesian model , with a single free parameter , accounts for a broad range of findings on surprise signals , sequential effects and the perception of randomness . Notably , it explains the pervasive asymmetry between repetitions and alternations encountered in those studies . Our analysis suggests that a neural machinery for inferring transition probabilities lies at the core of human sequence knowledge .
From bird song to music , sea waves , or traffic lights , many processes in real life unfold across time and generate time series of events . Sequences of observations are therefore often underpinned by some regularity that depends on the underlying generative process . The ability to detect such sequential regularities is fundamental to adaptive behavior , and many experiments in psychology and neuroscience have assessed this ability by appealing to tasks involving sequences of events . Various effects suggestive of local sequence learning have been consistently reported , even when experimental sequences are devoid of any regularity ( i . e . purely random ) and restricted to only two possible items or actions . Studies of “novelty detection” for instance show that the mere exposure to a sequence of stimuli elicits reproducible “novelty” brain responses that vary quantitatively as a function of the item infrequency and divergence from previous observations [1–11] . Behaviorally , studies using two-alternative forced-choices have revealed “sequential effects” , i . e . fluctuations in performance induced by local regularities in the sequence . For instance , subjects become faster and more accurate when they encounter a pattern that repeats the same instructed response , or that alternates between two responses , and they slow down and may even err when this local pattern is discontinued [12–23] . Finally , studies asking subjects to produce random sequences or to rate the apparent “randomness” of given sequences , show a notorious underestimation of the likelihood of alternations [24–29] . Here , we propose a model that provides a principled and unifying account for those seemingly unrelated results , reported in various studies and subfields of the literature quoted above . We adopt a Bayesian-inference approach [30–37] which relies on three pillars . The first one is that information processing in the brain relies on the computation of probabilities [30 , 38–43] . A second pillar is that these probabilistic computations closely approximate Bayes’ rule . This means that , in order to infer the hidden regularities of the inputs it receives , the brain combines the likelihood of observations given putative regularities and the prior likelihood of these regularities [44] . A third pillar is the predictive and iterative nature of Bayesian computations: once the hidden regularities of the inputs are inferred , the brain uses them to anticipate the likelihood of future observations . Comparison between expectations and actual data allows the brain to constantly update its estimates–a computational mode termed “active inference” [45 , 46] . To apply this general framework to sequences , one must identify the models that the brain computes when learning from a sequence . One possibility that we explore here is that there are core building blocks of sequence knowledge that the brain uses across many different domains [47] . Throughout this paper , our goal is to identify the minimal building block of sequence knowledge . By “minimal” , we mean that a simpler hypothesis would demonstrably fail to account for experimental effects such as surprise signals , sequential effects in reaction times and the biased perception of randomness . Our proposal can be succinctly formulated: the brain constantly extracts the statistical structure of its inputs by estimating the non-stationary transition probability matrix between successive items . “Transition probability matrix” means that the brain attributes a specific probability to each of the possible transitions between successive items . “Non-stationary” means that the brain entertains the hypothesis that these transition probabilities may change abruptly , and constantly revises its estimates based on the most recent observations . We formalized this proposal into a quantitative model , which we call the “local transition probability model” . As we shall see , this model predicts that expectations arising from a sequence of events should conform to several properties . We list these properties below and unpack them , one at a time , in the Results section . To test whether the proposed model is general , we simulated the results of five different tasks previously published [1 , 2 , 9 , 20 , 25] . They differ in the type of observable: either reaction times [2 , 20] , judgment of randomness [25] , functional MRI signals [2] or EEG signals [1 , 9] . They also differ in the experimental task: either passive listening [1] , two-alternative forced-choice [2 , 9 , 20] or subjective ratings [25] . Last , they also differ in the way stimuli are presented: sequential and auditory [1] , sequential and visual [2 , 9 , 20] or simultaneous and visual [25] . Yet , as we shall see , all of these observations fall under the proposed local transition probability model . We also tested the minimal character of the model , i . e . the necessity of its two main hypotheses , namely , that transition probabilities are learned , and that such learning is local in time . Instead of transition probabilities , simpler models previously proposed that subjects learn the absolute frequency of items , or the frequency of alternations [9 , 20 , 23 , 48] . We evaluated the predictive ability of these statistics , whose relationships are illustrated in Fig 1 . We also tested the non-stationarity hypothesis by comparing the local transition probability model with other models that assume no change in the quantity they estimate , or a simple forgetting rule ( as illustrated in Fig 2 ) . This model comparison is not exhaustive since many proposals were formulated over the past fifty years; however , it allows to test for the necessity of our assumptions . To anticipate on the results , we found that only a learning of local transition probabilities was compatible with the large repertoire of experimental effects reported here .
The local transition probability model assumes that several brain circuits involved in sequence learning entertain the hypothesis that the sequence of items has been generated by a “Markovian” generative process , i . e . only the previous item yt–1 has a predictive power onto the current item yt . Those circuits therefore attempt to infer the “transition probability matrix” which expresses the probability of observing a given item , given the identity of the preceding one . Further , the model is local in time in that it assumes that the transition probabilities generating the observations may change over time ( some theorists call this a model with a “dynamic belief” ) . More precisely , it assumes that there is a fixed , non-zero , probability pc that the full matrix of transition probabilities changes suddenly from one observation to the next ( see Fig 2A ) . Therefore , at any given moment , the current and unknown generative transition probabilities must be estimated only from the observations that followed the last change . Note that the occurrence of such changes is itself unknown–the model must infer them . Bayes’ rule and probabilistic inference allow to solve this challenging problem optimally . Intuitively , the optimal solution discounts remote observations and adjusts the strength of this discounting process to the trial-by-trial likelihood of changes . The estimation of transition probabilities is therefore “local” and non-stationary . In this paper , we contrast the local transition probability model with an alternative model which entertains a “fixed belief” , i . e . which assumes that the generative process never changes ( pc is exactly 0 ) . The fixed-belief assumption greatly simplifies the estimation of transition probabilities , which boils down to counting the occurrence of each transition between any two items–but it prevents the model from adapting to the recent history of events . We also consider models which only approximate the Bayes-optimal “dynamic belief” inference . One such model is a forgetful count that discards old observations or weights recent observations more than past ones [23] . The count may be forgetful because it is limited to a fixed window of recent observations ( the “windowed model” ) , or because it involves a leaky integration , such that previous observations are progressively forgotten . Importantly , the time scale over which forgetting occurs is fixed at a preset value and therefore cannot be adjusted to the trial-by-trial likelihood of changes , unlike the optimal solution . The leaky integration and the Bayes-optimal dynamic belief are two algorithms , each with a single free parameter , that result in local estimates of statistics . Both yield similar results in the present context , we therefore refer to both as the “local transition probability model” . We reported the results for the leaky integration in the main text and the results for the Bayes-optimal dynamic belief as supplementary information ( see S1 , S2 and S3 Figs ) . These different inference styles of transition probabilities–fixed belief , dynamic belief , leaky integration–are depicted in Fig 2B . For comparison , we also implemented variants that resort to the same inference styles but estimate a different statistic: either the absolute frequency of items , or the frequency of alternation between successive items . It is important to note that these statistics are simpler than transition probabilities , because the information about the frequency of items and the frequency of alternations is embedded in the larger space of transition probabilities ( see Fig 1A ) . Transition probabilities also dictate the frequency of ordered pairs of items ( see S1 Text for Supplementary Equations ) . Some of the models included in our comparison were proposed by others , e . g . the fixed-belief model that learns item frequencies was proposed by Mars and colleagues [4] and dynamic-belief models were also proposed by Behrens , Nassar and colleagues [49 , 50] for learning item frequencies , and by Yu and Cohen [23] for learning the frequency of alternations . The local transition probability model makes several predictions: In the following , we characterize these predictions in greater detail in specific experimental contexts , and we test them against a variety of data sets and by comparing with simpler models . The P300 is an event-related potential that can be easily measured on the human scalp and is sensitive to the surprise elicited by novel or unexpected stimuli . Squires et al . ( 1976 ) made a seminal contribution: they showed that , even in a random stream of items , the P300 amplitude varies strongly with the local history of recent observations . Even in purely random sequences ( like fair coin flips ) the amplitude of the P300 elicited by a given stimulus X increases when it is preceded by an increasing number of other stimuli Y , i . e . when it violates a “streak” of recent repetitions of Y ( e . g . XXXXX vs . YXXXX vs . YYXXX vs . YYYXX vs . YYYYX ) . The P300 amplitude also increases when a stimulus violates a pattern of alternations enforced by the recent history ( e . g . XYXYX vs . YXYXX ) . Squires et al . ( 1976 ) plotted these history effects as “trees” reflecting the entire history of recent stimuli ( Fig 3A ) . When they varied the overall frequency of items in the sequence ( from p ( X ) = 0 . 5 to 0 . 7 or 0 . 3 ) , they also found that the entire tree of local effects was shifted up or down according to p ( X ) . Altogether , their data show that the P300 amplitude reflects , in a quantitative manner , the violation of statistical expectations based on three factors: the global frequency of items , their local frequency and the local frequency of alternations . Importantly , these local effects emerged even in purely random sequences ( see the middle tree in Fig 3A ) . These effects correspond to properties #1 , #2 and #3 of the local transition probability model . Because the P300 wave seems to reflect the violation of expectations , rather than the expectations themselves , we quantified whether a given observation fulfills or deviates from expectations with the mathematical notion of surprise [51 , 52] . We computed theoretical levels of surprise , given the observations received ( and no other information ) , from the local transition probability model and we found that they quantitatively reproduce the data from Squires et al . ( Fig 3B ) . More precisely , the local transition probability model has a single free parameter , which controls the non-stationarity of the inference . It is crucial to avoid conflating the dimensionality of the estimated quantities ( which is two here , for the transition–probability matrix between two items ) and the number of free parameters constraining this estimation ( which is one for the local transition probability model ) . In the approximate model that we tested here , the only free parameter is the leak of the integration ( ω ) whose best fitting value was ω = 16 stimuli . This exponential decay factor means that the weight of a given observation is divided by two after a half-life of ω * ln ( 2 ) ≈ 11 new observations . We report in S1 Fig the results for the exact inference ( Bayes-optimal dynamic belief ) , for which the best fitting value of the a priori probability of change was pc = 0 . 167 . While the assumptions of the local transition probability model seem sufficient to account for the data , we can also demonstrate that each of them is actually necessary and that they can be distinguished from one anther ( see S4 Fig ) . Models with constant integration , i . e . without leak or a recent observation window , become increasingly insensitive to the recent history of observations as more observations are received . For such models , further details in the recent history have little impact on their expectations , as seen in the corresponding shriveled trees ( see Fig 3C ) . Models that learn simpler statistics are also not able to fully reproduce the data . The ones that learn the frequency of alternations show little effect of the global item frequency ( see the position of the roots of trees in Fig 3B ) . Those that learn the frequency of items capture the effect of global item frequency , but they fail to reproduce the specific arrangement of branches of the trees . For instance , in purely random sequences ( when p ( X ) = 0 . 5 ) , such models predict that the surprise elicited by YXYX patterns should be like the average response ( compare its position relatively to the root of the tree in Fig 3B ) whereas it is not the case in the data . A model that learns transition probabilities captures the lower-than-average activity for YXYX patterns because it detects the repeated alternation ( see Fig 3B ) . We quantified the superiority of the local transition probability model using the Bayesian Information Criterion ( BIC ) , which favors goodness-of-fit while penalizing models for their number of free parameters . The local transition probability model was better than the others: all ΔBIC > 9 . 46 ( see Table 1 ) . Cross-validation accuracy , another metric for model comparison , yielded the same conclusion ( see S5 Fig ) . We also included the model proposed by Squires et al . ( see Methods ) that achieves a similar goodness-of-fit , but at the expanse of a higher complexity . In addition , this model is descriptive ( a linear regression of several effects of interest ) and not principled . We tested the robustness of the local transition probability model with another dataset from Kolossa et al . ( 2013 ) . The authors introduced noticeable differences in the original design by Squires et al . : stimuli were visual ( instead of auditory ) and subjects had to make a two-alternative button press for each item of the sequence ( instead of listening quietly ) . Again , we could reproduce all qualitative and quantitative aspects of the data . Notably , we found almost the same best-fitting leak parameter ( ω = 17 instead of 16 ) . The BIC again favors the local transition probability model ( see Table 1 ) , even when compared to the model proposed by Kolossa et al . ( see Methods ) . Reaction time tasks submit subjects to long and purely random sequences of two items . Subjects are asked to press a dedicated button for each item , and response times typically vary with the recent history of button presses . We compared subjects' reaction times to the theoretical surprise levels computed from different leaky integration models in the same experiment . Huettel et al . ( 2002 ) were interested in the effects of streaks on reaction times and brain signals recorded with fMRI . Their data show that reaction times were slower for stimuli violating a streak ( see Fig 4A and 4B ) than for those continuing it . This was true both for repeating ( XXXXY vs . XXXXX ) and alternating streaks ( XYXYX vs . XYXYY ) , with a correlation with the streak length: the longer the streak , the larger the difference between violation and continuation . Importantly , the violation vs . continuation difference in reaction times increased more steeply with the length of repeating streaks compared to alternating streaks . This corresponds to property #5 of the local transition probability model . Importantly , this property is specific to a model that learns transition probabilities . A model that learns the frequency of alternations has identical expectations for repeating and for alternating streaks , because alternations and repetitions play symmetrical roles for this statistic . A model that learns the frequency of items has expectations in repeating streaks but not in alternation streaks . Indeed , as the streak length increases , the frequency of the repeated item increases but in alternating streaks , the frequency of either item remains similar . On the contrary , in a model that learns transition probabilities , expectations build up for both streak types by counting all possible transition types between successive items . In that case , an asymmetry emerges because repeating sequences offer twice the evidence about the current transition than do alternating sequences . For instance , in XXXXXXX , one may predict that the item following the last X should be another X , since six transitions X→X preceded without a single X→Y transition . In XYXYXYX , one can predict that the item following the last X should be a Y , since three transitions X→Y preceded without a single X→X transition . However , the ratio of evidence supporting the transition currently expected is stronger in the repeating sequence ( 6:0 ) compared to the alternating sequence ( 3:0 ) . One could argue that such an asymmetry is not a property of statistical learning but a simple consequence of motor constraints or motor priming . However , such a conclusion would be inconsistent with the EEG data recorded from passive subjects in Squires et al . study , in which the P300 difference between XXXXY and XXXXX was also larger than between XYXYY and XYXYX . In addition , Huettel et al . also recorded fMRI signals while participants performed the task in a scanner . Activity levels in several non-motor brain regions such as the insula and the inferior frontal gyrus showed the same sequential effects as the reaction times , again with a larger brain activation for violations of repetition patterns than for violations of alternation patterns . Cho et al . ( 2002 ) were interested not only in the effect of the preceding number of repetitions and alternations on reaction times , but also in their order . To do so , they sorted reaction times based on all patterns of five consecutive stimuli ( see Fig 4C ) . Each pattern contains four successive pairs , which can either be an alternation ( denoted A ) or a repetition ( denoted R ) of the same item . There are in total 24 = 16 possible patterns of repetition and alternation . Their analysis confirmed several effects already mentioned above , such as the effect of local frequency ( e . g . RRRR vs . RRAR vs . RAAR vs . AAAR ) . Their data also show clear evidence for property #4 of the local transition probability model: the same observations , in a different order , produce different expectations . Consider , for instance the sequences ARRR , RARR and RRAR . The local frequency of R is the same in these three patterns since they each contain a single discrepant observation ( A ) ; yet , the order of the observations matters: reaction times are slower when the discrepant A was observed more recently . In the local transition probability model , it is due to the non-stationarity of the estimation , which weights recent observations more than remote ones . This order effect could also be reproduced by a model that learns the frequency of alternation ( see Fig 4I ) . However , this model predicts that surprise levels should be symmetrical for alternations and repetitions . This contradicts property #5 , according to which expectations build up more rapidly for alternations than repetitions . The data conform to this property: reaction times for patterns ending with a repetition are lower than those ending with an alternation ( see Fig 4C ) , similarly to surprise levels in the local transition probability model ( see Fig 4L ) . Interestingly , the local transition probability model also captures additional aspects of the data that are left unexplained by a model that learns the frequency of alternations . When patterns are ordered as in Fig 4C , reaction times show gradual increases over the first eight patterns and gradual decreases for the last eight . There are also local deviations from this global trend: it is particularly salient for patterns RAAR and ARAA . The local transition probability model reproduces these local deviations . A model learning the frequency of alternations also predicts local deviations , but for other patterns ( RRAR and AARA ) . The observed deviations are thus specific to a learning of transition probabilities . RAAR corresponds to XXYXX where the last pair XX was already observed once , whereas in ARAR , which corresponds to XYYXX , the last pair XX was not observed . Surprise is therefore lower ( and not higher , as predicted by a model learning alternation frequency ) for RAAR than ARAR . A similar explanation holds for ARAA vs . RAAA . Finally , note that a model that learns the frequency of items fails to reproduce many aspects of the data ( see Fig 4F ) since it is completely insensitive to repetition vs . alternation effects . We obtained the results shown in Fig 4 by fitting the leak parameter ω of each model . The best-fitting value for Huettel / Cho data was: ω = 8 / ω = 4 with a learning of stimulus frequency , ω = 6 / ω = 1 with alternation frequency , and ω = 6 / ω = 3 with transition probabilities . However , simulations using the leak value fitted to the independent dataset by Squires et al . ( Fig 3 ) led to the same qualitative conclusions . Thus , a single set of parameters may capture both data sets . The asymmetry in expectation for alternation vs . repetition is probably the least trivial property of the local transition probability model ( #5 ) . This property is evidenced above in sequential effects and it entails a prediction in another domain: judgments of randomness should also be asymmetric . This prediction is confirmed: the human perception of randomness is notoriously asymmetric , as shown in particular by Falk & Konold ( 1997 ) ( see Fig 5A ) . Sequences with probabilities of alternations p ( alt . ) that are slightly larger than 0 . 5 are perceived as more random than they truly are . This is an illusion of randomness: in actuality , the least predictable sequence is when p ( alt . ) = 0 . 5 , i . e . when the next item has the same probability of being identical or different from the previous one . This bias in the perception of randomness is actually rational from the viewpoint of the local transition probability model . In order to quantify the perceived randomness of a sequence in the local transition probability model , we estimated the unpredictability of the next outcome . This unpredictability is formalized mathematically by the notion of entropy . The resulting estimated entropy level was maximal for sequences with p ( alt . ) larger than 0 . 5 ( see Fig 5D ) . This bias was all the more pronounced that fewer stimuli were taken into account in the estimation: a model with a stronger leak results in a larger bias . This aspect is specific to the local transition probability model . In contrast , a model that learns the frequency of alternation shows no bias because alternations and repetitions play symmetrical roles for such a model ( see Fig 5C ) . On the other hand , a model that learns the frequency of items shows an extreme bias: the maximal entropy level is reached for p ( alt . ) = 1 ( see Fig 5B ) . This is because when stimuli alternate , their observed frequencies are identical , closest to chance level ( 50% ) from the point of view of an observer that focuses solely on item frequency . To understand how the asymmetry emerges , one should note that , in the local transition probability model , expectations arise from both repeating transitions ( XX and YY ) and alternating transitions ( XY and YX ) . High expectations arise when one transition type is much more frequent than the other . The estimated entropy therefore decreases when p ( alt . ) approaches 1 , where alternating transitions dominate , and when p ( alt . ) approaches 0 , where repeating transitions dominate . However , remember that stronger expectations arise from repetitions than alternations in the local transition probability model ( property #5 ) . Therefore , expectations are not symmetric with respect to p ( alt . ) , but higher for p ( alt . ) < 0 . 5 than p ( alt . ) > 0 . 5 , so that the ensuing estimated entropy peaks at a value of p ( alt . ) that is slightly higher than 0 . 5 . This asymmetry is also dampened , without being abolished , when the leaky integration parameter of the local transition probability model is weaker . Indeed , experimental evidence confirms that the difference in expectations arising from repeating and alternating transitions is more pronounced for shorter sequences ( see the results from Huettel et al , Fig 4A and 4B ) .
We showed that learning non-stationary transition probabilities entails six properties . First , expectations derived from such a learning show effects of both the frequencies of items and their alternations because these statistics are specific aspects of transition probabilities ( #1 ) . Second , these effects emerge both globally and locally in the learning process because the inference is non-stationary ( #2 ) . Third , this non-stationarity also entails that local effects emerge even in purely random sequences ( #3 ) . Fourth , it depends on the exact order of observations within the local history ( #4 ) . Fifth , since the space of transition probabilities is more general than the frequencies of items and their alternations , the local transition probability model makes a non-trivial prediction , unaccounted for by simpler statistics: expectations build up more strongly from repetitions than from alternations ( #5 ) . Sixth , this asymmetry translates into a subjective illusion of randomness which is biased toward alternations ( #6 ) . We identified many signatures of expectations and their violation in human behavior ( such as reaction times ) and brain signals ( measured by electrophysiology and fMRI ) which conformed both qualitatively and quantitatively to these predictions . We therefore conclude that transition probabilities constitute a core building block of sequence knowledge in the brain , which applies to a variety of sensory modalities and experimental situations . Early studies [14 , 16] proposed that the information provided by stimuli modulates reaction times within sequences [12] . According to the information theory framework , an observation is informative inasmuch it cannot be predicted [51] . In line with this information-theoretic approach , the local transition probability model quantifies the extent to which an observation deviates from the preceding ones . The central role of expectations in cognitive processes has also been put forward by the predictive coding [7 , 53] and the active inference [46 , 54] frameworks , and applied , for instance , to motor control [55 , 56] or reinforcement learning [57] . However , some have claimed that sequential effects in reaction times arise from low-level processes such as motor adaptation . For instance , Bertelson wrote in 1961 “one must thus admit that the shorter reaction times [for repetitions] cannot depend on something which must be learnt about the series of signals–unless one assumes that this learning is fast enough to be completed and give already its full effect on performance in the first 50 responses” [13] . In contrast , the local transition probability model shows that , with optimal statistical learning , sequence effects can arise from a very local integration: our fit of Squires et al . ( 1976 ) data suggests a leak factor ω of 16 stimuli , meaning that the weight of a given observation is reduced by half after 16 * ln ( 2 ) ≈ 11 observations . In addition , facilitation of reaction times is observed for both streaks of repetitions and streaks of alternations , which speaks against a pure motor interpretation [17 , 18] . Moreover , similar sequential effects are also observed in electrophysiological and fMRI measures of brain activity in the absence of any motor task . Therefore , although motor constraints may also contribute to reaction times fluctuations , a parsimonious and general explanation for sequential effects is that they arise from learned statistical expectations . The non-stationary integration also explains why both local and global effects emerge and why local effects persist in the long run even within purely random sequences [20 , 23] . From the brain's perspective , the constant attempt to learn the non-stationary structure of the world could be a fundamental consequence of a general belief that the world can change at unpredictable times , as already suggested by others [23] . Many studies indeed show that the brain can perform non-stationary estimation and thereby efficiently adapt to changes in the environment [49 , 50 , 58–60] . Technically , the belief in a changing world can be captured in two different ways: either by the a priori likelihood of a sudden change ( a . k . a . volatility ) pc in the exact dynamic belief model , or by the leaky integration factor ω in the approximate model . The present data do not suffice to separate those two possibilities . This is because the latter ( leaky integration ) is such a good approximation of the former that both are difficult to disentangle in practice . Leaky integration is a popular model in neuroscience because it seems easy to implement in biological systems [23 , 58 , 61 , 62] . However , the dynamic belief model may not be less plausible given that neuronal populations have been proposed to represent and compute with full probability distributions [33 , 41] . Furthermore , only the full Bayesian model recovers an explicit probabilistic representation of change likelihood and change times . Several recent experimental studies suggest that the brain is indeed capable of estimating a hierarchical model of the environment , and that human subjects can explicitly report sudden changes in sequence statistics [60 , 63] . Our results suggest that , during sequence learning , the brain considers a hypothesis space that is more general than previously thought . We found that sequential effects in binary sequences are better explained by a learning of transition probabilities ( a 2-dimensional hypothesis space ) than of the absolute item frequencies or the frequency of their alternations ( which are one-dimensional spaces ) . Importantly , all of these models have the same number of free parameters , so that the local transition probability model is more general without being more complex or less constrained . The critical difference lies in the content of what is learned ( e . g . item frequencies vs . transition probabilities ) . More is learned in the latter case ( a 2D space is larger than a 1D space ) without resorting to any additional free parameter . The value of the learned statistic is not a free parameter , it is instead dictated by the sequence of observations and the assumptions of the model . In general , a Bayesian learner may consider a vast hypothesis space ( see the many grammars used by Kemp and Tenenbaum [64] ) and yet , as a model that attempts to capture human behavior , it may possess very few or even zero adjustable parameters . An alternative to the full 2D transition-probability model would be to combine two learning processes: one for the frequency of items and one for the frequency of alternations . However , such a model introduces a new free parameter compared to the local transition probability model: the relative weight between the predictions based on the frequency of items and the predictions based on the frequency of alternations . In addition , the distinction between learning transition probabilities vs . the frequency of items and their alternations is not a simple change of viewpoint: the correspondence between the two is extremely non-linear as shown in Fig 1A . Learning the frequency of items and the frequency of alternations is therefore not only less parsimonious than learning transition probabilities , it is also genuinely different . The difference between learning transition probabilities vs . the frequency of items and their alternation may have been overlooked in the past . However , the distinction is important since these learning strategies make distinct predictions about the asymmetry of expectations arising from repetitions and alternations . This asymmetry is a classical aspect of data , in particular response times [13] . In previous models , this asymmetry was simply assumed and incorporated as a prior [23 , 25] . We show here , to our knowledge for the first time , how this asymmetry follows naturally from first principles ( Bayes’ rule ) in the local transition probability model . Moreover , our account is also unifying since it addresses not only sequential effects but also judgments of randomness . We claim that the learning of transition probabilities is a core and general building block of sequence knowledge because we found supportive evidence in five representative datasets . There is also additional evidence from other fields . For instance , word segmentation in language relies on transition probabilities between syllables [65] . Moreover , neurons in the monkey inferior temporal cortex reduce their firing in direct proportion to the learned transition probabilities [66 , 67] . Ramachandran et al . ( 2016 ) , in particular , present single-cell recordings suggesting that the expectation about the next item does not depend on its absolute frequency or the absolute frequency of the pair it forms with the previous item , but instead on the conditional probabilities of items learned with a covariance-based rule . Additional sources of evidence that human subjects learn transition probabilities is provided by studies of “repetition suppression” [68 , 69] , choices in decision-making problems [70] and explicit reports of learned transition probabilities [60] . The study by Bornstein and Daw [70] in particular shows that humans can learn transition probabilities among 4 items . Our local transition probability model naturally extends from the binary case to a larger number of categories , a situation that is pervasive in every-day life . Learning only the frequency of items and of alternations becomes gradually inadequate when the number of items increases , because most environmental regularities are captured by various item-specific transition probabilities rather than absolute frequencies . For instance , the probability of imminent rain is typically high after a thunderstorm , but very low during a sunny day , and intermediate in case of strong wind . The learning of transition probabilities may even operate without awareness [71–73] . Our claim that a learning of transition probabilities accounts for a variety of experimental effects does not rule out the possibility that the brain also computes simpler statistics . Many studies report effects of item frequencies or alternation frequency . Electrophysiology in particular shows that these effects unfold across time and across brain circuits , as reflected in signals such as the mismatch negativity and the P300 [6 , 7 , 11 , 19] . In particular , Strauss et al . ( 2015 , experiment 2 ) identified two distinct time windows in magneto-encephalographic recordings , during which the absolute frequency of items and the frequency of alternations , respectively , affected the human brain responses to simple sounds . However , in most studies , it is not clear whether such effects are particular cases of a general learning of transition probabilities or whether they are genuinely limited to item frequency or alternation frequency . Both hypotheses are indistinguishable in most studies because of their experimental design . Therefore , it is not clear for the moment whether different brain circuits are tuned to these different statistics and compute them in parallel [48] , or whether most brain regions are equipped for the computation of transition probabilities . By contrast , it seems that more sophisticated building blocks of sequence knowledge , such as ordinal knowledge , chunking , algebraic patterns and tree structures are operated by specific brain circuits [47] . Future work should aim to incorporate these additional levels of representation to the local transition probability model , which we propose here as a minimal building block , likely to be duplicated in many brain regions and shared by humans and other animals alike .
The models are “ideal observers”: they use Bayes’ rule to estimate the posterior distribution of the statistic they estimate , θt , based on a prior on this statistic and the likelihood provided by previous observations , y1:t ( here , a sequence of Xs and Ys ) . Subscripts denote the observation number within a sequence . Different ideal observer models ( M ) estimate different statistics . The parameter θ can be the frequency of items , the frequency of alternations , or transition probabilities between items . The estimation of θ depends on the assumption of the ideal observer model: it can either consider that θ is fixed and must generate all the observations ( “fixed belief models” ) or that θ may change from one observation to the next ( “dynamic belief models” ) . For all models , we use a prior distribution that is non-informative: all possible values of θ are considered with equal probability . Note that a model estimating the frequency of alternations is equivalent to a model estimating the frequency of items after recoding of the stimuli as repetitions or alternations . Therefore , we only present below the derivation for the item frequency and transitions probabilities , in the case of both fixed belief and dynamic belief models . Squires et al . ( 1976 ) presented 7 subjects with sequences of two auditory stimuli ( pure tones of 1500 Hz and 1000 Hz , denoted X and Y ) during electroencephalogram ( EEG ) recordings . In separate sessions , the sequences were generated randomly with p ( X ) equals to 0 . 5 ( no bias ) or 0 . 7 ( biased condition ) . In the biased condition , p ( Y ) = 1 – p ( X ) = 0 . 3 . Because X and Y play symmetrical roles , the authors present results for a virtual condition “p ( X ) = 0 . 3” which actually corresponds to analyzing the responses to item Y in the biased condition . Subjects were not told about these exact probabilities . The stimulus duration was 60 ms and the stimulus onset asynchrony was 1 . 3 s . Subjects were asked to count the number of X items silently and report their count after each block of 200 trials . They were presented in total and in each condition , with 800 to 1600 stimuli . Squires et al . measured a P300 score for each stimulus . This score is a weighted combination of signals measured at central electrodes ( Fz , Cz , Pz ) and latencies corresponding to the N200 , P300 and slow-wave . The weights derive from a discriminant analysis to separate optimally the signals elicited by rare and frequent patterns ( XXXXY vs . XXXXX ) . The average scores are reported in each condition , for all patterns of five stimuli terminated by X . Kolossa et al . collected EEG data from 16 subjects who were presented with a stream of two visual stimuli ( red or blue rectangles , denoted X and Y; the mapping was counterbalanced across participants ) . In separate blocks , stimuli were generated with p ( X ) = 0 . 5 ( no bias condition ) or 0 . 7 ( biased condition ) . A virtual condition p ( X ) = 0 . 3 corresponds , as in Squires et al . , to the response to item Y in the biased condition . Subjects were not told about these exact probabilities . Subjects completed 12 blocks of 192 stimuli with 6 blocks in a row for each condition . The order of conditions was counterbalanced across participants . The stimulus duration was 10 ms and the stimulus onset asynchrony was 1 . 5 s . Subjects were asked to press a dedicated button for each item as quickly and accurately as possible . Kolossa et al . measured the P300 amplitude at electrode Pz . The exact latency of the measurement varied across trials and participants . The authors first identified subject-specific peak latencies for the difference between rare and frequent items in the biased condition . Then , in each trial they extracted the maximum value of the signal within a window of 120 ms centered on subject-specific peaks . P300 levels are reported for each condition , for all patterns of four stimuli terminated by X . We generated three sequences of 200 stimuli with probabilities p ( X ) equal to 0 . 3 , 0 . 5 and 0 . 7 . For each sequence , we computed the inference of the hidden statistics for each observer model and different values of their free parameters ( if any ) . We computed surprise , in bit of information , for each model and each stimulus in the sequences , as log2 ( p ( yt|y1:t–1 ) ) , where p ( yt|y1:t–1 ) is the likelihood of the actual observation . We sorted surprise levels by patterns of five stimuli terminated with X . We repeated this simulation 200 times and we averaged over simulations to reach stable results . To compare our simulation with the data from Squires et al , we extracted their values from figure 1 in [1] . For each model , we adjusted the offset and scaling to minimize the mean squared error ( MSE ) between simulated and experimental data . We repeated this procedure for different values of the free parameter ω in fixed belief models with leaky integration , and different values of pc in dynamic belief models . For all models , we fitted the data only for patterns of 5 stimuli since shorter patterns are not independent from longer ones: they are weighted averages of the data obtained for longer patterns . Including shorter patterns would have thus inflated some aspects of the data . For instance , the effect of item frequency can be seen for all pattern lengths , including length 1 , but by definition , the effect of alternations can be seen only in longer patterns . Therefore , including shorter patterns would have over-weighted the effect of global item frequency relatively to local alternations . Our fitting procedure gives the same weight to all patterns of 5 stimuli , although rare patterns are more likely to be corrupted by noise in the experimental data . However , our results are robust to this choice and are replicated when using a weighted MSE , taking into account the expected frequency of patterns . We also checked that the grids of values used for pc or ω were sufficiently dense around the global maxima , as shown in S1 Fig . We replicated this procedure with the data from Kolossa et al . , taking their values from figure 8 in [9] . The only difference was that we used patterns of length 4 , as reported by Kolossa et al . , instead of length 5 as Squires et al . For comparison , we implemented the models previously proposed by Squires et al . and Kolossa et al . These models are fully described in the related articles [1 , 9] . In short , the model by Squires et al . is a weighted sum of three factors ( the variables within brackets correspond to the notations by Squires et al . ) : The free parameters of this model are the relative weight of global frequency ( P ) vs . local frequency ( M ) , the relative weight of global frequency ( P ) vs . patterns of alternations ( A ) , and a decay factor for counting the local number of stimuli ( M ) . The model by Kolossa et al . is a sophistication of the model by Squires et al . It is a weighted sum of three factors , thought of as the output of digital filters computing probabilities . These filters correspond to: This model includes six free parameters ( we use the notations from Kolossa et al . ) : a decay factor for short-term memory ( βS ) , two normalized time constants for the dynamic long-term memory ( τ1 and τ2 ) , the relative weight of item probabilities computed from short- and long-term memory ( αS ) , the relative weight of probabilities computed for items and their alternations ( αΔ ) , and a parameter capturing the subjective distortion of probabilities ( γΔ , 2 ) . We use the best-fitting values of the free parameters reported by Squires et al . and Kolossa et al . in their respective article . To compare the fit provided by our models and by the models by Squires et al . and Kolossa et al . , we used the Bayesian Information Criterion ( BIC ) . The BIC favors the goodness-of-fit but penalizes model for their number of free parameters [74] . For maximum likelihood estimate of the model parameters and Gaussian residuals , BIC = n · log ( MSE ) + k · log ( n ) , with n the number of fitted data points and k the number of free parameters in the model . Note that here , k counts the scaling and offset parameter to adjust the model's data and the experimental data , and the internal free parameters of the model ( from 0 for fixed belief model with perfect integration , to 6 for the model by Kolossa et al . ) . Huettel et al . presented 14 subjects with a stream of two visual stimuli ( a square and a circle ) randomly generated with equal probability . The sequence length was 1800 . Each stimulus was presented for 250 ms , and the stimulus onset asynchrony was 2 s . Subjects were asked to press a dedicated button for each item as quickly as possible . Subjects performed the experiment in an MRI scanner for functional recordings . We extracted the data by Huettel et al . from their figure 2 in [2] . We simulated these results using the fixed belief model with leaky integration . We fitted the leak constant to the data using a grid search . We replicated the simulation results with the dynamic belief model , using pc = 0 . 019 for the estimation of frequencies and pc = 0 . 167 for the estimation of transition probabilities , which are the best fitting values for Squires et al . data . We generated a sequence of 105 stimuli with p ( X ) = 0 . 5 . This large number of observations ensured stable simulation results . Another solution is to generate many short sequences . Both options actually yield similar results because of the limited horizon of the non-stationary estimations used here . Similar to our fit of Squires et al . , we computed posterior inferences and surprise levels . The difference was that surprise levels were sorted based on whether local patterns of stimuli were alternating or repeating , whether the last item violated or continued the pattern , and the length of the pattern ( up to 8 ) . Cho et al . presented 6 subjects with a stream of two visual stimuli ( a small and a large circle ) , generated randomly with equal probabilities . Subjects were asked to press a dedicated button for each item as quickly and accurately as possible . Each stimulus was presented until a response was made within a limit of 2 s . The next stimulus appeared after a delay of 0 . 8 s . Subjects performed 13 series of 120 trials ( 1560 stimuli in total ) with a short break between series . We extracted the data by Cho et al . from their figure 1B in [20] . We used the same simulation as for Huettel et al . The only difference being that surprise levels were sorted based on all patterns of alternations and repetitions formed by 5 stimuli . Ass described in [25] , Falk presented 219 subjects with sequences of 21 binary visual stimuli ( “X” and “O” ) . The sequences had ratios of alternations ranging from 0 . 1 to 1 with 0 . 1 steps . The order of the sequences varied randomly across participants . Each sequence was printed as a row on a paper sheet: stimuli were therefore presented simultaneously . Subjects were asked to rate the apparent randomness of each sequence from 0 to 20 , with the indication that this judgment should reflect the likelihood of the sequence having been generated by flipping a fair coin . Ratings were later rescaled between 0 and 1 . We extracted the data by Falk from figure 1 , condition “ARI” in [25] . Following the original experiment , we generated sequences of 21 binary stimuli with various probabilities of alternations . For each sequence , we computed the posterior inference of the hidden statistics and the prediction about the next stimulus ( the 22th ) given the previous ones , for each observer model . We used different values for their leak parameter . To quantify the “randomness” of the sequence , we computed the entropy of the prediction: H ( p ) = – p * log2 ( p ) – ( 1 – p ) * log2 ( 1 – p ) . For each leak parameter and alternation frequency , we averaged over 104 sequences to reach stable results . Note that instead of focusing on the last prediction , one could average across successive predictions in each sequence . This alternative yields the same qualitative results as shown in Fig 5 . | We explore the possibility that the computation of time-varying transition probabilities may be a core building block of sequence knowledge in humans . Humans may then use these estimates to predict future observations . Expectations derived from such a model should conform to several properties . We list six such properties and we test them successfully against various experimental findings reported in distinct fields of the literature over the past century . We focus on five representative studies by other groups . Such findings include the “sequential effects” evidenced in many behavioral tasks , i . e . the pervasive fluctuations in performance induced by the recent history of observations . We also consider the “surprise-like” signals recorded in electrophysiology and even functional MRI , that are elicited by a random stream of observations . These signals are reportedly modulated in a quantitative manner by both the local and global statistics of observations . Last , we consider the notoriously biased subjective perception of randomness , i . e . whether humans think that a given sequence of observations has been generated randomly or not . Our model therefore unifies many previous findings and suggests that a neural machinery for inferring transition probabilities must lie at the core of human sequence knowledge . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"learning",
"medicine",
"and",
"health",
"sciences",
"diagnostic",
"radiology",
"functional",
"magnetic",
"resonance",
"imaging",
"reaction",
"time",
"brain",
"electrophysiology",
"social",
"sciences",
"electrophysiology",
"neuroscience",
"learning",
"and",
"memory",
"mag... | 2016 | Human Inferences about Sequences: A Minimal Transition Probability Model |
Elimination of kala-azar is planned for South Asia requiring good surveillance along with other strategies . We assessed surveillance in Gaffargaon upazila ( a subdistrict of 13 unions ) of Mymensingh district , Bangladesh highly endemic for kala-azar . In 4703 randomly sampled households , within nine randomly sampled villages , drawn from three randomly sampled unions , we actively searched for kala-azar cases that had occurred between January 2010 and December 2011 . We then searched for medical records of these cases in the patient registers of Gaffargaon upazila health complex ( UHC ) . We investigated factors associated with the medical recording by interviewing the cases and their families . We also did a general observation of UHC recording systems and interviewed health staff responsible for the monthly reports of kala-azar cases . Our active case finding detected 58 cases , but 29 were not recorded in the Gaffargaon UHC . Thus , only 50% ( 95% CI: 37%–63% ) of kala-azar cases were reported via the government passive surveillance system . Interviews with health staff based in the study UHC revealed the heavy reporting burden for multiple diseases , variation in staff experience , high demands on the staff time and considerable complexity in the recording system . After adjusting for kala-azar treatment drug , recording was found more likely for those aged 18 years or more , males , receiving supply and administration of drug at the UHC , and more recent treatment . Fifty percent of kala-azar cases occurring in one highly endemic area of Bangladesh were recorded in registers that were the source for monthly reports to the national surveillance system . Recording was influenced by patient , treatment , staff and system factors . Our findings have policy implications for the national surveillance system . Future studies involving larger samples and including interviews with health authorities at more central level and surveillance experts at the national level will generate more precise and representative evidence on the performance of kala-azar surveillance in Bangladesh .
Visceral leishmaniasis ( VL ) , caused by protozoan parasites transmitted by sandflies , is a systemic illness characterized by fever and splenomegaly . The disease is endemic in impoverished tropical areas globally [1–3] , and in South Asia is known as kala-azar ( black-fever ) and only affects humans . The governments of India , Bangladesh and Nepal in 2005 committed to eliminate kala-azar by 2015 . Elimination is defined as annual incidence of less than 1 per 10 , 000 population at the district or sub-district level [4] . Surveillance is one of the main strategies for kala-azar elimination [5] . In Bangladesh , surveillance provides estimates and trends of the nationwide occurrence of kala-azar [6] . Kala-azar is reported every month starting from the upazila or subdistrict level . We have studied surveillance in a highly endemic sub-district of Bangladesh and here we report our findings . We analyse factors influencing case recording and make recommendations to improve surveillance .
The study was conducted in Gaffargaon sub-district of Mymensingh district in Bangladesh , 80km north-east of Dhaka . Gaffargaon is divided into 15 unions and 214 villages . Around half of the population of Gaffargaon are male , half are literate and most are farmers . The sub-district , 398 sq . km . with a population of 430 , 000 , reported the third highest number of kala-azar cases in 2011[7 , 8] . We expected 1 . 4 kala-azar cases per 100 households ( based on an unpublished 2009 survey data ) . Also , using Epi Info StatCalc , we calculated that we would need a sample of 68 cases ( i . e . 4857 households ) to detect 50% under-reporting through the national surveillance system with 10% precision and 90% confidence [9] . First , we excluded two of the 15 unions that were part of central administration . Care seeking experience of kala-azar patients from these two unions would be different than that of the patients from the other unions because of the proximity to the Gaffargaon upazila health complex ( UHC ) . Then we randomly sampled three unions from the sampling frame of 13 unions . In each of these sampled unions we randomly selected three villages . We then sampled households in each village , aiming for about 600 households per village . If a village had more than 600 households we divided it into different paras or localities and randomly sampled paras until we got 600 households . If a village had less than 600 households , we sampled all the households of that village . This resulted in sampling of less number of households than expected . The resulting sample included a total of 4703 households from nine sampled villages from three sampled unions of Gaffargaon . Our case definition required diagnosis by a qualified health care provider based on clinical presentation and a positive confirmatory diagnostic test . Most of the study respondents could not mention the name of the confirmatory diagnostic test . But they mentioned that the providers who confirmed kala-azar did so based on positive diagnostic test results for kala-azar . Going from sampled house to house in the period December 2011 to May 2012 , we sought cases that had occurred between January 2010 and December 2011 . In each household the informant was a case or a senior family member . If there was more than one kala-azar case in a household , we studied the earliest case in order to avoid household clustering . Kala-azar cases and their families were interviewed using a structured questionnaire . For cases less than 18 years of age , a parent or carer was interviewed . Surviving family members of deceased cases were interviewed . We collected demographic and economic data , date of symptom onset , signs and symptoms , care seeking experiences and diagnosis . We also studied recording and reporting of kala-azar cases making general observations at the Gaffargaon UHC . As well , we interviewed three key health service staff from the Gaffargaon UHC who were involved in the kala-azar reporting system . We asked them to detail their role in kala-azar reporting . We asked about obstacles to reporting and how the reporting system could be improved . These semi-structured interviews were audio-recorded . We examined Gaffargaon UHC records to determine if cases we detected had been entered in the registers . Inpatient , outpatient and laboratory registers were reviewed . We sought record entries in the 2010–2011 periods for all kala-azar cases we detected in the community . We used the name and address to confirm identification of the patients in the register books . Quantitative data were digitized using Epi Info ( version 3 . 5 . 3 ) . Information from the patients and families was recorded in MS Excel in a file which also noted whether the patient was recorded in the UHC registers . Qualitative data included extensive notes on general observations . The recordings were transcribed in the Bengali language for the three UHC staff . Quantitative data were analysed using STATA version 8 [10] . We compared frequencies of patient and health system factors for those found and not found in hospital records . The factors were age , sex , place of diagnosis , treatment location , drug and year . Any difference with p<0 . 05 was considered statistically significant . Interview transcripts were contrasted for recurring themes and informative quotations related to the research questions . Quotations used in this publication were translated into English by author KMR . All analyses were supported by the general observations . Informed written consent was obtained from participants . For illiterate interviewees , the consent form was read out loud and the participants’ fingerprints were obtained on the consent forms . Before conducting the study of the kala-azar reporting system and interviewing staff of Gaffargaon UHC , written permission was obtained from the relevant authority based centrally in Dhaka and at Gaffargaon . Individual consent was also obtained from interviewed staff who were advised that they would not be identified . Ethical approvals were provided by the Human Research Ethics Committee of the Australian National University and the Ethical Review Committee of the International Centre for Diarrhoeal Disease Research , Bangladesh ( icddr , b ) . Both the ethics committees specifically approved the use of thumb print procedures .
We screened 4703 households and identified 89 people fulfilling our case definition of kala-azar . Before further analysis , we excluded cases whose circumstances did not represent the typical probability for notification from Gaffargaon UHC . These exclusions included those who migrated out ( 5 cases ) and those who received special clinical care or care from outside Gaffargaon ( 9 cases ) . As well , to avoid household clustering , we excluded 15 cases that arose in a family that had already provided a case for our sample . Two other cases refused to participate so 58 remained for analysis in our study of surveillance . The flow of kala-azar patients and related information is shown in Fig 1 . We identified three registers ( laboratory , kala-azar and admission ) where staff recorded the patient name and other details . Each month , using the kala-azar register as the final source of the information , the responsible nurse reports to the UHC statistician the consolidated data on the number of kala-azar cases ( by age , sex , treatment status ) . Then the statistician compiles a monthly kala-azar tally along with the routine ‘disease profile’ report and sends this to the Civil Surgeon’s office responsible for that district . After reviewing the UHC registers in Gaffargaon we concluded that there were no records for 29 of our 58 actively detected cases . Thus , overall , 50% ( 95% CI: 37%–63% ) were recorded by the government surveillance system—44% in 2010 and 59% in 2011 . Accordingly we can estimate that no more than 50% ( 95% CI: 37%–63% ) of kala-azar cases were reportable through the government mechanism that depends on UHC recording . Using the survey data we explored factors associated with recording in the UHC kala-azar register . Patient attributes showed some associations as follows: age 18 years or more and male sex increased the likelihood of being recorded by the UHC , although none of the associations were statistically significant ( Table 1 ) . We also found certain health system factors associated with recording in the UHC kala-azar register . First , place of diagnosis ( public vs . private ) was indicative ( OR = 0 . 5; 95% CI = 0 . 1–2 . 0 ) . As well , place of drug supply was indicative: among those recorded 100% were supplied their treatment at the UHC as compared to 76% of those who were not recorded ( p <0 . 01 ) . Actual administration of kala-azar drug at the UHC was associated with 80% higher odds of being recorded ( OR = 1 . 8; 95% CI = 0 . 3–11 . 5 ) . We also observed that the probability of being found in the UHC record further improved in 2011 relative to 2010 ( OR = 1 . 9; 95% CI = 0 . 6–6 . 5 ) . Interviews with staff indicated a number of factors influencing reporting . There were issues such as the burden of recording numerous notifiable diseases on the monthly ‘disease profile’ report . As well , the reporting system itself has changed , leading to new routines along with a new kala-azar reporting form . The informant indicated that reporting channels were sometimes complex . This reflected the need to get the information through quickly but this can be difficult in hard copy due to the need for a signature by the Upazila Health and Family Planning Officer ( otherwise known as the Thana Health Administrator or THA ) . In addition , different nursing staff recorded the patient data at different times . This may cause the quality or completeness of recording to vary . However , the system used was unlikely to lead to duplicate recording because of the unique registration number .
Only half of the kala-azar cases arising in Gaffargaon sub-district , a highly endemic area within Mymensingh district , were recorded in the upazila health complex records . Consequently the monthly tallies of kala-azar cases reported to the government of Bangladesh represent only about half of the actual cases that occur . Investigation of socio-demographic factors associated with non-recording was generally uninformative—all factors tested were not significantly associated and effect estimates had wide confidence limits revealing limited statistical power . However , health system factors had more influence . Recording in the UHC kala-azar register significantly associated with supply of drug ( UHC ) and place of administration of the drug ( UHC ) . We observed improvement of recording in 2011 compared to 2010 . Previous population based studies in Bangladesh have not reported surveillance performance in any way comparable to our study [2 , 11–13] . In India , Singh et al ( 2006 ) assessed kala-azar surveillance in Bihar in 2003 . They searched households for people with fever for over 15 days duration and confirmed kala-azar with microscopic parasite identification in spleen or bone marrow aspirates . Sixty-five cases were detected through their active case search and only 8 ( 12 . 3% ) were reported [14] . The same group in a later study found that only 17% of the cases were reported [15] . Those over 30 years of age were significantly less likely to be reported , but no other patient , family or health system factor was shown to be indicative . Our study was done on a two-year sample of kala-azar patients in a geo-demographically defined segment of the Bangladesh population in a highly endemic area . We achieved our principal aim of estimating with reasonable accuracy the proportion of incident kala-azar cases being recorded by the health system ( i . e . 50% ) . But our sample size did not have statistical power to enable conclusive sub-analyses of factors related to surveillance . However , the qualitative data we collected from the staff based at the study UHC were able to provide some perspective on the issue of preparing tallies and reporting to the government . The system in use involves recording in a kala-azar register which is the source for reporting kala-azar to the statistician to enable the monthly tally . The staff we interviewed made comments about the data flow and forms , and the figure we produced showed a rather complex system that could be simplified . We were not able to determine the actual proportion reported because we could not separate the contribution expected from our sample from the contribution expected from the rest of the population served by the same UHC . If we could have searched for cases in the entire population of Gaffargaon ( around 430 , 000 ) for a particular month then we would be able to compare the number of cases detected in the community and number of cases reported from the UHC . But this would require substantial resources . We learnt from our study that around 50% of kala-azar cases in a highly endemic area of Bangladesh are not yet detectable by the government passive surveillance system . Obstacles to reporting were related to the reporting system and the reporting burden for multiple diseases . Future studies involving larger samples and including interviews with health authorities at more central level and surveillance experts at the national level will generate more precise and representative evidence on the performance of kala-azar surveillance in Bangladesh . | Visceral leishmaniasis , a parasitic disease transmitted by sandflies , is known as kala-azar in South Asia , and has been targeted for elimination in that region . The aim is to reduce its incidence to a low level so that it is no longer a public health problem . Elimination strategies include good surveillance for occurrence of the disease . We assessed surveillance in Gaffargaon subdistrict of Mymensingh district in Bangladesh where occurrence of kala-azar is high . We randomly sampled 4703 households and searched for kala-azar cases that had occurred between January 2010 and December 2011 . We then searched for medical records of these cases in the patient registers of Gaffargaon upazila or subdistrict health complex ( UHC ) . We interviewed cases , their families and health staff responsible for the monthly reports of kala-azar . We also observed the UHC recording system . We detected 58 cases , but 29 ( 50% ) were not recorded in the Gaffargaon UHC . Problems include the heavy reporting burden for multiple diseases , variation in staff experience , high demands on the staff time and complexity in the recording system . Recording was more likely for adults , males , those given drugs at the UHC , and more recent treatment . Our findings have implications for kala-azar surveillance . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | Performance of Kala-Azar Surveillance in Gaffargaon Subdistrict of Mymensingh, Bangladesh |
Mitochondrial DNA ( mtDNA ) encodes proteins essential for ATP production . Mutant variants of the mtDNA polymerase cause mutagenesis that contributes to aging , genetic diseases , and sensitivity to environmental agents . We interrogated mtDNA replication in Saccharomyces cerevisiae strains with disease-associated mutations affecting conserved regions of the mtDNA polymerase , Mip1 , in the presence of the wild type Mip1 . Mutant frequency arising from mtDNA base substitutions that confer erythromycin resistance and deletions between 21-nucleotide direct repeats was determined . Previously , increased mutagenesis was observed in strains encoding mutant variants that were insufficient to maintain mtDNA and that were not expected to reduce polymerase fidelity or exonuclease proofreading . Increased mutagenesis could be explained by mutant variants stalling the replication fork , thereby predisposing the template DNA to irreparable damage that is bypassed with poor fidelity . This hypothesis suggests that the exogenous base-alkylating agent , methyl methanesulfonate ( MMS ) , would further increase mtDNA mutagenesis . Mitochondrial mutagenesis associated with MMS exposure was increased up to 30-fold in mip1 mutants containing disease-associated alterations that affect polymerase activity . Disrupting exonuclease activity of mutant variants was not associated with increased spontaneous mutagenesis compared with exonuclease-proficient alleles , suggesting that most or all of the mtDNA was replicated by wild type Mip1 . A novel subset of C to G transversions was responsible for about half of the mutants arising after MMS exposure implicating error-prone bypass of methylated cytosines as the predominant mutational mechanism . Exposure to MMS does not disrupt exonuclease activity that suppresses deletions between 21-nucleotide direct repeats , suggesting the MMS-induce mutagenesis is not explained by inactivated exonuclease activity . Further , trace amounts of CdCl2 inhibit mtDNA replication but suppresses MMS-induced mutagenesis . These results suggest a novel mechanism wherein mutations that lead to hypermutation by DNA base-damaging agents and associate with mitochondrial disease may contribute to previously unexplained phenomena , such as the wide variation of age of disease onset and acquired mitochondrial toxicities .
Mitochondrial DNA ( mtDNA ) maintenance is necessary for the majority of ATP production in eukaryotic cells . The inability to properly replicate mtDNA potentially impacts human health in several ways . The premature aging phenotype of POLG exonuclease deficient mice indicates that increased mtDNA mutagenesis can be detrimental [1]–3 . Also , mutations in genes encoding the mitochondrial replisome , including DNA polymerase γ ( pol γ , encoded by POLG ) , contribute to mitochondrial diseases characterized by mtDNA depletion , deletions , or point mutations [4]–[14] . Additionally , environmental changes can modify mitochondrial biology and potentially impact health . Chain-terminating nucleotide analogs used in anti-viral therapy impair mtDNA replication and can result in mitochondrial toxicity [15] . Antioxidants and exercise have been shown in model systems to improve mitochondrial function and suppress the premature aging phenotype associated with increased point mutations and deletions [3] , [14] . Therefore , environmental changes can be important for mitochondrial function , and the mechanisms that cause mtDNA mutations warrant further study . Currently hundreds of POLG mutations have been identified in patients with mitochondrial disease such as Alpers syndrome , progressive external ophthalmoplegia , and ataxia-neuropathy syndrome ( mutations listed in http://tools . niehs . nih . gov/polg/ ) [16] . Pol γ-related mitochondrial diseases display a wide variety of severities . For instance , Alpers syndrome manifests in infants and young children , and these patients rarely live through their first decade of life [17] . Alternatively , patients with progressive external ophthalmoplegia ( PEO ) and sensory ataxia neuropathy , dysarthria , and ophthalmoparesis ( SANDO ) often are asymptomatic until>20 years of age [4] , [18] . The catalytic subunit of pol γ contains DNA polymerase , 3′-5′ exonuclease , and 5′ dRP lyase activities , with known discrete polymerase and exonuclease domains [19]–[22] . Among the POLG mutations associated with mitochondrial disease , many have been characterized biochemically and shown to disrupt polymerase activity [5]–[10] , [13] , [14] , [23]–[27] . POLG polymerase variants H932Y , R943H , and Y955C alter dNTP-interacting side chains and are associated with less than 1% polymerase activity [26] . Polymerase variants G848S , T851A , R852C , and R853Q also reduce polymerase activity to <1% of wild type activity; in addition , G848S also exhibits a DNA-binding defect [24] . Although mutagenic effects of point mutations that disrupt exonuclease activity have been well established , disease-associated mutations in the human exonuclease domain surprisingly do not disrupt the exonuclease activity in Mip1 [9] , [10] . These disease associated exonuclease mutations have not been studied in the human enzyme . S . cerevisiae has been useful to characterize Pol γ functionality with mutations that alter amino acids within conserved stretches between human POLG and yeast MIP1 , most of which are in the polymerase domain [28] , [29] . Mitochondrial functionality and mtDNA point mutagenesis have been determined in various mutants using assays that measure frequency of petite colony formation ( ie , lacking mitochondrial function either with [rho−] or without mtDNA [rho0] ) and erythromycin resistance , respectively [30] . For instance , mutations that alter the catalytic aspartates ( eg , Asp171 and Asp 230 in yeast ) in the exonuclease domain are associated with 1440-fold and 160-fold increases in point mutagenesis [31]–[33] and deletions between direct repeats in mice , respectively [32] , [33]; the corresponding increases in yeast are 2000-fold for point mutagenesis [31] and 90-fold for deletions between direct repeats [12] , [33] . These increases in mtDNA mutagenesis establish Asp171 and Asp 230 as critical domains for “proofreading” against misinsertions . Surprisingly , disease-associated mutations in the exonuclease domain are associated with only modest increases in mutant frequency , suggesting that exonuclease activity is functionally sufficient to correct misinsertions [9] , [10] . Many of the disease-associated mutations to Mip1 eliminated the ability to replicate mtDNA and were associated with petite colony formation , including human variants R807C , R807P , R853W , N864S , G923D , H932Y , K947R , G1076V , R1096C , S1104C , and V1106I and Alpers-associated mutations G848S , T851A , R853Q , D930N , A957P , P1073A , and R1096H [9] . However , several strains that contain variants , including R853Q and Q308H ( R656Q and Q264H in yeast ) coexpressed with wild type MIP1 to maintain mtDNA showed significant increases in mutagenesis , and no mechanism has been described for this increase [9] . Environmental agents have also been shown to affect mitochondrial DNA replication both positively and negatively . The presence of antioxidants such as MitoQ and dihydrolipoic acid have been shown to improve mitochondrial function in mutants with disease-associated polymerase domain mutations by salvaging reactive oxygen species [14] . These results suggest an increase of oxidative damage in mtDNA in model systems with defective pol γ , a hypothesis supported by the increased levels of 8-oxo-dG in the mtDNA of a transgenic mouse model that overexpressed the Y955C mutant variant in cardiac tissue [34] . Methyl methanesulfonate ( MMS ) is an alkylating agent that is associated with increases in mtDNA base damage [35] . Interestingly , in embryonic fibroblasts , 2 mM MMS was associated with persistent mtDNA damage but not with loss of mtDNA or mitochondrial function [36] . In yeast , repair of alkylation damage of mtDNA by MMS involves Apn1 nuclease [37] , and Ntg1 [38] . Also , chronic exposure to trace amounts of the known human carcinogen , cadmium chloride , resulted in loss of mitochondrial function [39] . Because cadmium also results in extreme nuclear hypermutability [39] , the possibility that cadmium alters mtDNA replication warrants further study . Base excision repair is active in yeast and human mitochondria and protects cells against alkylation damage [40] . However , lesions on single-stranded DNA are not substrates for base excision repair because there is no complementary strand with which it can reanneal and are therefore highly mutagenic [41]–[43] . The proposed model of asymmetrical mtDNA replication of human mtDNA suggests that single-stranded mtDNA is exposed even in optimal conditions and could be more vulnerable under conditions of decreased replication efficiency [44] . To test whether reduction in mtDNA replication efficiency could leave the cell vulnerable to mutagenic base damage , mtDNA mutagenesis in previously characterized disease-associated mutants were tested in the presence of MMS .
Mutations in mip1 that result in changes in conserved amino acids previously have been shown to cause defective mtDNA replication and , in some cases , increased mtDNA mutagenesis when coexpressed with wild type MIP1 [9] . These experiments were performed in haploid heteroallelic strains with intact chromosomal wild type MIP1 and one of 31 mutant mip1 alleles on a centromeric plasmid with the endogenous promoter . This study interrogated some of the heteroallelic strains and newly created diploid heterozygotes to determine mtDNA point mutagenesis of the gene encoding the 16S ribosomal subunit that confers resistance to erythromycin and the fraction of cells unable to grow on glycerol which requires mitochondrial function . To test whether the presence of the catalytically defective mutant variant increases the vulnerability of mtDNA to base damage and mutagenic replication by the wild type polymerase , heteroallelic S . cerevisiae strains expressing either Q264H , R656W , R853H , or wild type MIP1 on a centromeric plasmid and chromosomal wild type MIP1 were grown in the presence of sublethal concentrations ( 3 mM ) of MMS ( see Table 1 for list of all genotypes ) . None of these mutant proteins are capable of maintaining mtDNA without the presence of wild type Mip1 [9] . MMS exposure caused a modest 2-fold increase in mutagenesis in the wild type control as compared to no exposure to MMS , whereas a greater increase in mtDNA mutagenesis—17-fold , 11-fold , and 6-fold—was observed in strains expressing Q264H , R656W , and R853H , respectively ( Figure 1 and Table 1 ) . Absolute mtDNA mutant frequencies after MMS exposures were associated with 30-fold , 18-fold , and 7-fold increases in strains with Q264H , R656W , and R853H mutant variants , respectively , compared with that of the wild type strain . The control strain with wild type MIP1 on both the centromeric plasmid and the chromosome was associated with only 2 . 7-fold increase in MMS-induce mutagenesis compared with no MMS exposure . To avoid the possibility of multicopy expression of the Mip1 variant from the plasmid , heterozygotes were created with mutations that encode Mip1 with defective exonuclease activity [31] or amino acid variants Q264H , R656Q , G651S , or D891A ( Table 1 ) . D891A is not associated with mitochondrial disease , but the conserved aspartate in human POLG ( Asp1135 ) is essential for binding catalytic Mg2+ in the active site [19] . Alanine substitution of this equivalent residue in other human DNA polymerases has been shown to disrupt binding of the catalytic Mg2+ , eliminating DNA polymerase activity but not binding to DNA binding [45] . With the exception of Q264H , biochemical characterizations have been reported for the remaining mutant variants [24] . In agreement with previous observations [9] , disruption of Mip1 exonuclease activity increased mutagenesis; however , there was no additional increase in MMS-induced mutagenesis ( Figure 2 and Table 1 ) . Compared with wild type , Q264H , R656Q , and D891A heterozygotes were associated with 7-fold , 8-fold , and 19-fold increases in absolute MMS-induce mutant frequency . In fact , the resulting increases in mutant frequency were approximately equal to or greater than that of the exonuclease defective variant ( Figure 2 ) . Strikingly , MMS exposure of the D891A variant resulted in an approximately 3-fold increase in mutant frequency compared with that of the exonuclease deficient variant . These results demonstrate that catalytically inactive or less active polymerases , whether generated through a site-directed mutation or a disease-associated mutation , participate in a mechanism that causes MMS-induced mutations . Mip1 variants R656Q and G651S are homologous to human disease variant R853Q and G848S , respectively , which both exhibit ≤1% catalytic activity [24] . However , G848S uniquely displayed an approximately 5-fold reduction in DNA binding [24] . Interestingly , Mip1 G651S variant resulted in fewer MMS-induced mutations compared with the other polymerase variants , suggesting that DNA binding may be important for the mechanism . To test this hypothesis , a R656Q/G651S double mutant was created , and MMS-induced mutagenesis was compared with each single mutant . Mutagenesis after exposure to MMS in the double mutant was indistinguishable from the G651S variant and lower than in a single R656Q mutant ( Figure 2 ) , suggesting that DNA binding is an important component of the R656Q mutator effect . To test whether Mip1 mutant variants participate in the bulk of mtDNA replication , mutagenesis was measured in heterozygous diploids that had one wild type MIP1 allele and one allele containing mutations that disrupt exonuclease activity and encode the Q264H and G651S variants in cis . Unlike exonuclease-deficient Mip1 without other mutations , eliminating exonuclease activity did not increase mutagenesis in either the Q264H or G651S variant , with or without MMS exposure ( Figure 3 and Table 1 ) . In fact , mutation frequency unexpectedly decreased when the exo− and Q264H were in cis . These results suggest minimal contributions of Q264H and G651S mutant variants to mtDNA replication . Resistance to erythromycin is conferred by one mutation at any of the following nucleotides: 1950 ( G to T or G to A ) , 1951 ( A to T , A to G , or A to C ) , 1952 ( A to T or A to G ) , 3993 ( C to G ) , or an insertion of G between nucleotide 1949 and 1950 of the 21S rRNA gene ( Gen Bank accession number L36885 ) [31] , [46] , [47] . To determine if there was a change in the spectrum of mutations associated with MMS exposure , PCR fragments containing nucleotides 1797–1995 and 3895–4107 of the 16S ribosomal subunit gene from erythromycin resistant mutants were sequenced . Only one mutant was taken from each original culture to ensure that each mutation represented a separate event . In the absence of MMS , this and prior studies [46]–[48] demonstrate that A:T→G:C and A:T→T:A were the most frequent mutations ( Figure 4 and Table S1 ) . In this and previous studies , C to G mutations were not detected in wild type strains and were detected in only 5% of Δrev1 strains [46] . Interestingly , we found that exposure to MMS was associated with a significant change in mutational spectrum , wherein the most common mutation was C:G→G:C transversions in both the wild type strain ( 40% ) and Q264H heteroallelic strain ( 45% ) . G:C→A:T was the only mutation other than C:G mutation detected , but it was only detected in the wild type strain and very low levels ( 7% ) . These results suggest that cytosine or guanine is especially sensitive to methylation by MMS , leading mostly to misincorporated cytosines or guanines , similar to previous studies [43] . Increased mutagenesis can arise by disrupting exonuclease activity and/or increasing the frequency of nucleotide misincorporation events . Previous work demonstrated that mitochondrial deletions between direct repeats of 21 nucleotides were rare events that were suppressed by exonuclease activity [12] . To test whether MMS promotes deletion formation , haploid deletion reporter strains that were heteroallelic for wild type , exonuclease-deficient , or Q264H variants of Mip1 were used to measure frequency of deletions between 21 nucleotide direct repeats that flank an ARG8 insertion in the mitochondrial genome . Frequency of deletions between direct repeats was increased ( 40-fold ) in the strain with the exonuclease-deficient Mip1 variant but was not significantly different in the strain with the Q264H variant compared with wild type ( Figure 5 and Table 1 ) . MMS had no significant effect on deletion formation in any of the three strains , suggesting that the effect of MMS is specific to point mutations . Mutations associated with MMS exposure occur in strains with mutants that affect mtDNA replication , suggesting that the mechanism requires suboptimal mtDNA replication . Therefore , it is possible that an environmental agent that reduces mtDNA replication may also be associated with MMS-induced mutagenesis in wild type cells . This was tested by treating wild type and mutant mip1 strains with CdCl2 . Exposure to 3 µM CdCl2 was associated with increased petite formation frequency of about 30% with stepwise increases at 4 and 5 µM of about 60% and 80% , respectively ( Figure 6 ) in both wild type and Q264H mutants . Exposure to 5 µM CdCl2 was associated with 3 . 6-fold reduction in mtDNA among rho+ cells , from 30 . 6±4 . 0 copies per cell without CdCl2 exposure to 8 . 5±0 . 9 copies per cell with 5 µM CdCl2 , suggesting that trace amounts of CdCl2 are associated with mtDNA depletion . Mitochondrial DNA mutagenesis was assayed in homozygous wild type diploid cells and Q264H heterozygotes to test if MMS-induced mtDNA mutagenesis occurs with CdCl2 . Exposure to 4 µM CdCl2 had no effect on mtDNA mutagenesis in either the wild type or Q264H mutant strains ( Figure 7 ) . Therefore , the mutagenic effect of CdCl2 is specific to nuclear DNA [39] . Exposure to both MMS and CdCl2 resulted in no increase in mtDNA mutagenesis . However , the mutagenic effect of MMS observed in the heteroallelic Q264H strain was completely suppressed by 3 µM or 4 µM CdCl2 . The lack of mtDNA mutagenicity and the suppression of the MMS-induced mtDNA mutagenesis by trace amounts of CdCl2 were recapitulated in heterozygotes expressing the D891A mutant variant ( Table 1 ) . Although reduced efficiency of mtDNA replication by a mutant variant is associated with MMS-induced mutagenesis , these results suggest that processes that reduce mtDNA replication suppress MMS mutagenesis .
This study demonstrates a novel mechanism of MMS-induced mtDNA point mutagenesis , mostly C:G→G:C transversions , in heterozygous strains with a disease-associated mutation that disrupts polymerase activity . The frequency of the mutagenesis appeared to be modulated by the activity of the mutant variant in that the possible DNA binding defect observed in the human homologue to G651S reduced MMS-induced mutagenesis . This study showed that alleles with added exonuclease defect were not associated with increased mutagenesis ( with or without MMS exposure ) , suggesting that the mutant variant replicated little or none of the mtDNA that remained and was propagated in the cell . Finally , chronic exposure to trace amounts ( 3–5 µM ) of CdCl2 resulted in the inability to replicate mtDNA , which surprisingly did not increase but instead suppressed MMS-induced mutagenesis . These results are the first to support a mechanism to understand the interplay between polymerases in heterozygous cells and reveal a novel pathway for environmentally-induced mtDNA mutagenesis . There are few known pathways that increase mtDNA mutations in yeast or humans . Mutations that disrupt the exonuclease activity of the mtDNA polymerase have been shown to cause increased mtDNA mutagenesis . One study used a reversion assay in yeast to identify several mitochondrial mutators including pos5 , a gene that encodes an NADPH kinase [49] . Other genes associated with increased mtDNA mutagenesis—such as hap2 , fen1 , and ntg1—have been identified , but their effect on mtDNA mutagenesis has been modest or requiring long incubation times [49]–[51] . Even base damaging agents such as H2O2 and MMS are associated with modest increases in mtDNA but only in strains without crucial repair pathways [52] . Previously , disease-associated mutations were shown to increase mtDNA mutagenesis , but these mutations also led to the inability to maintain functional mitochondria because of mtDNA depletion [9] . The increase in mutant frequency in some strains was evident in this study in the unexposed controls ( Figures 1 and 2 ) . Therefore , it was difficult to ascertain how mutant polymerases that in some cases were suggested to have little or no activity could significantly increase mtDNA mutagenesis . Exposing the heteroallelic strains to sublethal doses of MMS resulted in up to 30-fold increases in mtDNA mutant frequency ( Figure 1 ) . Interestingly , repeating the experiment in heterozygous strains recapitulated the MMS-induced increase , albeit at a lower frequency ( Figure 2 ) . The heteroallelic strain contains the mutant mip1 and its endogenous promoter on a centromeric plasmid that has been previously shown to have 1–2 copies of the gene per cell [13] . It is possible that MMS-induced mutagenesis is sensitive to differences in the number of Mip1 mutant copies . The fact that the heteroallelic strains are haploid whereas the heterozygotes are diploid could suggest that increased copy number of other replication proteins may alter the MMS-induced mutagenesis phenotype , although there is probably no difference in the protein concentration relative to the genome copy number . MMS-induced mutagenesis was shown in several disease-associated mutants and the polymerase defective mutant but not the exonuclease-deficient mutant . Although Q264H is an exonuclease domain disease-associated variant , it was previously shown to be detrimental to mtDNA replication . Mutations that result in the Mip1 R656Q and G651S variants are homologous to the mutations in the human pol γ thumb domain that were biochemically characterized to have <1% polymerase activity including a 5-fold reduction in DNA binding affinity in the G651S homologue [24] . Interestingly , G651S is associated with reduced mtDNA mutant frequency and suppression of R656Q when the two mutations are in cis . These results suggest that lower DNA binding affinity impedes the mechanism of MMS-induced mutagenesis . Gly651 is in a stretch of amino acids conserved between humans and yeast , making it likely that Gly651 is also involved in DNA binding . However , future studies will be necessary to show that G651S does not possess other characteristics ( eg , decreased stability ) that impair MMS-induced mutagenesis . Biochemical evidence suggesting that some disease-associated mutant variants were impaired for mtDNA replication was further supported by the observation that the mutant variants by themselves could not maintain mtDNA [9] . However , it has been unclear to what extent these polymerases function in the cell . It is well known that one of the catalytic aspartates ( Asp891 in Mip1 ) is necessary for mtDNA replication; therefore , D891A is unable to catalyze the polymerase reaction . Interestingly , the heterozygous strain with D891A was associated with the largest increase in MMS mutagenesis , approximately 3-fold more than the exonuclease-deficient strain . Because the generation of mutations requires DNA replication , this result indicates that the wild type polymerase , which is normally accurate , becomes more likely to incorporate the wrong nucleotide in the D891A strain upon MMS exposure . In the case of the disease-associated mutations , it is possible that the mutant variants contribute to the incorporation of the incorrect nucleotide . However , removal of exonuclease activity in cis with Q264H and G651S showed no increase in mutation frequency regardless of MMS exposure and even showed an unexpected reduction of mutagenesis in the Q264H strain . Considering that the mutant frequency in the Q264H/exo− strain was similar to the G651S mutant frequency , this study cannot discount the possibility that the combination of the two mutations may have similar characteristics to G651S ( eg , lower DNA binding affinity ) . Regardless , these results suggest that the mutant polymerase is not contributing directly to the mutations that drive the mutant frequency . It is possible that the mutant variants replicate mtDNA molecules that may be selected against or are not propagated , possibly because of incomplete replication . Resistance to erythromycin is associated with a limited mutational spectrum that rarely includes C:G→G:C transversions [46]–[48] . Interestingly , exposure to MMS dramatically changed the mutation spectrum such that 40–45% of the mutations were C:G→G:C transversions . These mutations could either arise from a cytosine incorporated opposite a cytosine or a guanine incorporated opposite a guanine . Previously , MMS exposure of artificially formed or random ssDNA in yeast was associated with increased frequency of all 3 kinds of cytosine substitutions , C→T , C→G and C→A [42] , [43] . The mutation spectra and strand bias suggested that N3-methyl cytosine is the prominent mutagenic lesion caused by MMS lesion in the yeast nuclear ssDNA [42] , [43] . Therefore , it is possible that C:G→G:C transversions that predominate after MMS exposure result from cytosine incorporation opposite of a methylated cytosine . These combined results support a model ( Figure 8 ) wherein disease-associated mutant polymerase variants bind to and temporarily stall mtDNA replication , an event that could result in ssDNA intermediates ( eg , because of polymerase-helicase uncoupling ) . Although mtDNA base excision repair would normally repair most damaged bases in dsDNA , there are no known repair systems that act on ssDNA in yeast . In the absence of MMS , the endogenous sources of DNA damage ( eg , oxidative damage ) impart a small but significant amount of DNA damage , whereas MMS magnifies the damage . Either the mtDNA continues to be stalled and degraded ( an event that would not yield a mutant colony ) or polymerase switching might occur , allowing the wild type polymerase access to the replication fork . In some cases the polymerase would incorporate the incorrect nucleotide leading to the development of a mutation . This model suggests that the mutant polymerase would stably bind DNA but be unable to replicate mtDNA efficiently . Recently it was shown that the exonuclease domain suppresses mtDNA deletions between 21-mer direct repeats up to 160-fold [12] . One hypothesis could be that MMS also inhibits exonuclease activity which would lead to an increase in mutagenesis . This hypothesis was unlikely because MMS exhibited a modest effect on the diploid wild type strain . This study showed that MMS did not increase the frequency of mtDNA deletion mutants in wild type or Q264H background suggesting that the mechanism that caused MMS-induced point mutations is different than that of mtDNA deletion formation . Furthermore , MMS does not alter the exonuclease activity involved in suppression of mtDNA deletion formation . A previous report showed that exposure to trace amounts of CdCl2 was associated not only with suppression of nuclear mismatch repair but also increase in petite colony formation frequency [39] . This study recapitulates this finding and further shows that CdCl2 does not significantly affect mtDNA mutant frequency . These results indicate that either there is no efficient mismatch repair system in yeast mtDNA or the amount of CdCl2 needed to observe a defect in mismatch repair is similar to the amount that is associated with a high frequency of dysfunctional mitochondrial . The only mismatch repair homologue associated with yeast mitochondria is Msh1 , and it has been proposed to play a role in base excision repair [48] . This study also showed that mtDNA content was reduced in rho+ cells exposed to CdCl2 suggesting that cadmium negatively affects mtDNA replication or maintenance . We tested whether the effect of inhibiting mtDNA replication could mimic the effect of a disease-associated mutation by increasing MMS-induced mutagenesis . Unexpectedly , 3 and 4 µM CdCl2 did not promote MMS-induced mutagenesis in wild type cells . It is possible that CdCl2 does not stall replication but rather affects another replication-related process , such as replication initiation or termination . Even more unexpectedly , CdCl2 suppressed MMS-induced mutagenesis associated with Q264H . It should be noted that mutant frequency is determined among only rho+ cells so increased petite colony formation frequency from CdCl2 exposure does not explain the suppression of mutant frequency . One possible explanation is that the presence of CdCl2 selects against the maintenance of mtDNA molecules that are damaged or stalled either through some direct inhibition of the enzymes involved or as a result of a cellular response to mtDNA stress . Another possibility is that damaged mtDNA is more sensitive to CdCl2-induced inhibition of mtDNA replication , and the replication of these mtDNA molecules are not completed or maintained during the growth of the colony . Therefore , the various combinations of environmental exposures and genetics may be a useful tool to understand different pathways that occur in response to DNA damage . This study shows a novel gene-environment interaction which greatly increases mtDNA mutagenesis and supports a polymerase-switching mechanism that has not been described in mtDNA replication . The interpretations of this study are limited in humans because there are key differences in yeast mtDNA maintenance compared with mammalian mtDNA ( eg , lower mtDNA copy number and high frequency of recombination ) . Also , the study assumes that the mutant frequency ( ie , mutants per culture ) is indicative of the mutation frequency ( mutations per mitochondrial division ) . True mutation frequencies would require information on mtDNA kinetics and mitochondrial dynamics that is currently unavailable . However , with the advent of high-throughput genome sequencing , similar studies in mtDNA mutagenesis will be possible in a model system that more closely mimics human mtDNA . Interestingly , this interaction involves heterozygotes containing mutations which are normally associated with disease [53] . It is interesting to consider that if a similar mechanism occurs in mammalian mtDNA , people who are heterozygous for a disease-associated mutation could be sensitive to environmental exposures that would impair mtDNA replication and promote symptoms involved in mitochondrial toxicity or disease .
S . cerevisiae strains were grown at 30°C in YP ( yeast extract 1% , peptone 2% ) with 2% glucose or glycerol as carbon sources or synthetic complete media . Escherichia coli strains were grown in standard LB media at 37°C . When appropriate , gentamicin and ampicillin were added to YPD ( 0 . 2 mg/ml ) and LB ( 0 . 1 mg/ml ) , respectively . The plasmid , pFL39 , which contains MIP1 on a centromeric plasmid was previously described [31] . Site-directed mutagenesis of plasmid-encoded MIP1 was performed using the QuikChange Site-Directed Mutagenesis Kit ( Invitrogen ) as described previously . The construction of plasmids containing ACT1 , COX1 , and COX2 fragments , used for mtDNA quantitation was previously described . All S . cerevisiae strains were derived from E134 ( MATα ade5-1 his7-2 lys2-A14 leu2-3 , 112 ura3-52 trp1-289 ) [54] and its MATa isogenic strain , YH747 . Heteroallelic mip1 strains were made by transforming pFL39-MIP1 or a mutant derivative into E134 by selecting for TRP . Strains that measure deletions between direct repeats were created by transforming TRP1-containing plasmids PFL39 [12] containing wild type MIP1 or mip1 encoding an exonuclease-deficient mutant variant ( mip1-exo [Het]; JSY114 ) into trp1::G418 NPY75 ( JSY77 ) [12] . Chromosomal mip1 mutations were created in haploid E134 with wild type MIP1 using a PCR-based delitto perfetto method as previously described . All strains were checked phenotypically for the absence of the CORE casette used in delitto perfetto ( ie , screening for Ura− and gentamicin sensitive cells ) and were sequenced to confirm the presence of the mutation and the absence of undesired nearby mutations . The mip1 haploid mutant strains were mated with the isogenic E134 with mating type a . Resistance to erythromycin is conferred by one of several missense mutations in the 21S rRNA gene in mitochondrial DNA [46] , [47] , [55] . Yeast strains were replica plated onto YPD plates with or without 1–5 mM MMS or 1–5 µM CdCl2 and grown at 30C for 2 days . In most cases 3 mM MMS was the highest exposure that allowed growth and was used for the experiments . After plating on CdCl2 , increased chromosomal mutagenesis was confirmed qualitatively by the presence of colonies on media lacking lysine . From these plates , 20–40 independent colonies from each strain were used to inoculate into 4 ml of synthetic media without tryptophan ( heteroallelic strains ) or YPD ( heterozygous strains ) , and these cultures were incubated to saturation ( for 2 days ) at 30°C . The cells were plated on YPEG ( 1 . 7% ethanol , 2% glycerol ) with 4 g/L erythromycin . A small aliquot of 5–10 cultures were used to titer the number of rho+ cells by plating 10−5 dilutions on YPG . Erythromycin resistant colonies were counted after 6 days of incubation at 30°C . The mutant frequency was the median number of erythromycin colonies per 108 rho+ cells plated . To determine the spectrum of erythromycin-resistant mutations , DNA was extracted from one mutant per culture and used for as a template in a PCR reaction to amplify two regions , approximately 200 nucleotides flanking nucleotides 1950 ( using 5′-GAGGTCCCGCATGAATGACG and 5′-CGATCTATCTAATTACAGTAAAGC ) and 3993 ( using 5′-CTATGTTTGCCACCTCGATGTC and 5′-CAATAGATACACCATGGGTTGATTC ) . The resulting amplified DNA was the template for the sequence reactions . To measure deletions between direct repeats , all strains were replica plated onto YPD with or without MMS exposure as described above . Independent cultures were grown from at least 20 colonies at 30°C in YP ( yeast extract 1% , peptone 2% ) with 2% glucose and adenine for 2 days . Appropriate dilutions of samples from the saturated cultures were plated on synthetic complete media lacking arginine to determine total number of cells with mtDNA . The cultures were plated onto YP with 2% glycerol and deletion mutants were counted after 4 days . The mutant frequency was determined as the median number of mutant colonies per 108 Arg+ cells . In all mutagenesis experiments , 95% confidence levels were determined using the method of the median . Petite frequencies are the frequency of rho− cells ( petites ) in the total population . Rho−cells are devoid of mitochondrial functions but are not necessarily devoid of mtDNA ( rho0 ) . To determine petite frequency in heteroallelic strains , at least 12 fresh transformants per strain were diluted in water and between 200–1000 colonies were plated onto YPD . For monoallelic strains , rho+ cells derived from several tetrad dissections were single colony purified on YPD and then assayed for petite frequency . Cells were incubated at 30°C for 2 days . Rho+ cells were identified either by the accumulation of red pigment as a result of mutations in the ade biosynthetic pathway or by the inability to grow on YPG . Using either or both method , at least 300 colonies per plate were counted , and petite colonies were identified . Frequencies were determined for each plate , and the median number of the frequencies was calculated . 95% confidence levels were determined by using the method of the median ( 52 ) . For monoallelic strains determined to be 100% petite , no rho+ cells could be isolated from cells derived from 10 different haploid spores . Mitochondrial DNA copy number was quantified relative to nuclear DNA copy number using real time PCR . Primers and probes designed to specifically amplify within the mitochondrial-encoded COX1 gene and the nuclear-encoded ACT1 gene . Real time PCR reactions using Taqman Universal PCR Master Mix ( Applied Biosystems ) were performed at 40 cycles of 95 for 30 sec and 50 degrees for 30 sec . Known concentrations of plasmid molecules containing COX1and ACT1 were quantified as a positive control [9] and real time PCR was performed on 7 different dilutions to determine a logarithmic equation of a curve ( R2 values>0 . 98 ) that represents numbers of molecules as a function of the critical threshold of every reaction . Every reaction was done in triplicate , and three replicates were tested for each experimental condition . Data represent the average ratio ( ± SEM ) of the number of COX1 molecules to the number of ACT1 molecules . | Thousands of mitochondrial DNA ( mtDNA ) per cell are necessary to maintain energy required for cellular survival in humans . Interfering with the mtDNA polymerase can result in mitochondrial diseases and mitochondrial toxicity . Therefore , it is important to explore new genetic and environmental mechanisms that alter the effectiveness and accuracy of mtDNA replication . This genetic study uses the budding yeast to demonstrate that heterozygous strains harboring disease-associated mutations in the mtDNA polymerase gene in the presence of a wild type copy of the mtDNA polymerase are associated with increased mtDNA point mutagenesis in the presence of methane methylsulfonate , a known base damaging agent . Further observations suggest that the inability of disease-associated variants to replicate mtDNA resulted in increased vulnerability to irreparable base damage that was likely to result in mutations when replicated . Also , this study showed that trace amounts of the environmental contaminant cadmium chloride impairs mtDNA replication but eliminates damage-induced mutagenesis in the remaining functional mitochondria . This interplay between disease-associated variant and wild type polymerase offers new insights on possible disease variation and implicates novel environmental consequences for compound heterozygous patients . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"mutagenesis",
"biochemistry",
"mitochondrial",
"genetics",
"medicine",
"and",
"health",
"sciences",
"dna",
"polymerase",
"polymerases",
"dna",
"replication",
"proteins",
"mitochondrial",
"dna",
"forms",
"of",
"dna",
"genetics",
"biology",
"and",
"life",
"sciences",
"... | 2014 | MMS Exposure Promotes Increased MtDNA Mutagenesis in the Presence of Replication-Defective Disease-Associated DNA Polymerase γ Variants |
Our investigations show that nonlethal concentrations of nitric oxide ( NO ) abrogate the antibiotic activity of β-lactam antibiotics against Burkholderia pseudomallei , Escherichia coli and nontyphoidal Salmonella enterica serovar Typhimurium . NO protects B . pseudomallei already exposed to β-lactams , suggesting that this diatomic radical tolerizes bacteria against the antimicrobial activity of this important class of antibiotics . The concentrations of NO that elicit antibiotic tolerance repress consumption of oxygen ( O2 ) , while stimulating hydrogen peroxide ( H2O2 ) synthesis . Transposon insertions in genes encoding cytochrome c oxidase-related functions and molybdenum assimilation confer B . pseudomallei a selective advantage against the antimicrobial activity of the β-lactam antibiotic imipenem . Cumulatively , these data support a model by which NO induces antibiotic tolerance through the inhibition of the electron transport chain , rather than by potentiating antioxidant defenses as previously proposed . Accordingly , pharmacological inhibition of terminal oxidases and nitrate reductases tolerizes aerobic and anaerobic bacteria to β-lactams . The degree of NO-induced β-lactam antibiotic tolerance seems to be inversely proportional to the proton motive force ( PMF ) , and thus the dissipation of ΔH+ and ΔΨ electrochemical gradients of the PMF prevents β-lactam-mediated killing . According to this model , NO generated by IFNγ-primed macrophages protects intracellular Salmonella against imipenem . On the other hand , sublethal concentrations of imipenem potentiate the killing of B . pseudomallei by NO generated enzymatically from IFNγ-primed macrophages . Our investigations indicate that NO modulates the antimicrobial activity of β-lactam antibiotics .
B . pseudomallei are endemic in tropical areas of Southeast Asia , Northern Australia and equatorial countries [1] . This Gram-negative , opportunistic pathogen is a saprophyte that inhabits water and soil , becoming infectious to humans and animals if inoculated through cutaneous abrasions , ingested in contaminated food and water , or inhaled through the respiratory mucosa . Melioidosis can present as an acute , chronic or latent infection [2] . Pneumonia accounts for about 50% of all the cases of B . pseudomallei infection [3] , [4] , whereas septic shock , often a fulminant complication of septicemia , kills 40% of melioidosis patients receiving therapy and 95% of those untreated . Despite recent advances in antibacterial therapy , management of melioidosis remains a challenge [4] . Antibacterial treatment of melioidosis often spans 20 weeks and requires combined antibiotic therapy . Ceftazidime is often used in the intensive phase , whereas trimethoprim-sulfamethoxazole ( TMP-SMX ) is used during the eradication phase of treatment [5] . Regardless of intense and vigorous treatment regimes , about 10% of melioidosis patients suffer from relapses [6] . B . pseudomallei are intrinsically resistant to most classes of antibacterials [7] . For example , B . pseudomallei growing in biofilms are phenotypically tolerant to doxicycline , ceftazidime , imipenem and TMP-SMX [8] , [9] . The efflux pumps BpeAB-OprB , BpeEF-OprC and AmrAB-OprA further increase the resistance of this opportunistic pathogen to β-lactams , aminoglycosides , macrolides , fluoroquinolones , chloramphenicol and polymyxins [10]–[12] . Class A and D β-lactamases add to the arsenal of enzymatic systems that protect B . pseudomallei against ampicillin , carbenicillin , ceftazidime and imipenem [13]–[15] . In addition to these well-characterized mechanisms of antibiotic resistance , changes in bacterial physiology in response to host environmental conditions may promote resistance to antibiotics . For example , anaerobiosis , which is normally attained in the hepatic , splenic and prostate abscesses of melioidosis patients , induces a population of B . pseudomallei remarkably refractory to several classes of clinically important antibacterials [16] . In addition to being an intrinsic component of the antimicrobial arsenal of vertebrate hosts [17] , the signaling properties of NO have been co-opted by prokaryotic and eukaryotic organisms . NO produced endogenously by bacterial NO synthase protects Bacillus subtilis against a wide spectrum of antibiotics [18] . This adaptive response of Bacillus might lessen the bactericidal activity of antibiotics produced by saprophytic microorganisms populating the soil . Modification of drugs and potentiation of antioxidant defenses have been evoked as mechanisms underlying the NO-induced antibiotic resistance of Bacillus [18] . NO produced in the inflammatory response has also been shown to shield Gram-positive and –negative bacteria against the antimicrobial activity of bactericidal antibiotics . Salmonella enterica survives exposure to members of the aminoglycoside family in response to the NO generated intracellularly by IFNγ-activated macrophages [19] , a situation that had previously been noted for Listeria with ampicillin [20] . Given the recently described role of NO in inducing resistance of phylogenetically diverse bacteria to different classes of antibiotics and the recent controversy attributing oxidative stress as the mechanism of action of bactericidal antibiotics [18] , [21]– , we tested whether NO generated chemically or enzymatically modifies the antimicrobial activity of β-lactams against B . pseudomallei and two representative members of the enterobacteriaceae family .
Strain K96243 , a clinical isolate of B . pseudomallei [24] , was grown in the BSL3 laboratory of the Department of Microbiology at the University of Colorado School of Medicine . This facility has been certified by the CDC for work with select agents . E . coli strain 3110 and S . enterica serovar Typhimurium strain 14028 s were also used in the course of these investigations . Where indicated , Salmonella strains AV0468 , AV07140 and AV07141 deficient in the flavohemoprotein hmp , acetate kinase ackA or phosphotransacetylase pta , respectively , were used . The bacteria were grown overnight to stationary phase in LB broth supplemented with 4% ( v/v ) glycerol ( LBG ) at 37°C and 315 RPM in a shaker incubator ( New Brunswick Innova , Edison , NJ ) . Where indicated , the bacteria were grown to log phase to OD600 of 0 . 6 . Log phase B . pseudomallei was grown from overnight cultures in a shaker incubator at 37°C in LBG broth to an OD600 of 0 . 6 . Log and stationary phase B . pseudomallei cultures were diluted to OD600 of 0 . 012 in 1 ml of LBG broth in 14 ml polypropylene tubes containing sterile stirrer magnets . The killing activity of imipenem was assessed in bacterial cultures at the indicated concentrations . The tubes were loosely capped and placed on a magnetic stirrer in a 37°C cell culture incubator . The anti-B . pseudomallei activity of ceftazidime was tested in 250 ml flasks as previously described [16] . Both antibiotics were purchased from Sigma-Aldrich , St . Louis , MO . Selected cultures were co-treated with spermine NONOate or DETA NONOate , which generate NO with half-lives of 39 min and 20 h , respectively , at 37°C , pH 7 . 4 . In selected experiments the susceptibility of log phase E . coli and Salmonella was also tested . Where indicated , E . coli and Salmonella were grown in EG medium [i . e . , E salts ( 0 . 2 g/L MgSO4 , 2 g/L C6H8O7-H2O , 10 g/L K2HPO4 , 3 . 5 g/L Na ( NH4 ) HPO4-4H2O ) supplemented with 0 . 4% glucose] . The number of surviving bacteria after antibiotic treatment was determined after culture on LB agar plates , and the fraction of bacteria that survived antibiotic treatment was calculated as ( cfu tn/cfu t0 ) ×100 . A mini-Mariner transposon [25] , [26] expressed from the suicide plasmid pTBurk1 was electroporated into B . pseudomallei strain K96243 . This plasmid contains the Himar1 transposable element with a kanamycin cassette flanked by inverted repeats ( IR ) . The Himar1 transposase was chosen because of its TA dinucleotide specificity . Bacteria with an integrated kanamycin cassette were selected on LB agar plates containing 50 µg/ml kanamycin and 100 µg/ml zeocin . The sequencing libraries were quantified using the Agilent Bioanalyzer DNA7500 chip , multiplexed , cluster amplified , and sequenced on the Illumina MiSeq platform . Sequencing reads containing the Himar1 IR sequence and the adjacent TA were isolated from the raw fastq file . The IR sequence was removed from the analysis . The processed reads were mapped onto the B . pseudomallei K96243 reference genome using the program Bowtie2 with local alignment settings and a k value of 1 [27] . Annotation of TA sites was accomplished using seqanno , a custom set of Python scripts ( https://github . com/brwnj/seqanno ) used to characterize specific genomic sequences . The publicly available code was used to quantify reads over a given sequence , annotate those counts at the gene level , compare results between samples , and annotate using a UniProt flat file for B . pseudomallei . B . pseudomallei grown overnight in LBG broth were subcultured 1∶100 in LBG broth at 37°C with shaking . The generation of H2O2 by B . pseudomallei was measured in stationary phase bacteria diluted to an OD600 of 0 . 5 . Selected bacterial cultures were treated for 1 h at 37°C with 12 . 5 µg/ml imipenem in the presence or absence of 100 µM spermine NONOate or 500 µM KCN . The cultures were continuously agitated with a magnetic stir-bar . The specimens were placed into a sealed , temperature-controlled chamber ( World Precision Instruments , Inc . , Sarasota , FL ) containing a small magnetic stir bar . To prevent artifacts associated with a possible loss of viability , the experiments were carried out for 1 h after exposure to imipenem before the onset of killing took place . The H2O2 accumulated in the bacterial cultures after 1 h incubation was measured pollarographically for about 2 min using an ISO-H2O2 sensor attached to an APOLLO 4000 free radical analyzer ( World Precision Instruments ) . The concentration of H2O2 produced by the bacterial cultures was calculated by regression analysis of a standard curve generated with known concentrations of H2O2 . B . pseudomallei was grown overnight in LBG broth at 37°C with shaking . Overnight cultures were diluted to OD600 of 0 . 5 in a volume of 1 ml . Where indicated , 100 µM spermine NONOate or 500 µM KCN were added to the bacterial cultures . The samples were transferred to a multiport temperature-controlled chamber , and the consumption of O2 by the bacteria was measured over a 5 min period using an ISO-OXY-2 O2 sensor attached to an APOLLO 4000 free radical analyzer . To ensure the even distribution of gases in the chamber , the samples were placed on a chamber containing a magnetic stir-bar . The data are expressed as µM of O2 . Nitrate reductase enzymatic activity was monitored by measuring the accumulation of NO2− in the cultures . Bacterial pellets of B . pseudomallei grown overnight in LBG broth diluted to an OD600 of 0 . 6 were moved into the anaerobic chamber , where they are aliquoted into 1 ml volumes in LBG broth or LBG broth supplemented with 50 mM NaNO3− . The O2 in the LBG broth had been eliminated by culturing the media in the anaerobic chamber for at least 24 h . The bacterial cells were allowed to reduce NO3− to NO2− for 2 . 5 h . NO2− concentrations were measured spectrophotometrically at 550 nm after mixing with an equal volume of Griess reagent ( 0 . 5% sulfanilamide and 0 . 05% N-1-naphthylethylenediamide hydrochloride in 2 . 5% phosphoric acid ) . NO2− concentrations were calculated by regression analysis using standard curves prepared with NaNO2 . The membrane potential of S . Typhimurium grown in LB broth and EG medium to OD600 of 0 . 5 was measured with the fluorescent probe DiSC3 ( 5 ) ( Molecular Probes , Eugene , OR ) . The pellet of 1 mL of cells grown to log phase in LB broth or EG medium was resuspended in 5 mM HEPES , pH 7 . 2 , supplemented with 5 mM casamino acids or 5 mM glucose , respectively . Samples were treated in 1 ml aliquots with 750 µM spermine NONOate for 15 minutes at 37°C . DiSC3 ( 5 ) was added to a final concentration of 1 µM from a stock solution made in DMSO . DiSC3 ( 5 ) was allowed to equilibrate in the cells before fluorescence measurements were collected in a Synergy 2 microtiter plate reader ( BioTek , Winooski , VT ) using excitation and emission wavelengths of 590 and 680 nm , respectively . J774 murine macrophage-like cells ( clone ATCC TIB-67 ) were grown in RPMI medium supplemented with 10% fetal bovine serum ( BioWhittaker , Walkersville , MD ) , 15 mM Hepes , 2 mM L-glutamine , 1 mM sodium pyruvate ( Sigma-Aldrich , St . Louis , MO ) , and 100 U·ml−1/100 mg·ml−1 of penicillin/streptomycin ( Cellgro ) . The macrophages were treated with 200 U/ml of recombinant murine IFNγ ( Peprotech , Rocky Hill , NJ ) 16 h before infection . The macrophages were infected for 2 . 5 h with B . pseudomallei at an MOI of 4 , after which the media was exchanged with RPMI+ containing 350 µg/ml kanamycin . One hour later , B . pseudomallei-infected cells were incubated for 4 h in fresh culture media containing 250 µg/ml kanamycin . The media was then replaced with fresh media containing increasing concentrations of imipenem in the presence or absence of 500 µM of the iNOS inhibitor aminoguanidine . In parallel experiments , macrophages were infected for 25 min with S . Typhimurium at an MOI of 2 . Extracellular Salmonella were killed after treatment for 1 h with 50 µg/ml gentamicin . The Salmonella-infected macrophages were then incubated in fresh RPMI media containing 10 µg/ml gentamicin in the presence or absence of aminoguanidine , which was maintained in the culture media for the rest of the experiment . After 8 h , the Salmonella-infected cells were washed and fresh media containing imipenem was added to the cultures . The B . pseudomallei and Salmonella burden in the cultures was determined 12–14 h after exposure to imipenem . The amount of nitrite , a terminal oxidative product of NO , synthesized by the macrophages was estimated by the Griess reaction . The data were analyzed using a Student's paired t test . Determination of statistical significance between multiple comparisons was achieved using one-way analysis of variance ( ANOVA ) followed by a Bonferroni post-test . Data were considered statistically significant when p<0 . 05 .
The β-lactam antibiotic imipenem has been used in the clinic to treat people with melioidosis [7] , [28] . Under the experimental conditions tested , stationary B . pseudomallei did not grow 2 . 5 h after subculture in LB broth supplemented with 4% glycerol ( figure 1A ) . Despite this lack of growth , 1 µg/ml imipenem reduced the viability of stationary phase B . pseudomallei by ∼1 , 000-fold ( figure 1B ) . These findings contrast with those reported earlier by Eng et al who found poor antimicrobial activity of imipenem against nongrowing bacteria [29] . Differences in bacterial species might account for these discrepancies . As expected , imipenem effectively killed log phase B . pseudomallei ( figure 1B ) , a population that double in numbers 2 . 5 h after culture in fresh LBG broth ( figure 1A ) . Together , our investigations indicate that imipenem can be equally efficient at killing both replicating and non-replicating B . pseudomallei . Our investigations also indicate that the imipenem-dependent inhibition of both peptidoglycan remodeling in stationary phase bacteria and de novo peptidoglycan biosynthesis in growing B . pseudomallei can exert profound antimicrobial activity . The NO donor spermine NONOate , which has a half-life of 39 min , was used to test whether this diatomic radical abrogates the antimicrobial activity of imipenem . Given that B . pseudomallei is extraordinarily susceptible to the antimicrobial activity of NO [30] , spermine NONOate was titrated in order to find conditions in which the viability of B . pseudomallei was not affected upon NO treatment . The addition of 100–200 µM spermine NONOate , which generates 2 moles of NO per mole of parent compound , failed to kill B . pseudomallei under the experimental conditions used in the course of these investigations . The concentrations of NO used in these experiments effectively inhibited growth of log phase B . pseudomallei ( not shown ) . The addition of 100 µM spermine NONOate completely abrogated killing by imipenem against both stationary and log phase B . pseudomallei ( figure 1C ) . We noticed that the colonies of B . pseudomallei treated simultaneously with imipenem and NO took even longer to grow that those of NO-treated controls , indicating that NO does not prevent imipenem from poisoning penicillin-binding proteins in the cell wall . The protective effects appear to be explained by the NO released by spermine NONOate and not the polyamine base , since spermine did not affect the imipenem-dependent killing of B . pseudomallei ( not shown ) . We also tested the effects of NO on the anti-B . pseudomallei activity of ceftazidime , which is the β-lactam antibiotic of choice in the acute phase of treatment of melioidosis [31] . Ceftazidime failed to kill B . pseudomallei under the same experimental conditions under which imipenem exerted profound bactericidal activity ( not shown ) . Therefore we adopted a system of long-term exposure to a high concentration of ceftazidime that has been shown to sustain cytotoxicity against the seemingly resistant strain of B . pseudomallei used in our studies [16] . B . pseudomallei was effectively killed 6 h after the addition of 64 µg/ml ceftazidime ( figure 1D ) . We used this in vitro culture system to test the effects of NO on ceftazidime-mediated killing of B . pseudomallei . To ensure long-term release of NO , these investigations made use of the slow NO donor DETA NONOate , which has an estimated half-life of 20 h at 37°C , pH 7 . 4 . The addition of 2 . 5 mM DETA NONOate abrogated most of the anti-B . pseudomallei activity associated with ceftazidime treatment ( figure 1D ) . Together , these findings indicate that chemically-generated NO protects B . pseudomallei against β-lactam antibiotics . The following experiments were performed in order to determine whether NO tolerizes Burkholderia against the cytotoxic actions of antibiotics or stimulates long-lasting genetic resistance . Imipenem was used to test these two models because 1 ) this β-lactam antibiotic is endowed with potent anti-Burkholderia activity , 2 ) imipenem-mediated killing occurs within a few hours of exposure , and 3 ) NO induces excellent protection against this drug . Two independent experimental approaches tested whether NO tolerizes Burkholderia or induces long-lasting genetic resistance . First , 100 µM spermine NONOate was added to the cultures 1 h after exposure to 25 µg/ml imipenem . As seen with Burkholderia co-exposed to NO and imipenem , the addition of NO 1 h after imipenem treatment abrogated killing of stationary phase B . pseudomallei ( figure 2A ) . Second , bacterial cultures were pretreated with spermine NONOate and imipenem for 1 h and then washed by centrifugation . The bacterial cells were then resuspended with fresh LBG broth containing 25 µg/ml imipenem . Again , these cultures were as protected as cultures receiving NO and imipenem during the full 2 . 5 h of challenge . These findings indicate that the protective actions afforded by NO are immediate and can occur after the bacteria have been exposed to imipenem . To test the duration of the protective effects associated with NO treatment , B . pseudomallei were treated with NO for 1 h , and then placed in fresh media containing 12 . 5 µg/ml imipenem for up to 5 additional hours . The protective effects associated with NO treatment were lost over time ( figure 2B ) . For instance , imipenem killed more than 99 . 99% of the bacteria in the population 5 h after NO was removed from the cultures . Cumulatively , our investigations indicate that the protective effects afforded by NO against imipenem are transitory and are best observed in cells actively undergoing nitrosative stress . To identify loci that may be associated with the NO-induced tolerance to imipenem , a B . pseudomallei mutant library was constructed using a mini-mariner transposon with an insertion specificity for a TA dinucleotide [25] , [26] . Sequencing of genomic DNA at the transposon-chromosome junctions allowed us to determine the coverage of the library . Overall , the library consists of 35 , 075 independent clones encompassing 28 , 543 and 6 , 532 intragenic and intergenic insertions , respectively . On average , each gene harbors 3–4 independent transposons . Disrupted genes were defined as those that sustained 6 or more sequence-reads per site within the internal 5–80% of gene length . Essential genes in Burkholderia were defined as those sustaining fewer than 6 sequence-reads within 3–97% of the open reading frame length . We estimate that 590 genes are essential for growth of B . pseudomallei under the experimental conditions tested ( table S1 ) . The estimated essential genes encode functions such as DNA replication , chromosome maintenance , lipid metabolism , translation , cell division , and energy and nucleic acid metabolism ( figure S1 ) . We used this transposon library to identify transposon mutants with increased resistance to imipenem . The transposon library was treated for 2 . 5 h with 12 . 5 µg/ml imipenem and/or 750 µM spermine NONOate at an OD600 of 0 . 5 in LBG broth . The specimens were then subcultured in 25 ml of fresh LBG broth until the bacteria reached an OD600 of 0 . 6 . The frequency of transposons in genomic DNA isolated from B . pseudomallei treated with either spermine NONOate , or spermine NONOate and imipenem was quantified by Illumina deep-sequencing as described previously for Heamophilus [32] . Sequencing data from 3 independent experiments were averaged , the fold change between samples calculated , and false discovery rate analysis determined . Sixteen genes with several transposons were found to be enriched in the group treated with imipenem and spermine NONOate ( table 1 ) . Among the positively selected genes were mutants with transposons in loci encoding cytochrome c oxidase function . In addition , the moaC and mogA genes involved in molybdenum utilization were also positively selected . Molybdenum is a common cofactor of enzymes such as nitrate reductases that allow bacteria to grow using NO3− for respiration . LBG broth and the autoxidation of the NO generated from spermine NONOate are likely sources of the terminal electron acceptor NO3− in our system . Together , the positive selection of clones bearing mutations in cytochromes and molybdenum utilization genes suggest that disruption of the electron transport chain provides B . pseudomallei a selective advantage against the killing of imipenem . In addition , the disruption of several genes associated with nucleotide metabolism , tRNA synthesis , β-lactamase processing and transcriptional regulation appear to provide a selective advantage to B . pseudomallei against the antimicrobial activity of imipenem . Given the selectivity of NO for metal prosthetic groups in the terminal oxidases of the electron transport chain and the fact that mutations in components of the respiratory chain provided a competitive advantage to Burkholderia in response to imipenem ( table 1 ) , it is possible that the antibiotic tolerance elicited in response to NO is associated with a loss in respiratory function . To test this hypothesis , we measured whether the addition of classical antagonists of terminal oxidases of the electron transport chain affects the susceptibility of B . pseudomallei to imipenem . As seen with NO , the respiratory inhibitor potassium cyanide ( KCN ) prevented the imipenem-dependent killing of B . pseudomallei ( figure 3A ) . To determine whether the concentrations of NO and KCN that protect B . pseudomallei against imipenem affect the respiratory activity of B . pseudomallei , we measured the consumption of O2 . Compared to untreated controls grown in LBG broth saturated with O2 ( figure 3B ) , bacterial cultures treated with 100 µM spermine NONOate or 500 µM KCN had reduced respiratory activity . Under the conditions tested , neither 100 µM spermine NONOate nor 500 µM KCN killed B . pseudomallei . These findings suggest that terminal cytochromes of the electron transport chain are critical molecular targets of NO-induced antibiotic resistance . Our investigations also suggest that β-lactam antibiotics require an active electron transport chain to exert their antimicrobial activity . This model is supported further by the fact that the antimicrobial activity of ceftizidime and imipenem was dramatically reduced in anaerobic cultures ( figure 3C ) . The enzymatic activity of terminal cytochrome oxidases and nitrate reductases of the electron transport chain help maintain an electrochemical gradient across the cytoplasmic membrane [33] . It is therefore possible that decreases of PMF in response to NO could mediate antibiotic tolerance . To test this idea , we evaluated the effect that collapsing the PMF has on the anti-B . pseudomallei activity of imipenem . The protonophore carbonyl cyanide 3-chlorophenylhydrazone ( CCCP ) and the ionophore valinomycin were chosen for these investigations , because these drugs dissipate the ΔH+ and ΔΨ components of the PMF by facilitating the transport of H+ and K+ down an electrochemical potential gradient . Remarkably , CCCP and valinomycin protected B . pseudomallei against the antimicrobial activity of imipenem ( figure 4A ) , supporting the model that the antimicrobial activity of this β-lactam antibiotic depends on a functional PMF . Next , we tested whether NO abrogates killing of other Gram-negative bacteria by imipenem . Surprisingly , exposure of S . enterica serovar Typhimurium or E . coli grown in LBG broth to 750 µM spermine NONOate had a small effect on the antimicrobial activity of imipenem ( figure 4B ) . To investigate whether the failure of NO to protect Salmonella against imipenem is associated with antinitrosative defenses , we compared the susceptibility of wild-type and an hmp mutant that lacks the main mechanism of NO detoxification known in Salmonella and E . coli [34] , [35] . NO induced remarkable levels of protection against imipenem in hmp-deficient Salmonella ( figure 4C ) . It should be noted that about 90% of the hmp mutants were still killed by imipenem , raising the possibility that , in addition to antinitrosative defenses , the metabolic pliability of enteric bacteria could prevent the protective effects associated with NO . We reasoned that the incomplete protection afforded by NO to Salmonella and E . coli might be related to the fact that these facultative anaerobes can energize the PMF using alternative electron acceptors . If this were the case , then we would predict that 1 ) mutations that favor metabolism through the TCA cycle could allow for a more complete NO-induced tolerance to imipenem , 2 ) classical PMF inhibitors may induce imipenem tolerance under conditions that NO fails to do so , 3 ) growth of Salmonella and E . coli with glucose as the sole carbon source might estimulate NO-induced antibiotic tolerance , and 4 ) NO may have different effects on the PMF according to the carbon source used for growth . The following experiments were performed to test these predictions . 1 ) We tested ackA and pta mutants unable to ferment pyruvate to acetate , thus forcing Salmonella to more fully utilize the TCA cycle and oxidative phosphorylation . Remarkably , NO completely protected ackA and pta mutants against the antimicrobial activity of imipenem ( figure 4D ) . 2 ) In contrast to NO , the addition of 50 µM CCCP protected Salmonella and E . coli grown in LBG broth against 12 . 5 µg/ml imipenem ( figure 4B ) , demonstrating that inhibition of PMF induces tolerance to imipenem under conditions that NO is unable to do so . 3 ) E . coli and Salmonella grown in E salts medium supplemented with 0 . 4% glucose ( i . e . , EG medium ) were efficiently killed by imipenem ( figure 4E ) . Moreover , the addition of 750 µM spermine NONOate or 500 µM KCN similarly protected most Salmonella and E . coli grown in EG medium against the bactericidal activity of 12 . 5 µg/ml imipenem . Salmonella exposed to suboptimal concentrations of spermine NONOate and CCCP ( i . e . , 250 and 10 µM , respectively ) became fully tolerant to imipenem ( figure 4F ) . Lastly , 4 ) we measured the PMF in the BSL2 pathogen S . Typhimurium with 3 , 3′-dipropylthiadicarbocyanine iodide [DiSC3 ( 5 ) ] , the fluorescence of which is inversely proportional to the PMF [36] . As expected , valinomycin increased DiSC3 ( 5 ) -mediated fluorescence ( figure S2 ) . DiSC3 ( 5 ) -mediated fluorescence increased ( p<0 . 001 ) in both Salmonella grown to OD600 of 0 . 5 in LBG broth or EG medium after exposure to 750 µM spermine NONOate ( figure 4G ) , suggesting that NO inhibits the PMF under both conditions . It should be noted , however , that DiSC3 ( 5 ) fluorescence was significantly ( p<0 . 001 ) lower in NO-treated Salmonella grown in LBG broth than NO-treated controls grown in EG medium , suggesting that NO is less efficient at inhibiting the PMF in cells grown in LBG broth . Our investigations indicate that the protection afforded by NO against imipenem is dependent on the degree of PMF inhibition . Together , these findings suggest that both metabolic activity and antinitrosative defenses modulate NO-mediated tolerance to imipenem . Our investigations demonstrate that NO induces tolerance to β-lactams by collapsing the PMF . It remains possible that the protection afforded by NO against β-lactam antibiotics could emanate from its ability to promote antioxidant defenses [18] . Following this line of reasoning , the limited antimicrobial activity of imipenem against anaerobic B . pseudomallei could be interpreted as a sign that oxidative stress is required for killing . Consequently , we tested whether imipenem induces oxidative stress in B . pseudomallei , and whether NO affects this response . Membrane soluble H2O2 , which arises by the spontaneous or enzymatic dismutation of O2− , was used as readout of overall production of reactive oxygen species . B . pseudomallei treated with sublethal concentrations of imipenem produced consistently lower concentrations of H2O2 than untreated controls ( figure 5A ) . The addition of 100 µM spermine NONOate to the bacterial cultures increased the amount of H2O2 generated by 3-fold . Similar to NO-treated cells , KCN-treated B . pseudomallei generated about 15 µM H2O2 . The addition of a sublethal concentration of imipenem significantly ( p<0 . 05 ) reduced the amount of H2O2 generated by B . pseudomallei in response to NO or KCN . These findings are consistent with those reported by Liu and Imlay , who conjectured that the effects of β-lactams on respiration could reflect damage of the cell envelope and dissipation of the back pressure of the proton motive force [22] . Our investigations indicate that NO-mediated antibiotic tolerance cannot be explained by diminished oxidative stress; nor does the imipenem-mediated killing of B . pseudomallei appear to be dependent on the elicitation of oxidative stress . Collectively , these data support recent investigations that have questioned oxidative stress as the mode of action of bactericidal antibiotics [22] , [23] . Our data also suggest that NO-induced antibiotic resistance takes place independently of its effects on antioxidant defenses . According to our proposed model that the NO-dependent collapse of the PMF induces tolerance to β-lactams , the poor antibiotic activity of imipenem against anaerobic B . pseudomallei could reflect a lack of respiratory activity . To shed light into this possibility , the antimicrobial activity of imipenem was tested in anaerobic B . pseudomallei grown in LBG broth supplemented with 50 mM of the terminal electron acceptor NO3− . Anaerobic cells respiring NO3− became susceptible to 50 µg/ml imipenem ( figure 5B ) . We tried to determine the effects of NO treatment on the killing of anaerobic Burkholderia by imipenem but , as previously noted [16] , anaerobic bacteria were found to be extraordinarily susceptible to 100 µM spermine NONOate . Therefore , KCN was used instead . The addition of KCN abrogated the antibiotic activity of imipenem against anaerobic B . pseudomallei cultured in LBG broth supplemented with 50 mM NO3− ( figure 5B ) . The concentrations of KCN that elicited protection also inhibited nitrate reductase activity ( figure 5C ) . These findings indicate that imipenem has antibiotic activity in the absence of O2 and derived reactive oxygen species if the electron transport chain is energized by the reduction of alternative electron acceptors such as NO3− . Our investigations indicate that chemically-generated NO protects B . pseudomallei , S . enterica and E . coli against the antimicrobial activity of β-lactams . Next , we studied whether NO generated through the enzymatic activity of NO synthases modulates the antimicrobial activity of imipenem . J744 macrophage-like cells were stimulated overnight with 200 U/ml recombinant murine IFNγ . The macrophages were infected with B . pseudomallei at an MOI of 4 . The data shown in figure 6A indicate that NO produced by IFNγ-treated macrophages enhances the antimicrobial activity of 2 . 5 µg/ml imipenem . Our investigations identify NO as the mechanism by which recombinant IFNγ enhances the antimicrobial activity of imipenem in vivo [37] . These findings , however , contrast with the protective effects observed for NO in vitro . Because B . pseudomallei are hypersusceptible to NO [30] , we tested the survival of B . pseudomallei after exposure to a bactericidal concentration of NO and a sublethal amount of imipenem . About 90% of B . pseudomallei in the cultures were killed after exposure to 500 µM spermine NONOate , whereas controls treated with 0 . 25 µM imipenem doubled in cell number during the 2 . 5 h of the experiment ( figure 6C ) . However , the simultaneous addition of 500 µM spermine NONOate and 0 . 25 µM imipenem killed ∼99 . 9% of B . pseudomallei in the cultures . Together , these investigations suggest that β-lactams can potentiate the antimicrobial activity of host defenses such as NO . We next tested whether NO produced by IFNγ-treated macrophages can modify the killing of intracellular Salmonella by imipenem . In contrast to B . pseudomallei ( figure 6A ) , NO produced by IFNγ-treated macrophages appears to protect Salmonella against this β-lactam ( figure 6D and E ) , since imipenem killed Salmonella in a concentration-dependent manner in macrophages treated with the iNOS inhibitor aminoguanidine .
Our investigations indicate that the antimicrobial activity of β-lactams can be modified by NO produced chemically or enzymatically in the inflammatory response of IFNγ-activated macrophages . At sublethal concentrations , NO protects B . pseudomallei , nontyphoidal Salmonella and E . coli from the antimicrobial activity of β-lactams , and NO generated by IFNγ-activated macrophages shields intracellular Salmonella from the cytotoxicity of imipenem . At higher concentrations , however , the bactericidal activity of NO itself against B . pseudomallei was potentiated by sublethal concentrations of imipenem . NO has been proposed to protect bacteria against different classes of antibiotics by promoting antioxidant defenses [18] . However , our investigations suggest that the mechanism by which NO induces resistance of B . pseudomallei and other Gram-negative bacteria to β-lactam antibiotics is mediated through the collapse of the PMF . The following independent lines of evidence support the proposed model . First , the concentrations of NO that protect B . pseudomallei against β-lactams also inhibit respiratory activity . Second , high throughput sequencing of a transposon library revealed that B . pseudomallei mutants in cytochrome oxidase are hyperresistant to β-lactam killing . Third , cyanide , a classical inhibitor of cytochromes , inhibits respiration and protects B . pseudomallei , E . coli and S . enterica against β-lactams . Fourth , the degree of the PMF appears to be inversely associated with the extent of β-lactam-mediated killing . Fifth , dissipation of ΔH+ and ΔΨ components of the PMF independently protect against β-lactams . And sixth , killing by imipenem is marginal in anaerobic B . pseudomallei cultures , unless the electron transport chain is energized with terminal electron acceptors such as NO3− . Collapse of the PMF can be added to β-lactamases , mutated penicillin-binding proteins and efflux pumps as strategies that protect bacteria against β-lactam antibiotics . The immediate antibiotic tolerance elicited in response to NO may provide a window of time required for the acquisition of mutations that mediate inheritable resistance to antibiotics . NO , KCN , CCCP and valinomycin blocked the imipenem-mediated killing of B . pseudomallei , E . coli and S . enterica . These findings indicate that the antimicrobial activity of β-lactams requires an energized membrane . Considering the high affinity of NO for terminal cytochromes of the electron transport chain and the advantage afforded by transposons in cytochrome c or cytochrome c oxidase for the survival of B . pseudomallei in the presence of imipenem , we propose that terminal cytochromes are the likely molecular switch by which NO induces tolerance to β-lactams . This model is independently supported by the fact that the concentrations of NO that elicited antibiotic tolerance in B . pseudomallei also inhibited O2 consumption . Repression of respiratory activity has already been shown to mediate NO-induced resistance of Salmonella to aminoglycosides [19] . Our investigations with Salmonella and E . coli indicate that the ability of NO to inhibit respiration is necessary but not sufficient for β-lactam drug tolerance . Ultimately , the ability of NO to induce antibiotic tolerance seems to be associated with the degree of inhibition of the PMF . For example , NO-dependent tolerance to imipenem in E . coli and Salmonella grown in EG medium or LBG broth is inversely proportional to the PMF of the bacteria . Interestingly , NO prevented most of the antibiotic activity of imipenem against the strict aerobe B . pseudomallei under growth conditions that failed to protect Salmonella or E . coli . This is likely explained by the fact that the PMF in B . pseudomallei is preferentially maintained by the enzymatic activity of cytochrome oxidases . The heavy dependence of B . pseudomallei on cytochrome oxidases to maintain the PMF may explain why β-lactam antibiotics are most efficient during the acute phase of therapy at a time when , in the absence of abscesses and NO-mediated immunity , cytochrome oxidases are expected to be fully functional . The NO blockage of energy-dependent drug uptake has been shown to protect Salmonella against aminoglycosides [19] . In an analogous fashion , NO could protect bacteria by interfering with the expression or function of the Omp38 porin that is required to transport β-lactams into the periplasmic space [38] . Blockage of drug uptake , however , may not explain why NO protects bacteria against β-lactam antibiotics , because B . pseudomallei treated simultaneously with NO and imipenem yielded even smaller colonies than NO-treated controls . Moreover , NO protected B . pseudomallei already exposed to imipenem . The NO-induced tolerance to β-lactams could be mediated by the negative impact that the collapse of the PMF has on metabolism and membrane function . At least three mechanisms could explain how the collapse of the PMF by NO lessens the antibiotic activity of β-lactam drugs . First , electrochemical gradients energize the transport of muropeptides across the cytoplasmic membrane [39]; thus , the negative impact of NO on the PMF may inhibit peptidoglycan biosynthesis . Second , rapid β-lactam-induced lysis requires successful assembly of the divisome [40] , [41] , an event that is initiated by the PMF-dependent localization of FtsA to the FtsZ septal ring [42] . FtsA serves as a scaffold for the assembly of several morphogenetic proteins , including penicillin-binding proteins [43] that are the targets of β-lactam antibiotics . Consequently , the NO-dependent inhibition of the PMF could delocalize morphogenetic proteins from the division septum , thereby contributing to resistance to β-lactams . Disassembly of the division ring could be a critical step by which NO protects rapidly growing bacteria from β-lactams , but may be of lesser importance in non-dividing stationary phase bacteria . And third , NO and the other chemical inhibitors of the electron transport chain could protect bacteria by stalling growth . According to this idea , imipenem exerted negligible antimicrobial activity against stationary phase Salmonella . However , inhibition of cell growth might not explain why sublethal concentrations of NO protect stationary phase B . pseudomallei against imipenem . We found it remarkable that the 1 , 000-fold reduction in viability of nongrowing B . pseudomallei was abrogated in the presence of sublethal concentrations of NO . NO did not select for intrinsically resistant populations , because the surviving bacteria remained susceptible to imipenem . Constant NO fluxes were required for the elicitation of tolerance to imipenem , and the effects were short-lived . The inhibition of metal centers in terminal oxidases of the electron transport chain , with the consequent reduction in PMF , provides a reasonable model for the fast and transient adaptation of B . pseudomallei , E . coli and Salmonella to β-lactam antibiotics . The transient protection afforded by NO can be better understood if we consider that the koff value for the dissociation of NO from cytochrome aa3 is 0 . 01 sec−1 [44] . Thus , the association of cytochrome c oxidase to NO would last about 1 min ( i . e . , t1/2 of 69 sec ) . In other words , the fast denitrosylation of metal prosthetic groups in the terminal cytochromes of the electron transport chain could explain the transient protection noted after NO is removed from the bacterial cultures . Oxidative stress has been proposed as a common killing mechanism of bactericidal antibiotics [21] , and the antioxidant defenses elicited by NO are thought to mediate resistance of B . subtilis to several classes of antibacterials [18] . Our investigations indicate that NO prevents antibiotic killing despite increasing the rate of H2O2 synthesis . The stasis of electrons that follows the nitrosylation of terminal cytochrome oxidases of the electron transport chain can facilitate the adventitious reduction of O2 by flavin cofactors or Q sites of NADH dehydrogenases [45] , [46] . The O2− generated in this process spontaneously or enzymatically dismutates to H2O2 . This model may explain why imipenem , similar to other β-lactam antibiotics that increase bacterial respiratory rates [22] , diminishes H2O2 synthesis . The increased H2O2 synthesis noted in NO-treated bacteria challenges the notion that NO enhances antibiotic resistance by eliciting antioxidant defenses [18] . Our investigations are consistent with recent work that has reported that bactericidal antibiotics can kill microorganisms in the absence of oxidative stress [22] , [23] . This idea is further substantiated by the fact that imipenem can kill B . pseudomallei in anaerobic cultures given that the bacteria are actively respiring the terminal electron acceptor NO3− . Although sublethal concentrations of NO reversed β-lactam-mediated killing of B . pseudomallei , NO produced by IFNγ-primed macrophages synergized with imipenem in killing intracellular B . pseudomallei . Our investigations identify NO as the mechanism by which recombinant IFNγ enhances the antimicrobial activity of imipenem in vivo [47] . We find it remarkable that inos , which is just one of over 150 loci regulated by IFNγ [48] , made such a difference in the outcome of imipenem treatment . Cumulatively , these investigations support the widely accepted concept that immunocompetent hosts respond better to antibiotic therapy than immunodeficient controls . A closer look at our investigations indicate , however , that imipenem potentiates the killing activity of NO in vivo and not vice versa . In fact , sublethal concentrations of imipenem augmented the bactericidal activity of NO in an exponential fashion . Weakening of the cell wall by β-lactams increases NO-mediated killing . This observation could be explained by a model in which the outstanding killing that NO exerts against B . pseudomallei is a direct consequence of membrane dysfunction . Because B . pseudomallei draw most energy from oxidative phosphorylation , nitrosylation of terminal cytochromes of the electron transport chain could have greater deleterious actions on the energetics and membrane function of the aerobe B . pseudomallei as compared to more metabolically pliable organisms such as Salmonella that can draw significant energy from fermentation . Degree or duration of the inhibition of the respiratory chain by NO could then rationalize why sublethal concentrations of NO protect against β-lactams , whereas higher NO fluxes become more lethal in the presence of imipenem . The clinical relevance of these findings can already be inferred from the observation that mice treated with recombinant IFNγ and ceftazidime clear acute B . pseudomallei infections [37] . It remains puzzling that during the natural course of meliodosis ceftazidime is largely ineffective during the eradication phase of treatment . Reduced respiratory activity imposed by growing abscesses or the sublethal amounts of NO could contribute to the more limited use of β-lactams in the eradication phase of therapy . | β-lactam drugs that inhibit peptidoglycan biosynthesis are often used in the treatment of bacterial infections , including melioidosis . Independent of their antibiotic activity , we have noted that submicromolar concentrations of β-lactams potentiate the killing of intracellular B . pseudomallei supported by NO generated by IFNγ-primed macrophages . The production of NO can nonetheless be a double-edged sword , as indicated by our observations that sublethal concentrations of nitric oxide ( NO ) , a diatomic radical produced by phylogenetically diverse organisms to regulate neurotransmission , vascular tone and host defense , tolerize B . pseudomallei , nontyphoidal Salmonella and E . coli against the antimicrobial activity of β-lactams . Accordingly , NO produced in the inflammatory response of macrophages protects nontyphoidal Salmonella against β-lactam antibiotics . NO mediates bacterial tolerance to β-lactam antibiotics by inhibiting the electrochemical gradient supported by terminal cytochrome oxidases of the respiratory chain , rather than by decreasing oxidative stress as previously thought . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"biology",
"and",
"life",
"sciences",
"medical",
"microbiology",
"microbiology"
] | 2014 | Nitric Oxide from IFNγ-Primed Macrophages Modulates the Antimicrobial Activity of β-Lactams against the Intracellular Pathogens Burkholderia pseudomallei and Nontyphoidal Salmonella |
Embryonic development is tightly regulated by transcription factors and chromatin-associated proteins . H3K4me3 is associated with active transcription and H3K27me3 with gene repression , while the combination of both keeps genes required for development in a plastic state . Here we show that deletion of the H3K4me2/3 histone demethylase Jarid1b ( Kdm5b/Plu1 ) results in major neonatal lethality due to respiratory failure . Jarid1b knockout embryos have several neural defects including disorganized cranial nerves , defects in eye development , and increased incidences of exencephaly . Moreover , in line with an overlap of Jarid1b and Polycomb target genes , Jarid1b knockout embryos display homeotic skeletal transformations typical for Polycomb mutants , supporting a functional interplay between Polycomb proteins and Jarid1b . To understand how Jarid1b regulates mouse development , we performed a genome-wide analysis of histone modifications , which demonstrated that normally inactive genes encoding developmental regulators acquire aberrant H3K4me3 during early embryogenesis in Jarid1b knockout embryos . H3K4me3 accumulates as embryonic development proceeds , leading to increased expression of neural master regulators like Pax6 and Otx2 in Jarid1b knockout brains . Taken together , these results suggest that Jarid1b regulates mouse development by protecting developmental genes from inappropriate acquisition of active histone modifications .
Embryonic development is characterized by a coordinated program of proliferation and differentiation that is tightly regulated by transcription factors and chromatin-associated proteins . As embryonic cells differentiate , certain genes are activated while others are repressed , resulting in a unique pattern of gene expression in each cell type . Histone H3 lysine 4 tri-methylation ( H3K4me3 ) localizes to transcription start sites with high levels present at actively transcribed genes [1] , [2] , even though H3K4me3 at promoters is not a definite indication for transcriptional activity [3] . Methylation of H3K4 is catalyzed by a family of 10 histone methyltransferases in mammals [4] . Five of these are members of the Trithorax group of proteins that were first described in Drosophila to be required for maintenance of Hox gene expression by counteracting Polycomb-mediated repression . In Mll1 and Mll2 mutant mice , target genes are properly activated but expression fails to be maintained leading to embryonic lethality [5] , [6] . In addition , H3K4 histone methyltransferases function in hematopoiesis [7] , [8] and neurogenesis [9] . H3K4me3 is found in a constant balance with Polycomb-mediated repressive H3K27me3 . Presence of both H3K4me3 and H3K27me3 at promoters is referred to as bivalency [10] . The category of bivalent genes is enriched in developmental regulators and is particularly abundant in embryonic stem cells ( ESCs ) that have the potential for several lineage choices [11] . Moreover , Polycomb proteins repress non-lineage specific gene expression , thereby ensuring developmental potency of embryonic and tissue stem cells during lineage specification , differentiation and development ( reviewed in [12] ) . Polycomb proteins are classified into two separate complexes referred to as Polycomb repressive complex 2 ( PRC2 ) , which mediates H3K27me3 , and PRC1 , which catalyzes mono-ubiquitylation of H2A ( H2AK119ub1 ) [13] , [14] . Classical models propose a sequential mechanism in which H3K27me3 creates a binding site for PRC1 leading to further repression [14] , [15] , even though emerging studies suggest that Polycomb function is more complex [16]–[18] . While histone methylation was initially viewed as a stable modification , the discovery of histone demethylating enzymes has changed this paradigm [19] . Demethylation of H3K4me3 is catalyzed by the JARID1 ( KDM5 ) family , which in mammals has four members: JARID1A , JARID1B , JARID1C and JARID1D [20] . The Drosophila JARID1 homologue LID ( Little imaginal discs ) is required for normal development [21] , and the C . elegans homologue RBR-2 ( retinoblastoma binding protein related 2 ) regulates vulva formation and lifespan [22] , [23] . Mice mutant for Jarid1a are viable , displaying only mild phenotypes in hematopoiesis and behavior [24] . A recent report suggests that Jarid1b mutant mice are embryonic lethal between E4 . 5 and E7 . 5 [25] . The molecular mechanisms underlying this phenotype were not addressed . In contrast , others obtained viable Jarid1b mutant mice [26] . However , the requirement of Jarid1b for the differentiation of ESCs along the neural lineage [27] , [28] suggests that Jarid1b may function in mouse development . In humans , JARID1B is highly expressed in several types of cancer , and it was shown to regulate proliferation of breast cancer cells and a slow cycling population of melanoma cells that promotes prolonged tumor growth ( reviewed in [20] ) . While the role of Jarid1b in mice remains controversial [25] , [26] , an understanding of its in vivo function is essential to direct future studies evaluating JARID1B as a potential drug target in cancer therapy . Jarid1b expression has been reported in various tissues during mouse embryogenesis whereas its expression becomes restricted in adults [29] . Here we report the first detailed analysis of the contribution of Jarid1b to mouse development . We show that Jarid1b is required for the proper development of several neural systems in the mouse and address the mechanisms underlying the observed defects .
To characterize the function of Jarid1b during mouse development , we generated constitutive Jarid1b knockout mice . Conditionally targeted Jarid1b mice containing a lacZ-Neo-reporter cassette flanked by FRT sites and in which Jarid1b exon 6 is flanked by loxP sites [28] were crossed with mice constitutively expressing Flp and Cre recombinase to obtain Jarid1b+/− mice . Jarid1b+/− mice were further intercrossed to generate Jarid1b−/− mice . Instead of the expected 25 percent of knockout mice , we only obtained 9 . 3 percent of adult Jarid1b knockouts ( Figure 1A ) , suggesting that Jarid1b−/− mice are sub-viable . Analysis of early and late embryos from Jarid1b+/− intercrosses showed expected ratios while an increased number of Jarid1b knockouts was present among pups found dead during the first day after birth ( Figure 1A ) , indicating that this might be the critical time for survival . We have previously shown that conditional deletion of Jarid1b using this construct in vitro results in complete loss of Jarid1b protein and no generation of truncated or alternatively spliced variants [28] . Loss of Jarid1b in vivo was confirmed in all Jarid1b−/− embryos tested ( see examples in Figure S1A and S1B ) , indicating that partial survival of Jarid1b knockouts in not due to incomplete deletion . Moreover , expression of other Jarid1 family members is unchanged both in vitro [28] and in vivo ( Figure S1C ) . To determine more precisely when Jarid1b−/− pups die , we performed caesarean deliveries and closely monitored the pups ( Figure 1B ) . While approximately 95 percent of wild-type pups survive , we found that 50 percent of the knockouts die within the first two hours after delivery and another approximately 20 percent die after 14 to 24 hours ( Figure 1C ) . All pups that survive the first day , develop normally until adulthood . Interestingly , survival of Jarid1b+/− pups is also slightly , even though not significantly , reduced during the first day . While most of the Jarid1b knockouts are grossly normal and not generally growth retarded ( Figure 1D ) , we observed an increased incidence of developmental defects like exencephaly and eye defects among Jarid1b knockouts ( Figure 1B , 1E and 1F ) . Taken together , loss of Jarid1b leads to major neonatal lethality of which only a small fraction can be explained by severe morphological abnormalities . There is a large spectrum of physiological systems whose defects can challenge neonatal survival including those affecting parturition , breathing , suckling and neonatal homeostasis [30] . The first extrauterine challenge for neonates is breathing and since the majority of Jarid1b−/− pups die immediately after birth , we studied the respiratory system in more detail . Analysis of lungs from E18 . 5 fetuses revealed a normal size and weight ( 3 . 34±0 . 26 versus 3 . 45±0 . 44 percent body weight in heterozygotes versus knockouts , respectively ) as well as a normal lobulation pattern ( data not shown ) . Next , we isolated lungs from Jarid1b−/− newborns that had died within 2 hours after delivery and had either not shown any sign of breathing or exhibited gasping respiration ( Figure 2A ) . While the wild-type lung showed saccular inflation , knockout lungs were compact and poorly inflated visible both from gross appearance and histology ( Figure 2B and 2C ) , suggesting that Jarid1b−/− neonates die due to an inability to establish normal breathing . Moreover , preterm ( E18 . 5 ) Jarid1b−/− lungs were abnormally compact compared to controls ( Figure 2D ) , which might indicate a failure of prenatal breathing activity [31] . Respiratory failure might be caused by delayed lung maturation characterized by reduced surfactant expression [32] . Therefore , we analyzed expression of surfactant proteins ( Sftpa1 , Sftpb , Sftpc and Sftpd ) in Jarid1b knockout mice ( Figure S2A ) . None of the four surfactants was reduced in Jarid1b knockout lungs at E18 . 5 , suggesting that respiratory failure is not due to pulmonary immaturity . In agreement with this , intrauterine administration of dexamethasone , a glucocorticoid that induces fetal lung maturation [33] , did not improve survival of Jarid1b knockout pups ( Figure S2B ) . We also examined other physiological systems that are required for neonatal survival including the rib cage , diaphragm , craniofacial appearance and the palate as well as the cardiovascular system [30] , but did not detect any abnormalities in the Jarid1b knockouts ( Figure S3A–S3E ) . We conclude that while the lungs , skeletal and cardiovascular systems are properly developed , Jarid1b−/− neonates are unable to reliably establish respiratory function . Immediate breathing after birth is also dependent on brainstem rhythmogenic and pattern forming neural circuits that develop before birth [34] . We therefore isolated brains from neonates after caesarean delivery , but found no gross abnormalities or differences in size of Jarid1b−/− brains compared to controls ( Figure S3F and S3G ) . Essential rhythmogenic networks regulating breathing are located in the brainstem . Therefore , we recorded spontaneous C3–C5 nerve activity in an in vitro brainstem-spinal cord preparation from E18 . 5 embryos . Surprisingly , given the respiratory defects in newborn Jarid1b knockouts , central respiratory rhythmogenesis was unperturbed in Jarid1b−/− embryos ( Figure S3H ) . To monitor neurological reflexes of newborn Jarid1b−/− pups , we tested their response to pinching stimuli [35] . As opposed to control neonates , Jarid1b mutants only weakly reacted to a tail pinch ( Figure S3I ) , suggesting that Jarid1b newborns show motosensory deficits characterized by hyporesponsiveness . These results together with our previous in vitro data showing that Jarid1b is required for the differentiation of ESCs along the neural lineage [28] prompted us to analyze the development of neural systems in more detail in Jarid1b−/− embryos . As a first step we analyzed cranial nerves , a pair of 12 nerves that are essential for sensory and motor functions and reside in the mid- and hindbrain [36] . Defects in cranial nerve development may compromise neonatal survival . Cranial and spinal nerves can be visualized by whole-mount immunostaining at E10 . 5 using an anti-neurofilament antibody . Comparison of Jarid1b−/− embryos with controls revealed that while all nerve pairs are present , several cranial and spinal nerves are dysmorphic in the Jarid1b knockouts ( Figure 3A ) . We used an arbitrary scoring system to quantify the differences between genotypes and found that Jarid1b knockouts are significantly affected while slight defects are already detectable in heterozygotes compared to wild-type ( Figure 3B ) . Cranial nerves are involved in a diverse range of functions including movement of the eye , innervation of muscles of mastication , facial expression and tongue , and in transmitting information from chemoreceptors to the respiratory center [36] , 37 , and thus , defects in cranial nerve development may be relevant to reduced survival of Jarid1b knockouts . For example , the hypoglossal nerve ( XII ) , which is dysmorphic in Jarid1b knockouts , innervates the muscles of the tongue , crucial for upper airway aperture during breathing . Next , we analyzed Jarid1b expression during the time of mouse development when cranial nerves are specified . From embryonic day 8 , the hindbrain becomes transiently partitioned along the anterior-posterior ( AP ) axis in a series of 8 rhombomeres that influence the spatial distribution of neuronal types [34] . Using the lacZ-Neo-reporter cassette present in the targeting construct [28] , we observed high ubiquitous expression of Jarid1b in embryonic but not extraembryonic tissues at E8 . 5 ( Figure S4 ) . Moreover , in agreement with previous reports [38] , at E12 . 5 and E14 . 5 , Jarid1b expression was observed in several neural tissues including the fore- and hindbrain , neural retina , spinal cord and dorsal root ganglia as well as other tissues ( Figures S5 and S6 ) , indicating that Jarid1b could be involved in the development of several organs . Cranial nerve development is imparted by genes involved in AP patterning and rhombomere specification , neuronal determination or survival and axonal migration [37] . Compartmentalization of the hindbrain , and in particular rhombomeres 3 and 4 , have emerged as territories for the maintenance of breathing frequency after birth [34] . Rhombomeres are characterized by specific patterns of Hox gene , Krox20 ( Egr2 ) and Kreisler ( Mafb ) expression , leading us to analyze expression of these genes by RNA in situ hybridization in Jarid1b−/− embryos . However , we did not observe any defects in the hindbrain patterning of E8 . 75 embryos ( Figure 3C ) , suggesting that other mechanisms are responsible for spinal nerve abnormalities in Jarid1b−/− embryos . In addition to sporadic cases of exencephaly , we frequently observed defects in eye development in Jarid1b−/− embryos and pups ( Figure 4A–4D ) . In the most severe cases , eyes were completely absent ( anophthalmia; Figure 4C ) . Other embryos exhibited microphthalmia ( Figure 4B and 4D ) or an incomplete closure of the optic fissure ( Figure 4A and 4D ) . Moreover , after birth , the eyelid was often found open in Jarid1b−/− pups while it was closed in control mice at this time ( Figure 4C ) . Altogether , externally visible eye defects were observed in approximately 22 percent of Jarid1b−/− embryos and pups ( Figure 4E ) , but never in the Jarid1b knockouts that survive to adulthood . Histological analysis of two microphthalmic Jarid1b−/− eyes at E18 . 5 revealed a misfolding of the neural retina and a much smaller lense ( Figure 4F and 4G ) . To test whether Jarid1b is expressed in the developing eye , we performed β-galactosidase stainings on sections of E12 . 5 and E14 . 5 eyes from targeted Jarid1b embryos ( Figure 4H ) . At both stages , Jarid1b is specifically expressed in the inner layer of the neural retina , which contains retinal ganglion cells . Thus , Jarid1b seems required for the proper development of a mouse neurosensory organ , the eye . We have previously shown that Jarid1b binds to the transcription start sites of many developmental regulators in mouse ESCs , many of which are also bound by Polycomb group proteins [28] . Therefore , we speculated that Jarid1b might also regulate Polycomb target genes in vivo . Hox genes represent classical Polycomb targets and their misexpression in Polycomb mouse mutants results in transformations of the axial skeleton [39] , [40] . To investigate whether such transformations are also present in Jarid1b mutants , we stained skeletal preparations of E17 . 5 embryos to visualize cartilage and bone . While we did not observe any defects in the anterior region of the vertebral column ( occipito-cervico-thoracic region ) , we found a transformation of the 26th vertebra , which is supposed to be the last lumbar vertebra ( L6 ) into the first sacral vertebrae ( S1 ) ( Figure 5A and 5B ) . Moreover , we also observed a transformation of the 34th vertebra ( Figure 5B and Figure S7 ) . Thus , Jarid1b−/− embryos display posterior transformations of the skeleton , which similar to Polycomb mutants are not completely penetrant [39] , [40] . To identify genes in addition to the Hox genes that might be misregulated in Jarid1b−/− embryos , we focused on an early embryonic stage ( E8 . 5 ) where morphological defects were not yet observed . We expected that several of the phenotypes observed in the Jarid1b mutants arise from misspecification events early in development , as genes involved in eye specification , neural tube closure and hindbrain patterning start to be expressed from E8 . 0 [41] , [42] . First , we performed chromatin immunoprecipitation ( ChIP ) followed by sequencing ( seq ) of head regions of E8 . 5 embryos ( Figure S8A ) for H3K4me3 and H3K27me3 to identify genes that change their chromatin state and thus might become misregulated in Jarid1b−/− embryos . By this analysis , we identified 492 peaks with increased H3K4me3 levels in Jarid1b knockouts versus heterozygotes , whereas only 27 peaks were detected in the reverse comparison ( Figure S8B ) . Representative examples of loci with increased H3K4me3 in the knockouts as well as loci with unchanged chromatin states are shown in Figure 6A and 6B , respectively . The results were validated in an independent experiment by ChIP-qPCR showing that the differences in H3K4me3 are reproducible ( Figure 6C ) . Comparison of genes with increased H3K4me3 in knockout embryos with all genes revealed an enrichment of repressed ( H3K27me3 positive ) and bivalent ( H3K4me3/H3K27me3 positive ) genes among genes with increased H3K4me3 ( Figure 6D and Figure S8C ) , suggesting that aberrant active histone marks accumulate mainly at genes that are usually not actively transcribed . Gene ontology analysis of genes with increased H3K4me3 in the Jarid1b−/− embryos identified regulators of transcription and development including genes involved in ectoderm , nervous system and skeletal development as significantly overrepresented ( Figure 6E and Figure S8D ) . To identify genes that are directly bound and regulated by Jarid1b , we also attempted ChIP experiments for Jarid1b in E8 . 5 embryos but unfortunately the results were of low quality due to very limited amounts of starting material . Instead , we compared genes with elevated H3K4me3 in Jarid1b−/− embryos with genes bound by Jarid1b in ESCs [28] and found that approximately one quarter was bound by Jarid1b in ESCs ( Figure S8E ) . Thus , it is likely that some of the genes with increased H3K4me3 are also Jarid1b targets during early mouse development . Next , we performed gene expression analysis of mRNA isolated from E8 . 5 Jarid1b heterozygotes and knockouts . Except for Jarid1b , we did not identify any genes that were more than 2-fold changed in the knockouts ( Figure S8F ) . We validated a number of genes by RT-qPCR and confirmed that Jarid1b was not expressed in the knockouts , whereas Jarid1a and Jarid1c as well as L1cam and Pax2 remained unchanged ( Figure S8G ) . Taken together , while we detected increased levels of H3K4me3 at a number of developmental regulators early in embryogenesis , these chromatin changes do not translate into detectable global transcriptional changes at this stage of development . Deletion of Jarid1b in ESCs leads to a global increase in H3K4me3 , while global H3K4me3 levels remain unchanged in Jarid1b depleted neural stem cells isolated from E12 . 5 embryos [28] . Likewise , depletion of JARID1B in MCF7 cells [43] or depletion of Jarid1a in mouse embryonic fibroblasts [24] did not result in a global elevation of H3K4me3 . To analyze the effect of Jarid1b depletion in vivo , we prepared protein extracts from different stages of embryos . We confirmed lack of Jarid1b protein in all knockout embryos analyzed ( Figure 7A ) . While we detected little change in H3K4me3 by immunoblotting in heads of E12 . 5 ( data not shown ) and E14 . 5 Jarid1b−/− embryos , global H3K4me3 levels were strongly increased in heads of late ( E17 . 5 ) Jarid1b−/− embryos and in forebrains of Jarid1b−/− newborns ( Figure 7A ) . These results suggest that H3K4me3 accumulates in Jarid1b knockouts as embryonic development proceeds , while H3K4me3 levels remain fairly constant during normal fetal development ( Figure S9 ) . Next , we wanted to know at which classes of genes H3K4me3 accumulates in brains of newborn mice . Since the brain is a complex and heterogeneous organ , we first determined whether Jarid1b expression is limited to specific regions at this stage of brain development . However , β-galactosidase stainings on sections of brains from newborns revealed high overall expression of Jarid1b ( Figure 7B ) . In addition , RT-qPCR analysis showed similar expression of Jarid1b in fore- and hindbrain , which is reduced in heterozygotes and lost in knockouts ( Figure 7C ) . Thus , we divided the brain into forebrain and hindbrain for ChIP experiments and selected a number of genes that represent different chromatin states ( Figure 7D–7G and S10 ) . We observed increased H3K4me3 at repressed ( H3K27me3-positive ) genes , including Otx2 , Pax9 , HoxB5 and Hesx1 , and at active ( H3K4me3-positive ) genes ( Sema5b ) , but not at unmodified genes in P0 forebrains . Some bivalent genes , for example Pax6 , showed increased H3K4me3 and slightly reduced H3K27me3 , while others remained unchanged ( e . g . Neurod2 ) . Similar results were obtained in independent ChIP experiments using forebrain or hindbrain . These data suggest that H3K4me3 is increased at transcription start sites in late stages of brain development . To test which genes are directly bound by Jarid1b , we also performed ChIP for Jarid1b ( Figure 7D–7G and Figure S10 ) . We detected Jarid1b binding at transcription start sites of bivalent ( Pax6 ) and H3K4me3-positive ( Sema5b ) genes , which is in agreement with our previous findings in ESCs [28] . Moreover , we detected low levels of Jarid1b binding ( 2- to 4-fold above background ) at several of the H3K27me3-positive loci with increased H3K4me3 in the Jarid1b knockouts ( Otx2 , Pax9 , Hoxb5 ) , suggesting that elevated levels of H3K4me3 at many of these loci are due to a loss of direct association of Jarid1b . To determine whether changes in chromatin modifications are accompanied by differences in expression , we performed RT-qPCR in P0 brains of controls and Jarid1b knockouts ( Figure 7D–7F and Figure S10 ) . We detected increased levels of the transcription factor Otx2 in forebrains of Jarid1b−/− newborns . Furthermore , in line with a shifted balance of H3K4me3 versus H3K27me3 , expression of the neural master regulator Pax6 was increased in Jarid1b−/− P0 brains . In contrast , expression of actively transcribed genes , like Sema5b , was unchanged despite higher levels of H3K4me3 . We conclude that Jarid1b mutants accumulate higher levels of H3K4me3 and show increased expression of genes important for regulating embryonic development . Next , we analyzed at which stage between E8 . 5 and P0 changes in gene expression arise in Jarid1b knockouts . While we did not detect transcriptional changes at E8 . 5 , expression of Otx2 , Pax6 and Sema5b was increased in heads of E12 . 5 Jarid1b knockout embryos compared to controls ( Figure S11A ) . Since the transcription factor Pax6 controls the balance between neural stem cell ( NSC ) self-renewal and neurogenesis [44] , we tested whether deletion of Jarid1b affected this balance . Sorting of NSCs and neuronal progenitor cells ( NPs ) from E12 . 5 brains ( Figure S11B ) revealed a slight ( but not significant ) increase in NSCs in Jarid1b knockouts and no change in NPs . Similarly , global levels of neuron and astrocyte markers remained unchanged in P0 brains ( Figure S9B ) , which is in agreement with normal gross morphology of Jarid1b knockout brains ( Figure S3F ) . Thus , the detectable changes in gene expression observed in Jarid1b knockout mice does not appear to be a result of abnormal numbers of NSCs or NPs . Finally , we tested whether increased expression of Otx2 and Pax6 correlated with survival . However , as shown in Figure S11C , we did not detect a significant difference in expression of Otx2 and Pax6 in brains of newborns that were alive 2 hours after caesarean delivery versus newborns that died immediately . In contrast , the expression of Otx2 was significantly higher in the adult brain of surviving knockout animals as compared to wild type ( Figure S11D ) , suggesting that transcriptional regulation by Jarid1b is not restricted to embryogenesis only , but affects selected genes rather than global transcription .
Embryonic development is regulated by transcription factors as well as chromatin-mediated processes resulting in tissue-specific gene expression . Here , we show that the histone demethylase Jarid1b is required for faithful mouse embryonic development ( see model in Figure S12 ) . Deletion of Jarid1b results in major neonatal lethality caused by an inability of the newborn mice to establish breathing . Jarid1b mutant embryos display a number of defects related to neural systems , including the misorganization of cranial and spinal nerves as well as increased incidence of exencephaly that might contribute to neonatal lethality . Respiratory rhythmogenic circuits in the brainstem of Jarid1b mutant embryos appear intact since a spontaneous motor output on cervical nerves was observed under in vitro conditions . In agreement , Krox20 and Kreisler , essential genes involved in specification of respiratory-related rhombomeres , are also not affected in mutant embryos . Thus , we speculate that the breathing problems of Jarid1b mutant neonates may stem from either compromised pattern forming circuits controlling airway patency , or an inability of the rhythmic motor output to reach respiratory muscles , caused by defects in cranial and spinal nerve development . Several other organs important after birth appeared undisturbed . However , the spectrum of physiological systems required for neonatal survival is large [30] and we cannot exclude that there are other subtle defects that manifest in secondary physiological problems interfering with survival of Jarid1b knockouts . Previous in vitro studies of ESCs with either reduced [28] or increased [27] levels of Jarid1b have reported a role for Jarid1b during differentiation of ESCs into neurons . This raises the question of why Jarid1b is specifically required in neural systems . During embryogenesis , Jarid1b is expressed in several neural organs including the brain , spinal cord and eye , but also in a number of other systems ( this study , [38] ) . Moreover , in ESCs , Jarid1b is targeted to transcription start sites of genes that regulate development , including genes involved in neurogenesis and ectoderm development [28] . Thus , a combination of tissue-specific expression and target gene selectivity might explain neural-specific phenotypes . It should be noted , however , that other systems are also affected by Jarid1b depletion , exemplified by homeotic transformation of the skeleton ( this study ) or slightly reduced expression of meso- and endodermal markers during embryoid body differentiation of ESCs [28] . Interestingly , knockdown of Jarid1b in the retina of newborn mice leads to abnormal morphology of rod photoreceptor cells and misregulation of rod-expressed genes [45] , supporting a role for Jarid1b in neuronal cells of the eye . In addition , other Jarid1 family members have reported functions in behavior and/or neurulation , suggesting that these processes are susceptible to changes in H3K4 methylation . Jarid1a knockout mice display abnormal clasping of the hindlimbs [24] , while mutations of human JARID1C occur in patients with X-linked mental retardation [46] . Knockout of Jarid1c in the mouse results in embryonic lethality due to defects in neurulation and cardiogenesis [47] . Taken together , while several Jarid1 family members are involved in the control of neural systems , they may regulate different aspects of development and cannot fully compensate for the absence of other Jarid1 members resulting in gene-specific phenotypes . To determine how Jarid1b contributes to the regulation of mouse development , we analyzed global as well as gene-specific histone methylation levels in Jarid1b−/− embryos . Consistent with the previously reported catalytic activity of Jarid1b [23] , [43] , H3K4me3 was increased around transcription start sites of developmental regulators already early during mouse development , particularly at genes that are normally in a repressed or poised state . At this early stage of development , we could not detect any global changes in gene expression levels , but we cannot exclude that expression of some genes might be affected in a subset of cells as the E8 . 5 mouse embryo is composed of many distinct cell layers . For example , the development of the eye initiates at E8 . 0 with the evagination of the optic pit from a subset of cells in the diencephalon [41] . In analogy , increased H3K4me3 at transcription start sites may reflect small increases in many cells of the early embryo or result from large increases in a subset of cells . Increased H3K4me3 around transcription start sites may render the associated genes more susceptible to later activation , especially during developmental time windows or in specific cell types where additional signaling molecules create a competent transcriptional state . As embryonic development proceeds , a global increase in H3K4me3 becomes detectable . Locally , similar to early embryos , increased H3K4me3 is present at the transcription start sites of genes in Jarid1b−/− brains . Even though altered H3K4me3 per se may not be sufficient to induce transcriptional changes or cell fate conversions [48] , it may have functional consequences when prompt responses to signaling events are required or for the fine control of steady state transcript levels [49] . Indeed , we detected increased expression of Pax6 and Otx2 , two master regulators of eye and neural development ( reviewed in [50] , [51] ) , in forebrains of Jarid1b−/− pups . For both Pax6 and Otx2 , it was shown that not only deletion but also overexpression affect eye and neural lineage development [44] , [52] , [53] . Binding of Jarid1b itself was found at H3K4me3-positive genes ( both active and bivalent ) , which is in agreement with previous ChIP-seq data [28] , [54] . Interestingly , the overlap between Jarid1b and Polycomb target genes was functionally supported in this study by the observation that Jarid1b knockout embryos show homeotic transformations of the skeleton , which is a hallmark of Polycomb mutant mice . Moreover , the observation that Jarid1b is bound to several repressed genes that are marked by aberrant H3K4me3 in the knockouts , suggests that Jarid1b is directly required to prevent aberrant accumulation of active chromatin modifications at developmental regulators during embryogenesis . Most of the phenotypes that we observed in Jarid1b knockouts occurred with incomplete penetrance . This is not uncommon and has been reported for other histone demethylases [55] but also transcription factors [56] . The penetrance is often affected by the genetic background of the mice [50] , [55] . This might also explain the difference in survival of our knockout mice compared to a previous study [25] . To test this hypothesis , we crossed our Jarid1b mutant mice that were derived on a C57BL/6 background into a mixed C57BL/6/129 genetic background ( Figure S13 ) . On the mixed background , 40% of knockouts die after birth , compared to 70% on a C57BL/6 background . Besides , we observed a similar frequency of exencephaly and a reduced response of the newborns to pinching stimuli , while no eye phenotypes were detected on the mixed background . These results suggest that different genetic background of mice strains used , could partly explain the divergence of obtained results . Moreover , many disease-causing mutations only have detrimental defects in a subset of individuals , and phenotypic discordance remains even in the absence of genetic and environmental variation . It was shown that feedback induction of genes with related functions differs across individuals leading to a buffering of stochastic developmental failure through redundancy [57] . In the Jarid1b mutants , we did not observe upregulation of other Jarid1 members at the transcript level . However , we cannot exclude that protein levels are increased , or that these proteins are preferentially recruited to Jarid1b target sites to compensate for lack of Jarid1b . In addition , systematic analysis of transcript levels and their correlation with phenotypes has shown that variability in gene expression underlies incomplete penetrance [58] . It was proposed that fluctuations in gene expression can be controlled by wild-type developmental networks . In contrast to transcription factors that can induce cell fate conversions , chromatin modification are thought to rather fine tune transcription . In Jarid1b mutant embryos , both repressed and bivalent genes acquire increased levels of H3K4me3 , which might render these genes more susceptible to unscheduled activation . Indeed , we observed increased expression of Pax6 and Otx2 in newborn knockout brains . Raj et al . [58] propose a model in which expression must surpass a threshold during a window of development . The same might be true for histone modifications . In plants , progressive increase in H3K27me3 was shown to be capable of switching a bistable epigenetic state of an individual locus [59] . While global levels of H3K4me3 are unchanged during early embryogenesis , H3K4me3 accumulates in the Jarid1b knockouts as embryonic development proceeds . For some genes or in specific cell types , this might lead to a switch in the balance of active versus repressive histone modifications , and if this coincides with a developmental window of transcriptional potency , it might affect phenotypic outcomes . Moreover , since Jarid1b binds to a large number of target genes [28] , it could be expected that a wide range of phenotypes with varying severity is observed in Jarid1b knockout embryos . The functions of histone demethylases in vivo are starting to emerge , however , in many cases the mechanisms of their action remain to be elucidated . Here , we present a detailed analysis of the role of the H3K4me2/3 demethylase Jarid1b during mouse development . In the adult organism , Jarid1b expression is also observed but it becomes more restricted . Since high expression of Jarid1b is detected during meiosis and in adult testis [29] as well as in several types of cancer [26] , Jarid1b has been proposed to belong to the family of testis-cancer antigens [60] . In future studies , it will be very interesting to characterize the function of Jarid1b in adult mice as Jarid1b presents a potential drug target for anti-cancer therapies and an understanding of its in vivo role will help to guide targeting efforts .
The derivation of targeted ( Jarid1bNeo/+ ) and conditional ( Jarid1bF/F ) Jarid1b mice has been previously described [28] . Conditional Jarid1b mice were crossed with Cmv-cre transgenic mice [61] to obtain heterozygous mice , which were further inter-crossed to generate Jarid1b knockouts . Jarid1b mice were maintained on a C57BL/6 background , unless otherwise stated . All mouse work was approved by the Danish Animal Ethical Committee ( “Dyreforsøgstilsynet” ) . For cesarean deliveries , timed matings were setup using Jarid1b+/− mice to generate experimental pups . Jarid1b+/+ mice were used for foster mothers . Pregnant Jarid1b+/− females were injected subcutaneously with 100 µl of Promon ( 50 mg/ml , Boheringer Ingelheim ) at E16 . 5 and E18 . 5 to prevent natural birth . Pups were delivered at E19 . 5 by caesarean , massaged gently to stimulate breathing and placed on a 37°C warm plate during initial examination . Pups were then placed with a foster mother and examined regularly during the first 24 hours after delivery . Dexamethasone ( Sigma ) or saline control was administered subcutaneous ( 0 . 4 mg/kg ) to pregnant females at E17 . 5 and E18 . 5 [33] . Histological analysis was performed according to standard procedures . Briefly , embryos or tissues were fixed over night in 4% buffered formaldehyde and subsequently incubated in baths of buffered formaldehyde , 96% ethanol , 99% ethanol , xylene and paraffin . Paraffin blocks were cut into 8–10 µm sections on a microtome ( Microm HM355S ) . Sections were deparaffinised , stained with hematoxylin and eosin , dehydrated and mounted using VectaMount ( Vector ) . The neuraxis in caesarean delivered E18 . 5 embryos was removed by dissection in an ice cold , oxygenated ( 95% O2 , 5% CO2 ) solution containing 250 mM glycerol , 3 mM KCl , 5 mM KH2PO4 , 36 mM NaHCO3 , 10 mM D- ( + ) -glucose , 2 mM MgSO4 and 0 . 7 mM CaCl2 . Brainstem-spinal cord preparations , which contained the entire brainstem and the cervical part of the spinal cord , were placed in a 2 ml recording chamber with a temperature of 29°C and was constantly superfused at a rate of 2 ml/min with preheated oxygenated ( 95% O2 , 5% CO2 ) artificial cerebrospinal fluid solution ( ACSF ) . The ACSF solution contained 130 mM NaCl , 5 . 4 mM KCl , 0 . 8 mM KH2PO4 , 26 mM NaHCO3 , 30 mM D- ( + ) -glucose , 1 mM MgCl2 and 0 . 8 mM CaCl2 . Glass-pipette suction electrodes ( tip-diameter of 40–160 µm , A-M Systems , Carlsborg , USA ) were placed on C3 , C4 , or C5 rootlets to record spontaneous respiratory-related nerve-activity . Nerve potentials were amplified by a custom-built nerve amplifier ( ×50 , 000 ) , filtered at DC-2 KHz , and digitized ( 2 . 5 KHz ) by a PCI–6289 , M Series A/D-board ( National Instruments , Austin , USA ) controlled by Igor Pro ( Wavemetrics , Lake Oswego , USA ) software . For whole-mount beta-galactosidase stainings , embryos were fixed in PBS with 0 . 25% glutaraldehyde for 10–30 min , washed in PBS and stained with PBS containing 0 . 02% NP40/IGEPAL , 0 . 01% sodium deoxycholic acid , 2 mM MgCl2 , 20 mM Tris-HCl pH 7 . 4 , 5 mM Potassium ferrocyanide and 5 mM Potassium ferricyanide until desired colour intensity . After post-fixation in 4% paraformaldehyde over night at 4°C , embryos were passed through a glycerol gradient incubating several days at each concentration , and finally stored in 100% glycerol . For beta-galactosidase stainings of cryo-sections , embryos were fixed in 0 . 2% paraformaldehyde over night at 4°C , incubated in PBS with 2 mM MgCl2 and 30% sucrose over night at 4°C , embedded in OCT ( TissueTek , Sakura ) and stored at −80°C . Samples were cut into 8 µm sections on a cryostat ( Leica CM3050 ) , post-fixed in 0 . 2% paraformaldehyde for 10 min on ice , washed in PBS with 2 mM MgCl2 , incubated in PBS with 2 mM MgCl2 , 0 . 01% deoxycholic acid and 0 . 02% NP40 for 20 min on ice , and stained as described above . Sections were washed , counter-stained with eosin , passed through an ethanol gradient , incubated in xylene and mounted in VectaMount ( Vector ) . Whole-mount in situ hybridization of mouse embryos was performed as previously described [62] . Embryos were isolated at E10 . 5 , fixed in 4% paraformaldehyde for 2 hours and stored in 100% methanol at −20°C . After bleaching with methanol/H2O2 , embryos were rehydrated , blocked in PBS with 2% milk and 0 . 1% triton ( PBSMT ) and stained over night at 4°C with anti-neurofilament antibody ( Developmental Studies Hybridoma Bank , 2H3 , 1∶50 ) . Embryos were washed in PBSMT , and incubated over night at 4°C with peroxidase-conjugated goat anti-mouse IgG ( Jackson ImmunoResearch Laboratories , 111-035-146 , 1∶500 ) . After several washes in PBSMT , embryos were incubated in 0 . 3 mg/ml DAB ( Sigma ) in 0 . 5% NiCl2 for 30 min , H2O2 was added to a concentration of 0 . 0003% and the embryos incubated until the desired colour intensity was obtained . Finally , embryos were dehydrated through a methanol gradient and cleared in 1∶2 benzyl alcohol∶benzyl benzoate in glass containers . For skeletal preparation , embryos were isolated at E17 . 5 , eviscerated and the skin removed . Embryos were fixed over night in 100% ethanol , rinsed in 95% ethanol and stained over night in 0 . 15 mg/ml Alcian Blue in 95% ethanol containing 20% glacial acidic acid . Embryos were washed in 95% ethanol , cleared in 1% KOH for 4 hours and stained with 50 mg/l Alizarin Red in 1% KOH over night at 4°C . Final clearing of embryos was performed in 1% KOH for 3 hours and through a gradient of 1% KOH/glycerol until final storage in 100% glycerol . Antibodies used in this study include anti-Jarid1b ( DAIN ) [28] , anti-H3K4me3 ( Cell Signaling , C42D8 ) , anti-H3K27me3 ( Cell Signaling , D18C8 ) , anti-H3 ( Abcam , 1791 ) , anti-H4 ( Millipore , 05-858 ) , anti-ß-III-tubulin ( Sigma , T8660 ) , anti-Glial Fibrillary Acidic Protein ( GFAP ) ( DakoCytomation , Z0334 ) , anti-Vinculin ( Sigma , V9131 ) and anti-ß-tubulin ( Santa Cruz , sc-9104 ) . Whole brain from E12 . 5 was dissociated , filtered , resuspended in PBS with 5% FBS and stained for 20 min on ice using the following antibodies: anti-Prominin-1-biotin ( MACS Miltenyi Biotec , 130-092-441 ) , anti-CD-15-FITC ( BD Biosciences 332778 ) , anti-A2B5-APC ( MACS Miltenyi Biotec , 130-093-582 ) and anti-CD24 ( BD Biosciences , 553262 ) [63] . Cells were washed , incubated with secondary fluorescent-conjugated antibodies for 20 min on ice , washed again and resuspended in buffer for viability dye staining containing 7AAD ( BioLegend 420404 ) . Cells were analyzed on a FACS Aria flow cytometer ( BD Biosciences ) . Single viable cells were gated into the following populations: Neural progenitors: CD15 low , Prominin low , CD24 high; Neural stem cells: CD15 high , Prominin high , CD24 low . Chromatin immunoprecipitation ( ChIP ) and ChIP-sequencing were performed as previously described [28] . MACS2 [64] was used to identify regions with increased histone methylation . For E8 . 5 embryos , five heads of embryos with 3–8 somites were pooled per ChIP , corresponding to approximately 100 , 000 cells in total . For ChIP and expression analysis of P0 brains , single brains were divided into fore- and hindbrain . Base-calling and demultiplexing of raw sequencing data was performed using the standard Illumina pipeline ( CASAVA , version 1 . 8 . 2 ) followed by alignment to the mouse genome ( mm9 assembly ) with bowtie ( version 0 . 12 . 7 ) [65] using the following parameters: -S -m 1 mm9 . Samtools ( version 0 . 1 . 18 ) [66] and bedtools ( version 0 . 1 . 18 ) [67] were applied for conversion of files between alignment formats . Furthermore , reads were extended to a total length of 250 bp ( estimated DNA fragement size ) in the 3′ direction . Various command-line utilities from UCSC ( http://hgdownload . cse . ucsc . edu/admin/exe/linux . x86_64/ ) were used to generate normalized bigwig track files for viewing in the UCSC genome browser . The files were normalized to tags per million after removing duplicate reads . Peak calling was performed in MACS2 [64] using the following settings: –broad -f BAM -g mm . Gene expression analysis was previously described [28] . For RNA isolation of E8 . 5 embryos , the RNAeasy Microkit ( Qiagen ) was used . For microarray analysis , four heads of E8 . 5 embryos were pooled per sample and three biological replicates analyzed . RNA was hybridized on mouse Gene 1 . 0 ST arrays by the RH Microarray Center at Rigshospitalet , Copenhagen , following Affymetrix procedures . Microarray data was analyzed using Gene Array Analyzer software [68] with default settings , P-value<0 . 05 and a log2 fold change of +/−1 . The ChIP-seq and microarray data have been submitted to the Gene Expression Omnibus ( GEO ) database ( GSE41174 ) . Primer sequences are provided in Table S1 . | Histone modifications are involved in transcriptional regulation and thus affect cellular identity , differentiation , and development . We study the histone demethylase Jarid1b ( Kdm5b/Plu1 ) , as it has been reported to be highly expressed in several human cancers and therefore might present a novel target for anti-cancer therapies . To gain insights into the physiological role of Jarid1b , we have generated a Jarid1b knockout mouse . We show that loss of Jarid1b affects survival of newborn mice and that Jarid1b is required for the faithful development of several neural organs . To understand how Jarid1b regulates embryogenesis , we identified genes with increased H3K4me3 at a genome-wide scale as well as Jarid1b target genes during development . In Jarid1b knockout embryos , master regulators of neural development are expressed at higher levels , underscoring the importance of Jarid1b in transcriptional regulation . Furthermore , we extend previous reports of overlapping Jarid1b and Polycomb target genes to show the functional relevance of this observation . Our results provide the first detailed analysis of the role of Jarid1b in normal development and provide a basis for further studies evaluating the contribution of Jarid1b to tumorigenesis . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biology"
] | 2013 | The Histone Demethylase Jarid1b Ensures Faithful Mouse Development by Protecting Developmental Genes from Aberrant H3K4me3 |
Copy-number variations ( CNVs ) constitute very common differences between individual humans and possibly all genomes and may therefore be important fuel for evolution , yet how they form remains elusive . In starving Escherichia coli , gene amplification is induced by stress , controlled by the general stress response . Amplification has been detected only encompassing genes that confer a growth advantage when amplified . We studied the structure of stress-induced gene amplification in starving cells in the Lac assay in Escherichia coli by array comparative genomic hybridization ( aCGH ) , with polymerase chain reaction ( pcr ) and DNA sequencing to establish the structures generated . About 10% of 300 amplified isolates carried other chromosomal structural change in addition to amplification . Most of these were inversions and duplications associated with the amplification event . This complexity supports a mechanism similar to that seen in human non-recurrent copy number variants . We interpret these complex events in terms of repeated template switching during DNA replication . Importantly , we found a significant occurrence ( 6 out of 300 ) of chromosomal structural changes that were apparently not involved in the amplification event . These secondary changes were absent from 240 samples derived from starved cells not carrying amplification , suggesting that amplification happens in a differentiated subpopulation of stressed cells licensed for global chromosomal structural change and genomic instability . These data imply that chromosomal structural changes occur in bursts or showers of instability that may have the potential to drive rapid evolution .
Copy number variations ( CNVs ) are regions of DNA either deleted or duplicated/amplified relative to a reference genome . CNVs constitute the most ubiquitous differences between individual or personal human genomes [1] , can be associated with many Mendelian and complex human diseases [2] because de novo events cause a significant fraction of sporadic birth defects [3] and are responsible for the selected rapid evolutionary changes accompanying animal domestication ( e . g . [4] ) . In human , CNV arises either through non-allelic crossing-over between repeated sequences , giving recurrent end-points , or at non-recurrent positions . Non-recurrent events show two conspicuous features: many of them show complexity [5] , often in the form of lengths of nearby sequence inserted at the novel junction , and second , the junctions tend to show microhomology of a few base-pairs , not sufficient to allow homologous recombination to occur ( reviewed by [6] , [7] ) . We and others have reported similar properties in our studies of amplification in Escherichia coli , namely that some of the events are complex , and the junctions show microhomology at the site of the joint making E . coli a useful model for studying the mechanisms that underlie human CNV [8] , [9] , [10] . Amplification at lac in the Lac assay system on an F′-plasmid in E . coli requires DNA polymerase I ( Pol I ) but not excision repair ( also involving Pol I ) , placing the event at replication forks [10] , [11] . Parenthetically , in yeast both break-induced replication ( BIR ) [12] and CNV [13] require the non-essential DNA polymerase subunit pol32 . Furthermore , in E . coli amplification is enhanced by 3′ single-stranded DNA ends , suggesting priming of DNA synthesis [10] . Based on these observations we proposed the long-distance template-switch model , in which the 3′ primer-end at a stalled replication fork switches template to a different replication fork and anneals at a site of microhomology [10] . Repeated switches would explain the complexity at the junctions , and a template switch to a region already replicated would produce a duplication that could be expanded into amplification by unequal crossing-over . However , amplification also requires TraI [14] , an endonuclease that nicks the F-plasmid at the origin of transfer , oriT , and this requirement is suppressed by double-strand cutting near lac on the F′-plasmid [14] . Taking these findings together with the report that BIR repair of collapsed ( broken ) replication forks in yeast shows frequent template switching [15] , we proposed that microhomology-mediated ( MM ) events might occur by a modification of BIR ( MMBIR ) whereby repair is achieved by annealing of the 3′-tail at a collapsed fork with any nearby single-stranded DNA [6] . Annealing would have lower homology requirements than homologous recombination , and hence explain the microhomology junctions . Another possible explanation for recombination at sites of microhomology is non-homologous end-joining ( NHEJ ) . NHEJ requires two double-strand breaks to make every heterologous junction , and consequently complex events would require multiple DNA double-strand breaks . NHEJ fails to explain the requirement for DNA polymerase I or the involvement of 3′ DNA ends in amplification . For these reasons we do not favor NHEJ as a mechanism for adaptive amplification in the Lac assay , nor is it our preferred mechanism to explain microhomology observed at human genomic deletion rearrangements with a single junction; the latter being explained more parsimoniously by a single template switch [1] , [5] . In the Lac assay in E . coli [16] , stationary phase Lac− cells carrying a +1 frameshift mutation are spread on lactose minimal medium . Lac+ colonies arise over days from the starving cells . The colonies carry either amplified arrays of the leaky lac allele or a compensating frame-shift mutation ( point mutants ) [17] . The point-mutant Lac+ colonies are found to carry secondary unselected mutations at a high frequency ( up to 10−2 for some loci ) [18] , [19] , [20] . Starved cells on the same plate that did not mutate to Lac+ carry a much lower frequency of unselected mutations [19] . Thus , some or all Lac+ colonies arise from a hypermutating subpopulation ( HMS ) while the majority of the starved cells do not take part in hypermutability . The HMS is defined by the stress responses that are activated in given cell [21] , [22] . It has not been established whether or not amplified Lac+ colonies arise from a chromosomally unstable subpopulation , though it has been shown that they do not arise from the HMS [17] . This study reports the use of array comparative genomic hybridization ( aCGH ) to analyze genome-wide changes in copy number . We sought , first , evidence of secondary unselected cell-wide chromosomal structural instability in those cells that carry amplification at lac . Evidence of secondary chromosomal structural change in amplified isolates that is not seen in controls constitutes evidence of a physiological difference that affects genome stability between cells undergoing amplification and those that do not . We found a significantly higher occurrence of unselected events that would not have bestowed a growth advantage among amplified isolates compared with stressed Lac− control cells . This demonstrates that amplification is happening in a differentiated subpopulation undergoing general chromosomal structural change , suggesting that this differentiation might be mediated by stress responses . Second , we sought further evidence that amplification in E . coli shows similar complexity to human non-recurrent CNV events . We found complexity in the amplification events in over 7% of amplified isolates , mostly in the form of inverted duplications within the amplicons ( units of amplification ) , confirming that there is a tendency for events that mediate chromosomal structural change to be complex .
The most common complexity was an inverted duplication embedded in the amplified region ( Figure 2a , PJH1490 ) . This was found in 16 of 300 amplified isolates ( 5 . 3% ) ( Table 1 ) . The same configuration was found to be common in the study by Kugelberg et al . with the Lac assay in Salmonella enterica [8] . In all 16 cases , the lac region was included in the embedded duplication . Detailed study of these events showed that the embedded inverted duplications vary in size from 5 . 2 to 42 . 6 kb . Two novel junctions were found in each case . The junctions showed microhomology of 3 to 30 bp ( Table 1 ) . We interpret these events as two inverted template switches that generate an inverted triplication , followed by unequal crossing-over that generates the amplified array ( Figure 3 , see Discussion ) . We identified two other inverted regions that generated a distinct pattern on aCGH data where part of the amplicon appears to be detached from the rest on the map of the parental strain based on the standard map of E . coli ( PJH39 and PJH2122 ) ( one example , PJH39 , is indicated in Figure 2b by an open arrow ) . When the map is corrected to include this inversion , the amplicon is seen to be contiguous . These events show only two novel junctions , the right end of the inversion and the amplification being the same junction . We therefore regard the inversion and the duplication as parts of the same event , and explain them below as a pair of inverted template switches followed by unequal crossing over ( Figure 3c , 3d ) . Another event of the same type , PJH2058 , that did not involve inversion or duplication of lac ( apart from the amplification ) is shown in Figure 2c . There is a short sequence within amplicon that is present in 2-fold less copy number than the rest of the amplicon ( open arrow in Figure 2c ) . This can also be explained by 2 switches , but neither of them is inverted ( Figure 3e , 3f , see below ) . A very large tandem duplication ( about 300 kb ) was found in an isolate ( PJH1475 ) in which the F′-factor was integrated into the chromosome , so that part of the F′ including lac , and part of the chromosome was duplicated ( Figure 1 ) . We have confirmed the HFR status of this isolate by showing that conjugational transfer of proAB , which is on the F′-plasmid in FC40 , is RecA-dependent in this isolate , whereas it would not be if it were situated on a plasmid . The duplication is flanked by IS5 sequences , and therefore was presumably formed by homologous recombination ( Table 1 ) . Similarly , the integration of the F′-plasmid occurred by homologous recombination between sequences that are in common between the chromosome and F′128 , because aCGH detected no other copy number change . Two other large duplications , PJH1477 and PJH1487 , were found that included lac and had one or both ends outside the chromosomal sequence on the F′ . The junctions were not found in the IS3 elements that span chromosomal sequence on the F′-plasmid as has been observed previously [8] , [9] . The same two events contained duplications within the amplified segment . The junction sequences of both duplications were found to be recalcitrant to amplification by PCR . Multiple primer pairs were used in all pair-wise orientations , but no product or only unspecific product was found . Similar results have been reported for some human non-recurrent copy number changes ( e . g . [25] ) . It is possible that these represent translocations , further unanticipated orientational complexities at the breakpoint junctions , or insertions of large genomic sequences/structures between the designed primers that do not correspond to a preconceived notion based on a reference genome sequence used for primer design . Array CGH provides copy number information , but neither positional nor orientational information . We were unable to characterize these further . These data establish that , like in human , a significant proportion of events of chromosomal structural change that generate amplification are complex in that more than one structural change occurred , apparently within the same event . This applies to 19 of 300 events resolved by our approach ( omitting large duplications that might have assisted amplification , but might not be part of the same event ) . In the same sample of 300 amplified isolates , we also found six that included a chromosomal structural change that was not apparently directly involved in the amplification . None was seen in the 240 stressed control isolates . The null hypothesis that the amount of that unrelated chromosomal structural change does not differ between amplified and stressed non-amplified isolates , can be rejected ( p = 0 . 036; Fisher's exact test [26] , [27] ) . Using the Peto Odds Ratio we can estimate the odds ratio ( OR = 6 . 2 ) and a corresponding 95% confidence interval ranging from 1 . 2 to 31 . 0 . [24] . Duplications should be unstable , so it is not surprising that we saw none that did not duplicate lac and thereby provide selection for maintenance of the duplication . Four of the unselected events were deletions ( 1 . 33% of 300 events ) : two on the F′-plasmid and two on the chromosome . One of the deletions ( PJH1474 ) was flanked by non-identical IS elements , and so might have occurred by homeologous recombination or alternatively might have utilized the shorter homology stretches to mediate a template switch . The other three show microhomology junctions ( 1 to 4 bp ) , and so probably happened by events similar to those generating amplification . The chromosomal deletions were 0 . 8 and 1 . 6 kb long , and are situated at about 1 . 4 and 1 . 6 megabases on the standard reference E . coli map ( PJH2116 and PJH1482 respectively ) . Deletions of 0 . 2 and 7 . 5 kb long ( PJH2030 and PJH1482 respectively ) were found on the F′ at about 44 kb and 50 kb from lac respectively ( Figure 1 ) . An example , PJH1474 , is shown in Figure 2D . We found one inversion because it made an apparent separation of the amplicon into two parts ( based on the standard map ) ( Figure 2e , PJH1479 ) . The endpoints of the inversion and the amplification are different , so we see no evidence that the events are related . The inversion presumably happened before the amplification , and the amplification then included part of the inverted region . Because most inversions would not be detected by aCGH , we searched all 300 amplified and 240 stressed control isolates for inversion within 20 kb to either side of lac by unidirectional PCR ( Figure 4 ) . When PCR primers point in the same direction , there is no PCR product unless the sequence at one of the primer binding sites has been inverted . We found one further inversion in an amplified isolate ( PJH1465 ) and none in the controls . These two inversions are described in Figure 4 . It is interesting that , although the exchanges were almost reciprocal , the junctions are not exactly in the same position , so that a mutation of a small deletion or insertion is made at either end of both inversions .
Kugelberg et al [8] , [9] , studying the Lac assay in Salmonella enterica , have proposed that amplification at lac is not induced by the stress of starvation , but is a product of selection for more β-galactosidase expression with parameters within those established for chromosomal structural changes in growing cells of E . coli . We regard these amplification events as stress-induced because it was not pre-existing [17] and has been shown to require two stress response regulators: the general and stationary-phase stress-response regulator σS ( RpoS ) [28] , [29] and the periplasmic misfolded protein stress-response regulator σE ( RpoE ) [30] . The strong requirement for σS would appear to be definitive , except that a few RpoS-controlled functions are expressed in growing cells [31] , so one might argue that it is growth-dependent functions that are required . This idea is refuted by the demonstration that the growth phase level of expression of σS is insufficient for adaptive mutation [14] . The strong requirement for the RpoE stress-response is for both formation and maintenance of amplification [30] . The requirement for two of the cell's major stress-response regulators is a strong argument for stress-induction of amplification . We report that a significant number of amplification events are complex in that they show more than one novel junction , indicating more than one non-homologous recombination event . The case that amplification events in the Lac assay reflect template switches during replication has been made in detail elsewhere [6] , [7] . The events described here are readily interpretable in terms of template-switching mechanisms , and support the concept . Figure 3 describes the template switch processes that we propose to have occurred to explain the complex events that we see , based on either the long-distance template switch model [10] or the MMBIR model [6] . Figure 3a and 3b show how two inverted template switches form an inverted triplication interspersed with direct and inverted duplications . Non-allelic homologous recombination ( or unequal crossing-over ) between directly duplicated regions will generate the complex amplicon that we see . Kugelberg et al . [8] use a very similar pattern of events to explain this configuration , which was also common in their data for amplification in S . enterica . Figure 3c and 3d shows how a different configuration , amplification overlapping an inverted region , which we saw twice ( Figure 2b , PJH39 and PJH2122 ) , can be derived very similarly from two inverted template switches followed by unequal crossing-over . The difference is only in the relative positions of the two template switches . If an inverted template switch occurs , the product will not be a viable Lac+ clone under the conditions of these experiments unless there is a second inverted template switch . This is because a single inverted switch will generate an incomplete F′-plasmid . However , this requirement for a second inversion cannot be the explanation for all complexity because , as shown in Figure 3e and 3f , we also see evidence of double template switches in direct orientation , where no consideration of viability exists . In the case portrayed in Figure 3e , PJH2058 , the amplicon consists of a direct duplication that has a deletion between the repeats . The events depicted in Figure 3e and 3f differ from those interpreted above that include inversions only in the positions and orientations of the switches . To determine whether a secondary structural change might play a role in the amplification process , we screened for loss of amplification in one strain carrying a secondary inversion . When this strain was used in a starvation-induced mutation experiment , the rate of amplification was unchanged from the control strain FC40 ( Figure S2 ) suggesting that there is no functional reason for the occurrence of this inversion in a strain that carries amplification at lac . Taking the four deletions and the two inversions that did not share a junction with amplification as events secondary to amplification , we see a significant occurrence of secondary events in cells that underwent amplification compared with starved cells not showing amplification ( p = 0 . 036; Fisher's exact test ) . This is clearly an under-estimate of structural changes because duplications would be expected to be unstable , so it is not surprising that we saw none that did not duplicate lac . Those duplications that include lac presumably provide selection for maintenance of the duplication [8] , [9] . We only found inversions that were close to lac because we did not look elsewhere . Using aCGH , we would detect only those unrelated inversions that overlapped the amplicon , and we found one of these . We also looked for inversion by unidirectional PCR , but only in the 40 kb surrounding the lac locus . The meaning of the finding that the sample of amplified isolates differs in the frequency of secondary events from non-amplified cells from the same plate is important . First , it means that amplifying cells differ from other cells in their propensity to undergo chromosomal structural change . Second , it shows that this happened in a subpopulation of starved cells rather than in the whole population of starved cells , because starved cells that did not undergo amplification provide the basis of comparison . The identification of chromosomal structural mutations that are secondary to the selected event is analogous to the finding in the Lac assay that lac+ point mutation is correlated with an elevated frequency of other unselected secondary point mutations [18] , [19] , [20] . Third , the discovery in the same cells of events that are apparently separate from the amplification events shows that the structural changes occurring during starvation on lactose medium are not targeted specifically or exclusively to the lac locus . The existence of this chromosomally unstable subpopulation is compatible with the concept of stress-induced differentiation to a condition permissive for chromosomal structural change and genomic instability , and is incompatible with models that seek to explain these events as normal change and selection in slowly growing cells . We suggest that this subpopulation is differentiated to a physiological condition that allows chromosomal structural change . This is suggested by the finding that some of the secondary events occurred on the chromosome , indicating that a diffusible factor is involved . Further , we suggest that this differentiation was induced by the stress of starvation . We made the unexpected finding that some inversions involve almost , but not quite , reciprocal non-homologous recombination so that the junctions show insertions or deletions of a few tens of base-pairs . We suggest that this might occur as follows: If a template switch occurred because a replication fork was stalled by secondary structure forming in template DNA , then the complementary sequence on the other template would also be capable of forming a similar secondary structure . If two such sites occurred in a short interval , within the dimensions of a single replication fork , then the series of template switches portrayed in Figure 5 might explain how the almost reciprocal recombination occurred to form the inversion . Uncoupling of lagging-strand synthesis , after leading-strand synthesis is stalled by secondary structure ( labeled “1” in Figure 5 ) , might allow both structures to form on both strands , and so expose one to two kb of single-stranded sequence within the same replication fork . An inverted switch of the nascent leading strand from “1” to where lagging-strand synthesis is blocked ahead of the second secondary structure “2” is followed by synthesis in inverted orientation as far as the complementary secondary structure to the first blockage “1R” . The second template switch is to downstream of the complement to the second structure “2R” , thus completing the not quite reciprocal exchanges that flank the inversion . This allows replication to escape the blockage imposed by secondary structures . Figure 5 shows the secondary structures that could form in the regions involved in one of the inversions . All four junction sequences are in positions that can form a stem or a stem/loop of secondary structure . We suggest that at least these two inversion events formed by template switches [32] , [33] within a replication fork induced by secondary structures in DNA . REP is a pseudopalindromic sequence of about 38 bp that occurs in clusters in intergenic regions [34] . Kugelberg et al . have noted that there is a tendency for junctions to occur at REP sequences [8] , [9] , and interpret this as evidence of homologous recombination . We found that 22 of 90 sequenced novel junctions ( 24% ) occur at REP sequences ( Table 1 and Table S1 ) . Of these , 14 are too short for homologous recombination ( 5 to 20 bp ) and 8 are in a range that might or might not allow homologous recombination ( 29 to 32 bp ) [35] , [36] , but could also allow microhomology-mediated template switches as has been proposed for Alu repetitive sequences in the human genome [5] . REP clusters are rich in potential to form secondary structures . Figure S1 shows secondary structure predicted in a cluster of REP sequences near lac that is involved in 20% of the junctions listed in Table 1 and Table S1 . We suggest that the propensity of the region to form secondary structures , rather than homology , is instrumental in forming this hotspot . Study of the positions of novel junctions of amplification show a preference for the stem of potential stem-loop structures . For 40 amplification junctions that we sequenced ( Table S1 ) , 16 are REP sequences , and therefore rich in potential secondary structures . Analyzing potential secondary structures in the regions close to the junctions of the other 24 amplicons , only one junction sequence is confined to predicted unstructured sequence , and only one is confined to a potential hairpin loop . Those in the commonest class ( 10/24 ) occur on the stems of predicted secondary structures , and 9/24 are on both a stem and a loop . We considered that secondary structure might target amplification by blocking the progress of replication forks , or it might function to provide single-stranded DNA to which a primer could anneal during template switching . Because a minority of junctions are situated on a predicted hairpin loop where single-stranded DNA might occur , we favor the hypothesis that secondary structures target amplification by blocking replication . Others have reported that secondary structures in DNA are involved in chromosomal structural change [37] . Direct evidence of fork stalling at inverted repeats in vivo strongly suggests that stalling is mediated by hairpin formation on the lagging-strand template at replication forks [38] . We have suggested above that some inversions are formed by template switching within a replication fork . Template switching is much more difficult to apply to other events reported here because most switches cover tens of kb , well beyond the 1 . 5 kb dimensions of a replication fork in E . coli . For this reason , we suggested previously that template switches occur between different replication forks: the long distance template-switch model [10] . Based on the evidence that double-strand breaks are involved in amplification at lac , we later suggested that the mechanism was a modification of break-induced replication ( BIR ) at collapsed replication forks , namely that in place of RecA-mediated strand invasion , the broken end annealed by microhomology to nearby single-stranded DNA ( MMBIR ) [6] . We know from experiments that use I-SceI endonuclease to make double-strand cuts near to lac that double-strand breaks increase amplification at lac [14] . From this we suggested that nicking at oriT by TraI provides a discontinuity in the DNA template that leads to replication fork collapse [6] followed by MMBIR . We present evidence that , among stressed cells , a small proportion enters a state of heightened genomic instability during which multiple chromosomal structural changes might occur anywhere in the genome . Many such changes would be expected to be disadvantageous , but rarely a change occurs that allows escape from the stress . Because the events that we studied here in a bacterial model system are similar to those described for copy number changes in human , this conclusion might apply generally throughout biology . This view suggests that genome evolution might occur in bursts of multiple simultaneous chromosomal changes induced by stress . This view also has implications for understanding cancer progression in the stressful tumor microenvironment and the stresses imposed by chemotherapy , both of which might induce showers of chromosomal structural changes .
Escherichia coli cells of strain SMR4562 [39] , isogenic with FC40 [16] carry the conjugative plasmid F′128 with a leaky lac +1 frameshift mutation were initially grown to stationary phase for 3 days at 32° [14] . We then followed the standard procedure [40] for adaptive mutation experiment in the Lac assay [16] . Lac+ colonies arise over several days , and are marked daily . Amplification was distinguished from point mutation by its instability as seen by blue and white sectoring of colonies grown on rich medium with X-gal . 284 Lac+ colonies from day 7 together with 16 previously published amplified strains [10] were collected for further study . We also studied 60 Lac+ colonies arising on day 7 that carried point mutations reverting the lac mutation . 180 Lac− stressed FC40 control cells were collected by taking plugs from the same lactose plate on day 5 . Sixty colonies derived from unstressed control cells were taken from the initial stationary phase culture . These 584 new isolates are identified by strain numbers PJH1458–PJH1642 and PJH2025–PJH2425 . Those described previously [10] are strains PJH2 , PJH5 , PJH6 , PJH7 , PJH19 , PJH20 , PJH22 , PJH26 , PJH27 , PJH39 , PJH59 , PJH64 , PJH79 , PJH80 , PJH81 and PJH165 . Total genomic DNA was extracted from exponential culture in M9 lactose medium for Lac+ isolates or M9 glycerol medium for Lac− isolates by using the QIAGEN DNA Purification kit . E . coli custom high-resolution genomic microarray ( 4×44K ) containing 44 , 000 unique sequence oligonucleotides spaced at about 100-bp intervals were obtained from Oxford Gene Technology ( OGT ) . Probe labeling and hybridization were performed following the manufacturer's protocol ( Agilent Oligonucleotide Array-based CGH for Genomic DNA Analysis ) . Slides were scanned on a GenePix 4000B Microarray Scanner ( Axon Instruments ) . Data extraction , normalization and visualization were achieved by using Agilent Feature Extraction Software A . 7 . 5 . 1 . Extraction data were analyzed for copy number differences by using Microsoft Excel software . All occurrences of two or more adjacent probes showing 2-fold or more increase or decrease in copy number relative to the reference FC40 DNA were investigated further , except those that mapped to repetitive elements or prophages . All deletion , inversion and duplication junctions were further validated by PCR and sequencing . Inward-facing primers for deletions and inversions and outward-facing primers for tandem duplication were designed based on sequence from National Center for Biotechnology Information ( NCBI ) Escherichia coli K-12 substr . MG1655 . Long-range PCR was performed using LongAmp™ Taq Master Mix ( New England Biolabs ) . The PCR products were purified with either a QIAquick PCR Purification Kit ( QIAGEN ) or a QIAEX II Gel Extraction Kit ( QIAGEN ) following the manufacturer's instructions , and sequenced by Lone Star Labs ( Houston , Texas , United States ) . DNA sequences were analyzed by comparison to reference sequences with the use of BLAST ( http://blast . ncbi . nlm . nih . gov/Blast . cgi ) . Possible secondary structures in DNA were found by use of DNAMAN version 6 ( Lynnon Biosoft ) . Deamplified lines were derived from amplified isolates by screening for sectors in colonies of amplified strains that showed a low level of β-galactosidase as seen on medium containing X-gal , but yet retained the ability to grow without proline . | Much of the difference between individual humans is in the number of copies of genes and lengths of genome . The mechanisms by which copy number variation arises are not well understood . We sought information on copy number change mechanisms by extensive use of array comparative genomic hybridization of whole genomes in bacteria selected for amplification of part of the genome . We report that about 10% of amplified isolates carried other chromosomal structural changes associated with the amplification , a result comparable to that seen in human copy number variants . Importantly , we found a significant occurrence of structural changes that were not involved in the amplification event . These were not seen in a control sample of stressed cells not carrying amplification . This establishes that chromosomal structural change happens in a subpopulation of cells apparently licensed to undergo these changes . Because the changes occur under the stress of starvation and require two of the cells' stress-response systems , we propose that licensing for cell-wide structural change in this subpopulation is a component of response to stress . This idea has implications for the mechanisms of evolution and cancer progression , suggesting that changes occur in a shower of events rather than as isolated random events . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cancer",
"genetics",
"genetics",
"molecular",
"genetics",
"biology",
"genetics",
"of",
"disease",
"genetics",
"and",
"genomics"
] | 2011 | Global Chromosomal Structural Instability in a Subpopulation of Starving Escherichia coli Cells |
Paracoccidioides brasiliensis and Paracoccidioides lutzii are the etiological agents of Paracoccidioidomycosis ( PCM ) , and are easily isolated from human patients . However , due to human migration and a long latency period , clinical isolates do not reflect the spatial distribution of these pathogens . Molecular detection of P . brasiliensis and P . lutzii from soil , as well as their isolation from wild animals such as armadillos , are important for monitoring their environmental and geographical distribution . This study aimed to detect and , for the first time , evaluate the genetic diversity of P . brasiliensis and P . lutzii for Paracoccidioidomycosis in endemic and non-endemic areas of the environment , by using Nested PCR and in situ hybridization techniques . Aerosol ( n = 16 ) and soil ( n = 34 ) samples from armadillo burrows , as well as armadillos ( n = 7 ) were collected in endemic and non-endemic areas of PCM in the Southeastern , Midwestern and Northern regions of Brazil . Both P . brasiliensis and P . lutzii were detected in soil ( 67 . 5% ) and aerosols ( 81% ) by PCR of Internal Transcribed Spacer ( ITS ) region ( 60% ) , and also by in situ hybridization ( 83% ) . Fungal isolation from armadillo tissues was not possible . Sequences from both species of P . brasiliensis and P . lutzii were detected in all regions . In addition , we identified genetic Paracoccidioides variants in soil and aerosol samples which have never been reported before in clinical or armadillo samples , suggesting greater genetic variability in the environment than in vertebrate hosts . Data may reflect the actual occurrence of Paracoccidioides species in their saprobic habitat , despite their absence/non-detection in seven armadillos evaluated in regions with high prevalence of PCM infection by P . lutzii . These results may indicate a possible ecological difference between P . brasiliensis and P . lutzii concerning their wild hosts .
The study of biological and ecological aspects of Paracoccidioides brasiliensis [1 , 2] and Paracoccidioides lutzii [3] has been developed by several research groups in recent years . In particular , investigators have been seeking to isolate and/or detect those pathogens in clinical and environmental samples in order to obtain more in-depth data on the ecological factors that determine their geographical distribution . Both species cause paracoccidioidomycosis ( PCM ) , the most prevalent systemic mycosis in Latin America , which can be acquired by non-immunocompromised hosts inhabiting and/or working mostly in endemic rural areas in South and Central Americas [4 , 5] . The disease is acquired by soil aerosolization and inhaling of infectious particles of the fungi [5] . It is known that Paracoccidioides spp . has its habitat ( physical and geographical distribution site ) located in the soil , but its ecological niche ( sum of all interactions of the microorganism with the biotic and abiotic factors of the environment ) has not been properly determined , demanding more environmental studies [4–6] . Studies on Paracoccidioides species distribution have been focused mainly on its isolation from human patients and scarcely from environment . Wild and domestic animals have been addressed due to the difficulties to retrieve the fungus from its habitat in laboratory conditions . The few cases of environmental pathogen isolation were from soil , foliage , dog food , bats and penguin feces , almost serendipitously , with little or no repeatability [4 , 7–10] . The frequent isolation of Paracoccidioides brasiliensis from armadillos , an animal whose home range is very limited , makes this mammal an excellent environmental source for mapping the geographic distribution of the fungus [5 , 11] . The difficulties faced by environmental studies limits the understanding about the ecology and the real distribution of this genus in endemic and non-endemic areas of PCM . For instance , the majority of the Paracoccidioides isolates used for molecular typing are from clinical specimens , which can be influenced by factors such as human migration and the long latency period of this mycosis , which may be greater than one or two decades [12 , 13] , so that it is very hard to specify the exact local of infection and the occurrence of each cryptic species in the endemic and non-endemic regions where the patients are from . Identification of risk areas associated with the different species and/or genotypes can also contribute to a better understanding of Paracoccidioides biogeography and also to help the clinical procedures for PCM diagnosis and treatment [11 , 14] . However , little has been done given its methodological difficulties . In this study , a new approach based on an in situ hybridization technique with species-specific DNA probes , as well as the previously established molecular technique of Nested PCR [15 , 16] , were carried out in order to detect the pathogen in different environmental samples , such as soil , aerosol and armadillo tissues . In this study , we expand the collection sites to include new areas of Southeastern , Midwestern and Northern Brazil ( prevalent areas of P . lutzii infection ) . The main objective was to search for new ecological and biogeographical information about these fungi , in order to have a better profile of P . brasiliensis and P . lutzii distribution and dynamics in nature .
Capturing of the armadillos was authorized by SISBIO-IBAMA for all of the Brazilian territory ( IBAMA—30585–1 and IBAMA—37333–2 ) , and the Animal Ethics Committee from Biosciences Institute of Botucatu—UNESP ( protocol number 528 ) also approved the procedures . Environmental samples from Rondônia ( RO ) , Goiás ( GO ) and Minas Gerais ( MG ) states of Brazil , which have been poorly or never sampled , were selected for the molecular detection of Paracoccidioides spp . [17–19] . Samples were obtained by collecting aerosol and soil from armadillo burrows , as well as the animal specimen ( Dasypus novemcinctus ) . The distribution of collection areas along the Brazilian map are shown in Fig 1 , highlighting the assessed municipalities: Monte-Negro ( RO ) , Santo Antônio de Goiás and Guarani de Goiás ( GO ) and Campina Verde ( MG ) . Georeferenced sites of the armadillo burrows in the collection areas are listed in Table 1 . About 50g of soil was collected from the armadillo burrows by using an iron spatula . The samples were carefully collected so that the burrows were not destroyed . The soil was placed in 50 mL sterile universal bottles , sealed , identified , stored at room temperature , for up to fifteen days , and processed for DNA extraction , accordingly with our previously experience [15 , 16] . During the collection procedure , the instruments were decontaminated with 70% v/v ethanol solutions in order to avoid cross contamination of samples from one location to another [15 , 16] . For soil samplings , the minimum established number was 10 samples for burrows and trails at each site studied . In addition to the collected samples , a soil called "Dark Earth" from the occidental Amazon region ( which is an anthropogenic soil and rich in organic matter ) was kindly provided by the Center for Nuclear Energy Research Group in Agriculture ( CENA ) —USP , under the responsibility of PhD . Siu Mui Tsai . These soil samples were processed with the same methods described above ( storage and DNA extraction ) . For the capture of armadillos in the field , track traps were used and we were assisted by local hunters in order to identify areas with high animal activity . Track traps were placed in the armadillo’s trails next to those burrows for which recent animal activity had been detected . In addition to the track traps , active capture took place in the evening , the period of the animal’s highest activity . Captured animals were placed in containers with fresh and dry straw to better accommodate them during transportation until the euthanasia procedure in the laboratory . The number of animals varied according to the season ( rain or dry ) , as well as to the difficulties for finding specimens and transporting them to the laboratory . The euthanasia for the evaluated animals in this study was performed by subcutaneous administration of Zoletyl 50 ( 0 . 2 mL/kg/IV , Virbac ) , followed by cardiac puncture and administration of potassium iodine to ensure the animal's death . Spleen , liver and mesenteric lymph nodes were removed and placed in sterile plates and soaked in alcohol 70% v/v for a brief cleaning followed by saline solution ( 0 , 9% w/v ) . Small fragments ( 1-2mm ) of the organs were then placed in a Mycosel Agar culture medium ( Difco ) supplemented with 50 μg/mL Gentamicin and incubated at 35°C during 45 days [11 , 16] . After the evaluation , these plates were properly sterilized and discarded . Air sampler model Cyclone 251 BC , developed by the Centers for Diseases Control—CDC ( Morgantown , WV , USA ) and certified by the National Institute for Occupational Safety and Health ( NIOSH ) coupled to vacuum pumps type 224-44XR Model SKC Universal Pumps was used for air sampling [20] . An air sampler was placed next to the armadillo burrows as well as in the areas where the animals had recently removed the soil searching for food . The vacuum pump has a rechargeable battery life up to 24 hours of operation at maximum flow , thereby facilitating the procedures for collection in remote locations without any power sources . Aerosol and soil samples were collected in the states of GO and MG during the dry/warm season , and in RO state during the cold/rainy season . Each aerosol collection was performed in a minimum period of 60 minutes with a flow rate of 3500 mL/min . At least four samples were collected at each site . Each soil sample was subjected to total DNA extraction in triplicate , using the commercial kit PowerSoil DNA Isolation Kit—MO BIO Laboratories , Inc . Total DNA was resuspended in 100 μl of Nuclease free water and quantified in NanoVue spectrophotometer equipment ( GE Healthcare ) . Fifty microliters of each one of the three replicates were mixed in a single 1 . 5 mL tube and concentrated to a final volume of 30μl in a concentrator ( Eppendorf ) and then quantified again to confirm the new concentration of each sample . The aerosol samples were directly used for PCR without any previous DNA extraction by washing the tubes from cyclonic sampler with the PCR reaction mix . The PCRs ( Polymerase Chain Reaction ) were carried out with ITS-4/5 primers for rRNA universal fungal region ITS1-5 . 8S-ITS2 ( Internal Transcribed spacer ) [21] . A Nested PCR was performed with the product of the first amplification using specific primers for the Paracoccidioides genus , annealing in the ITS-1 and ITS-2 regions , named PbITS-E ( 5’GAGCTTTGACGTCTGAGACC3’ ) and PbITS-T ( 5’GTATCCCTACCTGATCCGAG3’ ) [16] . Both PCR mixes were prepared using 12 . 5 μl of Nuclease Free Water ( Sigma ) , 0 . 5 μl of 0 . 2 mM dNTP mix , 5 . 0 μl of 5X GC buffer , 2 . 5 μl of 30% DMSO , 0 . 625 μl of each primer at 20 μM and 0 . 25 μl of 1000 units/μl Taq Phusion DNA Polymerase ( ThermoFisher ) for each reaction with 23 . 0 μl of PCR mix for 2 . 0 μl from a DNA soil sample of approximately 15 ng/μl . Twenty five microliters of PCR reaction was performed for aerosol samples , these mixes were prepared from 100 . 0 μl of reaction mix used for washing the aerosol collection tubes , so that four PCRs were carried out for each aerosol sample . The thermal cycling conditions for the first PCR were: an initial denaturation at 98°C for 30 seconds and 39 cycles of denaturation at 98°C for 10 seconds , followed by an annealing step at 55°C for 45 seconds , and extension at 72°C for 45 seconds , after that a final extension at 72°C for 10 minutes was applied . For aerosol samples , the first step of denaturation was longer ( 5 minutes ) than the one applied to the DNA soil samples , in order to break the spores and other fungal structures , releasing the genetic material in the PCR mix . For Nested PCR , the annealing step was adjusted to 58°C . After the PCR and Nested PCR reactions , PCR products were analyzed by electrophoresis in a 1 . 5% w/v agarose gel . The bands around 450bp ( Paracoccidioides spp . ) or that were best highlighted in the gel were cut out , purified by using the commercial Kit ( GE illustrates GFX PCR DNA and Gel Band Purification ) and quantified as described above . Purified samples were sent to the Laboratory for Molecular Diagnosis of the Department of Microbiology and Immunology ( UNESP , Botucatu/SP-Brazil ) for automatic capillary sequencing in ABI 3500 DNA Analyzer ( Applied Biosystems ) equipment . For low concentration of PCR products , a new amplification reaction with the PbITS-E/T primers were performed ( double PCR ) , followed by its purification and sequencing . The obtained sequences were aligned to the reference sequences with the help of the MEGA 6 . 0 program [22] and compared to an online database ( GenBank ) [23] , to verify their identity and phylogenetic clusterization with other deposited ITS sequences from P . brasiliensis and P . lutzii . The initial sequences obtained from environmental amplicons were checked using the Sequencing Analysis software from ABI 3500 DNA Analyzer ( Applied Biosystems ) in order to improve the quality of sequences . For phylogenetic analysis and comparison of other Paracoccidioides DNAs , we used the deposited sequences under the following access numbers: EU870314; EU870315; EU870316; AY631235; EU118561; EU118560; EU118548; EU118554; EU118553; EU118549; EU118546; EU118547; EU118545; EU118543; EU118542 for P . brasiliensis and EU870298; EU870303; EU870306; EU870309; EU870310; EU870311; AF092903; EU870299 for P . lutzii . Only the sequences , herein obtained from environmental samples , presenting ≥ 97% of similarity to Paracoccidioides spp . sequences in GenBank ( blastn analysis ) [24] were considered for phylogenetic analysis in MEGA 6 . 0 software . Sequences previously generated from soil and aerosol samples from the Southeast part of Brazil [16] were included in this dataset , access numbers KP636439 to KP636474 ( S1 Table ) . In addition , clinical strain sequences from GenBank , representing P . lutzii ( Pb01—EU870297 ) , P . brasiliensis species complex S1 ( Pb18—AF322389 ) , PS2 ( Pb3—EU870315 ) , and PS3 ( AY631237 ) were included into the final dataset . Sequences were aligned with the ClustalW [25] algorithm implemented in the Bioedit software [26] . Retrieved alignments were manually inspected in order to avoid mispaired bases . Neighbor-Joining [27] and Maximum Likelihood trees were inferred in the MEGA 6 software [22] using the Jukes-Cantor [28] nucleotide substitution model . One thousand bootstrap replicates [29] were used to estimate the monophyletic clades support , and values were displayed next to the branches . Probes used in this study were commercially synthesized targeting the rRNA region of P . brasiliensis and P . lutzii , specifically for the ITS-1-5 . 8S-ITS-2 region [14 , 30 , 31] . Sixty ITS sequences from different Paracoccidioides isolates were aligned in order to select conserved regions within species , exclusive and different between P . brasiliensis and P . lutzii ( Fig 2 ) . The probes were differentially labeled on their 5’ end , with Horseradish Peroxidase ( HRP ) for P . brasiliensis , and Texas Red for P . lutzii , ( Fig 2 ) . The use of this approach for detection of Paracoccidioides spp . as well as differentiation between the P . brasiliensis complex and P . lutzii species was previously standardized [32] . For those methods , each probe was tested against other Ajellomycetaceae fungi ( Histoplasma capsulatum ) and other clinically relevant ascomycetes ( Aspergillus flavus; Aspergillus fumigatus; Trichophyton mentagrophytes ) , as negative controls for both probes ( used in FISH and TSA-FISH techniques ) . Twelve aerosol samples were collected in armadillo burrows for in situ hybridization method: four from the state of GO , four from MG and four from RO . The detection of Paracoccidioides spp . with DNA probes by the TSA-FISH method in aerosol samples was applied for greater sediment volume in the cyclone sampler tubes ( stages ) . These samples from both stages of the cyclone sampler were fixed with 1 . 5 mL of 4% Paraformaldehyde plus 0 . 1 M Phosphate solution buffer . Series of 50% , 80% and 100% ethanol solutions were used to remove cell fixation solution and to dehydrate the cells , so that they have the ability to absorb the probes to be used in the hybridization step . After dehydration , 10 mL of pre-hybridization buffer [2 . 0 mL of ultra-pure water; 4 . 0 mL of 40% Formamide; 1 . 8 mL of 5 M NaCl; 200 μl of 1 M Tris ( pH 7 . 5 ) ; 100 μl of 1% SDS; 2 mL of 10% Buffer Blocking Agent] were added to the samples for stabilization and improvement of their permeability . After this first step , cells were hybridized with probes at a final concentration of 50 ng/μl in hybridization buffer . After 16–17 hours of incubation at 42°C , the slides with fungal controls were washed with 50 mL of Washing Buffer [47 . 54 mL of ultra-pure water; 460 μl of 5 M NaCl , 500 μl of 0 . 5M EDTA , 500 μl of 1% SDS and 1 mL of 1 M Tris ( pH 7 . 5 ) ] for removal of non-specific binding probes . After washing , the slides were stabilized with 250 mL of TNT buffer [217 . 315 mL of ultra-pure water; 25 mL of 1 M Tris ( pH 7 . 5 ) ; 7 . 5 mL of 5M NaCl and 0 . 185 mL of Tween 20] . After equilibrating and washing the slides with TNT buffer , 30 μl of TSA solution ( TSA Plus PerkinElmer Kit ) were added to each slide and incubated for 30 minutes in a humid dark chamber at room temperature . The slides were washed again , dried at room temperature , prepared with the addition of 4’ , 6-Diamidino-2-phenylindole dihydrochloride ( DAPI ) and covered with a cover slip to be observed under a fluorescence microscope . These slides were divided in two groups of four slides each: one was tested against P . brasiliensis probe and the other against P . lutzii probe . Two spare slides were used as controls during the hybridization phase for each of the two methods and probes used in this study .
Seven armadillos were captured and evaluated , three from GO , three from RO and one from MG states . Information about the gender , weight and cultivated organ fragments are listed in Table 2 . After 45 days of incubation , each plate was evaluated for fungal growth similar to Paracoccidioides spp . by micro-morphological analysis . All the colonies presented morphological structures of bacteria , despite the addition of antibiotics to the culture medium and no fungal structures were identified . After 45 days , the plates were then considered negative for Paracoccidioides spp . growth . Forty-four soil samples from armadillo burrows were obtained; 12 in GO , 10 in MG and 22 in RO . Twenty-eight aerosol samples from armadillo burrows were obtained; 10 in GO , 10 in MG and 08 in RO , 16 of these samples were set aside for molecular detection by Nested PCR and the rest of the 12 aerosol samples were used for in situ hybridization techniques . The sampled burrows in the areas of RO , GO and MG states were mostly located in deforested areas of pastures or in some riparian forest sites . Positive Nested PCR amplification for ITS region of Paracoccidioides spp . was observed in 67 . 5% of soil samples and in 81% of aerosol samples . No amplicons were observed for “Dark earth” soil samples . When compared to the GenBank database , sequences revealed SNPs specific for P . lutzii and/or P . brasiliensis in all the positive soil samples from RO , GO and MG . P . lutzii was found in aerosol samples in MG and GO states , while P . brasiliensis was only detected in GO ( Table 3 ) . Phylogenetic analyses were carried out using a total of 36 sequences obtained from the current study and 11 sequences collected from the GenBank were added to the final dataset . The majority of the soil and aerosol samples clusterized within P . lutzii species ( 27 out of 36 –Fig 3A ) . P . lutzii was detected in GO , RO and MG states , as well as in São Paulo ( SP ) as previously reported [16] . All these P . lutzii sequences were displayed in a single haplotype together with the referenced clinical strain Pb01 ( Fig 3B ) . Solely 4 samples ( AR_GO1 , AR_GO19 , AR_GO11 and AR_GO2D ) were clustered within P . brasiliensis , all obtained from aerosol samples collected in GO state ( Fig 3A ) . However , only the AR_GO1 and AR_GO19 samples are clustered with the clinical strains Pb3 , Pb18 and ATCC60855 in unique haplotype representing clinical isolates from the P . brasiliensis species complex S1 , PS2 and PS3 ( Fig 3B ) . The phylogenetic distribution and haplotypic network analysis revealed higher genetic variation in the environmental samples than reported so far for Paracoccidioides clinical samples . The samples AR_GO11 and AR_GO2D are disposed to polytomic branches within P . brasiliensis and constitute single haplotypes in the network ( Fig 3 ) . Moreover , two high supported clades were observed as being closely related to P . lutzii clinical/environmental samples ( Soil clades ) . The soil clade I is composed of the soil samples SO_GO10 and SO_GO19 from GO state , while the soil clade II is composed of soil samples SO_RO11 and SO_RO12 from RO state . In addition , the AR_MG16 sample collected from aerosol in MG also appears to be a genetic variant from P . lutzii , fallen into a paraphyletic branch in the tree ( Fig 3A ) . The detection and differentiation of Paracoccidioides spp . in aerosol samples was performed after validation of FISH and TSA-FISH in P . brasiliensis and P . lutzii in culture cells . Specificity and sensitivity control tests were applied for validating the positive detection in environmental samples [32] . The probes tested ( conjugated with HRP and Texas Red ) did not hybridize with the negative controls . Both P . brasiliensis and P . lutzii cells under culture conditions show specific nuclear staining , which merge with DAPI . For the in situ hybridization , the aerosol samples with higher sediment in the tubes were obtained in dry areas ( GO and MG ) where the burrow soil was easily aerosolized . For the RO state samples , which were obtained during an intense rainy season , the pellets were less visible to the naked eye . From the 72 slides prepared for the in situ hybridization , positive hybridization occurred in 36 ( 50% of the slides ) , revealing that ten out of 12 air samples from the armadillo burrows ( 83% ) were positive for Paracoccidioides spp . ( Table 3 ) . The in situ hybridization experiments using specific probes for P . brasiliensis and P . lutzii revealed the presence of both species in GO and RO states ( Fig 4A , 4B , 4E and 4F ) , and only P . brasiliensis species in MG state ( Fig 4 ) . Fig 5 summarizes the current data on Paracoccidioides spp . detection in soil and aerosol samples by the different methodologies applied so far , including the current results and those previously obtained [16] .
Soil and aerosol samples have shown to be excellent sources for mapping the fungus in endemic areas of PCM by molecular methods , since they are easy to obtain and handle in a laboratory , therefore corroborating previous studies of our group [15 , 16] . Both soil and aerosol samples were positive for the environmental detection of Paracoccidioides spp . DNA in the sampled areas , revealing the ubiquitous distribution of these pathogens in the Brazilian territory . Aerosol samples were collected in a smaller number compared with the soil samples , due to the methodological difficulties in the field and weather conditions of each sampled location . The soil collection methodology is faster and easier to perform than the aerosol sampling , which requires more field effort to become representative . On the other hand , the aerosol sampling mimics the fungal dispersion mechanism by which rural workers become infected with the mycelia and/or conidia particles . Currently , soil from endemic or non-endemic areas for PCM may contain fungal cells with infectious potential for human and/or animal population , although the course of infections can vary according to the biotic and abiotic factors of the environment . It has been stated that agricultural activities are the activities that most favor PCM infection in humans [34 , 35] . The aerosol samples are also methodologically simple to work with in the laboratory because DNA extraction is not required . Molecular detection in soil samples was positive in two of the sampled areas ( RO and GO ) , but it was negative in MG . The molecular detection of Paracoccidioides in RO soil showed higher amplification rate as visualized in the agarose gel electrophoresis ( S1 Fig ) , probably because of the rainy conditions during the collection , which according to Barrozo et al . , 2009 and 2010 [34 , 35] , may favor fungus maintenance in soil , as well as its dispersion , causing increased incidence of PCM . Positive samples were also obtained from northeastern Goiás , a warm and dry area , which might indicate some fungal resistance to adverse conditions . This work presented the first standardization of in situ hybridization techniques for environmental search of Paracoccidioides spp . , ( HRP-probe/TSA-FISH for P . brasiliensis and Texas Red-probe/FISH for P . lutzii ) , making its detection possible in aerosol samples from the three locations studied ( RO , MG and GO ) . This new approach showed sensitivity and specificity rates similar to the well-established Nested PCR technique . The advantage of in situ hybridization is the visualization of infective fungal structures of Paracoccidioides spp . directly in the environmental samples . The detection rate of Paracoccidioides spp . in soil and aerosol samples in GO , MG and RO was lower than the detection rate in endemic areas of PCM . In most cases , the fungus was detected in locations whose air humidity and temperature conditions were very similar to those found in endemic areas of PCM . However , new distribution nuances of P . brasiliensis and P . lutzii were revealed in our study , including a remarkable resistance to adverse environmental conditions , so that the spatial distribution of the Paracoccidioides species may be larger than previously defined based mainly on clinical isolates [30 , 31 , 36] . Both growth and dispersion of this pathogen seems to be greatly influenced by the climate . While high moisture levels increase fungal growth and maintenance in soil , a brief drought period dries the most superficial layer of the soil , making the dispersion of aerosols ( mycelia particles/conidia and other microorganisms ) easy and intense [37] . This was already observed in Coccidioides spp . [38 , 39] and could explain the negative detection of Paracoccidioides spp . in aerosol samples from RO , which faced one of the most severe rainy periods of the last few years [40] . This explains the higher positive molecular detection of Paracoccidioides spp . by Nested PCR in soil samples than in aerosol samples from RO , where the collection was carried out during the raining season . Therefore , different from soil samples , which directly demonstrate the presence of the fungus , and aerosol samples also reflect the spread of the fungus in the environment , and therefore its infective potential is extremely important in the epidemiological study of Paracoccidioides spp . , corroborating the growth and blow theory which reflects the crescent number of cases after rainy seasons in endemic areas [34 , 35] and the great recent number of new PCM cases in areas of North Brazil , as RO state [19] . In RO and GO , the detection of Paracoccidioides spp . was positive mostly in deforested pastures and in some riparian forest sites . In fact , such areas with increased agricultural activity present the greatest incidence of PCM cases in these states [17–19] . Deforestation of preserved areas exposes the soil and naturally or deliberately changes its chemical conformation , which can favor the infection of rural workers or other people living in these areas , leading to the emergence of PCM . Our studies indicate that P . lutzii and P . brasiliensis is often found in soil and aerosol samples in all four of the sampled regions . For this reason , we hypothesized that the geographic distribution of PCM caused by different Paracoccidioides species may be associated to the capacity of each fungal species to produce infective propagule in the current environmental conditions , which includes soil management for local agriculture activity ( sugar-cane , coffee and cattle breeding ) . For instance , despite the environmental molecular detection of P . brasiliensis in Goiás and P . lutzii in Minas Gerais , the majority of PCM cases in these states are caused by P . lutzii and P . brasiliensis , respectively . Considering the difficulty for animal capture and transportation , a reasonable number of armadillos for each sampled area were obtained . All armadillos evaluated in this study were negative for culture isolation of Paracoccidioides spp . However , molecular detection in soil and aerosol samples indicated the presence of Paracoccidioides spp . in these areas . This negative result for the non-endemic areas , where these animals were collected ( northeastern GO and MG ) , may point to the relationship of environmental conditions and the possibility of infection and/or disease in these animals , and probably in humans too . On the other hand , the negative isolation in armadillos from the endemic area ( RO ) is more intriguing . In this area , most of the sequences detected showed high similarity to P . lutzii species ( in all positive soil samples by Nested PCR , and in 50% of the positive aerosol samples by in situ hybridization ) . This may indicate that the relationship between armadillo and P . lutzii is different from the well-known interaction between armadillo and P . brasiliensis , which could have resulted in a speciation process ( P . brasiliensis complex X P . lutzii ) driven by hosts . P . brasiliensis S1 and PS2 species ( São Paulo state/Brazil ) and PS3 ( Colombia ) being highly recovered from the armadillos , and D . novemcinctus and Cabassous centralis [11 , 41] in endemic areas , while no P . lutzii has been isolated from these animals yet . According to the phylogenetic analysis , there was a prevalence of sequences belonging to the P . lutzii species in all evaluated areas , which reflect the distribution pattern of clinical isolates observed in recent works [14 , 16 , 18] . Despite the negative isolation of Paracoccidioides spp . from armadillos , data from molecular detection of these pathogens in soil and/or aerosol samples may be useful for delineating the geographic distribution of Paracoccidioides spp . Herein for the first time , environmental genetic variants were reported for Paracoccidioides genus ( Fig 3 ) . Phylogenetic and haplotype data revealed the presence of two well-supported clades in environmental soil samples , one from GO ( Soil Clade I ) and the other from RO ( Soil Clade II ) . The sample AR_MG16 collected from aerosol in MG also appears to be a genetic variant from P . lutzii , fallen into a paraphyletic branch in a tree . It’s worth noting that all environmental samples that clustered apart from the clinical referenced strains of both P . lutzii and P . brasiliensis are displayed in single haplotypes , reinforcing a higher diversity of the Paracoccidioides ssp . in the environment than in human hosts . This discrepancy shows that part of the environmental genotypes of Paracoccidioides spp . may not be able to infect and/or cause PCM . It is interesting to note that although Central-Western Brazil presents a prevalence of PCM caused by P . lutzii , P . brasiliensis was also detected in these areas in this work and in previous studies [30 , 31 , 42] . This observation could indicate different patterns of sporulation depending on soil constitution and weather , so that the conidia production/release of P . brasiliensis and P . lutzii could be different in distinct areas , explaining the current distribution pattern of PCM caused by both species . Previous studies have already pointed out differences in sporulation ( concerning conidia amount and morphology ) among the different cryptic species: S1 isolates produce a higher conidia amount compared to other P . brasiliensis and P . lutzii species . This observation may explain the higher isolation of S1 species ( from humans and armadillos ) in endemic areas for PCM caused by P . brasiliensis [42 , 43] . Molecular epidemiology studies of fungal pathogens are extremely important , since the genetic variability may reflect the existence of cryptic species , can have a geographic pattern and also be related to different clinical manifestations and antifungal drug response . For that reason , many efforts have been made to study the geographic limits of the different genotypes of some fungal pathogens , such as Histoplasma capsulatum , a complex of at least seven distinct clades , whose variability has been addressed by MLST ( Multi-locus sequencing type ) , as well as ITS and PRP8 intein genes as molecular markers [44–46] . Cryptic speciation event has also been recently described for another Ajellomycetacea member , Blastomyces dermatitidis , which is now considered two species , B . dermatitidis and Blastomyces gilchristii . Like P . brasiliensis and P . lutzii , these Blastomyces species seem to be sympatric in some of their distribution ranges [47] . Despite that ITS region is widely used as fungal barcoding [48] , most studies on cryptic speciation use the multi-locus sequencing approach . That is because ITS is , in some cases , useless to distinguish among very close cryptic species , such as those from the P . brasiliensis complex [31] . However P . brasiliensis and P . lutzii are clearly distinguished by ITS sequencing . Thus , our findings in fact confirm the coexistence of the P . lutzii and P . brasiliensis species ( S1 and PS2 species ) in the environment , even though the potential for human infection seems to be different for both species , depending on the region . This study presents important data regarding the eco-epidemiology of Paracoccidioides species , as well as their actual distribution over the Brazilian map . In addition , new environmental genetic variants of these pathogens should notify us about the actual scenario of the Paracoccidioides species diversity . Our results also pointed out the possible differential pathogen versus wild host interaction among Paracoccidioides species , generating new hypotheses and new work issues to be tested and studied by the scientific community . Thus , further studies based on data interacting with clinical and ecological aspects should be conducted in order to create a more reliable distribution map of this important systemic mycosis and its etiological agents . In addition , higher throughput sequencing methods should be addressed in order to get a better resolution of Paracoccidioides species diversity and distribution . | Paracoccidioides brasiliensis and Paracoccidioides lutzii are the fungal species responsible for one of the most important mycoses of Latin America , Paracoccidioidomycosis ( PCM ) . These fungi can grow in soil from forests , deforested areas , sugarcane , coffee , and rice plantations , as well as pasturelands , and they are strongly associated to armadillo burrows , which can explain their frequent isolation from this mammal’s tissues . The environmental detection of these pathogens in endemic and non-endemic areas of PCM is important for mapping risk areas , as well as for understanding the infection ability and clinical manifestations of these fungi . These pieces of information are not provided by isolates obtained from human patients , because these fungi have a long latency period and the human host can migrate , leading to a misinterpretation of the actual geographic distribution of these pathogens . By using two different molecular methodologies ( Nested PCR and in situ fluorescence ) , we detected both species of P . brasiliensis and P . lutzii in soil and in aerosol samples , even in areas where PCM is only associated to one of these two species . These data might indicate different habitat maintenance strategies between the species , which means that the infection ability may change according to the climatic and soil conditions . Despite contributing new information about the ecology of these important fungal pathogens , our molecular approach for the environmental detection of Paracoccidioides species may also be applied for their detection and differentiation in clinical samples , improving the diagnosis of this important systemic mycosis . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"in",
"situ",
"hybridization",
"molecular",
"probe",
"techniques",
"pathogens",
"microbiology",
"vertebrates",
"animals",
"mammals",
"xenarthra",
"fungi",
"materials",
"science",
"spe... | 2016 | Environmental Mapping of Paracoccidioides spp. in Brazil Reveals New Clues into Genetic Diversity, Biogeography and Wild Host Association |
A relatively small number of signals are responsible for the variety and pattern of cell types generated in developing embryos . In part this is achieved by exploiting differences in the concentration or duration of signaling to increase cellular diversity . In addition , however , changes in cellular competence—temporal shifts in the response of cells to a signal—contribute to the array of cell types generated . Here we investigate how these two mechanisms are combined in the vertebrate neural tube to increase the range of cell types and deliver spatial control over their location . We provide evidence that FGF signaling emanating from the posterior of the embryo controls a change in competence of neural progenitors to Shh and BMP , the two morphogens that are responsible for patterning the ventral and dorsal regions of the neural tube , respectively . Newly generated neural progenitors are exposed to FGF signaling , and this maintains the expression of the Nk1-class transcription factor Nkx1 . 2 . Ventrally , this acts in combination with the Shh-induced transcription factor FoxA2 to specify floor plate cells and dorsally in combination with BMP signaling to induce neural crest cells . As development progresses , the intersection of FGF with BMP and Shh signals is interrupted by axis elongation , resulting in the loss of Nkx1 . 2 expression and allowing the induction of ventral and dorsal interneuron progenitors by Shh and BMP signaling to supervene . Hence a similar mechanism increases cell type diversity at both dorsal and ventral poles of the neural tube . Together these data reveal that tissue morphogenesis produces changes in the coincidence of signals acting along orthogonal axes of the neural tube and this is used to define spatial and temporal transitions in the competence of cells to interpret morphogen signaling .
A large array of distinct cell types is generated during embryonic development in response to a relatively small number of inductive signals . A mechanism to explain this was described by C . H . Waddington in his influential book “Organizers and Genes” [1] . In this work he proposed that the specification of cell identity resulted from an interplay between “evocators , ” extrinsic inductive signals , and the specific intrinsic response of the tissue to the inductive signal , which he termed “competence . ” In this view inductive signals initiate cellular differentiation but the fate induced by the signal is intrinsic to the responding cell . Thus temporal shifts in a cell's competence provide a means to increase the diversity of cell types induced while maintaining control over the pattern in which they are generated . One example where this is relevant is the development of the vertebrate nervous system . In the spinal cord , this involves the well-ordered generation of a large variety of molecularly distinct cell types including the neurons that process sensory information and control motor movement and the migratory neural crest cells ( NCCs ) that form the peripheral nervous system [2]–[6] . The ventral part of the spinal cord contains motor neurons ( MNs ) and interneurons ( V0–V3 ) as well as the morphologically distinct nonneuronal cells of the floor plate ( FP ) [6] . These cell types are produced from domains of progenitors arrayed along the dorsal ventral axis , each of which is defined by the expression of transcription factors including Olig2 ( pMN ) , Nkx2 . 2 ( p3 ) , and Arx ( FP ) [7]–[10] . By contrast , NCCs and dI1–dI3 interneurons [3] are produced in the dorsal neural tube . Similar to the ventral neural tube , the progenitors of these cell types can be distinguished by their distinct gene expression programmes—Snail2 and Sox10 in NCCs and Olig3 in dI1–3 progenitors [11]–[13] . The stereotypic organization of neural tube cell types depends on secreted factors . Sonic Hedgehog ( Shh ) , emanating from the FP and the underlying notochord , is involved in patterning the ventral neural tube [14] . The dorsal neural tube is patterned by a distinct set of signals , prominent amongst these are members of the TGFβ family [15] . Several studies indicate that both dorsal and ventral signals function as morphogens to regulate differential gene expression in a graded manner [16]–[18] . Nevertheless a simple morphogen mechanism does not appear sufficient to explain the entirety of cell diversity produced by these factors . Importantly , the time at which cells are exposed to Shh or BMP has a significant influence over the cell types generated . For example , the induction of FP cells , which are situated in the most ventral part of the neural tube , require exposure to Shh at an early developmental time point [10] , [19] . Accordingly , progenitors exposed to similar amounts of Shh but at later developmental times differentiate into p3 progenitors of V3 neurons instead of FP cells [10] , [19] . Likewise , the differentiation of NCCs depends on the time-specific exposure to dorsal signals [20] . Neural cells exposed to BMP4/7 at early time points differentiate into the NCCs , whereas neural cells exposed to the same signals at later time points differentiate into the dorsal interneurons [20] . How neural cells change their competence to inductive signals over developmental time is unclear . It is notable , however , that when first generated in the posterior neural plate , neural progenitors are exposed to FGF signaling , but as development proceeds axis elongation interrupts FGF signaling and progenitors are exposed to retinoic acid ( RA ) secreted from the adjacent somites [21] . The switch from FGF to RA signaling has been suggested to control the timing of neuronal differentiation in the spinal cord [22] , [23] . Moreover , the repression by FGF signaling of Pax6 , Irx3 , and other transcription factors expressed in neural progenitors has been suggested to contribute to the maintenance of the undifferentiated state [21] . Whether this state provides cells with the competence to generate FP and NCC in response to appropriate inductive signals is unclear . Here we investigate the shift in generation from FP to p3 and from NCCs to dorsal interneurons to identify the mechanisms responsible for the change in competence . We provide evidence that FGF signaling , in early neural progenitors , provides cells with the competence to differentiate into FP and NCCs in response to Shh and BMP , respectively . Furthermore , we find that Nkx1 . 2 , a NK-1 transcription factor , which is regulated by FGF signaling [24] , mediates this competence and represses the expression of Pax6 and Irx3 [8] , [21] , [25] . In the case of FP , the coincidence of Nkx1 . 2 with Shh-induced FoxA2 expression defines the domain in which the FP will differentiate . Subsequently , axis elongation and the ensuing decline in FGF signaling result in the down-regulation of Nkx1 . 2 expression . This then allows cells to generate ventral and dorsal interneuron progenitors in response to Shh and BMP signaling . Hence the dynamics of cell movement drive temporal changes in signaling and gene expression in neural progenitors and these in turn control the transcription network that determines the intrinsic competence of cells to respond to morphogens acting along the orthogonal axis . Together the data reveal a molecular mechanism in which the interplay between cell competence and inductive signals increases the diversity cell types in the neural tube and determines their pattern of generation .
We previously showed that early exposure to Shh is required for neural progenitors to induce FP , characterized by Arx expression ( Figure 1A ) [10] . Forced expression of Shh , by in ovo electroporation , in the early [Hamburger Hamilton ( HH ) stage 9] [26] neural tube resulted in the broad ectopic induction of FP 48 h posttransfection ( hpt ) both in vivo ( 12/15; Figure 1A , B ) and in vitro ( Figure S1 ) . By contrast , forced expression of Shh later ( HH stage 12 ) did not induce ectopic FP ( 0/10; Figure 1C , D ) . Instead , progenitors expressed Nkx2 . 2 , characteristic of p3 neural progenitors that are normally situated in a progenitor domain dorsal and adjacent to the FP ( Figures 1A′ , B′ , C′ , D′ and S2B , C ) were induced . Moreover , longer incubation ( 72 h ) of embryos transfected at HH stage 12 with Shh did not lead to induction of Arx ( 0/8; Figure S2G , H′ ) . Similar results were obtained assaying Nato3 [27] and Nkx6 . 1 [28] , which are expressed in FP and p2–p3 domains , respectively ( unpublished data ) . Thus , neural progenitors lose their competence to generate FP in response to Shh between HH stage 9 and HH stage 12 . A similar change in competence was observed in ex vivo experiments . Intermediate [i] neural plate explants from HH stage 10 embryos treated with 4 nM Shh for 48 h expressed Arx in most of the cells ( Figure 1E , F , F′ , I ) [10] . By contrast , explants that were incubated in the absence of Shh for 12 h before the addition of 4 nM Shh induced little if any Arx expression ( Figure 1E , G , I ) . Instead Nkx2 . 2 expression was maintained in these explants ( Figure 1G′ , I ) . A longer culture , up to 72 h , did not change the expression profile ( unpublished data ) . The timing of the change in FP competence led us to focus on signals present in the neural plate of HH stage 10 embryos . To this end we tested the function of FGF [21] , Wnt , and RA [29] signaling in HH stage 10 [i] explants . Culturing [i] explants in the presence of Wnt or the RA inhibitor BMS493 for 12 h before replacing with media containing 4 nM Shh did not result in the induction of FP ( see Materials and Methods; unpublished data ) . By contrast , FP gene induction was observed if [i] explants were transiently exposed to 5 nM bFGF for 12 h in the absence of Shh and then transferred to 4 nM Shh for 48 h ( Figure 1E , H , I ) . Moreover , additional markers of the FP identity including Nato3 [Figure 1J ( j1 ) ] [27] , Shh [Figure 1J ( j2 ) ] [10] , HES1 [30] , and FoxP2 [31] were restored by this treatment ( Figure 1J; unpublished data ) . In contrast , expression of the p3 marker Nkx2 . 2 decreased [Figure 1H′ , I , J ( j3 ) ] . The expression of FoxA2 was similar in early and FGF-treated conditions ( Figures 1J ( j4 ) and S2D–F ) . FGF on its own did not induce Arx expression ( unpublished data ) at 48 h , nor did FGF induce Shh gene expression at 12 h [green arrow in Figure 1J ( j2 ) ] or Shh signaling ( green arrow in Figure S2I ) , indicating that FGF on its own is not sufficient to induce FP identity . We next asked whether FGF promotes FP differentiation in the presence of Shh in vivo . Surprisingly , the sustained expression of FGF8b in vivo , using in ovo electroporation , eliminated Arx expression and resulted in the ventral expansion of Nkx2 . 2 expression ( 5/5; Figure S2J–K′ ) . In light of the posteriorly restricted expression of FGF8 in vivo [21] we speculated that transient FGF signaling might be necessary for FP differentiation . To test this hypothesis we took advantage of a regulatable expression system to activate the FGF pathway in vivo for a limited time period [32] . We transfected a constitutively active version of the FGF-activated MAP kinase , MKK1 ( HA-MKK1-SD ) [33] , under the control of a Tamoxifen-regulated Gal4 transactivator ( ER-Gal4-VP16 ) . Following electroporation at HH stage 8+ , we stimulated HA-MKK1-SD using Tamoxifen for 8 h at HH stage 10 . The drug was then thoroughly washed out , and the embryos were cultured for an additional 40 h ( Figure 1K , M , M′ ) . This resulted in a transient up-regulation of luciferase activity ( Figure S2L ) and HA-MKK1-SD expression ( Figure S2M–Q′ ) . Moreover transient activation of MKK1 resulted in an expansion of Arx expression [6/8; Figure 1K ( ii ) , M , M′] . This suggested that transient , but not sustained , FGF signaling prolonged the competence period for FP induction and this facilitated the expanded FP induction in response to the increasing amplitude of the Shh gradient . Embryos without Tamoxifen did not show an expansion of Arx [0/6; Figure 1K ( i ) , L , L′] , and sustained treatment with Tamoxifen abolished FP induction , consistent with the unregulated activation of FGF [6/8; Figures 1K ( iii ) , N , N′ and S2J–K′ , W , W′ , Z , Z′ ) . Together these data indicate that transient exposure of neural progenitors to FGF provides competence for neural progenitors to differentiate into the FP in response to Shh . We next asked whether FGF/MAPK activity is necessary for FP induction in vivo . First , we in ovo electroporated a dominant-negative FGFR1 in which the intracellular portion of the protein containing the kinase signaling domain is truncated [34] . We targeted the posterior region of HH stage 8 embryos , comprising the preneural tube and stem zone [29] , [35] . Assaying embryos 48 hpt revealed Nkx2 . 2 expression in place of Arx ( 5/6; Figure 2B , B′ , b1–b3 ) . Transfection of a GFP control construct did not disrupt FP formation ( 0/10; Figure 2A , A′ , a1–a3 ) . Next , to inhibit FGF signaling downstream of the receptor , we transfected HH stage 8 embryos with MAP Kinase Phosphatase 3 ( MKP3; also known as DUSP6 ) [36] , [37] , which dephosphorylates ERK1/2 and thereby inactivates the MAP kinase pathway . As a result , Arx expression was down-regulated and replaced by Nkx2 . 2 expression ( 4/6; Figure 2C , C′ , c1–c3 ) . Neither perturbation led to a significant change in total number of Arx , and Nkx2 . 2-expressing cells greatly changed , suggesting that changes were mainly due to the alteration in gene expression ( Figure 2F ) . Together these data indicate that the FGF/ERK signaling pathway is required for the induction of FP in vivo . We next assessed the spatial-temporal requirement for FGF signaling . At spinal levels of the chick central nervous system , MAP kinase is active in the caudal neural plate and stem zones [37] . Nevertheless , FGF receptors are expressed throughout the neural tube [37] , and FGF signaling has been shown to play local roles even after the neural tube is closed [25] , [38] , [39] . To test whether the requirement for FGF signaling for FP differentiation was restricted to the regions of MAP kinase activity , we electroporated FGFR1ΔC or MKP3 into the neural tube and anterior preneural tube of HH stage 10- embryos , the region flanked by the posterior 4–5 somites . In contrast to the result of blocking FGF signaling in the preneural tube ( Figure 2A–C′ ) , Arx expression remained intact 48 hpt ( 6/7; 1/7 slightly down-regulated in the case of FGFR1ΔC and 0/6 for MKP3; Figure 2D , d1–d3 , E , e1–e3 ) . This result suggests that FGF signaling is required prior to prospective FP cells entering the neural tube . We confirmed the requirement of FGF/ERK signaling for FP induction using ex vivo experiments . Explants were prepared from HH stage 9 embryos that had been in ovo electroporated with MKP3 [37] 3 h prior to the dissection and cultured for 48 h with 4 nM Shh . In these explants , FGF target genes ETV5 [37] , [40] , IL17RD/SEF [41] , [42] , and Nkx1 . 2 ( see below ) were more rapidly down-regulated than in control explants ( Figure S3A , B ) . Compared to control explants ( Figure 2G–I , M ) , MKP3 blocked the ability of Shh to induce Arx and increased the expression of Nkx2 . 2 ( Figures 2J–L′ , M and S3C–G′ ) . A similar result was obtained when FGFR1ΔC was electroporated ( Figure S3H–J , P ) . We also assayed [i] explants exposed to 4 nM Shh for 36 h that had been incubated in the presence or absence of the MEK inhibitor PD184352 [Figure 2N ( b ) ] or the FGF Receptor inhibitor SSR128129E [43] , [44] [Figure 2N ( c ) ] for the first 12 h . Both treatments resulted in increased expression of Nkx2 . 2 ( Figure 2O′ , P′ , Q′ ) at the expense of Arx ( Figure 2O , P , Q ) . Together , this series of experiments indicate that neural progenitors require exposure to transient FGF/MAPK signaling at the time when Shh signaling is initiated in order to differentiate into FP . ETV4/5 ( also known as Pea3 and Erm , respectively ) are ETS family transcription factors expressed in neural cells competent to generate FP ( see Figure 3A ) and mediate transcriptional responses to FGF signaling [37] , [40] . We therefore investigated the requirement for ETV4/5 activity in FP induction . In ovo electroporation with a dominant inhibitory version of ETV5 ( EnR-ETV5DBD ) at HH stage 8 [45] blocked Arx expression assayed 48 hpt ( 5/5 ) ( Figure S3K–L′ ) . In addition , we cultured explants electroporated with EnR-ETV5DBD in the presence of 4 nM Shh for 48 h . As a result the induction of Arx was decreased and expression of Nkx2 . 2 increased ( Figure S3M–P ) . We also examined the expression of another FP marker , Nato3 , in explants electroporated either with MKP3 , FGFR1ΔC , or EnR-ETV5DBD by quantitative reverse transcription and polymerase chain reaction ( qRT-PCR ) . Consistent with the findings from the immunohistochemistry of Arx , Nato3 expression was significantly down-regulated compared to control explants and Nkx2 . 2 expression up-regulated ( Figure 2R ) . Together , these data indicate that FGF signaling through MAPK and ETV4/5 is necessary for FP development . To better understand the molecular mechanisms determining the competence for FP induction , we surveyed the transcriptome of FP-competent and -incompetent neural progenitors . For this purpose , we performed RNA-seq on samples extracted from competent explants harvested immediately after dissection ( sample “0” ) or explants treated with FGF for 12 h ( “FGF_12 h” ) and explants that had lost competence following incubation in vitro in control medium for 12 h ( “0_12 h” ) ( Figure 3A ) . From this analysis we selected 988 genes that displayed the greatest differences in their expression levels between explants assayed at the time of dissection [Figure 3A ( i ) ] and explants cultured in vitro for 12 h [Figure 3A ( ii ) ] . The genes expressed higher at 0 h than 12 h were categorized as Group A and further stratified using the expression level of genes in explants treated with FGF for 12 h [Figure 3A ( iii ) ] . The genes that were expressed lower in 0 h than 12 h samples were categorized as Group B . Group A contained several genes related to FGF ( including ETV4 and ETV5; Figure S3K–P ) , Wnt , and Eph/ephrin signaling as well as several transcription factors ( Figure 3A , Table S2 ) . To test whether any of the Group A genes mimicked the activity of FGF signaling to prolong the competence period for FP induction , we prepared explants from embryos electroporated with a selection of these candidates ( labeled blue in Figure 3A ) . Explants were incubated in control medium for 12 h and then exposed to Shh for an additional 48 h and assayed for Arx expression [see Figure 1E ( ii ) ] . This secondary screen led us to focus our attention on the NK-type transcription factor Nkx1 . 2 ( also known as Hox3 , Sax1 ) . Nkx1 . 2 is expressed in the caudal stem zone and preneural tube in chick and mouse embryos ( Figure S4A ) [46]–[48] and is induced by FGF signaling [24] . Shh signaling was not sufficient to maintain expression of Nkx1 . 2 in HH st10 [i] explants ( Figure 3B ) [24] , [47] , [48] . Hypothesizing that Nkx1 . 2 provides the FGF-dependent competence of cells to generate FP in response to Shh signaling , we devised a system in which Nkx1 . 2 expression could be manipulated to mimic the transient expression of Nkx1 . 2 during normal FP development . We prepared explants transfected with a construct encoding Nkx1 . 2 fused to the hormone-binding domain of Glucocorticoid Receptor ( GR-Nkx1 . 2 ) . In the absence of Nkx1 . 2 activation , no Arx expression was induced [Figure 3C ( i ) , D , D′ , d1 , d2] . By contrast , transient activation of Nkx1 . 2 in explants by Dexamethasone treatment ( DEX: a Glucocorticoid analogue ) for the initial 12 h followed by an additional 48 h with Shh resulted in the induction of a substantial number of Arx-expressing cells [Figure 3C ( ii ) , E , E′ , e1 , e2 , G ( ii ) ] . However , the presence of both GFP-positive/Arx-positive and GFP-negative/Arx-positive cells suggested that Arx has been induced non-cell-autonomously as well as cell-autonomously . A similar non-cell-autonomous induction of Arx was observed following the sustained expression of Nkx1 . 2 ( Figure S4B–D′ ) . This suggested that Nkx1 . 2 induced a secreted factor ( s ) . We therefore performed qRT-PCR in explants transfected with Nkx1 . 2 and found that FGF8 and its target gene MKP3 were induced at 12 h after Nkx1 . 2 induction ( Figure 3I ) . These findings suggest FGF8 and Nkx1 . 2 form a positive feedback loop and maintain the competence of cells to differentiate into FP . To test whether FGF signaling is necessary for FP competence in cells expressing Nkx1 . 2 , we repeated the 12 h activation of the Nkx1 . 2 experiment in the presence of the MAPK inhibitor PD184352 . In this condition , Arx continued to be induced and these cells were derived from cells that had expressed Nkx1 . 2 [Figure 3C ( iii ) , F , F′ , f1 , f2 , G ( iii ) ] . Other FP markers were also induced in this experimental regime [Figure 3C ( iii ) ] , as examined by qRT-PCR ( Figure 3H ) . The expression of Nkx2 . 2 expression was reciprocal to Arx in these explants ( Figure 3D′ , d1 , d3 , E′ , e1 , e3 , F′ , f1 , f3 ) . These data indicate that transient expression of Nkx1 . 2 immediately followed by exposure to Shh is sufficient to reconstitute FP induction even when FGF signaling is blocked . We next tested whether Nkx1 . 2 is required for FP differentiation . Nkx1 . 2 contains a Groucho-binding domain and appears to act as a transcriptional repressor [49] , [50] . We therefore hypothesized that an activator variant—Nkx1 . 2DBD-VP16—would function as a dominant negative . Consistent with this , forced expression of Nkx1 . 2DBD-VP16 repressed Arx ( 7/8 ) and Nato3 ( 6/8 ) expression in vivo ( Figure 3J , J′ and unpublished data ) and promoted the ventral expansion of Nkx2 . 2 ( 7/8; Figure 3K , K′ ) . In addition , in [i] explants expressing Nkx1 . 2DBD-VP16 , Arx induction by Shh was blocked cell-autonomously and cells expressed Nkx2 . 2 instead ( Figure 3L–N ) . Taken together , these findings suggest that Nkx1 . 2 and/or a closely related factor ( s ) is necessary for FP induction . FGF signaling has been shown to inhibit the expression of several neural progenitor expressed transcription factors , including Pax6 and Irx3 [Figures S2R–T′ and S4K ( iii ) ] and to perturb RA signaling ( Figure S5A ) [19] , [21] , [25] . The mRNA sequencing data were consistent with these findings ( Group B in Figure 3A ) . We therefore asked whether Nkx1 . 2 activity was responsible for this repression . Electroporation of the dominant-negative Nkx1 . 2DBD-VP16 up-regulated expression of Pax6 and Irx3 ( Figure 4A–F ) without affecting FGF8 expression ( Figure S5B–C′ ) . This suggested that Nkx1 . 2 mediates the repressive activity of FGF signaling . On the other hand , forced expression of either Irx3 , Pax6 , or a constitutive-active RA receptor ( RAR-VP16 ) in HH stage 8+ embryos repressed FP differentiation ( Figures 4G–I′ and S5D–I′ ) at 48 hpt . We therefore speculated that blocking the combined activity of these factors would prolong the competence of cells to generate FP . To test this , we electroporated , either alone or in combination , the dominant-negative versions of Irx3 ( Figure S5J–K′ ) , Pax6 [51] , and RA Receptor RAR-403 [25] . Forced expression of individual constructs was not sufficient to induce FP ( unpublished data ) . Electroporation of RAR-403 induced the expression of Nkx1 . 2 [Figure 4J ( j1 ) ( v ) ] , whereas the expression of FoxA2 was induced by dominant-negative Irx3 and Pax6 ( and their combination ) [Figure 4J ( j2 ) ( ii ) ( iii ) ( iv ) ] . Strikingly , the combined expression of Irx3DBD-VP16 or Pax6PD-EnR with RAR-403 maintained the competence of cells to differentiate into FP both in vivo ( 6/8; Figure 4K–N′ ) and in vitro ( Figure 4O–S′ ) . Consistent with this , the FP markers Nato3 and FoxJ1 were also substantially induced when RAR-403 was transfected together with either Irx3DBD-VP16 or Pax6PD-EnR ( Figure S5L ) . These results suggest that Nkx1 . 2 maintains FP competence by repressing the expression of transcription factors normally found in more mature neural progenitors . To extend these findings , we investigated if FP differentiation also requires the blockade of RA signal in another species . For this , we took advantage of the differentiation of mouse ES cells into Arx-expressing FP cells . ES cells , cultured in serum free media , expressed FGF8 and Nkx1 . 2 ( Figure S6E ) , and treatment with Shh for 60 h generated FP , as evidenced by the expression of Arx , FoxA2 , and Nato3 ( Figure S6A , B , D ) . In contrast , the differentiation of FP was substantially reduced in ES cells exposed to RA in addition to Shh , and instead Nkx2 . 2-expressing p3 cells were generated ( Figure S6A , C , D ) . In these assays , the addition of RA inhibited FGF8 and Nkx1 . 2 expression and up-regulated Pax6 expression ( Figure S6E ) . These findings are consistent with the idea that the blockade of RA is required for the FP differentiation in mouse as well as chick . The Shh target gene FoxA2 is critical for FP induction [52] . Cells lacking FoxA2 fail to form FP ( Figure S7A ) , and FoxA2 expression is initiated in prospective FP soon after neural induction and is maintained during FP differentiation [53] ( Figure 5V ) . This prompted us to investigate whether the induction of FP by FoxA2 was also dependent on the competence of neural progenitors . Forced expression of FoxA2 in the posterior neural tube at an early time point ( HH stage 11 ) induced Arx by 48 hpt , consistent with previous studies [10] ( 10/10; Figure 5A , A′ ) . FoxA2 also induced its target gene Shh , and therefore Nkx2 . 2 was induced non-cell-autonomously ( 10/10 for each; Figures 5B–C′ and FS7G , G′ for negative control ) . By contrast , transfection of FoxA2 at a later time ( HH stage 14 ) was not able to induce Arx ( 6/8;2/8 had sporadic expression; Figure 5D , D′ ) , although ectopic expression of Shh and Nkx2 . 2 were still induced ( 7/8 each; Figure 5E–F′ ) . Notably the induction of FP by FoxA2 was independent of Shh signaling , as the coelectroporation of PtcΔ , which inhibits the Shh signaling pathway [16] , with FoxA2 did not abrogate the induction of ectopic FP ( n = 8; Figure S7B–C″ ) . A requirement for early FoxA2 expression was also evident in explants . We constructed an inducible FoxA2 ( an estrogen-receptor-fused FoxA2; ER-FoxA2; Figure 5G ) and activated it for different time periods in explants prepared from transfected embryos . Induction of FoxA2 at the time explants were dissected and induced in Arx [Figures 5G ( i ) ( ii ) , H–I′ and S7H ) . However , when the induction of FoxA2 was delayed for 12 h after dissection , it failed to induce FP [Figure 5G ( iii ) , J , J′] . The ability of FoxA2 to induce FP after 12 h in culture could be restored by exposing the explants to FGF for the initial 12 h [Figures 5G ( iv ) , K , K′ and S7H] . Consistent with this , blocking FGF signaling in vivo at HH stage 11 , by transfecting MKP3 , inhibited the ectopic induction of FP cells by FoxA2 ( 6/8 inhibited; Figure 5L , L′ ) , although FoxA2 remained able to induce expression of Shh and Nkx2 . 2 ( n = 6 for each; Figure 5M–N′ ) . Thus , the timing and requirement for the competence of cells to induce FP in response to FoxA2 corresponds to the competence to induce FP in response to Shh . We next asked if Nkx1 . 2 was involved in maintaining the competence of cells to induce FP in response to FoxA2 . Forced expression of either Nkx1 . 2DBD-VP16 , Pax6 , or Irx3 with FoxA2 blocked induction of Arx , although ectopic Shh remained expressed , suggesting the FP induction by FoxA2 requires Nkx1 . 2 and the absence of Irx3 or Pax6 ( more than five embryos out of six for each; Figures 5O–Q′ and S7D–F′ and unpublished data ) . We sought to reconstitute FP induction in vitro by regulating the timing of Nkx1 . 2 and FoxA2 activity . To this end , we prepared explants from embryos coelectroporated with GR-Nkx1 . 2 and ER-FoxA2 . Explants were treated with DEX for the first 12 h , to maintain Nkx1 . 2 activity , and then media was replaced with Tamoxifen , to induce FoxA2 [Figures 5R ( vi ) , T–T′ and S7H] . This regime , but not conditions in which only GR-Nkx1 . 2 or ER-FoxA2 were activated [Figures 5G ( iii ) , J , J′ , R ( v ) , S , S′ and S7H] , resulted in the induction of Arx . Together these results suggest that the coincidence of FoxA2 and Nkx1 . 2 expression , which is determined by the intersection of Shh and FGF signaling , establishes the transcriptional code for FP induction . Finally , in order to map where the expression of Nkx1 . 2 and FoxA2 intersect in vivo , we performed whole mount in situ hybridization . Whereas Nkx1 . 2 was transiently expressed in the posterior stem and preneural tube ( Figure 5U , u1–u3 ) [24] , [46] , [48] , FoxA2 expression was initiated just anterior to Hensen's node , and continued to be expressed in midline cells and notochord ( Figure 5V , v1–v3 ) . Therefore , the midline cells anterior to the Hensen's node ( Figure 5u2 , v2 ) appear to simultaneously express , albeit at low levels , Nkx1 . 2 and FoxA2 . In light of the ex vivo data ( Figure 5K , K′ , T , T′ ) , it is highly likely that it is at this position cells acquire FP fate . Moreover the expression patterns of FGF8 ( Figure 5W , w1–w3 ) [37] and Shh ( Figure 5X , x1–x3 ) [54] are consistent with those of Nkx1 . 2 and FoxA2 , respectively . By contrast , Pax6 and Irx3 , which inhibit FP induction , are only expressed anterior to the limits of FGF8 and Nkx1 . 2 expression ( Figure 5Y , Z ) . These in vivo observations support the idea ( Figure 1E , H ) that transient FGF and subsequent Shh signaling are critical for FP differentiation , whereas Pax6 and Irx3 restrict FP differentiation . The induction of NCCs at the dorsal pole of the neural tube also depends on early exposure to inductive signals [20] . Consistent with this , ectopic BMP expression in the neural tube in vivo at early time points promoted the induction and delamination of NCCs [more than 7 embryos out of 10 for each , while Olig3 expression did not change significantly ( n = 10 ) ; Figure S8A–C′] . By contrast , later exposure to BMP favored the generation of dorsal interneurons instead ( n = 10; in all cases the expression of Sox10 and HNK1 were repressed while Olig3 expanded ventrally; Figure S8D–F′ ) [20] . This prompted us to address whether the switch in response to BMP dorsally is similar to the switch in response to Shh ventrally . NCCs express Snail2 , Sox10 , and HNK1 , whereas Olig3 is expressed in dorsal neural progenitors that generate dI1–dI3 interneurons [12] , [13] , [55] , [56] . Using these as markers , we assayed the generation of NCCs in vitro . Treatment of [i] explants with 0 . 25 nM BMP4 resulted in the induction of the Snail2 [Figure 6A ( ii ) ( iii ) , B , C , F , G] at 24 h and HNK1 and Sox10 at 36 h [Figure 6A ( ii ) ( iii ) , B″ , C″ , G] . Migratory cells were also apparent by 36 h . However , if BMP treatment was delayed for 12 h after [i] explants were placed in culture , the induction of Snail2 was lost [Figure 6A ( iv ) , D , G] . This did not appear to be due to the loss of responsiveness to BMP , as the induction of the dorsal neural progenitor marker Olig3 ( dP1–dP3 ) and the dorsal interneuron dI1 marker Lhx2 was maintained in these conditions [20] ( Figures 6B′ , C′ , D′ , F , G and S8G–I ) . Moreover , explants treated at early and late times generated comparable levels of signaling activity , assayed using a luciferase reporter with a BMP-responsive element ( Figure S8K ) [18] . Strikingly , NCC induction was restored if explants were exposed to bFGF for 12 h prior to the treatment with BMP4 [Figures 6A ( v ) , E–E″ , F , G and S8J] . Thus these data suggest that the competence to induce neural crest differentiation is determined by FGF signaling . Next we asked if the activity of FGF in the dorsal neural tube is also mediated by Nkx1 . 2 . Overexpression of Nkx1 . 2 on its own did not induce any gene expression characteristic of the neural crest [Figure S8L ( iii ) ] , suggesting the cells still require BMP signal for neural crest induction . To test for a direct effect of Nkx1 . 2 on the neural crest induction , we transiently expressed Nkx1 . 2 and blocked FGF signaling simultaneously [Figure 3C ( iii ) ] . We prepared [i] explants that had been electroporated with GR-Nkx1 . 2 and cultured these in the presence of PD184352 alone [Figure 6H ( a ) ] or together with DEX [Figure 6H ( b ) ] for 12 h . Media was then replaced with 0 . 25 nM BMP for an additional 24-h culture . Assaying NCC and dorsal neural progenitor markers revealed that the expression of Nkx1 . 2 promoted NCC induction [Figure 6H ( b ) , J–L] and blocked dI1–3 generation ( Figure 6L ) . Finally we asked if Nkx1 . 2 is necessary for the neural crest induction by expressing the dominant-negative Nkx1 . 2DBD-VP16 . This resulted in the cell-autonomous repression of Snail2 expression in vivo ( 4/6; Figure 6M–N′ ) and enhanced Pax6 expression ( 5/6; Figure 6P–Q′ ) . Moreover , the overexpression of RAR-VP16 in the dorsal area inhibited the Snail2 induction at the expense of that of Pax6 ( 6/6; Figure 6O , O′ , R , R′ ) . The electroporation of Pax6 and Irx3 also inhibited Snail2 expression ( 5/7 for Pax6 , 5/6 for Irx3; Figure S8M , M′ and unpublished data ) . Together these findings suggest that the competence of neural progenitors to generate NCCs is determined by the FGF-mediated expression of Nkx1 . 2 , via repressing the activities of RA , Pax6 , and Irx3 .
Classic embryological grafting studies provided the first evidence that a signal , later identified as Shh , produced by the notochord is responsible for FP induction [57]–[60] . These studies also found that the capacity for FP induction attenuated as neural cells matured . This loss of competence restricts the specification of FP to the ventral midline of the neural tube by limiting the homeogenetic induction of FP [57] . Our study reveals a molecular mechanism that explains these observations . The intersection of FGF and Shh signaling is restricted to regions of the neural plate immediately anterior to the regressing node ( Figure 5W , X ) . This function for FGF signaling complements its previously identified role as an inhibitor of neuronal differentiation in this region of the embryo [21] , [48] . In these regions the low levels of Shh emanating from axial mesodermal cells mean that the only Nkx1 . 2-expressing cells that receive sufficient Shh to induce FoxA2 are those in the ventral midline [10] , [61] . The function of FGF and Shh signaling in FP induction is also supported by data from the directed differentiation of ES cells to dopaminergic neurons [62] , [63] . This cell type is generated by FoxA2-expressing progenitors at the ventral midline of the midbrain , and their in vitro differentiation requires FGF signaling transiently during the period ES cells commit to a neural fate [63] , [64] . The coincidence of FGF and BMP signaling is required for NCC specification . This is in good agreement with studies that have implicated FGF signaling in the specification of NCC ( Pax7 , Zic1 , and Msx1 expression ) as early as gastrula stages [65] , [66] and the subsequent determination of neural crest fate by BMP and other signals ( e . g . , Wnt , Notch ) [67] . The functional difference between FGF , Wnt , and Notch in NCC specification remains unclear . Nevertheless , our data implicate a regulatory network at least between FGF and Wnt because FGF induces Nkx1 . 2 expression that in turn induces Wnt gene expression ( Figure S4K ) . Taken together , therefore , in both the dorsal and ventral neural tube , the intersection of anterior–posterior FGF signaling with dorsal–ventral morphogen signaling provides a spatial and developmental time window that determines the induction of the cell types characteristic of the poles of the neural tube . By contrast , a study from chick embryos suggested that ectopic expression of FGF inhibits NCC specification and emigration [56] . The lack of NCC induction in these experiments could be due to the prolonged activation of FGF signaling because our studies indicate that sustained FGF exposure inhibits NCC production ( Figure S8N–P′ ) . In this view , therefore , FGF signaling is transiently required to establish NCC competence , but its sustained activity blocks the elaboration of NCC identity [56] . FP induction also displays a similar requirement for transient FGF signaling ( Figures 1K–N′ and S2J , K ) . In vivo the transience of FGF signaling is determined by the posterior regression of the source of FGF driven by axis elongation [68] . The consequent down-regulation of FGF signaling in neural tissue as it becomes incorporated into the neural tube therefore allows the elaboration of FP and NCC identity that are specified earlier . This mechanism exploits tissue morphogenesis to coordinate progression in cell identity with the overall dynamics of the embryo's development . Loss of FGF signaling also prompts the down-regulation of Nkx1 . 2 expression and a change in the competence of progenitors not committed to FP or NCC identity . These cells now respond to the dorsal and ventral morphogens by acquiring identities of neuronal progenitors . In the ventral neural tube , the increasing levels of Shh production induce p3 identity in the cells dorsal to the FP and MN progenitors at a further distance [17] . Dorsally progenitors of dI1–3 neurons are induced by BMP signaling [20] . Taken together these data reveal how the cell movements that drive axis elongation provide a timing mechanism for changes in competence by controlling the combination of signals to which cells are exposed . This increases the diversity of the cell types generated in the neural tube and ensures their correct temporal and spatial generation . The competence to form FP and p3 progenitors appears to be mutually exclusive , as does the formation of NCC and dorsal interneurons . FGF has been shown to block both the induction of Shh-dependent neuronal subtypes in the ventral neural tube and the expression of transcription factors that define the progenitors of these neurons [21] , [25] . This does not appear to be a consequence of substantial changes in Shh signal transduction ( Figure S2I ) . Likewise BMP signaling in neural progenitors appears unaffected by FGF signaling ( Figure S8K ) . Moreover the induction of FP and NCC identity within cells receiving FGF signaling suggests that there is not a complete blockade in the specification of new cell identities . Instead , FGF signaling appears to act by regulating the expression of a set of transcription factors in neural progenitors that transform the transcriptional program induced by Shh or BMP . Our attention focused on Nkx1 . 2 , as this appeared to mediate the FGF-dependent competence for FP and NCC differentiation . In support of this , Xenopus Nkx1 . 2 gene ( Nbx ) [69] is expressed in the presumptive neural crest area and is essential for the neural crest differentiation . Nevertheless , mutation of Nkx1 . 2 in mouse embryos does not appear to affect FP or NCC generation [70] . Redundancy with Nkx1 . 1 , an Nkx1 . 2 paralogue , may explain this apparent discrepancy . Both Nkx1 . 1 and Nkx1 . 2 are expressed in similar regions of the caudal embryo and forced expression of Nkx1 . 1 had similar effects to Nkx1 . 2 ( Figure 5U and unpublished data ) . The generation of compound mutant mice lacking both genes would test this hypothesis . Alternatively there might be functional redundancy among a broader set of transcription factors in the caudal preneural tube , and it will be important to understand the function of these and the transcriptional network that connects them . A recent study has identified changes in higher order chromatin structure of specific genes as cells progress from the pre-eural tube to the neural tube [71] . How these changes are instated remains to be determined . It is possible Nkx1 . 2 regulates chromatin modifiers or factors that direct the chromatin modifiers to appropriate regions of the genome . Alternatively other targets of FGF signaling , independent of Nkx1 . 2 , could be responsible . Irrespective of the mechanism , the irreversible changes in chromatin structure might provide an explanation as to why cells that have lost their competence to differentiate into the FP do not regain it even if exposed to FGF . In this context , it is notable that in pancreatic development , the repression of Arx expression is controlled by methylation of a CpG island within the Arx gene locus [72] . This mechanism does not seem directly applicable in the neural tube , however , because neural explants treated with 5-aza-dC , a DNA demethylating agent , did not alter the expression of key patterning genes ( unpublished data ) . Identification and detailed analysis of the regulatory regions and epigenetic marks will be necessary to explore the relevant mechanism further . Expression of Nkx1 . 2 repressed expression of neural progenitor transcription factors , including Pax6 and Irx3 ( Figure S4E–K ) . Conversely , our experiments and previously published studies indicate that Pax6 , Irx3 , and RA signaling inhibit FP and/or NCC differentiation [19] , [21] , [73] , while promoting the establishment of neuronal progenitor identity . Although nonautonomous effects of Nkx1 . 2 might contribute , cell-autonomous mutual cross-repression between alternative transcription states is a reoccurring theme in developmental decisions and appears to be the most likely explanation for the spatial and temporal transition between the different competence states . Indeed cross-repressive interactions are apparent between the transcription factors that determine distinct progenitor domains along the dorsal–ventral axis of the neural tube [74] . Thus there appears to be a common logic that underlies the transcriptional mechanisms along both the dorsal–ventral and rostral–caudal axis of the neural tube . It is notable that as well as being induced by FGF signaling , Nkx1 . 2 also promotes the expression of FGF ( Figure 3I ) . This establishes a positive feedback loop that supports the FP/NCC competence state . This is reminiscent of the positive feedback loop between FoxA2 and Shh expression that is characteristic of the FP itself . In both cases the feedback loop functions to repress the expression of Pax6/Irx3 in a cell-autonomous manner ( Figure S4L–O″ ) and must be interrupted in order to limit the homeogenetic induction of FP cells [57] , [58] . In the case of the FoxA2-Shh loop , the change in progenitor competence mediated by the down-regulation of FGF signaling is responsible for ending the feedback loop . In the case of the FGF-Nkx1 . 2 loop , it seems likely that RA signaling terminates the positive feedback . RA emanating from somites adjacent to the maturing neural tube forms a rostral to caudal gradient in both neural tissue and paraxial mesoderm that counteracts the posteriorly produced FGF [21] . Consistent with this , a dominant-negative RAR effector is sufficient to induce Nkx1 . 2 expression in cells that would otherwise have down-regulated its expression [Figure 4J ( v ) ] . Thus axis elongation not only results in the posterior regression of the source of FGF but also the exposure of cells to RA signaling [21] . Once RA starts to be produced in the somites , the inductive effect on Pax6/Irx3 by RA overcomes the repressive effect of Nkx1 . 2 on their expression and cells change the competence in response to Shh or BMP , and this promotes the transition in competence , adding a further level of spatial and temporal control over the transition ( Figure 7 ) . The details of the transcriptional network controlled by Nkx1 . 2 that acts to induce FP or NCC fate in response to Shh or BMP signaling remains to be fully elucidated . The set of transcription factors responsible for defining premigratory NCC identity is known in some detail , and it will be interesting to examine how Nkx1 . 2 influences this network [75] . For FP differentiation the situation is not as clear . Our data suggest that FP is specified within 12 h of the initiation of FGF and Shh signaling . Nevertheless , it takes more than 30 h before expression of mature FP genes such as Arx , Nato3 , and HES1 ( Figure 1J and unpublished data ) . We anticipate that transcriptional mechanisms must relay the immediate targets of FGF signaling ( Nkx1 . 2 ) and Shh signaling ( FoxA2 ) to regulate the mature FP genes . Taken together , the current study provides new insight into how the interplay between cellular competence and inductive signals controls pattern formation and increases cell type diversity in the neural tube . A striking feature of this mechanism is that it combines the morphogenetic movements of the developing embryo with signals acting along orthogonal axes to position and time transitions in competence . The dynamics of these interactions offer a means to couple changes in the response of individual cells to the overall development of the embryo . More generally , the increasingly detailed knowledge of the gene regulatory network underpinning these events makes the neural tube a good model in which long established developmental concepts , such as competence and inductive signals , can be understood in mechanistic terms .
The Ensembl database ( http://www . ensembl . org/index . html ) and National Center for Biotechnology Information ( NCBI; http://www . ncbi . nlm . nih . gov ) accession numbers are as follows: chicken Arx ( ENSGALG00000025770 ) , chicken Nkx6 . 1 ( AF102991 ) , chicken Olig2 ( AF411041 ) , and chicken FoxP2 ( ENSGALG00000009424 ) . Refer to Table S1 for the other genes . All animal experiments were performed under a UK Home Office project license within the conditions of the Animals ( Scientific Procedures ) Act 1986 . All authors are personal license holders , and this study was performed under the project license PPL80/2528 , approved by the Animal Welfare and Ethical Review Panel of the MRC-National Institute for Medical Research . Unless otherwise stated , in ovo electroporation experiments were performed using the pCIG expression plasmid , which contains an IRES-GFP gene downstream of the gene of interest [76] . For the overexpression experiments of Shh , pCX-Shh-N ( an expression construct producing the amino-terminal region of Shh ) was used [77] . Electroporation was performed at the indicated stages using an ECM-830 electroporator ( BTX ) . For the early stage electroporations , DNA was applied with a glass capillary onto the open neural plate and electric pulses given from dorsal to ventral . Otherwise the DNA was placed in the lumen of the neural tube and electroporation was performed laterally . At the indicated time points , embryos were fixed with 4% paraformaldehyde ( PFA ) , subsequently treated with 15% of sucrose , embedded in gelatin , and 14 µm sections taken . The antibodies used in this study were against Arx ( rabbit , a gift from J . Chelly ) [78] , Irx3 ( rabbit , a gift from T . M . Jessell ) [25] , Nkx2 . 2 ( mouse , DSHB 74 . 5A5 ) , Olig2 ( rabbit , Millipore AB9610 ) and Pax6 ( rabbit , Millipore AB2237 ) , Olig3 ( mouse , Abcam ab168573; rabbit , SIGMA HPA018303 ) , Sox10 ( rabbit; Abcam ab27655 ) , Snail2 ( rabbit; Cell Signaling Technology 9585 ) , HNK1 ( mouse; BD 347390 ) , GFP ( sheep , Biogenesis 4745-1051; rabbit , Invitrogen A11120 ) , haemagglutinin ( rabbit , SIGMA H6908 ) , FoxA2 ( goat , Santa Cruz sc-6554X ) , FoxP2 ( rabbit , Abcam ab16046 ) , and Nato3 ( rat , a gift from Y . Ono ) [27] . For in situ hybridization of chick and mouse embryos , embryos were harvested at the indicated stages and fixed with 4% paraformaldehyde ( PFA ) overnight . Antisense RNA probes were synthesized with Digoxigenin ( DIG; Roche ) , and hybridization was performed at 70°C in the solution containing 1 . 3×SSC ( saline-sodium citrate ) pH 5 . 0 , 5 mM EDTA , 1 mg/ml torula yeast RNA ( SIGMA ) , 0 . 2% Tween 20 detergent ( SIGMA ) , 0 . 5% CHAPS detergent ( SIGMA ) , 100 µg/ml Heparin sodium salt , and 50% Formamide . Signals were developed by BM-Purple ( Roche ) . In situ hybridization on chick sections was performed as described previously [79] . The hormone-binding domain of the estrogen receptor ( ER ) was fused to Gal4-VP16 to make pCIG-ER-Gal4-VP16 [32] . The target gene HA-MKK1SD ( a haemagglutinin-tagged constitutive-active MKK1; [33] ) was placed under the control of 14 concatemerized Gal4 binding sequences ( the upstream activating sequences; UAS ) . For transient in ovo expression , pCIG-ER-Gal4-VP16 and 14×UAS-HA-MKK1SD were electroporated at HH stage 8 . For the induction of the target gene , Pluronic gel F-127 ( SIGMA ) was prepared to 20% ( w/v ) in Hank's Balanced Salt Solution ( HBSS; SIGMA ) , and 4-Hydroxytamoxifen ( 4-OHT; SIGMA ) was added to 50 µM . The gel was placed between the vitelline membrane and the embryo . After 8 h the gel was washed out with HBSS and the embryos were incubated until the indicated time point . For the sustained expression , the gel was replenished every 12 h . pCIG-ER-FoxA2 was generated by fusing the same ER domain to the coding region of mouse FoxA2 gene , and pCIG-GR-Nkx1 . 2 was generated by fusing the hormone-binding domain of human Glucocorticoid Receptor ( also known as NR3C1 ) [80] to the coding region of mouse Nkx1 . 2 gene . Explant assays were performed as described previously [81] , [82] . Briefly , explants were prepared from the posterior neural epithelial layer of HH stage 9+ embryos and maintained in Leibovitz ( L-15 ) medium ( Gibco ) during the preparation , embedded in a drop of collagen ( SIGMA ) buffered with DMEM ( SIGMA ) , and preincubation was performed under 5% CO2 and 37°C for 1 h to allow the collagen drop to harden . Explants were then cultured with F-12/Ham ( Gibco ) containing Glutamine supplemented with antibiotics ( 50 U/ml of Penicillin , 50 µg/ml of Streptomycin ) and Mitoserum ( BD ) . Mouse recombinant FGF2 was purchased from R&D , and mouse Shh ( C25II ) [54] , [83] was produced in house . In addition , in Figure 1E ( iii ) , we tried DKK ( R&D ) to inhibit Wnt , BMS 493 [84] , and 4-Diethylaminobenzaldehyde ( DEAB ) [85] to inhibit RA and valproate as a Notch activator [86] . In Figure 2N–Q′ , FGF signal was inhibited by PD184352 ( SIGMA ) or SSR128129E ( Selleckchem ) [43] , [44] . In these experiments , explants were prepared in L-15 in the presence of either of the inhibitors . Inhibitors were also added to the collagen . Explants were fixed with 4% PFA for 1 h and stained with antibodies as indicated . Data were collected with an SP2 confocal microscope ( Leica ) , and all explant images are 125 µm per side , except in Figure 6B , B″ , C , C″ , D , D″ , E , E″ , which are 375 µm . Quantitation was performed on at least three areas , each of which contained approximately 200–250 cells , randomly chosen from the explants . Data are presented as mean values ± s . e . m . The reporter constructs used in this study were as follows: the GBS-Luc reporter construct ( the Firefly Luciferase gene driven by 8 concatermized Gli-binding sites ) [87] , the 14×UAS-luc ( the luciferase gene controlled by 14×UAS sequences ) , the RARE-luc ( a construct with a triple repeat of the RA Responding Element; obtained from Addgene ) [88] , and the BRE-luc ( a gift from P . Ten Dijke ) [18] . In ovo reporter assays were performed by co-electroporating the reporter together with pRL-CMV ( Promega; used as the normalization control ) . The anterior thoracic levels of the embryos were harvested at indicated time points , and luciferase assays were performed following the manufacturer's instruction ( Dual luciferase assay kit; Promega ) using a Luminometer 9509 ( Berthold ) . The relative induction levels of the luciferase were calculated by comparison with control electropotation . At least five embryos were assayed at each condition and were represented as mean values ± s . e . m . For the reporter assays in explants , five explants were pooled for each measurement , and four pools were assayed for each condition . Relative induction levels were calculated compared to the luciferase activity in unstimulated control explants . RNA was extracted from a pool of 20–30 explants using Picopure RNA extraction kit ( LifeTechnologies ) to obtain 500 ng–1 µg of total RNA . cDNAs were synthesized by Superscript II reverse transcriptase ( Invitrogen ) , and qRT-PCR analyses were performed by 7900 HT Fast Real-Time PCR system ( Applied Biosystems ) . Where possible , the primers were designed to cross an intron by referring to Ensembl database in order to avoid amplifying any contaminating genomic DNA . Sequences of primers are presented in Table S1 . Each gene expression level was normalized to that of β-actin . Each data point contains at least two biological replicates and is presented as the mean values ± s . e . m . Nkx1 . 2DBD-VP16 and Nkx1 . 2DBD-EnR were made by fusing the transcription activator domain of the herpes simplex virus VP16 or drosophila Engrailed repressor domain and the DNA-binding domain of mouse Nkx1 . 2 ( amino acid numbers 131–218 ) . Likewise , the DNA binding domain of mouse Irx3 ( amino acid numbers 131–200 ) and mouse ETV5 ( amino acids numbers 290–489 ) [45] were used for Irx3DBD-VP16 , Irx3DBD-EnR , and EnR-ETV5DBD . A nuclear localization signal was added to Irx3 and ETV5 constructs . The construction of FGFR1ΔC [34] , Pax6PD-EnR [51] , RAR-VP16 , RAR-403 [25] , FoxA2-EnR [10] , and PtcΔ [16] were described previously . Libraries were synthesized with TruSeq™ RNA Sample Preparation kit according to the manufacturer's instruction ( Illumina ) . Fifty-five base-pairs paired-end sequencing was performed on a HiSeq 2000 ( Illumina ) . From 50 to 100 million clusters were obtained from each sample . The sequence data were mapped using Bowtie [89] to the 16 , 832 chick cDNAs annotated in the Ensembl database . Sequences mapped to each gene were then counted and each sample normalized to the total number of reads in the sample . The standard deviation of each gene across the samples was calculated , and the 988 genes that had the standard deviations of more than 0 . 5 were selected . These were divided into two groups: those with higher expression at time 0 than that in 0 nM_12 h ( 416 genes ) and a second group with lower expression at time 0 than in 0 nM_12 h ( 572 genes ) . The genes that restored by FGF ( 263 genes; Group A ) and the genes repressed by FGF ( 364 genes; Group B ) by up to 10-fold were selected for further study . The list of the categorized genes is available in Table S2 . The original sequence data have been deposited in ArrayExpress ( http://www . ebi . ac . uk/arrayexpress; E-MTAB-2393 ) . Mouse ES cells were maintained on feeders in LIF-supplemented medium . Cells were differentiated using a monolayer differentiation protocol with minor modifications [90] . Briefly , feeder-depleted ES cells were plated at high density on gelatin-coated CellBind dishes ( Corning ) and maintained in N2B27 medium . For FP differentiation , recombinant 2 µg/ml Shh was added to the medium at day 3 . 5 and cells were cultured for an additional 60 h . For p3 differentiation , 30 nM RA ( SIGMA ) was added at day 3 and replaced each day with media containing 30 nM RA and 2 µg/ml Shh for an additional 60 h . In qRT-PCR , the expression values of each gene were normalized to that of RhoA , which is expressed at the same level throughout the differentiation . | During embryonic development different cell types arise at different times and places . This diversity is produced by a relatively small number of signals and depends , at least in part , on changes in the way cells respond to each signal . One example of this so-called change in “competence” is found in the vertebrate spinal cord where a signal , Sonic Hedgehog ( Shh ) , induces a glial cell type known as floor plate ( FP ) at early developmental times , while the same signal later induces specific types of neurons . Here , we dissected the molecular mechanism underlying the change in competence , and found that another signal , FGF , is involved through its control of the transcription factor Nkx1 . 2 . In embryos , Shh and FGF are produced perpendicular to one another and FP is induced where the two signals intersect . The position of this intersection changes as the embryo elongates and this determines the place and time FP is produced . A similar strategy also appears to apply to another cell type , neural crest . In this case , the intersection of FGF with BMP signal is crucial . Together the data provide new insight into the spatiotemporal control of cell type specification during development of the vertebrate spinal cord . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biology",
"and",
"life",
"sciences",
"cell",
"fate",
"determination",
"developmental",
"biology"
] | 2014 | Integration of Signals along Orthogonal Axes of the Vertebrate Neural Tube Controls Progenitor Competence and Increases Cell Diversity |
The Haemophilus influenzae HMW1 adhesin is a high-molecular weight protein that is secreted by the bacterial two-partner secretion pathway and mediates adherence to respiratory epithelium , an essential early step in the pathogenesis of H . influenzae disease . In recent work , we discovered that HMW1 is a glycoprotein and undergoes N-linked glycosylation at multiple asparagine residues with simple hexose units rather than N-acetylated hexose units , revealing an unusual N-glycosidic linkage and suggesting a new glycosyltransferase activity . Glycosylation protects HMW1 against premature degradation during the process of secretion and facilitates HMW1 tethering to the bacterial surface , a prerequisite for HMW1-mediated adherence . In the current study , we establish that the enzyme responsible for glycosylation of HMW1 is a protein called HMW1C , which is encoded by the hmw1 gene cluster and shares homology with a group of bacterial proteins that are generally associated with two-partner secretion systems . In addition , we demonstrate that HMW1C is capable of transferring glucose and galactose to HMW1 and is also able to generate hexose-hexose bonds . Our results define a new family of bacterial glycosyltransferases .
Glycosylation of proteins is an essential process that plays an important role in protein structure and function and represents a strategy to fine tune cell-cell recognition and signaling . For a long period of time , glycosylation of proteins was believed to be restricted to eukaryotes . However , in recent years glycoproteins have been identified increasingly in prokaryotes as well , including pathogenic bacteria such as Pseudomonas aeruginosa , Campylobacter spp . , Neisseria spp . , and E . coli , among others [1]–[10] . Nonencapsulated ( nontypable ) Haemophilus influenzae is a human specific pathogen that is a common cause of localized respiratory tract and invasive disease and initiates infection by colonizing the upper respiratory tract [11] , [12] . Approximately 75–80% of isolates express two related high-molecular weight proteins called HMW1 and HMW2 that mediate high-level adherence to respiratory epithelial cells and facilitate the process of colonization [13] , [14] . The HMW1 and HMW2 adhesins are encoded by homologous chromosomal loci that appear to represent a gene duplication event and contain 3 genes , designated hmw1A , hmw1B , and hmw1C and hmw2A , hmw2B , and hmw2C , respectively [15] , [16] . HMW1 and HMW2 are synthesized as pre-pro-proteins ( Figure 1A ) and are secreted by the two-partner secretion system [17]–[19] . Amino acids 1–68 represent an atypical signal peptide and direct the pre-pro-proteins to the Sec apparatus , where they are cleaved by signal peptidase I [18] . The resulting pro-proteins are targeted to the HMW1B and HMW2B outer membrane translocators and undergo cleavage between amino acids 441 and 442 , removing the pro-pieces and generating mature species that are 125 kDa and 120 kDa , respectively [18]–[21] ( Figure 1A ) . Following translocation across the outer membrane , mature HMW1 and HMW2 remain non-covalently associated with the bacterial surface [18] , [19] . In recent work , we demonstrated that HMW1 is a glycoprotein and undergoes glycosylation in the cytoplasm in a process that is dependent upon HMW1C [22] . Functional analyses revealed that glycosylation of HMW1 protects against premature degradation , analogous to some eukaryotic proteins [22] . In addition , glycosylation appears to influence HMW1 tethering to the bacterial surface , a prerequisite for HMW1-mediated adherence [22] . Based on carbohydrate composition analysis of purified HMW1 using gas chromatography and combined gas chromatography-mass spectrometry , the modifying sugars include glucose , galactose , and possibly small amounts of mannose [22] . Analysis of HMW1 proteolytic fragments by mass spectrometry identified 31 sites of modification [23] . All of the modified sites were asparagine residues , in all except one case within the conventional sequence motif for eukaryotic N-linked glycosylation , namely NX ( S/T ) where X is any residue except for proline [23] . LC-MS/MS analysis , accurate mass measurement , and deuterium replacement studies established that the modifying glycan structures were mono-hexose or di-hexose units rather than N-acetylated hexosamine units that comprise the di-N-diacetyl chitobiose core of eukaryotic and many bacterial asparagine-linked glycans . These results suggested a novel N-linked carbohydrate-peptide transferase activity that does not require assembly of the monosaccharide units onto a lipid-linked intermediate [23] . In the present study , we studied the enzymatic mechanism responsible for the glycosylation of asparagine residues in HMW1 . We found that the HMW1C protein encoded in the hmw1 gene cluster is capable of transferring glucose and galactose to the HMW1 adhesin . In addition , HMW1C is capable of generating hexose-hexose linkages .
In earlier work we found that insertional inactivation of hmw1C in H . influenzae strain Rd-HMW1 resulted in a loss of glycosylation of HMW1 [22] , suggesting that HMW1C participates in the process of glycosylation . Further analysis revealed that amino acids 386–439 in HMW1C share 40–41% identity and 51–65% similarity with a domain conserved in a family of eukaryotic O-GlcNAc transferases , including human O-GlcNAc transferase , rat O-GlcNAc transferase , and a plant protein called Spy [22] , raising the possibility that HMW1C is a glycosyltransferase . To address the possibility that HMW1C is the glycosyltransferase responsible for N-linked glycosylation of HMW1 , we purified HAT-tagged HMW1C and Strep-tagged HMW1802–1406 ( Figure 1B ) . HMW1802–1406 corresponds to just over half of mature HMW1 ( HMW1442–1536 ) , contains 18 documented N-linked glycosylation sites , and was more amenable to purification than mature HMW1 ( Figure 1A ) . Subsequently , we incubated approximately equimolar quantities of HAT-HMW1C and Strep-HMW1802–1406 with both UDP-α-D-glucose and UDP-α-D-galactose at room temperature for 60 minutes , then examined the reaction mixture for reactivity with the DIG-glycan reagents . As shown in Figure 2A , we observed efficient glycosylation of HMW1802–1406 that was dependent on both HMW1C and the UDP-hexoses . To extend this result , we performed the same experiment with UDP-α-D-glucose by itself , UDP-α-D-galactose by itself , GDP-α-D-mannose by itself , UDP-α-D-N-Acetylglucosamine by itself , and UPD-α-D-N-Acetylgalactosamine by itself . As shown in Figure 2B , we observed glycosylation with UDP-α-D-glucose alone and UDP-α-D-galactose alone but not with GDP-α-D-mannose , UDP-α-D-N-Acetylglucosamine , or UPD-α-D-N-Acetylgalactosamine alone . To determine whether smaller amounts of HMW1C are associated with appreciable glycosylation of HMW1802–1406 , we repeated assays with a fixed amount of HMW1802–1406 , fixed amounts of UDP-α-D-glucose and UDP-α-D-galactose , and dilutions of HMW1C . Based on analysis using DIG-glycan reagents , we observed efficient glycosylation with molar quantities of HMW1C that were less than one-tenth the molar quantity of HMW1802–1406 ( data not shown ) . To address whether the glycosylation of HMW1802–1406 in in vitro reactions mimicked glycosylation of native HMW1 in whole bacteria and to gain further insight into which sugars modify which sites , we repeated reactions with purified Strep-tagged HMW1802–1406 , purified HAT tagged HMW1C , and UDP-α-D-glucose alone , UDP-α-D-galactose alone , GDP-α-D-mannose alone , or UDP-α-D-glucose plus UDP-α-D-galactose plus GDP-α-D-mannose and then examined the reaction mixtures by LC-MS/MS . As a positive control we examined purified HMW1802–1406 recovered from DH5α/pASK-HMW1802–1406 + pHMW1C , and as a negative control we examined HMW1802–1406 recovered from DH5α/pASK-HMW1802–1406 ( lacking pHMW1C ) . As summarized in Table 1 , we detected 10 of the 18 predicted sites of glycosylation and 11 distinct glycopeptides in HMW1802–1406 , including 10 glycopeptides with a single site of glycosylation and one glycopeptide with two sites of glycosylation ( KNITFEGGNITFGSR ) . Interestingly , of the 10 sites of glycosylation , all were modified in the in vitro reactions with UDP-α-D-glucose alone and with UDP-α-D-glucose plus UDP-α-D-galactose plus GDP-α-D-mannose . In contrast , only 6 of the 10 sites of glycosylation were modified in the in vitro reactions with UDP-α-D-galactose alone . Consistent with our observations using DIG-Glycan reagents , no sites were glycosylated in the in vitro reactions with GDP-α-D-mannose alone . As demonstrated by the collision-induced fragmentation spectra shown in Figure 3 and Figure S1 , the glycopeptide NLSITTNSSSTY ( HMW1 amino acids 946–958 , with glycosylation at N952 ) and the glycopeptide AITNFTFNVGGLFDNK ( HMW1 amino acids 909–924 , with glycosylation at N912 ) were present in two forms , including one with a mono-hexose at the predicted site of glycosylation and the other with a di-hexose at the predicted site of glycosylation . The forms containing a mono-hexose were detected in the in vitro reactions with UDP-α-D-glucose alone , UDP-α-D-galactose alone , and UDP-α-D-glucose plus UDP-α-D-galactose plus GDP-α-D-mannose , while the forms containing a di-hexose were detected only in the in vitro reactions with UDP-α-D-glucose alone and with UDP-α-D-glucose plus UDP-α-D-galactose plus GDP-α-D-mannose , suggesting that glucose must be the first hexose linked to asparagine in the glycopeptides containing di-hexose modification . Together these findings demonstrate that the HMW1C protein is a glycosyltransferase and has a novel activity capable of transferring glucose and galactose to asparagine residues in HMW1 and creating hexose-hexose bonds . In addition , they demonstrate that the di-hexosylated sites at N951 and N912 are initially modified with a glucose monosaccharide . To extend our understanding of glycosylation of HMW1 and confirm our observation that HMW1 is modified with glucose and galactose in in vitro glycosylation assays , we examined the effect of insertional inactivation of galU ( open reading frame HI0812 in strain Rd ) on glycosylation of HMW1 in strain Rd-HMW1 . The galU gene encodes glucose-1-phosphate uridyl transferase , which converts glucose-1-phosphate to UDP-glucose ( Figure S2 ) . UDP-glucose in turn can be converted directly to UDP-galactose by GalE ( UDP Gal-4-epimerase ) or can serve as the donor of UDP for conversion of galactose-1-phosphate to UDP-galactose . In assessing the effect of inactivation of galU , we incubated Rd-HMW1/galU in supplemented brain heart infusion broth [24] , which contains glucose as the primary carbon source . Interestingly , inactivation of galU mimicked the effect of inactivation of hmw1C described in our earlier work [22] , eliminating HMW1 glycosylation as assessed by DIG-glycan blots ( Figure 4A ) , virtually eliminating HMW1 tethering to the bacterial surface ( Figure 4B ) , and abolishing HMW1-mediated adherence ( Figure 4C ) . Consistent with our in vitro glycosyltransferase assays with purified HMW1802–1406 and HMW1C , these results indicate that UDP-glucose is required for glycosylation of HMW1 in H . influenzae under standard growth conditions in supplemented brain heart infusion broth .
In this study , we found that the H . influenzae HMW1C protein encoded in the hmw1 gene cluster is a glycosyltransferase and is capable of transferring glucose and galactose to asparagine residues in the HMW1 adhesin , providing the first example of a glycosyltransferase that transfers hexose units rather than N-acetylated amino sugars to asparagine residues in protein targets . Further analysis revealed that HMW1C is capable of creating both hexose-asparagine and hexose-hexose linkages , suggesting multi-functionality as a glycosyltransferase . All previously reported carbohydrate modification of asparagine residues in proteins in Eukarya and Bacteria involve the en bloc transfer of oligosaccharides from a lipid-linked intermediate by an oligosaccharyltransferase complex [25] . In Archaea , the mechanisms of N-glycosylation are less well understood . Glycosylation of asparagine residues with a trisaccharide moiety in the flagellin and S-layer proteins of Methanococcus voltae has been proposed to proceed via a lipid-linked intermediate [26] . More recently it has been shown that hexose units are attached directly to asparagine residues in an S layer glycoprotein of Haloferax volcanii [27] . A pentasaccharide with the structure Hex-X-hexuronic acid-HexA-HexA-Hex-peptide was identified at two glycosylation sites . Interestingly , these two sites were different from the conventional N-glycosylation sequence motif observed in eukaryotes and in HMW1 . It is currently unclear whether the H . volcanii Hex-Asn linkage is formed from a lipid-linked intermediate or via activated monosaccharides as we have found with HMW1 and HMW1C . In earlier work , we performed carbohydrate composition analysis on purified HMW1 and detected glucose , galactose , and small amounts of mannose [22] . Given the potential for contaminating sugars to be detected in this analysis , we were uncertain as to whether mannose was truly present as a modifying sugar in HMW1 , especially given that it accounted for only 2 . 5–3% of the total carbohydrate [22] . Our analysis in the current study argues that mannose is not present in HMW1 . In particular , in in vitro glycosyltransferase assays using purified HMW1802–1406 , HMW1C , and GDP-α-D-mannose , we were unable to detect modification of HMW1802–1406 using either DIG-Glycan reagents or LC-MS/MS . Based on assessment of the 10 glycopeptides that we detected in our in vitro glycosylation assays with HMW1802–1406 , which corresponds to just over half of mature HMW1 , we observed that HMW1C transfers glucose to all glyscosylated asparagines and transfers galactose to only a subset of glyscosylated asparagines . All of these glycosylation sites correspond to the conventional sequence motif of N-linked glycans , namely NX ( S/T ) , with X being any amino acid except proline . Examination of the primary amino acid sequence of the sites that are modified only with glucose and the sites that are modified with either glucose or galactose in in vitro assays reveals no apparent distinction , suggesting that factors beyond the amino acid sequence influence the specificity or potentially the efficiency of glycosylation . This observation is consistent with the fact that only a fraction of conventional sequences motifs are glycosylated in HMW1 purified from H . influenzae [23] . Further analysis of the glycopeptides detected after in vitro glycosylation revealed two peptides that were modified with a di-hexose . Interestingly , in both cases the glycopeptides were detected only in the reactions performed with UDP-α-D-glucose alone and with UDP-α-D-glucose plus UDP-α-D-galactose plus GDP-α-D-mannose , indicating modification with UDP-α-D-glucose . In contrast , the corresponding glycopeptides containing a single hexose at the asparagines in question were detected in the reactions performed with UDP-α-D-glucose alone , with UDP-α-D-galactose alone , and with UDP-α-D-glucose plus UDP-α-D-galactose plus GDP-α-D-mannose , indicating modification with either glucose or galactose . Considered together , these results suggest that glucose must be linked to asparagine in the glycopeptides containing di-hexose modification . At this point , it is unclear whether the di-hexose is generated prior to modification of the acceptor asparagine residue or whether instead a single hexose is linked to the target asparagine and then a second hexose is linked to the first hexose , although the conventional interpretation is that the hexose is added to the protein and then the chain is extended . In either event , it appears that HMW1C is responsible for creating the hexose-hexose bond . Interestingly , homology analysis reveals 42–68% identity and 58–83% similarity between the full-length HMW1C sequence and proteins in a number of other gram-negative bacterial pathogens , including the enterotoxigenic E . coli ( ETEC ) EtpC protein and predicted proteins in Yersinia pseudotuberculosis , Y . enterocolitica , Y . pestis , H . ducreyi , Actinobacillus pleuropneumoniae , Mannheimia spp . , Xanthomonas spp . , and Burkholderia spp , among others ( Table S1 ) . In ETEC , Y . pseudotuberculosis , Y . enterocolitica , and Y . pestis , these homologs are encoded by genes that are adjacent to known or predicted two-partner secretion loci . The H . ducreyi , Mannheimia succiniciproducens , and Burkholderia xenovorans genomes contain genes that encode predicted two-partner secretion proteins as potential targets for the HMW1C homologs , although these genes are in unlinked locations . The ETEC EtpC protein is encoded by a two-partner secretion locus called etpBAC and has been shown to be required for glycosylation of the EtpA adhesin , a high-molecular weight protein that has a predicted molecular mass of ∼177 kDa and promotes adherence to intestinal epithelial cells and colonization of the intestine in mice [28] , [29] . These observations suggest that that there is a family of bacterial HMW1C-like proteins with glycosyltransferase activity . To summarize , in eukaryotes N-linked glycosylation occurs in the endoplasmic reticulum and involves an oligosaccharyltransferase that catalyzes the transfer of the oligosaccharide from the lipid donor dolichylpyrophosphate to the acceptor protein . Similarly , in bacteria , N-glycosylation generally occurs in the periplasm and involves an oligosaccharyltransferase that transfers the glycan structure from a lipid donor to the acceptor protein . In contrast , in the case of the H . influenzae HMW1 adhesin , N-linked glycosylation occurs in the cytoplasm and involves direct transfer of hexose units to the acceptor protein by HMW1C , with no requirement for a lipid donor . In this study , we have established that the H . influenzae HMW1C protein is a multi-functional enzyme that is capable of transferring glucose and galactose to asparagine residues in selected conventional N-linked sequence motifs in HMW1 and is also capable of creating hexose-hexose linkages . Based on homology analysis , it is likely that a variety of other bacteria possess HMW1C-like proteins with similar enzymatic activity . In future work , we will examine whether these HMW1C-like proteins are identical to HMW1C in terms of the glycan units that they transfer and the acceptor protein sequence motifs that they recognize .
The strains and plasmids used in this study are listed in Table 2 . H . influenzae strain Rd-HMW1 is a derivative of strain Rd that contains the intact hmw1 locus and expresses fully functional HMW1 [22] . H . influenzae strain Rd-HMW1/hmw1C is a derivative of strain Rd-HMW1 that contains an insertionally inactivated hmw1C gene [22] . The H . influenzae Rd-HMW1 derivative harboring a kanamycin cassette in galU was constructed by transforming competent Rd-HMW1 with genomic DNA recovered from RdgalU and selecting for kanamycin resistance [30] . In order to overexpress HMW1802–1406 with a Strep tag at the N terminus , the fragment encoding HMW1802–1406 was amplified by PCR from pHMW1-14 using a 5′ primer that incorporated a BamHI site and a 3′ primer that incorporated a SalI site . The PCR amplicon was digested with BamI and SalI and then ligated into BamHI-SalI-digested pASK-IBA12 ( IBA , BioTAGnology ) , creating pASK-HMW1802–1406 . In order to overexpress the HMW1C protein with a HAT epitope at the N terminus , the hmw1C gene was amplified by PCR from pHMW1-14 using a 5′ primer that incorporated a BamHI site and a 3′ primer that incorporated an EcoRI site . The PCR amplicon was digested with BamHI and EcoRI and then ligated into BamHI-EcoRI-digested pHAT10 ( Clontech ) , creating pHAT-HMW1C . Plasmids were introduced into E . coli by chemical transformation [31] . DNA was introduced into H . influenzae using the MIV method of transformation described by Herriott et al . [32] . Transformants were selected by plating on agar containing kanamycin , and mutations were confirmed by PCR analysis using primers that anneal to regions flanking the target gene . To purify HMW1802–1406 , E . coli strain DH5α/pASK-HMW1802–1406 was grown at 37°C to an OD600 of 0 . 7 , then induced for 2 hrs with the addition of 100 µg/ml of anhydro-tetracycline ( Sigma ) . Cells were harvested , resuspended in 100 mM Tris pH 8 . 0 , 150 mM NaCl with Complete Mini protease inhibitor ( Roche ) , and lysed by sonication . Insoluble material was removed by centrifugation at 12 , 500 × g for 30 min . The supernatant was loaded onto a Strep-Tactin Superflow cartridge and eluted according to the manufacturer's instructions ( IBA , BioTAGnology ) . Eluted fractions were analyzed for purity by SDS-PAGE and were pooled . To purify HMW1C , E . coli strain DH5α/pHAT-HMW1C was grown at 37°C overnight . Cells were recovered , resuspended in 50 mM sodium phosphate buffer pH 7 . 0 , 300 mM NaCl ( bufferA ) , and lysed by sonication . Insoluble material was removed by centrifugation at 12 , 500 × g for 30 min . The supernatant was loaded onto a 1 ml Talon column ( Clontech ) and eluted with a gradient of 0 to 300 mM imidazole in Buffer A . Fractions were analyzed for purity by SDS-PAGE and were pooled . In standard in vitro glycosyltransferase assays , 1 . 5 µg ( 23 pmole ) of purified HMW1802–1406 was combined with a mixture containing 20 µl of 50 mM UDP-α-D-glucose , 50 mM UDP-α-D-galactose , 50 mM GDP-α-D-mannose , 50 mM UDP-α-D-N-Acetylglucosamine , or 50 mM UDP-α-D-N-acetylgalactosamine ( Calbiochem ) either as individual sugars or as mixtures . The reactions were initiated with addition of 1 . 5 µg ( 21 pmole ) of purified HMW1C in a final volume of 150 µl in 25 mM Tris pH 7 . 2 , 150 mM NaCl . Samples were incubated for 60 minutes at room temperature and then further incubated at 4°C overnight . To detect protein glycosylation , DIG Glycan reagents ( Roche ) were employed . Use of these reagents is based on the oxidation of hydroxyl groups in carbohydrates to aldehydes either in solution or bound to nitrocellulose membranes . Digoxigenin is then covalently linked to the aldehyde groups , and an anti-digoxigenin alkaline-phosphatase conjugated agent is used for detection of labeled carbohydrates . FACS analysis was performed by the Duke University Medical Center Cancer Research Center Flow Cytometry Shared Resource Center using a Becton Dickinson FACS Calibur instrument at a wavelength of 488 nm . Bacterial suspensions were fixed with 1% formaldehyde in PBS at room temperature for 30 min . After washing once with Tris buffered saline ( TBS ) , bacteria were resuspended in 1 ml of TBS , 50 mM EDTA , 0 . 1% bovine serum albumin , and a 1∶1000 dilution of guinea pig antiserum GP85 directed against HMW1 [33] and were incubated with gentle rocking at room temperature for 1 hr . Samples were then centrifuged , washed twice with PBS , and resuspended in 200 µl of PBS , 0 . 1% bovine serum albumin , and a 1∶200 dilution of Alexa Fluor488 anti-guinea pig antibody ( Molecular Probes ) . Samples were incubated with gentle rocking at room temperature for 1 hr . After two additional washes with PBS , bacterial pellets were re-suspended in 1 ml of PBS and were then analyzed . Data were analyzed with CELLQUEST software ( Becton Dickinson ) . To quantify histograms , markers were drawn on plots , and positive events within the markers were determined as a percentage of the positive control ( set at 100% ) . Adherence assays were performed with Chang epithelial cells ( human conjunctiva; ATCC CCL 20 . 2 ) ( Wong-Kilbourne derivative clone 1-5c-4 ) as described previously [16] . Percent adherence was calculated by dividing the number of adherent colony-forming units by the number of inoculated colony-forming units . All strains were examined in triplicate , and each assay was repeated at least two times . Whole cell sonicates were prepared by suspending bacterial pellets in 10 mM HEPES , pH 7 . 4 and sonicating to clarity . Proteins were resolved by SDS-PAGE using 10% polyacrylamide gels . Western blots were performed using guinea pig antiserum GP85 against the HMW1 protein [30] . Samples were precipitated using the 2D protein clean up kit ( GE Healthcare ) according to the manufacturer's instructions . Bovine serum albumin ( 100 ng ) was added to each sample as an internal standard . Pellets were dissolved in 40 µl 9 M urea and aliquoted into 0 . 5 ml microfuge tubes . Samples ( 20 µl in 9 M urea ) were reduced with 5 mM TCEP at pH 8 . 0 at room temperature for 30 min and were alkylated with 10 mM iodoacetamide ( Bio-Rad ) in the dark at room temperature for 30 min . TCEP and iodoacetamide were quenched with 5 mM dithiothreitol ( DTT ) at room temperature for 10 min . The reduced and alkylated proteins were digested with 1 µg of endoproteinase Lys-C ( Roche ) at 37°C overnight . Samples were diluted with 64 µl H2O to reduce the concentration of urea to 2 M and were then digested with 4 µg trypsin ( Sigma ) at 37°C overnight . Peptides were acidified with 5 . 5 µl formic acid ( Sigma ) and extracted 6 times with 10–200 µl NuTip porous graphite carbon wedge tips ( Glygen ) according to the manufacturer's directions and were then eluted into 1 . 5 ml autosampler vials with 60% acetonitrile ( Burdick & Jackson ) in 0 . 1% formic acid . The peptide digests were evaluated for quality and detergent contaminants using MALDI-TOF/TOF [34] prior to LC-MS analysis . For MALDI-TOF/TOF analysis , the peptide sample ( 0 . 5 µl ) was mixed with an equal volume of MALDI matrix solution ( Agilent Technologies ) prior to spotting . For nano-LC-FTICR-MS analysis , the peptide sample was dried and immediately dissolved in 10 µl aqueous acetonitrile/formic acid ( 1%/1% ) . The complex mixtures of peptides and glycopeptides from HMW1802–1406 were analyzed using high-resolution nano-LC-MS on a hybrid mass spectrometer consisting of a linear quadrupole ion-trap and an Orbitrap ( LTQ-Orbitrap XL , Thermo-Fisher ) . The liquid chromatographs were nanoflow HPLC systems ( NanoLC-1Dplus™ and NanoLC-Ultra™ ) that were interfaced to the mass spectrometer with a nanospray source ( PicoView PV550; New Objective ) . The in-house packed LC column ( Jupiter C12 Proteo , 4 µm particle size , 90 Å pore size [Phenomenex] ) was equilibrated in 98% solvent A ( aqueous 0 . 1% formic acid ) and 2% solvent B ( acetonitrile containing 0 . 1% formic acid ) . The samples ( 10 µL ) were injected from autosampler vials using the LC-systems autosamplers at a flow rate of 1 . 0 µL/min and were eluted using a segmented linear gradient ( 250 nL/min ) with solvent B: isocratic at 2% B , 0–2 min; 2% B to 40% B , 2–65 min; 40% B to 80% B , 65–70 min; isocratic at 80% B , 70–72 min; 80% B to 2% B , 72–77 min; and isocratic at 2% B , 77–82 min . The survey scans ( m/z 350–2000 ) ( MS1 ) were acquired at high resolution ( 60 , 000 at m/z = 400 ) in the Orbitrap , and the MS/MS spectra ( MS2 ) were acquired in the linear ion trap at low resolution , both in profile mode . The maximum injection times for the MS1 scan in the Orbitrap and the LTQ were 50 ms and 100 ms , respectively . The automatic gain control targets for the Orbitrap and the LTQ were 2×105 and 3×104 , respectively . The MS1 scans were followed by six MS2 events in the linear ion trap with wideband collision activation in the ion trap ( parent threshold = 1000; isolation width = 2 . 0 Da; normalized collision energy = 30%; activation Q = 0 . 250; activation time = 30 ms ) . Dynamic exclusion was used to remove selected precursor ions ( −0 . 25/+1 . 5 Da ) after MS2 acquisition with a repeat count of 2 , a repeat duration of 30 s , and a maximum exclusion list size of 200 . The following ion source parameters were used: capillary temperature 200 °C , source voltage 2 . 5 kV , source current 100 µA , and the tube lens at 79 V . The data were acquired using Xcalibur , version 2 . 0 . 7 ( Thermo-Fisher ) . The MS2 spectra were analyzed both by searching a customized protein database that contained the sequences of HMW802–1406 and by expert manual interpretation . The exact masses of the glycopeptides and fragmentation ions were calculated using the Molecular Weight Calculator , version 6 . 45 ( http://ncrr . pnl . gov/software/ ) . For database searches , the LC-MS files were processed using MASCOT Distiller ( Matrix Science , version 2 . 3 . 0 . 0 ) with the settings previously described [35] . The resulting MS2 centroided files were used for database searching with MASCOT , version 2 . 1 . 6 , and the following parameters: enzyme , trypsin; MS tolerance = 10 ppm; MS/MS tolerance = 0 . 8 Da with a fixed carbamidomethylation of Cys residues and the following variable modifications: Methionine , oxidation; Pyro-glu ( N-term ) ; Maximum Missed Cleavages = 5; and 1+ , 2+ , and 3+ charge states . | Decoration of proteins with carbohydrates has an important impact on protein function throughout biology and has been recognized increasingly in pathogenic bacteria . Haemophilus influenzae is a common cause of both bacterial respiratory tract disease and bacterial invasive disease and initiates infection by colonizing the upper respiratory tract . The Haemophilus HMW1 adhesin is a large protein that resides on the bacterial surface and mediates bacterial attachment to respiratory epithelial cells , an essential step in the process of colonization . In recent work , we discovered that HMW1 is decorated at multiple sites with short carbohydrate units that serve to prevent degradation and to stabilize association with the bacterial surface . In the current study we identify the enzyme responsible for adding carbohydrate units at specific sites of HMW1 . In addition , we demonstrate that this enzyme is capable of creating both carbohydrate-protein and carbohydrate-carbohydrate bonds . The amino acid sequence of this enzyme is similar to the sequences of proteins in several other bacteria , suggesting a new family of bacterial enzymes capable of creating carbohydrate-protein and carbohydrate-carbohydrate bonds . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases/bacterial",
"infections",
"microbiology/cellular",
"microbiology",
"and",
"pathogenesis",
"biochemistry/biocatalysis"
] | 2010 | The Haemophilus influenzae HMW1C Protein Is a Glycosyltransferase That Transfers Hexose Residues to Asparagine Sites in the HMW1 Adhesin |
Mammalian embryogenesis is a dynamic process involving gene expression and mechanical forces between proliferating cells . The exact nature of these interactions , which determine the lineage patterning of the trophectoderm and endoderm tissues occurring in a highly regulated manner at precise periods during the embryonic development , is an area of debate . We have developed a computational modeling framework for studying this process , by which the combined effects of mechanical and genetic interactions are analyzed within the context of proliferating cells . At a purely mechanical level , we demonstrate that the perpendicular alignment of the animal-vegetal ( a-v ) and embryonic-abembryonic ( eb-ab ) axes is a result of minimizing the total elastic conformational energy of the entire collection of cells , which are constrained by the zona pellucida . The coupling of gene expression with the mechanics of cell movement is important for formation of both the trophectoderm and the endoderm . In studying the formation of the trophectoderm , we contrast and compare quantitatively two hypotheses: ( 1 ) The position determines gene expression , and ( 2 ) the gene expression determines the position . Our model , which couples gene expression with mechanics , suggests that differential adhesion between different cell types is a critical determinant in the robust endoderm formation . In addition to differential adhesion , two different testable hypotheses emerge when considering endoderm formation: ( 1 ) A directional force acts on certain cells and moves them into forming the endoderm layer , which separates the blastocoel and the cells of the inner cell mass ( ICM ) . In this case the blastocoel simply acts as a static boundary . ( 2 ) The blastocoel dynamically applies pressure upon the cells in contact with it , such that cell segregation in the presence of differential adhesion leads to the endoderm formation . To our knowledge , this is the first attempt to combine cell-based spatial mechanical simulations with genetic networks to explain mammalian embryogenesis . Such a framework provides the means to test hypotheses in a controlled in silico environment .
How a complete embryo emerges starting from a single fertilized egg is an intriguing process in developmental biology , understanding of which has important clinical implications [1] . Recent advances in live imaging have allowed for the tracking of single cells as they grow and divide and subsequently form different tissues of the embryo [2] . Using fluorescent labeling one is able to monitor in real time the expression levels of key transcription factors in single cells as they move and divide . Recent experiments have shown significant correlations between the individual cell fates and specific gene expression patterns [3] , [4] . Studies with respect to early events in the morphogenesis of the mammalian embryo suggest that , although the combined interplay between gene expression and cell polarity perhaps determine the cell division rules , the mechanical properties of cells which may also depend on gene expression , collectively organize cells into different tissues [3]–[5] . The first developmental phase occurs when some of the cells from the morula differentiate to become part of the trophectoderm ( TE ) lineage , forming an outer layer surrounding the inner cell mass ( ICM ) [6] ( Figure 1 ) . After the TE layer is formed , cells secrete a fluid , which coalesces and expands as a single entity , the blastocoel [7] . The latter gradually pushes all ICM cells to one end of the protective outer envelope , the zona pellucida ( Figure 1 ) . At this stage a second developmental event occurs – the formation of the primitive endoderm ( PE ) . This is the covering which separates the ICM from the blastocoel . The analysis of molecular and mechanical processes , which ensure the robust patterning of these layers of cells [3] , is the subject of this work . Previous studies have identified specific gene expression with the three lineages , ICM , TE and PE . The inner cells which ultimately give rise to the three germ layers are pluripotent and express the well known embryonic stem cell transcription factors Oct4 , Sox2 and Nanog amongst several others [8] . The cells forming the trophectoderm exclusively express Cdx2 , whereas , the cells which are part of the endoderm lineage express Gata6 . There are several mutually antagonistic interactions between these key transcription factors . Cdx2 represses Oct4 and vice versa [9] . In addition , these transcription factors are also positively auto-regulating ( see [10] , [11] and references therein ) , ensuring that once turned on , they remain stably expressed . Recent work suggests that stochasticity is instrumental in the patterning process , in which key genes are initially expressed in a fluctuating manner and only later in the development does a pattern emerge [12] . When cells have decided upon particular lineages , the positive auto-regulation ensures that only the trophectoderm cells , which are the outer cells express Cdx2 ( simultaneously suppressing Oct4 ) , whereas the inner cells express Oct4 ( suppressing Cdx2 ) thereby enforcing mutually exclusive expression of lineage genes [13] . Similarly , Gata6 and Nanog are very likely mutually antagonistic [14] , the former is expressed in PE whereas Nanog , which is part of the trio of embryonic transcription factors , is expressed in the epiblast cells . In [11] a computational model , based upon several interactions of these key genes , was developed for the genetic circuit which determines cell fate , i . e . , TE , epiblast or PE . The main conclusion from [11] was that the network dynamics exhibited a switch-like behavior , as a function of an external signal . The question of interest in this work is how the circuit dynamics of these various components regulate cell fate , as cells become part of the PE , ICM and TE . After the fertilized egg has undergone three rounds of division ( Figure 1 ) , the outer cells get polarized along the apical-basal direction . If an outer cell undergoes a symmetrical division , both daughter cells retain the cell polarity , but if the division is asymmetrical , the inner cell looses polarity . Cell polarity and Cdx2 expression have been implicated to feedback onto each other [15] , thereby making the polarized cells increase Cdx2 expression . This ensures that the outer cells express Cdx2 . However , this begs the question as to which factor determines symmetrical/asymmetrical divisions . In [15] , the authors suggest that CDX2 levels themselves affect the division pattern . Cells , which express higher levels of CDX2 , divide symmetrically , whereas for lower levels of CDX2 , cells divide asymmetrically , such that upon division , the inner cell gives up most of its Cdx2 mRNA to the outer cell . Although the mechanism by which cells choose their division plane by reading out the levels of CDX2 is not known , it can be classified as implementing the “cell lineage determines cell position” rule . An alternative is the “cell position determines cell lineage” rule , which is thought to be connected to nuclear localization of YAP , which is a cofactor of Tead , a transcription factor upstream of Cdx2 . In [16] , the authors suggest that cells that are outside lack a signal , which necessarily allows YAP to be localized to the nucleus . In this way , the outside cells automatically express Cdx2 , whereas the inner cells do not express Cdx2 . In both of the above hypotheses , it can be assumed that the Oct4- Cdx2 mutual antagonism gradually fine tunes any small discrepancies in their levels , once they are determined to be expressed in cells . The next stage of development is the PE formation . The transcription factor Gata6 is expressed by the PE cells , which ensures through the mutual repressive interactions with Nanog , that the embryonic genes are shut off . However , initially , Gata6 and Nanog seem to be expressed in spatial “salt and pepper” pattern [12] . From this initial distribution the pattern changes such that , the cells occupying the outer layer of the ICM , which face the blastocoel , must express Gata6 . How cells get patterned in this manner , has been the subject of ample research [3] . Three different processes are thought to occur [17] . If the cells , which express Gata6 have slightly different adhesive properties from the cells expressing Nanog , the two populations of cells can get sorted out . However , for some of the Gata6 cells , which must move out from deeper layers of the ICM to occupy the outer layer , there could be some type of external “homing” signal . It is interesting to speculate if the fibroblast growth factor FGF signaling , which plays role in the endoderm development [3] , might provide such a cue . Finally , cells expressing Gata6 , which are unable to move through the deeper layers and emerge to the outer layer , can undergo apoptosis , and be hence removed from the entire population of cells . These processes can be combined to give a robust “movement” of GATA6 cells [18] , thereby implementing the “cell lineage determines cell position” rule . One of the aspects of embryo development is the formation of the embryonic-abembryonic ( eb-ab ) axis . An early proposal was that the eb-ab axis position was correlated to the first division plane of the fertilized egg [3] . Each of two cells would then contribute to different tissues ( TE and epiblast ) of the embryo . However , significant cell movement of individual cells occurs and the entire mass of cells can also rotate [2] . This makes it difficult to follow and assess the clonal expansion of cells . Another hypothesis is that the emergence of the eb-ab axis is entirely due to mechanical constraints [19] . The pellucid zone ( ZP ) is usually elliptical , and hence it is possible that cells move into one end of the long axis to minimize the elastic energy . We propose a mathematical framework which takes into account growing and proliferating cells , interacting through physical forces to understand the patterning of the blastocyst into the trophectoderm , epiblast and the primitive endoderm . Two hypotheses which we explicitly explore are , ( 1 ) gene expression determines the geometry of division ( 2 ) gene expression determines cellular motion through modification of cellular adhesion properties . The gene expression itself is determined by an underlying genetic network , which is coupled to both division and spatial location within the embryo . Faced with the complexity of the processes described above , we believe that such a computational approach , in which each hypothesis is simulated explicitly , provides the means to bridge intuition with understanding . Further , it provides novel predictions which can subsequently be tested .
The cell based model presented here aims at describing morphogenesis of the mammalian embryo in the early phases of development in terms of simplified mechanical interactions between blastomeres , which are coupled to gene network dynamics within each cell . The network dynamics feeds back on ( i ) division patterns and ( ii ) adhesive properties of cells . The mechanical part of our model is inspired by Dallon and Othmer [20] , who analyzed cell movement in the Dictyostelium discoideum slug . We model cells as elastic spheres interacting with each other and constrained by the pellucid zone . Spherical geometry faithfully reproduces the shape of the cells in early mammalian embryogenesis except for the flattened trophectoderm cells where the oblate elliptical shape is more appropriate . The blastomeres are treated as incompressible elastic bodies , whose mechanical response is confined to three orthogonal axes . This is equivalent to cells being represented as membranes whose deformation is restricted by springs in three orthogonal directions . These directions correspond to the principal directions of the stress tensor and all the external forces acting upon the cell are resolved to these coordinates . Cells come into contact and mechanically interact with each other ( Figure S1 ) . Keeping track of the attachment points of the forces on the surface of the cell allows the model to follow the changes in translative , compressive and tensile forces . The forces acting on the cell are summed up over all the neighbors of the cell . The neighborhood relation itself is determined from the Voronoi diagram of the cell centers , which is dynamically updated at each step of the simulation . Assuming that due to the low Reynolds number of the cell movement we can neglect accelerations in the dynamics we write the equations of motion for individual cells in the form , ( 1 ) In Eq . ( 1 ) we account for three types of mechanical forces: elastic interaction between cells ( ) , adhesive drag force of cells sliding against each other ( ) and an attractive adhesion force ( ) . The represents forces due to the active movement of cells or pressure from blastocoel , which do not originate from intrinsic mechanical interactions between the cells . The factor stands for the viscosity coefficient and is the three-dimensional position vector of cell ( see Text S1 for details ) . Each blastomere is defined by the genetic network ( see Text S1 for details ) , which evolves the mRNA and protein concentrations of the cell , a set of parameters ( Table S1 ) which define its mechanical properties , a polarization vector and the cell cycle length . These parameters can be different for different cell types ( i . e . TE , PE and ICM ) . Cell division in our model ( see Text S1 for details ) is a discrete event where a single mother cell is replaced by two daughter cells . The cells undergo division when the time elapsed from the last division exceeds the cell cycle length . The latter is randomized among the blastomeres according to a normal distribution [21] . The total cell volume is conserved during this step and initially overlapping daughter cells occupy the space inside the mother cell . During division , in addition to the obvious change of geometry , cells also need to partition their content to the daughter cells . This is accomplished in the model using two different recipes: ( i ) random symmetrical , where the direction of division is random and the content is distributed symmetrically , and ( ii ) polarized asymmetrical division , where the direction of division is correlated with the cell polarization vector and the content is partitioned asymmetrically . Note that the division gives rise to opposing forces for both daughter cells , in a direction perpendicular to the division plane . The mechanical equations of motion as well as genetic networks equations for the entire system are solved numerically using a fifth order Runge-Kuta differential equation solver , which makes adaptive time steps based on the requirement of keeping the error less than a given threshold . A typical movie of the resulting dynamics can be found in Video S1 .
We first analyze the correlation between the orientation of the blastocyst embryonic-abembryonic ( eb-ab ) axis and the axis of the two-cell embryo . Our motivation is to test the hypothesis that this phenomena occurs from the alignment of both of these directions with the long axis of elliptical pellucid zone ( ZP ) due to the mechanical constraints . This hypothesis has been analyzed experimentally in several studies providing data both in favor and against it [17] , [22]–[24] . The inconsistent results could be due to different strains of specimen , different experimental techniques , or difficulty in tracking cells given the considerable cell mobility in the early embryo and the embryo inside the ZP as a whole . However , this discrepancy could also be a result of different mechanisms that are involved in the formation of the eb-ab axes and the cleavage pattern of the two cell embryo . Since the shape of the ZP is not perfectly spherical , it provides a directionality , which could influence the orientation of the cells in the developing embryo . Here we test whether the mechanical constraints arising from the ZP geometry could be the underlying cause of the orientation of the embryo , both at two-cell and blastocyst stages . In [25] the authors developed a computational model of the embryo comprised of cells with a blastocoel , surrounded by the ZP to study mechanical effects on the eb-ab axis . They concluded that the cells would acquire the configuration of minimal energy orienting themselves along the corner of the ZP long axis . Hence , mechanical interactions would determine the eb-ab axis . Although their model demonstrates this interesting result , it does not take into account cell proliferation . Within our model , we are able to analyze the joint effects from mechanical interactions on orientation of both two cell embryo and the eb-ab axes together with cell proliferation . In our model , the ZP acts as a static barrier elastically repelling the spherical blastomeres in contact , while the cells interact mechanically with each other and favor configurations which minimize the elastic energy of the system ( Figure 2b ) . Based on the fact that isolated blastomeres attain spherical shapes , the elastic blastomere energy in the model increases with the overlap of the spheres representing their native shape . Therefore the two first blastomeres will position themselves along the long axis of ellipsoidal pellucid zone minimizing their overlap or deformation . In the model we assumed that there are no frictional forces between ZP and blastomers , since our analysis of the motion of the cells in experiments and test simulations including friction suggest that such interaction has a marginal effect ( see Text S1 for details ) . In such case , even tiny differences ( % ) in the axes length of the ellipsoid were sufficient to provide a positional cue for the two cell embryo . As expected , due to the increase in blastomere overlaps , the dynamics of the alignment process is faster and more robust with increasing difference of axes length . The random division of outer cells , which are close to the ZP , can create a torque , which rotates the entire cell mass ( Video S2 ) . This rotation is also observed in experimental movies [2] , which reaffirms that the mechanics within a confined region plays an important role in the arrangement of cells within the blastula . As the development of the embryo progresses , the interactions between blastomeres become more complex due to both their increased number and changes in their mechanical properties . Around the 32-cell stage , with the trophectoderm well defined , the fluid filled blastocoel cavity begins to form with secretion of intracellular vacuoles which coalesce . We model the blastocoel in a simplified manner , as a slowly expanding spherically shaped region inside the ICM , aiming to capture the behavior of spatially restricted ICM cells . The adhesion strengths of the cell-cell interactions are deduced by what is qualitatively known for different cell types [12] , [26] . Compacted ICM cells exhibit strong self-adhesion , trophectoderm cells adhere to each other through tight junctions and have decreased adhesion to ICM cells . While we do not expect any adhesion-like force between the cells and the blastocoelic fluid . The simulations are initialized with a small spherical blastocoel volume in the center of the ZP at the 32-cell stage that later expands to ( 20%–30% ) of the whole embryo . Depending upon the degree of ZP elongation , we observe preferential localization of the blastocoel to the one end of the ZP long axis . To ensure that our results do not depend upon the initial 32-cell configuration , for each simulation we used a different initial template obtained from a single blastomere by five rounds of cleavages with stochastic time and direction of the cell division ( see Text S1 for details ) . A 10% difference in the length of axes of the ellipsoidal ZP provides alignment of ab-eb axis to the long axis of ellipsoid ( Figure 3 , Video S3 ) in 76% ( n = 50 ) of the simulations , suggesting that this configuration is mechanically preferred and that the oblate shape of the ZP could influence alignment of both the two-cell embryo and the blastocyst axes . We consider two axes aligned , for the purpose of the simulation , if the angle between them is less than 10 degrees and we define embryonic-abembryonic axis as the line passing through blastocoel center and center of mass of ICM cells . One should mention that in vivo , the ZP is not essential for blastocyst formation [2] . In our model we can also form a blastocyst without the ZP . In that case we do not observe the alignment of its axis with the axis of two cell embryo ( Video S4 ) . We conclude that in cases where the ZP is present , its shape affects the relative position of ICM and blastocoel in agreement with findings from the model in [25] . The same mechanical constraint aligns axis of two cell embryo and , as cells with the same lineage tend to occupy nearby positions , it influences lineage allocation to trophectoderm and ICM . This we confirm by lineage tracking in our simulations . In relation to the orientation of eb-ab axis and blastocyst linage formation , timing and orientation of two- to four-cell divisions has been studied by several groups [21] , [27] , [28] . In particular , certain patterns of cleavages , meridional-equatorial ( M-E ) together with reversed equatorial-meridional ( E-M ) order , were found to occur more often ( 80% of cases ) and were associated with specific tetrahedral arrangement of blastomeres in the blastula . In our model we find that the configuration of blastomeres in four-cell embryos depends upon the size of the blastomeres relative to the ZP and upon the ZP shape even more than upon the division pattern . We characterize the size of four-cell embryo blastomeres with respect to the size of the initial spherical zygote , . After two rounds of cleavages and conserving the cell volume we obtain blastomeres of the radius . As a maximal radius of the zygote we consider the radius of the sphere of the same volume as the ZP . In simulations we used in range to and varied the elongation ratio of the ellipsoidal ZP within 20% . In the case of spherical ZP , due to the symmetry , cells always attain tetrahedral configuration in four-cell embryo simulations . However , even a slight deviation ( 5% ) from spherical ZP symmetry causes blastomeres to prefer different configurations minimizing the total elastic energy . By decreasing the blastomere sizes at this point , their mobility is increased since the drag force decreases and the elastic interaction between them is lowered . This is sufficient to rescue the tetrahedral arrangement at some point ( ) . The exact numbers depend upon simulation parameters but the trends are robust . In the regime of large blastomeres , when they tend to depart from a tetrahedral configuration , their mobility is lowered and specific patterns of cleavages become increasingly important for their final configuration . Our results suggest that geometrical factors like size of the blastomeres and specific shapes of the ZP , ignored in studies of blastocyst lineage so far , may be influencing positions of the blastomeres in the blastula . Similarly we analyzed the number of cells located inside and outside at the 32-cell stage as a function of the individual blastomere size . We found that the number of inner cells decreases when lowering the cell size ( Figure 2a ) . While the spherical approximation may not accurately describe the flattened shape of the trophectoderm cells and we cannot expect to reproduce observed ratio of inner to outer cells in this way , the result again confirms that mechanical and geometrical constrains very likely play prominent roles in blastocyst development . Early mammalian embryo formation is characterized by a sequential order of morphological events , such as morula compaction or blastocoel expansion , which take place at precise stages of development . Even if these events are under the genetic control of processes inside each blastomere , the exact mechanism governing them is unknown and it is possible that cues other than genetics can contribute to triggering those events . Our model offers the possibility to analyze mechanical interactions taking place during embryogenesis . We have measured the average energy per cell of elastic deformation of identical blastomeres , as a function of time , from 2- to 32-cell stage ( Figure 2b ) . High peaks in this energy are observed during cell division , because just after a division , the daughter cells are highly deformed from their native spherical shape . More interestingly , we find differences in the stress perceived by blastomeres at different stages . We see an overall decrease of average deformation energy as the simulation progresses . This is expected since as blastomeres are getting smaller during cleavages , they can fit the pellucid zone shape with less overlap on average . In Figure 2b we observe larger differences in deformation energy between 4- to 8-cell and 16- to 32-cell stages than between 2- to 4-cell and 8- to 16-cell stages . Also note that the average cell deformation energy is lowest for the 32-cell stage . Since morula formation and compaction of blastomeres happen precisely at the 8-cell stage and the secretion of vacuoles forming blastocoel takes place approximately at the 32-cell stage , these results raise the intriguing question: Can sensing of mechanical signals provide triggers for some of the important events during embryogenesis ? Trophectoderm is the first occurring specialized tissue distinguished from the embryo mass . Despite remarkable progress in our knowledge about this process , the exact mechanism of trophectoderm formation is still an area of active research . Here we evaluate the robustness of two conceptual models of Cdx2 expression pattern during this process . In the first , “position-based model” the position of the cell in the embryo dictates the Cdx2 expression level , with inner cells having lower expression levels of Cdx2 than the outer cells [13] . We have implemented this model with a simple switch-like genetic network based on the mutual repression between Cdx2 and Oct4 in each cell ( Figure S2 , Figure S3 , Text S1 , Table S3 ) [8] . We assumed that the outer cells receive additional signal from polarity genes which enhances Cdx2 expression in those cells ( Figure 4a ) . Starting with small and random CDX2 levels at the 4-cell stage , we evolve the system to 32 cells . Since the relation between position and CDX2 level in the cell is defined in a straightforward manner in this model , we expect the distribution of CDX2 concentration to be clearly separated between the inner and outer cell populations . Indeed , we consistently observe ( Figure 4b , c , Table S2 ) significantly higher levels of CDX2 at the 32-cell stage in outer cells , in agreement with experimental findings [13] . In the second , polarity-based model the spatial pattern of CDX2 is not determined by a direct relation to the position of the cell [15] . Instead , outer cells , which are known to be polarized , are assumed to polarize Cdx2 mRNA as well and can distribute it non-uniformly between daughter cells during asymmetric divisions . To simulate this behavior , the daughter cell located outside receives 90% of the mother cell's Cdx2 mRNA while the other daughter cell , placed during division inside the embryo mass , obtains the remaining small portion of original Cdx2 mRNA content . In addition , the probability of symmetric division is small for low- and large for high- Cdx2 expressing outer cells . Inner cells , due to the lack of polarization , always divide in a symmetric manner . Such a feedback loop between the CDX2 level and the inside-outside polarization has been proposed to produce high CDX2 content in outer cells . We tested this model taking into account cell movement and the physical constraints . In our simulations we adapted the gene network involving Cdx2 , Oct4 and the mRNA produced by these genes , as a bistable switch ( Figure 4d , Text S1 , Table S4 ) . We tested whether the unequal distribution in the Cdx2 mRNA levels in daughter cells after an outer cell divides , is sufficient to establish the observed pattern of Cdx2 expression . As before , assuming random CDX2 levels at the 4 cell stage , we evolve the model up to the 32-cell stage and performed an analysis of the Cdx2 expression level . The polarity-based model is , in principle , capable of creating distinct Cdx2 expression levels of inner and outer cells ( Figure 4e , f ) . However , since it results in larger overlap of Cdx2 distributions in inner and outer cells , it is less robust than the position-based model , which by definition produces the required pattern . This behavior of the polarity-based model is a consequence of more indirect relationship between the CDX2 concentration and the cell position . This complicates conditions for trophectoderm specification , as it requires coordination of several factors , including mechanical interactions , which in addition to division patterns influence the final positions of the cells in the embryo . For example , to minimize relocation of low level CDX2 cells from inside to the outside and high level CDX2 cells in the opposite direction , due to the geometrical and mechanical constrains , the division patterns must take into account the ratio of the inside and outside cells . We explored the parameter space of the probability of symmetric and asymmetric cell divisions as a function of the CDX2 concentration within our model ( Figure S5 ) and found that it had to be carefully tuned to obtain the optimal pattern . Another key point is that this model , in the form presented above , does not have a mechanism which could cope with high level CDX2 cells located inside , which indicates that it may need additional hypotheses in order to better match the observed CDX2 level distributions . We further show through a simplified population based model that this process is not as robust as the position-based model ( Text S1 , Figure S6 ) . Finally , we should note that assumptions of both position- and polarity-based models could work in unison to produce the correct spatial pattern of CDX2 expression . The next major developmental event is the formation of the primitive endoderm ( PE ) , which , from a patterning perspective , is very different from the trophectoderm patterning discussed above . Trophectoderm formation involves differentiating between inner cells and the outer cells that completely surround these . These cells take on very different fates . The endoderm is a layer of cells which separates the blastocoel from the ICM cells , and hence chooses a specific side , namely the one facing the blastocoel , thereby breaking the symmetry which is present in trophectoderm formation . Two genes are characteristic in specifying the cell fate . Nanog which specifies the ICM cells and Gata6 the cells which finally form the endoderm . Cells express these genes in a salt-and-pepper manner prior to the creation of the endodermic layer of cells next to blastocoel [8] , [29] . Cells expressing Gata6 are thought to migrate away from the Nanog expressing cells towards the blastocoel , through mechanisms of dynamic rearrangement to finally form the endoderm layer . There could be potentially several processes by which such a rearrangement is possible . We have used our mathematical framework to test each process individually as well as in combination so as to provide a comparison of different scenarios . The two mechanisms we have considered are differential adhesion , and active cell movement determined by a directional signal emanating from the blastocoel . Cell sorting with differential cell adhesion [30]–[32] has been shown to be an important mechanism to spatially separate two different cell types within heterogeneous populations . However , it is not obvious , how efficient such cell sorting is in a system the size of the ICM , where the motility of cells is considerably reduced due to the tight packing . To test these hypothesis , we make the assumption that gene expression determines mechanical properties through differential adhesive properties of the Nanog and Gata6 expressing cells . In addition , we test how the dynamics of interactions with blastocoel can affect the endoderm formation . This we do by making two alternative hypothesis . The first assumes that the blastocoel is a static surface which merely provides a barrier to moving cells . We also include a directional force on Gata6 cells towards the blastocoel , by assuming that an external signal informs these cells to preferentially move in that direction . Although there is no evidence for such a directional signal , the role of growth factors could be instrumental in providing cues for directional cell movement . The second assumes that the blastocoel acts dynamically by exerting a constant and dynamic pressure on cells , thereby explicitly simulating forces between the ICM cells and the blastocoelic fluid . In our model we consider an initial template of 10–14 cells constrained within a half-ellipsoidal space spanned by trophectoderm and blastocoel boundaries ( Figures 5a and 5b ) . The ICM cells express either Gata6 or Nanog , and are randomly distributed . Subsequently , cells move due to the processes described above and due to random motions which arise during cell divisions up to three division cycles , finally giving rise to 80–112 cells . During division the model assumes that the daughter cells retain the identity ( Nanog or Gata6 ) of the parental cell . Throughout this process we assume the trophectoderm does not take an active part in the PE formation , since it is already specified prior to this stage . However , we include the interaction of blastomers with trophectoderm cells explicitly via elastic , adhesion and drag forces . Efficient cell sorting with differential adhesion requires a hierarchy of self and cross adhesion strengths [33] . We assume that cells expressing Nanog adhere strongly to each other , whereas cells expressing Gata6 adhere less tightly to each other and the cross-adhesion between the two different cells types is the least . In the static blastocoel case , simulations of cell sorting suggest that ( compare Figures 5c and 5d , Video S5 ) differential adhesion does have a positive role in segregating cells , although , by itself , does not position Gata6 cells in endoderm layer . The patterning of the endoderm can be significantly improved on adding to differential adhesion a directional force attracting Gata6 cells towards the blastocoel ( as discussed above , Text S1 ) , which gives the most robust result ( Figure 5e , Video S6 ) . Finally , the directional force alone ( Figure 5f , Video S7 ) , would not be sufficient , pointing to the fact that both , differential adhesion as well as the directional force are required to correctly position the Gata6 cells adjacent to the static blastocoel barrier ( Table 1 ) . In fact the efficiency by which the endoderm layer is formed , can be increased considerably for a given directional force by tuning differential adhesion such that Nanog expressing cells adhere even more tightly while the adhesion between Nanog and Gata6 cells is reduced ( see Figure S7 ) . Considering the second hypothesis , in which the blastocoel is a dynamic entity and exerts pressure on blastomers , we first test the effects of differential adhesion alone . The Figure 6c shows that although Gata6 and Nanog expressing cells successfully segregate , the endoderm layer becomes tilted with respect to embryonic-abembryonic axis , such that Nanog expressing cells come into contact with blastocoel . Hence , the endoderm pattern is not successfully reproduced . If , however , we now assume a positional bias for Nanog cells such that they have stronger adhesion with the trophectoderm cells , we obtain the correctly patterned endoderm ( Figure 6d ) . The mechanical forces on the ICM cells coming from the blastocoel push all the cells into one side of the embryonic-abembryonic axis . The Gata6 cells , which neither adhere strongly to themselves , nor to the Nanog cells are pressed away from the strongly self-adhering and clustered Nanog cells , towards the blastocoel . We also analyzed the effects of additional random forces , simulating cumulative effect of active cell movements and interactions with the environment . As can be seen from the Figure 6b , random forces by themselves do not suffice for the formation of the endoderm pattern , as they cannot provide the directional signal and are inefficient in allowing the Nanog cluster to form . The main effect of the random forces is to increase the efficiency of the pattern formation . This is because the extra random motions allow cells to sample more locations in space , leading to a better search for energetically favorable positions , thereby forming the endoderm much faster . These results indicate that , even in the small sized system of ICM , differential adhesion is a crucial mechanism , which can sort ICM and endoderm cells . However , to do it efficiently and to position the endoderm in the correct place requires , in addition to differential adhesion , either a directional force on Gata6 cells , assuming that the blastocoel surface is a static barrier , or a dynamic interaction with the blastocoel and preferential adhesion of Nanog and TE cells . Comparison of these two different hypotheses , suggests that robust endoderm formation can occur without the presence of a directional chemical signal , purely though mechanical interactions arising from differential adhesion , but only if we include the dynamics of the entire embryo . Finally , although we have not explicitly modeled apoptosis , we expect that it would play a synergistic role in the endoderm layer formation along with differential adhesion and the active motion . Its function could be to eliminate cells that are placed in the wrong position and hence cannot move towards their target destination , due to the forces arising from intervening cells . It could also increase motility of cells by freeing up space , or ensuring the correct ratio of ICM to PE cells during endoderm formation .
The emergence of high quality experimental data of developing embryos is an opportunity to develop mathematical models which can be used to elucidate the different mechanisms of embryogenesis and their interplay . Our model describes how gene expression , cell proliferation and mechanical cell properties contrive to provide structure and patterns to the embryo as it morphs from a single cell to pattern of two tissue types - the trophectoderm and the endoderm . We first discussed how the observed correlation between the directions of the two cell embryo ( animal-vegetal ) and embryonic-abembryonic axes is explained as a mechanical effect of cell mass alignment with the long axis of the ellipsoidal pellucid zone . Cells find their position as they continuously move to decrease the mechanical stress that arises when cells are pushed against each other in a constrained space . Hence mechanical constraints are important in patterning the embryo . While this was shown in connection to blastocoel positioning , in [25] , we have here advanced the understanding by including cell proliferation , explicitly taking into account cell divisions . An advantage of this is that one can address alignment of the embryo continuously at all the stages of development and perform lineage tracking of individual blastomeres . This allows for more detailed comparisons with live imaging experiments . Next we considered the formation of the trophectoderm through two different mechanisms . The first is the position-based hypothesis , which asserts that cells on the boundary of the inner cell mass express Cdx2 , which commits these cells to form the trophectoderm . Hence the position of a cell determines its fate . We also considered the second hypothesis , the polarity based model , where cell division directions are regulated by CDX2 levels , such that cells on the outside of the inner cell mass having high CDX2 levels divide symmetrically , thereby positioning both daughter cells with high CDX2 levels on the outside . For cells with low CDX2 , division occurs asymmetrically , such that one cell is on the outside and inherits more Cdx2 mRNA , whereas the inner cell has less Cdx2 mRNA . In this way gene expression determines the fate of the cell , and ultimately its position within the embryo . Although both models produce enhanced CDX2 levels in outer cells , the position-based model is more robust . This is due to the deficiency of the polarity-based model in dealing with high CDX2 cells in the ICM . Consider in the polarity-based model an inner cell , which due to stochastic fluctuations expresses high CDX2 . Then this cell would divide symmetrically , thereby ultimately supplying inner cells with high CDX2 . Our simulations validate this hypothesis , suggesting that the polarity-based alternative , by itself , is less plausible as the mechanism by which cells find themselves as part of the trophectoderm lineage ( Figure 4 , Figure S4 , Figure S5 , see also discussion in Text S1 ) . Our model suggests that experiments in which either Cdx2 is transiently upregulated in inner cells or Cdx2 is transiently downregulated in outer cells , and the cell division patterns followed by live imaging , would test the hypothesis of this model . Further , in silico the ratio of inner and outer cells could be simulated , thereby elucidating the control of division patterns by CDX2 concentration . Finally we modeled the formation of the endoderm through processes which couple gene expression with the motility of cells . We first considered differential adhesion between cells which express Gata6 , ultimately forming the endoderm layer , and Nanog , which form the ICM . From our simulations , which included a range of adhesion strengths , we inferred that with strong NANOG-NANOG adhesion , weaker GATA6-GATA6 adhesion and still weaker cross adhesion , the two cell populations segregate . We found , however , that in the case of the static blastocoel , the robust formation of endoderm layer required some sort of directional force on Gata6 expressing cells , which could be postulated to arise from a hypothetical signal from the blastocoel . This we implemented through an extra force which moves Gata6 cells towards the blastocoel . We also found that another robust way of a directional bias guiding Gata6 cells towards the blastocoel could be obtained through purely mechanical means . Here we introduced a new model , in which cells in the ICM dynamically pushed in a direction away from the blastocoelic surface . We also found here , that active random motions of the cells allowed the endoderm pattern to be formed more quickly and robustly . Designing an experiment to test which of these two mechanisms does the embryo employ in actually forming the pattern is an interesting question for the future . We should point out that although we do not include apoptosis in our model , we expect that its inclusion would result in the elimination of remaining outlying cells which do not reach the endoderm and could facilitate movement by freeing up space for other cells . Mathematical modeling offers an unique opportunity to test and compare experimentally based hypotheses in a controlled in silico environment . As more detailed data becomes available accurate and predictive models of embryonic development would enhance our understanding of early mammalian embryogenesis . This work is based on different modeling techniques that have been used separately in other contexts , but it is to our knowledge , the first attempt to combine cell-based spatial mechanical simulation with a genetic network approach within the same computational framework to explain mammalian embryogenesis . Within this single framework , we are able to integrate seamlessly several very different temporally separated developmental events ( Video S8 ) . This enabled us to address not only intermediate stages but also the final pattern , as various genetic processes unfold in time . Our model for the formation of different tissue types can be advanced in several directions . We have assumed so far that the ZP is fixed in its geometry . This simplification has allowed us to avoid tracking changes in the shape of the ZP due to the forces between the latter and the cells inside . We aim to implement this feature in the future , which would be important for the later stages of development such as the transition from the blastocyst to the early egg cylinder [3] . In silico tests , such as the application of external forces on the ZP , and the effect on movement of cells and consequently formation of the endoderm layer , would be useful in comparison with experiments . The other challenging problem would be to simulate the changing morphology of trophectoderm cells as they morph from roughly spherical to slightly flattened out cells . The effects of their shape and also tight junctions between them could be an important factor in analyzing in the model . One future goal we plan to pursue is to model the formation and movement of the visceral endoderm [3] . Since our framework allows the implementation of signaling and gene expression networks within cells , it would be interesting to include signaling due to Nodal , bone morphogenic protein ( BMP ) and WNT and their roles in patterning the ICM . As pointed out by Cockburn et al [1] , the formation of the three lineages is crucial for the development of the fetus . We hope that our in silico approach of studying the dynamics of cells due to different hypothesis could ultimately prove useful in a clinical setting . | We elucidate by computational means the processes by which the development of the mammalian embryo during its first four to five days occurs , as it is transformed from a single stem cell into hundreds of cells of different tissue types . We are interested in understanding the fundamental processes of how gene expression dynamics within each cell is coupled to the mechanical forces between cells , such that cells move to take up their positions as part of different tissues depending on the genes they express . Recent experiments which track single cell movement and division in conjunction with their gene expression dynamics suggest various hypotheses as to how this coupling functions to pattern the embryo . We have developed a computational model which can test these hypotheses . The model consists of dividing cells , interacting with each other through mechanical forces , within a confinement of embryo boundary . Each cell contains a genetic network of specific genes which influence cell adhesion properties and cell division plane directions . We explicitly simulate the formation of the trophectoderm and endoderm layers of cells which illuminates the principles by which the embryo is robustly patterned . | [
"Abstract",
"Introduction",
"Model",
"Results",
"Discussion"
] | [
"developmental",
"biology/embryology",
"developmental",
"biology/stem",
"cells",
"developmental",
"biology/morphogenesis",
"and",
"cell",
"biology",
"computational",
"biology/transcriptional",
"regulation",
"biophysics/theory",
"and",
"simulation"
] | 2011 | Simulating the Mammalian Blastocyst - Molecular and Mechanical Interactions Pattern the Embryo |
Allergic reactions can be considered as maladaptive IgE immune responses towards environmental antigens . Intriguingly , these mechanisms are observed to be very similar to those implicated in the acquisition of an important degree of immunity against metazoan parasites ( helminths and arthropods ) in mammalian hosts . Based on the hypothesis that IgE-mediated immune responses evolved in mammals to provide extra protection against metazoan parasites rather than to cause allergy , we predict that the environmental allergens will share key properties with the metazoan parasite antigens that are specifically targeted by IgE in infected human populations . We seek to test this prediction by examining if significant similarity exists between molecular features of allergens and helminth proteins that induce an IgE response in the human host . By employing various computational approaches , 2712 unique protein molecules that are known IgE antigens were searched against a dataset of proteins from helminths and parasitic arthropods , resulting in a comprehensive list of 2445 parasite proteins that show significant similarity through sequence and structure with allergenic proteins . Nearly half of these parasite proteins from 31 species fall within the 10 most abundant allergenic protein domain families ( EF-hand , Tropomyosin , CAP , Profilin , Lipocalin , Trypsin-like serine protease , Cupin , BetV1 , Expansin and Prolamin ) . We identified epitopic-like regions in 206 parasite proteins and present the first example of a plant protein ( BetV1 ) that is the commonest allergen in pollen in a worm , and confirming it as the target of IgE in schistosomiasis infected humans . The identification of significant similarity , inclusive of the epitopic regions , between allergens and helminth proteins against which IgE is an observed marker of protective immunity explains the ‘off-target’ effects of the IgE-mediated immune system in allergy . All these findings can impact the discovery and design of molecules used in immunotherapy of allergic conditions .
Allergy is a hypersensitive immune reaction to environmental antigens from diverse sources such as foods , plants and innocuous organisms . The mechanism responsible for eliciting the allergic reaction involves components of the immune system , in particular the IgE antibody isotype , which also mediate the immune response against helminthic infection . Several significant studies have elucidated the mechanisms involved in the immune response to helminth infection and to allergen exposure , and have been comprehensively reviewed in the literature [1–3] . Extensive studies have correlated high levels of parasite-specific IgE antibody in the host with acquired immunity against both helminth endoparasites such as Platyhelminthes ( Schistosoma and Echinococcus ) [4–8] and nematodes ( hookworms , Trichuris and Ascaris ) [9–11] as well as arthropod ectoparasites ( tick , mites and insects ) [12–14] . Also , IgE cross reactivity has been well established between some allergenic proteins and certain metazoan parasite proteins [15–18] . These immunological assays further suggest that not only are similar immune system components involved in acquiring immunity against helminths and in allergic conditions , but that the molecular targets for these responses may also share key characteristics . Allergenic proteins from a wide variety of sources have been collated and documented in the Allergome database [19] and classified into protein domain families in the Allfam database [20] . The domain families that are populated predominantly by allergenic proteins , represent only around 2% of all protein domain families defined by Pfam [21] . Moreover just 10 protein domain families are reported to represent nearly half of all documented allergenic proteins . Although it has been previously proposed that allergenicity is associated with protease activity [22] and with toxic properties [23] , 9 of these 10 fall into neither category . Host IgE responses against several S . mansoni allergen-like proteins have been previously studied including members of the Tegumental-Allergen-Like ( TAL ) , Tropomysosin and Venom Allergen-Like ( VAL ) protein domain families [24–27] . During natural infection , antibody responses to S . mansoni antigens , including those to members of the TAL and Tropomyosin families have been found to depend on the expression patterns throughout the adult worm and eggs . Constant exposure to antigens may result in the induction of a regulated response against excessive IgE including the decreased production of IgE and increased production of anti-inflammatory IgG4 [27–29] . Such a switch in chronic helminth infections are indicated by a modified T-helper 2 ( Th2 ) cell environment , which is characterized by increased T-regulatory cell levels and a predominant IgG4 antibody profile [30] . Contrary to this , unregulated inflammatory responses are characterized by a hyper-responsive immune system , with higher levels of Th1 and reduced T-regulatory cell numbers accompanied by significantly high IgE levels [2] . However , hypo-responsiveness of the immune system can occur in cases of chronic helminth infections and this has been reported to be beneficial in reducing the inflammatory response caused by further infection of certain bacteria and eukaryotic parasites [31 , 32] . Indeed , infections of Schistosoma mansoni and haematobium have been observed to alleviate the symptoms of allergy [33 , 34] . These studies highlight the inverse relationship between helminth infection and atopy when tested against house dust mites and certain aeroallergens . Based on these observations , Fitzsimmons and Dunne [21] hypothesized that similarity between anti-metazoan parasite and allergic responses may be mirrored in the molecular similarity between allergenic proteins and proteins encoded in genomes of ecto- and endo-parasitic metazoans . Highly specialized immune system components have evolved to combat the effect of infecting metazoan parasites and provide immunity against the infection; however , in the absence of infection , with its attendant immunoregulation ( in atopic individuals ) , this system can switch to the collateral damaging mode and becomes hyper-responsive towards innocuous environmental proteins , possibly due to similar molecular features of the two . Noteworthy , but scarce , examples of studies establishing structure based homology between parasite and allergenic proteins ( e . g . dust mite group II allergen , Der p 2 and Der f 2 sharing structure based homology with carbohydrate-binding module of grass allergen expansin proteins ) provide modest support to our hypothesis [35] . In the quest to strengthen the hypothesis that the IgE-target proteins in metazoan parasites and allergenic molecules share similar molecular features , the systematic exercise of comparing allergenic and parasite proteins and assessing the epitopic regions of allergenic proteins and ‘epitope-like’ regions of parasite proteins becomes of primary importance . Here we present a workflow involving computational analyses supported by experimental verification for detecting putative IgE inducing structure/sequence motifs in proteins encoded in genomes of parasites that share molecular similarity with epitopic regions of members of protein families that are populated predominantly by allergenic proteins .
Two main datasets were generated: Though the Allfam database provides Pfam domain definitions for allergenic proteins documented in the Allergome database , it has not been updated since 2011 . To assign the latest Pfam domain definitions and to include the most recent additions to the Allergome database , we have derived protein domain definitions for dataset 1 and dataset 2 from the Pfam database 27 . 0 . Of these , 10 protein domain families/superfamilies that are reported to represent nearly 45% of all documented allergenic molecules and thus are highly populated with allergenic protein members are considered for further analysis [21] . These protein domain families/superfamilies with their brief description are shown in Table 1 . Homologs of allergenic proteins ( dataset 1a ) have been detected in parasite proteins ( dataset 2 ) by searching Hidden Markov Models ( HMM ) profiles of the above mentioned 10 Pfam protein domain families/superfamilies using HMMER3 [38] . We searched parasite proteins that are categorized in the same superfamily or same fold as allergenic proteins according to the CATH ( Class , Architecture , Topology , Homology ) [39] classification scheme in Gene3D database [40] and collated them . Gene3D assigns domains to regions of protein sequences with no known 3D structure based on the CATH classification scheme using more sensitive structure-based HMMs derived from a superfamily . Parasite proteins that did not show an association with any CATH homologous superfamily were searched in the Gene3D database to ascertain if they shared the same topology/fold as an allergenic protein to establish similarity/homologous relationships . Parasite proteins are considered to be a close homolog of the allergenic proteins if both the proteins belong to same family ( defined by Pfam ) or same homologous superfamily ( defined by CATH/Gene3D ) . However , parasite proteins are regarded as distant homologs if they do not share the same family/superfamily but share a common structure fold with an allergen protein . To establish the relationship of parasite proteins with allergenic proteins , we culled 3D coordinates of experimentally known structures ( if existing in the Protein Data Bank ( PDB ) ) or from homology based models of both proteins ( from Modbase [41] ) and superimposed corresponding structures using least square superimposition program LSQMAN ( http://xray . bmc . uu . se/usf/lsqman_man . html ) . Homology-based structural models that share more than 30% sequence identity with the template sequence ( from any source ) are considered for the analysis . The global structural similarity between two proteins was measured using TM ( template-modeling ) score . A TM score of ≥0 . 5 associated with structure alignment of a pair of proteins is considered to be statistically significant and the two proteins in the alignment are likely to share the same fold [42] . To observe whether these significantly similar epitopic and epitopic-like regions occur on equivalent positions on respective homologous proteins , a comprehensive analysis was performed by manual inspection/visualization comparing their topology on known 3D structure/models of allergenic and parasite proteins , respectively . A summary of the various steps described in the material and methods section is presented in Fig 1 . A Puerto Rican strain of S . mansoni was used in this study . Total RNA was isolated from adult worms and its integrity was verified using a Bioanalyser ( Agilent Technologies , Bracknell , UK ) as previously described [45] . cDNA was prepared from 1μg total RNA using random hexamers ( Sigma-Aldrich , Gillingham , UK ) and Superscript II reverse transcriptase according to the manufacturer’s instructions ( Life Technologies , Paisley , UK ) . SmBv1L was cloned into the pET15b vector ( Merck , UK ) . PCR to generate a gene-specific coding sequence was performed using Phusion High Fidelity DNA polymerase ( Thermo Fisher Scientific , Reading , UK ) using the following primers , 5’- TAGGATCCTATGAATGCATATATTATTCG and 5’-ATGGATCCTTATCTAGAGTCGGA at an annealing temperature of 62°C . The PCR product and plasmid were digested with FastDigest BamHI ( Thermo Fisher Scientific ) and plasmids were treated with FastAP ( Thermo Fisher Scientific ) . Ligation was performed using T4 DNA ligase ( Thermo Fisher Scientific ) and plasmids transformed into chemically competent DH5alpha E . coli cells by heat shock . Sanger Sequencing was used to confirm the CDS sequence that was uploaded to GenBank ( Accession No: KM281668 ) . Recombinant SmBv1L was produced by isolation from inclusion bodies and on-column refolding using an AKTAPrime+ according to the manufacturer’s instructions ( GE Healthcare ) . Briefly , overnight cultures of TG2 E . coli were diluted 1:10 , expanded to OD 0 . 4–0 . 8 and induced for 3 hours with 1mM IPTG . Pelleted cells were lysed using a French Pressure Cell at ≈10 , 000 psi and the pellet containing inclusion bodies were isolated by centrifugation at 15 , 000 x g for 1 hour . Pellets were washed 3 times in cold isolation buffer ( 2 M urea , 20 mM Tris-HCl , 0 . 5 M NaCl , 2% Triton-X 100 , pH 8 . 0 ) and then were incubated in binding buffer ( 6 M guanidine hydrochloride , 20 mM Tris-HCl , 0 . 5 M NaCl , 5 mM imidazole , 20 mM β-mercaptoethanol , pH 8 . 0 ) overnight at 4°C . Purification was performed by on-column refolding on a HisTrap FF 1 ml column ( GE Healthcare ) , after equilibration of the column and sample binding , the column was washed with solubilisation buffer ( 6 M urea , 20 mM Tris-HCl , 0 . 5 M NaCl , 5 mM imidazole , 1 mM β-mercaptoethanol , pH 8 . 0 ) and the bound protein refolded in a gradient of refolding buffer ( 20 mM Tris-HCl , 0 . 5 M NaCl , 5 mM imidazole , 20 mM β-mercaptoethanol , pH 8 . 0 ) and solubilisation buffer . Protein was eluted from the column in a gradient of elution buffer ( 20 mM Tris-HCl , 0 . 5 M NaCl , 0 . 5 M imidazole , 20 mM β-mercaptoethanol , pH 8 . 0 ) and collected in fractions 7–20 . Refolded SmBv1L was concentrated and buffer exchanged to resuspension buffer ( 20 mM Tris-HCl , pH 8 . 0 ) , attached to 1 mL washed HisTrap FF and protein eluted by incubation with 150 U/mg protein Thrombin ( GE Healthcare ) . Thrombin was removed by incubation with benzamadine-agarose beads ( Sigma-Aldrich ) and the protein buffer exchanged to PBS . Proteins were tested for bacterial contamination by in-house ELISA assay as previously described [24] . Venous plasma samples were collected in the fishing village of Namoni on the shores of Lake Victoria , Mayuge District , Uganda , an area of high year-round schistosomiasis ( S . mansoni ) transmission occurred . The 372 plasma donors ( 6–40 years , 57% female , mean age 18 ) were randomly selected from community members who had detectable S . mansoni eggs in a single stool sample ( microscopically examined in two Kato-Katz thick smears ) . Blood was collected in heparinized tubes at two time-points , immediately before anti-schistosomiasis treatment with praziquantel ( PZQ , 40mg/kg body mass ) and 5 weeks post-treatment . Plasma separated by centrifugation ( 2000 x g for 5 min ) was stored at -80°C until use . This analysis focuses on 222 individuals with full parasitological and serological information available both pre- and post-treatment ( age 6–40 years , mean age 16 years , 59% female ) . Antigen-specific IgG1 , IgG4 and IgE levels were measured in the plasma from infected individuals in duplicate by ELISA as detailed previously [24] with the following modifications . SmBv1L was applied to plates in sodium bicarbonate coating buffer at 4°C overnight at a concentration of 0 . 875 μg/ml . Human plasma was diluted 1:40 for IgE assays and 1:100 for IgG isotypes . Data handling and statistical analysis was performed using STATA for Mac version 12 . 1 . Area proportional Venn diagrams were drawn using eulerAPE version 1 . 0 [46] .
The distribution across different taxonomic groups of the 2712 allergenic molecules listed in the Allergome database for which IgE-binding antigen is reported suggests that allergenic/IgE-binding protein molecules are moderately represented in fungal ( 17 . 5% of all Allergome entries ) genomes and poorly represented in bacterial genomes ( ~1% ) ( Fig 2 ) and are consistent with the previously published studies [47 , 48] . A large proportion of these molecules are represented in plant ( ~38% ) and metazoan genomes ( 44% ) . Interestingly , these large proportions of allergens in metazoa and plants ( 1018 plant and 1188 metazoan allergens ) are encoded in genomes of a restricted number of species ( 201 plant and 373 metazoan species , respectively ) . The 2712 unique UniProt protein sequence entries corresponding to protein molecules listed in the Allergome database are distributed in 331 Pfam protein domain families . Protein domain families with allergenic entries are relatively few and account for merely 2 . 2% of total 14831 protein domain families specified in Pfam database ( Fig 3 ) . There are 128 protein families that are comprised of only one allergenic protein member . A total of 1389 out of 2712 allergenic molecules ( ~51% ) are members of just 20 protein domain families ( Fig 4 ) . Among these families , Tropomyosin ( Pfam accession: PF00261 ) constitutes 217 allergenic molecules ( ~8% of all allergens ) . A large proportion of known epitopic regions in proteins have been determined by employing a 'peptide fragment approach' . The IEDB database provides a curated list of these epitopic regions along with a limited set of epitopes that are determined by analysing structures of antigen-antibody complexes . The majority ( 58% ) of linear epitopic regions in the IEDB database demonstrated to bind to IgE/IgG4 are within peptides of 10 and 15 amino acid residues length . This is consistent with the average length of an epitope determined based on analysing structures of antigen and antibody complexes [49] . We have chosen 10 protein domain families/superfamilies that have been referred to in the ‘Opinion article’ co-authored by Fitzsimmons and Dunne [21] for our analysis . These families/superfamilies , encompassing nearly 40% of all allergenic proteins , differ in terms of their distribution across different taxonomic groups as well as experimental details available , and can be categorized into three groups ( A , B , C ) depending on: ( i ) whether protein families/superfamilies include parasite protein representatives as well as known allergens; and ( ii ) whether IgE-binding immunoassays have/have not been performed on parasite proteins ( Table 2 ) . Illustrative examples of newly detected metazoan parasite protein which are potential IgE target are discussed in the following section . The plant-specific Bet v1 domain family represents the second large family which is highly populated with allergens ( see Fig 4 ) . However , no Bet v1-like IgE target proteins have been reported in metazoan parasites so far . We tested the potential of Bet v 1-like protein ( SmBv1L ) from S . mansoni predicted by our computational analysis to bind IgE and IgG4 in plasma from individuals infected with this parasite . The Bet v 1-like protein from S . mansoni ( Uniprot accession: G4VE06 ) and the plant birch pollen allergen ( Uniprot accession: P15494 , PDB accession: 1BV1 ) are categorised in the same homologous superfamily ( CATH accession: 3 . 30 . 530 . 20 ) as described by the CATH database . SmBv1L was cloned and expressed as a recombinant protein in E . coli . Insoluble recombinant protein was extracted from inclusion bodies , purified and refolded on column to give soluble antigen . The presence of a single protein species of approximately 18 kDa , as predicted in silico , was confirmed by a Coomassie blue-stained SDS-PAGE gel . Antibody responses against SmBv1L were measured in a population of 222 S . mansoni-infected individuals from a parasite endemic area in Uganda . Levels of antigen specific responses were measured against IgG1 , IgE and IgG4 ( Fig 8 ) . SmBv1L was found to be antigenic with 38 . 7% of the population responding with IgG1 , IgG4 or IgE . Antibody responses against SmBv1L with IgG1 , IgG4 and IgE were found in 23 . 9% ( 53 ) , 19 . 8% ( 44 ) and 16 . 7% ( 37 ) of the population respectively . The geometric mean magnitudes of responder antibody levels were as follows IgG1; 70 . 2 μg/ml ( 95% CI 57 . 9 , 85 . 1 μg/ml ) , IgG4; 0 . 7 μg/ml ( 95% CI 0 . 5 , 0 . 9 μg/ml ) and IgE; 56 . 6 ng/ml ( 95% CI 40 . 4 , 79 . 5 ng/ml ) . As shown by Fig 9 , individuals were capable of producing responses against IgG1 , IgG4 or IgE alone or any combination of the three . There were no significant differences between the overlaps of any combination of the three antibody isotypes .
The Ugandan National Council of Science and Technology provided ethical clearance for the human serology studies . Consent forms were translated in the local language and informed written consent was obtained from all adults and from the parents/legal guardians of all children under 15 . | Allergy is an increasingly widespread clinical problem that leads to various conditions such as allergic asthma and susceptibility to anaphylactic shock . These conditions arise from exposure to a range of environmental and food proteins ( ‘allergens’ ) that are recognised by a form of immune system antibody called IgE . This part of the immune system is thought to have evolved to provide mammals with additional rapid response mechanisms to combat metazoan parasites . Here , we address the pertinent question , ‘what makes an Allergen an Allergen’ as , although they constitute a very small percentage of known proteins , they appear to be diverse and unrelated . Using computational studies , we have established molecular similarity between parasite proteins and allergens that affect the nature of immune response and are able to predict the regions of parasite proteins that potentially share similarity with the IgE-binding region ( s ) of the allergens . Our experimental studies support the computational predictions , and we can present the first confirmed example of a plant pollen-like protein in a worm that is targeted by IgE . The results of this study will enable us to predict likely allergens in food and environmental organisms and to help design protein molecules to treat allergy in the future . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | Comparisons of Allergenic and Metazoan Parasite Proteins: Allergy the Price of Immunity |
Resistance of the snail Biomphalaria glabrata to the trematode Schistosoma mansoni is correlated with allelic variation at copper-zinc superoxide dismutase ( sod1 ) . We tested whether there is a fitness cost associated with carrying the most resistant allele in three outbred laboratory populations of snails . These three populations were derived from the same base population , but differed in average resistance . Under controlled laboratory conditions we found no cost of carrying the most resistant allele in terms of fecundity , and a possible advantage in terms of growth and mortality . These results suggest that it might be possible to drive resistant alleles of sod1 into natural populations of the snail vector for the purpose of controlling transmission of S . mansoni . However , we did observe a strong effect of genetic background on the association between sod1 genotype and resistance . sod1 genotype explained substantial variance in resistance among individuals in the most resistant genetic background , but had little effect in the least resistant genetic background . Thus , epistatic interactions with other loci may be as important a consideration as costs of resistance in the use of sod1 for vector manipulation .
Although vector-borne diseases account for approximately one-sixth of the global human disease burden [1] , [2] , we still lack effective drugs and vaccines for many of these diseases . Even when effective drugs are available , high-risk populations often cannot be adequately treated due to a lack of funding and infrastructure in the heavily impacted countries [1] , [3] . Therefore , in the absence of vaccines , eradication efforts that include both drug therapy and vector control can be the most effective approach [4] . Vector control methods most often utilize chemicals for eradication [1] , [4] . This approach has obvious drawbacks because it results in habitat degradation and risk of human exposure to pesticides . Also , recurrent pesticide application is often necessary because it is nearly impossible , with a single treatment , to completely remove all possible vector individuals from an epidemiologically relevant site [5] . Recent advances in understanding the genetics of host-parasite interactions have led to increased interest in driving resistance genes into susceptible vector populations [6]–[11] . In this context , the term “resistance” describes a continuously varying trait we define as the probability of becoming infected after being challenged by a parasite , rather than to mean the absolute inability to become infected ( i . e . a population or genotype can have high or low average resistance ) . Making vector populations more resistant to infection could be a better long-term solution and an ecologically safer way of breaking transmission cycles . Unfortunately , this approach faces major population-genetic hurdles . A non-exhaustive list includes: ( 1 ) genotype-by-environment ( GxE ) , where the performance of a gene or gene ( s ) of interest depends on environmental conditions such that interactions can affect how a resistance gene performs in the field versus in the lab [12]–[16] , ( 2 ) parasites and hosts are genetically more variable in the field , and there can be interactions between host genotypes and parasite genotypes ( genotype-by-genotype ( GxG ) interactions; [16]–[19] ) , ( 3 ) genetic background can influence how a resistance gene performs in a natural versus a lab population . In other words , the gene of interest may perform differently depending on the genomic context in which it is interacting ( epistasis ) , and ( 4 ) , there may be a cost of resistance such that natural selection in the absence of parasites favors the “wild-type” alleles that we wish to replace . Cost of resistance may be a particularly vexing problem for resistance-gene introduction programs . Such costs have been demonstrated in many host-parasite systems ( reviewed in [20]–[26] ) . Nevertheless , costs of resistance are not universal [8] , [27]–[31] , and they may be context dependent ( e . g . revealed only in stressful environments; [12] , [32]–[36] ) . Costs of resistance presumably involve a reallocation of metabolic resources between one or more of the following life-history components: reproduction , growth , and somatic maintenance/immune function [24] , [26] , [37] , [38] . Also , the severity of the cost should depend on the particular mechanism of resistance [29] , [39] . For example , it was predicted that mechanisms involving over-expression of particular genes might be among the most costly [39] . This study was designed to measure costs of resistance and epistatic effects of genetic background associated with a single locus in Biomphalaria glabrata , a snail vector of the human pathogen Schistosoma mansoni . Schistosomiasis is responsible for approximately 200 , 000 deaths yearly , with 200 million people infected worldwide [40]–[42] . B . glabrata is a facultative , hermaphroditic freshwater pulmonate snail that occurs throughout much of the New World tropics [43]–[45] . The B . glabrata/S . mansoni system is a well-established model for investigating host-parasite interactions in a controlled laboratory setting [46] . Resistance to S . mansoni infection in B . glabrata is highly heritable in many lab and field populations , and is almost certainly controlled by multiple loci [47]–[52] . The expression patterns of known immune-related genes have been found to differ between individuals from more resistant and less resistant strains when each is challenged with the same strain of parasite [53]–[59] . However , to date only a single locus has been identified at which allelic variation clearly associates with resistance to the parasite: copper-zinc superoxide dismutase ( sod1 ) [60] , [61] . SOD1 is a ubiquitous protein involved in several cellular functions including signaling and immune response [62]–[65] . Among the various functions of SOD1 , it catalyzes the reduction of highly reactive superoxide ( O2− ) to hydrogen peroxide ( H2O2 ) . Hydrogen peroxide is a known cytotoxic component of the oxidative burst , which is the primary defense mechanism for parasite clearance in molluscs [46] , [66] , [67] . When a schistosome invades a snail , hemocytes surround the invading parasite and are thought to generate H2O2 as part of the killing mechanism [46] , [66] , [68] . Consistent with this hypothesis , increased H2O2 production was correlated with the difference in resistance between snails from the M-line strain and the more resistant 13–16-R1 strain [46] , [68] . An sod1 allele present in the 13–16-R1 strain was over-expressed relative to the other alleles , and correlated with a more effective defense against parasite infection [46] , [61] , [69] . More recently , Moné et al . [70] demonstrated a correlation between the ability of certain strains of B . glabrata to produce reactive oxygen species and the anti-oxidant defenses of their respective compatible S . mansoni strains . Thus , loci involved in the oxidative burst , such as sod1 , may be very important in the evolution of schistosome-snail interactions . Therefore , sod1 is a promising candidate locus for driving resistance alleles into susceptible natural populations of snails . Although sod1 seems a favorable candidate for genetic manipulation of snail populations , there are two reasons why one might expect a cost of resistance associated with the allelic polymorphism at sod1 . First , increased expression of any gene is likely to be costly [39] . Second , increased expression of sod1 might incur a cost due to increased oxidative stress on the host [71] , [72] . Therefore , investigating the fitness costs associated with allelic variation at sod1 is an important first step in evaluating the potential use of sod1 for creating highly resistant vector populations in the field .
We used a population of the 13–16-R1 strain of B . glabrata that has been maintained as a large population ( hundreds ) in C . J . Bayne's lab at Oregon State University since the mid-1970s . The 13–16-R1 strain was reportedly created by crossing highly resistant strains of snails isolated from Brazil and Puerto Rico [47] but it has been in culture for so long in so many laboratories that its history is not entirely clear . Our population has been maintained in the absence of parasite exposure , and therefore under relaxed selective pressure in regards to resistance to S . mansoni . B . glabrata is a facultative self-fertilizing hermaphrodite such that snails will preferentially outcross when given access to a mate , but when isolated will usually reproduce through self-fertilization ( e . g . [73]–[75]; our laboratory population is in Hardy-Weinberg Equilibrium for sod1 and microsatellite loci: [61] , [69]; unpub . data ) . We recently created 52 inbred lines: we started with haphazardly picked juvenile snails and completed three generations of selfing using a single offspring from each self-fertilization event to begin the next generation . The inbred lines are mostly fixed for one of three alleles of sod1 A , B and C , as described in [61] . These lines also vary substantially for resistance within each sod1 genotypic class ( AA , BB , and CC ) . That there are highly resistant and highly susceptible lines within each sod1 class suggests that other loci besides sod1 have a large effect in determining resistance . These inbred lines can be used to compare directly the fitness effects of carrying a specific genotype at sod1 and the effects of genetic background on the association between resistance and sod1 genotype . Several inbred lines were used to create three outbred F2 populations , each of which was segregating for the B and C allele ( Figure 1 ) . We hereafter refer to these three F2 populations as “genetic backgrounds” because we wanted to know if the phenotypic effects of variation at sod1 depend on the genomic context in which those alleles are expressed . These F2 individuals were then used to evaluate the effects of sod1 allele on life history traits and resistance . Inbred lines were chosen so that the three populations differed in average resistance . BB and CC fixed lines were chosen because the B allele confers the highest resistance and the C allele the lowest [61] . Additionally , in hemocytes ( the defense cells ) the B allele is constitutively over-expressed relative to the other two alleles [69] . To create the three F2 populations , we paired an individual from an inbred line fixed for the B allele with an individual from an inbred line fixed for the C allele ( BB×CC ) , which resulted in offspring that were heterozygous at sod1 ( BC ) . Three unique BB and CC inbred lines were used , and each cross was completed in triplicate with unique individuals ( n = 9 crosses ) . To compare directly the effects of carrying the BB and CC genotypes within a family and among different backgrounds , we paired heterozygous offspring from each initial cross with a heterozygous individual from a different initial cross using a factorial design . This resulted in three different F2 populations of outbred individuals that had the same sod1 genotypes , but in different genetic backgrounds ( Figure 1 ) . The F2 individuals in each of the three populations carried the BB , BC and CC genotypes in the expected ( 1∶2∶1 ) Mendelian ratios ( sod1 genotypes were verified by sequencing ) . We used these F2 individuals to test the effects of sod1 genotype on fecundity , growth and resistance in each of the three genetic backgrounds . Our three populations ( genetic backgrounds ) differed in overall resistance ( 77 . 8% , 63 . 8% , and 38 . 9% ) , which strongly correlated with the resistance of their grandparents ( the original inbred lines ) ( Figure 2 ) . For each F2 population ( genetic background ) , a total of 72 individuals were haphazardly chosen from a pool of offspring from the final set of crosses . We exposed single juvenile snails ( 4–5 mm diameter ) to five S . mansoni strain PR-1 miracidia in 3 mL of artificial spring water ( ASW; [76] ) for two hours at 26°C , in 12-well culture plates . The PR-1 strain has been maintained in Syrian hamsters and the M-line ( Oregon ) strain of B . glabrata snails by the Bayne lab for 36 years . Challenged individuals were then reared in moderately dark tubs in groups of 24 , with three replicate tubs for each background ( n = 72 ) . We examined the snails for infection at six , nine , and eleven weeks ( we rarely see shedding after 11 weeks ) . Each examination week we induced cercarial shedding ( parasite emergence ) by exposing snails individually in 3 mL of ASW to direct fluorescent light for two hours at 26°C in 12-well culture plates . The presence of cercarial shedding indicated a positive infection . Infected snails were preserved in 95% ethanol ( EtOH ) , and non-infected snails were returned to rearing tubs after each assay . After the final cercarial shedding attempt ( eleven weeks ) we preserved the remaining snails , and all tissue samples were processed for sod1 genotyping ( described below in ‘Molecular Methods’ section ) . Resistance to parasite infection was scored in each tub group as the percentage of snails that did not shed cercariae by eleven weeks post-challenge . Snails that died prior to shedding assays were excluded from the experiment . Average mortality observed from the parasite challenge ranged from 8–12% among tubs , and did not differ among genetic backgrounds ( One-way ANOVA , p = 0 . 442 ) . We collected single egg masses ( n = 58 ) from Styrofoam substrate within 48-hours of egg mass deposition from individual pairs of the final set of crosses ( i . e . embryos in the eggs are F2s ) . The single egg masses were reared individually and allowed to hatch . We measured offspring size ( diameter of the shell ) twelve weeks after egg mass deposition . All snails were then preserved in 95% ethanol for subsequent sod1 genotyping . Clutch sizes ( the numbers of eggs/embryos in single egg masses ) ranged from 2 to 34 ( n = 58 ) . Initial analysis revealed that average offspring size was correlated with clutch size , ( adjusted R2 = 0 . 363 , P<0 . 001 ) suggesting a strong density-dependent effect of number of snails per bowl on growth ( same effect across all genetic backgrounds ) . Therefore , we restricted our analysis of effects of sod1 genotype to the offspring of clutch sizes between 13–17 eggs/embryos ( there was no association between clutch size and snail size within that limited range of clutch sizes; adjusted R2 0 . 001 , P = 0 . 28 ) . We compared snail growth from 3–4 clutches in each genetic background ( background 1: n = 45 , background 2: n = 57 , background 3: n = 58 ) . We also measured growth ( shell diameter ) in snails that were raised individually for 32 weeks as part of the egg production and hatch success experiments described below ( hereafter referred to as “late growth” compared to the “early growth” measures described in the above experiment ) . As in the growth study , we collected egg masses from individual pairs of the final crosses ( i . e . the F2 offspring ) . From each population , we haphazardly chose 50 sexually immature offspring ( 4–5 mm shell diameter ) . Each snail was reared singly and a portion of a tentacle was excised to determine its sod1 genotype . We then randomly chose ten juveniles of each genotype ( BB , BC , and CC ) from each set of 50 genotyped snails , and reared them individually for subsequent fecundity comparisons ( i . e . n = 30 per genetic background ) . Because B . glabrata is a facultative self-fertilizing hermaphrodite , we provided a mate to each snail prior to measuring egg production and hatch success to ensure offspring were not the result of selfing ( because inbreeding depression is expected to affect egg survival ) . We chose to mate the genotyped individuals with snails from an isogenic inbred population to keep consistent the relative contribution of the “male-acting” snail to egg production . The isogenic inbred individuals were from a population of inbred M-line strain of B . glabrata established at the University of New Mexico through 32 generations of selfing ( Si-Ming Zhang pers comm . ) . Because the M-line and F2 offspring look morphologically similar , we marked the M-line snails with a white dot using nail polish 24 hours prior to mating . All snails were individually reared until reproductively active , as determined by the presence of well-formed egg masses containing developing embryos . B . glabrata preferentially use allosperm for fertilization and store sperm for up to 10 days [74] . Consequently , each snail was paired with a size-matched , painted , inbred M-line individual for one week , then separated and allowed to lay eggs for one week in a new cup . These eggs were thus presumably fertilized by allosperm , even though layed in the absence of a partner [73]–[75] . Egg numbers were counted at the end of each 1-week laying period , after which snails were re-paired with a different mate . We continued the mating/laying schedule for ten weeks , resulting in five one-week accumulated egg production measurements from each snail . We present the sum of the five one-week egg accumulation measures as the total egg production for each snail over five weeks . We examined egg hatch success in the same set of genotyped individuals in which we surveyed egg production . Each snail was paired with a size-matched painted inbred M-line individual for 48 hours , and then isolated in a new cup . Two egg masses from each snail were carefully collected 72 hours post-transfer and reared individually ( n = 180 ) . Egg masses were surveyed for total egg count upon collection , and final hatch counts were conducted six weeks later . Hatch success ( percent of eggs hatched at six weeks ) from the two egg masses was averaged for each snail . In addition to measuring egg production and egg hatch , we also monitored mortality at eight and twelve months in the same set of F2 snails used for the egg production and hatch success experiments . Mortality was measured as percent of individuals from each sod1 genotype alive at the time of census for each genetic background . All snails were reared in an environmentally controlled room kept at 26°C and on a 12 hr day/12 hr night light cycle with full spectrum light . Snails were fed green leaf lettuce ad libitum throughout all experiments . In experiments other than those in which we measured resistance , egg masses and snails were reared , mated , and maintained in 500 mL cups with 300 mL of ASW . Complete water changes were carried out weekly . When generating the three different populations ( i . e . the three different genetic backgrounds ) for the fecundity experiments , the egg masses ( and offspring ) were reared in 2 L of ASW in aerated , lidded 1-gallon , clear plastic boxes ( IRIS , USA ) . The egg masses monitored in the hatch success experiment were reared in petri-dishes ( 100×15 mm ) with 5 mL of ASW . Finally , in the resistance assay we reared exposed snails in moderately dark , lidded 3-gallon plastic tubs ( Dark Indigo Rubbermaid Roughneck boxes ) . Each contained 7 . 5 L of aerated dechlorinated water supplemented with 10 mL of calcium carbonate shell hardening solution ( 30 mg Ca++/L ) . Half of the water was changed with dechlorinated water between each infection assay . We extracted genomic DNA from snail head foot tissue following the CTAB protocol [77] , and used chelex extraction methods for tentacle tissue . sod1 genotype was determined using fragment analysis on an ABI 3730 capillary sequencer following amplification with AmpliTaq ( Applied Biosystems , Inc . ) ( F- ( VIC ) - TCA TTG GTC GCA GCT TAG TG , R - GTC CTG TCA TGT AGC CAC CA ) . The B and C alleles are differentiated by a two base-pair ( bp ) insertion/deletion in the fourth intron that is fully resolved by the capillary system ( the full sequences for the fourth intron are available for the B and C allele on NCBI GenBank from [61] ) . Sequence analysis of a subset of samples corroborated fragment analysis methods . Fragment analysis peaks were visualized using GENOTYPER ( Applied Biosystems , Inc . ) , and sequence data were analyzed using SEQUENCHER ( GeneCodes , Inc . ) . Data were assessed for normality ( Shapiro-Wilk ) and equal variance . To examine the effects of genetic background on the association between carrying the B allele and resistance to parasite infection we used generalized linear models ( logit function ) to compare resistance ( coded as a binomial response for each snail , infected = 1 , not infected = 0 ) among genetic backgrounds and sod1 genotypes . We used regression coefficients from individual logistic regressions to quantify the relative effect sizes of substituting one allele for another in each of the genetic backgrounds . We compared fitness measures ( growth rate , egg production , and hatch success ) among genetic backgrounds and genotypes using two-way ANOVAs and Tukey post-hoc tests . For mortality we used generalized linear models ( logit function , surviving snail at time of census = 1 , dead snail = 0 ) . No transformations were needed to normalize any of these data . We defined significance at the level of alpha = 0 . 05 . For data analyses , we used the statistical packages SPlus version 8 . 1 for Windows ( TIBCO Software , Inc ) and SigmaPlot for Windows version 11 . 0 ( Systat Software , Inc ) .
We found main effects of genotype and genetic background , and a background-by-genotype interaction ( logit GLM; background: P = 0 . 09 , genotype: P = 0 . 003 , background×genotype: P = 0 . 022 ) . As expected , the B allele was most protective . However , the strength of the association between sod1 genotype and resistance to infection depended on genetic background . The association was strongest in genetic background 1 and there was a similar but non-significant trend in background 2 . In contrast , allelic variation at sod1 explained little of the variance in resistance in background 3 ( Figure 3 ) . Substituting a B allele for a C allele decreased the odds of infection by 6 . 2 in genetic background 1 , and by 2 . 5 in genetic background 2 ( logit GLM; P = 0 . 0027 and 0 . 0477 , respectively ) . In genetic background 3 there was no significant additive effect . Thus , the effect of allelic variation at sod1 on resistance to infection was most important in predicting infection in the genetic background having high average resistance , and was largely irrelevant in the low-resistance genetic background . With regard to early growth ( size at 12 weeks ) , we found significant main effects of genetic background and sod1 genotype , but no interaction effect . Surprisingly , individuals with the CC genotype were smaller , on average , than those with BB and BC genotypes ( two-way ANOVA; background: F2 , 151 = 11 . 07 , P<0 . 001; genotype: F2 , 151 = 8 . 11 , P<0 . 001; background×genotype: F4 , 151 = 0 . 68 , P = 0 . 991 ) ( Figure 4A ) . Thus the B allele was associated with faster growth and appeared almost completely dominant to the C allele for this trait ( Figure 4A ) . For late growth ( size at 32 weeks ) , we again found significant main effects of genetic background and genotype , and no interaction ( two-way ANOVA; background: F2 , 75 = 39 . 8 , P<0 . 001; genotype: F2 , 75 = 3 . 68 , P = 0 . 030; background×genotype: F4 , 75 = 1 . 54 , P = 0 . 20 ) . The CC individuals were still smaller than the BC and BB individuals , and the B allele appeared to act dominantly ( Figure 4B ) . In regard to egg production , we found a main effect of genetic background , but no main effect of sod1 genotype and no significant interaction ( two-way ANOVA; background: F2 , 73 = 6 . 11 , P = 0 . 0035; genotype: F2 , 73 = 0 . 533 , P = 0 . 59; background×genotype: F4 , 73 = 0 . 472 , P = 0 . 756 ) . The BB genotype had the lowest estimated fecundity in genetic backgrounds 1 and 2 , but the CC genotype had the lowest in background 3 ( Figure 4C ) . However , we examined only 10 individuals per genotype within each genetic background , and thus had low power to detect all but strong main or interaction effects , as evidenced from a post-hoc power analysis . Our calculated effect size for the main effect of genetic background was 0 . 432 , while effect sizes for the main effect of genotype and interaction were only 0 . 15 and 0 . 17 , respectively . Additionally , our calculated power was 0 . 95 for the main effect of genetic background but only 0 . 22 and 0 . 27 for the main effect of genotype and for the interaction , respectively . Thus , an effect of sod1 genotype on fecundity would have had to be much stronger than observed to be detected with our sample sizes . Average hatch success across all genetic backgrounds was 49% , and varied from 35% to 62% among genotypes ( Figure S1 ) . We did not find a significant main effect of genetic background or genotype on hatch success ( two-way ANOVA; background: F2 , 60 = 0 . 47 , P = 0 . 62; genotype: F2 , 60 = 1 . 52 , P = 0 . 23; background×genotype: F4 , 60 = 0 . 99 , P = 0 . 42 ) . Thus , the B allele does not incur an obvious fitness cost associated with egg production ( Figure 4C ) or offspring hatch success . We note that although our average hatch rate of 49% is on the low side of rates reported in the literature , it is not unusually low ( e . g . [78] ) . At the 8-month census we found significant main effects of both genetic background and genotype on mortality ( logit GLM , background: P = 0 . 002 , genotype: P = 0 . 04 ) , but no interaction ( drop-in-deviance test , P = 0 . 19 ) . CC individuals exhibited greater mortality , averaging 37% across genetic backgrounds , whereas BB and BC average 17% and 13% respectively ( Figure 4D ) . At 12 months we again found a significant main effect of genetic background , but the genotype effect was no longer significant ( logit GLM , background: P = 0 . 02 , genotype: P = 0 . 18 ) , and there was no interaction ( drop-in-deviance test , P = 0 . 39 ) . These results suggest there is no cost to having the B allele in terms of increased mortality , and a possible advantage in early survival ( Figure 4E ) .
The association between allelic variation at sod1 and resistance to infection varied substantially among genetic backgrounds . The three genetic backgrounds differed in average resistance ( 78% , 64% , and 39%; Figure 2 ) . sod1 genotype was most predictive in the genetic background having the highest average resistance , and had a negligible effect in the genetic background having the lowest average resistance ( Figure 3 ) . Thus , sod1 appears to interact epistatically with other genes that influence resistance , a result that might help us identify those other loci . That there are other resistance loci segregating in the 13–16-R1 population is evident because inbred lines having identical sod1 genotypes vary substantially in resistance ( Bender and Larson , unpublished observations ) . Through gene expression studies , several other loci have been identified in B . glabrata as being potentially immune relevant [53]–[59] , and various physiological differences have been noted between snail strains having high or low resistance to trematode parasites ( reviewed in [67] ) . However , candidates that seem particularly likely to interact with sod1 as observed here include loci encoding proteins involved in non-self recognition and loci that control other steps in the oxidative burst pathways . Recognition loci are suggested because , as part of the effector mechanism used by the host to attack the parasite , sod1 would come into play only after the parasite has been recognized . Thus , sod1 genotype would be irrelevant in a low-recognition background , but very important in a high-recognition background . Possible recognition loci include lectin-like molecules such as FREPs [79] . Loci affecting numbers or some other property of hemocytes might also behave epistatically with sod1 in a similar manner such that if hemocytes were incompetent ( or insufficient in number ) to encapsulate the parasite , their ability to produce H2O2 would be irrelevant . Costs of resistance have been demonstrated in many systems [21]–[26] . Even in B . glabrata , there is some evidence that strains with higher resistance to schistosomes differ from strains with lower resistance in components of fitness [49] , [50] , [80]–[85] . Furthermore , relative to the A and C alleles , the B allele of sod1 is over-expressed . The SOD1 protein produces H2O2 , a highly reactive species with the potential to damage host tissue as well as the parasite [69] . Thus , it would be no surprise to see a cost of resistance associated with the B allele at sod1 . Nevertheless , here we failed to detect any disadvantage due to the B allele in terms of reproduction , and observed an advantage over the C allele in terms of growth rate and survival to 8 months post-hatch ( Figure 4 ) . Furthermore , there were no significant interactions between sod1 genotype and genetic background with regard to life history traits . It is also interesting that the B allele acted dominantly to the C allele for growth rate ( Figure 3 ) , a result that might be expected if the difference really results from over-expression of the B allele . Given our data suggest that the B allele may confer a slight advantage in terms of growth and early survival , one might wonder why our population has not become fixed for the B allele . Possible explanations include: ( 1 ) this laboratory maintained population is not in equilibrium and the selection pressure is not strong enough to have driven the allele to higher frequency yet ( we have no data on allele frequencies of sod1 at the founding of this laboratory population ) ; ( 2 ) there may be costs to having the B allele in other components of fitness that we did not measure; ( 3 ) perhaps there are complex interactions among the three major alleles in the population ( A , B , and C ) that prevent the B allele from increasing in frequency ( e . g . see p 223–225 in [86] ) . We showed the promising result of no obvious cost , and perhaps a life history trait advantage for the more-resistant allele at sod1 . Obvious caveats include that our experiments were conducted in a ( presumably benign ) laboratory setting , and would need to be replicated under field conditions . Other studies have found that costs of resistance are more likely to manifest under specific environmental conditions , such as low food and temperature stress [12] , [32] , [35] , [36] . Of perhaps greater concern is the strong epistatic effect on resistance between sod1 and other loci in the genome . Defeating an attempted infection is a complex process that involves many steps including recognition , signaling and implementing the effector ( killing ) mechanisms . SOD1 can participate in both signaling and effector mechanisms , and the products of many loci may need to interact properly to sufficiently clear an infection . Thus , it will be essential to assess the performance of sod1 in the field and in a variety of other genetic backgrounds . There are also a number of basic questions , unrelated to those addressed here , about sod1 and resistance to S . mansoni that need to be answered before one could seriously consider using sod1 for vector manipulation in the field . We still need to prove that the association between resistance and sod1 alleles is actually causal , and if so , if the protective effect of allele B is really owing to its overexpression . It is theoretically possible that sod1 is not the actual causal locus , but is just in strong linkage disequilibrium with a closely-linked locus that actually controls resistance . This seems unlikely given the association between sod1 genotype and resistance was discovered using a functional approach ( e . g . knocking down H2O2 production in B . glabrata hemocytes increases their susceptibility to infection [66] ) , but the functional basis of the association still needs to be proven . Additional work to test the causality of the association is underway . In the unlikely event it turns out that another locus is actually causal , then the results of this study are still quite relevant , but for the new locus of interest . We also do not know yet if the effect of sod1 we observed is generalizable to other populations/strains of S . mansoni . We have only studied the PR-1 strain of S . mansoni in interaction with the 13–16-R1 population of B . glabrata . It is possible that the protective effect of sod1 alleles depends on the strain of parasite in addition to the strain of snail . In a similar vein , we also have no data on if , or how , sod1 genotype affects resistance to other pathogens . A field population of snails interacts with many pathogens in addition to S . mansoni , and there could be fitness tradeoffs associated with other pathogens that render the use of sod1 for vector manipulation ineffective in some environments . In summary , we have here shown that , in a laboratory setting , there was no obvious cost to having the most protective allele at sod1 , and perhaps a slight advantage . The generality of this result will need to be verified in other environments , and for other components of fitness . We also demonstrated an effect of genetic background on the association between sod1 genotype and resistance , a result that points to strong epistatic interactions with other loci in the genome . Clearly sod1 is not the only locus in the genome that influences resistance . So perhaps vector manipulation will require changes at several interacting loci to insure success . Further work of this sort on sod1 and other resistance-associated loci will be essential for evaluating the prospects for vector manipulation as a way to control transmission of S . mansoni . | Driving resistance genes into vector populations remains a promising but underused method for reducing transmission of vector-borne diseases . Understanding the genetic mechanisms governing resistance and how resistance is maintained in vector populations is essential for the development of resistant vectors as a means of eradicating vector-borne diseases . We investigated the utility of one gene ( cytosolic copper-zinc superoxide dismutase - sod1 ) for driving resistance associated alleles into populations of the snail Biomphalaria glabrata , a vector of the trematode parasite of humans , Schistosoma mansoni . Under controlled laboratory conditions we found no evidence for costs of resistance associated with carrying the most resistant allele at sod1 ( in terms of growth , fecundity , or mortality ) . However , we did find a strong effect of genetic background on how strongly sod1 genotype influences resistance . Thus , epistatic interactions with other loci may be as important a consideration as costs of resistance in the use of sod1 for vector manipulation in the field . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"genetics",
"of",
"disease",
"trait",
"locus",
"heredity",
"genetics",
"biology",
"evolutionary",
"biology",
"evolutionary",
"selection",
"genetic",
"determinism",
"evolutionary",
"processes",
"genetics",
"and",
"genomics"
] | 2012 | Effects of Cu/Zn Superoxide Dismutase (sod1) Genotype and Genetic Background on Growth, Reproduction and Defense in Biomphalaria glabrata |
Model organisms have played an important role in the elucidation of multiple genes and cellular processes that regulate aging . In this study we utilized the budding yeast , Saccharomyces cerevisiae , in a large-scale screen for genes that function in the regulation of chronological lifespan , which is defined by the number of days that non-dividing cells remain viable . A pooled collection of viable haploid gene deletion mutants , each tagged with unique identifying DNA “bar-code” sequences was chronologically aged in liquid culture . Viable mutants in the aging population were selected at several time points and then detected using a microarray DNA hybridization technique that quantifies abundance of the barcode tags . Multiple short- and long-lived mutants were identified using this approach . Among the confirmed short-lived mutants were those defective for autophagy , indicating a key requirement for the recycling of cellular organelles in longevity . Defects in autophagy also prevented lifespan extension induced by limitation of amino acids in the growth media . Among the confirmed long-lived mutants were those defective in the highly conserved de novo purine biosynthesis pathway ( the ADE genes ) , which ultimately produces IMP and AMP . Blocking this pathway extended lifespan to the same degree as calorie ( glucose ) restriction . A recently discovered cell-extrinsic mechanism of chronological aging involving acetic acid secretion and toxicity was suppressed in a long-lived ade4Δ mutant and exacerbated by a short-lived atg16Δ autophagy mutant . The identification of multiple novel effectors of yeast chronological lifespan will greatly aid in the elucidation of mechanisms that cells and organisms utilize in slowing down the aging process .
Model eukaryotic organisms such as Drosophila and C . elegans have played important roles in the identification of genes and the molecular characterization of cellular and biochemical pathways that affect the aging process [1] . For example , large-scale systematic RNAi knockdown screens for lifespan extension with C . elegans have implicated multiple genes that regulate metabolism , signal transduction , protein turnover , and gene expression [2] , [3] . The budding yeast , Saccharomyces cerevisiae , has also been particularly useful , especially in characterizing the NAD+-dependent protein deacetylase , Sir2 , as a replicative lifespan ( RLS ) factor [4] . RLS is defined by the number of mitotic cell divisions that a mother cell undergoes prior to senescencing [5] . Yeast lifespan can also be measured chronologically , where the time that non-dividing cells remain viable is monitored [6] . This chronological lifespan ( CLS ) is typically measured in cells that have entered stationary phase ( G0 ) . Both types of yeast aging share multiple effectors of lifespan related to nutrient signaling . Deletion of SCH9 extends both RLS and CLS [6] , [7] . Sch9 is related to the serine/threonine kinase ( Akt ) , that in higher eukaryotes functions in insulin-like growth factor ( IGF ) signaling pathways that have been linked to lifespan regulation [6] . Mutations in the Target of Rapamycin ( TOR ) signaling pathway also extend both types of lifespan in yeast [8]–[10] , as well as in C . elegans [11] . The overlap between CLS and RLS extends to the effects of calorie restriction ( CR ) , a dietary regimen shown to extend the mean and maximum lifespan of rodents [12] . In the yeast system , CR consists of reducing the glucose concentration in the growth medium from the non-restricted ( NR ) level of 2% ( w/v ) to the CR level of 0 . 5% or lower [13] , [14] . CR extends both RLS and CLS [13]–[16] , consistent with the general theme of conserved nutrient signaling pathways playing major roles in longevity . CR , sch9Δ , and tor1Δ conditions all cause a shift in glucose metabolism from fermentation toward respiration in both lifespan systems [10] , [16] , [17] , revealing a strong link with mitochondrial function . Despite the numerous similarities in nutrient-mediated responses between RLS and CLS , there are also significant differences . One of the most striking is that while SIR2 promotes RLS and is reported to be required for lifespan extension by CR [14] , deletion of SIR2 mildly extends CLS and is not required for CR-mediated lifespan extension in this system [15] , [16] . Instead , Sir2-mediated deacetylation of the gluconeogenesis enzyme Pck1 limits the large extension of CLS caused by extreme CR conditions [18] . Due to its simplicity , CLS has been amenable to genome-wide functional aging screens . A previous screen for long-lived mutants used the yeast knockout ( YKO ) collection of individual diploid deletion mutants to individually test each mutant for CLS while incubating in 96-well plates . Several deletion mutants downstream of the TOR signaling pathway were identified , thus implicating TOR signaling in lifespan control [8] . In our study we have utilized the YKO collection to identify additional genetic factors that influence CLS through a different approach . A microarray-based genetic screen was performed on the collection , with the goal of determining which deletion mutants shorten or extend lifespan under NR or CR growth conditions . We report the identification of several classes of short-lived mutants , including those that affect mitochondrial function and the autophagy pathway . We also identify and characterize long-lived mutants in the highly conserved de novo purine biosynthesis pathway that generates IMP , AMP , and GMP . Deletion of genes in this pathway extended lifespan equally to the effect of CR , and CR did not further extend the lifespan of the mutants , suggesting that there are overlapping mechanisms between these two conditions that promote longevity . We show that the de novo purine biosynthesis mutants alter the surrounding growth medium in a way that extends the lifespan of WT cells , pointing to a cell-extrinsic component of CLS regulation .
We took advantage of the YKO collection of gene deletion mutants [19] , in which each individual gene is replaced by the selection marker ( kanMX4 ) and flanked by specific UPTAG and DNTAG sequences ( Figure 1A ) . Viable mutants from the haploid collection were pooled together and grown in synthetic complete ( SC ) medium that contained either 2% glucose ( non-restricted/NR ) or 0 . 5% glucose ( calorie restricted/CR ) . On days 1 , 9 , 21 , and 33 , aliquots were removed and spread onto YPD plates to recover mutants that remained viable ( Figure 1B ) . The TAG sequences present in the recovered cells were PCR amplified using universal primers labeled with Cy3 for day 1 , or Cy5 for days 9 , 21 , and 33 ( Figure 1B ) . Following microarray co-hybridizations , the relative abundance of each mutant was determined by the ratio of Cy5 signal ( days 9 , 21 , or 33 ) to the Cy3 signal ( day 1 ) . ( see Table S1 for ratios ) . Under- or over-representation of a particular mutant's DNA in the aging population was predicted to be indicative of its CLS relative to the other mutants . As expected , the abundance ratios of the TAG signals for most mutants decreased over time in the NR culture ( Figure 1C ) , indicating that most mutants in the population lost viability ( aged ) . By day 33 , when the WT strain was completely dead ( Figure 1B , spot assay ) , there were a limited number of viable mutants in the population that could potentially be extremely long-lived ( Figure 1C , data shown for the NR population ) . The viability of most mutants at day 33 was greater in the CR growth condition than in the NR condition ( Figure 1D ) , suggesting that most mutants respond to CR by extending their CLS . To conservatively choose a subset of mutants for retesting the predicted short CLS phenotype , we set two separate threshold criteria . First , the abundance ratios at day 9 for both TAGs had to be ranked in the bottom 200 . Second , the abundance ratio at day 21 had to be less than 0 . 3 for both TAGs , which represented the bottom quartile for this time point ( Figure 1C ) . The day 33 abundance ratios were not considered because most mutants were dead by then ( Figure 1C ) . The result was 117 candidate mutants predicted to be short-lived ( Table S2 ) . Out of this list of 117 mutants , we individually retested 16 of them for CLS , and found 13 ( 81 . 3% ) to actually be short-lived ( Table S2 ) . Interestingly , 42 of the 117 candidate genes were related to mitochondrial function in some way ( Table S2 ) , most likely because respiration defects prevent cells from properly transitioning through the diauxic shift , thus reducing stationary phase viability [20] . Another major sub-class from the 117 candidates included 10 of the “ATG” genes involved in autophagy . As shown in Figure 2A , the autophagy mutants that we directly tested generally caused a short CLS in 2% glucose as predicted by the screen . The CLS of these mutants was fully extended by the CR condition ( Figure 2A ) , which was somewhat surprising because earlier work in C . elegans showed that autophagy was required for dietary restriction ( DR ) -mediated extension of lifespan [21] , [22] . All mutants that were tested for various reasons in this study and found to have a short CLS in 2% glucose , including the atg mutants , are listed in Table S3 . We were also interested in identifying mutants whose lifespan was not extended by CR . Such mutants were predicted to have similar abundance ratios in the NR and CR conditions across the time course . Many mutants initially appeared to fit this category , which required them to have average NR and CR log rations within 10% of each other ( see Materials and Methods ) . However , only 2 of 41 mutants retested ( 4 . 9% ) were actually confirmed as being CR-unresponsive . These two affected genes were NFU1 and FET3 , both of which encode proteins involved in iron homeostasis . The CLSs of these two mutants were slightly shorter than WT when grown under NR conditions , and , as predicted from the screen , were not extended by CR ( Figure 2B and 2C , and data not shown ) . NFU1 encodes a mitochondrial matrix protein thought to be involved in iron-sulfur complex biogenesis [23] , an important part of the electron transport cascade within the mitochondrial membrane . Its close link with respiration could explain why the nfu1Δ mutant had a shorter lifespan in the CR condition than in the NR condition ( data not shown ) . FET3 encodes a multicopper oxidase , that along with the iron permease ( Ftr1 ) , comprises a high affinity iron uptake system [24] , initially suggesting that high affinity transport of iron is required in CR-induced CLS determination . However , even though an ftr1Δ mutant exhibited a slight shortening of CLS in the NR condition similar to the fet3Δ mutant , CR still induced full CLS extension ( Figure 2D ) . Another protein , Fit3 , is one of three secreted mannoproteins that functions in the retention of siderophore-iron in the cell wall , which can be released and then imported by the Fet3/Ftr1 transport system [25] . Deletion of FIT3 had no affect on CLS , and like the ftr1Δ mutant , its CLS was extended by CR ( Figure 2D ) . Taken together , these results suggest that Fet3 may have a function independent of Ftr1-mediated iron transport at the plasma membrane that is important for CLS during CR growth conditions . To identify long-lived mutants , we again defined conservative thresholds in which the day 33/day1 signal ratio had to be in the top 500 for both the UP-and DN-tags . The day 21/day1 ratio also had to be greater than 1 . 0 for both TAGs , resulting in a list of 40 mutants ( Table 1 ) . Twelve out of the 39 mutants retested ( 30 . 7% ) had a long CLS ( several shown in Figure 3A ) . Isolation of the de novo NAD+ biosynthesis gene , BNA2 , was consistent with the long CLS of a strain lacking BNA1 [16] . YPL056C , YLR104W , and YGL085C , were previously uncharacterized and have now been named based on their Long Chronological Lifespan phenotype as LCL1 , LCL2 , and LCL3 , respectively . The lcl1Δ mutant was previously shown to be resistant to the antifungal drug fluconazole [26] , and the lcl2Δ mutant has deficient levels of mannosylphosphate in the cell wall [27] , suggesting that both of these genes may function in cell wall integrity . DCW1 encodes a putative mannosidase involved in cell well biosynthesis [28] , again pointing to the importance of cell wall structure and function in longevity . LCL3 encodes a protein with homology to Staphylococcus aureus nuclease [29] . Three of the long-lived mutants were involved in either de novo purine biosynthesis ( ADE3 and ADE4 ) or purine import ( FCY2 ) [30]–[32] . Ade4 catalyzes the first step of the pathway , while Ade3 functions in one-carbon metabolism , which donates tetrahydrofolate-linked carbon units for synthesis of the purine ring ( see Figure 3B ) . Fcy2 is a purine/cytosine permease that mediates transport of purine bases ( adenine , guanine , hypoxanthine ) , and a specific pyrimidine base ( cytosine ) across the plasma membrane into the cell . Additional mutants were analyzed for CLS outside of the selection criteria . Those mutants that exhibited an extended lifespan under NR conditions are listed in Table S4 , while those with a normal lifespan under NR conditions are listed in Table S5 . The effects of the de novo purine biosynthesis pathway on aging have not been well studied . In Drosophila melanogaster , mutations in the pathway cause pleiotropic effects due to general purine deficiency , one of them being a short lifespan [33] . In yeast , the pathway was not previously implicated in lifespan regulation . The de novo purine biosynthesis pathway is highly conserved and consists of ten consecutive reactions catalyzed by the ADE gene products that convert 5-phosphoribosyl 1-pyrophosphate ( PRPP ) to inosine monophosphate ( IMP ) , which is then used for AMP and GMP synthesis ( Figure 3B ) . There are also purine salvage pathways that either import extracellular purines via Fcy2 or utilize endogenous purines to synthesize IMP , GMP or AMP through only a few enzymatic steps ( Figure 3C; for review see [34] ) . Deleting other genes in the de novo synthesis pathway such as ADE1 , ADE2 , ADE5 , 7 , ADE6 , or ADE12 significantly extended CLS ( Figure 3C and data not shown ) . ADE13 is essential and ade8Δ was not available in our KO collection , so they were not tested . The lone exception encountered was an ade17Δ mutant , which had a lifespan modestly , but reproducibly , shorter than WT ( Figure 3C ) . Ade17 , as well as Ade16 , catalyzes the conversion of 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside ( AICAR ) into 5′-phosphoribosyl-5-formaminoimidazole-4-carboxamide ( FAICAR ) . The major enzyme in this step is Ade17 , being responsible for ∼90% of AICAR transformylase activity [35] . Mutants in the purine salvage pathways ( AAH1 , APT1 , or HPT1 ) or the one-carbon metabolism pathway ( MTD1 , SHM2 , or SHM1 ) also extended CLS , but to a lesser extent than mutants in the de novo pathway ( Figure 3C ) . The effects of these two pathways on CLS may , therefore , be mediated by a secondary effect on regulation of the de novo pathway . The de novo synthesis of purine nucleotides is regulated at the genetic and enzymatic levels . Enzymatically , the first step of the pathway catalyzed by Ade4 is feedback-inhibited by the end products ADP and ATP [36] . Genetically , excess adenine has a repressing effect on ADE regulon genes , while depletion of adenine results in transcriptional up-regulation due to the activity of transcription factors Bas1 and Bas2/Pho2 [37] , [38] . Regulation of all the de novo pathway genes , with the exception of ADE16 , is achieved via the Bas1/Pho2 complex [39] . It is proposed that the AICAR or SAICAR intermediates promote Bas1-Pho2 dimerization , resulting in the up-regulation of ADE-gene transcription [36] , [38] , [40] . Since we observed CLS extension in ade mutants lacking an enzyme upstream of the AICAR intermediate and CLS shortening for the ade17Δ mutant that likely accumulates AICAR [40] , we generated an ade4Δ ade17Δ double mutant and tested CLS . As shown in Figure 4A , the ade4Δ mutation was epistatic to the ade17Δ mutation for lifespan in the double mutant , initially consistent with a hypothesis that accumulation of AICAR shortens CLS of the ade17Δ mutant . However , completely blocking the AICAR to FAICAR step of the de novo pathway with an ade16Δ ade17Δ double mutant , surprisingly resulted in CLS extension ( Figure 4A ) . Since excess adenine represses the de novo purine synthesis pathway , we next tested whether excess adenine would extend CLS . The SC medium contained either our standard limiting concentration of adenine ( 30 mg/L ) or a 4-fold excess ( 120 mg/L ) , which represses the de novo pathway . Surprisingly , excess adenine did not extend the CLS of a WT strain , but instead suppressed the long CLS phenotype of ade2Δ , ade3Δ , or ade4Δ mutants ( Figure 4B and data not shown ) . This effect was specific to the long-lived ade mutants , because excess adenine did not shorten the CLS of two long-lived mutants with inhibited TOR signaling , tor1Δ and gln3Δ ( Figure 4B ) . The fcy2Δ mutation blocks adenine transport , so the addition of excess adenine did not affect CLS . The long CLS of the ade mutants was reminiscent of the CR effect , suggesting there could be some degree of overlap between the two . To test this idea , CLS of the WT and ade4Δ mutant was measured using the semi-quantitative spot growth assay ( Figure 5A ) , and a quantitative colony forming unit assay that can detect more subtle changes in CLS ( Figure 5B ) . Both assays showed there was no additive effect on CLS when combining the genetic factor ( ade4Δ ) and the environmental factor ( CR ) , at least for the duration of the experiment ( 36 days ) . This was consistent with some overlap in function or involved pathways . To further test this possibility , we also examined the effect of deleting ADE4 on the CLS of an autophagy mutant ( atg16Δ ) . While the CR growth condition fully extended CLS of the atg16Δ mutant ( Figure 2A ) , deleting ADE4 from the atg16Δ mutant only resulted in a partial extension of CLS ( Figure 4A ) . Therefore , one of the differences between CR and the ade4Δ mutant in CLS extension is a differential requirement for autophagy . An earlier large-scale screen for long-lived yeast mutants did not uncover the de novo purine biosynthesis pathway genes [8] . We noticed that one of the differences between our study and the earlier study was the media composition used for the CLS assays . In general , the SC medium used in our study ( Hopkins mix ) is relatively rich in most amino acids compared to the SC medium used in the earlier study , which is described in Current Protocols in Molecular Biology [41] , and abbreviated here as “CPMB” mix ( Table S6 ) . We compared the effects of each SC mix on the CLS of WT , ade4Δ , atg16Δ , and fet3Δ strains . As shown in Figure 5C , CLS of the WT strain was significantly longer in the CPMB media than in the Hopkins medium , even though glucose was 2% in both . As a result , the WT and ade4Δ lifespans were indistinguishable in the CPMB medium . Interestingly , the CPMB media did not extend the short CLS of the atg16Δ mutant ( Figure 5C ) , even though reducing the glucose concentration in Hopkins medium fully extended its CLS ( Figure 2A ) . Similar results were observed with several other autophagy mutants ( data not shown ) , consistent with autophagy being required for mediating the effects of amino acid restriction on CLS . In contrast , the short CLS of the fet3Δ mutant , which was not extended by glucose CR ( Figure 2B ) , was also not extended by the CPMB media ( Figure 5C ) , making Fet3 important for mediating the effects of both glucose restriction and amino acid restriction on CLS . Considering the large effects of media composition on CLS , we next investigated whether any of the mutants isolated from the screen could influence longevity via cell-extrinsic factors that are secreted or released into the growth media . For example , secreted purine compounds such as adenine and hypoxanthine have previously been implicated in the regulation of meiosis within a sporulating yeast culture [42] . Additionally , we noticed during this study that expired medium from NR cultures would reverse the long CLS of CR-grown cells , and expired medium from CR cultures would extend CLS of NR-grown cells ( D . L . Smith Jr . , unpublished data ) . A similar finding was recently published by the Kaeberlein lab , who reported that acetic acid secreted into the medium during NR growth conditions correlated with the short lifespan , and that CR conditions prevented acetic acid secretion [43] . Reduced exposure to acetic acid in the CR cultures was specifically shown to extend CLS , therefore providing a possible mechanism for how CR extends CLS . Interestingly , other conditions that extend CLS such as high media osmolarity or deletion of SCH9 have been proposed to make the cells more resistant to the acetic acid accumulation , rather than blocking organic acid production and secretion [43] . Taken together , these observations raised the question of whether any mutants isolated from our screen could affect CLS through a similar cell extrinsic mechanism . To test for cell extrinsic effects we grew WT , ade4Δ , and atg16Δ strains in SC 2% glucose ( NR ) medium for 5 days into stationary phase . The cells were then pelleted and the expired medium was filtered and swapped in various combinations ( Figure 6A ) . For example , the WT cells received expired medium from the ade4Δ or atg16Δ cells , and vice versa . The media-swapped cultures were then followed through a standard CLS assay ( Figure 6B ) . Interestingly , the CLS of WT and atg16Δ cells was extended when incubated in expired medium from the long-lived ade4Δ cells . In the reciprocal swap , medium from the WT cells largely suppressed the long CLS of the ade4Δ mutant , but had no effect on the atg16Δ mutant . Expired medium from the short-lived atg16Δ mutant did not shorten CLS of the WT strain , but shortened CLS of the ade4Δ mutant ( Figure 6B ) . The expired atg16Δ medium also tended to induce an adaptive regrowth effect , as shown in Figure 6B for the ade4Δ mutant , where nutrients released by dying cells in the stationary phase culture allow some of the remaining viable cells to regrow and populate the culture [44] . The ade4Δ and atg16Δ mutants therefore do alter the growth media in a way that can impact CLS . The secretion of organic acids ( including acetic acid ) and CO2 into the growth medium during fermentation results in a reduction of pH . The toxicity of acetic acid on yeast cells requires a low pH [43] . Therefore , we next tested whether CLS of these mutants correlated with changes in media pH . WT , ade4Δ , ade17Δ , and atg16Δ strains were grown in SC medium containing 2% glucose ( NR ) or 0 . 5% glucose ( CR ) , and the pH of the media was measured over time . As expected , the pH of NR medium for WT cells decreased from ∼3 . 9 to ∼3 . 15 during the first 24 hr of growth and then leveled off . For WT cells in CR medium , the pH still decreased , but only to ∼3 . 5 by day 5 . Media from the short-lived atg16Δ and ade17Δ mutants had pH profiles across the time course that were similar to the long-lived ade4Δ mutant regardless of the starting glucose concentration , indicating that CLS did not correlate with overall pH of the media . However , the lack of a correlation between pH and CLS did not rule out the possibility that acetic acid could still be involved in the extrinsic CLS regulation , especially since the pH remained relatively low ( <4 . 0 ) in each conditions . Furthermore , an acidic environment is not sufficient to chronologically age yeast cells in the absence of acetic acid [43] . If acetic acid was involved in the extrinsic CLS effects , then raising the medium pH close to neutral should suppress the relatively short CLS of the WT and atg16Δ strains . Indeed , raising the medium pH to 6 . 0 either at the time of inoculation ( D0 ) or after two days of growth ( D2 ) ( Figure 7A ) , resulted in a dramatic extension of CLS for the WT and atg16Δ strains that was at least as strong as the ade4Δ mutant effect or the CR growth condition ( Figure 7B ) . To determine whether the ade4Δ and atg16Δ mutants had any effect on acetic acid accumulation in the growth medium , the acetic acid concentration was measured from log phase , day 2 , or day 5 cultures . As shown in Figure 8A , acetic acid accumulated to ∼3 mM in the WT culture on day 5 . For the atg16Δ mutant , acetic acid accumulated earlier ( day 2 ) and at a higher concentration by day 5 ( ∼11 mM ) , which was consistent with the short CLS of this mutant . In contrast , the long-lived ade4Δ mutant did not accumulate acetic acid at all compared to WT , which was very similar to the effect of CR on blocking acetic acid accumulation ( Figure 8A ) . Therefore , the amount of acetic acid secreted into the medium for these two mutants was inversely correlated with their respective CLSs . Since the short CLS phenotype of the atg16Δ mutant was rescued by raising the pH to 6 . 0 ( Figure 7 ) , we were curious whether the higher pH was accompanied by a decrease in acetic acid concentration . The pH was again adjusted to 6 . 0 at the time of inoculation for WT and atg16Δ strains , and then acetic acid concentration measured at day 2 and day 5 ( Figure 8B ) . Unexpectedly , the acetic acid concentration was elevated in the WT strain and reduced in the atg16Δ strain at both time points when the pH was adjusted to 6 . 0 at the time of inoculation ( Figure 8B ) . Such variations in acetic acid accumulation apparently have no effect on CLS because the pH is too high to support the toxicity . Since a long-lived sch9Δ mutant was previously shown to make yeast cells more resistant to acetic acid [43] , we tested whether the ade4Δ and atg16Δ mutations affected cell survival when cultures grown for 2 or 5 days were challenged with 300 mM acetic acid for 200 minutes ( Figure 8B ) . In the day 2 cultures , the ade4Δ mutant was significantly more resistant to acetic acid than the WT strain , again consistent with the long CLS of this mutant . However , resistance of the atg16Δ mutant was indistinguishable from WT . The CR condition made all three strains highly resistant to the acetic acid treatment . In the day 5 NR cultures ( the time of the media swaps in Figure 6 ) , there were no significant differences in the acetic acid resistance between the three strains , and surprisingly , the CR growth condition no longer made the cells more resistant . Resistance to acetic acid could potentially play a role in CLS extension for the ade4Δ mutant , which would be consistent with its ability to survive in the pooled mutant culture used for the screen , where many mutants would secrete acetic acid . In contrast , the short CLS of the atg16Δ mutant may not be due to acetic acid hypersensitivity . These results suggest that secreted acetic acid can commonly impact CLS through a cell extrinsic mechanism that is dependent on media pH .
Autophagy is a multi-step process in which a portion of the cytoplasm is sequestered into a de novo-formed double membrane vesicle called the autophagosome . These vesicles fuse with a lysosome ( the vacuole in yeast ) and release the inner single-membrane vesicle called the autophagic body . Any sequestered organelle or other cellular matter in the autophagic body is degraded and recycled into amino acids , fatty acids , sugars , etc . [45] . This process is especially important during times of stress when cellular components can become damaged and aggregate , or when nutrients are depleted . Chronological aging of yeast cells is characterized by the ability to survive during extended incubation in starvation phase , making the ability to recycle resources critical . The identification of multiple deletion mutants in the autophagy pathway that shorten CLS therefore makes sense , not only because of the need to regenerate cellular components , but to potentially eliminate damaged proteins that arise as the cells age . Our results are consistent with results in Drosophila where mutation of the ATG7 gene shortens lifespan [46] , and a more recent study in yeast showing that atg1Δ and atg7Δ mutants have a short CLS in synthetic growth medium [47] . The atg7Δ mutant was one of the autophagy mutants also isolated from our screen . Surprisingly , most autophagy gene deletion mutants have a normal RLS in rich YPD medium [48] . Similarly , the atg16Δ mutant had a normal RLS when we tested it in SC medium ( Figure S1 ) , making it a CLS-specific longevity factor . Disruption of autophagy in C . elegans prevents the extension of lifespan caused by a daf-2 mutation or dietary restriction [21] , [22] , [49] . Deleting ATG15 , but not the other autophagy genes , blocks CR-mediated RLS extension in yeast [48] . ATG15 was not isolated from our screen , and hence not tested for CLS , but every other autophagy mutant we tested responded to CR with CLS extension ( Figure 2A ) . Interestingly , we found that deleting ATG16 , ATG2 , or ATG6 ( VPS30 ) prevented CLS extension induced by the CPMB variety of SC media used in the Powers et al . screen , which has a normal 2% glucose level but generally has lower concentrations of amino acids compared to the Hopkins mix ( Figure 5C ) . This result is consistent with the strong stimulation of autophagy triggered by nitrogen limitation or amino acid depletion [50] , [51] . Indeed , maintenance of amino acid homeostasis via the general amino acid control system is important for proper CLS [47] . Furthermore , the long CLS of a tor1Δ mutant requires the autophagy gene ATG16 ( data not shown ) . Similarly , autophagy was recently shown to be required for the extension of CLS induced by low concentrations of rapamycin [52] , an inhibitor of the TOR signaling pathway . Future studies on the links between autophagy , amino acid depletion , and lifespan extension are clearly warranted . Two proteins ( Fet3 and Nfu1 ) involved in iron homeostasis/metabolism were isolated as mutants whose CLS was not extended by the CR growth condition . Iron accumulates to high levels in the vacuole of yeast cells where it can be accessed during times of need , such as low iron growth conditions . Another key time of iron release from the vacuole is during the diauxic shift when the balance of iron is shifted to the mitochondria , where it is needed for mitochondrial biogenesis . The iron is incorporated into iron/sulfur complexes within multiple mitochondrial proteins , including aconitase and components of the electron transport chain . A defect in iron homeostasis could affect mitochondrial processes . One of the phenotypes observed during chronological aging is an accumulation of intracellular iron . Much of this iron is likely tied up in lipofuscin , an insoluble aggregate of proteins and lipid that is high in iron and accumulates in aging cells . Interestingly , CR reduces this accumulation of lipofuscin and iron [53] . The reduction in iron could contribute to the corresponding reduction in reactive oxygen species because a major source of reactive oxygen species is generated via iron through the Fenton reaction . It is not clear why a fet3Δ mutant would block the CR effect , but perhaps the iron oxidase activity of Fet3 has an additional function in iron homeostasis beyond its role in high affinity transport . Interestingly , a recent report showed that FET3 is one of several iron related genes that are up-regulated in response to increasing strength of CR [54] . FET3 was also required for the extension of CLS induced by the low amino acid CPMB medium ( Figure 5C ) , pointing to iron and possibly mitochrondrial function being important for both glucose and amino acid restriction effects on CLS . The de novo purine biosynthesis pathway is familiar to yeast researchers because the AIR intermediate that accumulates in ade2 mutants takes on a red pigmentation when it is oxidized and concentrated in the vacuole of respiring cells . Multiple genetic assays have taken advantage of this visual phenotype [55] , [56] . Limiting the amount of adenine in the growth medium promotes development of the red color by increasing flux through the pathway . The 30 mg/L of adenine in Hopkins mix SC is limiting in this context . Excess adenine suppresses the red color by reducing flux through the pathway , thus reducing AIR formation . Excess ( 4X ) adenine also suppresses the long CLS of the ade2Δ , ade3Δ , and ade4Δ mutants , but had no effect on CLS of the WT strain . One possible mechanism for a block in this pathway to regulate CLS is that reduced production of AMP and/or IMP leads to lifespan extension . Consistent with this idea , deletion of the adenylate kinase 1 gene ADK1 , which leads to a large increase in cellular AMP concentration [57] , also shortens CLS ( Table S3 ) . AMP is an allosteric effector of multiple enzymes in metabolism , including phosphofructokinase ( PFK ) , a key regulatory step in the glycolytic pathway who's activity is enhanced by AMP binding . CR has been shown to reduce PFK activity in mouse liver [58] . Lower AMP levels could mimic CR by reducing glycolytic flux . This model also fits the extended CLS of the fcy2Δ mutant , which would also reduce AMP production by blocking the import of extracellular adenine . The compensatory increase in AMP production by the de novo purine synthesis pathway would partially suppress the effect , resulting in the more modest increase in lifespan for this mutant compared to the ade4Δ mutant . Since the de novo purine biosynthesis pathway and Fcy2-mediated transport of guanine also regulate GMP production ( and subsequently GTP/GDP levels , reduced GMP levels could also contribute to the lifespan extension via effects on the Ras/cAMP/PKA pathway , as inhibition of Ras2 results in extension of CLS [59] . Consistent with this possibility , we have found that deletion of BCY1 , which constitutively activates PKA , shortens CLS ( Table S3 ) . A second possible mechanism for the de novo purine biosynthesis pathway to regulate CLS is through the control of AICAR concentration . Severe accumulation of AICAR induced by ADE4 over-expression in an ade16 ade17 double mutant causes synthetic lethality [40] . The less severe accumulation predicted for an ade17Δ mutant is not lethal , but instead leads to a short CLS ( Figure 3C ) . However , any putative negative effect of AICAR accumulation from a defect in this step of the pathway is overcome , in terms of CLS , by a double deletion of ADE16 and ADE17 . This double mutant behaves like any other deletion mutant in the de novo pathway ( long-lived ) , suggesting that effects on IMP/AMP production or other unknown mechanisms are dominant to the AICAR effect . If AICAR does have a negative effect on CLS , then it is modest and opposite of that observed in higher eukaryotes . In metazoans , AICAR acts as an agonist of AMP-activated protein kinase ( AMPK ) [60] , an enzyme that functions in mediating some aspects of longevity in C . elegans [61] , [62] . The yeast paralog of AMPK , Snf1 , is not activated by AMP or AICAR [63] . Furthermore , the snf1Δ mutant was found to have a short CLS ( Table S2 ) , a phenotype that is likely due to the roles of Snf1 in promoting respiration and autophagy [64] , [65] . Given the complex nature of purine biosynthesis regulation and its links to the regulation of other metabolic pathways , including amino acid biosynthesis , other mechanisms leading to lifespan extension are certainly possible . For example , secreted adenine-related compounds could contribute to the cell-extrinsic effects of the ade mutants on CLS . In fact , the temporal secretion of various purines into the media and their subsequent uptake and utilization is a key signal that synchronizes the sporulation process between cells in a dense culture [42] . Acetic acid accumulates to low millimolar concentrations in stationary phase yeast cultures that are grown in SC medium with 2% glucose ( NR ) . Exposure to this acetic acid , coupled with the acidic environment of the expired medium contributes to chronological aging [43] . CR growth conditions block the acetic acid accumulation , and long-lived mutants such as sch9Δ and ras2Δ tend to be resistant to acetic acid toxicity , suggesting that resistance to acetic acid may be a general property of chronologically long-lived yeast cells [43] . We found that the long-lived ade4Δ mutant blocked acetic acid accumulation in the growth medium as effectively as CR , while the short-lived atg16Δ mutant accumulated significantly higher concentrations of acetic acid than did the WT strain ( Figure 8A ) . In addition to greatly reducing acetic acid levels ( Figure 8A ) , we found that CR makes all three strains more resistant to acetic acid when the exposure occurs after 2 days growth , but is no longer effective with 5-day cultures ( Figure 8B ) . While the ade4Δ mutant was moderately resistant to acetic acid at day 2 when compared to WT , by day 5 there was very little difference in sensitivity between the two mutants and WT . This is an important point , because the expired media swaps between the WT , ade4Δ , and atg16Δ strains were performed with 5-day old cultures . Perhaps chronologically aged yeast cells are simply programmed to be more resistant to acetic acid as a defense against this by-product of fermentation . These were short-term acetic acid exposures ( 200 minutes ) , so it is possible that prolonged exposure of the day 5 cultures , or lack of exposure for the ade4Δ expired media , could still affect CLS . This would also correlate well with the extension of CLS induced by raising the pH to 6 . 0 ( Figure 7B ) , which would neutralize the toxicity of acetic acid . The ade4Δ mutation therefore both suppresses acetic acid accumulation and promotes acetic acid resistance , a phenotypic combination also induced by the CR growth condition . It remains unclear why a defect in autophagy ( atg16Δ ) results in hyper-accumulation of acetic acid , while a block in de novo purine biosynthesis prevents acetic acid accumulation . An important function of autophagy is the turnover of organelles , including mitochondria . In mice deficient for Atg7 , mitochondrial dysfunction has been observed that is accompanied by elevated reactive oxygen species [66] . Perhaps a defect in mitochondrial function would promote fermentation during NR conditions by preventing the yeast cells from fully transitioning from fermentation to respiration at the typical diauxic shift , and thus favoring acetic acid production . This would also account for the large number of mitochondria-related mutants that were isolated from the screen as being short-lived . Given the similarities of the ade4Δ CLS phenotype to CR , it is possible that the ade4Δ mutant could also enhance a shift from fermentation toward respiration , which could reduce acetic acid production . For the various mutants isolated from the screen , it will therefore be interesting to further compare the relative CLS contributions of their actual cellular defects with their acetic acid secretion and toxicity profiles . Specific combinations of intracellular and extracellular effects are likely going to be CLS determinants . The microarray-based genetic screen performed in this study was successful in identifying several novel longevity genes , but its quantitative ability to predict long-lived mutants based on the abundance ratios from the arrays was modest . Similar difficulties were previously observed using the YKO collection in a different type of longevity screen , in which each mutant was individually grown in a 96-well plate , and ability to re-grow was tested over time . In that screen , only 5 of 90 predicted long-lived mutants ( 5 . 6% ) were confirmed when retested [8] , [67] . In our case , 12 of the 39 candidate mutants ( 30 . 8% ) were confirmed as long-lived when retested ( Table 1 ) . Not surprisingly then , only 4 of the 12 confirmed long-lived mutants isolated from our screen ( LCL1 , DCW1 , LCL2 , and MUM2 ) were ranked in the top 1000 long-lived candidates from the earlier Powers et al . CLS screen . These results are likely indicative of inherent variability in large-scale screens for long CLS , as well as subtle differences in the growth conditions . Large-scale screening for short-lived mutants is much more efficient , which is reflected in the fact that 68 of the 117 short-lived candidates from our screen ( 58 . 1% ) are also in the bottom 1000 short-lived candidates from the Powers et al . screen ( Table S2 ) . Having multiple screening approaches is advantageous , as mutants not detected by one method may be detected by another . There are several possible reasons for the variability associated the microarray-based longevity screen , especially for long-lived mutants . One possibility is the adaptive regrowth phenomenon , in which a subpopulation of cells in an aging stationary phase culture adapts to utilize the nutrients released by dead cells to re-grow and populate the culture [44] . If a mutant underwent gasping during aging of the pooled collection , then it would register an artificially high abundance ratio , and fail to be long-lived when individually retested . Another possibility that would be unique to the mixed population approach is the introduction of competition between the strains , where mutants with improved overall fitness could have an advantage that is lost when they are retested individually . In a related scenario , certain mutants in the mixed population are likely highly resistant or overly sensitive to changes in medium composition ( such as acetic accumulation ) that occurred as the cultures were aging . Certain mutants could directly influence the medium composition , thus altering the lifespan of the highly sensitive mutants in the process . A good example is the ade4Δ mutant , whose expired SC medium extended the lifespan of the WT and atg16Δ strains ( Figure 6 ) , possibly through the reduction of acetic acid accumulation ( Figure 8A ) . In applying the microarray/barcode approach to other aging or age-related problems , it is likely that the amount of variability would be more limited with the addition of duplicate or triplicate screens . However , even with the inherent variability , this microarray screen successfully identified several novel longevity regulators that will be the subject of future studies .
Yeast strains used in this study were isogenic to the haploid strain BY4741 ( MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 ) , and were obtained from the yeast gene knockout collection [19] . The ade16Δ ade17Δ mutant strain ( Y1093 ) was kindly provided by Bertrand Daignon-Fornier [68] . Most in vivo assays were performed in synthetic complete ( SC ) medium following the recipe provided in the Cold Spring Harbor Yeast Genetics Course Manual [69] , and sold by QBioGene as “Hopkins mix” . The alternative SC medium is derived from Current Protocols in Molecular Biology [41] , which we refer to as “CPMB” mix . Chemical compositions of the various SC media types are listed in Table S6 . Glucose was added to the SC media to a final concentration of either 0 . 5% ( CR-Calorie Restricted ) or 2% ( NR-Non Restricted ) . Where indicated , the Hopkins mix SC medium was buffered to pH 6 . 0 with a citrate phosphate buffer ( 6 . 42 mM Na2HPO4 and 1 . 79 mM citric acid , final concentration ) , as previously described [43] . For buffering the medium at day 2 , a 10× concentrate of the citrate phosphate buffer was added to SC . For pH measurements of expired media , small aliquots were removed from the cultures and then discarded to prevent contamination of the long-term culture . To begin the screen , 1 ml ( 15 OD600 units ) of the pooled haploid knockout collection was inoculated into 200 ml SC medium containing either 2% glucose ( NR ) or 0 . 5% glucose ( CR ) . The next day ( day 0 ) , aliquots of 100 µl were transferred into 10 ml of fresh SC-NR and SC-CR media , respectively . Twenty such cultures were inoculated for each glucose concentration and allowed to age at 30°C in the roller drum to provide aeration [16] . Starting with day 1 ( D1 ) , 100 µl of each culture was plated onto YPD plates every 3 days to allow viable cells in the population to re-grow . These YPD plates were incubated at 30°C for 2 days and the cell lawns harvested by scraping and pooled together , then washed with ice cold water and stored at −80°C . Once the time course was completed ( day 33 ) , genomic DNA was isolated from the cell pellets [41] . The UP- and DNTAGs were labeled with Cy5 ( day 1 ) or Cy3 ( days 9 , 21 , and 33 ) by PCR amplification of genomic DNA using primer pairs U1/U2 and D1/D2 , respectively , as previously described [70] . The Cy5-labeled UP- and DNTAGs from day 1 were then co-hybridized with the Cy3-labeled UP- and DNTAGSs on custom-designed “Hopkins TAG-arrays” from Agilent Technologies ( AMADID 011443 ) as previously described [70] . Fluorescence signal intensities were measured by scanning the arrays with a Genepix 4000B instrument coupled with GenePix Pro software . The signal intensity ratios were then calculated for days 9 , 21 , and 33 compared to day 1 as the control using Microsoft Excel . The signal ratios for all essential genes on the array were averaged and considered the background . Any non-essential genes with up- or down-tag ratios lower than this background average were eliminated from the analysis , thus ensuring that only genes with signals from both tags were included ( 2715 genes , which included most of those in the DNTAG list in Table S1 ) . Box plots of the ratios in Figure 1 were assembled from the 3478 genes in the UPTAG list ( Table S1 ) using R Software . Mutants with similar average NR and CR log ratios were identified by applying two criteria to their values at every time point: ( 1 ) ratios were within 10% of each other and ( 2 ) the null hypothesis that were the same according to a t-test . In the case of ( 1 ) , we calculated the fractional difference between the average NR and CR log ratios ( i . e . , difference between these values divided by their average ) . The absolute value of the fractional difference was required to be less than 0 . 1 . We then applied a t-test to the NR and CR log ratios and required their p-value to be less than 0 . 05 ( i . e . , their means are not significantly different ) . Quantitative ( colony forming unit ) and semi-quantitative ( 10-fold serial dilution spot-test ) chronological life span ( CLS ) assays were performed as previously described [16] . For the media swap experiments , the 10 ml cultures were grown for 5 days . The cultures were then pelleted in a swinging bucket rotor ( 2500 RPM ) at room temperature in an Eppendorf 5810R tabletop centrifuge . The supernatants were removed and passed through a 0 . 2 micron syringe filter prior to the swap . For the measurement of acetic acid concentration in growth media , cells were grown in the appropriate SC medium ( 10 ml in culture tubes ) to the indicated time points . Log phase cells ( OD600 of 0 . 8 ) and cells grown to day 2 and day 5 were pelleted by centrifugation , and the clarified media was passed through a 0 . 2 micron syringe filter . The filtrate was used for measuring the acetic acid concentration using an Acetic Acid Kit ( R-Biopharm AG , Darmstadt , Germany ) , following the manufacturer's directions . Three biological replicas were assayed for each condition to provide mean millimolar concentrations and standard deviations . To determine sensitivity/resistance of the mutant strains to exogenously added acetic acid , cultures were challenged for 200 minutes with 300 mM acetic acid either at day 2 or day 5 of the CLS assay . Cells were diluted in water and then spread onto YPD plates to allow viable cells to grow into colonies , which were then counted . The percent survival was calculated by dividing the colony forming units ( CFU ) of the treated samples by the untreated samples . Three biological replicates were tested for each condition . | The aging process is associated with the onset of several age-associated diseases including diabetes and cancer . In rodent model systems , the dietary regimen known as caloric restriction ( CR ) is known to delay or prevent these diseases and to extend lifespan . As a result , there is a great deal of interest in understanding the mechanisms by which CR functions . The budding yeast , Saccharomyces cerevisiae , has proven to be an effective model for the analysis of genes and cellular pathways that contribute to the regulation of aging . In this study we have performed a microarray-based genetic screen in yeast that identified short- and long-lived mutants from a population that contained each of the viable haploid gene deletion mutants from the yeast gene knockout collection that were pooled together . Using such an approach , we were able to identify genes from several pathways that had not been previously implicated in aging , including some that appear to contribute to the CR effect induced by restriction of either amino acids or sugar . These results are expected to provide new groundwork for future mechanistic aging studies in more complex organisms . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/gene",
"discovery",
"genetics",
"and",
"genomics",
"molecular",
"biology"
] | 2010 | A Microarray-Based Genetic Screen for Yeast Chronological Aging Factors |
Soil may serve as an environmental reservoir for prion infectivity and contribute to the horizontal transmission of prion diseases ( transmissible spongiform encephalopathies [TSEs] ) of sheep , deer , and elk . TSE infectivity can persist in soil for years , and we previously demonstrated that the disease-associated form of the prion protein binds to soil particles and prions adsorbed to the common soil mineral montmorillonite ( Mte ) retain infectivity following intracerebral inoculation . Here , we assess the oral infectivity of Mte- and soil-bound prions . We establish that prions bound to Mte are orally bioavailable , and that , unexpectedly , binding to Mte significantly enhances disease penetrance and reduces the incubation period relative to unbound agent . Cox proportional hazards modeling revealed that across the doses of TSE agent tested , Mte increased the effective infectious titer by a factor of 680 relative to unbound agent . Oral exposure to Mte-associated prions led to TSE development in experimental animals even at doses too low to produce clinical symptoms in the absence of the mineral . We tested the oral infectivity of prions bound to three whole soils differing in texture , mineralogy , and organic carbon content and found soil-bound prions to be orally infectious . Two of the three soils increased oral transmission of disease , and the infectivity of agent bound to the third organic carbon-rich soil was equivalent to that of unbound agent . Enhanced transmissibility of soil-bound prions may explain the environmental spread of some TSEs despite the presumably low levels shed into the environment . Association of prions with inorganic microparticles represents a novel means by which their oral transmission is enhanced relative to unbound agent .
Bovine spongiform encephalopathy , human Creutzfeldt-Jakob disease and kuru , sheep scrapie , and chronic wasting disease of deer , elk , and moose belong to the class of fatal , infectious neurodegenerative diseases known as transmissible spongiform encephalopathies ( TSEs ) or prion diseases [1 , 2] . The precise nature of the etiological agent of these diseases remains controversial , but most evidence points to a misfolded isoform of the prion protein ( PrPTSE ) as the major , if not sole , component of the pathogen [3] . Sheep scrapie and cervid ( deer , elk , and moose ) chronic wasting disease are distinct among TSEs because epizootics can be maintained by horizontal transmission from infected to naïve animals [4–6] , and transmission is mediated , at least in part , by an environmental reservoir of infectivity [7–10] . The presence of an environmental TSE reservoir impacts several epidemiological factors including contact rate ( the frequency animals come in contact with the disease agent ) , duration of exposure ( time period over which animals come in contact with the pathogen ) , and the efficiency of transmission ( the probability that an exposed individual contracts the disease ) . The oral route of exposure appears responsible for environmental transmission of chronic wasting disease and scrapie [6 , 11]; the propagation of bovine spongiform encephalopathy epizootics ( feeding TSE-infected meat and bonemeal to cattle ) ; the appearance of variant Creutzfeldt-Jacob disease in humans and feline spongiform encephalopathy in cats ( presumably by consumption of bovine spongiform encephalopathy–infected beef ) [12 , 13]; the spread of kuru among the Fore of Papua New Guinea ( ritualistic endocannibalism [14–16] ) ; and outbreaks of transmissible mink encephalopathy ( TME ) in farm-reared mink [17] . Following consumption , TSE agent is sampled by gut-associated lymphoid tissue , such as Peyer's patches or isolated lymphoid follicles , and accumulates in lymphatic tissues before entering the central nervous system via the enteric nervous system [18–20] . While ingestion is a biologically relevant TSE exposure route , oral dosing is a factor of ~105 less efficient than intracerebral inoculation in inducing disease in rodent models [21] . The amounts of TSE agent shed into the environment are presumably small . The assumed low levels of TSE agent in the environment and the inefficiency of oral transmission have led to uncertainty about the contribution of environmental reservoirs of infectivity to prion disease transmission . We and others have hypothesized that soil may serve as a reservoir of TSE infectivity [8 , 9 , 22 , 23] . Deliberate and incidental ingestion of soil by ruminants can amount to hundreds of grams daily [24 , 25] . Prions enter soil environments via decomposition of infected carcasses [8 , 26] , alimentary shedding [11 , 27 , 28] , deliberate burial of diseased carcasses/material [29] , and possibly , urinary excretion [30] . TSE agent persists for years when buried in soil [26] . The disease-associated prion protein sorbs to soil particles [22 , 31 , 32] , and the interaction of PrPTSE with the common aluminosilicate clay mineral montmorillonite ( Mte ) is remarkably avid [22] . Despite this strong binding , PrPTSE–Mte complexes are infectious when inoculated into brains of recipient animals [22] . For TSEs to be transmitted via ingestion of prion-contaminated soil , prions bound to soil components must remain infectious by the oral route of exposure . We therefore investigated the oral infectivity of Mte- and soil-bound prions . We examined the effects of prion source ( viz . infected brain homogenate [BH] and purified PrPTSE ) and dose on disease penetrance ( proportion of animals eventually exhibiting clinical TSE symptoms ) and incubation period ( time to onset of clinical symptoms ) in experiments with Mte . We investigated the oral infectivity of soil particle–bound prions to Syrian hamsters using four dosing regimes: ( 1 ) infected BH mixed with Mte ( BH–Mte mixtures ) , ( 2 ) isolated complexes of purified PrPTSE bound to Mte ( PrPTSE–Mte complexes ) , ( 3 ) purified PrPTSE mixed with Mte ( PrPTSE–Mte mixtures ) , and ( 4 ) PrPTSE mixed with each of three whole soils ( PrPTSE–soil mixtures ) . The rationale for each dosing regime is described below . Survival analysis was used to assess risk of clinical disease manifestation and quantify differences in effective titer . Application of survival analysis to oral bioassays of TSE transmissibility is discussed in Figure S1 and Text S1 .
To examine the effect of Mte on the oral transmissibility of prions in BH , we incubated infected BH with clay particles for 2 h to allow sorption of the agent; controls lacking Mte were treated identically [22] . Three doses of 10% BH ( 30 , 3 , and 0 . 3 μL ) were assayed . Diminished gastrointestinal bioavailability was expected to be evidenced by significant lengthening of incubation period , reduced disease penetrance , or both . Binding of either 30 or 3 μL of brain material to Mte yielded disease penetrance and incubation periods similar to BH alone ( Figure 1A and 1B ) , a finding consistent with our previous report that a substantial fraction of PrPTSE in clarified BH binds to Mte and that Mte-bound prions remain infectious [22] . Surprisingly , at the lowest BH dose ( 0 . 3 μL , Figure 2 ) , sorption of TSE agent to Mte enhanced transmission , increasing disease penetrance and shortening incubation period . Adjusted for the amount of BH administered and combined across doses , Mte significantly enhanced oral transmissibility ( p < 0 . 0001 ) . Survival analysis indicated the risk of clinical disease manifestation relative to Mte-free controls was 3 . 03 ( 95% confidence interval [CI]: 1 . 68 , 5 . 45 ) , signifying an increase in the effective titer of TSE agent . While the influence of Mte was significant when tested across all BH doses , the effect was most readily observed at 0 . 3 μL . The dose-dependent difference in the influence of Mte on transmissibility may be attributable to competition between macromolecules in BH ( e . g . , lipids , other proteins , nucleic acids ) with PrPTSE for sorption sites on the clay surface . Such competition was evidenced by detection of unbound PrPTSE and other proteins in incubations of Mte with 30 and 3 μL BH ( unpublished data ) . To examine the influence of Mte on oral transmissibility without the interference of other macromolecules from brain homogenate , we purified PrPTSE and inoculated hamsters using two different dosing regimes . The first dosing regime ( PrPTSE–Mte complexes ) was designed to directly assay the infectivity of PrPTSE sorbed to Mte surfaces ( i . e . , the amount of unbound PrPTSE was minimized in treatments containing Mte ) . Purified PrPTSE was clarified to remove large aggregates , and after 2-h incubation with Mte , PrPTSE–Mte complexes were separated from unbound protein by centrifugation through a sucrose cushion [22] . Hamsters were orally challenged with the isolated PrPTSE–Mte complexes [22] or an amount of unbound clarified PrPTSE ( 200 or 20 ng ) equivalent to that introduced into the clay suspension ( Table 1 ) . Immunoblot analysis of the inocula ( Figure S2A ) demonstrated that the amount of PrP in the unbound samples was not less than that in PrPTSE–Mte complexes . Sorption of PrPTSE to Mte dramatically enhanced prion disease transmission ( Table 1 ) . Approximately 38% of animals receiving 200 ng of unbound clarified PrPTSE exhibited clinical symptoms with an incubation period for infected animals of 203 ± 33 ( mean ± standard deviation ) days post inoculation ( dpi ) . In contrast , all animals orally dosed with an equivalent amount of Mte-bound PrPTSE manifested disease symptoms ( incubation period = 195 ± 37 dpi ) , an enhancement of transmission comparable to that observed for the lowest BH dose ( Figure 2 ) . Animals inoculated with Mte alone or 10-fold less unbound clarified PrPTSE ( 20 ng ) remained asymptomatic throughout the course of the experiment ( >365 dpi ) , whereas 20 ng of clarified PrPTSE adsorbed to Mte produced TSE infection in 17% of animals . These data establish not only that the Mte-bound prions remain infectious via the oral route of exposure , but that agent binding to Mte increases disease penetrance , enhancing the efficiency of oral transmission . The second oral dosing regime using purified PrPTSE ( PrPTSE–Mte mixtures ) was designed to ensure that treatments with and without Mte contained equivalent PrPTSE doses . These experiments differed from those above in two important aspects . First , PrPTSE–Mte complexes were not separated from suspension prior to inoculation so that comparable amounts of infectious agent were administered to both treatment groups . In the first dosing regime , some PrPTSE may have been lost during sedimentation of PrPTSE–Mte complexes ( Figure S2A ) . Second , the purified prion preparation was not clarified and therefore contained a range of PrPTSE aggregate sizes . The sizes of PrPTSE aggregates attached to Mte particles were expected to be more heterogeneous than those in the first dosing regime . Compared to Mte-free controls , administration of purified PrPTSE mixed with Mte increased disease penetrance at all doses and shorted incubation times in the 1-μg PrPTSE treatment ( Figure 3A ) . At the two lower doses ( 0 . 1 and 0 . 01 μg PrPTSE ) , binding of the agent to Mte dramatically increased disease penetrance ( 31% ) at PrPTSE doses failing to yield clinical infection in 31 of 32 animals in the absence of the clay mineral ( Figure 3B and 3C ) . Comparison of the survival curves in Figure 3A and 3C indicates that the 0 . 01-μg PrPTSE–Mte mixture was at least as infectious as 1-μg PrPTSE Mte-free samples , suggesting that sorption of purified PrPTSE to Mte enhanced transmission by a factor of ≥100 . To quantify the contributions to changes in relative risk of prion dose and agent sorption to Mte , we constructed a multivariate Cox proportional hazards model with two covariates: log10 PrPTSE dose and Mte presence ( Table 2 ) . Each log10 increase in PrPTSE dose multiplies the relative risk by a factor of ~2 ( i . e . , a 10-fold increase in dose approximately doubles the risk of infection ) . Notably , sorption of purified PrPTSE to Mte multiplies the relative risk by a factor of ~8 . These values allowed computation of a multiplicative equivalence factor between PrPTSE dose and Mte presence in the inoculum . Expressed in terms of PrPTSE dose , addition of Mte to the inoculum is equivalent to multiplying the PrPTSE dose by a factor of 680 ( 95% CI 16 , ∞ ) ; that is , inclusion of Mte increases the effective titer of a given PrPTSE dose by 680-fold . Estimates of effective titer span a wide range ( 95% CI 16 , ∞ ) , and the present data do not allow us to place an upper bound on the increased risk associated with the presence of Mte in a sample . At a minimum , effective titer increased by 1 . 2 orders of magnitude , but the effect could be substantially larger . The best estimate of the Cox analysis represents a 2 . 8 order-of-magnitude increase in effective titer . Oral administration of Mte-bound PrPTSE did not appear to alter strain properties . Following limited proteinase K ( PK ) digestion , many PrPTSE strains can be discriminated by the size and glycoform pattern of PK-resistant core of PrPTSE ( PrP-res ) [33–36] . Strain differences are also manifested in specific clinical symptoms . At the conclusion of the oral transmission experiments described above , the brains of clinically infected animals were assayed for PrP-res by immunoblotting ( Figure S3 ) . Differences in the molecular mass and glycoform distribution of PrP-res were not apparent between the treatment groups . Furthermore , clinical presentation of disease ( symptoms or length of clinically positive period ) did not differ between treatments . The experiments described above were conducted using the Hyper ( HY ) strain of hamster-adapted TME agent ( PrPHY ) . To further examine the strain stability of Mte-bound PrPTSE , we employed the Drowsy ( DY ) strain of hamster-passaged TME agent ( PrPDY ) to investigate the molecular mass of PrP desorbed from Mte and the effect of this clay mineral on oral transmissibility [35 , 36] . We previously reported the N-terminal cleavage of PrPHY extracted from Mte yielding a product similar in size to PK-digested PrPHY [22] . PK digestion of PrPHY and PrPDY results in products of characteristically different molecular masses [35 , 36]: the length of the PrPHY digestion product exceeds that of PrPDY by at least ten amino acids [35 , 36] . We found that extraction of bound PrPDY from Mte resulted in a product similar in molecular mass to PrPDY cleaved by PK ( Figure 4 ) . These data are consistent with the idea that strain properties are preserved when PrPTSE binds to Mte . DY agent is not orally transmissible [37] , and we find that sorption of DY to Mte does not facilitate oral transmission ( Text S1 ) . Natural soils are composed of a complex mixture of inorganic and organic components of various particle sizes . Smectitic clays such as Mte are important constituents of many natural soils and contribute significantly to their surface reactivity [38] . In natural soils , metal oxide and organic matter often coat smectite surfaces and may alter their propensity to bind PrPTSE . Furthermore , additional sorbent phases may be important in the binding of TSE agents to whole soils . We previously demonstrated that PrPTSE binds to whole soils of varying texture , mineralogy , and organic carbon content [22] . To examine the impact of agent binding to whole soil on oral TSE transmission , we incubated 1 μg of purified PrPTSE with each of three whole soil samples ( Elliot , Dodge , and Bluestem soils ) to allow sorption , and then orally dosed hamsters with the PrPTSE–soil mixtures . Soil-bound TSE agent remained infectious perorally , and two of the soils significantly enhanced oral disease transmission ( Figure 5 ) . Hazard ratios between Elliot ( 4 . 76 [95% CI: 1 . 38–16 . 4] , p = 0 . 019 ) and Bluestem ( 6 . 04 [95% CI: 1 . 59–22 . 9] , p = 0 . 013 ) soils and unbound PrPTSE indicate a significant increase in transmissibility , but no difference for the Dodge soil ( 1 . 66 [95% CI: 0 . 52–1 . 66] , p = 0 . 578 ) . The hazard ratios for the Elliot and Bluestem soils did not differ from one another ( 0 . 79 [95% CI: 0 . 19–3 . 25] , p = 0 . 543 ) indicating statistical equivalence in transmissibility . The limited numbers of animals in the treatment groups precluded derivation of a multiplicative equivalence factor to equate the presence of Elliot or Bluestem soil with dose of infectious agent; however , substantially more animals in the Elliot and Bluestem treatment groups ( 14 of 16 animals , 87 . 5% penetrance ) displayed clinical symptoms compared to the unbound PrPTSE treatment group ( two of eight animals , 25% penetrance ) .
These experiments address the critical question of whether soil particle–bound prions are infectious by an environmentally relevant exposure route , namely , oral ingestion . Oral infectivity of soil particle–bound prions is a conditio sine qua non for soil to serve as an environmental reservoir for TSE agent . The maintenance of infectivity and enhanced transmissibility when TSE agent is bound to the common soil mineral Mte is remarkable given the avidity of the PrPTSE–Mte interaction [22] . One might expect the avid interaction of PrPTSE with Mte to result in the mineral serving as a sink , rather than a reservoir , for TSE infectivity . Our results demonstrate this may not be the case . Furthermore , sorption of prions to complex whole soils did not diminish bioavailability , and in two of three cases promoted disease transmission by the oral route of exposure . While extrapolation of these results to environmental conditions must be made with care , prion sorption to soil particles clearly has the potential to increase disease transmission via the oral route and contribute to the maintenance of TSE epizootics . Two of three tested soils potentiated oral prion disease transmission . The reason for increased oral transmissibility associated with some , but not all , of the soils remains to be elucidated . One possibility is that components responsible for enhancing oral transmissibility were present at higher levels in the Elliot and Bluestem soils than in the Dodge soil . The major difference between the Dodge soil and the other two soils was the extremely high natural organic matter content of the former ( 34% , [22] ) . The Dodge and Elliot soils contained similar levels of mixed-layer illite/smectite , although the contribution of smectite layers was higher in the Dodge soil ( 14%–16% , [22] ) . The organic matter present in the Dodge soil may have obstructed access of PrPTSE to sorption sites on smectite ( or other mineral ) surfaces . The mechanism by which Mte or other soil components enhances the oral transmissibility of particle-bound prions remains to be clarified . Aluminosilicate minerals such as Mte do not provoke inflammation of the intestinal lining [39] . Although such an effect is conceivable for whole soils , soil ingestion is common in ruminants and other mammals [25] . Prion binding to Mte or other soil components may partially protect PrPTSE from denaturation or proteolysis in the digestive tract [22 , 40] allowing more disease agent to be taken up from the gut than would otherwise be the case . Adsorption of PrPTSE to soil or soil minerals may alter the aggregation state of the protein , shifting the size distribution toward more infectious prion protein particles , thereby increasing the specific titer ( i . e . , infectious units per mass of protein ) [41] . In the intestine , PrPTSE complexed with soil particles may be more readily sampled , endocytosed ( e . g . , at Peyer's patches ) , or persorbed than unbound prions . Aluminosilicate ( as well as titanium dioxide , starch , and silica ) microparticles , similar in size to the Mte used in our experiments , readily undergo endocytotic and persorptive uptake in the small intestine [42–44] . Enhanced translocation of the infectious agent from the gut lumen into the body may be responsible for the observed increase in transmission efficiency . Survival analysis indicated that when bound to Mte , prions from both BH and purified PrPTSE preparations were more orally infectious than unbound agent . Mte addition influenced the effective titer of infected BH to a lesser extent than purified PrPTSE . Several nonmutually exclusive factors may explain this result: ( 1 ) other macromolecules present in BH ( e . g . , lipids , nucleic acids , other proteins ) compete with PrPTSE for Mte binding sites; ( 2 ) prion protein is more aggregated in the purified PrPTSE preparation than in BH [45] , and sorption to Mte reduces PrPTSE aggregate size , increasing specific titer [41]; and ( 3 ) sorption of macromolecules present in BH to Mte influences mineral particle uptake in the gut by altering surface charge or size , whereas the approximately 1 , 000-fold lower total protein concentration in purified PrPTSE preparations did not produce this effect . We previously showed that other inorganic microparticles ( kaolinite and silicon dioxide ) also bind PrPTSE [22] . All three types of microparticles are widely used food additives and are typically listed as bentonite ( Mte ) , kaolin ( kaolinite ) , and silica ( silicon dioxide ) . Microparticles are increasingly included in Western diets . Dietary microparticles are typically inert and considered safe for consumption by themselves , do not cause inflammatory responses or other pathologies , even with chronic consumption , and are often sampled in the gut and transferred from the intestinal lumen to lymphoid tissue [39 , 46 , 47] . Our data suggest that the binding of PrPTSE to dietary microparticles has the potential to enhance oral prion disease transmission and warrants further investigation . In conclusion , our results provide compelling support for the hypothesis that soil serves as a biologically relevant reservoir of TSE infectivity . Our data are intriguing in light of reports that naïve animals can contract TSEs following exposure to presumably low doses of agent in the environment [5 , 7–9] . We find that Mte enhances the likelihood of TSE manifestation in cases that would otherwise remain subclinical ( Figure 3B and 3C ) , and that prions bound to soil are orally infectious ( Figure 5 ) . Our results demonstrate that adsorption of TSE agent to inorganic microparticles and certain soils alter transmission efficiency via the oral route of exposure .
Syrian hamsters ( cared for according to all institutional protocols ) were experimentally infected with the HY or DY strain of hamster-adapted TME agent [48] . Brain homogenate , 10% w/v , was prepared in 10 mM NaCl . PrPTSE was purified to a P4 pellet from brains of hamsters infected with the HY strain using a modification of the procedure described by Bolton et al . [49 , 50] . The P4 pellet prepared from four brains was resuspended in 1 mL of 10 mM Tris ( pH 7 . 4 ) with 130 mM NaCl . In the subset of experiments using PrPTSE–Mte complexes , larger prion aggregates were removed from the preparation by collecting supernatants from two sequential 5-min centrifugations at 800 g ( clarification ) . Protein concentrations were determined using the Bio-Rad ( http://www . bio-rad . com ) DC protein assay as directed by the manufacturer's instructions . Four types of Mte- or soil-containing inocula were prepared: BH–Mte mixtures , PrPTSE–Mte mixtures , PrPTSE–soil mixtures , and PrPTSE–Mte complexes ( see below ) . To prepare mixtures of BH or PrPTSE with Mte , the indicated amount of 10% brain homogenate ( Figures 1 and 2 ) or PrPTSE ( Figure 3 ) was added to 500 μL of 10 mM NaCl in the presence or absence of 500 μg of Na+-saturated Mte ( particle hydrodynamic diameter = 0 . 5–2 μm ) ( prepared per [51] ) . Mixtures of PrPTSE and whole soils ( Figure 5 ) were prepared by adding 1 μg of PrPTSE to 500 μL of 5 mM CaCl2 in the presence or absence of 1 mg of each soil type . Samples were rotated at ambient temperature for 2 h , like samples were pooled , and the equivalent of 500 μg of Mte or 1 mg of whole soil was orally inoculated into each hamster . We previously showed that absorption of purified PrPTSE to Mte was complete within 2 h [22] . Isolated PrPTSE–Mte complexes were prepared as previously described [22] . Briefly , the indicated amount of clarified PrPTSE ( 200 or 20 ng , Table 1 ) was added to 500 μg of Mte in 10 mM NaCl ( 500 μL final volume ) per sample . Mixtures were rotated at ambient temperature for 2 h . Each PrPTSE–Mte suspension was placed over a 750-mM sucrose cushion prepared in 10 mM NaCl and centrifuged at 800 g for 7 min to sediment mineral particles and adsorbed PrPTSE . PrPTSE–Mte complexes were resuspended in 500 μL of 10 mM NaCl and pooled . The equivalent of 500 μg of Mte was orally inoculated into each hamster . To control for potential sedimentation of unbound PrPTSE , “mock” samples lacking Mte were processed identically , and any sedimented material was inoculated into hamsters . As a positive control , unbound PrPTSE ( 200 or 20 ng ) was orally administered to hamsters . All oral inoculations were via pipette and voluntary consumption . Following oral dosing , hamsters were observed twice weekly for the onset of clinical symptoms [48] for at least 300 d , a period of time found sufficient to observe most or all clinical cases [52] . Immunoblotting was performed as previously described [22] . Briefly , proteins were separated by SDS-PAGE ( 4%–20% gradient for analysis of inocula , 15% for analysis of brain PrP ) , transferred to polyvinyl difluoride membranes , and immunoblotted with the PrP-specific antibody 3F4 ( 1:40 , 000 dilution ) . Detection was achieved with HRP-conjugated goat anti-mouse immunoglobulin G . The quantity and characteristics of PrPTSE dosed in Table 1 and Figure 3 were compared by immunoblot analysis ( Figure S2A and S2B ) . For both unbound and Mte-bound PrPTSE inocula , a 50- μL aliquot ( one-tenth the total volume ) of each 200-ng or 1-μg sample of PrPTSE ( Figure 3 and Table 1 , respectively ) was removed following the 2-h incubation . Samples with 20 ng PrPTSE were not consistently detectable by immunoblot analysis . Mte was sedimented by 1-min centrifugation at 14 , 000 g , and PrP was extracted for 10 min in 5 μL of 10× sample buffer ( 100 mM Tris [pH 8 . 0] , 10% SDS , 7 . 5 mM EDTA , 100 mM dithiothreitol , and 30% glycerol ) at 100 °C . While still hot , Mte was sedimented by brief centrifugation , and the supernatant containing extracted PrP was diluted with 10 mM NaCl to a total volume of 50 μL . Sample buffer was added to the unbound PrPTSE samples to a 1× final concentration , and samples were heated at 100 °C for 10 min prior to SDS-PAGE and immunoblotting . Analysis of the sorption of PrPHY and PrPDY from brain homogenate to Mte was performed as previously described [22] . Brains from hamsters orally dosed with unbound PrPHY or Mte-bound PrPHY were homogenized to 10% w/v in PBS . For samples without PK , 10 μL of BH was mixed 1:1 with 10× sample buffer and heated at 100 °C for 5 min . Other samples ( 30 μL ) were treated with PK ( 50 μg·mL−1 ) for 30 min at 37 °C . Phenylmethylsulfonyl fluoride was added to achieve a concentration of 1 mM to block PK activity , and samples were diluted 1:1 with 10× sample buffer and heated at 100 °C for 5 min prior to SDS-PAGE and immunoblotting . Multivariate Cox proportional hazards regressions [53] were used to estimate the effects of PrPTSE dose ( using log10 PrPTSE dose as a continuous variable ) and Mte inclusion on times to onset of clinical symptoms [54] . Several diagnostic procedures were performed to assess the validity of the Cox regressions . First , interaction in the statistical model between Mte and PrPTSE dose was tested and found to be far from significant ( p = 0 . 92 ) ; this interaction was therefore excluded from further consideration . Second , comparison of the linear fit with the three-level dose factor indicated that the nonlinearity of the log10 prion dose covariate was nonsignificant ( p = 0 . 21 ) ; log10 prion dose was therefore retained as a continuous covariate . Last , cumulative hazard curves were approximately parallel and simple diagnostics for proportionality [53] showed the assumption of linearity to be appropriate . Equivalence factors ( the dose multiplier equal in effect to adding Mte to a sample ) can be derived as 10 raised to the inverse ratio of the Mte and dilution coefficients . A 95% CI for the ratio was generated using Fieller's method and then exponentiated to produce a CI for the factor [53] . All Cox analyses were performed using S-PLUS version 3 . 4 [55] .
The GenBank ( http://www . ncbi . nlm . nih . gov ) accession number for PrP is M14054 . | Transmissible spongiform encephalopathies ( TSEs ) are a group of incurable neurological diseases likely caused by a misfolded form of the prion protein . TSEs include scrapie in sheep , bovine spongiform encephalopathy ( “mad cow” disease ) in cattle , chronic wasting disease in deer and elk , and Creutzfeldt-Jakob disease in humans . Scrapie and chronic wasting disease are unique among TSEs because they can be transmitted between animals , and the disease agents appear to persist in environments previously inhabited by infected animals . Soil has been hypothesized to act as a reservoir of infectivity and to bind the infectious agent . In the current study , we orally dosed experimental animals with a common clay mineral , montmorillonite , or whole soils laden with infectious prions , and compared the transmissibility to unbound agent . We found that prions bound to montmorillonite and whole soils remained orally infectious , and , in most cases , increased the oral transmission of disease compared to the unbound agent . The results presented in this study suggest that soil may contribute to environmental spread of TSEs by increasing the transmissibility of small amounts of infectious agent in the environment . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods",
"Supporting",
"Information"
] | [
"infectious",
"diseases",
"public",
"health",
"and",
"epidemiology",
"mathematics",
"none",
"science",
"policy",
"neurological",
"disorders",
"gastroenterology",
"and",
"hepatology",
"in",
"vitro",
"mus",
"(mouse)",
"mammals"
] | 2007 | Oral Transmissibility of Prion Disease Is Enhanced by Binding to Soil Particles |
A common theoretical view is that attractor-like properties of neuronal dynamics underlie cognitive processing . However , although often proposed theoretically , direct experimental support for the convergence of neural activity to stable population patterns as a signature of attracting states has been sparse so far , especially in higher cortical areas . Combining state space reconstruction theorems and statistical learning techniques , we were able to resolve details of anterior cingulate cortex ( ACC ) multiple single-unit activity ( MSUA ) ensemble dynamics during a higher cognitive task which were not accessible previously . The approach worked by constructing high-dimensional state spaces from delays of the original single-unit firing rate variables and the interactions among them , which were then statistically analyzed using kernel methods . We observed cognitive-epoch-specific neural ensemble states in ACC which were stable across many trials ( in the sense of being predictive ) and depended on behavioral performance . More interestingly , attracting properties of these cognitively defined ensemble states became apparent in high-dimensional expansions of the MSUA spaces due to a proper unfolding of the neural activity flow , with properties common across different animals . These results therefore suggest that ACC networks may process different subcomponents of higher cognitive tasks by transiting among different attracting states .
To fully understand how neural processes give rise to cognitive operations , it is essential to reconstruct the underlying neural network dynamics from electrophysiological or neuroimaging measurements in relation to behavior . A common theoretical idea is that these dynamical properties of the nervous system , like the convergence of activity to specific stable population patterns ( attractors ) , are what ultimately implement the computational operations that link inputs to outputs [1]–[6] . For instance , different attracting states may represent different active memories or cognitive entities , and movement between these states may correspond to the recall of a memory sequence or the execution of a behavioral or motor plan . Attractor states as a basis for cognition received particular attention in the context of working memory [2] , [4] , [7]–[9] and decision making tasks [5] , [10]–[12] . Especially in recent years , along with the advances in multiple single-unit recording techniques [13] , there has been a dramatic rise in the attempts to reconstruct cognitively relevant aspects of the population dynamics . Many of these relied on methods from multivariate statistics and machine learning ( as reviewed in [14] , [15] ) . These studies gave a number of valuable insights into mechanisms of neural information processing like the information content of the transient dynamics connecting steady states [16] , [17] , the representation or processing of stimuli by reproducible sequences of states [18] , or the sudden nature of transitions among representational states during learning [19] . Several experimental studies also suggested that spatial representations in the rodent hippocampus [6] , [20]–[22] or olfactory representations in zebrafish [16] , [23] may have attractor-like properties with sometimes stochastic transitions among them [24] , [25] . In these studies , attractor states were indicated by discrete switches in the population activity patterns eventually attained ( after some transient ) when stimulus parameters were continuously varied . Strictly speaking , however , these studies did not attempt to explicitly demonstrate a convergent flow of neural trajectories ( as sometimes pointed out by the authors themselves , [23] ) , as another important signature of attracting states . Moreover , they mostly focused on ( stimulus-driven ) sensory or spatial representations rather than on presumably intrinsically-driven higher cognitive processes . In addition , since most of these previous approaches worked directly in the space of observed variables , i . e . the recorded units' firing rates or spike times , they could potentially miss some important structural details of neural space organization , especially in high-noise situations , as they try to infer the dynamics of a large complex system by selecting only a few of its dimensions ( recorded neurons ) . Thus , experimental evidence for the hypothesis that higher cognitive processes proceed by moving among attracting states is still sparse . Here we combined and adapted two approaches well established in statistical learning theory [26] , [27] and nonlinear time series analysis [28] , [29] in an attempt to move beyond some of the limitations that could arise in previous analyses of electrophysiological data . These methods were applied to multiple single-unit recordings from the rat anterior cingulate cortex ( ACC ) during a complex memory-guided decision making task in a radial arm maze ( Figure S1 ) . The ACC is assumed to play a key role in higher-level cognitive processes like monitoring of behavior [30] , processing error feedback [31] , making choices [32] and dissecting task structure [33] . Thus , the ACC is a brain area with complex intrinsic dynamics and computational properties that presumably demand a sophisticated multivariate analysis to much larger degree than comparatively simpler early sensory systems ( e . g . [16] , [23] ) . The present analysis was designed to be more sensitive to potential state space structure , suggesting previously unrecognized convergence properties of ACC neural ensemble states associated with cognitive processing steps and stable across multiple trials .
A state space is a coordinate map spanned by all relevant dynamical variables of a system ( e . g . the membrane voltages or firing rates of neurons ) . A single ( vector ) point in this space represents the whole state of the recorded neural system at a given point in time ( e . g . the current firing rates of all neurons ) , while a trajectory in this space charts how its state changes over time . Most computational theories of the brain work by linking geometrical objects in these spaces ( e . g . attractors ) and the temporal evolution of neural activity ( the trajectories ) to specific computational and cognitive functions ( e . g . [2] , [4] , [34]–[37] ) . However , inferring the dynamics of a large complex system from experimental data by selecting only the observable dimensions ( recorded neurons ) can lead to incorrect conclusions [28] , [29]: Neural trajectories may not be sufficiently “unfolded” , i . e . may follow apparently convoluted patterns where they frequently “intersect” themselves and exhibit ambiguities with regards to their direction of flow ( Figure 1A , left ) . This is due to the fact that other dimensions along which the flow would have been disambiguated are missing ( e . g . the third axis in Figure 1A , top left; [28] , [38] ) . Thus , a state space construed solely from the activities of the simultaneously recorded units ( termed multiple single-unit activity , MSUA , space in the following ) is not guaranteed to properly represent the geometry of the underlying dynamical system's attractors . A potential solution to this problem was provided by time series embedding theorems [28] , [38] which demonstrated that the structure of the underlying attractor dynamics could be fully recovered ( under ideal , noise-free conditions ) if the dimensionality of the space is expanded by adding a sufficient number p of time-lagged versions ν ( t-τi ) of the present observations ν ( t ) as new variables to the space , where the time lags τi are determined such that these new variables do not contain redundant information with respect to the original MSUA axes , i . e . , are only weakly correlated with them ( Figure 1A , center ) . In principle , the optimum number of delay axes is constrained by the dimensionality of the underlying attractor of the system [28] . Unfortunately , however , due to the sparseness of the MSUA spaces and the noise levels in these data it cannot be reliably computed . Moreover , given that for neural systems the true dimensionality could be ( much ) higher than the number of dimensions one has experimentally access to , the number of time lags required for a statistically optimal disambiguation of trajectory flows may be so high that it cannot be accommodated by the ( experimentally ) limited length of the time series ( Materials and Methods ) . Therefore , it may be necessary to consider also other types of state space expansion that allow to effectively discern the neural dynamics associated with different cognitive events . Adding interactions between units' firing rates as dimensions to the space seems a particularly suitable choice since neuronal cross-correlations have often been postulated to play an important role in cognitive processes ( e . g . [39]–[43] ) . From a mathematical point of view products of neural firing rates would correspond to terms of a multinomial basis expansion frequently employed in statistical classification procedures [27] . Hence , such an expansion would have both a neuroscientific meaning and a theoretical foundation . Therefore , in our approach the delay-coordinate ( DC ) map of the MSUA space ( DC-MSUA space ) is further expanded by adding pairwise and higher order cross-products of the recorded units' firing rates , up to some order O , as new dimensions . For example , an expanded state space of 3rd order will contain all the original MSUA axes , plus time-delayed versions of the firing rates of all n recorded units , ν1 ( t-τ1 ) , ν2 ( t-τ2 ) , … , νn ( t-τn ) as well as new axes corresponding to third order products like ν1 ( t-τ1 ) ν2 ( t-τ2 ) ν3 ( t-τ3 ) or ν1 ( t-τ1 ) 2 ν3 ( t-τ3 ) . Vectors in these high-dimensional spaces will be denoted by Φ ( t ) – each such vector corresponds to a specific ( spatio-temporal ) pattern of neural firing rates and firing rate correlations up to the order set by the expansion . Since the dimensionality of such spaces can be extremely high , specialized algorithms ( so-called kernel-methods [44]–[46] ) were used for the statistical analyses , as discussed below . As illustrated in Figure 1A ( right ) , adding these cross-product terms can help to further disentangle neural trajectories by amplifying small differences present in the DC-MSUA space . Why this 2-stage process in expanding the original MSUA space ? If trajectories in the originally recorded MSUA space are already nicely disentangled and noise levels are very low , no further expansion may be necessary . However , many of the simultaneously recorded neurons may fire very sparsely , or may otherwise be non-informative about the system's dynamics , or there may simply not be enough of them which access “sufficiently different aspects” of the system's dynamics . Adding delay coordinates ( with delays chosen such as to minimize cross-correlations among the firing-rates of different neurons , see Materials and Methods ) will increase the amount of information about the neural dynamics captured by the space by removing ambiguities in the neural flow which may occur in the MSUA space ( Figure 1A , center ) . Adding product terms , on the other hand , may not add further information about the dynamics to the space ( although it may make information contained in neuronal correlations explicitly accessible ) , but it will help to pull trajectories apart and thus enhance task-related differences in the activity flow in situations of high noise ( Figure 1A , right; see also Materials and Methods ) . It may also take care of the fact that putative attractor geometries may be highly nonlinear structures that are not easily captured by linearly separating hyperplanes . Hence , by combining these two types of expansion we arrive at a space which should be both , more informative due to the addition of delay coordinates , and at the same time “less noisy” and more apt for detecting nonlinear structures . Here we show that the identification of ensemble dynamics for different animals and behavioral performance levels will , in general , indeed significantly improve by combining both types of expansion . As an example , Figure 1B shows a single trial of an animal performing a higher cognitive task explained in the next section . A type of principal component analysis ( PCA ) suitable for very high-dimensional Oth order spaces , termed kernel-PCA [47] ( for O = 1 equivalent to conventional PCA ) , was used to visualize the neural dynamics in the 3 most variance-explaining dimensions . While for both the MSUA and O = 5 spaces the two illustrated task phases ( blue and red dots in Figure 1B ) can be clearly discerned , the actual trajectories ( the lines connecting the dots ) are quite entangled in the MSUA space but are nicely unfolded for high-order expansion spaces , exposing attracting orbits and properties of the two task phases ( Figure 1B; see also Video S1 ) . The techniques introduced above were used to analyze MSU recordings obtained from the rat ACC ( Figure S1 ) while the animals were situated in a radial-arm-maze decision-making task with temporal delay ( Figure 2A ) . This task is considered to be ecologically valid in the sense that it mimics key aspects of rats' natural foraging , food-hording , and retrieval behavior ( e . g . [48] , [49] , [33] ) . The entire time on task was divided into six epochs with differing cognitive demands as illustrated in Figure 2A ( see Materials and Methods for precise definition of the cognitive epochs ) . Two data sets were available for the present analyses: 1 ) Three animals recorded for up to 15 trials solely for the purposes of the present study . From these , only trials with good performance were selected ( = less than 3 test phase errors; median errors across all trials were 1 , 0 . 5 and 2 for respectively for each animal ) , with an error defined as re-entrance into an arm from which food was already retrieved . 2 ) Six animals recorded for one or two trials from a previous study [33] , which will be used to further confirm the results obtained with the “multiple-trial animals” and to conduct an explicit comparison of high ( <2 errors ) vs . low ( >4 errors ) performance trials . Average trial duration ( ±SEM ) was 159 . 3±19 . 7 s across all trials and animals . With a standard binning for the spike density estimates of 0 . 2 s , this resulted in an average of 797±99 firing-rate vectors per trial ( see further below for a discussion on data size effects ) . To provide a direct comparison with previous approaches for constructing neural state spaces , Figure 2 shows three dimensional projections obtained in different ways from the first five trials of one of the multiple-trial animals which performs the task with less than three errors per trial . Consistent with our previous observations [33] , the MSUA space shows a visually apparent segregation among the different task epochs ( indicated by the color-coding ) , using either PCA ( Figure 2B , left ) or multi-dimensional scaling ( MDS; Figure 2B , right ) for the 3-dimensional reconstruction . Figure 2C shows the same data projected into a 3-dimensional space using a Fisher discriminant analysis technique ( FDA; see e . g . an application to MSUA spaces in [19] ) . Like PCA , FDA amounts to just a linear transformation of the original variables . However , unlike PCA , the directions sought are such that the differences between group means are maximized while at the same time within-group jitter is minimized along them ( for the Oth order higher-dimensional spaces we used a regularized kernel-FDA which is equivalent to a standard ( regularized ) FDA for O = 1 [50]; see Materials and Methods and Text S1 ) . The figure displays the flow field in addition to the data points , i . e . the speed and direction of movement of the neural population state at each time bin ( computed as the difference between temporally consecutive vector pairs ) . While the flow field in the FDA-reduced original MSUA space may appear relatively disordered ( Figure 2C , left ) , in the expanded space ( Figure 2C , right ) a consistent movement into each of the task related clusters at points far from any cluster center appears to occur ( as will be statistically confirmed below ) . In summary , these 3-dimensional visualizations seem to suggest that different cognitively defined task epochs are associated with different population states which exhibit attractor-like properties ( convergence of flow ) , a phenomenon that becomes apparent only after expanding the spaces to sufficiently high dimensionality using the techniques outlined in the previous section . We stress that , in principle , expansion of spaces to much higher dimensionality is a well-known technique in statistical classification approaches to improve the linear separability of classes [27] . However , a serious statistical issue with such approaches is the potential problem of “over-fitting” the data: For instance , n+1 points can always be perfectly linearly separated in a n-dimensional space , even if their configuration is purely random . To circumvent this problem , two approaches which are standard in statistics ( e . g . [46] ) and machine learning ( e . g . [44] ) were employed here: First , a regularization term ( fixed throughout the study; Eq . S3 in Text S1 ) , which penalizes model complexity and thus reduces the efficient dimensionality of the fitted classifier ( typically way beyond the nominal dimensionality ) , was included in the optimization criterion for the kernel-FDA . The technique of cross-validation ( e . g . [44] , [46] ) is used in the next section for deriving this regularization term and the expansion order optimal for across-trial predictions ( Materials and Methods ) . Over-fitting would imply poor generalization to new data sets not used for fitting the classifier , i . e . a high out-of-sample prediction error across trials . Second , the performance of the classification statistics on the original data was compared to bootstrap data in which the relation between neural population vectors and cognitive-class labels has been randomized . Such bootstrap samples have to be devised carefully such that they retain features of the original time series ( like their temporal autocorrelations ) which are not necessarily related to task-imposed structure , as explained in the sections to follow . For determining the optimal state space we assessed whether the assignment of population-interaction patterns to task epochs could be correctly predicted in a test set of trials based on information obtained solely from a non-overlapping training set of trials , or , from another perspective , how stable the task-epoch-specific clusters in Oth-order expansion space are across multiple trials . To these ends , state spaces were reconstructed exclusively from the first set of 4 to 8 well-performed trials , and data points from the ( non-overlapping ) set of the last 4–8 well-performed trials were projected into this space ( “forward predictions” ) . Vice versa , “backward predictions” from the last to the first trials were also obtained . If the neural dynamics remain largely invariant across multiple trials , then vector points on any subsequent trial should fall into the same clusters derived only from the first few trials . This analysis was performed for any pair of task epochs using the most discriminating direction as obtained by kernel-FDA within the expanded high-dimensional spaces . Assuming that the projections of the Oth-order population vectors from any two task epochs onto this maximally separating direction are normally distributed ( which will almost inevitably be the case due to the central limit theorem , as the projections are sums of many random variables ) , for each population pattern ν ( t ) the probabilities P ( ν ( t ) |C1 ) and P ( ν ( t ) |C2 ) that it comes from one task-epoch or the other can be evaluated . Assigning population vectors to task epochs based on these probabilities yields a segregation error ( SE ) for each pair of task epochs defined as the relative number of misclassified population patterns ν ( t ) ( see Materials and Methods for discussion of further advantages this brings over other kernel-based approaches ) . By chance this misclassification rate will be 50% since we fixed the prior probabilities P ( C1 ) and P ( C2 ) at 0 . 5 for any pair of epochs , such that the results would not be biased towards the longer-lasting epochs . Note that all time bins ( population vectors ) from a given task epoch class were entered into this analysis , regardless of whether they came from the same or from different trials . For checking predictability across trials , the crucial aspect now is that the optimal discriminant direction was solely obtained from the first ( or last ) couple of ( reference ) trials , and then fixed and used for out-of-sample predicting the corresponding misclassification rate SEpredic ( for “predicted SE” ) of population interaction patterns to task-epochs for the non-overlapping set of last ( or first , respectively ) prediction trials ( see Materials and Methods for more details ) . To evaluate the significance of the observed SEpredic , bootstrap data were constructed by randomly shuffling stretches of the ν ( t ) vector time series that retained entire trajectories form a given specific task epoch , i . e . each bootstrap replication preserved all temporal autocorrelations up to the length of the relevant task epochs . Consistent with the visual displays presented above , for O∼5 SEpredic was significantly lower ( p<0 . 01 ) in the original as compared to the bootstrap data ( Figure 3A; see Figure S2 for a schema on bootstrap construction ) . Note , however , that SEpredic for the bootstraps is also less than what would be expected by chance , i . e . <0 . 5 , such that prediction accuracy in the bootstraps is above chance level . This is because the bootstraps retain original auto-correlations as indicated above , which by themselves may induce some state space clustering , irrespective of task-epoch membership . Surprisingly , in contrast to the case O∼5 , for O = 1 ( i . e . , within the DC-MSUA space ) predictability across trials was not significantly better in the original than in the bootstrap data . Thus there does not seem to be sufficient information in the lower-dimensional state spaces to allow prediction of population pattern assignments across trials . Rather , given the experimental noise and the potentially nonlinear state space structures , neural interactions have to be included to establish stable associations between task epochs and population patterns , or , in other words , further trajectory separation beyond the one achieved by delay-coordinates is indeed necessary to reveal across-trial stability . Specific comparisons for each pair of task epochs are shown in Figure 3B . Finally , for O>5 predictability starts to deteriorate again . Hence , it seems that there is a maximum order of activity products which would be required to optimally resolve task-epoch-related structure in the neural state spaces , a finding consistent across the different data sets studied ( Figure 3C ) . We emphasize that this result does not imply that neural activity interactions up to some precise order ( 3rd–5th ) are important– it only shows that below or above a certain expansion order generalization performance degrades , which can be the case for purely statistical reasons ( i . e . , simply because there are too few data or too few simultaneously recorded neurons to reliably estimate the optimum order of interactions ) . On the other hand , the optimal orders we obtained do not seem to be completely arbitrary ( in the sense of being determined purely by the number of data points and recorded units ) : First , similar optimal orders were also observed for the other two animals ( Figure 4 ) which differed in the number of recorded units ( 18 , 13 and 21 , respectively ) and the size of the training and prediction sample sets ( 5 , 8 and 4 trials , respectively ) . Second , we performed additional controls by including subsets of neurons of differing size ( Figure 4 , upper left ) and by artificially augmenting or decimating the data sets in a way that preserved the original distributions ( Figure 4 , right ) . Hence , we conclude that there is an organization of task-related population interaction patterns predictable across many trials which is optimally revealed by expanding the MSUA space by taking higher orders of activity interactions into account . In a previous study [33] we had compared animals performing well on the task to animals which committed a lot of behavioral errors . We observed that in animals performing poorly state space segregation ( task-epoch-dependent clustering ) was generally comprised compared to trials on which only few ( 0 or 1 ) errors were committed . Here we re-addressed this issue using the methods developed above ( Figure 5 ) . Data from 8 trials ( coming from 4 different animals ) performing with less than two errors ( = “good performers” ) and 8 trials ( coming from 5 animals ) with more than four errors ( = “bad performers” ) were used . These two groups of trials were combined into two separate data sets for analysis ( termed “single-trial” datasets ) . This works since the basic structure of the cognitively-defined classes was the same for all animals , i . e . , the task obviously was the same for all animals , and population patterns specific for different task episodes like choices , rewards , or the delay phase , were a common feature of ACC activity . Since only a single trial with electrophysiological recordings , however , was generally available from each of these animals , results were cross-validated by removing each single one of the animals from the data set in turn ( i . e . , a jackknife validation [51] ) . Consistent with our previous observations [33] , discriminability in the MSUA space is significantly worse ( Wilcoxon rank-sum test T13 = 113 , p<0 . 05 ) for “bad performers” ( Figure 5A , dark curve for O = 1 ) when compared to “good performers” ( Figure 5A , gray curve for O = 1 ) . However , as Figure 5A shows , for both groups discriminability significantly increases just up to expansion orders of about 5 , i . e . the segregation error ( SE ) as defined further above ( computed from FDA with the same regularization as above , see Materials and Methods ) significantly decreases ( Wilcoxon ranksum tests , p<0 . 03; see details in Figure 5 legend ) . Thus , as the maximum order O of the reconstructed state space is increased , cognitively relevant features of the neural dynamics are increasingly better resolved to the extent that an organized dynamics becomes evident even in situations where previous methods had failed ( see [33] ) . However , as for the multiple-trials data analyzed in the previous section , SE for O>5 grows again for both groups ( Figure 5A ) , suggesting once again that there may be a maximum order of activity interactions for which trajectories are optimally resolved . Finally , and again consistent with previous results [33] , although SE decreases for both groups , there still remains a significant difference between the low and the high performance groups even for O>3 ( Wilcoxon test , p<0 . 04 ) , confirming that still some of the state space organization is corrupted in bad performers . Detailed task-epoch comparisons are shown in Figure 5B . Similar results were obtained with information-theoretic measures of task-epoch segregation like the relative entropy ( Kullback-Leibler divergence , e . g . [44] ) between the conditional probability distributions of task-epochs given a specific firing-rate vector ( Figure 5C; see Materials and Methods section ) . Moreover , further control analyses indicated that results are not significantly altered by using state spaces constructed by using different types of expansion , other classification criterions , or other smoothing parameters for the spike trains ( as shown in Figure S3 ) . The most interesting aspect of the present methodological approach is that it permits to examine the flow of neural trajectories during performance of a cognitive task , dynamical properties that may not be well accessible in the unprocessed representation of MSU activity as demonstrated in the previous sections ( Figure 3B , left ) . Here we analyzed the attracting behavior suggested by the three-dimensional visualizations more systematically . First , a simple statistical approach was taken . Activity flows were evaluated in the low-dimensional kernel-PCA projections of task epochs , since velocity vectors cannot be reliably obtained in the extremely high-dimensional expanded spaces ( for similar reasons for which we used kernel methods before; see Figure S4 and Text S2 for further discussion ) . Figure 6 displays the speed of movement at each data point in these projections as a function of the likelihood of a population pattern given the task epoch to which it belongs , i . e . p ( ν ( t ) |correct task-epoch classification ) , evaluated using FDA in the high-dimensional Oth-order spaces for the prediction set of trials ( see Figure 3 ) . If the task-epoch states have indeed attracting properties , one would expect that vector points which exhibit little movement should have a high likelihood of correct classification , reflecting the fact that these points should be found close to the cluster centers . Consistent with the idea that in low-order spaces trajectory flows should appear convoluted and disordered , for O = 1 velocities were evenly distributed across all regions of the state space , i . e . the velocity of movement of the neural state was largely independent of the likelihood of correct classification ( Figure 6 , left-top; O = 1 ) . In contrast , for higher-order expansions the likelihood of correct classification rapidly falls off as the speed of neural state changes increases ( Figure 6 , left-bottom; O = 5 ) , confirming that regions where trajectories move quickly are on average far from the cluster centers . Although these results are suggestive , they by themselves do not conclusively rule out alternative explanations unrelated to the potentially attracting nature of the task-specific ensemble states , e . g . the tendency of extreme values to be followed by values closer to the mean simply by laws of probability ( “regression to the mean” ) , auto-correlative properties of the time series , or by systematic deformations of the flow field induced by PCA . To statistically control for such alternatives , we performed a bootstrap test . The right column of Figure 6 shows results from the same analysis as performed on the bootstrap data when the temporal sequence of binned firing rates was inverted for all neurons within task-epochs . Therefore , task-epoch-specific lengths are preserved , but any causal relationships in the original time series are destroyed . For O = 1 , the correct classification likelihood as a function of velocity behaves similar for bootstrap and original time series , but at higher expansion orders the fall-off of correct classification likelihood with vector velocity is significantly less steep in the bootstrap than in the original time series ( paired t-test between the two slopes , p<0 . 001 for O = 5 , see Figure 6 caption ) as demonstrated by the linear fits to the log-linear graphs . In summary , different cognitively defined task epochs may potentially act as attracting states of the neural dynamics , i . e . regions of state space towards which all trajectories tend to converge with high likelihood and within which they remain bounded for some time . While this analysis suggests attracting behavior related to the task epochs , it was performed on a three-dimensional representation in which velocity vectors could still be reliably determined . We therefore next sought to precisely quantify within the full high-dimensional spaces to which degree the ( mathematical ) conditions defining attracting states were met in the empirical data , with the statistical analysis based on the task-epoch boundaries defined previously . As the definition of these boundaries did not include any knowledge about putative attractor states , there is no a-priori reason why there should be strong convergence over time towards the center of these states . Attracting state conditions are illustrated in Figure 7A which shows a schema of different kind of convergent trajectories in the high-dimensional state spaces . Figure 7B shows within the 3-dimensional PCA projections some empirical examples of such trajectories which either cycle within or return to the task-epoch-specific population states . Figure 7C precisely quantifies , both for the single-trial data sets ( red bars , left y-axis ) and for the prediction-sets of trials in the multiple-trial data ( blue bars , right y-axis ) , the fraction of trajectories which escaped again from the task-epoch specific clusters without returning to them within the given period ( i . e . trajectories which are not of the kind “a” or “b” in Figure 7A ) . For O≈3–5 , consistently across all task epochs this was only the case for ∼15% of the trajectories ( across all 3 animals ) when escape behavior was determined in the prediction trials while event boundaries were those defined in the non-overlapping reference set of trials , as shown in Figure 7C ( blue bars , right y-axis; and ∼8% of the escaped trajectories when assessed within the reference set of trials , see red bars , left y-axis ) . Thus , these results further support the hypothesis that the task-epoch clusters constitute regions of convergence with >80% of trajectories returning to these states or bound within them . In summary , the quantitative analysis of trajectory flows in the optimal state spaces seems to confirm that different cognitively defined task epochs of the present memory-based decision making task act as high probability regions of convergence .
We observed that unfolding of trajectories and separation of task-epoch clusters became stable across trials when higher-order activity products were taken into account , but did not improve further when moving to arbitrarily high expansion orders . This , in other words , seems to imply that considering the joint activity constellations of a couple of neurons will still add information about the neural dynamics not easily or directly available from single unit activities , while still higher-order interactions may not be relevant: For sub-optimal state spaces the clustering into task-epoch-specific patterns was either unclear ( O = 1 ) or had no predictive power across trials ( O>6; cf . Figure 3 ) . Note , however , that higher-order activity products are used here mainly as a statistical tool for disentangling trajectory flows and not for assessing the cognitive relevance of neural correlations . Thus , we cannot conclusively rule out , for instance , that adding many more neurons and data points to the state spaces than were available in the present study would shift the optimal expansion dimensionality to different orders . The specific value for the optimal expansion order obtained here may just reflect the well-known ( in statistics; e . g . [46] ) “bias-variance tradeoff” for our data set ( in the sense of yielding low generalization errors , i . e . without over-fitting the data ) . Nevertheless it is still remarkable that for all the different types of data sets studied here ( multiple-trials vs . many animals ) , different numbers of recorded units , and different numbers of trials ( and hence data points ) a similar order of activity interactions appeared to be optimal . Similarly , the control studies reported in Figure 4 suggest that sample size effects cannot completely account for the specific optimality value obtained here . Indeed , a recent study , performed in visual cortex , revealed the importance of higher-order correlations in local neural ensembles like recorded here , while only second-order correlations seemed to be the relevant for information transmission across larger cortical distances [65] . The importance of higher-order correlations among neurons for information processing has also been stressed by many previous authors [43] , [66] , [67] , e . g . by relating multiple-spike coincidence statistics to significant behavioral events [40] , [42] , [68] , or by computing the information gained from correlations while decoding the current stimulus from the neural activity [69] . Some research had suggested that higher than second order correlations are redundant , at least in some preparations like the retina which may strongly differ in their structural and computational properties from the neocortex [67] . On the other hand , most recently it was suggested that some of the low bounds found in earlier studies may be an artifact of the limited number of experimentally accessed units [69] . Finally , studies in somatosensory cortex also found similar bounds on the maximum order of perceptually relevant neural activity interactions as suggested here [66] . Within the optimal order expansion spaces , the stable and attracting nature of the task-epoch-specific states became apparent ( cf . Figure 7 ) : The neural dynamics progressively slows down as trajectories approach the cluster centers ( Figure 6 ) and the majority of trajectories cycles within or returns towards these states ( Figure 7 ) , indicating that there should be bounded regions of the neural state space which capture and contain neural trajectories . Just like in most previous studies indicating attractor-like dynamics ( e . g . , [16] , [20] , [23] ) , we cannot rule out , however , that these states are stimulus-driven , i . e . become attracting states only under the influence of certain ( sensory or motor ) stimulus conditions , rather than being a property of the intrinsic ( autonomous ) dynamics . For instance , in Wills et al . [20] or in Niessing and Friedrich [23] the different “categorical” steady state population responses which reflect attracting dynamics are observed for different types of external stimuli ( spatial layout of a maze in the first and olfactory composite stimuli in the second case ) . Likewise , in our case specific spatial , motor , olfactory , or visual properties may be associated with the choice and reward periods . There are three observations , however , which make it less likely that only external factors account for establishing different attracting states: First , also the delay period where the animals are confined to one arm of the maze and lights are switched off approximately acts as an attracting set of the dynamics , just like the other task epochs ( Figures 3B and 5B ) . Second , the training and test epoch choice periods act as separate attracting states although they should share all sensory and motor features , but differ only in their memory requirements . Third , task-epoch specific states break down if the animals commit a lot of behavioral mistakes in the test period , yet one would assume that they experience similar sensory input and perform similar movements at each choice point . Thus , there must be some internal component in the generation of task-epoch specific states . Nevertheless , true attractor states as mathematically defined ( e . g . [70] ) may be unlikely to exist in such an extremely non-stationary and high-dimensional complex system like the neocortex – rather , it seems more likely that neural information processing proceeds by stochastically itinerating among “semi-attracting” states which , for instance , may attract trajectories along most dimensions yet allow them to escape again along others [71] . This idea underlies many more recent conceptualizations of neural information processing ( e . g . [35] , [72] ) , and has also been advanced as a theoretical explanation of experimental results on sensory processing in locusts [16] , [73] . For instance , a specific population activity pattern may be temporarily stable until some slow negative feedback mechanism has build up sufficiently to inhibit this currently active configuration [74] , or until noise has driven the system out of this state again , i . e . until a stochastic transition between states has occurred [24] , [25] , [75] . It will be very difficult or even impossible to experimentally prove in such a high-dimensional and almost never stationary system under constant bombardment from external sources that any neural activity configuration is formally an attractor . Moreover , whether physiological phenomena as the ones reported here really match formal definitions of attracting states may be largely irrelevant from a computational perspective [35] . Rather , neural objects with semi-attracting properties as shown here could serve equally well ( or even better , e . g . with regards to sequence processing ) in most computational ideas about cognitive processing . Does the high expansion order needed to fully reveal the converging dynamics of neural trajectories imply that the attracting states are very high-dimensional ? Not necessarily: The key point of the delay embedding is to add more dimensions which are informative about the dynamics; many of the single-unit firing rate dimensions may be non-informative , i . e . may not contribute much to disentangling trajectories [28] , and thus in principle could be omitted . The multinomial expansion on the other hand primarily serves to optimally pull apart noisy trajectories [45] . In a purely deterministic , noise-free system these dimensions would not be needed either to reveal the attractor . Indeed , the fact that convergent properties of the dynamics could be reasonably well evaluated in the 3-dimensional projections obtained by kernel-PCA suggests that the attracting states may in fact live in much lower dimensional subspaces [57]; which however were only fully revealed by properly expanding the space first before reducing it to the most informative dimensions by using kernel-PCA [76] . Finally , we stress that methods like the ones introduced here are widely applicable to almost any multivariate neural time series , including those obtained from various optical or functional imaging techniques , EEG , MEG [60] , [63] or electrochemical techniques generating spatio-temporal time series . Thus , they may allow to address a number of previously unanswered questions about neural dynamics in many fields that require a proper unfolding and detailed resolution of trajectories not aided by across-trial averaging . Such techniques may also aid the discovery of common dynamical phenomena across tasks , species , and recording techniques . Here they revealed that ACC networks move among different state space regions , defined by specific population constellations of neural firing rates and their interactions , with a high likelihood of attracting neural trajectories . In this manner ACC networks may parse experience into meaningful task-relevant subcomponents .
All animals in this study were treated in accordance with the ethical guidelines set forth by the University of British Columbia and Canadian Council for Animal Care . Briefly , animals were placed on a reverse light cycle upon arrival and given ad libitum access to food for one week . Surgery was then performed and the animals were allowed two weeks of recovery before maze training . For an in depth description of the multi electrode array fabrication and surgical procedures see Lapish et al . [33] . After recovery from surgery , all animals were trained on the delayed spatial win shift run on an eight arm radial maze . Each trial consisted of a training and test phase separated by a one minute delay phase . Prior to the task , the terminal end of all eight arms were baited with a sugar pellet ( Research Diets , Inc . , New Brunswick , NJ , USA ) . The training phase commenced by opening four of eight arms . Upon retrieval of the fourth sugar pellet in the training phase , the animal was locked in the last arm visited and the lights were extinguished for the delay . After the delay , the test phase began by allowing access to all eight arms and errors were scored as re-entries into previously visited arms . Upon completion of the trial by retrieving all eight sugar pellets , all arms were closed and the animal was re-confined to the center of the maze . Animals received one trial per day until they made one error or less for two days in a row , and then received a minimum of 10 trials per day . Data sets for the multiple trials analysis were selected from animals that were able to remain vigilant and attend to the task for ∼15 trials as evidenced by uninterrupted foraging . In order to assess the population dynamic as the cognitive demands of the task vary , the whole time on task was divided into the following six epochs ( Figure 2A ) : reward epochs ( dark gray and red dots ) during the training or test phases , respectively , correct choice epochs during training and test phases ( blue and green , respectively ) , incorrect arm choice periods ( yellow ) during the test phase; and the entire delay period ( light gray ) . Reward epochs were defined as the 1 s periods starting 200 ms before the points where the animal's nose reached a food cup during the training and test phase , respectively . Choice epochs were defined as periods starting 1 . 5 s before the arm choice and finishing 500 ms after it or before the reward period starts ( assessed by visual video inspection ) . Behavioral data were captured via a video camera ( Cohu , Poway , CA , USA ) , recorded in Noldus Ethovision ( Noldus , Leesburg , VA , USA ) , and exported via voltage in real time as Cartesian coordinates to the Neuralyx recording system and then scored offline . All data was acquired with arrays of 24 single-wire tungsten ( diameter = 25 µm , impedance = 150–300 kΏ , California Fine Wire ) electrodes implanted into the ACC ( Figure S1A ) . Recordings were sampled at ∼30 kHz , band-pass filtered from 600–6000 Hz , and stored off-line for sorting and analysis . Spike channels were amplified 5 , 000–10 , 000 times and thresholds for detection were set to ∼50 µV , which corresponded to >5 times the root mean squared noise amplitude for the system . Spike sorting and classification was performed in Neuralynx Spikesort 3D ( Neuralynx , Bozeman , MT , USA ) . Spike cluster assignments were based upon investigation of numerous principle components of the waveform ( ) , and clusters lacking a well-defined boundary were excluded After classification , unrealistically low ISIs ( ≤10 ms ) were removed as well as neurons with unrealistically high cross-correlations indicating the same neuron may have been captured on two different channels . An intuitive introduction to our statistical methodology was provided at the beginning of the Results section , while most of the mathematical details can be found in the Supplementary Material . Spike-trains from the n simultaneously recorded units were convolved with Gaussian functions to obtain statistically reliable estimates of spike densities from single trials ( checking the range σ = 5–200 ms , see Figure S3 for values from 5–50 ms ) , normalized to the length of the whole trial ( to yield a true probability density ) and then summed and binned at 200 ms ( approximately the inverse of the average single unit firing rate ) . Single unit spike densities were then combined into n-dimensional population vectors with components νi ( t ) for each unit i ( e . g . [33] , [59] as a function of time bin t . Small bin sizes ( <50 ms ) produce extremely sparse νi ( t ) series which became computationally prohibitive for the exact algorithm described below , and numerical approximations were required [45] Units for which 〈νi〉 <2% of the most responsive unit were excluded . For the across-trial analysis , three different datasets consisting of 15 trials recorded on the same day were obtained from 3 animals . For each animal , only trials with ≥20 responsive units ( see criterion above ) were selected ( 10 trials from animal #1 , 16 trials from animal #2 , and 8 trials from animal #3 ) . The first set of trials obeying above criteria constituted the reference set ( trials 1–5 for animal #1 , trials 1–8 for animal #2 , and trials 1–4 for animal #3 ) , while the last set of trials in the sequence formed the prediction set selected such that it had no overlap with the reference set . Furthermore , for each task-epoch , time series from the reference and prediction sets were constrained to have about the same length ( number of vectors ) . For each of these two ( reference and prediction ) data sets , for each animal firing-rate vectors were then concatenated across trials to yield the two data matrices which entered into the analysis described further below . From our previous study [33] where animals were run only on a single trial after reaching criterion , two separate data sets from six animals were constructed from 8 trials performing with less than two errors ( “good performers” ) and 8 trials with over four errors ( “bad performers” ) . Neurons from different networks were ordered according to their mean firing-rate , while low-responsive units were excluded as in the multiple-trial dataset . For standard parametric testing , statistical test details can be found in the corresponding figure captions . For testing attracting properties of the task-epoch sets , nonparametric tests were used based on conservatively designed bootstrap data ( 100 replications used for one-sided comparisons at p = 0 . 01 ) as explained in the corresponding text sections and in Figure S2 . For the control analyses shown in Figure 4 , original task epochs were artificially augmented 5–20 times ( generating ∼104 data points ) and decimated by a factor of 0 . 8-0 . 6 . This process did not significantly alter the original distributions , auto- and cross-correlations for all units . An instantaneous population firing rate vector in MSUA space , obtained by convolution of the spike trains with Gaussian functions as described above , is given by [33] . For univariate time series ν ( t ) , delay embeddings are usually constructed by forming vectors from temporally delayed values with delays ( lags ) τi . These are typically chosen to correspond to p successive minima of the autocorrelation function ( or mutual information ) where p would be high enough to unfold ( pull apart ) trajectories within this delay-coordinate space [38] , [77] . In general , the reconstructed spaces should have dimensionality p = 2×D+1 , where D is the attractor dimension [28] . Similar ideas can be applied to multivariate systems [29] . The attractor dimension is often estimated via the correlation dimension , which , however , will not provide sensible answers in sparse high-dimensional spaces as the ones examined here ( see [29]; Figure S4 and Text S2 ) . Moreover , the use of large delay coordinate maps would result in an extensive loss of data and hence poor statistics . Therefore , additional non-delay variables are sought to effectively disentangle noisy trajectories . The first step in our approach is to construct a reduced multivariate delay-coordinate map which should simply ensure that trajectories do not significantly cross each other . This auxiliary DC-MSUA space , defined by vectors , contains only a single lag for each unit optimized to be the first minimum of the average cross-correlation between pairs of units' firing-rates . The resulting lags ranged from just one time bin to <5% of the task-phase length . Note that the main purpose of these lagged variables is to add axes to the space which contain information about the system dynamics not captured by the current state of the firing rate variables , therefore the choice of lags such as to minimize cross-correlations . In fact , the use of more than one delay per unit did not improve across-trial predictions ( data not shown ) . After this step , differences between trajectories were further amplified by combining these variables into new functional forms , in accordance with ideas from statistical learning theory [27] . As we were specifically interested in functional forms with a biological meaning , this was done by adding higher-order products of the units' firing rates as new coordinates to the neural state space . The oth-order interaction of n-units omitting lags for notational convenience , is defined by ( 1 ) By construction of the smoothed firing rate vector ( see above ) each axis φ ( t ) is the net sum of probabilities of o multiple spikes independently occurring across n neurons ( e . g . [78] ) . For the frequent case that a single spike is contained within a single bin , and for small smoothing windows σ , φ ( t ) tends to represent a pattern of multiple spike-co-occurrences ( a “poly-synchronous” pattern [79] ) . Now , the Oth-order delay-interactions coordinate map consists of all oth-order firing-rate products with o = 1…O . Vectors in this high-dimensional space will be denoted by Φ ( t ) . For instance , a vector in the space corresponding to O = 2 is defined by ( 2 ) The dimensionality p ( O ) of such a space is typically p∼105–109 , much larger than the number of task-epoch vectors which is on the order of ∼103 . Note that this approach sharply contrasts with other methods where the MSUA space dimensionality is instead further reduced by exploiting correlations among units ( e . g . [14] , [15] , [80] ) . As was noted further above , the delay-coordinate map suffices to remove overlap between trajectories in an ideal , purely deterministic system . On the other hand , the multinomial basis expansion defined above helps to achieve an optimal separation in a statistical learning sense when dealing with highly noisy systems ( e . g . Figure 3 ) . Explicit computations in such extremely high-dimensional spaces are associated with numerical and computational problems which can be solved by the so-called “kernel trick” [45] . In this context a kernel is a function which represents a vector product in a high-dimensional space without explicitly computing the dot product of the vectors . Here , for any two high-dimensional vectors Φ ( t ) from the expanded Oth-order space occurring at times ta and tb , respectively , the kernel function is given by ( 3 ) Thus , the function on the right hand side operating on the low-dimensional firing rate vectors ν ( t ) is mathematically equivalent to ( and uniquely defined for ) a dot product between vectors Φ ( t ) from the much higher-dimensional Oth-order space [45] . See Text S1 and [81] for further motivation for the use of this kernel . Within the mathematical framework of kernel algorithms , high-dimensional covariance matrices are replaced by kernel matrices in the reformulation of classical statistical procedures like PCA or FDA . Kernel matrices were computed for each possible pair of task epochs ( such that ta and tb in Equation 4 may correspond to two different time points of the same epoch , or to time points from two different epochs ) , and then used to build a classifier using Fisher's discriminant ( FD ) criterion . FD analysis works by maximizing the difference between task-epoch means while minimizing within-task-epoch ( co- ) variances , i . e . , by finding the direction Ω of the high-dimensional Oth-order space along which the overlap between two task-epoch distributions is minimized [45] , [50] . Since in the expanded spaces the number of dimensions ( variables ) d is extremely high , in fact much higher than the number of observations m , means and covariance matrices cannot be explicitly computed , as stated above , and thus for the FD analysis all computations on high-dimensional vectors are reformulated in terms of a kernel matrix K of much smaller dimensionality ( equal to m2<<d2; see Text S1 ) . By usage of the kernel matrix K , the projections x ( ti ) of high-dimensional vectors Φ ( ti ) onto the optimally discriminating direction Ω are obtained by ( 4 ) where the m elements of the vector α are derived as e . g . explained in Schölkopf and Smola , [45] and in Text S1 . Since the projected values x ( ti ) on the most discriminating axis represent linear combinations of up to 109 random variables ( one variable per dimension ) , the projected data will be approximately normally distributed according to the central limit theorem ( e . g . [44] ) . Hence , building on this assumption of approximate normality , a Bayes-optimal classifier ( the one with theoretically best performance ) can be defined on this most-discriminating axis ( where equal priors were used here for not biasing the results according to the lengths of the sampled task epochs ) . From this , classification ( separation ) errors ( SE; cf . Figure 5 ) , likelihoods p ( ν ( t ) |C ) of classification into task-epoch C , posterior probabilities P ( C|ν ( t ) ) ( using Bayes criterion ) , and 99% confidence intervals are straightforward to obtain . The ( discretized ) Kullbach-Leibler divergence [e . g . 44] was computed as a measure of the distance between these Gaussian posterior distributions corresponding to any two tasks epochs C1 and C2 . It was estimated for each Oth order expansion ( Figure 5C ) and is given by ( 5 ) The utilization of normal probability theory represents a fundamental advantage over other approaches specialized for high-dimensional spaces ( e . g . support-vector-based classifiers [27] ) which may have similar classification performance [45] but do not easily permit other aspects of the present statistical analysis . For the across-trial analyses , optimal directions Ω for each task epoch pair were obtained using exclusively the first set of ( reference ) trials , Φref . This direction was then fixed for computing the projections xpredic ( ti ) of vectors Φpredic ( ti ) from the prediction set onto Ω to yield the predicted SE ( SEpredic ) : ( 6 ) where the vector αref is the one obtained from the reference set and K represents projections of prediction set vectors into the reference space . A brief summary of these algorithms can be found in Text S1 [45] . A regularization penalty was furthermore added to the kernel matrices to ensure a low generalization error ( loosely speaking , a regularization factor automatically constrains the number of free parameters to reduce out-of-sample prediction errors; e . g . [27] , [45] ) . This regularization was optimized such that SEpredic was minimal for animal #1 and then it was fixed for all other analyses ( because of this regularization , for instance , in-sample SE never decreases to zero for the expanded spaces in Figures 3A and 5A ) . Prediction errors were found to be invariant for large enough values of this regularization penalty ( as demonstrated in Figure S3 ) . The robustness of the present approach with regards to different basis functions used in the expansion ( and thus different definitions of the kernel ) is also discussed in Figure S3 . Finally , we also investigated how unsupervised clustering approaches perform on the DC-MSUA spaces , and noticed that they reliably pick up only the delay vs . training/test phase differences in this lower-dimensional representation ( see Figure S5 for an example ) . Kernel-FDA [50] and kernel-PCA [47] were used to obtain three-dimensional visualizations for each high-dimensional task-epoch state . Three-dimensional projections were also used for determining velocity vectors ( cf . Figure 6 ) , as these cannot be efficiently computed in high dimensions ( a problem running under the label “curse of dimensionality”; e . g . [44] ) . Kernel-PCA proceeds in much the same way as ordinary PCA , except that – like kernel-FDA – it works on the kernel matrices defined above instead of directly on the high-dimensional covariance matrices ( see brief summary in Text S1 ) . Thus , the three orthogonal dimensions capturing the largest amount of data variance in the high-dimensional spaces were obtained . Additional discussion about the adequacy of these three-dimensional velocity vectors as obtained by kernel-PCA can be found in Figure S4 and in Text S2 . Finally , note that , apart from the convergence analysis shown in Figure 6 , these three-dimensional reductions served only for the purpose of visualization , while all statistical analyses were performed on the full high-dimensional spaces ( see Figure 7C ) . Analysis software was implemented in MatLab ( Mathworks Inc . , MA , USA ) and is freely available in http://www . bccn-heidelberg-mannheim . de under the terms of the general public license ( http://www . gnu . org/licenses/ ) . | For understanding how neural processes give rise to cognitive operations , it is essential to understand how aspects of the underlying neural network dynamics reconstructed from neurophysiological measurements relate to behavior . For instance , different actions may be represented by neural states characterized by stable population patterns to which activity converges in time , called attractors in the language of dynamical systems . However , experimental demonstrations of neural attractors associated with cognitive entities have been rare so far . One problem may have been that in behaving animals , in-vivo one can access only a relatively small fraction of the total number of neural units comprising the whole system , even with modern multiple single-unit ( MSU ) recording techniques . Therefore , the neural activity dynamics are necessarily projected from a very high-dimensional into the empirically accessible much lower-dimensional space in which attractor properties may be lost due to ambiguities and entanglement in the flow of trajectories . In the present study , principles from nonlinear time series analysis and statistical learning are applied to MSU recordings from the rat's prefrontal cortex during decision-making tasks . By expanding the empirically accessed neural state space ( semi- ) attracting properties of neural states corresponding to cognitively defined task-epochs became apparent , in line with many neuro-computational theories . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"computational",
"neuroscience",
"biology",
"computational",
"biology",
"neuroscience"
] | 2011 | Attracting Dynamics of Frontal Cortex Ensembles during Memory-Guided Decision-Making |
The models in statistical physics such as an Ising model offer a convenient way to characterize stationary activity of neural populations . Such stationary activity of neurons may be expected for recordings from in vitro slices or anesthetized animals . However , modeling activity of cortical circuitries of awake animals has been more challenging because both spike-rates and interactions can change according to sensory stimulation , behavior , or an internal state of the brain . Previous approaches modeling the dynamics of neural interactions suffer from computational cost; therefore , its application was limited to only a dozen neurons . Here by introducing multiple analytic approximation methods to a state-space model of neural population activity , we make it possible to estimate dynamic pairwise interactions of up to 60 neurons . More specifically , we applied the pseudolikelihood approximation to the state-space model , and combined it with the Bethe or TAP mean-field approximation to make the sequential Bayesian estimation of the model parameters possible . The large-scale analysis allows us to investigate dynamics of macroscopic properties of neural circuitries underlying stimulus processing and behavior . We show that the model accurately estimates dynamics of network properties such as sparseness , entropy , and heat capacity by simulated data , and demonstrate utilities of these measures by analyzing activity of monkey V4 neurons as well as a simulated balanced network of spiking neurons .
Activity patterns of neuronal populations are constrained by biological mechanisms such as biophysical properties of each neuron ( e . g . , synaptic integration and spike generation [1 , 2] ) and their anatomical connections [3] . The characteristic correlations among neurons imposed by the biological mechanisms interplay with statistics of sensory inputs , and influence how the sensory information is represented in the population activity [4–6] . Thus accurate assessment of the neural correlations in ongoing and evoked activities is a key to understand the underlying biological mechanisms and their coding principles . The number of possible activity patterns increases combinatorially with the number of neurons analyzed . The maximum entropy ( ME ) principle and derived ME models—known as the pairwise ME model or the Ising model—have been used to explain neural population activities using fewer activity features such as event rates or correlations between pairs of neurons [7 , 8] . This approach has been employed to explain not only the activity of neuronal networks but also other types of biological networks [9–11] . For large networks , however , exact inference of these models becomes computationally infeasible . Thus researchers have employed approximation methods [12–18] . While they successfully extended the number of neurons that could be analyzed , it was pointed out that the pairwise ME model might fail to explain large neural populations because the effect of higher-order interactions may become prominent [19–21] . Another fundamental problem of the conventional ME models is that these models assume temporarily constant spike rates for individual neurons . The assumption of stationary spike-rates is invalid , e . g . , when in vivo activity is recorded while an animal performs a behavioral task . Ignoring such dynamics might result in erroneous model estimates and misleading interpretations on their correlations [22–26] . Moreover neural correlations themselves likely organize dynamically during behavior and cognition , which can be independent from changes in the spike rates of individual neurons [27–29] . The time-dependence of neural activity may be explained by including stimulus signals in the model , e . g . , for analyses of early sensory cells [30] . However , the approach may become impractical when analyzing neurons in higher brain areas in which receptive fields of neurons are not easily characterized . Thus it remains to be examined how much the pairwise ME model can explain the data if the inappropriate stationary assumption is removed . The state-space analysis [31] offers a general framework to model time-series data as observations driven by an unobserved latent state process . The underlying state changes are uncovered by a sequential estimation method from the noisy measurements . While observations of neuronal activity are often characterized by point events ( spikes ) , a series of studies have established the nonlinear recursive Bayesian estimation of the underlying state that drives the event activity [32–34] . The method successfully estimated an animal’s position from population activity of hippocampal place cells [32] , or estimate arm trajectories from neurons in the monkey motor cortex [35 , 36] . Recently , this framework has been extended to the analysis of population activity [37–39] . In addition to the point estimates of interaction parameters suggested by earlier studies [40–42] , the state-space analysis provides credible intervals of those estimates through the recursive Bayesian fitting algorithm . Nevertheless , as previously mentioned , the state-space model of a neural population was restricted by its computational cost . Therefore , it could be utilized to analyze only small populations ( N ≤ 15 ) . Recent advances in electrophysiological and optical recording techniques from a large number of neurons in vivo under free moving or virtual reality settings challenge these analysis methods . Thus the challenge is to make it possible to fit the exponentially complex state-space model to such large-scale data . For this goal , we need to incorporate approximation methods into the sequential Bayesian algorithm . More specifically , we need good approximations of mean and variance of the model parameters required in the approximate Bayesian scheme . These approximation methods must be analytical to avoid impractical computation time . By doing so we will be able to directly estimate all time-varying interactions of a large neural population . Such a model will serve as benchmark for alternative unsupervised methods that aim to capture low-dimensional , time-dependent latent structure of the pairwise interactions [43–45] ( see also [46–48] for other dimension reduction methods for neuroscience data ) . Here by combining the state-space model proposed in [37–39] with analytic approximation methods , we provide a framework for estimating interactions of neuronal populations consisting of up to 60 neurons . To find the mean we used the pseudolikelihood approximation method . To approximate the variance , we provide two alternative methods: the Bethe or the mean-field approximation . The Bayesian analysis methods for larger networks of neurons allow us to better understand macroscopic states of a neural population , such as entropy , free energy and sensitivity , all in a time-resolved manner and with credible intervals . Thus the model provides a new way to investigate effects of stimuli and behavior on activity of neuronal populations . It is expected to provide observations that give us insights into the underlying circuitry and its computation .
In the following subsections , we demonstrate the fit of the state-space model of neural population activity to artificially generated data of 40 neurons with dynamic couplings for T = 500 time bins . To be able to compare it to the ground truth we construct 4 populations each consisting of 10 neurons . Individual parameters θ1:T of the underlying submodels are generated as smooth independent Gaussian processes , where the mean for the first order parameters θ i t increases at t = 100 and then decreases more slowly shortly after that . The interaction parameters θ i j t are generated as Gaussian processes whose mean is fixed at 0 . In total , 500 trials of spike data are sampled from this generative model . Note that the sampled individual parameters differ and vary over time although we use homogeneous means . The increase of the mean for θ i t increases spiking probability followed by a decrease back to baseline ( Fig 1A ) . In the resulting data neurons spike with time averaged probabilities ranging from 0 . 10 up to 0 . 21 . Supposing bin width Δ = 10 ms these are in a physiologically reasonable range . This exemplary scenario may mimic a population that independently receives an external input elicited by e . g . , a sensory stimulus . For details of the generation of the data see S3 Text . Next we fit the state-space model of neural population activity to the generated data with the combination of pseudolikelihood and Bethe approximation . This combination is chosen for the demonstration because it provides the best estimates of the underlying model as we will assess later in this section . Top panel of Fig 1B shows snapshots of the smoothed estimates of the inferred network at different time points ( t = 50 , 150 , 300 ) . The color of the nodes indicate the smoothed estimates of the first order parameters θ i t | T and the one of the edges interactions θ i j t | T . Visual inspection of the fitted network suffices to identify that there are 4 independent subpopulations of correlated neurons ( one in each quadrant ) . To check whether the inferred changes over time match those of the underlying generative model , credible intervals of three fitted couplings are compared with their underlying values ( Fig 1B Bottom ) . The fit follows the dynamics , and correctly identifies the parameter that is constantly 0 ( the lowest panel ) . One of the main motives to model joint activities of a large population of neurons is to assess macroscopic properties of the network in a time-dependent manner with credible intervals . The macroscopic measures obtained for this example are shown in Fig 1C , and in the following we introduce them one by one . The first and simplest macroscopic property shown in the top left panel of Fig 1C is the probability of spiking in a network ( population spike rate ) . We define it as p spike ( t ) = 1 N ∑ i = 1 N η i t , ( 29 ) where η i t is the spike rate of ith neuron at time t . Considering the smoothed estimate η i t = η i t | T , the method recovers correctly the empirical rate obtained from the data ( Fig 1A Bottom ) . The shaded area in the panel indicates the 98% credible interval of the population spike rate obtained by resampling the natural parameters from the smoothed posterior density 100 times at each bin . The underlying spike probability for N = 40 neurons is obtained by calculating the marginals η i t independently for each subpopulation and averaging over all neurons . Next from the state-space model of neural population activity one can estimate the probability of simultaneous silence ( i . e . , the probability that no neuron elicits a spike , Fig 1C bottom left ) p silence ( t ) = exp ( - ψ t ) . ( 30 ) The approximation methods allow us to evaluate the log partition function ψt ( Eqs 24 and 25 ) . Here we use smoothed estimates to compute the log partition function . Thus we immediately obtain the probability of simultaneous silence . The expected simultaneous silence for N = 40 neurons is obtained as multiplication of the silence probabilities of the 4 subpopulations . The entropy of the network ( i . e . , expectation of the information content , 〈−log p ( x|θt ) 〉θt ) can be also calculated from the model as S ( t ) = - θ t ′ η t + ψ t . ( 31 ) Estimation of this information theoretic measure allows us to quantify the amount of interactions in the network by comparing the pairwise model to the independent one ( see following analyses and Eq 36 ) . Since it is an extensive quantity , the entropy of N = 40 neurons is obtained by addition of the entropies from the 4 independent subpopulations . The entropy increases while the individual activity rates of neurons also increases ( Fig 1C top right ) . The last measure shown in the bottom right panel of Fig 1C is the heat capacity , or sensitivity , of the system . It is the variance of information content: C ( t ) = 〈{−log p ( x|θt ) }2〉θt − {〈−log p ( x|θt ) 〉θt}2 , where the brackets indicate expectation by p ( x|θt ) . It is also the variance of the Hamiltonian - θ t ′ F ( x ) . Thus we can obtain it by introducing a nominal dual parameter β to the Hamiltonian in the model , assuming that it is 1 for real data . The log partition function of the augmented model is ψ t ( β ) = log ∑ x exp ( β θ t ′ F ( x ) ) . ( 32 ) The variance of Hamiltonian is given as the Fisher information w . r . t . β , i . e . , the second derivative of the log partition function . This allows us to use the approximate ψt to assess the heat capacity . Then we further approximate the second derivative by its discrete version C ( t ) = ∂ 2 ψ t ∂ β 2 β = 1 ≈ ψ t ( 1 + ϵ ) - 2 ψ t ( 1 ) + ψ t ( 1 - ϵ ) ϵ 2 , ( 33 ) and ϵ is chosen to be 10−3 . The heat capacity measures sensitivity of the network , namely how much the network activity changes due to subtle changes in its network configuration ( i . e . , to changes of the θt parameters ) . Networks with higher sensitivity are more responsive to changes than those with lower sensitivity . Similarly to the entropy , the heat capacity is an extensive quantity . For the simulated data , the heat capacity decreases while activity rates of neurons are increased ( Fig 1C bottom right ) . Next we examine the goodness-of-fit of the model fitted by the pseudolikelihood and Bethe approximation methods . In particular , we ask how the fitting performance changes with increasing network size . For this reason we generated 6 dynamic models for populations of 10 neurons as described previously ( 500 time bins , 500 trials ) . Then we construct smaller or larger populations by concatenating the independent groups . The model is fitted by the pseudolikelihood and Bethe approximation methods to the first subnetwork , then two subnetworks , and so on , until we fit the model to a network containing 60 neurons composed of 6 independent groups . We obtain estimates of the macroscopic measures from the smoothed estimates of the model parameters at each time bin . Fig 2A shows values of these measures averaged over time . The results show extensive properties of macroscopic measures ( except for the population spike rate ) , and that the estimates may slightly deviate for larger number of neurons . To assess quality of the fit , first the root mean squared error ( RMSE ) for the natural parameters averaged across time bins is calculated RMSE ( θ t | T ) = 1 T ∑ t = 1 T θ t | T - θ t 2 , ( 34 ) where θt|T is the smoothed estimate of the underlying model θt . ‖v‖ denotes the L2-norm of vector v . For the data sets with 500 trials , the RMSE increases linearly with network size ( Fig 2B Left ) . Furthermore , the error for the macroscopic measures is assessed by Error [ f ( θ t | T ) ] = RMSE ( f ( θ t | T ) ) 1 T ∑ t = 1 T f ( θ t ) , ( 35 ) where f ( θt|T ) is any function of the macroscopic measures . The RMSE is defined similarly to Eq 34 while substituting the parameters θt|T by the function f ( θt|T ) . Besides the population rate these errors also increase as the network size increases ( Fig 2B ) . We observe non-monotonic behavior in some of the macroscopic properties ( e . g . , average spike rate and the entropy’s error ) , which can be explained by fluctuations from the data generation process . To understand whether these errors increase primarily due to the approximation methods used for the fit or because of the finite amount of data , the fit is repeated but now to spiking data with 1000 trials . The error of the fit is reduced particularly for larger network size ( Fig 2B dashed lines ) , suggesting that the limited amount of data is mainly responsible for the estimation error . In general , the estimation error is largest at time points where the parameters θt change rapidly . This is a general problem of smoothing algorithms , including spike rate estimation , which depend on fixed smoothness parameter ( s ) ( i . e . , here λ ) optimized for an entire observation period ( see e . g . , [63] for optimizing a variable smoothness parameter to cope with such abrupt changes ) . To this end , only the Bethe approximation was used in combination with the pseudolikelihood to fit the model approximately . However , as discussed previously , the TAP approximation constitutes a potential alternative . To assess the quality of both approximations , we investigated a small network ( 15 neurons , 500 time bins , 1000 trials ) . The data was generated as described for Fig 1 . The smaller network is considered because it allows to fit the model by an exact method without the Bethe or TAP approximations . Here the exact method refers to the method in which the expectation parameters are calculated exactly at the gradient search for the MAP estimates of model parameters ( Eq 13 ) . It should be noted that we approximate the posterior density by the Gaussian distribution even for the “exact method” in the recursive Bayesian algorithm . Comparison of the approximation methods with the exact method determines the error that is caused by the approximation methods and not by the finite amount of data . First , investigation of three exemplary time points ( Fig 3A ) reveals that both the pseudolikelihood-Bethe and the pseudolikelihood-TAP approximation recover the underlying parameters . We examine the error across time bins by the RMSE . Comparing RMSE of the approximation results with the exact fit ( Fig 3B ) demonstrates that the both approximations perform worse in the same range . To examine the approximations also for large networks ( N = 60 ) we sampled 1000 trials ( as for Fig 2 ) . In Fig 3C we observe that errors of the approximations are comparable . Furthermore , we compare running times required for fitting the network of the two methods ( Fig 3D ) . The pseudolikelihood-TAP approximation turns out to be faster than Bethe . We observed that the EM algorithm required more iterations for the Bethe approximation . Furthermore , the occasional use of the CCCP contributed to the long fitting time of the pseudolikelihood-Bethe procedure . Since both , Bethe and TAP , provide an approximation for the log partition function ψt ( Eqs 25 and 24 ) , we assess their performance for the same data as in Fig 3 . The time evolution of simultaneous silence ( directly linked to ψ by Eq 30 ) is recovered by exact , Bethe , and TAP ( Fig 4A ) . The results show that the TAP approximation slightly overestimated the probability in this example . This is also reflected in the Error [ ψ ( { θ ^ t | T } t ) ] ( Fig 4B ) , where the Bethe approximation performs better than the TAP method . However , the error for the Bethe approximation increases compared to the exact method . The relation between the two approximation methods persists also for large networks ( Fig 4C ) . Another disadvantage of the TAP approximation is that the system of non-linear equations occasionally could not be solved . This happens more frequently when fitting larger networks and/or networks with stronger interactions . Therefore , it seems that the pseudolikelihood-Bethe approximation exhibits more accurate estimates; hence we will use it again for the following analysis . However the faster fitting of pesudolikelihood-TAP can be advantageous elsewhere . We now apply the approximate inference method to analyze activity of monkey V4 neurons recorded while the animal performed repeatedly ( 1004 trials ) the following behavioral task . Each trial began when the monkey fixated its gaze within 1 degree of a centrally-positioned dot on a computer screen . After 150 ms , a drifting sinusoidal grating was presented for 2 s in the receptive field area of the neuronal population that was recorded , at which time the grating stimulus disappeared and the fixation point moved to a new , randomly chosen location on the screen , and the animal made an eye movement to fixate on the new location . Data epochs from 500 ms prior to grating stimulus onset until 500 ms after stimulus offset were extracted from the continuous recording for analysis . The spiking data obtained by micro-electrode recordings includes 112 single and multi units identified by their distinct wave forms . The experiment was performed at the University of Pittsburgh . All experimental procedures were approved by the University of Pittsburgh Institutional Animal Care and Use Committee , and were performed in accordance with the United States’ National Institutes of Health ( NIH ) Guide for the Care and Use of Laboratory Animals . For details on experimental setup , recording and unit identification see [64] . The recorded units are tested for across-trial stationarity ( which is the assumption of the model ) : The mean firing rates for each trial are standardized and if more than 5% of the trials were outside the 95% confidence interval the unit is excluded . After this preprocessing 45 units remained . To obtain the binary data , the spike trains are discritized into time bins with Δ = 10 ms resulting into 300 time bins over the course of the trial . Exemplary data are displayed in Fig 5A Top . We note that the following conclusions of this analysis do not change even if we use smaller and larger bin size ( Δ = 5 and 20 ms ) . After the data are preprocessed , we analyze the network dynamics of the 45 units during the task period by the state-space model for the neural population activity . Inference is done by using the pseudolikelihood-Bethe approximation . The results of fitting the state-space model are displayed in Fig 5B . Before presenting detailed results , we note that considering dynamics in activity rates and neural correlations better explains the population activity while avoiding overfitting , compared to assuming that they are stationary . To assess this , we compared the predictive ability of the state-space model with that of the stationary model , using the Aikake ( Bayesian ) Information Criterion ( AIC ) [65] defined as −2l ( X1:T|w ) + 2k , where k is the number of free parameters in w . To obtain the latter , we fitted the state-space model once more but now fixing λ−1 = 0 , which results in a stationary model since the state model in Eq 3 no longer contains variability . The result confirms that the dynamic model better predicts the data ( AICdyn = 4467026 for the dynamic model and AICstat = 4576544 for the stationary model ) . We observe stimulus locked oscillations in the population firing rate that are also captured by the model ( Fig 5A Bottom ) . The average of the estimated natural parameters ( Fig 5B Bottom ) show that these oscillations are explained by the first order parameters θ i t | T . We note that these oscillations are mainly caused by two units with high firing rates and they should not be considered as a homogeneous property of the network . Investigation of the network states before , during , and after the stimulus ( Fig 5B Top ) reveals that the interactions θ i j t | T are altered over time . This is also reflected in an average over the all pairwise interactions ( Fig 5B Center ) , where the mean decreases during the stimulus presentation as well as the standard deviation . Thus neurons are likely to decorrelate during the stimulus presentation whereas the population rate increases and oscillates at the same time . Similarly to the analysis of artificial data ( Fig 1 ) , we measure the macroscopic properties of the fitted model over the task period ( see Fig 5C for credible intervals ) . To test the contribution of interactions in the recorded data , the model is once again fitted to trial shuffled data [23] , which should destroy all correlations among units that do not occur due to chance . Comparison of the macroscopic measures between the models fitted to the original data and to the trial shuffled data shows how interactions among units alter the results . In the following , we will refer to the two models as “actual” and “trial shuffled” model . The probability of simultaneous silence shows again the stimulus locked oscillations , and decreases during the stimulus period . The difference between the actual and trial shuffled model before the stimulus is larger than during and after the stimulus , suggesting that the observed positive interactions contributed to increasing the silence probability in particular before and after the stimulus period . The entropy reflects the oscillations and shows a strong increase ( ∼1/3 ) during the stimulus period . This is reasonable because we observe an increase in activity rates and a decrease in correlations—both effects should result in an increase in entropy . Next , we examine how much of the entropy is explained by the interactions among the neurons . To do so , at each time point we calculate the corresponding independent model by projecting the fitted interaction model to the independent model ( i . e . , the model with the same individual firing rates η i t but with all θ i j t = 0 ) . The entropy of the independent model Sind should always be larger than Spair , the entropy of the model with interactions . Hence , a fraction of entropy explained by the interactions can be calculated as S ind - S pair S ind . ( 36 ) In general , contribution of interactions to the entropy is small for these data ( ≤ 2% ) . However , the contribution is less during stimulus presentation , compared to the period before the stimulus . Only in the beginning of the stimulus presentation , two peaks of correlated activity can be observed . The observed reduction of the fractional entropy for interactions could be caused by the increase of the first order parameters θ i t and/or by the decrease of the interactions θ i j t during the stimulus period . The decorrelation observed during the stimulus period is successfully dissociated from the oscillatory activity: Previously observed oscillations are absent in this measure of interactions . This result is important because ignoring such firing rate dynamics often leads to erroneous detection of positive correlations among neurons . A clear exception is the first peak appeared during the stimulus presentation , which was also observed in the trial-shuffled model . Indeed , the first sharp increase of the spike rates was not faithfully captured by the models , which caused spurious interactions in the trial-shuffled model . Last , the sensitivity ( heat capacity ) of the network over time is obtained . While for the artificial data in Fig 1 the sensitivity showed a drastic decrease , such reduction is not observed in the V4 data . The sensitivity of the network is maintained at approximately the same value before and during the stimulus period . This is interesting since we already observed that before and during the stimulus the network seems to be in two qualitatively different states ( low vs . high firing rate and strong vs . weak interactions ) . After stimulus presentation the sensitivity drops . Overall , neural interactions contribute to have higher sensitivity ( see light vs . dark credible intervals ) . Networks with balanced excitation and inhibition have been used to describe cortical activity [66 , 67] . To see whether the balanced network model can reproduce the findings from the recorded V4 , we simulate spiking data using the balanced spiking network following [24] , and analyze these data with the state-space model . The network consists of 1000 leaky integrate-and-fire neurons ( 800 excitatory , 200 inhibitory ) ( For details see S4 Text ) . Connection probability is 20% , between all neurons . The network receives input from 800 Poisson neurons . Each input neuron has a Gaussian tuning curve , where the preferred direction is randomly assigned . We choose an experimental paradigm which resembles one of the V4 data . 1000 trials of 3 s duration are simulated . Before each trial , the simulation runs for 500 ms under random Poisson inputs such that the network state at the beginning of each trial is independent . Then the trial starts at −500 ms . At 0 ms a 90° is shown for 2 s followed again by a 500 ms period of stimulus absence . The activity of 140 neurons are recorded for investigation . From the recorded subpopulation , we further selected 40 excitatory and 20 inhibitory neurons with the highest firing rates for the following analysis . Binary spike trains were obtained by binning with Δ = 10 ms . Exemplary data are shown in Fig 6A ( top spike trains are from excitatory , and bottom spike trains from inhibitory neurons ) . We then fitted the state-space model to these data . As for the V4 data , we show in Fig 6B 3 snapshots of the network ( N = 60 ) ( Top ) , as well as mean and standard deviation of θ i t | T and θ i j t | T ( Bottom ) . In contrast to the V4 network there are numerous significant non-zero couplings . However , similarly to the monkey data , we observe an increase for θ i t and a decrease of θ i j t during the stimulus period . We also assess the macroscopic states for the balanced network ( Fig 6C ) . As in the V4 data the probability of silence decreases during the stimulus period . Furthermore , compared to the trial shuffled result , the difference is larger before and after the stimulus than during the stimulus , suggesting a larger contribution of the couplings to silence when no stimulus is present . The entropy increases during the stimulus period . The credible interval for the trial shuffled data is narrower than for actual model and the entropy tends to be larger . Up to this point we did not find , in the macroscopic properties , significant qualitative differences between the V4 data and the simulated data from the balanced network . However , the entropy that is explained by the couplings increases during the stimulus , while in the V4 data a decrease is observed ( Fig 6C , third panel ) . Hence , the interactions in the balanced network become stronger during the stimulus , even though the mean of the couplings θ i j t | T decreases for this period . This can be explained by more negative values in estimated couplings during the stimulus period . The sensitivity slightly decreases when the stimulus is shown and , as for the V4 data , couplings contribute to higher sensitivity . Observing the dynamics in the model parameters poses the question how the actual synaptic connectivity structure of the network is reflected in the inferred interactions . Do positive values correspond to excitatory synapses , and negative to inhibitory ones ? While for the V4 data this is impossible to assess , we compare the values of θ i j t | T of pairs , that are at least connected by one excitatory synapse and those that are connected by at least one inhibitory synapse ( Fig 7A , red and blue histograms respectively ) . In general , excitatory connected pairs show more positive values , while inhibiting ones tend to be negative . The most negative values are almost exclusively explained by inhibiting pairs . However , compared to all θ i j t | T ( gray histogram ) many positive couplings θ i j t | T do not represent excitatory connected pairs . Thus it is difficult to identify excitatory synapses from the inferred couplings . The result that inhibitory pairs showed stronger negative couplings , while excitatory pairs were mostly represented by weak positive couplings , can be explained by on average much stronger conductance of inhibitory synapses . Finally we compare the mean values of couplings between different network sizes ( Fig 7B ) . To do so networks of size N = 15 , 30 , 60 are fitted , where the network always consisted of one third inhibitory and two thirds excitatory neurons . However , neither for excitatory , inhibitory or all couplings we could identify dependency on the network sizes that can be analyzed by our model .
This study provides approximate inference methods for simultaneously estimating neural interactions of a large number of neurons , and quantifying macroscopic properties of the network in a time-resolved manner . We assessed performance of these methods by using simulated parallel spike sequences , and demonstrated the utility of the proposed approach by revealing dynamic decorrelation of V4 neurons and maintained susceptibility during stimulus presentations . Furthermore we compared those findings with data from a simple balanced network of LIF neurons , which suggested that further refinements were necessary to reproduce the observed network activity . Accurate assessment of correlated population activity in ongoing and evoked activity is a key to understand the underlying biological mechanisms and their coding principles . It is critical to model time-dependent firing rates to correctly assess neural interactions . If we apply a stationary model of neural interactions to independent neurons with varying firing rates , we may erroneously observe excess of correlations [22–24 , 26 , 68] . Such an apparent issue of a stationary model can introduce considerable confusion in search of fundamental coding principles of neurons . Several related studies accounted for the nonstationary activity by modeling time-dependent external fields ( c . f . , { θ i t } in Eq 1 ) while fixing pairwise interactions [26 , 30] . In addition to the external fields , however , we consider that modeling dynamics of correlations are important particularly for analyses of neurons recorded from awake animals because neural correlations are known to appear dynamically in relation to behavioral demand to the animals [27–29 , 38 , 69] . Indeed , we found dynamic decorrelation of V4 neurons during stimulus presentation ( Fig 5C 3rd panel ) , which may reflect asynchronous neural activities under stimulus processing of an alert animal [70 , 71] . In general , it is important to compare the result with that of surrogate data in which one destroys correlations to examine potentially short-lasting time-varying interactions in relation to behavioral paradigms . The current state-space model presumes that the neural dynamic follows a quasistatic process . At each time t , we assumed that population activity is sampled from the equilibrium joint distribution given by Eq 1 across trials while the state of population activity smoothly changes within a trial . This is of course a simplified view of neuronal dynamics . Most notably , dependency of the neurons’ activity on their past activity makes the system a nonequilibrium one . Such activity is captured by models via the history effect , e . g . , using the kinetic Ising model [25 , 26 , 72 , 73] or generalized linear models ( GLM ) of point and Bernoulli processes [35 , 74–76] . Given the past activities , these models construct the joint activity assuming their conditional independence . The equilibrium and non-equilibrium models thus assume different generative processes , even though the pseudo-likelihood approximation for our equilibrium Ising model used similar conditional independence given the activity of other neurons at the same time . It is an important topic to include both modeling frameworks in the sequential Bayes estimation to better account for dynamic and nonequilibrium properties of neural activity [39] . The model goodness-of-fit may be additionally improved by including sparseness constraints on the couplings as was done in the stationary models [40 , 77 , 78] . In this study , we employed the classical pseudolikelihood method to perform MAP estimation of interactions ( i . e . , natural parameters ) without computing the partition function . For the inverse problem without the prior , we may use alternative approximation methods such as Bethe and TAP approximations , and further state-of-the-art methods such as the Sessak-Monasson [12] , minimum-probability-flow [15] , and adaptive-cluster expansion [17] method . However , here we chose the pseudolikelihood method because it was not trivial to apply the other methods to the Bayesian estimation . Alternatively , the Bethe and TAP approximation methods may be used to approximate the expectation parameters during the iterative procedure of the exact MAP estimation ( Eq 13 ) because these methods allow us to estimate the expectation parameter from the natural parameters ( the forward problem ) . However , as we found in the estimation of the Fisher information , TAP may occasionally fail and Bethe approximation by BP may not converge . Thus we rather used these methods after the MAP estimation was found by the pseudolikelihood method . The framework , however , is not limited to these approximation methods , and new methods may be incorporated into the state-space model to further increase the number of neurons that can be analyzed . It should be noted that the current model does not include higher-order interactions to explain the population dynamics . While neural higher-order interactions are ubiquitously observed in vivo [38 , 79–81] as well as in vitro [20 , 21 , 82 , 83] conditions , it remains to be elucidated how they contribute to characterizing evoked activities . It is an important step to include higher-order interactions in the large-scale time-dependent model . However , the proposed method that includes up to pairwise interactions can be used as a null model for testing activity features involving higher-order interactions . For example , both experimental and modeling studies showed that simultaneous silence of neurons constitutes a major feature of higher-order interactions of stationary neural activities [83 , 84] . It remains to be tested , though , if silence probability of all neurons recorded from behaving animals exceed prediction by the pairwise model . Such sparse population activity may be expected when animals process natural scenes , compared to artificial stimuli [85] . The limiting factor for the current model on the network size is rather the lack of data than the performance of the approximation methods ( Fig 2 ) . Hence , the state-space or other time-resolved methods that include dimension reduction techniques will be important approaches to explain activity of much larger populations than analyzed here . While there is still room for improvement , the currently proposed method already allows researchers to start testing hypotheses of network responses under distinct task conditions or brain states . These observations will serve to construct biophysical models of neural networks by constraining them , therefore revealing their coding principles . | Simultaneous analysis of large-scale neural populations is necessary to understand coding principles of neurons because they concertedly process information . Methods of thermodynamics and statistical mechanics are useful to understand collective phenomena of the interacting elements , and they have been successfully used to understand diverse activity of neurons . However , most analysis methods assume stationary data , in which activity rates of neurons and their correlations are constant over time . This assumption is easily violated in the data recorded from awake animals . Neural correlations likely organize dynamically during behavior and cognition , and this may be independent from the modulated activity rates of individual neurons . Recently several methods were proposed to simultaneously estimate dynamics of neural interactions . However , these methods are applicable to up to about 10 neurons . Here by combining multiple analytic approximation methods , we made it possible to estimate time-varying interactions of much larger neural populations . The method allows us to trace dynamic macroscopic properties of neural circuitries such as sparseness , entropy , and sensitivity . Using these statistics , researchers can now quantify to what extent neurons are correlated or de-correlated , and test if neural systems are susceptible within a specific behavioral period . | [
"Abstract",
"Introduction",
"Results",
"Discussion"
] | [
"action",
"potentials",
"medicine",
"and",
"health",
"sciences",
"neural",
"networks",
"applied",
"mathematics",
"population",
"dynamics",
"membrane",
"potential",
"electrophysiology",
"random",
"variables",
"neuroscience",
"covariance",
"simulation",
"and",
"modeling",
"... | 2017 | Approximate Inference for Time-Varying Interactions and Macroscopic Dynamics of Neural Populations |
Diagnosis of leptospirosis by the gold standard serologic assay , the microscopic agglutination test ( MAT ) , requires paired sera and is not widely available . We developed a rapid assay using immunodominant Leptospira immunoglobulin-like ( Lig ) proteins in a Dual Path Platform ( DPP ) . This study aimed to evaluate the assay's diagnostic performance in the setting of urban transmission . We determined test sensitivity using 446 acute and convalescent sera from MAT-confirmed case-patients with severe or mild leptospirosis in Brazil . We assessed test specificity using 677 sera from the following groups: healthy residents of a Brazilian slum with endemic transmission , febrile outpatients from the same slum , healthy blood donors , and patients with dengue , hepatitis A , and syphilis . Three operators independently interpreted visual results without knowing specimen status . The overall sensitivity for paired sera was 100% and 73% for severe and mild disease , respectively . In the acute phase , the assay achieved a sensitivity of 85% and 64% for severe and mild leptospirosis , respectively . Within seven days of illness onset , the assay achieved a sensitivity of 77% for severe disease and 60% for mild leptospirosis . Sensitivity of the DPP assay was similar to that for IgM-ELISA and increased with both duration of symptoms ( chi-square regression P = 0 . 002 ) and agglutinating titer ( Spearman ρ = 0 . 24 , P<0 . 001 ) . Specificity was ≥93% for dengue , hepatitis A , syphilis , febrile outpatients , and blood donors , while it was 86% for healthy slum residents . Inter-operator agreement ranged from very good to excellent ( kappa: 0 . 82–0 . 94 ) and test-to-test reproducibility was also high ( kappa: 0 . 89 ) . The DPP assay performed acceptably well for diagnosis of severe acute clinical leptospirosis and can be easily implemented in hospitals and health posts where leptospirosis is a major public health problem . However , test accuracy may need improvement for mild disease and early stage leptospirosis , particularly in regions with high transmission .
Leptospirosis , caused by >200 pathogenic serovars of Leptospira interrogans , is an increasingly important cause of morbidity worldwide with >500 , 000 cases annually [1] , [2] . Most urban infections in Brazil and other emerging economy countries occur in densely populated , resource-poor slums that lack adequate sanitation , cultivate rodent reservoirs , and foster the environmental persistence of Leptospira 3–[6] . Although few ( 5–10% ) infections progress to severe disease , typified by jaundice , acute renal failure , and hemorrhage ( Weil's disease ) and/or respiratory compromise , the case fatality may exceed 15% when severe disease develops [7] , [8] . Early antimicrobial therapy reduces illness duration and severity [9] , [10] . However , leptospirosis is often clinically confused with other acute febrile illnesses [11] , [12] . Accurate early detection therefore remains urgently needed to avert the significant consequences of leptospirosis . Culturing Leptospira is difficult and growth success is diminished in patients already initiated on antimicrobial therapy . The gold standard diagnostic assay for leptospirosis , the microscopic agglutination test ( MAT ) , requires skilled technicians , maintenance of live cultures , and paired sera for confirmation . Application of these standard confirmatory techniques is limited and prolonged [13] , [14] , thus hindering patient management , community-based surveillance , and outbreak response . Polymerase chain reaction ( PCR ) is ≤60% sensitive in the acute phase and is consistently outperformed by serological tests [15] , [16] . Current PCR and enzyme-linked immunoassay ( ELISA ) systems further require sophisticated equipment . Agglutination , dipstick , and lateral flow assays are among other diagnostic technologies for leptospirosis whose performance has been described [17]–[24] . Collectively , these assays demonstrated insufficient sensitivity in early acute disease and some require basic laboratory support . Most rapid serological tests to date relied on genus-wide cross-reactivity to detect antigenically diverse pathogens , most commonly utilizing whole-cell antigen from the saprophytic serovar Patoc I [7] . The novel Dual Path Platform ( DPP ) ( Chembio Diagnostic Systems , Medford , New York , USA ) assay for leptospirosis incorporates high concentrations of recombinant leptospiral immunoglobulin-like ( rLig ) proteins as antigens . It thereby avoids the cross-reactivity observed in whole-cell assays with nonspecific cell surface components , such as lipopolysaccharides , that are common to other pathogens . Lig proteins are key markers for the serodiagnosis of acute-phase leptospirosis because they elicit a robust humoral immune response [25] , [26] , are conserved among pathogenic species [27] , [28] , and are active in natural infection as they are preferentially expressed at physiological osmolarity [29]–[31] and contribute to cell adhesion [32]–[34] . We rationally selected the most seroreactive combination of rLig proteins for use as antigens in the DPP assay for leptospirosis using a multi-antigen print immunoassay ( MAPIA ) ( unpublished data ) . The DPP has been successfully applied to the diagnosis of other human diseases , including syphilis [35] , and utilizes a variation of lateral flow technology , whereby the biological sample and the colorimetric marker are delivered on separate , perpendicular nitrocellulose membranes . This design increases assay sensitivity by circumventing non-specific interference between the assay's embedded marker proteins and immunoglobulin in the patient sample . In this study , we assessed the diagnostic performance of the DPP assay in the setting of urban leptospirosis transmission using the MAT as the gold standard to determine the primary outcomes of sensitivity , specificity , and reproducibility . Secondarily , we compared its diagnostic accuracy with a commonly used IgM-ELISA and correlated DPP performance with severity and duration of illness .
We adhered to comprehensive diagnostic accuracy evaluation standards ( Table S1 ) [36] and received IRB approval from FIOCRUZ , New York Presbyterian Hospital , and Yale University . Leptospirosis case-patients , non-leptospirosis febrile outpatients and healthy slum residents provided written consent and blood donors consented to its use in biomedical research . We procured sera for hepatitis A , dengue , and syphilis as anonymous reference specimens . We measured sensitivity using 446 serum samples from 378 individuals with either mild or severe leptospirosis from two urban Brazilian populations . We collected acute sera at enrollment and convalescent samples after approximately 15 days . Case-patient sera from all sites were well characterized according to clinical presentation , clinical and diagnostic laboratory results , epidemiological risk factors , and clinical outcomes using standardized data collection tools based on active case detection protocols [37] . We designated hospitalized case-patients as having severe leptospirosis , regardless of clinical syndrome , and non-hospitalized case-patients as mild leptospirosis . Both mild and severe leptospirosis case-patients were included solely on the basis of serological confirmation by the following MAT criteria: i ) seroconversion ( undetectable acute titer and convalescent titer ≥1∶200 ) , ii ) ≥four-fold rise in acute to convalescent titers , or iii ) single sample titer ≥1∶800 . We calculated specificity from 677 control sera . We ordered all samples in random sequence and assigned a blinded unique numerical code prior to testing . We double entered and cross-validated all data elements and analyzed the data with SAS v9 . 2 ( SAS Inst . ; Cary , NC , USA ) using α = 0 . 05 .
Severe disease case-patients providing acute-phase sera from Salvador were older and more frequently male than those with mild disease , whereas demographics between severe disease groups were similar ( Table S3 ) . In comparison to those providing acute-phase sera ( Table S3 ) , the 110 severe leptospirosis cases-patients from Salvador providing convalescent-phase sera less frequently died ( 0%; chi-square P<0 . 001 ) . Acute sera for mild case-patients were collected earlier than for severe disease ( Table S3 ) ; mild disease sera were collected within two days of symptoms onset for 70% compared to <4% for severe disease from Salvador ( chi-square P<0 . 001 ) . Case-patients designated as mild leptospirosis had objectively less severe disease than those designated as severe leptospirosis per several clinical indicators ( Table S3 ) , which correlated with mild disease less frequently diagnosed clinically as leptospirosis ( 95% for severe vs . 7% for mild; chi-square P<0 . 001 ) . Among those with severe disease , Salvador case-patients were sicker according to clinical jaundice , oliguria , tachypnea , elevated serum creatinine ( ≥4 mg/dL ) , and total serum bilirubin ( >1 . 5 mg/dL ) ( Table S3 ) . Most case-patients from Salvador ( 96% of severe and 93% of mild ) had infections presumptively caused by the locally dominant serogroup , L . interrogans Icterohaemorrhagiae , compared with 70% from Recife ( chi-square P<0 . 001 ) . Finally , case-patient groups differed by MAT confirmation criteria . Few Recife case-patients had convalescent specimens available and consequently a significantly greater proportion was confirmed with a single titer ≥1∶800 ( Table S3 ) . The overall sensitivity for the 42 severe disease and 26 mild disease patients with paired sera evaluated by DPP was 100% ( 95% CI 92–100% ) and 73% ( 52–88% ) , respectively . Sensitivity did not differ significantly between laboratories ( data not shown ) . We measured higher DPP sensitivity in the acute phase for severe disease from Salvador ( 85% ) and Recife ( 78% ) compared to mild disease ( 64% ) ( Table 1 ) . Sensitivity was lower for sera collected <7 days after disease onset: 77% for severe disease from Salvador , 43% from Recife , and 60% for mild disease . For severe case-patients from Salvador collected <7 days of onset , the acute-phase sensitivity for DPP ( 77% , 66–85% ) was superior to the 1:00 MAT screening titer ( 46% , 35–58%; P<0 . 001 ) and showed a trend toward superiority over the IgM-ELISA ( 65% , 54–76%; P = 0 . 12 ) . In convalescence , the sensitivity was 98% for severe disease from Salvador and 50% for mild disease . Of 18 DPP-positive mild acute sera , seven ( 39% ) were negative in convalescence , despite an increase in MAT titer from the acute phase for six ( data not shown ) . DPP specificity was >95% except among Brazilian blood donors ( 93% ) and slum residents ( 86% ) , for which IgM-ELISA outperformed DPP ( chi-square P = 0 . 001 ) ( Table 2 ) . Among the 86 slum residents for whom MAT titers were known , DPP specificity ( 87% , 78–93% ) was also inferior to the MAT screening titer 1∶100 ( 97% , 90–99%; P = 0 . 05 ) ; rather , it was equivalent to the titer 1∶50 ( 86% , 77–93% ) . Sensitivity for both DPP and IgM-ELISA was positively correlated with duration of symptoms in both severe disease from Salvador and Recife ( combined in Figure 1A ) and mild disease ( Figure 1B ) . In the 14 days after onset for severe leptospirosis , regression on prevalence analysis ( chi-square = 10 . 1 , P = 0 . 002 ) estimated a daily increase in DPP sensitivity of 2 . 7% . Notably , the DPP assay outperformed IgM-ELISA early in both severe and mild disease , when treatment initiation is critical . We similarly found a positive relationship between symptom duration and MAT titer for the combined cases of severe disease from Salvador and Recife , and for the mild disease cases ( Spearman ρ = 0 . 24 , P<0 . 001 ) . The proportion of all severe disease acute specimens with high MAT titers ( ≥1∶800 ) was 17% on days 2–3 after onset and then rose to 98% after day 11 ( data not shown ) . Severe leptospirosis case-patients with more serious clinical manifestations had an increased likelihood of a positive DPP result ( data not shown ) . Sensitivity varied according to higher serum creatinine ( 91% for creatinine ≥4 mg/dL vs . 79% for creatinine <4 mg/dL , chi-square P = 0 . 03 ) and clinical jaundice ( 88% for jaundiced vs . 60% for not jaundiced , chi-square P<0 . 001 ) , but we found no difference by presumptive infecting serogroup ( 85% for each Icterohaemorrhagiae and other serogroups; chi-square P = 0 . 95 ) . A logistic regression model incorporating days of illness ( OR 1 . 25 , 95% CI 1 . 06–1 . 47 ) , jaundice ( OR 2 . 94 , 95% CI 1 . 10–7 . 84 ) , and serum creatinine ≥4 mg/dL ( OR 1 . 24 , 95% CI 1 . 01–1 . 54 ) ( global Wald chi-square = 19 . 1 , P<0 . 001 ) suggested that duration of illness and disease severity independently influenced DPP performance . Based on the pre-test probability of 90% , we estimated PPV and NPV of the DPP assay for severe acute clinical leptospirosis to be 98% and 39% , respectively . Using the pre-test probability of 50% for mild disease in the outpatient setting , the estimated PPV and NPV were 81% and 69% , respectively . Inter-operator reproducibility across three operators was very good to excellent ( kappa 0 . 82 , 95% CI 0 . 76–0 . 89 to kappa 0 . 94 , 95% CI 0 . 92–0 . 96 ) and test-to-test reproducibility was very good ( kappa 0 . 89 , 0 . 80–0 . 98 ) . Upon repeat testing , 92% of originally positive and 97% of originally negative assays were interpreted concordantly . Diagnostic performance correlated with the intensity of the assay's reaction . When no colored band was visualized at the test line ( Figure 2A ) , the probability of a properly assigned negative result was 89–91% . Similarly , the probability of correctly identified case-patient status was 95–100% for assays with moderate to strong reactivity ( Figure 2B ) . The probability of an accurate classification was lower , however , with weakly reactive ( Figure 2C ) interpretations ( 71–83% ) .
We measured the diagnostic performance of a novel point-of-care immunoassay for leptospirosis developed from rLig protein fragments . The DPP assay , which detects both IgM and IgG , is sensitive for acute-phase severe leptospirosis and was superior to IgM-ELISA in the first week of illness . We previously reported the superior immunoblot IgM detection against recombinant Lig proteins compared with whole-cell IgM-ELISA and other recombinant proteins [25] . The DPP assay further improves on existing technology by independently delivering the biological sample and the antibody-detecting conjugates to the test line . This method thereby reduces interference between the immunoglobulins in the biological sample and their conjugate proteins that may occur with conventional , single path lateral-flow assays . Lower sensitivity for mild illness was observed for both DPP and IgM-ELISA , perhaps due to earlier patient presentation in the outpatient setting when immunoglobulin development is underway . Alternatively , an altogether weaker antibody response to mild leptospirosis may occur [44] . Similarly low sensitivity in mild disease convalescence was noted for other rapid serological tests [45] and in our previous work using an rLig membrane-based assay among febrile outpatients from Thailand . Like mild case-patients in the present study , we found that those from Thailand did not require hospitalization , presented earlier , and had lower MAT titers ( unpublished data ) . Lastly , mild case-patients had nondistinctive clinical presentations , were more frequently confirmed with a single MAT titer ≥1∶800 , and resided in a high-risk area for previous exposures . Some of the mild case-patients included in this study therefore may have presented with other diseases erroneously attributed to acute clinical leptospirosis . The sensitivities for severe leptospirosis from Salvador ( 85% ) and Recife ( 78% ) were not statistically different , a comparison limited in power by the small sample of confirmed case-patients from Recife . However the trend toward lower sensitivity in Recife may be explained by disease severity . Both recruitment sites for severe leptospirosis used the same inclusion criteria and both serve as state reference hospitals for severe leptospirosis , yet severe leptospirosis case-patients from Recife were less acutely ill than those from Salvador . The DPP assay specificity was good in both sick ( i . e . , illnesses other than leptospirosis ) and healthy populations . The DPP assay satisfactorily excluded diseases that may exhibit clinical presentations similar to leptospirosis and cross-reacting antibodies , establishing its suitability for the acute care setting . It also performed well in at-risk Brazilian blood donors . Specificity was relatively low in samples from healthy residents of a slum population highly exposed to Leptospira [5] compared with that for samples from residents of the same slum that presented to clinic with acute febrile illnesses . We included sera from healthy slum residents without regard to MAT status , whereas sera from residents with acute fever were screened for negative MAT titers in both the acute and convalescent phases . These observations suggest that the presence of low-level agglutinating antibody titers , which may persist for months to years [4] , [46] , even after mild disease [47] , affected test performance in this group . In the hospital setting , a positive DPP result predicted disease status with high probability . However , a negative result did not effectively exclude leptospirosis among severe disease suspects . Further diagnostic evaluation for leptospirosis should be pursued in hospitalized patients with high suspicion for leptospirosis , particularly at early stages of infection . In outpatient settings where the prevalence of leptospirosis is typically low , clinicians should use clinical and epidemiological reasoning in selecting patients for DPP testing and thereby enhance pre-test probability for leptospirosis . We showed that stratifying severe disease case-patients by end organ injury , manifested as jaundice and elevated serum creatinine , correlated with DPP positivity . Even though the model was biased toward the sickest patients , these findings suggest a potential means for stratifying PPV on clinical criteria . Ours is the first evaluation of a field-ready rapid assay for leptospirosis to stratify test performance simultaneously by both disease severity and duration of symptoms . The results suggest that performance of serological assays for leptospirosis should ideally be evaluated in the context of both . The DPP assay was developed principally for earlier diagnosis of acute clinical leptospirosis and we established its utility in that respect . We expect the assay to also provide more timely diagnostic information for public health surveillance . Nonetheless , our study has limitations . The DPP assay relies on subjective visual interpretation for diagnosis and weakly reactive assays may be ambiguous . Further , we included some case-patients in this study without paired sera and , although we conservatively confirmed them with a high single-titer MAT threshold ( ≥1∶800 ) , therefore we did not observe a rise in titers in these individuals . The lack of convalescent samples may have also contributed to the wider variation in presumptive infecting serogroup for Recife cases . The referral process for specialty care of severe leptospirosis was more centralized in Salvador than in Recife during the study period , thereby possibly making Recife case-patients less representative of the regional severe leptospirosis patient population . Lastly , we defined leptospirosis cases using an imperfect gold standard test , probably resulting in an underestimate of the DPP assay's diagnostic performance [48] . In summary , the field-ready DPP assay displayed acceptable diagnostic performance for severe leptospirosis , was highly reproducible , and can be easily implemented in hospitals where leptospirosis is a major public health problem . The next generation assay must improve detection of mild and early-phase illness , and previous work suggests that increased accuracy may be achieved with independent measurement of IgM and IgG antibodies in areas of high endemic transmission [25] , [49] . The results from this study may be generalizable throughout urban Brazil where the epidemiology of leptospirosis is similar [37] , [50] , yet the diagnostic value of the DPP assay should be evaluated in other epidemiological settings and in serial patients with clinical syndromes consistent with leptospirosis to validate its point-of-care efficacy using whole blood . | Leptospirosis is an important cause of acute fever in the tropics and the mortality rate may exceed 15% in patients with severe disease manifestations . The gold standard serological test for diagnosing leptospirosis , the microagglutination test or MAT , requires significant laboratory resources and results are not timely . Improved diagnostics are therefore critically needed to identify patients and outbreaks earlier and to thereby prevent unnecessary deaths . The need for a rapid diagnostic test is particularly acute in resource-poor settings where leptospirosis is a major public health problem and sophisticated laboratories are unavailable . In this study , we measured the diagnostic accuracy of the novel Dual Path Platform ( DPP ) for leptospirosis using serum from patients with mild and severe disease . The DPP assay detected up to 85% of severe leptospirosis and 64% of mild leptospirosis patients using the initial clinical specimen collected at hospital presentation and its diagnostic performance was comparable to a commonly used IgM-ELISA . Furthermore , the DPP assay produces a result in 20 minutes and can be more easily implemented in field settings than existing diagnostic technologies . The commercially available DPP kit offers the simple , accurate , and quick diagnosis of leptospirosis and , consequently , more timely clinical and public health decision-making . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"immunology",
"bacterial",
"diseases",
"test",
"evaluation",
"clinical",
"epidemiology",
"global",
"health",
"neglected",
"tropical",
"diseases",
"immunologic",
"techniques",
"infectious",
"diseases",
"epidemiology",
... | 2012 | Accuracy of a Dual Path Platform (DPP) Assay for the Rapid Point-of-Care Diagnosis of Human Leptospirosis |
The information processing abilities of neural circuits arise from their synaptic connection patterns . Understanding the laws governing these connectivity patterns is essential for understanding brain function . The overall distribution of synaptic strengths of local excitatory connections in cortex and hippocampus is long-tailed , exhibiting a small number of synaptic connections of very large efficacy . At the same time , new synaptic connections are constantly being created and individual synaptic connection strengths show substantial fluctuations across time . It remains unclear through what mechanisms these properties of neural circuits arise and how they contribute to learning and memory . In this study we show that fundamental characteristics of excitatory synaptic connections in cortex and hippocampus can be explained as a consequence of self-organization in a recurrent network combining spike-timing-dependent plasticity ( STDP ) , structural plasticity and different forms of homeostatic plasticity . In the network , associative synaptic plasticity in the form of STDP induces a rich-get-richer dynamics among synapses , while homeostatic mechanisms induce competition . Under distinctly different initial conditions , the ensuing self-organization produces long-tailed synaptic strength distributions matching experimental findings . We show that this self-organization can take place with a purely additive STDP mechanism and that multiplicative weight dynamics emerge as a consequence of network interactions . The observed patterns of fluctuation of synaptic strengths , including elimination and generation of synaptic connections and long-term persistence of strong connections , are consistent with the dynamics of dendritic spines found in rat hippocampus . Beyond this , the model predicts an approximately power-law scaling of the lifetimes of newly established synaptic connection strengths during development . Our results suggest that the combined action of multiple forms of neuronal plasticity plays an essential role in the formation and maintenance of cortical circuits .
The computations performed by cortical circuits depend on their detailed patterns of synaptic connection strengths . While the gross patterning of connections across different cortical layers has been well described in some cases [1] , [2] , the detailed connectivity structure between groups of cells and its relation to information processing have been notoriously difficult to investigate [3] . This detailed structure could either be largely random – the product of somewhat arbitrary growth processes , or it could be highly organized . On the one hand , randomly structured networks have been shown to possess powerful computational properties [4]–[6] and they are easy to generate . On the other hand , a precise non-random organization could be the product of network self-organization , where network structure determines neural activity patterns and activity patterns in turn shape network structure through plasticity mechanisms . At the macroscopic and mesoscopic scales , models based on self-organization have already explained fundamental features of brain networks . Examples are the formation of topographic mappings [7] or properties of orientation preference maps in primary visual cortex [8] , [9] . Here we show that fundamental aspects of the microscopic structure of cortical networks can also be understood as the product of self-organization . Self-organization typically relies on a combination of self-reinforcing ( positive feedback ) processes that are combined with a competition for limited resources . In the context of Neuroscience , an example of a self-reinforcing process may be that correlated firing of two groups of neurons may strengthen synaptic connections between them according to Hebb's postulate of synaptic plasticity , while the strengthened connections will in turn amplify the correlated firing of the neurons . An example for competition for a limited resource may be a synaptic scaling mechanism that limits the sum of a neuron's synaptic efficacies such that one synapse can only grow at the expense of others . The combination of self-reinforcing mechanisms with limited resources often gives rise to the formation of structural patterns , which may or may not have specific functional advantages . Here , we will offer an explanation for fundamental aspects of the fluctuations of synaptic strength and the distribution of synaptic efficacies based on self-organization . Specifically , recent evidence shows that the distribution of synaptic efficacies is highly skewed [10] , [11] , having an approximately lognormal distribution [12]–[14] . Only around 20% of synapses are responsible for 50% of total synaptic weight . Importantly , synaptic contacts are constantly being created and destroyed and sizes of dendritic spines are fluctuating over time scales of hours and days [14] , [15] . In the face of this highly dynamic network structure , stable long-term memories are thought to be based on subsets of synapses with long life times [16] , [17] , which may also be comparatively strong [16] . In line with this , the daily fluctuations of dendritic spine sizes , which are closely related to synaptic efficacies , are such that weak synapses can change their size by as much as a factor of 6 , while strong synapses are much more stable [15] . To investigate whether and how these properties can arise from self-organization induced by neuronal plasticity mechanisms , we have developed a self-organizing recurrent network ( SORN ) model . It extends a previous model [18] , and consists of noisy binary threshold spiking neurons ( 80% excitatory and 20% inhibitory ) and uses five different forms of plasticity ( see Materials and Methods for details ) . Connections between excitatory neurons are subject to an additive spike-timing dependent plasticity ( STDP ) rule that changes synaptic strength in a temporally asymmetric causal fashion as reported experimentally [19] , [20] . A synaptic normalization mechanism keeps the sum of all excitatory weights to a neuron constant and models classic findings on multiplicative synaptic scaling of synaptic efficacies [21] , [22] . An intrinsic plasticity mechanism adjusts the firing thresholds of excitatory neurons to maintain a low average firing rate . This mechanism models homeostatic changes in neuronal excitability through modification of voltage gated ion channels observed experimentally [23] , [24] . Connections from inhibitory neurons onto excitatory neurons are subject to an inhibitory spike-timing dependent plasticity ( iSTDP ) rule that balances the amount of excitatory and inhibitory drive that the excitatory neurons receive as reported in recent studies [25]–[27] . Finally , a structural plasticity rule generates new synaptic connections between excitatory cells at a small rate . This models the constant generation of new synaptic contacts observed in cortex and hippocampus [15] , [28] .
We simulated networks of 200 excitatory and 40 inhibitory neurons for 10 , 000 time steps and observed the resulting activity patterns ( Fig . 1 ) and distributions of synaptic strength ( Fig . 2 ) . The network shows irregular activity patterns reminiscent of cortical recordings ( Fig . 1A ) . Inter-spike interval ( ISI ) distributions are well fitted by an exponential function ( Fig . 1B ) and coefficient of variation ( CV ) values are close to one ( Fig . 1C ) as would be expected from a Poisson process . Neurons show only very weak correlations of their firing during this phase of network development ( Fig . 1D ) . To estimate the probability distribution governing excitatory-to-excitatory synaptic strengths we bin connection strengths and divide the number of occurrences in each bin by the bin size . The bin sizes are uniform on the log scale . To mimic experimental procedures [15] , very small synapses ( ) are excluded . Fig . 2A–D shows the distribution of synaptic connection strengths after 10 , 000 time steps and compares it to EPSP data from rat visual cortex [12] . With distinctly different initial conditions ( Fig . 2E ) , the network faithfully develops a long-tailed distribution of connection strengths that is similar to the biological data ( see Text S1 for details ) . Experimental data and model results are both well fit by lognormal distributions . As the network evolves it goes through different phases ( Fig . 3 ) . The initial phase is characterized by a decay of connectivity , where a substantial fraction of the excitatory-to-excitatory synaptic weights get eliminated ( Fig . 3A ) . In the subsequent growth phase , the network connectivity recovers through the integration of newly created synapses produced by the structural plasticity . Eventually , the degree of connectivity stabilizes and the network enters into a stable regime . Here , connectivity fluctuates very little ( Fig . 3A inset ) . Newly created synapses tend to quickly disappear and there is a large stable backbone of connections with extremely long life times ( as long as we simulated ) . The distribution of excitatory-to-excitatory connection strengths is lognormal-like throughout most of the network's evolution ( Fig . 3B–D ) . ( see Fig . S2 in Text S2 for more results with different parameters ) . An exception is the transition from the decay to the growth phase , where large deviations from the lognormal shape are observed ( not shown ) . However , the distribution of synaptic weights maintains a long tail and a positive skewness throughout its development . The thresholds of the excitatory units in the network develop an approximately Gaussian distribution . In the stable regime of the network , this distribution is exhibiting only small fluctuations . As a next step , we assessed the dynamics of synaptic connection strengths in SORN . Fig . 4A shows traces of 6 synaptic connection weights as a function of time . The distribution of life times of newly created synapses is well described by a power law with an exponent close to −3/2 during this phase as expected for random walk behavior ( Fig . 4B ) . We next compared the weight changes occurring in SORN over 3000 time steps with experimental data from time lapse imaging of dendritic spine sizes in rat hippocampus [15] . In both SORN and the experimental data , strong synapses are found to have comparatively small fluctuations ( Fig . 4C–F ) . This is not a simple ceiling effect , since synaptic weights could , in principle , grow much larger than the typical values for very strong synapses we observe in the model , which lie between 0 . 2 and 0 . 3 . There exists a small population of synaptic connections in both model and experimental data which decays completely ( horizontal lines in Fig . 4C , D and oblique lines in Fig . 4E , F ) . The population of synapses clustered on the Y-axis in Fig . 4E , F represents newly established synaptic connections . The big fluctuations are mostly seen in decay phase and imply that the network is far from stability in this regime ( see Fig . S6 in Text S2 for additional results with different parameters showing weight fluctuations during different phases of network evolution ) . To better understand the mechanism through which the network self-organizes its connectivity and dynamics , we examined how the strength of a synaptic connection influences its probability of undergoing further growth or decline . Among all the plasticity mechanisms , only STDP and synaptic normalization adjust the weights of EE connections . While synaptic normalization will only scale all incoming excitatory-to-excitatory connections linearly , STDP has the power to change the shape of the distribution of synaptic weights impinging onto a neuron . When we recorded the isolated effect of STDP , i . e . independently of the synaptic normalization , we found that over a large range of synaptic weight strengths , the expected increase in strength of a connection due to STDP grows approximately linearly with the strength of the synapse ( Fig . 5A ) . The fraction of connections undergoing depression depends much less on connection weight ( Fig . 5B ) . Thus , the net effect is that stronger synaptic connections have a higher chance to be potentiated by STDP establishing a rich-get-richer behavior ( Fig . 5C ) . This mechanism is kept in check by the synaptic normalization mechanism , which scales weights in a multiplicative fashion . We estimated the mean absolute change of synaptic connection strengths due to STDP and synaptic normalization over 200 time step intervals during the initial 10 , 000 time steps . The mean absolute sizes of fluctuations grow roughly linearly with weight ( Fig . 5D ) as observed experimentally [14] . Note that this approximately linear dependence on weight strength occurs despite the additive STDP rule we are using and does not require a multiplicative STDP rule [12] . With all forms of plasticity present , the network will show irregular firing activity and develop a lognormal-like weight distribution . These results are stable over a large range of parameter values ( see Text S2 for details ) . To investigate the extent to which the different forms of plasticity contribute to these results , we performed simulations where we switched off individual plasticity mechanisms . When synaptic normalization is switched off , the network will show bursts of high activity separated by long periods of inactivity . As shown in Fig . 4 , the network keeps eliminating synapses as a result of STDP . The structural plasticity counteracts this process . If we switch off the structural plasticity , a large number of neurons eventually lose all their postsynaptic targets . No lognormal-like weight distribution will emerge if one or both forms of plasticity are missing . Intrinsic plasticity and inhibitory STDP both try to maintain a low average firing rate of excitatory cells and both are important to keep healthy network dynamics . If both are switched off , some units will exhibit very high firing rates while others remain essentially silent and all the phenomena shown in Fig . 1–5 will disappear . To study the individual effects of intrinsic plasticity and iSTDP , Fig . 6 shows a scatter plot of the fraction of active excitatory units in subsequent time steps . With all plasticity mechanisms active , the network activity is confined within a small area . Activity never dies out or becomes very big . When either intrinsic plasticity or inhibitory STDP is switched off , the network activity exhibits big fluctuations and can temporarily die out completely . In certain parameter regimes the network may function without one or the other , but with both mechanisms being present , we obtain robust results over a large range of parameter values . We conclude that all five plasticity mechanisms are important for proper self-organization .
Understanding the structure and dynamics of neural circuits and reproducing them in neural network models remains a major challenge . Classic models of STDP have been shown to lead to physiologically unrealistic bimodal weight distributions under certain conditions [29] . This has lead to the proposal of a number of modifications to STDP rules to remedy the problem . Specifically , multiplicative STDP rules have received much interest recently [30] , [31] . Here we have shown that an additive STDP rule when operating together with other plasticity mechanisms in a recurrent network is sufficient to explain both the statistics and fluctuations of synaptic connection strengths observed in cortex . Associative synaptic plasticity induces a rich-get-richer dynamics of synaptic weights , while homeostatic mechanisms induce competition . With distinctly different initial conditions , the ensuing self-organization faithfully develops Poisson-like irregular firing patterns , lognormal-like weight distributions and the characteristic pattern of fluctuations of synaptic strengths reminiscent of cortical recordings . Beyond this , our model predicts a power-law scaling of the lifetimes of newly established synaptic connections during development . Our results suggest that the statistics and dynamics of neural circuits are the product of network self-organization , and that the combined action of multiple forms of neuronal plasticity plays an essential role in the formation and maintenance of cortical circuits . It is important , however , to also consider alternative explanations . One of the simplest ways to obtain lognormal distributions is by virtue of Gibrat's law , which was originally developed in Economics . It describes the growth of companies by random annual growth rates which are independent of the companies' sizes . This process by itself , when applied to the growth of synaptic connections , would predict that the variance of the synaptic weight distribution would grow without bounds , which is clearly at odds with biological reality . Adding a multiplicative normalization mechanism such as our synaptic normalization rule to Gibrat's proportionate growth process retains the development of a lognormal-like distribution while avoiding the problem of unbounded growth . However , this model does not reproduce the pattern of weight fluctuations observed experimentally . Furthermore , such a model is purely phenomenological and does not describe the mechanism that causes the synaptic fluctuations in the first place . Similarly , the models proposed in [15] and [14] describe the fluctuations of synaptic weights as independent random walk processes , but do not explain what causes the synaptic fluctuations . In contrast , our model offers a mechanistic account that explains the patterns of weight fluctuations and the distribution of synaptic strength in terms of fundamental processes of neuronal plasticity in a recurrent network . This approach is consistent with the finding in [15] that the fluctuations of dendritic spine sizes seem to strongly depend on activity-driven synaptic plasticity . Specifically , they found strongly reduced fluctuations of spine sizes and fewer spine eliminations when inhibiting NMDA receptors with APV or MK-801 . Interestingly , the generation of new spines was unaffected by this manipulations . This is consistent with our model's assumption that the generation of new spines occurs via a process of structural plasticity that is independent of activity-driven synaptic changes . A further advantage of our model is that it can also be used to derive predictions regarding the emerging network topology in terms of clustering , network motifs , etc . This topic is left for future work . If our model is essentially correct , despite its very abstract formulation , then one should be able to replicate the present results in more realistic network models of spiking neurons . As a first step in this direction , we have constructed a version of the model using leaky-integrate-and-fire neurons with realistic parameter values . We have also adapted the plasticity mechanisms for this network . Initial explorations show that major features such as the lognormal-like weight distribution and the pattern of synaptic fluctuations can also be found in this less abstract network model . Future work will elaborate on these preliminary results . Since the structure of cortical circuits determines the dynamics of neuronal activity , it also determines how information is encoded and propagated . The existence of a small number of very strong synaptic connections may greatly facilitate the highly reliable propagation of signals along pools of neurons [32] . In fact , SORN networks have previously been shown to spontaneously develop encoding strategies based on trajectories through their high-dimensional state space of unit activations [18] . In this work , the networks were fed with structured time series of input letters and were shown to learn internal representations of these input sequences that allowed large performance increases in prediction tasks . This was found to be due to the ongoing self-organization in the network driven by the network's plasticity mechanisms . They were shown to effectively increase the separation of network states belonging to different input conditions . More recently , we have found evidence that such networks may naturally self-organize to perform computations resembling Bayesian inference processes [33] . Further work is needed to better understand how the network's self-organization enables it to behave this way . Many computational models of local cortical circuits assume random network structure [4]–[6] , sometimes with distance-dependent or layer-dependent connection probabilities [34] . Such random network structure is at odds with recent evidence that changes to the connectivity structure such as the generation of stable new spines are associated with the formation of new memories [35] . Hence , we believe that the study of random networks where only connection statistics are matched to those in the brain , may be quite misleading when the goal is to understand processing in cortical circuits . Instead , self-organizing networks , which can faithfully develop brain-like activity and connectivity patterns , seem a much more promising subject of study .
We use a SORN ( self-organizing recurrent neural network ) model [18] that uses noisy units , incorporates additional plasticity mechanisms , and receives no external input . The network is composed of excitatory and inhibitory threshold neurons connected through weighted synaptic connections . is the connection strength from neuron to neuron . We distinguish connections from excitatory to excitatory neurons ( ) , excitatory to inhibitory connections ( ) and inhibitory to excitatory connections ( ) . Connections between inhibitory neurons and self-connections of excitatory neurons are forbidden . The connections onto excitatory cells ( and ) are subject to synaptic plasticity mechanisms described below . and connections have sparse random initial connectivity with connection probabilities of 0 . 1 and 0 . 2 , respectively . The remain fixed at their random initial values . They have all-to-all topology and are drawn from the interval and subsequently normalized such that the incoming connections to an inhibitory neuron sum up to one: . The network's activity state , at a discrete time , is given by the binary vectors and corresponding to the activity of the excitatory and inhibitory neurons , respectively . The evolution of the network state is described by: ( 1 ) ( 2 ) The and are threshold values for the excitatory and inhibitory neurons , respectively . They are initially drawn from a uniform distribution in the interval and . The Heaviside step function constrains the activation of the network at time to a binary representation: a neuron fires if the total drive it receives is greater then its threshold , otherwise it stays silent . and represent white Gaussian noise with and . The time scale of a single iteration step in the model corresponds to typical membrane time constants and widths of spike-timing dependent plasticity ( STDP ) windows — lying roughly in the range of 10 to 20 ms . Note that in order to save computation time the homeostatic plasticity mechanisms described below are simulated to be much faster than in reality . The network relies on several forms of plasticity: STDP of EE and EI connections , synaptic scaling and structural plasticity of EE connections , and intrinsic plasticity regulating the thresholds of excitatory neurons . The set of synapses adapts via a causal STDP rule that strengthens the synaptic weight by a fixed amount whenever neuron is active in the time step following activation of neuron . When neuron is active in the time step preceding activation of unit , is weakened by the same amount ( or set to zero if necessary to prevent it from becoming negative , which triggers synapse elimination ) : ( 3 ) Synaptic normalization proportionally adjusts the values of incoming connections to an excitatory neuron at each time step so that they sum up to one: ( 4 ) This rule does not change the relative strengths of synapses established by STDP but regulates the total incoming drive a neuron receives and limits weight growth . It leads to a competition among excitatory-to-excitatory connections impinging onto the same neuron such that growth of some connections is compensated by the decay of others . An intrinsic plasticity rule maintains a constant average firing rate in every neuron . To this end , a neuron that has just been active increases its threshold while an inactive neuron lowers its threshold by a small amount: ( 5 ) where sets the target firing rate . For simplicity , one can also set the same target firing rate for all the excitatory neurons . Note that the synaptic normalization and intrinsic plasticity mechanism operate faster in the model than they would in biological brains . This choice is warranted because of a separation of time scales and speeds up the simulations . Compared to the original SORN model , we introduce two additional forms of plasticity . Structural plasticity adds new synaptic connections between excitatory cells to the network at a small rate , which balances the synapse elimination induced by STDP . With probability a new connection is added between a random pair of excitatory cells that are unconnected . The strength of this weight is set to 0 . 001 . Inhibitory spike-timing dependent plasticity ( iSTDP ) adjusts the weights from inhibitory to excitatory neurons to balance the amount of excitatory and inhibitory drive a neuron is receiving . If the inhibitory neuron spikes and the excitatory neuron remains silent in the subsequent time step ( the inhibitory spike was “successful” in preventing the excitatory cell from spiking ) , the inhibitory weight is reduced by an amount ( or set to a small positive value of 0 . 001 if necessary to prevent it from being eliminated ) . If , however , the inhibitory neuron spikes and the excitatory neuron also spikes in the subsequent time step ( the inhibitory spike was “unsuccessful” in preventing the excitatory cell from spiking ) , the inhibitory weight is increased by the larger amount . In all other cases the weight remains unchanged: ( 6 ) Equivalently , we can write: ( 7 ) Unless otherwise specified , the initial weights of , and are drawn from a uniform distribution as shown in Fig . 2E , and the simulations are conducted using the following parameters . , , , , , , , , . | The computations that brain circuits can perform depend on their wiring . While a wiring diagram is still out of reach for major brain structures such as the neocortex and hippocampus , data on the overall distribution of synaptic connection strengths and the temporal fluctuations of individual synapses have recently become available . Specifically , there exists a small population of very strong and stable synaptic connections , which may form the physiological substrate of life-long memories . This population coexists with a big and ever changing population of much smaller and strongly fluctuating synaptic connections . So far it has remained unclear how these properties of networks in neocortex and hippocampus arise . Here we present a computational model that explains these fundamental properties of neural circuits as a consequence of network self-organization resulting from the combined action of different forms of neuronal plasticity . This self-organization is driven by a rich-get-richer effect induced by an associative synaptic learning mechanism which is kept in check by several homeostatic plasticity mechanisms stabilizing the network . The model highlights the role of self-organization in the formation of brain circuits and parsimoniously explains a range of recent findings about their fundamental properties . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"circuit",
"models",
"computational",
"neuroscience",
"biology",
"neuroscience"
] | 2013 | Network Self-Organization Explains the Statistics and Dynamics of Synaptic Connection Strengths in Cortex |
Positive strand RNA viruses , such as dengue virus type 2 ( DENV2 ) expand and structurally alter ER membranes to optimize cellular communication pathways that promote viral replicative needs . These complex rearrangements require significant protein scaffolding as well as changes to the ER chemical composition to support these structures . We have previously shown that the lipid abundance and repertoire of host cells are significantly altered during infection with these viruses . Specifically , enzymes in the lipid biosynthesis pathway such as fatty acid synthase ( FAS ) are recruited to viral replication sites by interaction with viral proteins and displayed enhanced activities during infection . We have now identified that events downstream of FAS ( fatty acid desaturation ) are critical for virus replication . In this study we screened enzymes in the unsaturated fatty acid ( UFA ) biosynthetic pathway and found that the rate-limiting enzyme in monounsaturated fatty acid biosynthesis , stearoyl-CoA desaturase 1 ( SCD1 ) , is indispensable for DENV2 replication . The enzymatic activity of SCD1 , was required for viral genome replication and particle release , and it was regulated in a time-dependent manner with a stringent requirement early during viral infection . As infection progressed , SCD1 protein expression levels were inversely correlated with the concentration of viral dsRNA in the cell . This modulation of SCD1 , coinciding with the stage of viral replication , highlighted its function as a trigger of early infection and an enzyme that controlled alternate lipid requirements during early versus advanced infections . Loss of function of this enzyme disrupted structural alterations of assembled viral particles rendering them non-infectious and immature and defective in viral entry . This study identifies the complex involvement of SCD1 in DENV2 infection and demonstrates that these viruses alter ER lipid composition to increase infectivity of the virus particles .
Phospholipids are critical for membrane structure , function and stability of eukaryotic cells . Specific distributions of lipids within these membranes define their characteristics such as curvature , fluidity , leakiness and the interactions between membranes and membrane-bound protein complexes . A key approach to alter the architecture of a membrane is to incorporate unsaturated fatty acyl chains , to induce curvature and fluidity in a lipid bilayer , altering its functional capacity [1 , 2] . Unsaturated fatty acids ( UFA ) are generated in the cytoplasm and after their initial desaturation they are further elongated , desaturated and shunted towards triglyceride , cholesterol ester or phospholipid synthesis . This initial desaturation event is the rate-limiting step in UFA biosynthesis and is catalyzed at the Δ9 position in the carbon chain by stearoyl CoA desaturase ( SCD ) [3 , 4] . In humans , it has two isoforms: SCD1 is ubiquitously expressed and preferentially converts stearic and palmitic acids into oleic and palmitoleic acids , respectively . SCD5 , is restricted to the brain and pancreas [5] . SCD1 is a 40 kD integral membrane protein in the endoplasmic reticulum ( ER ) and is highly conserved from bacteria to mammals [6] . It regulates the balance between saturated and monounsaturated fatty acids ( MUFA ) in the cell . Flaviviruses are obligate intracellular pathogens that hijack lipid metabolic pathways for their energy and substrate requirements . As enveloped viruses , they rely heavily on host phospholipid membranes at every stage of their life cycle and alter the architecture and composition of these membranes to fit their replicative needs [7 , 8 , 9 , 10] . Specifically , flaviviruses target membranes of the ER to generate a scaffold for the assembly of viral protein complexes , concentrate substrates required for genome replication , and protect the double-stranded RNA replicative intermediates from detection by the cellular immune response [11] . As a result , the architecture of the ER membrane is altered to form structures known as convoluted membranes ( CM ) , vesicles ( Ve ) and vesicle packets ( Vp ) [12 , 8 , 13] . The CM are considered to be sites for viral protein translation . The Vp/Ve are sites for viral RNA replication [12] . Additionally , these viruses co-opt the ER membrane as a structural component of the virus particle ( envelope ) and use the ER for virus particle assembly and egress . Embedded in this ER-derived lipid envelope are the viral transmembrane glycoproteins , pre-membrane ( prM ) and envelope ( E ) . Structural transitions between these proteins are critical for virion maturation and infectivity . Flaviviral dependence on cellular membranes is reflected in alterations in cellular fatty acid metabolism during infection [14 , 15] . This has also been observed for other viruses [16 , 17 , 18 , 19] . The specific physiochemical properties of the required fatty acids and their influence on specific steps of the flaviviral life cycle are not known . In this study , we investigated the importance of fatty acid desaturation on the flavivirus life cycle . We evaluated required enzymes in the UFA pathway using an siRNA library and identified key restrictions that reduced replication of the flavivirus , dengue virus type 2 ( DENV2 ) . The enzymatic activity of SCD1 in particular , was required for viral replication and was regulated in a time-dependent manner . SCD1 protein expression levels were inversely correlated with the concentration of viral dsRNA ( Replicative Intermediate , RI ) in the cell . This modulation of SCD1 , coinciding with the stage of viral replication , highlighted its function as an enzyme that controls alternate lipid requirements during early and advanced infections . Loss of function of this enzyme adversely altered the maturation and infectivity of released virions . This study highlights the importance of the UFA biosynthesis pathway in flaviviral genome replication and virion infectivity .
We hypothesized that enzymes in the UFA biosynthesis pathway are important for the DENV2 lifecycle and interrogated this pathway with siRNAs ( S1 Table ) to determine effects of their transient knock-down on viral replication ( Fig 1 and S1 Fig ) . An irrelevant siRNA ( IRR ) controlled for off-target effects of siRNA treatment , while an siRNA targeting the DENV2 genome was a positive control for reduction in viral replication . We identified two enzymes in this pathway that represent host-viral interaction points , SCD1 and peroxisomal trans-2-enoyl-coA reductase ( PECR ) ( Fig 1A ) . These observations were made in two human cell lines ( Huh7 and A549 ) , yielding similar effects on viral replication ( S1A and S1B Fig ) . Cytotoxic effects of siRNA treatment were not significant ( S1C–S1E Fig ) . We found that knockdown of SCD1 expression significantly reduced DENV2 replication in all cell lines and conditions tested when compared to an irrelevant siRNA ( Fig 1A , B and 1C and S1A and S1B Fig ) . Since the initial screen was carried out with a pool of four siRNAs against SCD1 , we also validated the results with a single siRNA against SCD1 and showed a similar reduction in virion release ( Fig 1B ) . This siRNA was further tested against a luciferase expressing DENV2 replicon [14] , and we found that viral RNA replication was significantly reduced ( Fig 1C ) . Therefore , the effect of SCD1 knockdown on the release of infectious virus is at least partly mediated at the RNA replication step . We confirmed that siRNA treatments were effective at reducing SCD1 mRNA and protein levels using qRT-PCR and western blot analyses . The SCD1 mRNA expression was reduced by 90% and protein levels were below the level of detection ( S1F and S1G Fig ) . These data suggest that UFA synthesis is critical for DENV2 replication . Based on the above observations , we hypothesized that DENV2 requires a stable or increased level of the SCD1 enzyme and its products to regulate the cellular lipid repertoire for its replicative advantage , and that the activity of this enzyme may be controlled by viral infection . Since SCD1 expression is regulated at the transcriptional level [20] , we first examined SCD1 mRNA levels in infected cells and found them to be increased at early time-points post infection compared to mock-infected cells ( Fig 1D ) . To ensure that the protein is translated and active during viral replication , we quantitatively examined the enzymatic activity of SCD1 in DENV2-infected cells over time . Consistent with the mRNA expression profile at the early time points , we found that cells infected with DENV2 had an initial increase in SCD1 activity , as measured by the conversion of radiolabeled stearic acid to oleic acid , as early as 6hr post infection ( Fig 1D ) . This coincides with early replication and translation of the viral genome . Also , consistent with changes in its mRNA profile , later during infection we found a decrease in SCD1 expression and activity . SCD1 is the only enzyme that can produce oleic acid . Our observations suggest that DENV2 infection specifically up-regulates SCD1 activity early in infection to expand the pool of MUFAs available for its replicative needs . Since mRNA , protein and enzymatic activity of SCD1 measured above collectively contribute to producing oleic acid , we measured oleic acid using high-resolution , liquid chromatography-mass spectrometry ( LC-MS ) . Specifically , we observed that pools of oleic acid were increased in Huh7 cells infected with DENV2 compared to two controls; UV-inactivated DENV2 exposed cells and mock-infected cells ( Fig 1E ) . The UV-inactivated DENV2 is capable of binding to and entering cells , but cannot replicate [15] . These data indicate that actively replicating virus is required to activate SCD1 . Previous studies with DENV2 have shown that lipid biosynthetic enzymes ( such as fatty acid synthase , FAS ) are recruited to viral replication complexes , to increase local synthesis of lipids at sites of viral RNA replication and virus assembly [14] . Since SCD1 is immediately downstream of FAS in the biosynthetic pathway , we investigated whether SCD1 was also re-localized to viral replication complexes during DENV2 infection . We processed mock-infected or DENV2-infected cells at 6 , 9 and 24hr for immunofluorescence studies , using antibodies against DENV2 NS3 , dsRNA ( RI ) and human SCD1 ( Fig 2 ) . Interestingly , unlike what we previously observed for FAS in Huh7 cells [14] , there is a temporal progression in marker distribution ( best observed in Fig 2A , 24hr dsRNA and SCD1 panel ) . Uninfected cells show normal SCD1 signal ( S2A Fig and Fig 2C ) , while cells with a low level ( early ) infection show that SCD1 can be found within viral replication complexes . Using Manders co-localization values [21 , 22] , we can see a correlation between RI and SCD1 in all infected cells ( Fig 2B ) . However , cells with high levels of viral RI ( late infection ) show very low levels of SCD1 ( Fig 2C ) . Quantification of the mean fluorescent intensity in a representative image of these cells demonstrated an inverse correlation between levels of SCD1 and viral markers ( Fig 2D and 2E ) . These data correspond with our findings of SCD1 expression and enzymatic activity , where Huh7 cells with high levels of RI at late time points of infection have reduced levels of SCD1 . The initial spike in SCD1 activity from SCD1 localized around viral replication and assembly sites early during infection likely generates concentrations of oleic acid sufficient for the metabolic needs throughout replication . The specificity of the SCD1 antibody was tested in SCD1 siRNA-treated cells ( S2B and S2C Fig ) . We did not see this distribution of SCD1 in DENV-infected A549 or human embryonic lung ( HEL ) cells . Rather we see a uniform co-localization of SCD1 and NS3 ( S2D and S2E Fig ) . This suggests that changes in SCD1 expression levels are cell-type specific . We used a pharmacological inhibitor to characterize the enzymatic requirement for SCD1 during DENV2 replication . The piperidine-aryl urea-based inhibitor , A939572 , which we will refer to as the SCD1 inhibitor has been shown to be effective [23] . We tested the SCD1 inhibitor in our activity assay and found that it abolished the formation of oleic acid ( Fig 3A ) . In DENV-infected cells , SCD1 inhibition resulted in a dose-dependent reduction in viral titers ( up to 2 logs ) without significant toxicity ( Fig 3B ) . Analysis of the effectiveness of the SCD1 inhibitor as an antiviral compound gave a therapeutic index of 2 . 1 . The SCD1 inhibitor was also effective against DENV2 replication in A549 cells but it had no effect on mosquito cells ( S3A and S3B Fig ) , suggesting that the Δ9 desaturase in arthropods may differ from the mammalian version . Inhibition of SCD1 halts the desaturation of stearic acid ( C18:0 ) , acid leading to a decrease in cellular concentrations of oleic acid ( C18:1 ) . Oleic acid is a key building block for more complex phospholipids , cholesterol esters and triglycerides that function as constituents of cellular and virus-induced membranes . We hypothesized that addition of exogenous oleic acid would rescue the effect of SCD1 inhibition on virus replication . To accomplish this we added oleic acid conjugated to BSA in serum free medium combined with the SCD1 inhibitor and measured viral replication . We found that addition of oleic acid restored viral replication , implying that the product of SCD1 enzymatic activity is critical for DENV2 replication ( Fig 3C ) . The rescue was not complete ( 95 . 3% , compared to the rescue with DMSO ) , likely because exogenous fatty acids have many destinations in the cell and are often shunted to β-oxidation ( Fig 3E ) [24] . Therefore , they may be minimally incorporated into the ER where they could be used for virus replication . Next we tested the SCD1 inhibitor for its effects on the replication of other enveloped , mosquito-borne viruses . We used a non-cytotoxic concentration that was effective against DENV2 and found that the replication of all other DENV serotypes and Kunjin virus ( KUNV ) , yellow fever virus ( YFV ) , Zika virus ( ZIKV ) and Sindbis virus ( SINV ) was significantly reduced ( Fig 4A–4D and 4I ) . To confirm that this was not due to off-target effects of the inhibitor , we knocked down SCD1 with siRNA and found similar effects on viral replication ( Fig 4E–H and 4I ) . These data indicate a common need for SCD1 enzymatic activity and incorporation of MUFAs into complex lipid species to aid in the replication of enveloped viruses . To further characterize the impact of SCD1 inhibition on the DENV2 life cycle , we carried out a time of addition experiment . We found that addition of the inhibitor prior to infection or during attachment followed by removal of the inhibitor had no impact on viral replication ( S4A and S4B Fig ) . However , addition of the inhibitor at any time point after infection resulted in a decrease in viral replication compared to the vehicle control ( S4C Fig ) . Addition of the siRNA after 24 hr of viral replication , however had no impact on viral replication ( S4D and S4E Fig ) . Taken together SCD1 activity is important for viral replication at multiple stages of the virus life cycle , but may be more important at the early time points and less critical at later time points . Having determined that SCD1 is important for the DENV2 life cycle , we investigated its requirement at specific stages of viral replication . Using a viral replicon , we found that inhibition of SCD1 reduced DENV2 RNA replication ( Fig 3D ) . Viral RNA replication and assembly are tightly coordinated [25] . We examined the release of infectious virus particles by quantifying the intra- and extracellular virus at 24 and 48hr post infection and found that intra- and extracellular virus from untreated cells had equivalent titers at 24hr with an increase in extracellular virus titer at 48hr ( Fig 5A ) . However , virus grown in the presence of SCD1 inhibitor lagged in release of extracellular infectious virus compared to intracellular virus at 24hr . Titers reached equivalence at 48hr but were below those of untreated controls ( Fig 5A ) . Three-way ANOVA confirmed a significant interaction between inhibitor treatment and intracellular vs . extracellular virus location ( Fig 5B; p = 1 . 600e-06 for the interaction term ) . These analyses indicate that inhibition of SCD1 affects the release of infectious virus . We next investigated the effect of SCD1 inhibition on the ratio of infectious and non-infectious virus particles released . We measured the ratio of total particles released to infectious-particles released ( the specific infectivity ) for virus grown in cells exposed to the SCD1 inhibitor compared to untreated cells . We found a small but significant reduction in total particles released from SCD1-inhibited cells as measured by genome equivalents ( GE ) ( Fig 5C ) . However , the titer of infectious particles ( as measured by plaque assay ) was reduced at both time points by almost 100-fold , similar to our previous results ( Fig 5D ) , and indicated a reduction in the specific infectivity of virus released from inhibitor-treated cells ( Fig 5E ) . Hence , when cells are treated with the SCD1 inhibitor , the total numbers of viral particles released was only slightly lowered compared to control cells , but fewer of these viral particles were infectious ( Fig 5C–5E ) . Similar results were observed when loss of function studies were carried out using an siRNA against SCD1 ( Fig 5F–5H ) . This observation was also confirmed with ZIKV , another flavivirus that also showed reduced particle release from Huh7 cells treated with the SCD1 inhibitor ( S5A Fig ) . This was not observed in C6/36 mosquito cells treated with SCD1 inhibitor ( S5B Fig ) . For DENV2 , GE in virus particles released from SCD1 inhibitor-treated Huh7 cells were compared to GE from cells treated with two other lipid synthesis inhibitors , C75 ( inhibits FAS ) and Lovastatin ( inhibits cholesterol synthesis ) ( Fig 5I ) , that had previously been shown to be effective against DENV2 [14 , 15 , 26] . We observed a defect in the release of infectious particles with all treatments ( S5C and S5D Fig ) , however , C75 treatment resulted in a larger decrease in the GE ratio compared to the other treatments ( Figs 5F and compare S5E to S5F ) . FAS activity is critical for DENV2 genome replication [14] , while cholesterol ( altered by Lovastatin ) is critical for maturation of DENV2 and generation of infectious particles [27] . Inhibition of SCD1 was similar to inhibition of cholesterol biosynthesis ( Figs 5 and S5 ) , further demonstrating its significance in the virus life cycle . We evaluated the virions released from inhibitor-treated cells for defects in infectivity . DENV2 was passaged in Huh7 cells in the presence of the SCD1 inhibitor or DMSO , and released virus particles were collected , titrated , and used to infect new Huh7 cells in the absence of inhibitor ( Fig 6A ) . During adsorption , the same concentration of inhibitor ( 10μM ) was added to control supernatants to mimic remaining , un-metabolized inhibitor in the treated-cell supernatant . Virus isolated from inhibitor-treated cells had reduced infectivity compared to virus from control cells as determined by intracellular viral RNA ( Fig 6B ) and a reduction in released infectious virus ( S5G Fig ) . This was confirmed in A549 cells ( S5H Fig ) . We confirmed this result with loss of function studies using an siRNA specific to SCD1 ( Fig 6C ) . To control for possible interference by defective particles in drug-treated cell supernatants , we UV-inactivated virus in supernatants from drug-treated and control cells and added it to equal titers of untreated virus stock to observe possible inhibition of infection . We found no difference in the resulting production of infectious virus , indicating that defective particles from cells treated with the SCD1 inhibitor did not interfere with subsequent virus infection ( Fig 6D ) . Therefore , the defect in initiating a second round of infection is unique to virus particles released from cells lacking SCD1 activity . We examined the early kinetics of viral infection with the virus from inhibitor-treated cells versus virus from control cells . We used a viral entry assay to determine the amount of virus internalized or endocytosed at given time points after attachment . Virus still external to the cell was inactivated at the indicated time point . The virus from cells treated with SCD1 inhibitor was slower to enter new cells ( 0 . 133 PFU/min versus 0 . 243 PFU/min for the control virus; Fig 6E ) . This defect was found both in the rate at which virus entered cells as well as the number of virus particles that entered the cells at equal titer of infection . Inhibition of SCD1 lowers the quantity of and changes the characteristics of infectious particles . These data indicate that there is an attachment or fusion defect that is generated in cells with decreased SCD1 enzymatic activity . To determine if the defect was in the physical structure of the virus particle ( either the prM/E glycoprotein shell or virion lipid envelope ) or in the genome encapsidated within virus particles released from SCD-inhibited cells , we isolated and transfected the RNA from these virus particles into BHK cells and measured the ability of the RNA to initiate infection . We found that both viral RNA populations were able to initiate infections with the same efficiency ( Fig 6F ) . These data suggest that there is a change to the physical structure of the virion when it is released from cells lacking SCD1 activity . We initially tested if the virions were less thermally stable when grown in the presence of the SCD1 inhibitor compared to controls . The virus from SCD1-inhibitor treated and control-infected cells were diluted to the same infectious virus concentration , heated to the indicated temperatures ( S6A Fig ) , and titrated . Nonlinear regression demonstrated that the control virus lost infectivity at 44 . 03°C and the SCD1 inhibitor-treated virus lost infectivity at 43°C . An F-test to determine the difference between the two models indicated no significant difference ( S6A Fig ) . We also carried out freeze/thaw cycles on virus samples and measured infectivity and found that control virus maintained its infectivity for 6 or more freeze/thaw cycles , but virus grown in the presence of the SCD1 inhibitor lost its infectivity after 3 freeze/thaw cycles ( S6B Fig ) . This effect was also observed with Zika virus grown in the presence of the SCD1 inhibitor ( S6C Fig ) . We limited the defect in SCD1 inhibitor-treated virus infectivity to the structure of the virion envelope , which is more susceptible to freeze-thaw transitions than the control . During infection , cells produce a range of structurally diverse particles with varying levels of infectivity . The main structural classes are immature , partially mature or fully mature and they are defined by the amount of uncleaved prM protein retained on the virion [28 , 29] . Although , the precise role of these various particles in the infectious cycle and immune modulation is not well understood we sought to determine whether virions grown in the presence of the SCD1 inhibitor had uncleaved prM protein similar to immature virus . We purified viruses from Huh7 cells infected with DENV2 at an MOI = 3 under four conditions: untreated ( WT ) , immature virus ( treated with 20mM NH4Cl ) , and virus from cells treated with the SCD1 inhibitor or with vehicle ( DMSO ) . Virus from DMSO or SCD1 inhibitor-treated cells at 24hr post-infection were pelleted through a sucrose cushion and purified by sedimentation velocity in a potassium tartrate step gradient . Visible bands ( S6D Fig ) were analyzed for viral RNA and infectious virus . A majority of the viral RNA and infectivity from both treatments sedimented in fractions 5–7 ( S6E and S6F Fig ) . To characterize the physical properties of the virus particles in each gradient we collected and similarly processed cell culture supernatants from all four conditions at 72hr post-infection , a time-point with sufficient virus for purification and analysis . The bands observed in gradients for all samples were similar to those obtained from the gradient analysis of virus harvested at 24hr with the exception of the top band ( fraction 2 at the top of the 10% interface ) that was more prominent after 72hr , indicating the presence of low molecular weight material ( S6G Fig ) . We collected each band separately and characterized them for viral RNA ( S6H Fig ) , prM , capsid and envelope proteins ( Figs 7 and S6I ) . Quantification of the western blots is shown in Fig 7C–7F . The WT virus primarily sedimented at the 15–20% interface ( fraction 6 ) and the 20–25% interface ( fraction 8 ) . Both of these fractions had similar levels of viral RNA . Typically , DENV2 purified from mosquito cells sediments in the 20–25% fraction , which was the primary fraction previously used for structure elucidation [30] . DENV2 produced in Huh7 cells show differences in sedimentation profiles . Immature virus sedimented at the same densities as WT virus ( fractions 6 and 8 ) and showed the expected enrichment in prM protein compared to envelope protein ( Fig 7C and 7D , fraction 6 ) , but had less capsid protein compared to the WT virus . The virus isolated from SCD1 inhibitor-treated cells showed similar patterns of enrichment of the prM protein as the immature virus , and this enrichment was confined to the virus population in fraction 6 ( S6G and S6I Fig ) . The virus population that sedimented in fraction 8 had no detectable prM . The virus isolated from vehicle-treated ( DMSO ) cells that sedimented in fraction 6 had a similar protein content to WT virus , but there was a prominent population that sedimented in fraction 4 that did not have high GE ( S6H Fig ) . Based on these analyses , the fractions with the highest GE ( fractions 6 and 8 ) demonstrated the clearest differences between the four conditions . The virus isolated from SCD1 inhibitor-treated cells was similar to immature virus in prM content and was distinctly different from WT virus .
Previously we showed that lipid biosynthesis was upregulated in DENV-infected cells through the activation and relocalization of fatty acid synthase ( FAS ) , an enzyme critical to the production of palmitic and stearic acids that are structural components of complex lipids [14] . Here , we investigated the next step following FAS-catalyzed fatty acid production in the lipid biosynthesis pathway and demonstrated that desaturation of these fatty acids plays a critical role in the viral life cycle . Specifically , DENV2 infection resulted in upregulated monounsaturated fatty acid ( MUFA ) biosynthesis , catalyzed by SCD1 at early time points post-infection . Inhibition of this process impaired virion maturation and particle infectivity and stability . Two distinct scenarios were observed ( quantitatively and visually ) in this study: early during infection when low levels of viral RNA and protein were present , SCD1 transcripts , protein and enzymatic activity levels were elevated . However , late during infection , when high concentrations of viral RNA were present , SCD1 mRNA , protein and activity levels declined . As summarized in Fig 7F , we hypothesize that the metabolic environment required to progress from early to advanced infections changes and SCD1 activity could function as an enzyme that modulates these changes . For instance , at early stages of viral RNA replication , increased amount of fatty acids , specifically MUFAs , are generated ( through activity of SCD1 ) for construction of virus replication compartments . The unique architecture of positive-strand RNA virus replication compartments likely requires certain lipid components that can induce membrane curvature to provide extensive membrane contact sites and increased fluidity and leakiness to acquire substrates for genome replication [8 , 16 , 17 , 18] . Energy and lipid substrates are provided by increased glycolysis and activation of the pentose-phosphate pathway [31] . Increased SCD1 activity also results in the build-up of storage lipids , which reduces levels of β-oxidation . The high content of MUFAs in the membranes ( resulting from SCD1 activity ) also ensures appropriate assembly and maturation of the virus particles being released during early infection . However , during advanced infections , when viral RNA replication is at maximum efficiency , the cell has excess complex fatty acids that feed back to inhibit SCD1 expression [32] . By the time this occurs , the virus has already constructed its replication compartments and sites of assembly and does not require synthesis of new components by SCD1 . At this later time point , the focus is on producing a massive explosion of virus particles from the already assembled viral replication factories . This massive output of virus compromises quality control and increases the probability of producing mixed populations ( structurally diverse ) of virus particles . In these advanced infections , the decrease in MUFAs through the inhibition of SCD1 negatively influences virus particle quality , resulting in the production of higher ratios of non-infectious particles . Viral replication compartments are tightly coordinated with sites of virion assembly . For flaviviruses , only replicated viral RNA is packaged into newly assembled virions [33] . Therefore , the lipid membrane environment surrounding viral RNA replication complexes and that involved in virion assembly must be physically coordinated to transfer newly replicated viral genomes to sites of virion assembly [8] . Essentially , any lipid alterations that occur in these complex , virus-induced membranes must retain the capacity to support multiple functions . This is especially important since intracellular membranes become structural components of virus particles ( virion envelope ) that require assembly of a specific stoichiometry of viral glycoproteins . Conformational transitions that occur between these glycoproteins in lipid membranes define the state of maturation and infectivity of virus particles . Accordingly , we found that virion infectivity decreased when SCD1 was inhibited early in infection . At the given dose of the SCD1 inhibitor , DENV2 was able to replicate and produce particles with intact genomes , however a higher proportion of these particles were non-infectious . We and others have provided evidence that SCD1 inhibition results in lowering MUFAs in cellular membranes . Our observations suggest that the physical properties of the virus envelope influenced by the proportion of MUFAs may be critical for virus infectivity . DENV2 populations grown in the presence of the SCD1 inhibitor were found to be as stable to increased temperature as control virus , however , when the virus was subjected to multiple freeze-thaw cycles , it lost its infectivity faster than untreated virus . Inhibition of SCD1 results in changes in lipid content of the ER , resulting in an increase in saturated fatty acids [34 , 35] . At lower temperatures , saturated fatty acids pack together tightly , forming a rigid membrane [36 , 37 , 38] . Rigid membranes are not able to curve well and this may impair interactions with the trans-membrane glycoproteins [36 , 38] . Hence we observed a greater loss of functionality or infectivity of viral particles from SCD1-inhibited cells when transitioning from freezing to ambient temperature versus transitioning from higher temperatures , where lipids may achieve greater fluidity [39] . The defect in infectivity of virus particles from SCD1-inhibited cells apparently resulted from incomplete particle maturation . After acquiring its ER-derived lipid envelope with inserted , stoichiometrically assembled prM and E glycoprotein heterodimers , the newly formed virions traverse the Golgi apparatus to complete a maturation process prior to exiting the cell . Maturation includes pH-dependent conformational transitions between the prM and E glycoproteins embedded in the envelope that are pre-requisite for the cleavage of the prM protein to M protein in mature virions by a Golgi-resident furin protease . If the conformational changes are inhibited or incomplete , furin cannot access the cleavage site on prM to complete the maturation process . This results in prM retention on the virions; these virions are non-infectious . Based on our observations in this study , the virus requires MUFAs in its lipid envelope to allow the necessary structural transitions for virion maturation . SCD1 inhibition results in decreased proportions of MUFAs in intracellular membranes destined to become virion envelopes , resulting in impaired conformational shifts necessary for maturation and increased release of prM containing virions . We demonstrated that virions with higher prM content were defective in viral entry in subsequent infections . This establishes for the first time that lipids incorporated into the virion envelope are critical for particle infectivity . Enveloped viruses must acquire their lipid envelope from a specific organelle membrane [40 , 41 , 42 , 43] . Our data suggest that DENV2 assembly may occur preferentially and successfully at ER-membrane regions with a high content of MUFAs . Future research will explore the content of specific lipid species at sites of viral assembly and the consequences of their alterations . The lipid composition of the DENV2 virion is currently undetermined , thus the content of MUFAs in the infectious virion envelope is unknown . Studies of the lipid composition of other enveloped viruses have focused on lipid classes such as phospholipids and sphingolipids but have not looked at fatty acid content or saturation levels . However , it is clear that certain lipid species are enriched in viral envelopes and are functionally relevant for virion infectivity [44 , 45 , 46 , 47 , 48] . This study provides insights on how fatty acid biosynthesis , specifically unsaturated fatty acids impact flavivirus genome replication and assembly of infectious viral particles in human cells . It sheds light on host metabolic pathways that enhance viral replication success and provides an unique avenue for antiviral intervention .
The cell lines used were as follows: Human embryonic Lung epithelial cells ( HEL 299 ) ( ATCC CCL-137 , male ) , adenocarcinomic human alveolar basal epithelial cells ( A549 ) ( ATCC CRM-CCL-185 , male ) , C636 ( ATCC CRL-1660 , larva , sex unknown ) , Clone 15 ( ATCC CCL-10 ) of the Baby Hampster Kidney Clone 21 cells ( BHK-21 ) , and Human hepatoma ( Huh7 ) ( From Dr . Charles Rice , sex unknown , [49] . Huh7 , HEL and A549 cells were maintained in Dulbeccos Modified Eagle Medium ( DMEM ) ( Gibco , LifeTech ) , while BHK and C636 were maintained in Minimum Essential Media ( MEM ) ( Gibco , LifeTech ) , both supplemented with 0 . 1 mM nonessential amino acids , and 0 . 1 mM L- glutamine , and 10% Fetal Bovine Serum ( Atlas Biologicals ) at 37°C with 5% CO2 . The virus strains used are as follows: DENV1 ( 16007 ) [50 , 51] , DENV2 ( 16681 ) [50 , 52] , DENV3 ( 16562 ) [53 , 54 , 55] , DENV4 ( 1036 ) [53] , YFV 17D [56] , and KUNV [57] these viruses were passaged in C6/36 cells . Additionally , ZIKA ( PRVABC59 ) [58] was passaged in African Green Monkey Kidney Epithilial cells from the Vero lineage ( Vero ) ( ATCC CRL-1586 ) . A DENV luciferase reporter replicon containing only the nonstructural proteins was also used [14] . Virus titers were determined by plaque assay on BHK cells as described previously [59] . Infection of cells was carried out at room temperature for one hour to allow virus to adhere to cells . Virus was then removed , cells rinsed with 1XPBS , overlaid with the indicated media and transferred to the 37°C incubator for required periods of time . RNA was extracted from cells using Trizol ( ThermoFisher ) and from virus in supernatant using Trizol LS ( ThermoFisher ) . A one-step qRT-PCR kit with SYBR green from Agilent was used . Reactions were set up according to the manufacturer’s protocol and run on a LightCycler 96 real-time PCR machine ( Roche ) . The cycling parameters were: 20 mins at 50°C for reverse transcription , then 5 mins at 95°C followed by 45 two-step cycles of 95°C for 5 seconds and 60°C for 60 seconds . This was followed by a melt curve starting at 65°C and ending at 97°C . DENV primers [60] were used to quantify viral RNA copies in the supernatant as well as in cells . A standard curve of in vitro transcribed viral RNA from a DENV2 cDNA subclone was generated and used to quantify the genome copies in the supernatant [61] . Copies of viral RNA in the cell as well as copies of SCD mRNA transcripts were both normalized to glyceraldehyde 3-phosphate dehydrogenase ( GAPDH ) RNA using the delta delta ct method [62] . For this method: the fold change in gene expression = 2^ ( - ( Infected samples ( ( Ct value of gene of interest ) – ( Ct value of control gene ) ) ) – ( Uninfected samples ( ( Ct value of gene of interest ) – ( Ct value of control gene ) ) ) ) . The Ct values were generated from the Light Cycler software and the gene of interest was either SCD1 or DENV and the control gene was GAPDH . Cells were transfected with pooled siRNAs ( S1 Table , Fig 1A ) or single siRNAs ( S2 Table , Fig 1B and 1C ) using RNAiMax ( Invitrogen ) similar to previous experiments [14] and allowed to incubate for 48hr . Cells were then infected with DENV2 , or collected for knockdown confirmation or cytotoxicity tests ( described below ) . Virus was collected and titrated with plaque assays . To confirm knockdown of mRNA transcripts RNA was extracted and qRT-PCR performed to measure SCD1 levels relative to GAPDH in SCD1 siRNA treated samples and compared to IRR treated samples using the delta delta ct method [62] described above . To confirm knockdown of protein levels cellular protein was collected in radioimmunoprecipitation assay buffer ( RIPA ) [150 mM sodium chloride , 1 . 0% NP-40 , 0 . 5% sodium deoxycholate , 0 . 1% SDS ( sodium dodecyl sulfate ) , 50 mM Tris , pH 8 . 0] . and separated by electrophoresis on an SDS-Page gel . They were then transferred to a nitrocellulose membrane ( Bio-Rad ) , blocked in 5% milk and probed with SCD N-20 ( Santa Cruz Biotechnology ) and β-actin ( Cell Signaling Technology ) . Secondary antibodies were IRDye 680RD and IRDye 800CW ( Li-Cor ) . The blot was imaged on an Odyssey IR Imaging system ( Li-Cor ) and quantification of the signals measured with Image Studio 5 . 2 ( Li-Cor ) . The inhibitors used were A939572 ( the SCD inhibitor , MedChem Express ) , C75 ( Cayman Chemicals ) and Lovastatin ( Sigma-Aldrich ) . Each was diluted in DMSO , added to DMEM and filtered through a 0 . 2μM filter before being added to cells . Inhibitors were added to cells following virus attachment . Oleic acid was acquired from Sigma and came dissolved in bovine serum albumin ( BSA ) at 200mM . It was further diluted in 1% fatty acid free BSA ( Gold Biotechnology ) in 1x phosphate buffered saline ( PBS ) to 50 μM . Huh7 cells were infected with virus as described above and overlaid with the indicated treatments diluted in DMEM . Supernatants were collected at 24hr and plaque assays performed . Cytotoxicity was measured with alamar blue ( ThermoFisher ) diluted 1:10 in DMEM incubated on cells for 2–4 hr and read on a Victor 1420 Multilabel plate reader ( Perkin Elmer ) with excitation at 560 nM and emission at 590 nM . Virus that was used for re-infection and entry assays was grown in Huh7 cells at an MOI of 3 in 10 μM SCD1 inhibitor or DMSO ( 0 . 02% ) . The virus samples were titrated by plaque assay and diluted to equivalent titers . The SCD1 inhibitor was added to the DMSO sample and the virus samples ( equivalent pfu/ml ) were allowed to attach to new cells for 1 hr at room temperature before being removed . The cells were then washed with 1xPBS and incubated with DMEM + 2% FBS for the indicated times at 37°C . For siRNA confirmation of these results , virus was grown with siRNA treatment as described above . Virus was collected at 48hr , titrated and equal PFU were used to infect new Huh7 cells at MOI = 0 . 3 and overlaid with DMEM + 2% FBS . After 48hr of replication virus supernatant was collected and titrated . Each virus was diluted to 1000 pfu/ml in DMEM . The SCD inhibitor was added to the virus that was grown in DMSO ( 0 . 02% ) at 10 μM and then filtered . Virus was allowed to attach to BHK cells in 6 well plates for 2 hr at 4°C . The virus was then aspirated and cells were rinsed with 1XPBS to remove any unbound virus . The cells were then overlaid with MEM with 10% FBS and transferred to 37°C . At given time points after the temperature shift , cells were removed from the incubator rinsed with 1XPBS and treated with acid-glycine ( 8 g of NaCl , 0 . 38 g of KCl , 0 . 1 g of MgCl2 6H2O , 0 . 1 g of CaCl2 2H2O , and 7 . 5g of glycine/L , pH adjusted to 3 with HCl ) for 1 minute at room temperature to inactivate any extracellular virus . The cells were again rinsed with 1XPBS and overlaid with 1% agarose and MEM with 5% FBS , plaques were counted at 6 days . Cells were grown on a sterilized cover slip and maintained in 10% DMEM . Cells were infected with DENV2 ( M0I = 100 ) or mock infected ( 1XPBS ) . Cells were fixed in ice-cold methanol ( Fisher Chemical ) at room temperature and permeabilized with 0 . 1% TritonX ( Fisher Chemical ) in 1XPBS with 1% BSA ( GoldBiotech ) at room temperature and blocked with 0 . 01% TritonX in 1XPBS with 1% BSA overnight at 4°C . Cells were then probed with the indicated primary antibodies including dsRNA ( English and Scientific Consulting Bt . ) , NS3 ( gifted by Richard Kuhn , Purdue University ) and SCD N-20 ( SantaCruz Biotechnology ) . Secondary antibodies were Alexa-fluor 488 or 647 . The coverslips were fixed to slides with FluoroSave ( Calbiochem ) and imaged on an Olympus inverted IX81 FV1000 ( Olympus ) confocal laser scanning microscope with a 100x oil objective using FV10-ASW 4 . 2 ( Olympus ) . Digital images were processed with Volocity 6 . 3 ( Perkin Elmer ) and colocalization coefficients calculated by encircling each individual cell and using the measurement function with internal thresholds . Values represent averages of at least 30 cells from different image frames of the same slide . Huh7 cells were infected with DENV2 ( MOI = 0 . 5 ) and overlaid with DMEM . At the indicated time-point the inhibitor ( 10μM ) or vehicle were added to cells . Virus supernatants were collected after 48hr and titrated . To look at the pre-infection and attachment phases , we added the inhibitor to Huh7 cells and then infected cells after 12hr of treatment . After infection we washed cells and overlaid with either the inhibitor , vehicle control or DMEM . We also added the inhibitor or vehicle control to the virus inoculum , allowed attachment to occur for 1 hr , then washed cells and either added new inhibitor or removed it to determine an impact on the attachment stage . To confirm findings with the siRNA we infected Huh7 cells with DENV2 ( MOI = 0 . 1 ) and overlaid with DMEM + 2% FBS . After 24hr of replication , the cells were transfected with the indicated siRNAs . Viral supernatant was then collected and titrated at 48 and 72hr post infection . The indicated cell samples were collected at given time points and prepared in order to preserve enzymatic activity [63] . Briefly , cells were washed in the wash buffer ( 35 mM Hepes , pH 7 . 4 , 146 mM NaCl , 11 mM glucose ) 3 times , then incubated in a hypotonic solution ( 20 mM Hepes pH 7 . 4 , 10 mM KCl , 1 . 5 mM MgOAc , 1mM DTT ) for 20 minutes to allow the cells to swell . Then they were passed through the dounce homogenizer 25 times to break apart the membranes . A post lysis buffer ( 20 mM Hepes pH 7 . 4 , 120 mM KOAc , 4mM MgOAc , 5 mM DTT ) was added . Nuclei were spun down at 1000xg for 5 minutes at 4°C and the cytoplasmic extract was flash frozen in liquid N2 . Protein content was measured by BCA ( Pierce ) and equal protein content was used for activity assays . Activity assays were performed similar to previous studies [64] . Briefly , 1 . 5 mg/ml protein was incubated with 0 . 01 μCi of stearoyl [1-14C]-coA ( American Radiolabeled Chemicals , 55mCi/mM ) at 37°C for 5 minutes . The reactions were stopped by adding 150 μL of methanolic HCl ( Sigma-Aldrich ) for 1 hour at 72°C , which generated fatty acid methyl esters that were extracted from the samples in 1 mL of chloroform . One aliquot of these samples was measured on a Beckman LS 6500 liquid scintillation counter ( Beckman ) and equal counts were spotted on a 0 . 5% AgNO3 impregnated normal phase thin layer chromatography plate . Fatty acids were separated by a mobile phase of hexane: acetone ( 50:1 ) . Plates were exposed on a phosphoimager screen and scanned on a Typhoon Trio 9400 ( GE Healthcare ) . Pixel intensity was measured with ImageQuant TL ( GE Health Care ) software . Standards were sprayed with 5% phosphomolibic acid in 100% ethanol and heated to visualize fatty acids . Huh7 cells were infected with DENV2 ( MOI = 10 ) , a UV-inactivated virus or mock-infected . Cells were collected at the indicated time points and metabolites were extracted from an equal number of cells per sample . A mixture of 2∶1 chloroform∶methanol , 0 . 1% acetic acid and 0 . 01% butylated hydroxy toluene ( BHT ) were added to the cell suspension in ammonium bicarbonate to generate a 4∶1 ratio of organic solvent to cells . The non-polar phase was removed , dried down under N2 stream resuspended in 75 μl of ice-cold methanol and vortexed for 10 s . The samples were then centrifuged at 13 , 400× g for 5 min to remove any particulates . The samples were run on a LTQ Orbitrap XL instrument ( Thermo Scientific , Waltham , MA ) . It was coupled to an Agilent 1100 series LC ( Agilent Technologies , Santa Clara , CA ) . An Xterra C18 column ( Waters Corp . , Milford , MA ) was used in reverse phase . Solvent A consisted of water + 10mM ammonium acetate + 0 . 1% formic acid . Solvent B was acetonitrile/isopropyl alcohol ( 50/ 50 v/v ) + 10mM ammonium acetate + 0 . 1% formic acid . The flow rate was 300 μL/minute . A sample volume of 10 μL was loaded onto the column . The gradient was as follows: time 0 minutes , 35% B; time 10 minutes , 80% B; time 20 minutes , 100% B; time 32 minutes , 100% B; time 35 minutes , 35% B; time 40 minutes 35% B . The LC-MS analysis was run twice , with negative polarity ESI . The acquired data were evaluated with Thermo XCalibur software ( version 2 . 1 . 0 ) . The raw data was converted to mzXML format with msConvert ( 46 ) . Peak picking was accomplished with the XCMS package using centWave ( 47–49 ) and the IPO package to optimize parameters ( 51 ) . Intensities for peaks were determined and normalized using the median fold change method ( 52 , 53 ) . Features were annotated with the mummichog software version 2 . 0 ( 21 ) . Virus purification and identification of prM were carried out similar to previous studies [30 , 65] . Briefly , Huh7 cells were infected with DENV ( MOI = 3 ) and left untreated , treated with 10 μM of the SCD inhibitor or DMSO similar to other experiments presented here . The supernatant was collected at the indicated time points and was replaced with fresh DMEM plus the indicated treatment . At 24hr after infection 20mM NH4Cl was added to generate the immature virus samples . The supernatants from the SCD inhibitor and DMSO samples were collected at 24hr . Cellular debris was removed from the supernatant and the virus was run through a 22% sucrose cushion at 32 , 000 xg for 2 hr at 4°C in a Sorvell WX ultracentrifuge ( ThermoFisher Scientific ) . The pellet was loaded onto a potassium-tartrate step gradient consisting of 10 , 15 , 20 , 25 , 30 and 35% potassium-tartrate in TNE buffer ( 50 mM Tris–HCl ( pH 7 . 4 ) , 100 mM NaCl , 0 . 1 mM EDTA ) and spun at 32 , 000 xg for 2 hr at 4°C in a Sorvell WX ultracentrifuge ( ThermoFisher Scientific ) . Ten different fractions were collected from each gradient and used to titrate the virus and quantify viral genomes similar to other experiments . At 72 hr virus from all 4 samples were collected , cellular debris removed and virus was PEG precipitated at 4°C . The virus was then run through a 22% sucrose cushion and loaded onto a potassium-tartrate gradient same as mentioned above . Visible band of virus were noted in the gradients and the 4 fractions were collected and buffer exchanged with TNE buffer . Viral genome copies were measured with qRT-PCR . Protein content was measure with BCA ( Pierce/ThermoFisher ) and equal amounts of protein were loaded on an SDS-page gel and transferred to a PVDF membrane and probed for prM , capsid and E . Blots were imaged on an Odyssey and densitometric measurements were made . Western blots for each fraction ( i . e . : fraction 6 ) had equal total protein loaded for the four conditions . Conditions between fractions cannot be compared due to differences in protein sedimentation between fractions ( i . e . : total protein in fraction 6 cannot be compared to total protein in fraction 8 ) . Results were expressed as mean values with standard deviation . The statistical details are noted in the figures and/ or in the corresponding figure legends . Statistical significance was primarily determined using either an unpaired Student's t-test or a one or two way Analysis of Variance ( ANOVA ) with a Bonferroni , Dunnets or Tukey’s multiple-comparison depending on the experimental design in the GraphPad Prism version 7 . 00 for Mac OS x ( GraphPad Software , La Jolla California USA ) . Drug inhibition studies were analyzed with a non-linear regression using GraphPad Prism version 7 . 00 for Mac OS x ( GraphPad Software , La Jolla California USA ) to calculate an IC50 . The assay for viral release ( Fig 5A ) was analyzed with a 3-way ANOVA and included an interaction term for each . This was performed in R studio version 1 . 0 . 136 [66] . Finally , the nonlinear regression models for the thermostability of the viruses ( S6A Fig ) was also performed in R studio version 1 . 0 . 136 [66] . The model used was y = a+ ( ( b-a ) / ( ( 1+10^ ( ( c-temp ) *d ) ) ) . An F-test was used to determine the difference between the curves with the formula F = ( ( SStotal-SS pooled ) / ( ( m+1 ) ( K-1 ) ) ) / ( SSpooled/dfpooled ) . Each study shown is representative of at least two independent experiments . Additional key resources are included in S2 Table . | Dengue viruses are aggressive mosquito-borne viruses causing over 350 million infections annually . There are no antivirals to combat infection and the only vaccine available is suboptimal . Here , we have investigated how these viruses hijack critical metabolic pathways in cells to convert the host environment to support their replicative needs . Specifically , we have identified an enzyme , Stearoyl CoA desaturase ( SCD1 ) , that is critical to virus replication and infectious particle formation . SCD1 is responsible for making unsaturated lipids in the cell . When these lipids are incorporated into cellular membranes , they allow these membranes to be more fluid and increase their ability to exchange nutrients between cellular compartments . Inhibition of SCD1 disrupts viral genome replication and blocks structural rearrangements in the virus particles that are required to make them infectious . As a result , SCD1 inhibition causes non-infectious particles to be produced . Additionally , we show that SCD1 enzymatic activity is critical at early stages of virus replication and is shut off during late stages of virus replication . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"oleic",
"acid",
"gene",
"regulation",
"rna",
"extraction",
"microbiology",
"viral",
"structure",
"viruses",
"rna",
"viruses",
"extraction",
"techniques",
"research",
"and",
"analysis",
"methods",
"small",
"interfering",
"rnas",
"lipids",
"gene",
"expression",
"viral"... | 2018 | Stearoly-CoA desaturase 1 differentiates early and advanced dengue virus infections and determines virus particle infectivity |
Ion homeostasis is essential for plant growth and environmental adaptation , and maintaining ion homeostasis requires the precise regulation of various ion transporters , as well as correct root patterning . However , the mechanisms underlying these processes remain largely elusive . Here , we reported that a choline transporter gene , CTL1 , controls ionome homeostasis by regulating the secretory trafficking of proteins required for plasmodesmata ( PD ) development , as well as the transport of some ion transporters . Map-based cloning studies revealed that CTL1 mutations alter the ion profile of Arabidopsis thaliana . We found that the phenotypes associated with these mutations are caused by a combination of PD defects and ion transporter misregulation . We also established that CTL1 is involved in regulating vesicle trafficking and is thus required for the trafficking of proteins essential for ion transport and PD development . Characterizing choline transporter-like 1 ( CTL1 ) as a new regulator of protein sorting may enable researchers to understand not only ion homeostasis in plants but also vesicle trafficking in general .
The root is highly specified and well organized for the uptake and transport of water and mineral nutrients from the highly heterogeneous soil environment . Plants have evolved complex gene networks to fine tune the activities of different transporters , as well as the functions of different cell wall structures and processes associated with cell-to-cell communication , to match the nutrients supplied by the soil with the normal growth- and development-related needs of their cells . These processes are regulated at both the transcriptional and posttranscriptional levels . At the transcriptional level , the expression of transporters and genes that function in root patterning and development is orchestrated by a series of transcription factors . Ectopic expression of a transporter may result in disorders of ion homeostasis . For example , enhanced expression of A . thaliana high affinity K+ transporter 1 ( AtHKT1 ) , specifically in the stele leads to reductions in shoot sodium ion ( Na+ ) concentrations; however , constitutive overexpression of AtHKT1 throughout the plant leads to significant increases in shoot Na+ concentrations [1] . At the posttranscriptional level , transporters can be regulated by processes such as protein modification , degradation , and trafficking . For example , the subcellular localization of iron regulated transporter 1 ( IRT1 ) , which transports bivalent cations including iron ( Fe2+ ) , zinc ( Zn2+ ) , manganese ( Mn2+ ) , cobalt ( Co2+ ) , and cadmium ( Cd2+ ) ions , is mediated by monoubiquitin- and clathrin-dependent endocytosis involving the IRT1 degradation factor1 RING-type ( IDF1 RING ) E3 ligases [2 , 3] . FYVE domain protein required for endosomal sorting 1 ( FYVE1 ) , a phosphatidylinositol-3-phosphate ( PI3P ) -binding protein , interacts with IRT1 and mediates the recycling of IRT1 from the endosome back to the plasma membrane ( PM ) [4] , suggesting that membrane lipids and lipid-binding proteins play important roles in protein trafficking and ion homeostasis . Phosphatidylinositol ( PI ) sphingolipids and phosphatidic acids ( PAs ) have all been found to participate in the regulation of vesicle trafficking [5 , 6] . However , as the most abundant phospholipids in the PM of eukaryotic cells [7 , 8] , phosphatidylcholines ( PCs ) have not been shown to be involved in either cargo trafficking or protein subcellular localization , although they have been found to play a crucial role in signal transduction [9] . These important phospholipids incorporate choline as a head group and can be used as a substrate to produce choline in reactions catalyzed by phospholipase D ( PLD ) . In animals , choline is an essential nutrient , because it serves not only as a component of membrane lipids and lipoprotein but also serves as a precursor of many essential molecules such as the neurotransmitter acetylcholine and the osmoregulator betaine [10] . The choline transporter is required for choline homeostasis , as well as secretion of acetylcholine [11–13] . Choline is also produced in plants , but its biological function was rarely studied . A recent study showed that the trans-Golgi network ( TGN ) -localized choline transporter choline transporter-like 1 ( CTL1 ) is involved in the formation of the phloem sieve plate and sieve pore; however , the mechanism underlying its role in these processes remains unknown [14] . Here , we identified a new CTL1 allele by screening ethyl methanesulfonate ( EMS ) -mutagenized plants to identify those with an altered leaf ionome . Our results showed that CTL1 is required for vesicle trafficking and that CTL1 mutations result in the disruption of plasmodesmata ( PD ) and the normal localization of PM proteins , including ion transporters and PD callose-binding ( PDCB ) proteins .
A previous study screened fast neutron mutagenized A . thaliana plants for mutants with an altered leaf ionome to help unravel the gene networks regulating plant ion homeostasis [15] , several of which have since been well characterized [16 , 17] . Using the same strategy , we isolated a new mutant that displays significantly increased leaf concentrations of Na and decreased leaf concentrations of Mn , Fe , Zn , and molybdenum ( Mo ) when grown in artificial soil ( Fig 1A , S1 Table ) . We named this mutant significant ionome changes 1 ( sic1 ) . The leaf ionomic phenotypes of sic1 observed in plants grown in hydroponics were similar to those observed in plants grown in soil ( Fig 1B ) ; however , the mutant displayed significant alterations in the concentrations of more elements—as it displayed increases in lithium ( Li ) , boron ( B ) , and sulfur ( S ) concentrations and decreases in phosphorus ( P ) , potassium ( K ) , calcium ( Ca ) , Co , nickel ( Ni ) , and copper ( Cu ) concentrations ( S2 Table ) —compared with the wild-type plant . The hydroponically cultured root of sic1 also displayed impaired ion homeostasis , as well as ionomic changes similar to those observed in sic1 leaves ( Fig 1C , S3 Table ) , suggesting that the ionomic changes characteristic of sic1 may result from defects in ion uptake or short-distance transport in root . In addition to exhibiting ionomics phenotypic abnormalities , sic1 also exhibited clear developmental defects . Specifically , sic1 displayed defects in leaf and root elongation ( Fig 1D , S1 Fig ) and had fewer leaves than its wild-type counterpart ( S1 Fig ) ; however , there was no significant difference in flowering time between the two plants . To test the hypothesis that the ionomic phenotypes of sic1 resulted from defects in ion uptake or short-distance transport by the roots , we performed reciprocal grafting experiments involving wild-type and sic1 plants . Principle component analysis ( PCA ) showed that the leaf ionomic phenotypes of grafted plants with a sic1 scion and wild-type root were similar to those of nongrafted and self-grafted wild-type plants . However , grafted plants with a wild-type scion and sic1 root had a leaf ionome that was indistinguishable from that of sic1 ( Fig 1E , S4 Table ) . Such evidence indicates that the leaf ionomic phenotype is driven by the root . Interestingly , the shoot developmental phenotypes of sic1 were also partially driven by the root ( S2 Fig ) . These data suggested that sic1 root function is defective . Consistent with this finding , we also found that the cells in the root of sic1 were organized in an irregular manner , probably as a result of irregular cell division ( Fig 1F ) . To identify the causal gene of sic1 , we constructed an F2 mapping population derived from a cross between sic1 and Landsberg erecta-0 ( Ler-0 ) . The ionomic and developmental phenotypes of the F2 individuals co-segregated with wild-type and mutant individuals at a segregation ratio of 3:1 ( X2 = 0 . 56 < X20 . 05 , 1 = 3 . 84 ) , demonstrating that the ionomic and developmental phenotypes of sic1 were controlled by the same single recessive locus . Based on the analysis of 2 , 030 F2 individuals , we mapped the casual locus to a 100-kb region on chromosome 3 ( Fig 2A ) . We sequenced the entire candidate region and identified a single G-to-A mutation on the fourth exon of gene At3g15380 . This mutation causes a predicted 247Glycine-to-247Glutamate substitution in a conserved domain of this protein ( Fig 2B and S3 Fig ) , which suggests that At3g15380 is a good candidate for SIC1 . To confirm the above finding , we isolated a homozygous transfer DNA ( T-DNA ) insertion mutant known as cher1-4 [14] , which featured a T-DNA sequence that had been inserted into the first intron of At3g15380 ( Fig 2B , S4 Fig ) . The T-DNA insertion mutant cher1-4 displayed ionomic and growth phenotypes indistinguishable from those of sic1 ( Fig 2C , S5 Fig and S5 Table ) . In addition , the F1 progenies of a cross between sic1 and cher1-4 displayed phenotypes identical to those of sic1 and sic1-2 ( Fig 2C , S5 Fig and S5 Table ) , indicating that sic1 and cher1-4 are two alleles of the same locus . In addition , we introduced a functional pSIC1::SIC1-GFP construct into sic1 and found that this construct could fully complement both the ionomic and visible phenotypes of sic1 ( Fig 2D , 2E , S5 Fig and S6 and S7 Tables ) . These data indicated that the mutation identified in At3g15380 is responsible for both the ionomic and the growth defects of sic1 . Sequence analysis showed that sic1 is a new allele of the recently identified CTL1 gene , which encodes a choline transporter localized at the TGN and nascent cell plates [14] , and the mutation occurs at the second transmembrane domain of the protein ( S3 Fig ) . The SIC1 gene was therefore renamed CTL1 . CTL1 is expressed in both root and shoot , as revealed by quantitative real-time PCR ( qRT-PCR ) and the green fluorescent protein ( GFP ) signals in pCTL1::CTL1-GFP transgenic plants ( S6 Fig ) . In root , CTL1-GFP was observed in all cell types of the root tip , whereas it is mainly distributed in the stele of the maturation zone ( S6C Fig ) . It has been reported that CTL1 is required for sieve plate and sieve pore formation [14] . As the sieve pore is a special type of PD that plays important roles in ion transportation , we hypothesized that the disorder of ion homeostasis characteristic of sic1 may be caused by defects in PD . We examined the PD morphology of root cortex cells , wherein PD plays an important role in ion transportation . Using a transmission electron microscope , we observed that most PDs in sic1 are blocked or shrunken , whereas those in Columbia-0 ( Col-0 ) exhibit typical morphology ( Fig 3A ) . A similar result was also observed in a recent study [18] . To examine PD function in sic1 further , we examined the localization of two PD markers that play an important role in PD development [19] , namely , PDCB1 and PDCB2 , by introducing two constructs , pPDCB1::PDCB1-GFP and pPDCB2::PDCB2-GFP , into Col-0 and sic1 . We analyzed five independent transgenic lines for each genotype and construct , and observed well consistent results . Both PDCB1-GFP and PDCB2-GFP were found to be localized predominantly on the cell wall in a spotty manner in wild-type roots , and only a small amount of intracellular fluorescence was observed ( Fig 3B and 3C ) , findings consistent with those of previous studies [19] . In contrast , in the mutant root , either PDCB1-GFP or PDCB2-GFP was localized mostly , if not entirely , in some intracellular compartments , as only a weak PDCB1-GFP signal was observed on the cell wall ( Fig 3B and 3C ) . The intracellular aggregation of PDCB proteins not only confirmed that PDs are defective in sic1 but also indicated that PDCB could not localize properly in the mutant . Given that PDs are essential for mineral element transportation and that Casparian strips block the apoplast pathway , it is plausible that PD defects may result in disorders of ion homeostasis . However , we noticed that the levels of some elements were decreased , whereas those of other elements were increased in sic1 ( Fig 1A and 1C , S1–S3 Tables ) , indicating that impaired PD-mediated symplastic transportation is not the only cause of the defects in ion homeostasis in sic1 . A previous study reported that shrunken PD apertures reduce cell-to-cell communication and disrupt processes such as protein and microRNA ( miRNA ) intercellular translocation and that such changes systematically disturb organism patterning and cell identities [20 , 21] . Given that sic1 mutants display a root cell-patterning disorder phenotype ( Fig 1F ) , we wondered whether the intercellular movement of shoot root ( SHR ) , a key transcription factor that moves from the stele to the endodermis to regulate root development , was affected by the shrunken PDs . To validate this hypothesis , we introduced a pSHR::SHR-GFP construct into Col-0 and sic1 , and obtained six to ten independent transgenic lines for each genotype for analysis . Confocal microscopic analysis showed that SHR-GFP can move from the stele to the endodermis correctly in all Col-0 transgenic lines , as previously reported [22] . However , PD-mediated intercellular movement was defective in all sic1 lines , as SHR-GFP was expressed normally in all stele cells but was hardly visible or even vanished in most endodermal cells ( Fig 3D ) . As SHR plays a central role in the initialization and architecture of the cortex , endodermis , and stele [22–24] , the reduced movement of SHR noted above may be an important cause of the disordered root patterning characteristic of sic1 and may result in the abnormal expression of transporters in the wrong cell types . Given that cell-type–specific transporter expression is important for ion homeostasis , we investigated whether the ionomic phenotype of sic1 was also a result of the ectopic expression of a series of transporters . HMA4 is a Zn2+ transporter expressed in the pericycle and xylem parenchyma of roots and is responsible for transporting Zn2+ from the symplast to the xylem [25] . As sic1 exhibits defective Zn2+ transport , we assessed the expression patterns of HMA4 in sic1 and Col-0 . We expressed pHMA4::HMA4-GFP in sic1 and introduced the construct into Col-0 by crossing . As previously reported [25 , 26] , HMA4-GFP was clearly present in the pericycle and xylem parenchyma of Col-0 ( Fig 4A ) . However , HMA4-GFP was also found to be ectopically expressed in the sic1 epidermis , as well as the root hairs of the mature region of the root ( Fig 4A ) . This result was further confirmed by the observation of β-glucuronidase ( GUS ) signals in 5–8 independent transgenic lines of Col-0 and sic1-expressing pHMA4::GUS . The GUS signal was only present in root steles of Col-0 lines , in which it was also observable in epidermis and root hair cells in addition to the steles of sic1 lines ( S9 Fig ) . Because HMA4 mediates Zn efflux , we surmised that ectopically expressing HMA4 in the epidermis of sic1 would reduce Zn uptake in the root . Interestingly , we found that the total expression level of HMA4 , as determined by qRT-PCR , was downregulated in sic1 ( Fig 4C ) , suggesting that Zn loading into the xylem may also be affected in this mutant . To confirm this hypothesis , we stained the roots of Col-0 and sic1 plants with the membrane-permeant Zn2+ fluorescent sensor Zinpyr-1 , and found that the distribution of Zn was significantly altered in sic1 roots compared with Col-0 roots . In contrast to Col-0 roots , which exhibited Zn accumulation mainly in the stele , sic1 roots exhibited Zn accumulation mainly in the cortex and endodermal cells ( Fig 4E and 4F ) , findings fully consistent with those of the experiments in which disturbances in the expression patterns and reductions in the expression levels of HMA4 in sic1 mutants were noted . HKT1 functions as a Na+ transporter in xylem parenchymal cells and is responsible for retrieving Na from the xylem transpiration stream [27 , 28] . We attempted to express pHKT1::HKT1-GFP in Col-0 and sic1 to determine whether , similar to HMA4 , HKT1 is ectopically expressed in epidermal cells . However , we failed to observe a GFP signal in either Col-0 or sic1 . Alternatively , we introduced a pHKT1::GUS construct into Col-0 and sic1 mutants . The GUS staining results from six independent lines of each genotype showed that HKT1 is expressed in the stele of Col-0 roots , as previously reported [29] , but is expressed in the epidermis and root hair cells of sic1 ( Fig 4B ) . Consistent with these findings , we found that HKT1 was down-regulated in sic1 ( Fig 4D ) . These results indicated that HKT1 was also expressed ectopically in sic1 . It has been shown that ectopic expression of HKT1 results in increased accumulation of Na in both the roots and the shoot , as it mediates Na influx [1] . We therefore assessed the distribution of Na in Col-0 and sic1 seedlings by staining CoroNa™ Green Sodium Indicator ( CoroNa ) , a membrane-permeant Na fluorescent indicator . We found that Na levels are apparently higher in the root epidermal and cortical cells of sic1 than in those of Col-0 ( Fig 4G ) . Given that sic1 had low concentrations of most of the divalent cations , including Fe2+ , Mn2+ , and Zn2+ , in either the leaves or the roots , we assessed the expression pattern of natural resistance-associated macrophage protein 1 ( NRAMP1 ) , a divalent cation transporter responsible for Fe2+ and Mn2+ uptake and long-distance transportation [30 , 31] . We expressed a pNRAMP1::NRAMP1-GFP construct in Col-0 and introduced the construct into sic1 by crossing sic1 with transgenic Col-0 plant . Surprisingly , we did not observe alterations in the expression pattern of NRAMP1 in sic1 , as both Col-0 and sic1 expressed NRAMP1 in the epidermis , cortex , and xylem ( Fig 5A–5D ) , findings consistent with those of a previous report [30] . However , we found that the NRAMP1-GFP protein in sic1 was intracellularly retained in the epidermis and root hair cells , whereas in Col-0 it was localized predominantly on the PM ( Fig 5B , 5D , 5F and 5H ) . Statistical analysis showed that the PM NRAMP1-GFP signal intensity of Col-0 was 2-fold higher than that of sic1 ( Fig 5I ) . In addition , the intensity of the intracellular NRAMP1-GFP signal in sic1 was 4-fold higher than that in Col-0 ( Fig 5J ) , and that the proportion of PM-localized NRAMP1-GFP was 82% lower in sic1 than in Col-0 ( Fig 5K ) . As NRAMP1 functions in Mn2+ and Fe2+ uptake and transportation , defects in NRAMP1 trafficking may explain the reduced Mn and Fe concentrations observed in both leaves and roots of sic1 . The recycling of iron transporter IRT was previously reported to be mediated by vesicle trafficking [4] . We failed to examine the subcellular localization of IRT1 in sic1 . But interestingly , we observed that IRT1 expression was up-regulated up to 9-fold in sic1 ( S10A Fig ) . Similar to that , the expression of Arabidopsis H +-ATPase 2 ( AHA2 ) , another key gene for iron uptake , is also extremely upregulated in roots of the sic1 mutant ( S5B Fig ) . These data might reflect that the recycling of IRT1 probably also requires CTL1 , and the up-regulation of these genes is just a feedback of mislocalization of their encoding proteins . A previous study reported that CTL1 localizes to the TGN , the hub of vesicle trafficking [14] . Consistent with this finding , we found that CTL1 co-localized with N- ( 3-Triethylammoniumpropyl ) -4- ( 6- ( 4- ( diethylamino ) phenyl ) hexatrienyl ) pyridinium dibromide ( FM4-64 ) , an endocytosis tracer ( S7 Fig ) . This evidence , as well as the observations that PDCB proteins and NRAMP1 are intracellularly aggregated in sic1 , suggests that CTL1 is probably involved in vesicle trafficking . We stained 6-day-old Col-0 and sic1 seedlings with FM4-64 to observe endocytosis . Our results showed that FM4-64 internalization was clearly observed after 30 minutes of staining in Col-0 but after 60 minutes of staining in sic1 ( Fig 6A and 6B ) , indicating that endocytosis was strongly repressed in sic1 . Recently , it was reported that the localization of NRAMP1 to the PM was mediated by vesicle trafficking [32] . Considering the intracellular aggregation of NRAMP1 and the suppression of endocytosis in sic1 , we wondered if CTL1 is involved in regulating the recycling of NRAMP1 . To examine this , we treated the transgenic lines expressing pNRAMP1::NRAMP1-GFP in both Col-0 and sic1 background with a vesicle trafficking inhibitor brefeldin A ( BFA ) [33] . BFA inhibits the function of ADP-ribosylation factor GTPases ( ARF GTPases ) by interacting with their associated guanine nucleotide exchange factors ( GEFs ) and thereby results in membranous aggregates known as BFA compartments [33] . With BFA treatment for 1 . 5 hours , the NRAMP1-GFP aggregation in BFA compartments was observed in the epidermal cells of both Col-0 and sic1 ( Fig 6C ) , supporting that subcellular localization of NRAPM1 is mediated by vesicle trafficking . However , after 2 hours BFA washing out , NRAMP1-GFP aggregation in the BFA compartments only remained in 3 . 0% of Col-0 epidermal cells but that number in sic1 is 58 . 3% ( Fig 6C and 6D ) , suggesting that the recycling of NRAMP1-GFP from endosomes back to PM is altered in sic1 . In root , BFA primarily inhibits post-Golgi traffic rather than endoplasmic reticulum ( ER ) to Golgi traffic due to that this process heavily relies on GNOM-like 1 ( GNL1 ) , which is BFA resistant [34] . It thus might be contentious that the aggregation of NRAMP1-GFP caused by BFA treatment is from recycling or neosynthesis . To address this question , we first treated the roots of Col-0 and sic1 expressing pNRAMP1::NRAMP1 with a protein synthesis inhibitor cycloheximide ( CHX; 50 μM ) for 30 minutes followed by co-treatment with BFA ( 50 μM ) and CHX ( 50 μM ) for 90 minutes . We found that the intracellular BFA bodies in sic1 were significantly reduced either in amount or in size compared with that in Col-0 after such treatments ( Fig 6E and 6F ) . This observation established that the NRAMP1 recycling is impaired in sic1 and further confirmed that CTL1 is required for endocytosis . In addition , when the plants were pretreated with CHX ( 50 μM ) for 30 minutes and co-treated with BFA ( 50 μM ) and CHX ( 50 μM ) for 90 minutes followed by 2-hour washing out with water , only 13 . 9% of BFA bodies remain in Col-0 cells , whereas 35 . 5% of BFA bodies still remain in sic1 ( Fig 6E and 6F ) . This data further revealed that recycling of NRAMP1 between endosomes and PM requires CTL1 , independently of the de novo delivery . To further examine the role of CTL1 in vesicle trafficking , we examined recycling of PIN-formed 1 ( PIN1 ) , an auxin efflux carrier of which polar PM localization is mediated by vesicle trafficking [35] . We pretreated Col-0 and sic1 plants expressing pPIN1:PIN1-GFP with CHX for 30 minutes and then co-treated them with CHX and BFA for 90 minutes , and observed BFA bodies in both genotypes—which is consistent with previous studies [36] . Then we washed out the BFA with water to see the protein trafficking of PIN1 . After this process , only 14 . 7% of the BFA bodies still existed in Col-0 cells , whereas 34 . 8% of the BFA bodies were presented in sic1 cells ( S11 Fig ) . This result further demonstrated that CTL1 is required for vesicle trafficking and recycling of PM proteins . As CTL1 localizes on TGN and is involved in vesicle trafficking , we wondered if the intracellularly aggregated NRAMP1 and PDCB proteins in sic1 are in TGN/early endosomes ( EEs ) . To figure this out , we stained the transgenic line of sic1 expressing pNRAMP1::NRAMP1-GFP with FM4-64 for 1 hour , which stains the TGN/EE given 1 hour staining for sic1 is equivalent to 30 minutes staining for wild type . As expected , we found that the NRAMP1-GFP compartments were indeed co-localized with FM4-64 ( S8A Fig ) , suggesting the aggregation of NRAMP1 is in TGN/EE . To further confirm this , we co-expressed NRAMP1-GFP and the TGN marker VHA-a1-mCherry in sic1 driven by promoters of NRAMP1 and 35S , respectively . Consistently , we observed that the intracellular NRAMP1 compartments were co-localized with the VHA-a1-mCherry , further confirming that NRAMP1 is retained in TGN of sic1 ( S8D Fig ) . These data suggest that CTL1 is involved in recycling of some cargos from endosomes back to PM . Consistently , when we examined the Golgi apparatus in sic1 by using a transmission electron microscope , we observed that the numbers of secretory vesicles , a major component of TGN [37] , are significantly reduced in sic1 compared to that in Col-0 ( Fig 6G and 6H ) , suggesting that the TGN structure or its volume might be altered in sic . However , when we studied the nature of the intracellular aggregation of PDCBs , we found that they are not the same case as NRAMP1 . After staining the transgenic lines of sic1 expressing PDCB1 with FM4-64 , we found the compartments are not co-localized with FM4-64 ( S8B and S8C Fig ) , suggesting that CTL1 might be involved in not only vesicle trafficking-mediated PM protein secretion pathway but also some other unknown secretion pathways . CTL1 is a known choline transporter [14] . Therefore , we investigated whether choline or the transporter itself participates in vesicle trafficking by treating 6-day-old Col-0 seedlings with 1 mM choline for 2 hours and then staining the seedlings with FM4-64 dye to observe endocytosis . Interestingly , we found that endocytosis was significantly suppressed under choline treatment ( Fig 7A and 7B ) , suggesting that choline homeostasis is crucial for endocytosis . To investigate the role of choline in protein recycling further , we treated Col-0 seedlings expressing NRAMP1-GFP with 50 μM BFA . After 1 . 5 hours , we washed the seedlings with either water ( control ) or 1 mM choline . After 2 hours , 50% of the BFA bodies were still visible in the seedlings washed with choline , whereas 96% of the BFA bodies had vanished in the seedlings washed with water ( Fig 7C and 7D ) . This result suggested that choline homeostasis is important not only for endocytosis but also for membrane protein recycling . In plant cells , choline serves mainly as a precursor of PC , the most abundant phospholipid component of the PM . We thus measured the absolute contents and relative proportions of eight major membrane lipids , including monogalactosyldiacylglycerols ( MGDGs ) , digalactosyldiacylglycerols ( DGDGs ) , PA , PC , phosphatidylethanolamines ( PEs ) , phosphatidylglycerols ( PGs ) , PI , and phosphatidylserines ( PS ) . We found that the absolute contents of all these lipids in sic1 plants decreased at different levels ( Fig 7E and 7F ) , indicating that the generation or steady status of endomembrane system is impaired in sic1 . This result is consistent with the role of CTL1 in vesicle trafficking , as vesicle trafficking is required for membrane generation and communication [38] . Interestingly , though the absolute PC and PE contents in sic1 were decreased , the proportions of PC and PE relative to the total membrane lipids were significantly increased in the mutant , with PC levels increasing by 47% and PE levels increasing by 71% ( Fig 7G ) . This is to say , the compositions of PC and PE in the membrane system of sic1 increase though they decrease relative to the whole plant . As both PC and PE are substrates of PLDs that cleave PC to form choline , this result may indicate that CTL1 mutations inhibit PLD activation . To confirm this , we examined the responses of the sic1 mutant and Col-0 to PLD-specific inhibitor 1-butanol . As expected , sic1 is more resistant to 1-butanol than Col-0 , as the sic1 root is significantly shorter than Col-0 root when grown on 1/2 Murashige and Skoog ( MS ) plate , meanwhile there is no significant difference between the roots of the 2 genotypes when treated with 1-butanol ( Fig 8A–8C ) . This result suggested that the PLD activity is inhibited by the mutation of CTL1 , or in other words , PLD is a downstream component of CTL1 . PLDs have been reported to positively regulate vesicle trafficking . We therefore proposed that CTL1 regulates vesicle trafficking through the effects of choline on PLD activity or membrane PC and PE content ( Fig 8D ) .
In this study , we employed a forward genetic-based ionomics approach [16] to determine that the choline transporter CTL1 is a new component of the vesicle trafficking machinery . Interestingly , we found that CTL1 is highly conserved among different species ( S3 Fig ) , indicating that CTL1 plays a crucial and fundamental role in all organisms . Transmembrane domain analysis showed that CTL1 has ten transmembrane helices ( S3 Fig , TMHMM Server v . 2 . 0 ) . And the mutation site lies in the second transmembrane helix , where the mutated glycine is conserved in plants ( S3 Fig ) . Coincidently , a recent study identified the same mutation in CTL1 . According to that report , such a mutation results in abnormal subcellular localization of CTL1 , which probably explains why this mutation leads to loss of function of CTL1 [18] . Loss of function of CTL1 in the root significantly disrupted the leaf homeostasis of a series of mineral nutrients , including Na , Mn , Fe , Zn , and Mo , suggesting that CTL1 has a systemic impact on ion transportation ( Fig 1A–1C ) . CTL1 was previously shown to be involved in the development of sieve pores , which are a special type of PD [14] . As PDs play an important role in ion transport in the root , it is plausible that the disruption of ion homeostasis characteristic of sic1 may result from defects in PD . We found that PD development in sic1 is severely impaired , probably due to PDCB protein mislocalization or some other reasons ( Fig 3B and 3C ) . However , we realized that the direct effects of PDs on ion transport are not specific , as different elements of sic1 were differentially affected by PD defects . For example , Na+ concentrations were increased 3-fold , while Mn2+ concentrations were decreased by 77% in sic1 compared with Col-0 ( Fig 1A , 1B and 1C ) , suggesting that the ionome disorders characteristic of sic1 may be the ultimate result of several phenomena that exerted direct and indirect effects on ion transport . Abnormal PDs not only affect ion transport but also block the cell-to-cell communication required for root cell identity , patterning , and development ( 23 ) . SHR is a key transcription factor that initiates root patterning , and its movement from the stele to the endodermis via PDs is essential for its function . We established that SHR cell-to-cell communication is severely blocked in most sic1 endodermal cells ( Fig 3D ) , leading to abnormal root patterning ( Fig 1F ) . The cell-type specific expression of transporters is generally orchestrated by cell identity and organization , which are driven by certain common transcription factors . Thus , it is reasonable to predict that the expression of some transporters may be altered in sic1 because of defects in SHR or other mobile transcription factors . We found that both HMA4 and HKT1 were ectopically expressed in the epidermis of sic1 ( Fig 4A , 4B and S9 Fig ) . As a consequence , Zn2+ and Na+ transport was also disordered ( Fig 4E–4G ) and may have been an indirect cause of the ionomic phenotype of sic1 . Interestingly , not all the transporter genes were misexpressed in the sic1 mutant , suggesting that the impact of CTL1 on transcription is gene-specific . We noticed that both HKT1 and HMA4 are stele-specific genes in Col-0 , and their ectopic expression is in epidermal cells of sic1 , whereas NRAMP1 is originally expressed in the epidermis . One possibility could be that the expression of the transporter genes like HKT1 and HMA4 are inhibited in epidermal cells because of some factors requiring CTL1 . In such a case , loss of function of CTL1 thus could result in disinhibition of HKT1 and HMA4 in the epidermis , meanwhile NRAMP1 is not affected . Consistent with this hypothesis , the establishment of the cell identity of the epidermis also requires factors that move through PDs , such as CAPRICE [39] . In the future , it would be interesting to address whether expression of HKT1 and HMA4 are related to these factors and if these transcription factors require CTL1 . However , we also noticed that the expression patterns of HKT1 and HMA4 are somewhat overlapped with CTL1 in the mature zone of Col-0 , but it is hard to judge if the expression pattern of CTL1 is related to its gene-specific effect on transcriptional regulation or if that’s just a coincidence . Although the expression pattern of NRAMP1 was not changed , we observed that its trafficking to the PM was severely affected in sic1 . As NRAMP1 is responsible for Mn and Fe uptake and long-distance transportation , defective NRAMP1 trafficking is certainly at least partially responsible for the low Mn and Fe phenotypes characteristic of sic1 shoots . CTL1 localizes to the TGN ( S7 Fig ) [14] , and CTL1 mutations affect the subcellular localization of various PM proteins , including NRAMP1 and PDCB proteins . These data inspired us to assess whether CTL1 is involved in regulating vesicle trafficking . Our studies of the endocytosis tracer FM4-64 and NRAMP1 and PIN1 trafficking confirmed our hypothesis . Supporting this conclusion , the numbers of secretory vesicles are much less in sic1 than in Col-0 ( Fig 6G and 6H ) , which indicates that the TGN structure or volume is changed by the mutation of CTL1 . The localization of NRAMP1 to the PM was recently reported to be mediated by vesicle trafficking [32] . Our results confirmed this finding and established that this process involves CTL1 . In contrast , little is known regarding how PDCB proteins are specifically directed to PDs , which are special PM sites . PDCBs were previously found to be localized to PD domains , where callose deposit and play important roles in PD formation [19] . In this study , we found for the first time that the localization of PDCBs on PDs requires CTL1 , which was initially identified as a regulator of the sieve plate and sieve pore and xylem development [14] . As a sieve pore is a special PD in the phloem , the sieve pore phenotype of ctl1 may also be a result of defects in the subcellular localization of PDCBs or their partner ( s ) . The sic1 mutant exhibits a comprehensive ionomic defect , which may indicate that CTL1-mediated vesicle trafficking affects the recycling of large numbers of PM proteins , including various ion transporters . For example , the subcellular localization of the iron transporter IRT1 , which is essential for iron uptake of A . thaliana , was previously found to be regulated by endosomal cycling . Although we did not get direct evidence about the IRT1 localization , we found that the expression of IRT1 , as well as AHA2 , was dramatically up-regulated in sic1 even at Fe sufficient condition ( S10 Fig ) . This elevated IRT1and AHA2 might reflect that their subcellular localization is also regulated by CTL1 . In animals , choline is essential for neuron signaling and choline transporter plays important roles in choline homeostasis [12 , 13 , 40] . The vesicle-localized choline transporter was believed to function in secretion of acetylcholine , but it remains unknown if it plays some other roles . It was well-documented that phospholipid compositions of the membrane system play important roles in vesicle trafficking . For example , PA , one of the products in the hydrolysis of PC catalyzed by PLD , is enriched at the budding position of PM to form vesicles [41] . The involvement of PLD in vesicle trafficking was therefore believed to be associated with its role in phospholipid homeostasis on membrane system . Interestingly , as the most abundant phospholipid and the substrate of PLD , PC has never been found to play a role in vesicle trafficking . PC is synthesized on ER or Golgi apparatus and delivered to PM through membrane trafficking . Along this membrane trafficking process , PC contents gradually decreased [42] . In addition , the distribution of PC is asymmetric on PM and endosomes , which predominantly lies on the exoplasmic leaflet of PM and the luminal side of the endosomes . Meanwhile , PC is symmetrically distributed on both sides of the ER membrane [42] . It is unclear what is the significance of these phenomena , but it would be attractive if they represent a mechanism for vesicle trafficking . At least , we might hypothesize that the gradual reduction of PC content during vesicle trafficking is mediated by PC hydrolysis on the cytoplasmic side of the vesicle membrane catalyzed by cytoplasmic PLD . The hydrolysis of PC releases hydrosoluble choline into cytoplasm and leaves PA on the cytoplasmic side of the vesicle membrane . The vesicle-localized choline transporter could be important in this process , given that free choline produced by PC hydrolysis in the cytoplasmic side would inhibit PLD activity . Sequestration of choline into vesicles could help to shape asymmetric distribution of PC on vesicle membrane and PM , and thus to improve vesicle trafficking . CTL1 was characterized as a TGN-localized choline transporter that regulates choline homeostasis in plant cells [14] . Our findings that CTL1 is involved in vesicle trafficking provide a substantial evidence to support above hypothesis . Interestingly , we observed that the proportions of PE and PC were increased in the cellular membrane system of sic1 . As PE and PC are both substrates of PLDs , the high proportions of PE and PC in sic1 may suggest that PLD activity was inhibited in sic1 mutants , which was then supported by our PLD inhibitor experiment . Thus , it is plausible that CTL1 functions in sequestration of choline into TGN/endosomes to maintain high PA/low PC proportion on the cytoplasmic leaflet and high PC proportion on the luminal side of the vesicles , which might be required for vesicle trafficking . Consistent with this hypothesis , we found that high concentrations of choline inhibit vesicle trafficking . Based on these results , we surmised that the involvement of CTL in vesicle trafficking may be explained by the fact that CTL1 functions in compartmentalizing choline in vesicles to create a low-choline cytosol environment for high PLD activity to promote vesicle trafficking . Though the relative proportions of PC and PE in the membrane system were increased in sic1 mutants , the absolute contents of them in the whole plant were both reduced . This result is consistent with previous observation that the absolute content of choline and PC were reduced in cher1-4 [14] . However , we also found that not only PC , but all other major membrane lipids are decreased in sic1 . This uniform reduction of all membrane lipids suggests that the mutation of CTL1 leads to a shrink in producing cellular membrane system . Different membrane organelles are connected through vesicle trafficking . The reduced membrane lipids thus might be a result of vesicle trafficking defects of sic1 . Of course , we could not exclude the possibility that the reduction of total membrane lipids in sic1 is caused by lipid synthesis problems , if choline homeostasis could affect synthesis of all membrane lipids . Except for vesicle trafficking process , the PD development also has a close connection with the lipid composition of the PM-lining PD ( PD-PM ) domain [43] . In a previous study , it was found that the major phospholipids in PD-PM were PE ( approximately 45% ) and PC ( approximately 20% ) [43] . As CTL1 regulates PC and PE homeostasis and their delivering to PM , the mislocalization of PDCBs in sic1 could also be a result of composition changes of phospholipids on PM . Based on this hypothesis , the intracellular retention of PDCBs in sic1 might not be a result of vesicle trafficking defect . Otherwise , the subcellular localization of PDCBs in PD should be mediated by some unknown secretory pathway , for example , directly via Golgi apparatus , which has been shown to play an essential role in formation of cell plate and PD [44–46] . Overall , both vesicle trafficking and ion homeostasis are fundamental biological processes . Here , we identified a new participant in both vesicle trafficking and ion homeostasis that links these two critical processes . Furthermore , we provided credible evidence of the existence of a direct chain of molecular events linking vesicle trafficking , ion transporter sorting , PD development , cell-to-cell communication , root patterning , and ion homeostasis . However , the detailed molecular mechanisms underlying the involvement of CTL1 in these processes need to be investigated further .
The A . thaliana plants used in this study were the Col-0 accession , and the T-DNA insertion mutant SALK_065853 , which was used to generate sic1-2 , was obtained from the Nottingham Arabidopsis Stock Centre ( NASC ) . To grow seedlings on agar solidified growth medium , we pretreated A . thaliana seeds with 75% ethanol for 1 minute , surface-sterilized them by immersing them in 10% NaClO for 10 minutes , and then washed them with distilled water at least five times . The surface-sterilized seeds were sowed on medium containing 1/2 MS salt and 1% sucrose solidified with 0 . 6% phytalgel ( Sigma-Aldrich , St . Louis , MO ) . After 3 days of stratification at 4°C in the dark , the plates were maintained under 16 hour light/8 hour dark cycles at 23°C . The plants used for leaf elemental analysis as part of the screening of the EMS-mutagenized plants were grown in a climate-controlled room for up to five weeks , as previously described [15] . The plants were bottom-watered twice a week with modified 0 . 25 × Hoagland’s Type 2 with 1 mL/L Fe-HBED [15] . Mutagenized A . thaliana seeds were obtained from Lehle Seeds ( Round Rock , TX ) . To screen 1 , 554 M2 plants from 17 different parental plants , we grew seed packets and analyzed them by inductively coupled plasma mass spectrometry ( ICP-MS ) . Of these plants , 233 were identified as putative mutants based on their leaf elemental profiles and were allowed to self-fertilize . The seeds were collected , and 11 plants per M3 family were re-screened by ICP-MS after growing in soil . Fifty-six mutants with altered leaf ionomes were identified by this second round of screening . For hydroponic culture , seeds of A . thaliana ( Col-0 and sic1 ) were stratified for three days at 4°C in water , and then put on a pipe cover with a pore in the middle of a 1 . 5-mL Eppendorf tube containing Hoagland solution in a climate-controlled room for up to two weeks , as previously described [15] . The seedlings with the pipe cover were then transferred to a new container for 3 additional weeks . The medium was refreshed every three days . The leaves and roots of 5-week-old plants were collected for elemental analysis . A . thaliana leaf tissue elemental analysis via ICP-MS has been previously described [47] . Briefly , 2 healthy rosette leaves from one 5-week-old plant were cut with a scalpel , and the leaf was held with plastic tweezers . The collected leaves were subsequently rinsed with 18 MΩ water in a 1 , 000-ml beaker four times to wash off external impurities . The rinsed samples were then placed in glass tubes and pushed to the bottoms of the tubes with a glass rod to ensure that none of them were left on the tube walls . The tubes were then transferred to an oven for 20 hours at 92°C . After cooling , seven to ten samples were weighed on an analytical balance . All the samples , including the blank controls , were then digested with 1 ml of concentrated nitric acid containing indium ( In ) as an internal standard for 4 hours at 110°C before being diluted with 18 MΩ water in a final volume of 10 ml . Elemental analysis of Li , B , Na , Mg , P , S , K , Ca , Mn , Fe , Co , Ni , Cu , Zn , As , Se , Rb , Sr , Mo , and Cd was performed with an ICP-MS ( NexION 350D; PerkinElmer , Waltham , MA ) coupled with an Apex desolvation system and an SC-4 DX autosampler ( Elemental Scientific Inc . , Omaha , NE ) . All the samples were normalized with a heuristic algorithm using the best measured elements as previously described [15] . Reciprocal grafting was performed as previously described [47] . After the graft unions were established , the grafted plants were examined under a stereoscopic microscope before being transferred into potting mix soil to observe the formation of any adventitious roots from the graft unions or above . Healthy grafted plants without adventitious roots were transferred to potting mix soil and grown in the controlled environment described above . After four weeks , leaf samples were harvested for ionomic analysis . After harvesting , the plants were examined again , and those with adventitious roots or without a clear graft union were excluded from the subsequent analysis of the ionomic data . PCR-based genotyping was used for mapping of the causing mutation in sic1 . Using two SSLP markers and a population of 262 F2 plants from the Ler-0 Χ sic1 cross , we were able to map the gene to within a 560-kb genomic region on chromosome 3 . Further SSLP markers were developed within this 560-kb mapping interval and used to screen 1 , 768 further Ler-0 Χ sic1 F2 plants to identify informative recombinants to further narrow the mapping interval to a 100-kb region between SSLP markers of GM510 and GM524 . Overlapping fragments of approximately 0 . 7 to approximately 1 kb each , covering this 80-kb candidate region , were amplified from the genome of sic1 and sequenced . The sequence of these fragments was alignment with the wild-type sequence using KOD neo plus ( TOYOBO , Osaka , Japan ) . The markers used in identifying informative recombinants were shown in S8 Table . For the sic1 complementation test , we constructed a pSIC1::SIC1-GFP fusion vector . A 4 . 4-kb genomic DNA fragment including a 1 . 3-kb gene promoter and a gene body without a TGA stop codon was amplified using the primers SIC1-GFP-L and SIC1-GFP-R , and the GFP fragment was amplified from the pHAC1::HAC1-GFP vector [47] using the primers GFP-L and GFP-R . Thereafter , the two fragments were fused together by overlapping PCR using the primers SIC1-GFP-L and GFP-R . The fused fragment was inserted into a pHMS plant expression vector modified from a pHB vector [48] via HindIII and PstI restriction sites using a Hieff Clone one-step PCR cloning kit ( Yisheng Co . Ltm , Shanghai , China ) . To construct the pHMA4::HMA4-GFP fusion vector , we amplified an approximate 10-kb HMA4 genomic DNA fragment from Col-0 by PCR using a KOD Hot Start DNA Polymerase ( TOYOBO ) . The fragment included a 4-kb gene promoter and gene body but lacked a TGA stop codon . Amplification was performed using the primer pair HMA4-GFP-L and HMA4-GFP-R . Because of the large size of the HMA4 genomic DNA fragment , the GFP coding sequence fused with five GGA repeats at the N-terminal , where it served as a linker , was cloned into the pHMS vector backbone between the Pst I and Xba I restriction sites in advance . This modified pHMS vector was renamed as pGHMS . The amplified HMA4 fragment was cloned into pGHMS using Hind III and PstI restriction sites using a Hieff Clone one-step PCR cloning kit ( Yisheng Co . Ltm ) . For the pNRAMP1::NRAMP1-GFP construct , we prepared the NRAMP1 genomic DNA , including the 2 . 5-kb native promoter , from Col-0 plants by PCR amplification using the primers NRAMP1-GFP-L and NRAMP1-GFP-R . The amplified fragment was cloned into a pGHMS vector with Hind III and Pst I restriction sites using a Hieff Clone one-step PCR cloning kit ( Yisheng Co . Ltm ) . For the 35S::VHA-a1-mCherry construct , we first modified the pHB vector [48] by inserting the mCherry fragment with a five-glycine linker between the Pst I and Xba I restriction sites , and renamed the modified pHB vector as pHB-mCherry . We then amplified the CDS region of VHA-a1 from the Col-0 cDNA using the primers VHA-a1-mCherry-L and VHA-a1-mCherry-R . The amplified CDS fragment was then cloned into pHB-mCherry vector with Hind III and Pst I restriction sites using a Hieff Clone one-step PCR cloning kit ( Yisheng Co . Ltm ) . To construct the pHKT1::GUS vector , we amplified the HKT1 promoters with the primer pair HKT-GUSL and HKT-GUSR using Col-0 genomic DNA as a template , after which we inserted them into the pCAMBIA1303 vector to drive uidA expression . GUS histochemical staining was performed as previously described [30] . The procedure with which the pSHR::SHR-GFP and pPDCB1::PDCB1-GFP and pHMA4::GUS vectors were constructed has been previously described [19 , 49] . All the expression vectors were transformed into the Agrobacterium tumefaciens strain GV3101 and introduced into Col-0 or sic1 using the floral dip method [50] . The primers used in the study are shown in S8 Table . The roots of 3-week-old hydroponically cultured Col-0 and sic1 plants were used to extract total RNA by using TRNzol A+ RNA Purification reagent ( DP421; Tiangen Biotech , Beijing , China ) . Two micrograms of total RNA were used to synthesize first-strand cDNA with TransScript one-step gDNA removal and cDNA synthesis super mix ( AT311-02; TransGen Biotech , Beijing , China ) . qRT-PCR was performed using SYBR green PCR master mix ( TRT-101; TOYOBO ) with the first-strand cDNA as a template on a real-time PCR system ( CFX thermocycler; Bio-Rad , Hercules , CA ) . Primers for qRT-PCR were designed using Primer Express software version 3 . 0 ( Applied Biosystems , Foster City , CA ) . The primers UBCF and UBCR were designed for ubiquitin-conjugating enzyme 21 ( At5g25760 ) , which was used as the reference gene . The primer sequences are shown in S8 Table . Expression data analysis was performed as previously described [51] . Before choline treatment , choline chloride was dissolved in water to make a 100-mM choline solution . In the experiment , the choline solution was diluted to a concentration of 1 mM , and the seedlings were treated for the desired time . For BFA treatment , 6-day-old Arabidopsis seedlings were immersed in 50 μM BFA for 1 . 5 hours , and then the BFA bodies were observed by confocal laser scanning microscopy ( Leica TCS SP8 ) . In the washout experiments , the BFA-treated seedlings were incubated in water for 2 hours , after which the BFA bodies were examined by confocal laser scanning microscopy ( Leica TCS SP8 ) . For CHX and BFA treatment , 6-day-old Arabidopsis seedlings were pretreated with 50 μM CHX for 30 minutes , followed by being immersed in a solution with 50 μM CHX and 50 μM BFA for 1 . 5 hours , and then the BFA bodies were observed by confocal laser scanning microscopy ( Leica TCS SP8 ) . In the washout experiments , above seedlings were incubated in water for 2 hours , and the BFA bodies were then examined by confocal laser scanning microscopy ( Leica TCS SP8 ) . For 1-butanol treatment , the Arabidopsis seeds were grown on 1/2 MS salt medium plate for six days , and the seedlings were then transplanted to the 1/2 MS salt medium plate containing 0 . 4% 1-butanol and grew for another five days . The root length was measured by Image J . Seven-day-old Col-0 and sic1 plants grown on 1/2 MS plates were treated with PBS solution containing 20 mM Zinpyr-1 ( Cayman Chemical , Ann Arbor , MI ) at room temperature in darkness for 3 hours . Then , the seedlings were washed with PBS solution twice . The stained plants were observed with a confocal laser scanning microscope ( Leica TCS SP8 ) using excitation at 490 nm and emission at 530 nm . For CoroNa staining , 6-day-old Col-0 and sic1 plants grown on 1/2 MS plates were treated with 5 μM CoroNa water solution in the dark for 20 minutes . The stained plants were washed in water twice and then observed under a confocal laser scanning microscope ( Leica TCS SP8 ) with an excitation wavelength of 492 nm and an emission wavelength of 516 nm . Confocal laser scanning microscopy was performed on Leica TCS SP8 and Olympus FluoView FV1000 confocal microscopes . To observe the GFP fusions , we illuminated 5- to 7-day-old plants with an excitation wavelength of 488 nm with an Argon laser , and emission was detected at 505–550 nm . The images used for the comparisons of signal intensity between Col-0 and sic1 were captured using a confocal microscope with the same settings . For the FM4-64 internalization assay , 5- to 7-day-old Col-0 and sic1 mutant seedlings grown on solidified growth medium were incubated with 2 μM FM4-64 solution for 5 minutes and then rinsed twice with water . The concentration of the FM4-64 stock solution in DMSO was 4 mM . The FM4-64 in seedlings was visualized at 10 minutes , 30 minutes , and 60 minutes after staining . For propidium iodide staining , 5-day-old seedlings were incubated in a fresh solution of 15 mM ( 10 mg/ml ) PI dissolved in water in the dark for 5 minutes , after which they were rinsed twice in water . The fluorescence excitation and emission wavelengths of FM4-64 and PI were 561 nm and 610–630 nm , respectively . The fluorescence signal intensity was calculated use Leica Application Suite X ( version 1 . 1 . 0 , Leica , Wetzlar , Germany ) . The PDs were imaged at 80 kV with a Hitachi H-7650 transmission electron microscope . The extraction and analysis of membrane lipids were performed as described previously [52] . Briefly , the seedlings were quickly immersed in isopropanol preheated to 75°C with 0 . 01% butylated hydroxytoluene ( BHT ) for 15 minutes . Then chloroform and water ( 5:2 ) were added , and the mixture was incubated at room temperature for 1 hour . The lipid extract was transferred into glass tubes with Teflon-lined screw-caps . Chloroform/methanol ( 2:1 ) with 0 . 01% BHT was added into the extracts and shaken for 30 minutes . This process was repeated until all samples became white . The extracts for each sample were combined and washed with 1 M KCl followed by water washing . The final lipid extractions were evaporated under a gentle flow of N2 gas and then stored at −80°C . The membrane lipid analysis was performed by automated electrospray ionization–tandem mass spectrometry as previously described , with the standards including di14:0-PC , di24:1-PC , 13:0-lysoPC , 19:0-lysoPC , di14:0-PE , di24:1-PE , 14:0-lysoPE , 18:0-lysoPE , di14:0-PG , di24:1-PG , 14:0-lysoPG , 18:0-lysoPG , di14:0-PA , di20:0 ( phytanoyl ) -PA , di14:0-PS , di20:0 ( phytanoyl ) -PS , 16:0–18:0-PI , di18:0-PI , 16:0–18:0-MGDG , di18:0-MGDG , 16:0–18:0-DGDG , and 0 . 71 nmol di18:0-DGDG [53] . | Ion transporters play a key role in mineral nutrients uptake and transport of plants . Their cellular and subcellular localization is essential for fulfilling their functions and are therefore generally fine-tuned . However , the molecular mechanisms underlying this process remain largely unclear . In this study , we analyze the role of the choline transporter CTL1 in A . thaliana and find that it controls dynamic cell trafficking , a fundamental process that plays vital roles in cell signaling , development , and protein sorting . We also show that CTL1 regulates the expression pattern of different ion transporters through the modulation of vesicle trafficking . These results suggest that CTL1 is a new component of the vesicle trafficking machinery that is also required for ion homeostasis , which link these processes and shed light on the underlying molecular mechanisms . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"protein",
"transport",
"b",
"vitamins",
"cell",
"physiology",
"plant",
"anatomy",
"medicine",
"and",
"health",
"sciences",
"vesicles",
"chemical",
"compounds",
"cell",
"processes",
"green",
"fluorescent",
"protein",
"organic",
"compounds",
"cholines",
"physiological",
... | 2017 | A new vesicle trafficking regulator CTL1 plays a crucial role in ion homeostasis |
Previous studies in Saccharomyces cerevisiae established that depletion of histone H4 results in the genome-wide transcriptional de-repression of hundreds of genes . To probe the mechanism of this transcriptional de-repression , we depleted nucleosomes in vivo by conditional repression of histone H3 transcription . We then measured the resulting changes in transcription by RNA–seq and in chromatin organization by MNase–seq . This experiment also bears on the degree to which trans-acting factors and DNA–encoded elements affect nucleosome position and occupancy in vivo . We identified ∼60 , 000 nucleosomes genome wide , and we classified ∼2 , 000 as having preferentially reduced occupancy following H3 depletion and ∼350 as being preferentially retained . We found that the in vivo influence of DNA sequences that favor or disfavor nucleosome occupancy increases following histone H3 depletion , demonstrating that nucleosome density contributes to moderating the influence of DNA sequence on nucleosome formation in vivo . To identify factors important for influencing nucleosome occupancy and position , we compared our data to 40 existing whole-genome data sets . Factors associated with promoters , such as histone acetylation and H2A . z incorporation , were enriched at sites of nucleosome loss . Nucleosome retention was linked to stabilizing marks such as H3K36me2 . Notably , the chromatin remodeler Isw2 was uniquely associated with retained occupancy and altered positioning , consistent with Isw2 stabilizing histone–DNA contacts and centering nucleosomes on available DNA in vivo . RNA–seq revealed a greater number of de-repressed genes ( ∼2 , 500 ) than previous studies , and these genes exhibited reduced nucleosome occupancy in their promoters . In summary , we identify factors likely to influence nucleosome stability under normal growth conditions and the specific genomic locations at which they act . We find that DNA–encoded nucleosome stability and chromatin composition dictate which nucleosomes will be lost under conditions of limiting histone protein and that this , in turn , governs which genes are susceptible to a loss of regulatory fidelity .
Twenty-five years ago , Michael Grunstein's laboratory began a series of experiments in which histones H2B or H4 were depleted in vivo in S . cerevisiae ( yeast ) . After histone gene silencing , the yeast cells complete a single round of DNA replication , reducing their histone-DNA ratio by a factor of approximately two , and enter cell-cycle arrest . A number of conditionally expressed genes , including PHO5 , GAL1 , CYC1 , CUP1 , and HIS3 , were reported to be transcriptionally de-repressed following histone depletion [1]–[4] . In 1999 , Richard Young's laboratory revisited the transcriptional effects of H4 depletion , this time on a genomic scale using microarrays . Roughly 15% ( 888 ) of yeast genes were de-repressed 3-fold or greater , and another 10% of genes ( 569 ) were repressed at least 3-fold following H4 depletion [5] . Southern blots revealed changes in chromatin structure upstream of the PHO5 promoter during H4 depletion , but at the time these experiments were performed , high-resolution genomic methods were not available to measure the widespread chromatin changes hypothesized to underpin the observed transcriptional changes . We therefore revisited these classic experiments using RNA-seq and MNase digestion coupled with next-generation sequencing ( MNase-seq ) to monitor RNA and chromatin changes after histone depletion . These experiments also bear on the recent debate regarding the degree to which DNA-encoded elements or other cellular factors control nucleosome position and occupancy [6]–[10] . The raw RNA-seq and MNase-seq data are available at GEO accession number GSE29294 .
Two genes encode histone H3 in wildtype S . cerevisiae , HHT1 and HHT2 . We obtained a strain in which HHT1 had been deleted and HHT2 had been placed under control of the GAL1 promoter ( Figure 1A ) [11] . When grown in galactose , this “H3 shutoff” strain ( DCB200 . 1 ) grows similarly to wildtype yeast ( YEF473A ) , whereas cultivation in dextrose results in growth arrest in the G2/M phase of the cell cycle after a single round of DNA synthesis as large-budded cells ( Figure S1A , S1B ) [11] . After 3 hours in dextrose , RNA-seq shows that HHT2 transcription is reduced to near zero , and Western blot analysis shows that histone H3 protein levels have dropped by a factor of two , as expected ( Figure S1C , S1D ) . This evidence , coupled with previous cytological evidence using this strain [11] , shows that the switch to dextrose successfully depleted histone H3 . We mapped nucleosome position and relative nucleosome occupancy using MNase-seq in the wildtype and H3 shutoff strains at 0 and 3 hours after transitioning the cells from galactose media to dextrose media ( Figure S2A–S2C ) . While previous studies examined transcriptional changes 6 hours after the shift [5] , we chose 3 hours for our study based on our characterization of the rate of histone depletion ( Figure S1A–S1D ) and to decrease the time the cells were under stress . Consistent with previous studies [2] , [12] , the chromatin was generally more sensitive to MNase digestion following H3 depletion , resulting in increased background smearing ( Table S1 , p<0 . 05; Figure S2D–S2G ) . Mono-nucleosome bands were extracted from the gel and sequenced , and the resulting reads were mapped back to the genome ( Materials and Methods ) . For the four possible strain and growth combinations , we aligned all of the genes by their +1 nucleosome relative to the transcription start site ( TSS ) [13] and calculated the smoothed dyad density across the gene body ( Materials and Methods ) . The genes were then sorted by length from shortest to longest ( Figure 1B ) . This plot reveals several key features of our dataset . First , the nucleosome organization of the wildtype strain in galactose ( 0 hours ) is very similar to the nucleosome organization of the H3 shutoff strain in galactose ( 0 hours ) . Second , the nucleosome organization of the wildtype strain in galactose is very similar to the nucleosome organization of the wildtype strain in dextrose ( 3 hours ) . Third and most importantly , there are dramatic changes in nucleosome organization in the H3 shutoff strain between galactose ( normal histone levels ) and dextrose ( depleted levels of H3 ) . As histone availability decreases , the overall nucleosome positioning at any given position is weaker . Nucleosome positioning is still evident at the +1 and +2 nucleosomes but then decays rapidly , leading to a nearly complete loss of visible positioning until the gene's transcription termination site ( TTS ) is reached . There , one can still see a nucleosome at the −1 position relative to the TTS , although it is much more weakly positioned than in the wildtype strain . To assess the overall similarity of our experimental replicates , we compared the overall nucleosome occupancy for each base pair among the replicates and found the experiments to be highly correlated ( Figure S2H–S2K ) . As expected from the positioning data , the wildtype and H3 shutoff strains had very highly correlated nucleosome occupancy profiles when grown in galactose . When both strains were grown in dextrose , which results in histone H3 depletion in the shutoff strain , the correlation between the nucleosome occupancy profiles of the wildtype and H3 shutoff strains was lower , as expected ( Figure 1C ) . To identify changes in specific nucleosomes , we determined the position and occupancy of individual nucleosomes in each replicate at each time point ( Materials and Methods ) . We used a slightly modified version of previously proposed definitions for measuring nucleosome positions and occupancies [14] that relies upon the midpoint of the paired-end reads or the center of the extended single-end reads ( the “nucleosome dyad” ) when examining nucleosome position and occupancy ( Materials and Methods; Figure 2A ) . This method reduces positional noise that might be introduced from MNase producing fragments of varying lengths . Changes in each nucleosome's position between the 0- and 3-hour time points were assessed using the t-test on the distribution of the read centers . The nucleosome position was then classified as being “altered” or “unchanged” based on the resulting p-value . Nucleosomes with borderline p-values were placed in a “no call” category and not used for the downstream analyses ( Materials and Methods; Figure S3 ) . The use of the t-test allowed us to account statistically for the “fuzziness” of each individual nucleosome . We note that p-values represent the confidence that an observation differs from the null expectation , not the degree to which an observation is biologically relevant . Changes in each nucleosome's occupancy were assayed using the binomial distribution test on the number of read centers that fell within a 100-bp window centered on the nucleosome's consensus dyad . We stress that our experimental approach did not allow absolute nucleosome occupancy calculations , so the occupancy at a given location is in effect measured relative to all other nucleosomes in the same experiment . Nucleosomes were classified as “preferentially reduced” ( decreased occupancy relative to other nucleosomes ) , “unchanged” ( similar occupancy relative to others ) , or “preferentially retained” ( increased relative occupancy ) . Nucleosomes with borderline p-values were placed in a separate “no call” category , similar to the position classification ( Figure S3 ) . Nucleosomes that were classified identically in two or more replicates were selected as being “reliably characterized” ( Materials and Methods; Figure S3 ) . By comparing nucleosomes that were reliably characterized in both the wildtype and H3 shutoff experiment , we were able to remove nucleosomes for which the behavior could be attributed to the change in carbon source ( Figure 2B and Figure S3 ) . All further analysis was performed only with the “H3-depletion dependent” set of nucleosomes . The source code for all of the custom nucleosome calling tools can be found at http://sourceforge . net/p/callnucleosomes . As an illustration of our nucleosome calls , we produced a graphical representation of the nucleosome structure surrounding PHO5 , which had been previously shown to have altered chromatin and increased transcription following H3 depletion [1] , and IDP2 , a gene with increased transcription following H3 depletion ( Figure S4 and Figure S5 ) . After H3 shutoff , a new nucleosome configuration was observed upstream of both genes . We sought to understand the factors that were responsible for the nucleosomes that were preferentially lost or retained or that responded to histone depletion by altering their positions . We compared each of the five classes of nucleosome changes ( “position altered” , “position unchanged” , “occupancy reduced” , “occupancy unchanged” , and “occupancy retained” ) to genome annotations and to previously published ChIP-chip or ChIP-seq experiments , including in vivo maps of histones , histone modifications , histone variants , and DNA-associated proteins ( Figure 3 and Figure S6 and Table S1 ) . Nearly all of the published data we used was obtained under standard growth conditions in dextrose , which matches the growth conditions of our strains upon the initiation of the H3 transcriptional shutoff ( Materials and Methods ) . Therefore , the published datasets offer a reasonable representation of the chromatin landscape during histone depletion . We note that while our data has single-base pair technical resolution , the resolution of the published datasets varies ( see Figure 3 ) . Therefore statements about nucleosomes lacking or harboring a specific histone modification or other property should be interpreted with this in mind . Upon histone H3 depletion , reduced-occupancy nucleosomes tended to occur in promoter regions , while nucleosomes in the gene bodies were associated with unchanged occupancy ( Figure 3A ) . Consistent with this , a mark associated with transcriptional elongation , H3K36me2 , was strongly enriched on nucleosomes classified as “position unchanged” or “occupancy unchanged” ( Figure 3B ) . H3K36me2 has been proposed to contribute to increased nucleosome stability in the wake of RNA Polymerase II ( RNA Pol II ) transcription via the recruitment of the histone deacetylase ( HDAC ) Rpd3 [15] , [16] . Consistent with nucleosome destabilization due to histone acetylation , “altered position” or “reduced occupancy” nucleosomes were enriched for 14 of the 16 of the histone acetylation marks we examined ( p<0 . 01 ) ( Figure 3B ) . Acetylated nucleosomes are generally less stable than those that are not acetylated [17]–[19] , so it follows that acetylated nucleosomes would be more susceptible to changes following H3 depletion . We also found that nucleosomes with reduced occupancy or altered position tended to exhibit high replication-independent turnover rates ( Figure 3E and Figure S7 ) [20] , suggesting that the more dynamic histone-DNA interactions at these locations result in a relative loss under histone-limiting conditions . We compared the H3-depletion responsive nucleosomes to previously published genome-wide data for several transcription factors , including Bdf1 , Rap1 , Reb1 , and Tup1 [21] , [22] . Sites of Bdf1 , Rap1 , and Reb1 binding were all associated with reduced nucleosome occupancy following H3 depletion , which is consistent with the known ability of these proteins to displace nucleosomes ( Figure 3C ) . In contrast , Tup1 is a known transcriptional repressor that can recruit histone deacetylases ( HDACs ) [23]–[26] . Sites of Tup1 binding were associated with “position unchanged” and “occupancy retained” or “occupancy unchanged” nucleosomes ( Figure 3C ) . These results support the hypothesis that Tup1 represses transcription by stabilizing nucleosome position and occupancy [21] , [27] . Isw2 is an ATP-dependent chromatin remodeler that positions nucleosomes at the 5′ and 3′ ends of genes by binding to both the histone octamer and DNA [28] , [29] . ISW2 is known to catalyze the centering of a nucleosome on a DNA substrate in vitro [30] , [31] . Based on the distribution of catalytically inactive Isw2 enzyme ( Isw2K215R ) [32] , Isw2 is associated with the unique combination of “occupancy retained” and “position altered” nucleosomes in our experiment . This is in contrast to most other factors associated with “occupancy retained” nucleosomes , which are typically classified as “position unchanged” nucleosomes ( Figure 3D ) . Based on its biochemical nucleosome-centering activity [30] , [31] , we hypothesized that Isw2 may help retain nucleosome occupancy by stabilizing histone-DNA interactions , while at the same time causing the position of bound nucleosomes to be especially sensitive to the loss of an adjacent nucleosome . Adjacent nucleosome loss could provide free DNA for the nucleosome centering activity of ISW2 . Consistent with this hypothesis , of the 653 nucleosomes bound by Isw2K215R and classified as “position altered” , 429 move in the same direction in all 4 replicates , while another 142 shift in the same direction in 3 of the 4 replicates , indicating that for over 85% of the affected nucleosomes there is a clear directionality to the Isw2-associated shift in vivo ( Figure S8A ) . The shift in position of these Isw2-bound nucleosomes was strongly associated with a decrease in nucleosome occupancy within 600 bp of the direction of the positional shift ( p = 6 . 9E−11 ) . In other words , the Isw2-bound nucleosomes consistently shifted specifically in the direction of a nearby , lost nucleosome . In contrast , Isw2-bound nucleosomes that do not change position are associated with adjacent “occupancy retained” nucleosomes ( p = 0 . 059 ) and are not associated with a reduced occupancy of adjacent nucleosomes ( p = 1 ) , suggesting that a loss of adjacent nucleosomes is required for ISW2-mediated positional changes ( Figure S8B , S8C ) . Fully consistent with the known in vitro activity of ISW2 , the “linker” length ( distance to the next mapped nucleosome ) increased both upstream and downstream of Isw2-bound nucleosomes following nucleosome depletion such that the Isw2-bound nucleosomes became centered on the new local DNA substrate . While the ability of Isw2 to slide nucleosomes in vivo has been demonstrated previously [33] , our result suggests that under conditions of lowered nucleosome density , Isw2 acts to create regularly-spaced nucleosomal arrays in vivo . We note that position-altered nucleosomes not bound by the mutant Isw2K215R as described in [32] were also centered , suggesting that wildtype Isw2 or another remodeling enzyme may center nucleosomes on newly-created gaps during genome-wide histone depletion ( Figure S8D ) . “Position altered , ” Isw2-bound nucleosomes tended to contain the histone variant H2A . z ( p<0 . 01 ) relative to all Isw2-bound nucleosomes [34] , while Isw2-bound nucleosomes that were stably positioned were enriched for Tup1 ( p<0 . 001 ) and Rsc8 ( p<0 . 0001 ) binding relative to all Isw2-bound nucleosomes . Tup1 has previously been shown to localize independently to Isw2-bound regions , suggesting that the two proteins may work independently to maintain nucleosome position [35] . Taken together , these patterns suggest that Isw2's centering function in vivo may be aided by incorporation of H2A . z and restricted by Tup1 and Rsc8 . Nhp6a is an HMG-group protein known to associate with chromatin ( reviewed in [36] ) . Loss of Nhp6a interferes with conditional gene activation [37] and has been reported to stabilize nucleosomes at promoters [38] . Human HMGB1 aids in depositing nucleosomes on a DNA template in vitro [6] , and deletion of both Nhp6a and Nhp6b in S . cerevisiae results in a 20–30% decrease in histone levels in vivo [6] , suggesting that Nhp6a/b functions in regulating nucleosome stability or deposition . We found that previously measured Nhp6a binding was associated with nucleosomes that were preferentially lost in our experiments following H3 depletion ( Figure 3D ) [38] . This is generally consistent with a role for Nhp6 in nucleosome destabilization or deposition . To investigate if Nhp6a has different functions in promoters and gene bodies , we divided nucleosomes affected by H3 depletion into nucleosomes that fell into intergenic or genic regions using annotations from a recent study on the transcribed portion of the yeast genome [13] . For 23 of the 40 data sets shown in Figure 3 , the pattern of association with nucleosome behavior was similar between the intergenic and genic regions . That is , factors that were enriched or depleted in a given classification category were enriched or depleted , respectively , in both the intergenic and genic regions . However , the chromatin remodelers Rsc8 , Isw2 , and Nhp6a were among those that that showed the most striking differences in nucleosome behavior in intergenic regions versus gene bodies ( Figure S9 and Table S2 ) . Nhp6a binding was weakly associated with the “occupancy reduced” class in intergenic regions but was strongly associated with the “position altered” and the “occupancy reduced” classes in transcribed regions . This is consistent with a function for Nhp6a in nucleosome incorporation [6] and suggests that areas to which Nhp6a is recruited in transcribed regions may be especially sensitive to a reduction in the available histone pool . We found that the nucleosomes with reduced occupancy following Nhp6a/b deletion according to [6] were not the same as the nucleosomes classified as “reduced occupancy” in our study after H3 depletion . Only 299 out of the ∼7000 nucleosomes with reduced occupancy following Nhp6a/b deletion in [6] were held in common with the ∼2000 nucleosomes classified as reduced occupancy in our study , suggesting that the mechanisms underlying nucleosome loss in the two experiments are distinct . To examine this more closely , we compared the average Nhp6a binding from [38] to changes in nucleosome occupancy due to H3 depletion ( this study ) or to changes in nucleosome occupancy due to the absence of Nhp6a/b [6] . While there is a connection between Nhp6a binding and changes in occupancy following H3 depletion in our study ( Figure S10A ) , a connection between Nhp6a binding and nucleosomes lost in in the Nhp6a/b deletion was not apparent ( Figure S10C ) . Similarly , a significant enrichment for Nhp6a binding was found at nucleosomes that belonged to our “reduced occupancy” class ( compared to all of the nucleosomes with classified occupancies following H3 depletion; p<1E−15; Figure S10B ) . However , there was only a weak association between Nhp6a binding and nucleosomes lost in in the Nhp6a/b deletion strain ( p = 0 . 004; Figure S10D ) . The influence of DNA sequence on nucleosome occupancy has been determined directly by in vitro reconstitution of nucleosomes using naked yeast DNA [8] , [39] . We compared genome-wide nucleosome occupancy in wildtype and H3 shutoff cells to nucleosome occupancy measured in nucleosome reconstitution experiments . At 0 hours , both wildtype and H3 shutoff cells showed similar correlations to in vitro nucleosome reconstitution ( r = 0 . 69 and 0 . 67 , respectively ) . After 3 hours in dextrose , the wildtype correlation to the in vitro data decreased ( r = 0 . 59 ) , but the H3 shutoff strain's correlation increased ( r = 0 . 75; Figure 4A ) . To confirm that the decreased correlation in dextrose-grown wildtype cells was not specific to our experiment , we compared the correlation between cells grown in galactose and dextrose to the in vitro data from an independent strain and data set [8] . A similar decrease in correlation with in vitro data was observed between galactose-grown ( r = 0 . 774 ) and dextrose-grown ( r = 0 . 716 ) yeast ( Figure S11 ) . This suggests that nucleosome occupancy during growth in galactose ( as opposed to dextrose ) is more similar to the organization observed in vitro . More relevant to our study , H3 depletion in vivo ( which in this case occurs in dextrose ) results in a chromatin organization that is much more similar to the in vitro configuration than cells with the normal complement of nucleosomes grown in either dextrose or galactose . We next used a previously published DNA sequence-based model of nucleosome occupancy [8] to determine if the changes in a given nucleosome's position or occupancy after H3 depletion were influenced by the underlying DNA sequence . By definition , occupancy predictions based on DNA-sequence can change only if the nucleosome's position changes . Therefore , we calculated the change in a nucleosome's predicted occupancy based on its position before and after H3 depletion . As expected , the DNA-predicted occupancy value did not change at nucleosomes classified as having unchanged positions because the coordinates for the before and after locations were virtually identical . However , among the “position altered” nucleosomes , there was an average increase in the predicted occupancy at the position following histone depletion relative to the starting position ( Figure 4B ) . Approximately 25% of the occupancy-altered nucleosomes ( reduced or retained ) also showed a significant position change following H3 depletion . At these nucleosomes , the average predicted occupancy for the “before” and “after” positions corresponds with the actual observed change in nucleosome occupancy ( Figure 4B ) . To see if this trend extended to all of the nucleosomes with altered occupancy , we examined the change in the actual and predicted occupancies for all of the nucleosomes classified as having altered occupancy and having any degree of shift in position ( this group included all position classifications other than “position unchanged” , including “no call” ) . For both reduced and retained occupancy nucleosomes , the degree of nucleosome reduction or retention in vivo correlated with the degree of change in the DNA sequence-predicted nucleosome affinity ( Figure 4C ) . In a recent study that examined nucleosome loss due to Nhp6a/b depletion , the underlying nucleosome-affinity of the DNA also corresponded with changes in nucleosome occupancy [6] . Thus , there is strong evidence that under conditions of limiting histone concentrations , DNA sequence contributes directly to changes in nucleosome occupancy in vivo . We next revisited the hypothesis that gene expression changes in response to histone depletion are rooted in changes in chromatin organization [1] , [3]–[5] . We used RNA-seq to quantify the relative abundance of transcripts from wildtype and H3 shutoff cells after 3 hours in dextrose . Our RNA-seq measurements of relative expression correlated well with previously published microarray experiments from H4-depleted cells , despite differences in histone-depletion methodology and RNA detection methods ( r = 0 . 66 for expression arrays vs . RNA-seq RPMK ) . We used a BioConductor package , EdgeR , to analyze the RNA-seq data [40] . We detected 2453 de-repressed genes following histone depletion , compared to the 888 that were previously identified [5] . The number of genes with significantly decreased transcript levels following H3 depletion was 753 in this study , compared to the 569 reported previously for H4 depletion ( Table S3 ) [5] . Thus , the increased sensitivity of RNA-seq identified nearly half of yeast genes as having higher expression due to H3 depletion , three times the previous estimate . We divided the yeast genes into three groups based on expression changes: increased , normal , and decreased . In the promoters of genes with increased or normal expression following H3 depletion , we found an over-representation of “occupancy reduced” nucleosomes ( Figure 5 and Figure S12 ) . In contrast , nucleosomes in the promoters of genes with decreased or normal expression were not significantly associated with any class of H3-depletion response . In the gene body , genes with decreased expression tended to harbor “occupancy retained” nucleosomes , while those with increased expression harbored significantly fewer “occupancy retained” nucleosomes than expected ( Figure 5 and Figure S12 ) .
The experiments described above support the following main conclusions: ( 1 ) Depletion of histone H3 levels causes a defined subset of nucleosomes to alter their position and/or occupancy in vivo . ( 2 ) Nucleosomes that are preferentially lost tend to be located at promoters , and this , in turn , leads to de-repression of downstream genes . ( 3 ) Isw2 , an important ATP-dependent chromatin remodeler , is associated with stable nucleosome occupancy but altered position , especially when an adjacent nucleosome is destabilized . Such nucleosomal positioning shifts in the direction of the adjacent loss event are consistent with a nucleosome-centering activity for Isw2 in vivo , which to this point has been observed only in vitro . ( 4 ) Following nucleosome loss , the intrinsic DNA sequence preferences of nucleosomes have a greater influence on occupancy profiles , presumably due to reduced steric hindrance from adjacent nucleosomes . ( 5 ) Nhp6a is associated with preferentially lost nucleosomes in gene bodies . Deletion of Nhp6a/b causes a 20–30% reduction in histone abundance and altered transcription of approximately 10% of the yeast genome ( fold-change >1 . 5 and p<0 . 05 ) [6] . However , the sets of nucleosomes that we identify as being sensitive to H3 depletion are largely separate from the set identified as being destabilized by the loss of Nhp6a/b . This implies that the nucleosome loss events observed in the two studies may occur by independent mechanisms . Our approach used MNase-seq to map nucleosome position and relative occupancy before and after H3 depletion . One concern with using MNase to measure nucleosome position and occupancy is that the enzyme's ability to digest DNA can be influenced by the local histone occupancy , with regions of lower histone occupancy being more susceptible to MNase digestion . Following H3 depletion , we observed increased smearing in the MNase-digested DNA fragments but saw no change in the average fragment length or the number of nucleosomes called by our algorithm . Thus , despite the increased sensitivity to MNase , the mononucleosome properties were equivalent to wildtype cells after adjusting the MNase concentration . The strong correlations to biologically relevant annotations provide additional evidence that the position and occupancy comparisons we make based on the data are informative . We conclude that a combination of DNA-encoded nucleosome preference and chromatin composition regulate nucleosome occupancy and positional stability under conditions of limited histone protein availability . This , in turn , dictates which genes are most susceptible to a loss of regulatory fidelity . Most importantly , our data point to the factors likely to influence nucleosome stability under normal growth conditions , and to the specific genomic locations at which they are likely to act . This information serves as a platform for more detailed investigations into the mechanisms of nucleosome regulation .
Wildtype ( YEF473A ) and H3 shutoff ( DCB200 . 1 ) cells were maintained on agar plates containing 2% galactose [11] . For experiments , the cells were grown in liquid media with the indicated carbon source . Logarithmically growing cells were diluted to an OD600 of 0 . 0375 in 1 . 25 L of fresh YPGal media ( 1% yeast extract , 2% peptone , 2% galactose ) and grown for 16 hours . Cells were collected via suction filtration with a 0 . 2 µm filter and washed with approximately 100 mL YPD ( 1% yeast extract , 2% peptone , 2% dextrose ) before being resuspended in 1 . 25 L of YPD media . After switching the carbon source , 1 L of cells was immediately transferred to a fresh flask containing formaldehyde for our “0 hour” MNase digested time point ( next section ) . We added 750 mL of fresh YPD to the remaining 250 mL , and the cells were grown at 30°C with shaking for 3 hours prior to collection for the final time point . After collection , the cells were crosslinked with formaldehyde , the cell wall was digested with lyticase , and the DNA was digested with a titration of MNase . The resulting DNA was electrophoresed on an agarose gel , and the mono-nucleosome band was excised for sequencing . See Text S1 for additional details . The raw sequencing data is available at GEO accession number GSE29291 ( single-end sequences ) and GSE29292 ( paired-end sequences ) . RNA was isolated using the hot acidic phenol method [41] from wildtype and H3 shutoff cells grown as described above and transitioned to dextrose-containing media for 3 hours . RNA was further purified using the RNEasy Mini Kit ( Qiagen 74104 ) to remove trace amounts of phenol . Ribosomal RNA was removed using the RiboMinus system ( Invitrogen K155003 ) . The quality of the RNA and the absence of ribosomal RNA were confirmed by gel electrophoresis . We then fragmented 4 . 5 µg of RNA using Ambion RNA fragmentation reagent ( Ambion AM8740 ) at 70°C for 5 minutes and used the resulting RNA fragments as input for double-strand cDNA synthesis using a double-stranded cDNA synthesis kit ( Invitrogen 11917-010 ) with random priming ( Invitrogen 48190-011 ) . The resulting cDNA was then prepared for Illumina sequencing . The raw RNA-seq data is available at GEO accession number GSE29293 . Samples were prepared for either single-end ( two H3 shutoff , one wildtype MNase-seq replicates , and three replicates each of H3 shutoff and wildtype RNA data ) or paired-end ( two H3 shutoff and two wildtype replicates ) sequencing using established protocols ( Text S1 ) . All libraries were sequenced using an Illumina Genome Analyzer IIx . Reads were aligned to the sacCer1 build of the yeast genome using Bowtie v . 0 . 12 . 6 with default settings . Single-end reads were extended to the average fragment size in each experiment , which was calculated based on the distribution of reads on the Watson and Crick strands . Paired-end reads were required to have matching ends within 100–200 bases . Reads aligning to the rDNA locus ( chr12 450 , 000–472 , 500 ) were removed . For analysis , all replicates were normalized with regard to sequencing depth by randomly selecting 5 million aligned reads per sample . A read center density map was created for each experiment using the center point of each extended ( single-end ) or paired ( paired-end ) read . The read-center density map was Gaussian smoothed with a standard deviation ( s . d . ) of 10 bp and a window of 3 s . d . The overall dyad density was visualized for Figure 1 using Matrix2png [42] . To identify discrete nucleosomes , we repeated the following process: 1 ) The smoothed read-center density maximum was set as the center of a nucleosome . 2 ) The size of the region protected by the nucleosome was calculated as the average length of all extended/paired reads that covered the center base . 3 ) The s . d . of the nucleosome center in bases ( the “fuzziness” ) of the nucleosome was calculated using the number and locations of read centers that fell within the nucleosome's protected region . 4 ) The smoothed read-center density for all bases within one protected region of the nucleosome center was set to zero to prevent calling overlapping nucleosomes in successive rounds of nucleosome calling . 5 ) The nucleosome's occupancy was defined as the number of read centers falling within 50 bp on either side of the nucleosome center . This process was repeated until no additional nucleosomes could be called . Nucleosome positions at 0 and 3 hours in each replicate were required to overlap by 30 bp , and the coverage in each case was required to be greater than 5 reads . The nucleosome center positions at 0 and 3 hours were subjected to a t-test , and a p-value was calculated . The occupancy was compared using the binomial distribution test on the number of read centers in the 100-bp window centered on the nucleosomes' respective centers . In both cases , the resulting p-values were Bonferroni corrected based on the total number of nucleosomes compared in that replicate . Nucleosomes for each replicate were classified as having either unchanged ( p>0 . 2 ) or altered ( p<0 . 01 ) position and reduced ( p<0 . 01 ) , unchanged ( p>0 . 2 and <0 . 8 ) , or retained ( p>0 . 99 ) occupancy relative to the 0 hour position and occupancy . Nucleosomes with p-values between 0 . 01 and 0 . 2 or 0 . 8 and 0 . 99 were left unclassified ( “no call” ) and were not used in downstream analyses . Nucleosome classification was compared between replicates , and only nucleosomes that were similarly classified in two or more replicates were used for further analyses . In addition , occupancy categories were restricted to nucleosomes that were not oppositely classified in any replicates ( i . e . , no nucleosomes were considered that were classified as occupancy reduced and occupancy retained in different replicates ) . Nucleosomes exhibiting behavior that was dependent on histone H3 depletion were identified by comparing the wildtype and H3 shutoff strains and removing any nucleosomes that were similarly classified in the wildtype experiments . The majority of data sets used for comparative analysis were also generated from cells grown in glucose at an OD600 of ∼1 . If the growth condition in the study was substantially different , it is noted in parentheses . We compared the nucleosome classes to available genome-wide data sets for the following marks that were downloaded from ChromatinDB ( www . bioinformatics2 . wsu . edu/cgi-bin/ChromatinDB/cgi/downloader_select . pl ) as bulk histone occupancy normalized data: H2AK7ac , H2BK11ac , H2BK16ac , H3K9ac , H3K14ac , H3K18ac , H3K23ac , H3K27ac , H4K8ac , H4K12ac and H4K16ac [43]; H3 N-terminal ac and H4 N-terminal acetylation [44]; H3K36me2 [45]; H2A . zK14ac [46]; H3K56ac [34] , [47]; and H3 and H4 occupancy [48] . We also used H3K36ac [49] and H3K4me3 [50] data normalized to H3 occupancy from [48] . H3K4me2 , H3K4me4 , and H3R2me2a H3-normalized data is from [51] . H3 turnover data is from [20] ( grown in raffinose and galactose ) . Sir2 data is from [52] . Tup1 data is from [21] ( OD600∼0 . 6–0 . 8 ) . Rpo21 , Bdf1 , Rap1 , Reb1 , Vps72 , and Srm1 data is from [22] . Rsc8 data is from [53] . Isw2K215R data is from [32] ( OD600∼0 . 7 ) . Nhp6a data is from [38] . Rpd3 data is from [54] . High resolution H2A . z data is from [55] . For Figure 3 and Figure S7 , the average enrichment of the comparison data for nucleosomes in each class was calculated . To calculate significance , the average enrichment for an identical number of nucleosomes selected from the entire set of H3 depletion-dependent nucleosomes was calculated 100 times , and the standard deviation of the random average was used to assign a z-score for the actual value . The data is reported as the −log10 of the p-value of the z-score ( raw z-scores are available in Table S2 and Table S3 ) . H3-depletion affected histones identified in this study were compared to nucleosomes identified in a previous study in which Nhp6a and Nhp6b were deleted in a yeast strain [6] . To compare changes in occupancy between the two experiment sets , we used the log2 ratio of the Nhp6a/b deletion strain's score to the wildtype score for the Celona et al . data [6] and the previously described classification of H3-depletion dependent occupancy . RNA-seq reads from three biological replicates each of H3 shutoff and wildtype cells were mapped to the S . cerevisiae sacCer1 genome using Bowtie ( v . 0 . 12 . 6 . 0 ) and TopHat ( v . 1 . 1 . 0 ) with a maximum intron size of 1 kb . Samtools ( v . 0 . 1 . 8 . 0 ) was used to determine read pileups at each base in the genome . The total coverage in bases reported as transcribed [13] was divided by the read length ( 28 bp ) and was used as the coverage for determining differential gene expression using EdgeR [40] . Genes with a reported p-value<0 . 001 were considered to be differentially expressed . The number of nucleosomes in each class that overlapped with each gene and upstream , non-transcribed region was determined . To calculate significance , a number of nucleosomes equal to those in the class were randomly chosen from all classified nucleosomes , and this subset was tested for overlap with the region . This was repeated 100 times , and the −log10 p-value of the z-score is reported . | Chromatin is formed by wrapping 146 bp of DNA around a disc-shaped complex of proteins called histones . These protein–DNA structures are known as nucleosomes . Nucleosomes help to regulate gene transcription , because nucleosomes compete with transcription factors for access to DNA . The precise positioning and level of nucleosome occupancy are known to be vital for transcriptional regulation , but the mechanisms that regulate the position and occupancy of nucleosomes are not fully understood . Recently , many studies have focused on the role of DNA sequence and chromatin remodeling proteins . Here , we manipulate the concentration of histone proteins in the cell to determine which nucleosomes are most susceptible to changes in occupancy and position . We find that the chromatin-associated proteins Sir2 and Tup1 , and the chromatin remodelers Isw2 and Rsc8 , are associated with stabilized nucleosomes . Histone acetylation and incorporation of the histone variant H2A . z are the factors most highly associated with destabilized nucleosomes . Certain DNA sequence properties also contribute to stability . The data identify factors likely to influence nucleosome stability and show a direct link between changes in chromatin and changes in transcription upon histone depletion . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genome",
"expression",
"analysis",
"genomics",
"functional",
"genomics",
"molecular",
"cell",
"biology",
"molecular",
"biology",
"gene",
"regulation",
"gene",
"expression",
"molecular",
"genetics",
"biology",
"computational",
"biology",
"chromatin",
"genetics",
"and",
... | 2012 | In Vivo Effects of Histone H3 Depletion on Nucleosome Occupancy and Position in Saccharomyces cerevisiae |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.