pmcid stringlengths 6 6 | title stringlengths 9 374 | abstract stringlengths 2 4.62k ⌀ | fulltext stringlengths 167 106k | file_path stringlengths 64 64 |
|---|---|---|---|---|
518965 | A human in vitro model system for investigating genome-wide host responses to SARS coronavirus infection | Background The molecular basis of severe acute respiratory syndrome (SARS) coronavirus (CoV) induced pathology is still largely unclear. Many SARS patients suffer respiratory distress brought on by interstitial infiltration and frequently show peripheral blood lymphopenia and occasional leucopenia. One possible cause of this could be interstitial inflammation, following a localized host response. In this study, we therefore examine the immune response of SARS-CoV in human peripheral blood mononuclear cells (PBMCs) over the first 24 hours. Methods PBMCs from normal healthy donors were inoculated in vitro with SARS-CoV and the viral replication kinetics was studied by real-time quantitative assays. SARS-CoV specific gene expression changes were examined by high-density oligonucleotide array analysis. Results We observed that SARS-CoV was capable of infecting and replicating in PBMCs and the kinetics of viral replication was variable among the donors. SARS-CoV antibody binding assays indicated that SARS specific antibodies inhibited SARS-CoV viral replication. Array data showed monocyte-macrophage cell activation, coagulation pathway upregulation and cytokine production together with lung trafficking chemokines such as IL8 and IL17, possibly activated through the TLR9 signaling pathway; that mimicked clinical features of the disease. Conclusions The identification of human blood mononuclear cells as a direct target of SARS-CoV in the model system described here provides a new insight into disease pathology and a tool for investigating the host response and mechanisms of pathogenesis. | Background The causative agent for SARS has been identified as a novel coronavirus [ 1 - 3 ] with genome sequence revealing no strong homology to existing known coronaviruses [ 4 - 6 ]. Coronaviruses belong to the family of enveloped viruses called Coronaviridae , and have the largest known single-stranded viral RNA genomes (27 to 32 kb). Coronaviruses, have both "early" and "late" phases of gene expression. Regulatory proteins are synthesized as "early" non-structural proteins, while the structural proteins are synthesized as "late" proteins. "Late" structural proteins are usually required in greater amounts thus, there is a necessity to regulate the expression of the viral genes quantitatively. After the viral entry via endocytosis or through specific receptors, the 5'-end of the viral genome is translated directly giving rise to twenty-three viral proteins, including the RNA dependent RNA polymerase (RdRp), and other functional products involved in transcription, replication, viral assembly and cell death. Coronaviruses can be classified into species and three major antigenic groups based on, serology, natural hosts, monoclonal antibody recognition and nucleotide sequencing [ 7 ]. Most coronaviruses have restricted host ranges as they infect only one host species or, at most, a few related species, they are an important group of animal pathogens. Group one (I) includes human coronavirus 229E (HCoV), porcine transmissible gastro-enteritis virus (TGEV) and feline enteric coronavirus (FECoV). Group two (II) includes bovine coronavirus (BCoV), murine hepatitis virus (MHV), and HCoV-OC43; and Group three (III) includes avian infectious bronchitis virus (IBV) [ 7 ]. Some coronaviruses like HCoV have restricted tissue tropism, including macrophages [ 8 ], although most strains that infect humans cause only mild respiratory infections. However, SARS has rapidly caused a world-wide problem. The earliest known cases of SARS was reported in Guandong Province, China in November 2002, becoming more widespread by March 2003, when it was introduced to Canada, Singapore, Taiwan and Vietnam via Hong Kong. The largest number of infected patients has been in China with a worldwide incidence totalling more than 8,400 by July 2003. Infection by the virus induces high morbidity and mortality, the latter being estimated at 15% by the World Health Organisation. SARS is characterized by high fever, non-productive cough or dyspnea and in many cases may progress to generalized, interstitial infiltrates in the lung, thus needing intubation and mechanical ventilation [ 2 ]. The characteristic compression of alveolar sacs seen in atypical pneumonia is largely due to fluid build up outside the alveoli. One possible cause of this could be interstitial inflammation, following a localised host response. To date, the details of the host response to SARS-CoV infection is still largely unknown and consequently the most appropriate treatment regime remains to be established. Typically a pro-inflammatory cytokine profile (up regulated TNFα, Il1, IL6 and IFNs) is seen in viral infections such as influenza [ 9 ], together with perhaps limited amounts of IL8 and other chemokines [ 10 ] that may depend on which cell type is infected [ 11 ]. In experimental systems the immediate innate immune response has been shown to be directed by the monocyte-macrophage-dendritic lineage to a range of different organisms [ 12 , 13 ] and consists of a core set of pathways common to all, together with pathogen specific pathways. This data points to critical time points in the response, with the first 12 hours representing primary events while longer periods the consequence of this activity and a secondary (perhaps larger) cascade of responses. We postulated that the pulmonary damage in SARS may not be a direct effect of the virus on the alveoli, but represents a secondary effect of cytokines or other factors proximal to but not from the lung tissue mediated by the host either as the primary or secondary response [ 2 , 14 ]. In this study, we have addressed this question by developing a human in vitro model system that will in the future allow detailed investigations of the host response to be made. Methods Cell culture and virus infection PBMCs were obtained by Ficoll-Hypaque separation of whole blood. 2 × 10 5 PBMCs were seeded into each well of a 24-well culture plate, 0.5 ml of complete RPMI-1640 (Life Technologies-Invitrogen, USA) added to each and cultured overnight at 37°C (5% CO 2 ). A seed stock of SARS-CoV (strain SIN 2774) passaged in Vero E6 cells was used for infection. Vero E6 culture supernatants were added to each well in 50 μl volume at a concentration of 0.1 or 0.01 MOI (based on plaque forming units) and a control plate (media only). Each culture was set up in duplicate. After 4 hours of incubation, one set of the duplicate wells of the control plate and a 4 hours incubation plate were harvested while the rest received 0.5 ml media top-up and incubated for a further 2, 4, 6 and 8 days. Cell harvest and RNA isolation Harvesting was performed by gently flushing the wells with a Pasteur pipette to removed non-adherent cells, followed by a rinse of 1 ml RPMI. The rinsed fraction was pooled with the first harvest aliquot and spun at 1500 rpm. The cell pellet was washed twice with 2 ml RPMI to remove virus in the supernatant. 1.5 ml Trizol (Invitrogen, USA) was then added to the adherent cell fraction as well as the non-adherent cell fraction to lyse cells and stabilize the RNA. Extraction of total RNA was then performed following manufacturer's protocol and the resultant RNA dissolved in 40 μl water. Real-time quantitative polymerase chain reaction (PCR) The amount of SARS-CoV in each cell fraction was measured by real-time quantitative PCR assay. 2 μl of RNA was reverse transcribed and amplified in 20 μl using 0.9 μM each of forward (5'-GGTTGGGAT TATCCAAAATGTGA-3') and reverse (5'-AGAACAAGAGAGGCCATTATCCTAAG-3') primers, and 0.25 μM of TaqMan ® MGB probe: 5'-(6-FAM)AGAGCCATGCCTAACAT(NFQ)-3') in a one step PCR using master mix from Applied Biosystems (USA) according to manufacturers' recommendations. Reactions were performed using an ABI PRISM 7900 sequence detection system (48°C for 30 min, followed by 95°C for 10 min and 40 cycles of 95°C for 15 sec and 60°C for 1 min) and quantitation achieved using standard curves generated from in vitro transcribed RNA. High density oligonucleotide array hybridization At each time point (4 hours, 8 hours, 12 hours, 24 hours), 5 × 10 7 cells of mock-infected and infected cells were harvested and lysed using Trizol (Invitrogen, USA). Total RNA was isolated according to the manufacturer's recommendation. Quality of the total RNA was judged from the ratio between 28S and 18S RNA after agarose gel electrophoresis. 20 μg of total RNA was labeled with Cy-3 or Cy-5 using the Superscript II reverse-transcription kit (Invitrogen, USA) and hybridization was carried out overnight (16 hours) at 42°C on high-density oligonucleotide arrays (~19,200 gene features, Compugen) using universal human reference (Stratagene, USA) as a reference. Hybridized arrays were scanned at 5 μm resolution on a GenePix 4000A scanner (Axon Instruments) with variable photo-multiplier tube voltage to obtain maximal signal intensities, and the resulting images were analyzed via GenePix Pro v4.0 (Axon Instruments) as described in the manual. Microarray data analysis Raw data were analyzed on GenePix analysis software version 4.0 (Axon Instruments) and uploaded to a relational database. The logarithmic expression ratio for a spot on each array was normalized by subtracting the median logarithmic ratio for the same array. Data were filtered to exclude spots with a size of less than 25 μm and any poor quality or missing spots. Since the correlation of the overall data from reciprocal labeling was good, values obtained from reciprocal labeling experiments were averaged. In addition, the data were distilled to the set of gene features that were present at all 4 time points in both the viral infected samples and the negative controls. The results were represented as the logarithmic ratio of gene expression between the viral infected samples and their corresponding negative controls at the various time points. Application of these filters resulted in the inclusion of ~12,900 of the total ~19,200 gene features in subsequent analyses. To discover patterns of gene expression, the values associated with each gene feature f were translated so that their means were zero. Similar genes, whose translated gene-features exhibited same induction-repression pattern, were grouped together. Genes g i , g j were said to be similar if they satisfied the following condition: , where U ( x ) = 1 if x > 0, U ( x ) = 0 otherwise, where g it and g jt denote the translated values of gene features g i , g j at time t respectively; and N is the number of time points for which the expression of a gene was observed. Similar genes, based on the above criteria, are grouped together. Within each group, the genes were ordered in the descending order of their expression range (defined as the difference between the maximum and minimum ratios of gene expression). This algorithm is a special case of the Friendly Neighbor algorithm currently under development. The final plots were generated using the original expression ratios while preserving the clustering and ordering discovered by the above algorithm. To determine whether a gene observed to be responsive could appear merely by chance, 100,000 expression profiles were generated by randomly sampling the expression ratios from the entire dataset with replacement. The P value of a gene is the fraction of the random profiles whose logarithmic expression range is as good as, or better than that of the selected gene. Results and Discussion Kinetics of SARS-CoV replication We obtained PBMCs from 6 healthy volunteers by Ficoll-Hypaque separation of whole blood. Of the 6 donor PBMCs tested, all were able to support SARS-CoV replication when infected with multiplicity of infection (MOI) of 0.1. The first sampling, taken from cells infected for 4 hours, showed an average copy number of 32 × 10 3 (Fig. 1A ) and represents the initial inoculum level. Over the course of the next 8 days, there was a steady rise in viral load, reaching as high as 480 × 10 3 copies per well in one donor, which could only be explained by active replication of the SARS-CoV intracellularly. This work is supported by recent in vivo evidence suggesting that SARS-CoV may have infected and replicated within PBMCs of SARS patients [ 15 ] and cells from humans and animals [ 16 ]. An indication of the PBMCs lineage involvement was provided by repeating the experiment using the monocyte-macrophage cell line THP-1 [ 17 ], in which viral replication was similar to the primary cell culture over the first 4 days (Fig. 1B ). In the primary cultures, the non-adherent cell fraction which comprises mainly lymphocytes and granulocytes showed dramatically less viral replication in our assay as did all cells infected at MOI of 0.01 (Fig. 1A ). The kinetics of viral replication was variable among the 6 donors (Fig. 1C ). There was a lag phase of 2 days in the case of donors a, c and e; and 4 days for donors b, d and f before any significant increase could be detected. The viral replication generally peaked at either day 4 or day 6. The exception was donor b, in which the virus seemed to replicate at a much slower pace compared to the other 5 donor samples. Equally interesting was the different levels of virus attained. Donor d seems to stand out from the rest, reaching a peak of 480 × 10 3 copies per well which is 4 times more than that attained by donor e, with 120 × 10 3 copies per well. Such variation strongly suggests that there is an underlying host-pathogen interaction influencing the kinetics of SARS-CoV replication efficiency. These in vitro observations may reflect the wide range of patient outcomes after SARS-CoV infection [ 18 ]. Antibody blocking experiments were also performed in which SARS-CoV was pre-incubated with convalescent patient sera for 30 minutes before introduction to the PBMCs and after a 4 day incubation period, the adherent cell fraction was harvested and assayed for SARS-CoV viral titer. Results clearly showed that even at high dilution, convalescent sera inhibited SARS viral replication (data not shown), presumably by blocking viral entry. This supports other reports indicating that SARS-CoV is not endocytosed through antibody mediated mechanisms and confirms a protective role for antibodies elicited either by the infection or through immunization [ 19 , 20 ]. SARS-CoV specific gene expression changes To further elucidate the molecular processes of SARS-CoV infection, PBMCs from 3 healthy individuals were infected separately in vitro with SARS-CoV (0.1 MOI) and harvested at 4 hours, 8 hours, 12 hours and 24 hours time intervals post-infection. As controls, uninfected aliquots of the same PBMCs were also harvested at the corresponding time points. Total RNA extracted from the PBMCs of the 3 individuals were pooled, labeled and hybridized to human oligonucleotide arrays consisting of ~19,200 gene features. Reciprocal dye swap replicate hybridizations were performed to minimize technical noise. Analysis of variance in expression levels for each gene across all the time points indicated the ~1200 genes which showed the largest variability (Fig. 2A and 2B ). In order to focus the analysis, we queried the entire data set for genes related to the immune response by keyword searches on their gene ontology descriptions with the aim of describing the specific host-pathogen interaction. In common with other studies of respiratory pathogens [ 9 - 13 ], our data points towards two critical time points in the response, with the first 12 hours representing a primary pro-inflammatory cytokine profile while longer periods represent the consequence of this activity and a secondary cascade of responses [ 9 - 13 ]. We observed that within the first 12 hours of SARS-CoV infection, evidence of this monocyte-macrophage activation was seen, indicated by enhanced expression of CD14, TLR9 plus NFKβ1 and GATA signaling (Fig. 2C and Table 1 ). In addition, the MRC2 endocytotic receptor was upregulated as was the complement pathway (C1q, C6). Taken together, these data suggest an early activation of the innate immunity pathway. This activation was accompanied by an unusual cytokine transcriptional profile (Fig. 2C and Table 1 ). While IL1β (up regulated for the first 12 hours) would be expected following macrophage activation [ 21 ], TNFα, IFNγ and IL6 were noted by their surprisingly low level of expression. This is in spite of the presence of elevated IL19 which is thought to enhance their up regulation [ 22 ]. In some clinical investigation, concentrations of TNF and IL 6 measured during active disease were found to be relatively low [ 23 , 24 ], reflecting our findings. This paper did not report on IFN levels, however, we found them to be low (Supplementary figure [see Additional file 1 ]). This is of particular interest as IFNs have been shown to have significant anti-SARS-CoV effects [ 25 ]. Such effects suggest that alteration of the IFN response and perhaps other immune modulators might provide opportunity for novel treatment and management regimes for SARS patients to be developed. A number of CC chemokines (CCL4, CCL20, CCL22, CCL25, CCL27) and their receptors (CCR4 and CCR7) were highly expressed in response to the infection (Fig. 2C and Table 1 ), indicating a rapid mobilization and increased trafficking, in particular of the monocyte-macrophage lineage very early on in the infection [ 26 ]. CXC chemokines (CXCL9, CXCL12) were also highly expressed suggesting significant increase in neutrophil homing as well. These are likely to be lung directed as IL8 and IL17 were also highly expressed [ 27 - 32 ]. Specific trafficking of these cells to the lung may account for the localized nature of the response [ 33 ]. Surprisingly, a number of blood coagulation genes were highly expressed early during our in vitro infection (Fig. 2C and Table 1 ), in particular TBXAS, which has been implicated in vasoconstriction, platelet aggregation, membrane lysis and increased permeability [ 34 , 35 ]; fibrin (FGB and FGG) and the coagulation pathway directly (SERPINs D1 and A3 together with Factors 10, 3 and 2). This gives a pro-coagulation profile, which mimics the clinical-pathological observations: at autopsy, many SARS patients have unusually disseminated small vessel thromboses in the lungs without evidence of disseminated intravascular coagulation [ 1 , 36 ]. Again, these expression profiles provide an experimental framework to explore an important aspect of SARS pathobiology and treatment. It is interesting to note that the TLR9 was highly expressed in comparison to other TLR receptors, implying some degree of TLR specificity for the virus (Fig. 3A ). TLR9 is known to respond to CpG signaling motifs (GTCGTT) [ 37 - 39 ] and one possibility is that the virus is activating directly through this mechanism. In support of this, we found that the SARS-CoV viral sequence contains the highest number (7 copies) of such specific signaling motifs compared to other coronaviruses and significantly more than several other viruses involved in respiratory diseases (Fig. 3B ). It is conceivable that TLR9 may be aiding host recognition of the virus via the CpG groups and contributing to the initiation of the innate host inflammatory response. An alternative explanation is that TLR9 is being stimulated by mechanisms unrelated to CpG recognition. The emerging picture from this study implicates a central role for the immune response in the pathobiology of a SARS infection. While detailed in vivo studies of the host response are now required, the in vitro model described here will allow responses to specific modulators (such as therapeutics) to be investigated. In future developments of the model, it will be interesting to compare the host response to different SARS-CoV isolates with inactivated preparations of the virus. In other diseases, in vitro models have revealed a number of key processes relevant to the clinical diseases [ 9 , 12 , 13 ] and it is likely that the responses identified here will prove to be equally important. Although some clinical parameters have now been used as prognostic markers [ 40 - 42 ], further study of the regulatory mechanisms for chemokine-cytokine production will likely improve their accuracy and perhaps allow development of new treatment protocols. Competing interests None declared. Authors' contributions LFPN, MLH, ETL and REC conceived the study, its design and coordination, results analysis and drafted the manuscript. EEO and KFT carried out the virus infections. SYN, KRKM, VNV and JMC were involved in the array and statistical analysis. JT carried out the real-time PCR assays. Pre-publication history The pre-publication history for this paper can be accessed here: Supplementary Material Additional File 1 List of all immune related genes after SARS-CoV infection Comprehensive list of 1087 immune related genes that were altered in PBMCs in response to SARS-CoV infection at 4 hours, 8 hours, 12 hours, and 24 hours. Genes were grouped and ordered using the algorithm described in Methods. Rows represent individual genes, columns represent individual time points. Each cell in the matrix represents the mean expression level from 3 subjects for a gene feature at a particular time point (non-infected PBMCs responses have been subtracted from infected responses). The red and green color bars reflect high and low expression levels respectively, while black indicates equivalent expression level. The magnitude of the log-transformed ratio is reflected by the degree of color saturation. The line graph indicates the average expression ratios for each group. The area above the axis indicates upregulation, while the area under the axis means downregulation. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC518965.xml |
314479 | Evolution of Primate Sense of Smell and Full Trichromatic Color Vision | null | Conventional wisdom says that people deficient in one sense—such as vision or hearing—often acquire heightened acuity in another. And some studies support this notion by showing that areas of the brain known to control vision can respond to other forms of sensory stimuli in persons without sight. These adjustments, of course, take place over the lifetime of an individual. Now it appears that similar adjustments may occur over evolutionary time. Investigating the deterioration of olfactory receptor (OR) genes in primates, Yoav Gilad and his colleagues at the Max Planck Institute for Evolutionary Anthropology in Germany and the Weizmann Institute in Israel found a correlation between the loss of OR genes and the acquisition of full trichromatic color vision. OR genes—the molecular basis for the sense of smell—form the largest gene superfamily in mammalian genomes. But a high percentage of these genes are “pseudogenes,” DNA sequences that are remnants of genes that are no longer functional. Following an evolutionary “use-it-or-lose-it” rule, pseudogenes tend to evolve in larger gene families where there's no selective advantage in having, say, 100 versus 120 genes. While humans, nonhuman primates, and mice have roughly the same number of OR genes, in humans a much higher percentage of these are pseudogenes, at 60%, while nonhuman apes have about 30%, and the mouse has about 20%. Reliance on the sense of smell, it appears, decreases for animals that develop a dependence on other senses, such as hearing or sight, to survive. In characterizing this high proportion of pseudogenes, Yoav Gilad et al. asked: Is this characteristic of all primates? If not, at what point in primate evolution did the increase occur? Looking at 19 primate species—including one human, four apes, six Old World monkeys, seven New World monkeys, and one prosimian—Gilad et al. randomly sequenced 100 distinct OR genes from each of the species. The team found that Old World monkeys had roughly the same percentage of OR pseudogenes as nonhuman apes, but a much higher percentage than New World monkeys—except for one, the howler monkey. The percentage of OR pseudogenes in the howler monkey was much closer to that seen in the Old World monkeys and apes than in its New World cousins. The sense of smell, it appears, deteriorated both in the ape and Old World monkey lineage and in the howler monkey lineage. Since Old World monkeys, apes, and the howler monkey do not share an exclusive common ancestor, this deterioration must have evolved independently in both groups. Surprisingly, howler monkeys share another sensory feature with apes and Old World monkeys: trichromatic color vision. In trichromatic color vision, three retinal protein pigments, called opsins, absorb various wavelengths of light, which the brain processes to produce full-color images. Apes and Old World monkeys carry three opsin genes, and most New World monkeys carry only two, though females can sometimes have three. Only howler monkeys routinely have three genes occurring in both sexes. Thus, full trichromatic vision evolved twice in primates—once in the common ancestor of apes and Old World monkeys, about 23 million years ago, and once in the howler monkey lineage, about 7–16 million years ago. The evolution of color vision, the authors propose, coincided with a growing complement of OR pseudogenes and a deterioration of the sense of smell. Gilad et al. suggest that investigating the types of visual cues required for finding food may shed light on the nature of this connection. Howler monkey | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC314479.xml |
555736 | A configuration space of homologous proteins conserving mutual information and allowing a phylogeny inference based on pair-wise Z-score probabilities | Background Popular methods to reconstruct molecular phylogenies are based on multiple sequence alignments, in which addition or removal of data may change the resulting tree topology. We have sought a representation of homologous proteins that would conserve the information of pair-wise sequence alignments, respect probabilistic properties of Z-scores (Monte Carlo methods applied to pair-wise comparisons) and be the basis for a novel method of consistent and stable phylogenetic reconstruction. Results We have built up a spatial representation of protein sequences using concepts from particle physics (configuration space) and respecting a frame of constraints deduced from pair-wise alignment score properties in information theory. The obtained configuration space of homologous proteins (CSHP) allows the representation of real and shuffled sequences, and thereupon an expression of the TULIP theorem for Z-score probabilities. Based on the CSHP, we propose a phylogeny reconstruction using Z-scores. Deduced trees, called TULIP trees, are consistent with multiple-alignment based trees. Furthermore, the TULIP tree reconstruction method provides a solution for some previously reported incongruent results, such as the apicomplexan enolase phylogeny. Conclusion The CSHP is a unified model that conserves mutual information between proteins in the way physical models conserve energy. Applications include the reconstruction of evolutionary consistent and robust trees, the topology of which is based on a spatial representation that is not reordered after addition or removal of sequences. The CSHP and its assigned phylogenetic topology, provide a powerful and easily updated representation for massive pair-wise genome comparisons based on Z-score computations. | Background Past events that gave birth to biological entities can be tentatively reconstructed based on collections of descriptors traced in ancient or present-day creatures. Using genomic sequences, an estimate of the relative time separating branching events, previously supported by geological records, could be formalized using mathematical models. The use of proteins for evolutionary reconstructions was vastly explored as soon as the first amino acid sequences were made available [ 1 - 9 ]. The rich biological information contained in protein sequences stems from their being, on the one hand, translation of genes that reflect the history of genetic events to which the species has been subjected, and on the other hand, effectors of the functions constituting a living creature [ 10 ] Since protein sequences are encoded in a 20-amino acid alphabet, they are also considered to embody more information-per-site than DNA or RNA [ 11 ]; they also exhibit smaller compositional trends [ 12 , 13 ]. When compared, sequences that share substantial features are considered as possible homologues [ 14 ], based on the fundamental postulate that can be simply stated as "the closer in the evolution, the more alike and conversely, the more alike, probably the closer in the evolution". As summarized by Otu and Sayood [ 15 ], the techniques of molecular phylogenetic analyses can be divided into two groups. In the first case, a matrix representing the distance between each pair of sequences is calculated and then transformed into a tree. In the second case, a tree is found that can best explain the observed sequences under evolutionary assumptions, after evaluation of the fitness of different topologies. Some of the approaches in the first category utilize distance measures [ 16 - 19 ] with different models of nucleotide substitution or amino acid replacement. The second category can further be divided into two groups based on the optimality criterion used in tree evaluation: parsimony [ 20 , 21 ] and maximum likelihood methods [ 22 , 23 ]. For a detailed comparison of these methods see [ 24 ]. In phylogeny inference based on distance methods, features separating related proteins are used to estimate an observed distance, also called the p-distance, the simplest measure of which is just the number of different sites between proteins. Divergence time ( t ), also called genetic distance or evolutionary time, is calculated from the p-distance, depending on assumptions derived from evolutionary models [ 11 , 24 ]. For example, the assumption that mutational events happen with equal probability at each site of any sequence leads to the molecular clock model [ 2 ]. Although widely used, it is well-known to be unrealistic and numerous corrections have been proposed to refine it [ 19 , 25 , 26 ]. By definition, the distance matrix is given as T = ( t ab ) where a and b represent the homologous sequences from the analyzed dataset. Tree reconstruction algorithms are then applied to these matrices [ 11 , 24 ]. Eventually, phylogenetic trees corresponding to the classified sequences are statistically evaluated with bootstrap methods and, when available, calibrated using dated fossils [ 25 , 26 ]. Doolittle et al. [ 27 ] have proposed methods for converting amino acid alignment scores into measures of evolutionary time. Similarity between amino acids [ 28 - 30 ] provides a way to weight and score alignments [ 31 ]. In practice the optimal alignment of two sequences ( a and b ) is determined from the optimal score s ( a , b ) [ 25 , 27 ], computed with a dynamic programming procedure [ 32 , 33 ]. In aligned sequences, conservation is measured at identical sites, whereas variation is scaled at substituted sites. To estimate the variation/conservation balance, the p-distance can be given as a function of f id , the fraction of identical residues: p-distance = 1 - f id . To take into account that multiple mutations can happen at the same site, an expression of f id was proposed by Doolittle et al. [ 27 ] using s ( a *, b *), the score obtained from randomized a and b sequences [ 34 ] and s id , the average score of the sequences compared with themselves [ 19 , 25 , 27 ]: To connect pair-wise alignments and phylogeny, divergence time has been approximated: t ( a , b ) = - λ log[ f id ( a , b )] (2) introducing a Poisson correction [ 2 ] as a reasonable stochastic law relating amino acid changes and elapsed time. As mentioned earlier, adjustments and corrections of equation (2) were proposed to fit more realistically the complexity of evolution [ 11 , 25 , 35 ]. This attempt of unification helped reconstructing phylogeny of major lineages [ 27 ]. However, detailed phylogenic trees obtained from evolutionary close sequences are not satisfactory. In practice, phylogenies are reconstructed based on multiple alignments. Multiple alignment based (MAB) trees are re-calculated when incremented with additional sequences; although MAB methods are usually considered accurate, numerous cases of inconsistencies (incongruence) between observed data and deduced MAB trees are recorded (see [ 15 , 36 ]). Here, we re-examine the estimate of the p-distance between two homologous sequences, based on f id , as a source for geometric positioning, divergence time calculations and evolutionary reconstruction. We based our model on mathematical properties that alignment scores should respect; i) information theory [ 37 , 38 ] applied to sequence similarity, ii) algorithmic theory applied to alignment optimization [ 28 ] and iii) alignment probability, particularly in conformity with the TULIP theorem [ 39 ]. We used these properties as a framework of constraints to build a geometric representation of a space of probably homologous proteins and define a theoretically explicit measure of protein proximity. This unified model conserves information in the way physical models conserve matter or energy. The obtained representation of protein sequences is unaltered by adding or removing sequences. Applications include therefore the reconstruction of evolutionary consistent and robust trees, the topology of which is based on a spatial representation that is not reordered after addition or removal of sequences. Results and discussion Pair-wise sequence alignment scores in information theory Criteria to measure the variation/conservation balance between proteins should embody as much as possible the structural and functional potentiality within sequences of amino acids. In the absence of explicit physical criteria, amino acid similarity was solved empirically by measuring amino acid substitution frequencies in alignments of homologous sequences [ 30 , 40 ]. Given two amino acids i and j , the similarity function s ( i , j ) was set as: where is the observed frequency of substitution of i by j or j by i , and π i and π j are the frequencies of i and j in the two aligned sequences. The frequency is the estimate of the probability of substitution of i by j in real alignments; whereas π i π j is the estimate of the probability of substitution under the independency hypothesis. The similarity function gives a 20 × 20 similarity matrix usable to score protein sequence alignments, that can be interpreted in the information theory [ 37 , 38 ] according to the following proposition. Proposition 1 Amino acid substitution matrix values are estimates of the mutual information between amino acids in the sense of Hartley [ 37 , 38 ]. Consequently, the optimal alignment score computed between two biological sequences is an estimate of the optimal mutual information between these sequences. Proof Given a probability law P that characterizes a random variable, the Hartley self-information h is defined as the amount of information one gains when an event i occurred, or equivalently the amount of uncertainty one loses after learning that i happened: h ( i ) = -log( P ( i )) (4) The less likely an event i , the more we learn about the system when i happens. The mutual information I between two events, is the reduction of the uncertainty of one event i due to the knowledge of the other j : I j → i = h ( i ) - h ( i / j ) (5) Mutual information is symmetrical, i.e. I j → i = I i → j , and in the following will be expressed by I ( i ; j ). The self and mutual information of two events i and j are related: h ( i ∩ j ) = h ( i ) + h ( j ) - I ( i ; j ) (6) If the occurrence of one of the two events makes the second impossible, then the mutual information is equal to - ∞. If the two events are fully independent, mutual information is null. The empirical measure of the similarity between two amino acids described in equation (3) can therefore be expressed in probabilistic terms: where P ϖ is the joint probability to have i and j aligned in a given alignment and P π the measure of probability that amino acids occur in a given sequence. From equations (4) and (6), equation (7) becomes: s ( i , j ) = h ( i ) + h ( j ) - h ( i ∩ j ) (8) that is s ( i , j ) = I ( i ; j ) (9) As a consequence, the similarity function (or score) is the mutual information between two amino acids. Additionally, the score between sequences (the sum of elementary scores between amino acids, [ 32 , 33 , 41 , 42 ]) is, according to the hypothesis of independence of amino acid positions, the estimated mutual information between the two given biological sequences. Once two sequences are aligned, we pose the question whether the alignment score is sufficient to assess that the proteins are conceivably alike and thus evolutionarily related? The theorem of the upper limit of a sequence alignment score probability (TULIP theorem, [ 39 ]), sets the upper bound of an alignment score probability, under a hypothesis less restrictive than the Karlin-Altschul model [ 43 ]. Given two real sequences a and b ( a = a 1 a 2 ... a m and b = b 1 b 2 ... b n ), where s = s ( a , b ) the maximal score of a pair-wise alignment obtained with any alignment method, b * the variable corresponding to the shuffled sequences from b , and given P { S ( a , b *)≥ s } the probability that an alignment by chance between a and b * has a higher score than s , then whatever the distribution of the random variable S ( a , b *) the TULIP theorem states: with k > 1, μ the mean of and σ its standard deviation. The unique restriction on S ( a , b *) is that it has a finite mean and a finite variance. A first corollary of the TULIP theorem is that the Z-score is a statistical test for the probability of a sequence alignment score. We additionally state the following new corollary. TULIP corollary 2 Given the TULIP theorem conditions, let be the Z-score [ 44 ]. Then, z ( a , b *) is the greatest possible value for k ( k ∈]1,+∞[), which holds relation (10) true. In consequence, with k = z ( a , b *), then The best upper bound value for P { S ( a , b *)≥ s } is termed . From the TULIP theorem and corollaries, the comparison of a protein to a given reference a , weighed by an alignment score, is characterized by a bounded probability that the alignment is fortuitous. Question of the proximity between protein sequences in the light of information theory Since the optimized alignment score of two protein sequences allows an access to both the mutual information between proteins and an upper bound that the alignment is not fortuitous, one would expect that it is an accurate way to spatially organize proteins sets. A simple relation would be "the higher the mutual information, the nearest". There are three ways to assess the proximity between two objects a and b in a given space E [ 41 ]. The first is dissimilarity, a function f ( a , b ): E × E → + such that f ( a , b ) = 0 ⇔ a = b and f ( a , b ) = f ( b , a ); the second is the distance per se , that is a dissimilarity such that the triangle inequality is respected: ∀ a , b , c ∈ E , f ( a , c ) ≤ f ( a , b ) + f ( b , c ); and the third is the similarity defined as a function f ( a , b ): E × E → such that and f ( a , b ) = f ( b , a ). Representing objects in a space is convenient using the notion of distance. When the optimal alignment is global, i.e. requiring that it extends from the beginning to the end of each sequence [ 32 ], it is theoretically possible to define a distance per se , that is to spatially organize the compared sequences [ 41 ]. However, from a biological point of view, global alignment algorithms are not reliable to assess homology of protein domains. Local alignments are better suited, using scoring matrices to find the optimum local alignment and maximizing the sum of the scores of aligned residues [ 28 , 31 ]. In contrast with global alignments, local alignments do not allow any trivial definition of distances [ 41 ]. Although amino acid similarity is a function f ( i , j ): E × E → , owing to the local alignment optimization algorithms, the computed score is a function f ( a , b ): E × E → + , requiring the existence of at least one positive score in the similarity matrices. Thus, when constructing an alignment with the Smith and Waterman [ 33 ] method, the constraint that s ( a , b )>0 ( i.e. I(a;b) > 0) is imposed. This condition is consistent with proposition 1: if two sequences are homologous, knowledge about the first has to bring information about the second, that is to say, the mutual information between the two sequences cannot decrease below zero: I(a;b) > 0 ( i.e. s(a,b) > 0). As a consequence, in the following geometric construction we sought a refined expression for the proximity of proteins. Geometric construction of a configuration space of homologous proteins (CSHP) conserving mutual information In a set of homologous proteins, any sequence a can be selected as a reference, noted a ref , in respect to which the others are compared. A geometric representation of objects relatively to a fixed frame is known as a configuration space (CS). In physics, a CS is a convenient way to represent systems of particles, defined by their positional vectors in some reference frame. Here, given n similar sequences, it is therefore possible to consider n references of the CSHP. In a given (CSHP, a ref ), each amino acid position aligned with a position in the a ref sequence, corresponds to a comparison dimension (CS dimension). Proteins are simply positioned by a vector, the coordinates of which are given by the scores of aligned amino acids. Gaps are additional dimensions of the CS. When considering that local algorithms identify the space of biological interest, i.e. a CSHP, the gap penalty is a parameter that maximizes the shared informative dimensions. Thus, given the amino acids mutual information, alignment optimization methods define the relative positions of proteins. At this point in our construction, a first important property of the CSHP can be deduced. Since mutual information with a ref is sufficient for the full positioning, then positioning of proteins in a given (CSHP, a ref ) is unambiguous, unique, and is not altered when proteins are added or removed. In other words, a (CSHP, a ref ) is a univocal space. Given two sequences a and b , if b occurs in (CSHP, a ref ), then a also occurs in (CSHP, b ref ). The pair-wise alignment of a and b having no order (symmetry of the mutual information), the positions of b in (CSHP, a ref ) is dependent of the position of a in (CSHP, b ref ). Thus, once a (CSHP, a ref ) has been built, ∀ b ∈ (CSHP, a ref ), part of the geometry of (CSHP, b ref ) is learnt. Thus, in a CSHP, information needed for the position of n sequences is totally contained in the geometry of the n (CSHP, a ref ). This geometric stability is not observed with multiple alignments, which can be deeply modified by addition or removal of sequences. In the CSHP, protein position is unaltered by additions or removals of other proteins. In practice, the construction of CSHP is therefore completely deduced from any all-by-all protein sequence comparison [ 45 , 46 ] and can be easily updated. The q-dissimilarity, a proximity notion for a geometric representation of the CSHP In the CSHP, the definition of a distance per se based on mutual information is reduced ad absurdum (For demonstration, see methods). To define a proximity function i) sharing properties of distance, i.e. increasing when objects are further apart, ii) deriving from similarity and iii) relying on mutual information, particularly the property " f ( a , a ) ≠ f ( b , b ) is possible", we introduce a fourth notion of proximity. Such proximity was called q-dissimilarity (for quasi-dissimilarity), a function f ( a , b ): E × E → + is defined such that ∀ a ∈ E , ∀ b ∈ E , f ( a , b ) = f ( b , a ) (13) Let s be a similarity, then q = e - s is a q-dissimilarity, named the 'canonical q-dissimilarity' associated to s . Accordingly, the TULIP theorem allows a statistical characterization of q ( a , b ) the canonical q-dissimilarity between two sequences a and b . TULIP corollary 3 From the TULIP corollary 2, relation (14) is simply deduced: with Q ( a , b *) being the random q-dissimilarity variable associated with S ( a , b *). Given a (CSHP, a ref ), each sequence b aligned with a is characterized by a q-dissimilarity q ( a , b ). In geometric terms, b can be represented as a point contained in a hyper-sphere B of radius q ( a , b ). The representation of a (CSHP, a ref ) shown in Figure 1 is therefore in conformity with all constraints listed earlier and can also serve as a Venn diagram for the setting of events realized following a continuous random variable Q ( a , b *). When a is compared to itself, it is set on a hyper-sphere A of radius q ( a , a ), which is not reduced to one point. In the context of information theory, it is therefore possible to express that the proximity respects the property " q ( a , a ) ≠ q ( b , b ) is possible". Considering Figure 1 , is the probability for a random sequence b* to be in the hyper-sphere B . In conclusion, the q-dissimilarity is therefore a proximity notion that allows a rigorous geometric description of the configuration space of homologous proteins, real or simulated, (CSHP, a ref , q ). Unification of pair-wise alignments theory, information theory, p-distance and q-dissimilarity in the CSHP model A geometric space is a topological space when endowed with characterized paths that link its elements. Here, paths can be defined as the underlying evolutionary history separating sequences [ 11 ]. Given u the common unknown ancestor, then the divergence time t ( a , b ) is theoretically the summed elapsed times separating u to a and to b . Without any empirical knowledge of u , the simplest approximation for t ( a , b ) was sought as a function of the fraction of identical residues f id , thus of the p-distance. With the hypothesis of the molecular clock, this function can be given as equation (2), where the transmutation of a and b is a consequence of a Poisson process. By using relation (9) on the equivalence between score similarity and mutual information, then the fundamental postulate "the closer in the evolution, the more alike and conversely, the more alike, probably the closer in the evolution" can be reformulated: Fundamental postulate Given two homologous proteins a and b , the closer in the evolution, the greater the mutual information between a and b ( i.e. the optimal computed score s ( a , b )) and conversely, the greater the mutual information between a and b , probably the closer in the evolution. Whereas the first part of the postulate is a consequence of the conservational pressure on mutual information, the second assertion founds the historical reconstruction underlying a set of biological sequences on statistical concepts. A corollary is that evolution of two homologous proteins is characterized by a loss of mutual information. In the CSHP, this formulation of the fundamental postulate allows a novel mathematical formalization of the p-distance in probabilistic terms. Basically, the p-distance is the divergence observed between two sequences knowing that they share some features (the observed sequences a and b ) and that they were identical before the speciation event (sequence u ). Looking back to equation (1), we can re-formulate f id in probabilistic terms, considering the fraction of shared features (identical sites) knowing the observed data and the existence of a common ancestor. Given two proteins a and b , let us consider the random variable Q (a,b*), defined in TULIP corollary 3. In (CSHP, a ref ), shown in Figure 1 , one can define the probability law P { Q ( a , b *)≤ ρ } as the probability that the q-dissimilarity between b * and a ref is lower than ρ . The hyper-sphere of radius ρ contains therefore the b * random sequences sharing informative features with a accordingly. The probability p id / a that b * shares identity with a , knowing that the q-dissimilarity between b * and a is lower than that between the real sequences b and a , is: p id / a ( b *) = P { Q ( a , b *) ≤ q ( a , a ) / Q ( a , b *) ≤ q ( a , b )} (15) which is a probabilistic expression of f id in respect to the reference a ref . According to the Venn diagram in Figure 1 : p id / a ( b *) = P ( A / B ) Using the Bayes theorem, equation (15) can be expressed as: In consequence: which can be expressed as Assuming that substitution rates are independent of lineages [ 35 ], then random sequence models a* and b* are equivalent, that is to say Q ( a , a *) ≈ Q ( a , b *) and Thus p id / a , and symmetrically p id / b , provide a probabilistic expression of f id knowing the data, i.e. the observed mutual information between a and b expressed as Q ( a , b ). Given two homologous sequences a and b , when their optimal score is s ( a , b ) ≥ μ + ψ with ψ being a critical threshold value depending on the score distribution law (See Methods for the demonstration for the critical threshold), owing to the TULIP corollary 2, we can state that p id / a is bounded above: This expression can also be developed as: where μ 1 , σ 1 , μ 2 and σ 2 are the mean and the standard deviation of S ( a , b *) and S ( a , a *) respectively. The right term in relation (21) exhibits analogies with f id given by equation (1), showing that the pragmatic approach by Feng and Doolittle [ 19 ] could be supported and generalized in a theoretical elaboration. Using the Poisson correction, an expression of t ( a , b ) is given as the linear combination of the two corrections of the p-distance deduced from p id / a and p id / b : t ( a , b ) = -[log( p id / a ( b *)) +log( p id / b ( a *))] (22) with a * and b * the random variables corresponding to the shuffled sequences of a and b respectively. The sum of the logarithms corresponds to the product of the two probabilities, an expression of the hypothesis of independence of lineage. Interestingly, equation (22) provides an expression of the symmetric effect of time on the variations that independently affected a and b . From relation (20), t ( a , b ) appears as a function of Z-score ratios. For any set of homologous proteins, it is therefore possible to measure a table of pair-wise divergence times and build phylogenetic trees using distance methods. Reconstruction of protein phylogeny: first example, case study of the glucose-6-phosphate isomerase phylogeny We compared the trees we obtained, called TULIP trees, to phylogenetic trees built using classical methods, for instance the popular PHYLIP [ 47 ] or PUZZLE-based [ 48 ] methods, termed here MAB trees (for multiple alignment-based trees). Firstly, because MAB trees are constructed from multiple alignments, removals or additions of proteins modify the multiple alignments. Inclusion of sequences is considered as a way to improve the quality of multiple alignments and to increase the sensitivity of the comparison of distant sequences [ 49 , 50 ]. By contrast, the protein space used to build TULIP trees is not reordered when data sets are incremented or decremented (drawing of the TULIP tree may apparently change due to the tree graphic representation methods; nevertheless the absolute tree topology is not reordered). This remarkable property is due to both the geometrical construction by pair-wise comparison and the convergence of the distance matrix elements estimated by equation (21). Indeed, the estimate of the right-hand term of equation (21) relies on a Monte Carlo method, after randomization of the biological sequences [ 39 , 44 , 51 ] and is therefore dependent on the sequence randomization model [ 52 ] and convergent in respect to the weak law of large numbers [ 53 ]. Convergence is proportional to , where numb rand is the number of randomizations. In the case studies presented here, we set num rand = 2000 (see Methods). By contrast, stability of MAB trees is sought by bootstrapping approaches and consensus tree reconstruction. MAB trees appear as the result of a complex learning process including possible re-adjustment of the multiple alignments after eye inspection pragmatically applied to assist the reconstruction. Alternatively, Bayesian analyses have been recently proposed for phylogenetic inference [ 54 ], estimating posterior probability of each clades to assess most likely trees. Still, in a recent comparative study, Suzuki et al. [ 55 ] and Simmons et al. [ 56 ] provided evidence supporting the use of relatively conservative bootstrap and jacknife approaches rather than the more extreme overestimates provided by the Markov Chain Monte Carlo-based Bayesian methods. In the absence of any decisive methods to assess the validity of the trees obtained after such different approaches, no absolute comparison with the TULIP classification trees can be rigorously provided. Whenever a TULIP classification was achieved on a dataset that led to a consensual MAB tree, both were always consistent. For example, Figure 2 shows the phylogenetic PHYLIP [ 47 ] and TULIP trees obtained for glucose-6-phosphate isomerases (G6PI). Phylogeny of the G6PI enzyme has been studied by Huang et al. [ 57 ] in order to demonstrate the horizontal transfer of this enzyme in the apicomplexan phylum due to a past endosymbiosis [ 57 ]. Owing to the neighbor-joining analysis used by Huang et al. [ 57 ] (see methods) Figure 2A shows that apicomplexan G6PI is "plant-like". The TULIP tree shown in Figure 2B is consistent with this conclusion. Interestingly, differences between the two trees are found only when the bootstrap values on the MAB tree are not strong enough to unambiguously assess branching topology. Reconstruction of protein phylogeny: second example, case study of the enolase phylogenic incongruence TULIP classification tree further helps in solving apparent conflicting results obtained with MAB methods. In a comprehensive study from Keeling and Palmer [ 36 ] the PUZZLE-based reconstruction of the enolase phylogeny led to incongruent conclusions. Enolase proteins from a wide spectrum of organisms were examined to understand the evolutionary scenario that might explain that enolases from land plants and alveolates shared two short insertions. Alveolates comprise apicomplexan parasites, known to contain typical plant features as mentioned above, particularly a plastid relic. In this context, the shared insertion in apicomplexan and plant enolases (Figure 3 ) has been interpreted as a possible signature for some evolutionary relationship between apicomplexans and plants [ 58 , 59 ] and a likely sign of a lateral transfer. From the distribution of this insertion in enolases from several key eukaryotic groups, Keeling and Palmer [ 36 ] postulated that lateral transfer had been an important force in the evolution of eukaryotic enolases, being responsible for their origin in cryptomonads, Chlorarachnion and Arabidopsis . However, they could not conclude about alveolates, finding a conflict between the distribution of the insertion and the MAB phylogenetic position (Figure 4A ). The authors had to admit that lateral gene transfers failed to explain apicomplexa enolases, and were compelled to suppose that the lack of congruence between insertion and phylogeny could be because of a parallel loss of insertions in lineages, or to more complex transfers of gene portions. Based on our theoretical model, we constructed the corresponding TULIP tree. TULIP trees given with BLOSUM 62 or PAM 250 matrices, Fitch-Margoliash or neighbor-joining methods led indistinctly to a unique tree topology (Figure 4B ). Separation of great phyla (Archaebacteria, Eubacteria, Diplomonads, Trypanasomes, Animals, Fungi and Amoeba) is recovered. A plant-like cluster is additionally reconstructed, in which a distinct separation occurred between {Rhodophytes ; Cryptomonads} and {Land Plants ; Charophytes ; Chlorarachnion ; Alveolates} main clusters. It is remarkable that this latter cluster is that characterized by the enolase insertion. This topology corresponds to the observed distribution of the enolase short insertions and provides therefore a solution to the apparent enolase phylogeny incongruence: the phylogenetic position of alveolates is not in conflict with the distribution of enolase insertion and the apicomplexa enolase is possibly a consequence of a lateral transfer, like in cryptomonads. Large scale phylogeny based on a CSHP built from massive genomic pair-wise comparisons A CSHP containing large sets of protein sequences can be built after any all-by-all massive comparison providing Z-score statistics. Because the space elaboration is explicit, then quality of the mutual information conservation depends on the choice of the scoring matrix, the geometric positioning depends on the local alignment method, the homology assessment depends on the alignment score and probabilistic cutoffs and the phylogenetic topology on the choice of the stochastic law correction. Eventually, genome-scale pair-wise comparisons [ 39 , 36 ] find in the present CSHP a robust, evolutionary consistent and easily updatable representation. Methods Glucose-6-Phosphate Isomerase sequences The 41 Glucose-6-phosphate isomerase (EC 5.3.1.9) sequences studied in the paper are taken from several representative groups, as provided from the Swiss-prot database. Group I: Archae ([Swiss-prot:G6PI_HALN1], Halobacterium sp.; [Swiss-prot:G6PI_METJA], Methanococcus jannaschii ). Group II: Bacteria Actinobacteria ([Swiss-prot:G6P1_STRCO], Streptomyces coelicolor ; [Swiss-prot:G6PI_COREF, Corynebacterium efficiens ; [Swiss-prot:G6PI_MYCTU], Mycobacterium tuberculosis ). Group III: Bacteria Cyanobacteria ([Swiss-prot:G6PI_ANASP], Anabaena sp. ; [Swiss-prot:G6PI_SYNEL], Synechococcus elongates ). Group III: Bacteria Bacillus ([Swiss-prot:G6PI_LACFE], Lactobacillus fermentum ; [Swiss-prot:G6PI_BACHD], Bacillus halodurans ; [Swiss-prot:G6PI_BACSU], Bacillus subtilis ; [Swiss-prot:G6PI_CLOPE], Clostridium perfringens ). Group IV: Bacteria Proteobacteria ([Swiss-prot:G6PI_BIFLO], Bifidobacterium longum ; [Swiss-prot:G6PI_ECOLI], Escherichia coli ). Group V: Bacteria Chlamydiae ([Swiss-prot:G6PI_CHLTR], Chlamydia trachomatis ; [Swiss-prot:G6PI_CHLCV], Chlamydophila caviae ; [Swiss-prot:G6PI_CHLMU], Chlamydia muridarum ). Group VI: Others Bacteria ([Swiss-prot:G6PI_CHLTE], Chlorobium tepidum ; [Swiss-prot:G6PI_DEIRA], Deinococcus radiodurans ; [Swiss-prot:G6PI_BORBU], Borrelia burgdorferi ; [Swiss-prot:G6PI_THEMA], Thermotoga maritime ). Group VII: Fungi ([Swiss-prot:G6PI_SCHPO], Schizosaccharomyces pombe ; [Swiss-prot:G6PI_YEAST], Saccharomyces cerevisiae ; [Swiss-prot:G6PI_NEUCR], Neurospora crassa ; [Swiss-prot:G6PI_ASPOR], Aspergillus oryzae ). Group VII: Eukaryota Viridiplantae ([Swiss-prot:G6PI_ARATH], Arabidopsis thaliana ; [Swiss-prot:G6PI_MAIZE], Zea mays ; [Swiss-prot:G6PI_SPIOL, Spinacia oleracea ; [Swiss-prot:G6PA_ORYSA], Oryza sativa ). Group VIII: Eukaryota Alveolata Apicomplexa ([Swiss-prot:G6PI_PLAFA], Plasmodium falciparum ; [Swiss-prot:Q9XY88], Toxoplasma Gondii ; [Swiss-prot:269_185], Cryptosporidium parvum ). Group IX: Animals ([Swiss-prot:G6PI_DROME, Drosophila melanogaster ; [Swiss-prot:G6PI_MOUSE], Mus musculus ; [Swiss-prot:G6PI_HUMAN], Homo sapiens ; [Swiss-prot:G6PI_PIG], Sus scrofa ; [Swiss-prot:G6PI_RABIT], Oryctolagus cuniculus ; [Swiss-prot:G6PI_TRYBB], Trypanosoma brucei brucei ). Group X: Other Eukaryota ([Swiss-prot:AY581147], Entamoeba histolytica ; [Swiss-prot:G6PI_LEIME], Leishmania mexicana ; [Swiss-prot:AY581146], Dictyostelium discoideum ; [Swiss-prot:Q968V7], Giardia intestinalis ). Enolase sequences Enolase sequences used for the case-study presented in this paper were taken from eight major groups previously studied by [ 36 ]. Group I: Land Plant, Charophytes, Chlorophytes, Rhodophytes and Cryptomonads ([Swiss-prot:CAA39454], Zea mays ; [Swiss-prot:Q42971], Oryza sativa ; [Swiss-prot:Q43130], Mesembryanthemum crystallinum ; [Swiss-prot:P42896], Ricinus communis ; [Swiss-prot:Q43321], Alnus glutinosa ; [Swiss-prot:Q9LEJ0], Hevea brasiliensis 1; [Swiss-prot:P25696], Arabidopsis thaliana ; [Swiss-prot:P26300], Lycopersicon esculentum ; [Swiss-prot:AF348914], Chara corallina ; [Swiss-prot:AF348915], Nitella opaca ; [Swiss-prot:AF348916], Nitellopsis obtusa ; [Swiss-prot:AF348918], Pycnococcus provasolii 2; [Swiss-prot:AF348919], Bigelowiella natans – Chlorarachnion -; [Swiss-prot:AF348920], Mastocarpus papillatus 1; [Swiss-prot:AF348923], Prionitis lanceolata 1; [Swiss-prot:AF348931], Rhodomonas salina 1; [Swiss-prot:AF348933], Guillardia theta ; [Swiss-prot:AF348935], Pedinomonas minor ). Group II : Animals and Fungi ([Swiss-prot:P04764], Rattus norvegicus ; [Swiss-prot:P51913], Gallus gallus A; [Swiss-prot:P07322], Gallus gallus B; [Swiss-prot:Q9PVK2], Alligator mississippiensis ; [Swiss-prot:P06733], Homo sapiens A; [Swiss-prot:P13929, Homo sapiens B; [Swiss-prot:P15007], Drosophila melanogaster ; [Swiss-prot:AF025805], Drosophila pseudoobscura ; [Swiss-prot:O02654], Loligo pealeii ; [Swiss-prot:AF100985], Penaeus monodon ; [Swiss-prot:Q27527], Caenorhabditis elegans ; [Swiss-prot:Q27877], Schistosoma mansoni ; [Swiss-prot:P33676], Schistosoma japonicum ; [Swiss-prot:Q27655], Fasciola hepatica ; [Swiss-prot:P00924], Saccharomyces cerevisiae 1; [Swiss-prot:Q12560], Aspergillus oryzae ; [Swiss-prot:P42040], Cladosporium herbarum ; [Swiss-prot:P40370], Schizosaccharomyces pombe 1; [Swiss-prot:AF063247], Pneumocystis carinii f.). Group III: Amoebae ([Swiss-prot:P51555], Entamoeba histolytica ; [Swiss-prot:Q9U615], Mastigamoeba balamuthi ). Group IV: Alveolates ([Swiss-prot:AF348926], Paramecium multimicronucleatum ; [Swiss-prot:AF348927], Paramecium tetraurelia ; [Swiss-prot:AF348928], Colpidium aqueous ; [Swiss-prot:AF348929], Tetrahymena thermophila I; [Swiss-prot:AF348930], Tetrahymena bergeri ; [Swiss-prot:Q27727], Plasmodium falciparum ; [Swiss-prot:AF051910], Toxoplasma gondii ). Group V: Trypanosomatidae ([Swiss-prot:AF159530], Trypanosoma cruzi eno1 partial; [Swiss-prot:AF152348], Trypanosoma brucei complete). Group VI: Hexamitidae ([Swiss-prot:AF159519], Hexamita inflata eno1 partial; [Swiss-prot:AF159517], Spironucleus vortens partial). Group VII: Archaebacteria ([Swiss-prot:Q9UXZ0], Pyrococcus abyssi ; [Swiss-prot:O59605], Pyrococcus horikoshi i; [Swiss-prot:Q60173], Methanococcus jannaschii ; [Swiss-prot:Q9Y927], Aeropyrum pernix ). Group VII: Eubacteria ([Swiss-prot:O66778], Aquifex aeolicus ; [Swiss-prot:P37869], Bacillus subtilis ; [Swiss-prot:Q9K717], Bacillus halodurans ; [Swiss-prot:P77972], Synechocystis sp.; [Swiss-prot:P33675], Zymomonas mobilis ; [Swiss-prot:P08324], Escherichia coli ; [Swiss-prot:P47647], Mycoplasma genitalium ; [Swiss-prot:Q8EW32, Mycoplasma penetrans ; [Swiss-prot:P74934], Treponema pallidum). Demonstration that distance of a protein to itself cannot be defined in the CSHP In the simplest case, building a distance between amino acids (that would lead to distance between sequences) on the basis of computed similarity values would have to respect the condition: ∀ i ∈ E , ∀ j ∈ E , d ( i , j ) = 0 ⇒ i = j (a) for i and j , two given amino acids and d the distance function. Using this condition in the proposition, any organization of the CSHP with a geometric distance is reduced ad absurdum . Proposition Building a distance between amino acids derived from the composed function d ( i , j ) = ( φ ○ s )( i , j ), where s is a similarity function and φ a bijection, is impossible without a loss of mutual information. Moreover, two proteins from distinct organisms can have the same configuration, being like "twins", and d ( i , j ) = 0 does not imply i = j . Proof Condition (a) implies that φ ( s ( i , i )) = φ ( s ( j , j )) = 0. This equality imposes that s ( i , i ) = s ( j , j ) and, following equation (7) of main text, that I ( i ; i ) = I ( j ; j ). Considering for example tryptophan (W) and glutamic acid (E), if W occurs in a sequence, the mutual information gained about the occurrence of W at the aligned position would be the same as that gained in the case of E about the occurrence of E at the aligned position in the homologous protein. This statement is easily rejected on the basis of biochemical concerns. On one hand, aspartic acid (D) shares common biochemical properties with E, particularly a carboxylic acid, and easily substitutes in homologous sequences. By contrast W, exhibiting a unique biochemical feature, is less substitutable without altering the function. Thus the mutual information I(E;E) is necessarily lower than I ( W ; W ). This that can be checked in scoring matrices such as BLOSUM 62 [ 30 ] where I(E;E) = 5, I(D;D) = 6 and I(W;W) = 11. Condition d ( i , i ) = 0 leads to an obvious loss of information. The second assertion of the proposition is obvious. Determination of the threshold value ψ , for topological reconstructions in the CHSP based on pair-wise alignment score probabilities An important basis of the reconstruction of a probabilistic evolutionary topology in the CSHP is based on the demonstration that, given S the random variable corresponding to the alignment scores of pairs of shuffled sequences and μ the mean of S , given two homologous sequences a and b , when their optimal score is s ( a , b ) ≥ μ + ψ (with ψ a critical threshold value depending on the score distribution law), owing to the TULIP corollary 2, we can state that p id / a is bounded above To the purpose of this demonstration, we considered the cumulative distribution function F ( s ) = P ( S ≤ s ), its derivative f ( s ) known as the probability density function defined as dF ( s ) = f ( s ) ds , and the positive delta function δ ( s ) = ( s - μ ) 2 (1 - F ( s )). Since δ ( s ) = ( s - μ ) 2 (1 - F ( s )) is null for s = μ and , the Rolle's theorem implies that ∃ s 0 ∈] μ ,+∞[ such as [ 60 ]; s 0 corresponds to a maximum of δ ( s ) and is therefore the solution of the equation 2(1 - F ( s )) - ( s - μ ) f ( s ) = 0 (b) one can express as The term corresponds to a continuous function. Interestingly, is known as the hazard function [ 61 ], that is the probability of s , per score unit ( i.e. mutual information), conditional to the fact that the pair-wise alignment score is at least equal to s . The hazard function is also defined by . A critical hypothesis is that φ ( x ) function is strictly increasing and conversely that is strictly decreasing. Considering , equation 2 has only one solution s 0 and this solution is bounded above: In consequence, δ ( s ) reaches its maximum for a s 0 ( s 0 ≤ μ + ψ ) and it is strictly decreasing on ] μ + ψ ,+∞[. The estimation of s 0 is not trivial because it depends on the knowledge of the cumulative distribution function. Extensive studies provided experimental and theoretical supports for an extreme value distribution of alignment scores [ 31 , 43 , 44 ]. Using the extreme value distribution of type I, i.e. the Gumbel distribution [ 62 ], the cumulative distribution is given by with θ and β ( β > 0) the location and scale parameters. The probability density function g ( s ) is defined by dG ( s ) = g ( s ) ds . We observe with that . Using the Taylor's polynomial formula, i.e. : In consequence, for a Gumbel score probability distribution: A graphical determination of ψ from a Gumbel distribution is illustrated in Figure 5 . If a pair-wise alignment score of two sequences a and b is relatively high, that is s ( a , b ) ≥ μ + ψ , then the trivial inequality s ( a , a ) ≥ s ( a , b ) implies ( s ( a , b ) - μ ) 2 (1 - F ( s ( a , b ))) ≥ ( s ( a , a ) - μ ) 2 (1 - F ( s ( a , a ))) (h) that is to say From inequality (i), we deduce that p id / a is bounded above. Construction of PHYLIP multiple alignment based trees and pair-wise alignment based TULIP trees To build PHYLIP trees, multiple sequence alignments were created with ClustalW [ 63 ]. PHYLIP trees where constructed using the protpars and neighbor modules from the PHYLIP package [ 47 ] and the BLOSUM 62 substitution matrix. Bootstrap support was estimated using 1000 replicates. To build TULIP trees, for each couple of sequences a and b , alignment was achieved with the Smith-Waterman method and the BLOSUM 62 scoring matrices, using the BIOFACET package from Gene-IT, France [ 64 ]. We computed estimated z-scores z ( a , b *), z ( a , a *), z ( a *, b ), z ( b *, b *), with 2000 sequence shuffling. For all computations, an estimation of the Gumbel parameters θ and β was made using the computed μ and σ of any S ( a , b *) and the formula and θ = μ - β Γ'(1), where Γ'(1) ≈ 0.577216 is the Euler constant. In all computations, both Gumbel parameters were very close (in the case of enolases, mean ( θ ) = 35.04, SD ( θ ) = 0.12, mean ( β ) = 3.92, SD ( β ) = 0.08). As a consequence, the assumption Q ( a , a *) ≈ Q ( a , b *) was verified for any pair of sequences. We used the parameters to estimate μ = θ + β Γ '(1) (in the case of enolases, μ = 37.33), and μ + ψ ≈ μ + 10.5178 ≈ 47.85. As any pairs of computed scores are higher than this critical threshold, we used relation [ 20 ]. Estimation of evolutionary time was achieved according to equations [ 20 ] and [ 22 ]. Trees were constructed using Fitch-Margoliash and Neighbor-Joining methods [ 47 ]. List of abbreviations CSHP, configuration space of homologous proteins, TULIP, theorem of the upper limit of a score probability Authors' contributions OB conceived the main theoretical model, designed and developed the method to build phylogenetic trees and drafted the manuscript. SR and PO participated in the theoretical model refinement and in the design and development of computational methods to build TULIP trees. EM contributed to the conception of this study, participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC555736.xml |
516784 | Exploratory disease mapping: kriging the spatial risk function from regional count data | Background There is considerable interest in the literature on disease mapping to interpolate estimates of disease occurrence or risk of disease from a regional database onto a continuous surface. In addition to many interpolation techniques available the geostatistical method of kriging has been used but also criticised. Results To circumvent these critics one may use kriging along with already smoothed regional estimates, where smoothing is based on empirical Bayes estimates, also known as shrinkage estimates. The empirical Bayes step has the advantage of shrinking the unstable and often extreme estimates to the global or local mean, and also has a stabilising effect on variance by borrowing strength, as well. Negative interpolates are prevented by choice of the appropriate kriging method. The proposed mapping method is applied to the North Carolina SIDS data example as well as to an example data set from veterinary epidemiology. The SIDS data are modelled without spatial trend. And spatial interpolation is based on ordinary kriging. The second example is included to demonstrate the method when the phenomenon under study exhibits a spatial trend and interpolation is based on universal kriging. Conclusion Interpolation of the regional estimates overcomes the areal bias problem and the resulting isopleth maps are easier to read than choropleth maps. The empirical Bayesian estimate for smoothing is related to internal standardization in epidemiology. Therefore, the proposed concept is easily communicable to map users. | Background As with the analysis of any set of data, it is always good practice to begin by producing and inspecting graphs. A feel for the data can then be obtained and any outstanding features identified. In spatial epidemiology this is called disease mapping. Bithel [ 7 ], Diggle [ 14 ] and Lawson [ 25 ] provide recent reviews of disease mapping. Spatial epidemiology comprises at least three types of study focus [ 17 , 25 ]. These are (i) disease mapping, (ii) disease clustering and (iii) geographical correlation analysis but these distinctions are not strict. For example a disease map is also used for reporting the results of a geographical correlation study or to highlight areas of high or low disease incidence, i.e. cluster locations in a cluster study [ 4 , 14 ]. But in the following, disease mapping is considered as exploratory analysis used to get an impression of the geographical or spatial distribution of disease or the corresponding risk. For this the disease map should be based on smoothed estimates, clean of noise and adjusted for variation in 'at-risk' population [[ 24 ], p. 163]. The resulting disease map should provide insight into possible causes, effects and trends in the vast amount of data. This will provide an invaluable starting point for epidemiologic enquiry. There are three basic types of disease maps corresponding to certain types of data. These are dot maps for point (or case-event) data, choropleth maps for regional data (also called lattice- or census-tract data), and lastly, isopleth maps for geostatistical data (also called point measurements) representing spatially continuous phenomena at a limited number of sampling locations. Spatial epidemiology is mainly concerned with the analysis of two types of data: case-event data and regional summary data, generally leading to dot maps and choropleth maps, respectively. However, epidemiology is also concerned with the identification of unknown risk factors, which may be part of the environment such as air pollution, radiation or magnetic fields. And such factors vary spatially continuously. Thus it is important to produce an isopleth map of disease occurrence or risk. Furthermore, as has been often pointed out in the literature [[ 24 ], p. 131], choropleth maps must be interpreted with caution, as the grey scale grouping is arbitrary and such a choice can affect interpretation, although choropleth maps imply a constant risk (incidence or prevalence) over regions with discontinuities at the border lines. Another point of criticism concerns the areal bias [ 12 ], as the varying shape and size of geographical regions make the patchy maps difficult to interpret. The visual impact of larger areas is higher and may dominate the map, leading to biased visual perception, whereas in human epidemiology it is the smaller, urban areas, and not the rural surroundings, that are of primarily interest due to population sizes. These objections may be circumvented by the use of continuous surface mappings, i.e. isopleth maps. Methods Review of interpolation methods For spatial point case-control data Bithell [ 6 ] introduced a spatial interpolation method based on kernel density estimation. The resulting map is called the "spatial relative risk function". Lawson and Williams [ 28 ] and Kellsall and Diggle [ 20 ] proposed modifications of this procedure. Regional data arise from summarising individual information for administrative regions such as census tracts. The basic model for the individual data, i.e. spatial point data, is the spatial Poisson process. In case of a rare disease, aggregation of the data over distinct regions results again in Poisson distributed data. For a more common disease the regional counts may be binomially distributed. When spatial interpolation of regional data is the objective of the disease mapping study a grid of surface interpolant co-ordinates must be provided. Then a number of techniques can be used to automatically interpolate the data by use of deterministic methods or to predict the values statistically at the grid co-ordinates. Among the possible choices are kernel smoothing, splines, loess and running medians; see [ 19 ] for a discussion of these methods. Kernel smoothing has the advantage of preserving the positivity condition implied in rate data. Brillinger [ 8 ] used kernel-type smoothing in the context of birth rate data, and Müller, Stadtmüller and Tabnak [ 33 ] applied locally weighted least squares adapted for spatial aggregation to AIDS incidence maps for the San Francisco Area. Another approach for the interpolation of regional data onto a continuous surface is the geostatistical prediction method of kriging. Some implementations of kriging have been proposed to obtain a risk surface [ 10 , 23 , 31 , 34 , 35 , 38 ]. Kelsall and Wakefield [ 21 ] have proposed a more complex geostatistical approach based on generalised linear modelling (GLIM) of regional data, which is similar to the work by Diggle, Tawn and Moyeed [ 15 ]. The relative merits of different interpolation approaches haves not so far been systematically investigated [[ 24 ], p. 131]. Kriging the spatial risk function The problems with the kriging method for generating isopleth maps of disease occurrence, i.e. the spatial risk function, are: (1) heterogeneous variances in the regional estimates and (2) the potential of negative interpolations. The first problem can be ameliorated by the use of empirical Bayes estimation to smooth the data prior to kriging. Appropriate geostatistical modelling can solve the second problem. Let the study area be divided into N disjunctive sub regions indexed by i ( i = 1 ,..., N ). Define n i as the size of the 'at-risk' population of the i -th region, and denote the number of cases by m i . The proportion p i = m i / n i is the crude estimate for the parameter of interest θ i . Smoothing by empirical Bayes estimation Mapping raw estimates of disease occurrence can lead to spurious spatial features. To overcome this problem Cressie and Read [ 13 ] have explored the use of several variance stabilising transformations, but the results on the transformed scale are difficult to interpret. Furthermore, empirical Bayesian methods have been developed based on the idea of pooling information across regions. The resulting smoothed regional estimates have a variance stabilising side effect by borrowing strength from (local or global) neighbourhood information. The outline of the empirical Bayes approach for smoothing regional rates of rare diseases is based on the Poisson model. For the case of more common diseases Martuzzi and Elliott [ 30 ] adapted the approach to the Binomial model. Poisson model for rare diseases Clayton and Kaldor [ 11 ] proposed empirical Bayes or shrinkage estimation for smoothing regional data along with maximum likelihood estimation (MLE) of the unknown prior parameters. Marshall [ 29 ] modified this approach using method of moments estimators (MME) instead of MLE. The resulting estimates provide starting values for the iterative maximum likelihood procedure; or they could be used for exploratory mapping purposes, which is the main interest of this work. Assume the cases in every region i are independently Poisson distributed with the unknown parameter θ i which has an unknown prior distribution associated with expectation E( θ i ) = π and variance Var( θ i ) = φ 2 . Then the totals of cases from the i -th region, e.g. m i ( i = 1 ,..., N ), are distributed as follows m i | θ i , n i ~ Po( n i θ i ) θ i ~ ( π , φ 2 ) The MME of the unknown hyperparameters are for the prior mean and for the prior variance, where denotes the mean regional 'at-risk' population and the summation is over the range of i . The empirical Bayesian estimates then becomes with shrinkage weights . Tendency of variance homogeneity As stated above, the smoothing of the crude regional estimates p i has the side-effect of stabilising the variances for the regional empirical Bayes estimates, i.e. . Heuristically this is clear, because the empirical Bayes estimates are based on the whole sample information and not just on the individual regional sample, and thus are more stable. But this is difficult to prove, because this requires an analytical expression for the corresponding variance or an approximate variance estimate. Morris [ 32 ] proposed an approximate variance estimate for the empirical Bayes estimate in the Gaussian/Gaussian setting. However, an extension of this idea to the non-Gaussian setting is awkward [[ 9 ], p. 80]. Therefore, two facts are used to claim the conjecture that the empirical Bayes estimates of the regional estimates show up the tendency of variance homogeneity. First, it should be noted that Bayes estimates, i.e. , generally have a smaller associated variance than the corresponding frequentist estimates, i.e. p i [ 32 ]. This means the Bayesian smoothing generally results in reduced variances, thus reducing the absolute differences between regional variances, i.e. the variance heterogeneity. Secondly, by virtue of the shrinkage property of the empirical Bayes estimate, it follows that the variance will be shrunk back to the global variance in the case of small regional samples, which is responsible for a large part of the unstable estimates. Both points together make up what is called "borrowing strength from the ensemble" [ 32 ]. Geostatistical modelling The geostatistical method of kriging is widely accepted for the purpose of spatial prediction, i.e. interpolation and (moderate) extrapolation. It is proposed here to predict the smoothed regional data onto a fine meshed regular grid of points for isopleth mapping purposes. Spatial linear model Kriging of spatial data, say Z = ( Z 1 ,..., Z N )' at sample sites s i , i = 1,.., N , takes place within the framework of the spatial linear model [ 1 , 12 ] Z = μ + δ , E( Z ) = μ , δ ~ Gau(0, Σ) An integral part of this model is spatial correlation, which must be taken into account to draw valid scientific inferences. Here Σ is a variance-covariance matrix, spatially structured according to the position and direction between sampling sites. The basic assumption for spatial data is that near things are more related than distant things. This was neglected for smoothing of the regional epidemiological measures (prevalence or incidence) via empirical Bayes estimators. Semivariogram For geostatistical modelling, the structure of spatial variation will be estimated through the semivariogram , where h denotes the translation between any two arbitrary sites s i and s j within the study region. See [ 12 , 37 ] for diverse parameterised semivariogram models and estimation methods. Kriging In geostatistics kriging is used synonymously with optimal spatial prediction. However the optimality depends on the appropriateness of the spatial model. Cressie [ 12 ] gives a review on diverse kriging approaches. There are different models with respect to the knowledge and estimation of the spatial mean function, i.e. μ ( s ). Ordinary kriging is concerned with an unknown but constant mean function, i.e. μ ( s ) = μ . Furthermore, universal kriging is based on a polynomial trend surface model which is to be removed prior to estimation of the semivariogram from the residuals, i.e. δ ( s ). This technique may be the most widely-used in practice. An outlier-resistant alternative is median polish kriging. This method starts by the robust and non-parametric estimation of the non-constant mean surface via median polishing followed by robust semivariogram estimation. Berke [ 2 ] proposed a modification of median polish kriging for larger spatial data sets. Generally, kriging surfaces are the sum of an estimate for the trend surface μ ( s ) plus the kriging prediction for the residual process δ ( s ), formally . Kriging and smoothing Kriging is sometimes termed a smoothing method. This is due to the fact that the predicted residuals are in absolute not larger than the model residuals δ ( s ) = Z ( s ) - μ ( s ), i.e. the variability of the predictions around the estimated mean surface is smaller than the variability of the observations Z ( s ). When the semivariogram is modelled without nugget effect, i.e. without small-scale variability at spatial scales smaller than the observational scale, then kriging leads to direct interpolation at the sampling sites. In this case the prediction equals the observation at the sample sites and thus the predicted residuals are equal to the model residuals, i.e. . Predictions at any other sites have the tendency to shrink towards the value of the estimated trend surface at that place. On the other hand, when a semivariogram with nugget effect is appropriate, than the prediction tends to be closer to the mean surface, which gives smaller residuals , i.e. a smoother prediction surface . Thus the prediction of invalid values for risk or for other epidemiologic measures, as has been criticised in the past, is a consequence of inappropriate spatial modelling. This could only happen when universal kriging is based on an estimated trend surface model , which exceeds the range of valid values. Results The mapping technique proposed to generate isopleth risk maps from regional count data is now applied to two example data sets. Example 1 is based on the SIDS mortality rates in North Carolina that show no spatial trend. Example 2 is based on spatial trend contaminated data of tapeworm infections among red foxes in Lower Saxony. Example 1: sudden infant death syndrome (SIDS) in North Carolina This by now classical spatial data set on SIDS mortality rates in North Carolina from 1974 to 1984 has been analysed by many researchers. Cressie [ 12 ] gives an introduction to the problem, earlier references and results on data modelling, mapping and cluster detection. More recently Kulldorff [ 22 ] applied the spatial scan test to detect clusters of disease and Lawson and Clark [ 27 ] applied a spatial mixture modelling approach to map the standardized mortality ratio (SMR) for the period 1974 – 1978. For the years 1974 to 1984 the number of live births ranges from 567 to 52345 over North Carolina's 100 counties. The total number of reported SIDS cases is 1503 out of 753354 live births, which results in an annual mortality rate of approximately 2 per 1000 live births. The mean of the counties boundary files coordinates in longitude and latitude were used here as the geographic coordinates for the regions centres. Figure 1 shows the shrinkage effect by using parallel box plots for the raw rates and the empirically Bayesian smoothed rates under the Poisson model for rare phenomena. Figure 1 Comparison of raw and Bayesian smoothed SIDS mortality rates. Parallel box plots for the raw annual SIDS mortality rate per 1000 live births ( r ) and the corresponding shrinkage estimates or empirical Bayesian estimated rates ( eber ) from 100 counties of North Carolina, 1974–1984. The smoothed rates are more appropriate for disease mapping than the raw rates and hence used here for choropleth mapping in Figure 2 . Cut points of the grey scale shading are the 5%, 50% and 95% quantiles of the empirical distribution. These are used to highlight the upper and lower five percent of the distribution of the Bayesian smoothed mortality rates and to distinguish between higher and lower values. The smoothed rates vary from about 1.2 to 3.5 cases per 1000 live births. Visual map perception reveals some potential high and low risk areas in the north and southwest but no striking spatial trend pattern. Figure 2 Choropleth map of Bayesian smoothed SIDS mortality rates of North Carolina. Choropleth map of the empirical Bayesian smoothed mortalities per 1000 live births from 100 counties of North Carolina, 1974–1984. Cut points of the grey scale shading are the 5%, 50% and 95% quantiles of the empirical distribution. Geostatistical prediction, i.e. kriging of the empirical Bayesian smoothed rates, is based on appropriate modelling of the data. Here, the constant mean assumption for the spatial mean surface is chosen and justified by visual inspection of the empirical semivariogram which levels out and reaches a sill. The spatial dependence structure is modelled by an isotropic exponential semivariogram without nugget effect, which is fitted by weighted least squares estimation to the robustly estimated empirical semivariogram; see [ 12 ] for technical details. The result of this procedure is summarised in Figure 3 . The close fitting of the empirical semivariogram to the model indicates appropriate model choice for the dependence structure as well as for the constant mean surface. See [ 1 ] for diagnostic methods and its applications in geostatistical modelling. Figure 3 Empirical semivariogram from smoothed SIDS mortality rates and fitted exponential model. Exponential model (sill = 0.22, range = 0.37) fitted by weighted least squares (WLSE) to the robust empirical semivariogram from empirical Bayesian smoothed annual SIDS mortality rates per 1000 live births from 100 counties of North Carolina, 1974–1984. Due to the constant mean assumption ordinary kriging is the appropriate spatial prediction method and without nugget effect this leads to direct interpolation of the data at the counties centres and to smoothed values shrunken towards the global mean for the rest of the study area. The resulting isopleth map or risk surface map is given in Figure 4 . Now the grey scale shading is almost continuous, i.e. the patchy nature of Figure 2 is replaced by a surface. Additional isolines are drawn for the same cut points as for the grey scale shadings in Figure 2 (i.e. at the 5%, 50% and 95% level of the empirical distribution of the empirical Bayesian estimates) and will be useful to support map interpretation and comparison. The risk surface map may be more useful to identify potential environmental risk factors than the choropleth map. Figure 4 Isopleth map from kriging the smoothed SIDS mortality rates of North Carolina. Isopleth map based on kriging predictions of the empirical Bayesian smoothed annual SIDS mortality rates per 1000 live births from 100 counties of North Carolina, 1974–1984. Cross-validation can be used to explore the predictive performance of the kriging model. The cross-validation residuals (weighted with respect to the spatial dependence structure given by the semivariogram) should be approximately Gaussian distributed. Normal probability plots can be used to disclose grossly model inadequacies [[ 12 ], p. 498] as well as for outlier identification. Figure 5 shows the normal probability plot of the cross-validation residuals based on the refitted model with the outliers removed. Figure 5 Normal probability plot of cross-validation residuals from kriging smoothed SIDS mortality rates. Normal probability plot of the cross-validation residuals from kriging the empirical Bayesian smoothed regional SIDS mortality rates. Observations 4, 42, 83, 8 and 24 as well as 77 and 59 may be outliers with respect to the spatial model. Aside from some positive and negative extreme values the general appearance of the residual process is Gaussian. The potential outliers are regions with steep gradients in risk and hence part of disease clusters (or their surroundings) which were previously identified [ 22 ] and are located in the north and south-west of North Carolina. Of course, disease clusters are of interest and the proposed modelling approach reveals their existence and points to the respective high risk areas. This is in line with the gross aims of exploratory disease mapping. Example 2: tapeworm infections in red foxes in Lower Saxony Echinococcus multilocularis (E.m.) is a tapeworm occurring in the northern hemisphere, including endemic regions in central Europe, most of northern and central Eurasia and parts of North America. In central Europe the red fox is the main definitive host with rodents such as mice or muskrats serving as intermediate hosts [ 16 ]. The parasite E.m. causes the zoonosis alveolar echinococcosis (A.E.), which has a potentially high fatality rate. Recent studies reflect an alarmingly wide geographic range of the parasite in foxes, with average prevalences varying up to 60% for central Europe. However, the spatial distribution of E.m. in foxes is complex and insufficiently known. There are indications of emerging risk factors for human A.E. such as increasing parasite prevalences in red foxes, growing fox population and progressive spread of foxes to cities. The federal state of Lower Saxony is part of northern Germany and contains the federal city-state of Bremen as an enclave. During the period from 1991 to 1997, 5365 red foxes were sampled in Lower Saxony and examined for infections with E.m.. The data are given in Table 1 [ 3 ]. Table 1 Regions and statistics for the fox tape worm example. The 43 investigated regions in Lower Saxony with their co-ordinates (x, y), the number of red foxes tested (n) and found positive for E. multilocularis (m), the raw period prevalence in % (PP) and the empirical Bayes smoothed period prevalence in % (BPP). The regions no. 5, 8, 13, 14 and 15 form a previously identified disease cluster [4]. Regions Statistics Nr. Name x y n m PP BPP 1 Braunschweig 7,57 -4,30 25 1 4 6 2 Salzgitter & Wolfenbüttel 7,47 -6,07 115 9 8 8 3 Wolfsburg 9,33 -2,83 22 6 27 23 4 Gifhorn 8,15 -1,37 158 13 8 8 5 Göttingen 3,44 -12,92 157 84 54 51 6 Goslar 6,68 -8,81 152 20 13 13 7 Helmstedt 10,11 -4,30 66 7 11 10 8 Northeim 2,86 -10,28 186 96 52 49 9 Osterode/Harz 6,68 -11,07 94 23 24 23 10 Peine 5,80 -4,11 115 10 9 9 11 Hannover 2,17 -2,64 327 41 13 12 12 Diepholz -4,39 0,88 143 10 7 7 13 Hameln-Pyrmont -0,17 -6,36 99 41 41 39 14 Hildesheim 3,74 -6,36 202 60 30 29 15 Holzminden 1,19 -8,62 60 16 27 24 16 Nienburg/Weser -1,74 -0,58 325 20 6 6 17 Schaumburg -1,25 -4,11 90 12 13 13 18 Celle 4,81 0,59 255 8 3 3 19 Cuxhaven -3,51 10,69 73 14 19 18 20 Harburg 3,64 7,06 285 17 6 6 21 Lüchow-Dannenberg 11,68 3,92 225 20 9 9 22 Lüneburg 7,37 6,37 278 22 8 8 23 Qsterholz-Scharmbeck -3,80 6,57 32 8 25 22 24 Rothenburg/Wümme -0,17 6,47 114 6 5 5 25 Soltau-Fallingbostel 2,46 2,84 137 8 6 6 26 Stade 0,41 10,00 84 5 6 6 27 Uelzen 7,76 3,53 214 14 7 6 28 Verden -1,45 3,44 107 11 10 10 29 Delmenhorst -4,88 4,31 8 1 13 12 30 Emden -14,74 8,19 4 0 0 9 31 Oldenburg -6,84 4,12 103 15 15 14 32 Osnabrück -9,19 -3,12 290 21 7 7 33 Wilhelmshaven -8,55 10,30 10 0 0 6 34 Ammerland -8,99 6,28 76 4 5 6 35 Aurich -13,11 9,42 111 15 14 13 36 Cloppenburg -9,78 2,84 110 8 7 7 37 Emsland -13,60 1,07 242 21 9 8 38 Friesland -9,10 9,32 15 1 7 9 39 Bentheim -15,95 -1,46 36 1 3 4 40 Leer -12,42 6,67 39 2 5 6 41 Vechta -8,12 0,20 83 2 2 3 42 Wesermarsch -6,54 7,55 45 9 20 18 43 Wittmund -10,76 10,10 53 4 8 8 Sum 5365 706 Median (in %) 8,2 9,0 Figure 6 is a choropleth map of the empirical Bayes smoothed period prevalences from 43 regions in the study area. The cut points of the grey scale shading are again the 5%, 50% and 95% quantiles of the empirical distribution or the smoothed data. The period prevalences range from 3% to 51% with the median at 9%. Figure 6 Choropleth map of Bayesian smoothed prevalences in Lower Saxony. Choropleth map of the empirical Bayesian smoothed period prevalences from 43 regions in Lower Saxony. Cut points of the grey scale shading are the 5%, 50% and 95% quantiles of the empirical distribution. The map indicates high period prevalences in the south and north of Lower Saxony. Extraordinarily high prevalences were observed in the southern regions, which indicate the presence of a positive disease cluster that is identified by use of the spatial scan statistic [ 4 ]. Figure 7 shows the corresponding isopleth map resulting from universal kriging, with overlaid isolines. The isopleth map gives an impression of gradual changes instead of jumps at the regional borders. Furthermore, the missing region of Bremen in the central north of Lower Saxony has also been supplied with predicted values. Figure 7 Isopleth map from kriging the smoothed prevalences in Lower Saxony. Isopleth map with overlaid isolines of the kriging interpolated choropleth map in Figure 6. Isolines are at the 5%, 50% and 95% quantiles of the empirical distribution of the empirical Bayesian smoothed regional period prevalences. Figure 6 points out the presence of heterogeneity in the spatial mean. Here, universal kriging is based on the assumption that the spatial mean function of the data could be represented by an incomplete quadratic polynomial in the spatial co-ordinates. Other potential explanatory variables are not at hand. Let s = ( x , y )' denote an arbitrary location in the study region, where x and y are the co-ordinates to the east and north. By inspection of the variogram cloud [ 12 ] based on trend residuals, two observations (Sites 3 and 13) were identified as outliers and removed from the structure analysis. In the second step of an iterated modelling approach then the trend polynomial fitted by ordinary least squares (OLSE) is given by μ ( s ) = β 0 + β 1 y + β 2 y 2 , with , , . Instead of OLSE one could use iterated WLSE, but iteration may lead to biased estimates. The residuals of the trend surface fit to the empirical Bayes smoothed regional period prevalences were then used to model a spherical semivariogram by weighted least squares estimation of the robustly estimated empirical semivariogram. Figure 8 shows the empirical semivariogram as well as the fitted model for both, the detrended and trend contaminated data (sill = 0.0039, range = 5.29 and sill = 0.0047, range = 7.83 respectively), to indicate the benefit from the trend model as measured by the 20% decrease of the sill value. Figure 8 Empirical semivariograms and fitted exponential models from detrended and trend-contaminated smoothed prevalences. Robust empirical semivariograms of the detrended (black dot) and trend-contaminated data (circle) along with the WLSE fitted spherical models for the detrended (solid line) and the trend-contaminated (dashed line) data with observations 3 and 13 removed. The normal probability plot of the kriging cross-validation residuals in Figure 9 draws attention to the presence of three or four outliers (Observations 3, 8 and 13 and possibly 15). Observation 3 and 15 were previously identified as extremes and excluded from the analysis. Regions 8 and 13 are the nearest neighbours of Region 15. All three are part of a positive spatial cluster [ 4 ], a spatial structure that could not be captured by the spatial linear model used for universal kriging. However, the general appearance of the data is Gaussian. This in turn justifies the appropriateness of the modelling and prediction approach based on the empirical Bayesian smoothed regional data. Figure 9 Normal probability plot of cross-validation residuals from kriging smoothed prevalences. Normal probability plot of the cross-validation residuals from kriging the empirical Bayesian smoothed regional period prevalences. Observations 3, 8 and 13 may be spatial outliers with respect to the model based on the data set with observations 3 and 15 removed. Regions 8 and 13 are nearest neighbours of region 15, all three of which are part of a positive cluster. Geostatistical modelling of the empirical Bayesian smoothed regional prevalences allows the calculation of an error map representing the kriging standard errors. The darker the error map in Figure 10 the larger are the kriging standard errors, which range up from 0 to 0.06. Figure 10 Error map. Error map based on the universal kriging standard errors. Summary and discussion In this paper two sophisticated statistical methods were combined to solve an open problem in disease mapping. Mapping regional data using choropleth maps holds on to several problems that are overcome by isopleth mapping. To interpolate the regional, i.e. spatially discrete information, the geostatistical method of kriging is used here. Kriging requires variance homogeneous data. However, regional risk estimates are generally based on varying sample sizes and consequently turn up spatially varying standard errors. Therefore the spatial risk estimates are smoothed using linear empirical Bayes estimation. By borrowing-strength-from-the-ensemble, the impact of outliers is reduced and standard errors are stabilized over space. This cannot achieve variance homogeneity entirely nor could the tendency be shown analytically. Bootstrapping methods for spatially dependent data [ 18 ] promise an empirical justification and will be a source for future research. The purpose of exploratory disease mapping is to provide insight, as opposed to precise estimates of location, spread or trends [ 19 ]. Emphasise shall be on easy and intuitive statistical mapping methods. The method proposed here for exploratory disease mapping is on one hand based on empirical Bayesian estimation for smoothing which is related to internal standardization in epidemiology [[ 9 ], p. 260] and could thus be viewed as a natural choice for adjusting the data for spatially varying variances. On the other hand, the kriging predictions are weighted moving averages, where weights are chosen with respect to the spatial autocorrelation structure exhibited by the sample data. This concept is easily communicable to map users. Only monochrome colours or shades of grey should be used along with isopleth mapping [ 36 ], as has been done here in order to avoid the natural preference of the human eye for bright colours and in deference to colour-blind map users. Kriging is known to be a smoothing method and it may be argued that the proposed mapping method results in double or over-smoothing. But here kriging is based on a semivariogram model without nugget effect, which is known to be a direct interpolation method [ 12 ]. Therefore the risk map shows the Bayesian smoothed regional risks in the respective regional centres. The proposed disease mapping approach is a two-step procedure. The corresponding error map (Figure 10 ), however, neglects the error from the first step, i.e. the smoothing step. This is in line with error maps used along with similar sandwich predictors obtained via median polish kriging [ 3 , 12 ]. Thus the error map is useful to investigate the predictive performance of the kriging step but less useful for analytical inferences. The empirical Bayes method as described here does not take into account spatial autocorrelation, but a modification to overcome this lack has been proposed [ 29 ]. The method described thus far is called global smoothing, because the Bayes estimates are shrunk towards the global mean of all regions. The modification consists of using the local mean based on neighbouring regions instead of shrinking the estimates, which is called local smoothing. In any case, a simulation-based evaluation of a wide range of estimation methods [ 26 ] shows that the global smoother performs better overall than the local smoother. (This result may be related to the fact, that also regions with higher or lower population density cluster into urbanized regions and rural surroundings.) Only the full Bayesian approach to hierarchical modelling of regional disease data [ 5 ] outperforms the global smoother. But this modelling approach is certainly not suited for exploratory work. Conclusions Interpolation of the regional disease risk estimates overcomes the areal bias and related problems of choropleth disease mapping. Consequently, isopleth maps are easier to read and interpret than choropleth maps. The geostatistical method of kriging is appropriate for this task when based on linear empirical Bayesian smoothed data. Unlike non-parametric and mathematical interpolation methods, the spatial model underlying the exploratory spatial risk map offers ways of interpretation. Finally, the proposed concept for exploratory spatial risk mapping is easily communicable to map users. The Bayesian smoothing estimator is related to internal standardization in epidemiology. Also kriging can be viewed as weighted moving averaging. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516784.xml |
503389 | Stimulation of MMP-11 (stromelysin-3) expression in mouse fibroblasts by cytokines, collagen and co-culture with human breast cancer cell lines | Background Matrix metalloproteinases (MMPs) are central to degradation of the extracellular matrix and basement membrane during both normal and carcinogenic tissue remodeling. MT1-MMP (MMP-14) and stromelysin-3 (MMP-11) are two members of the MMP family of proteolytic enzymes that have been specifically implicated in breast cancer progression. Expressed in stromal fibroblasts adjacent to epithelial tumour cells, the mechanism of MT1-MMP and MMP-11 induction remains unknown. Methods To investigate possible mechanisms of induction, we examined the effects of a number of plausible regulatory agents and treatments that may physiologically influence MMP expression during tumour progression. Thus NIH3T3 and primary mouse embryonic fibroblasts (MEFs) were: a) treated with the cytokines IL-1β, IL-2, IL-6, IL-8 and TGF-β for 3, 6, 12, 24, and 48 hours; b) grown on collagens I, IV and V; c) treated with fibronectin, con-A and matrigel; and d) co-cultured with a range of HBC (human breast cancer) cell lines of varied invasive and metastatic potential. Results Competitive quantitative RT-PCR indicated that MMP-11 expression was stimulated to a level greater than 100%, by 48 hour treatments of IL-1β, IL-2, TGF-β, fibronectin and collagen V. No other substantial changes in expression of MMP-11 or MT1-MMP in either tested fibroblast culture, under any treatment conditions, were observed. Conclusion We have demonstrated significant MMP-11 stimulation in mouse fibroblasts using cytokines, matrix constituents and HBC cell lines, and also some inhibition of MT1-MMP. Our data suggest that the regulation of these genes in the complex stromal-epithelial interactions that occur in human breast carcinoma, is influenced by several mechanisms. | Background Integral to both normal and pathological tissue remodeling, the matrix metalloproteinase (MMP) family of proteolytic enzymes collectively degrades laminin, collagen, gelatin, and other protein components of the extracellular matrix [ 1 , 2 ]. MMP expression allows degradation of the basement membrane, an event essential to the process of tumour metastasis [ 3 ]. The induction of specific MMPs such as MT1-MMP and MMP-11 have been shown to correlate with tumour invasiveness and resultant metastasis in several human cancers, including breast carcinoma [ 4 - 7 ]. Thus, gene expression analysis of these two MMPs may have clinical application in breast cancer diagnosis, management and therapy. Little is known about the mechanisms underlying MT1-MMP and MMP-11 induction. Previous studies have revealed induction of MT1-MMP in human fibroblasts grown on interstitial collagen (collagen I), which is abundant around tumours and during wound healing [ 8 ]. Furthermore, MMP-11 has been shown to be induced by retinoic acid [ 9 ]. This study aimed to investigate possible regulatory agents and treatments inducing MT1-MMP and/or MMP-11 expression. Specifically, MMP induction was measured in NIH3T3 and primary mouse embryonic fibroblasts (MEFs) subjected to each of the following potentially regulatory agents: collagen I, IV and V, concanavalin-A, fibronectin, and several exogenous cytokines. As induction by such agents could be transitory, treatment times were varied from 3, 6, 12, 24 and 48 hours, and were immediately followed by RNA extraction. As collagen I has been reported to induce MT1-MMP expression, detection of such induction by quantitative RT-PCR would provide an ideal basis for testing further regulatory agents. Collagen IV and V appeared to be ideal candidates for further analysis of MMP induction as they are present in the basement membrane and would therefore be released specifically during early tumour invasion and metastasis, which coincides with MT1-MMP and MMP-11 expression in vivo . Type IV collagen has been shown to mediate pro-MMP-2 activation in HT1080 cells, a human fibrosarcoma cell line, without inducing either a transcriptional modulation of MMP-2 or MT1-MMP expression nor any alteration of MT1-MMP protein synthesis or processing [ 10 ]. Concanavalin-A is a plant lectin reported to stimulate activation of MMP-2 [ 11 , 12 ]. Fibronectin encourages cell adhesion and HT1080 cells cultured on fibronectin have exhibited elevated levels of MT1-MMP protein [ 13 - 15 ]. We investigated the potential role of the following cytokines, interleukin (IL)-1β, IL-2, IL-6, IL-8 and transforming growth factor (TGF)-β in MT1-MMP and MMP-11 regulation. All of these cytokines are reportedly released by neoplastic cells and are capable of affecting transcription in fibroblasts [ 16 - 19 ]. IL-2 is also released by tumour infiltrating lymphocytes (TILs) and IL-6 is released by tumour associated macrophages (TAMs), TILs and fibroblasts [ 16 , 20 - 22 ]. Moreover, MT1-MMP and MMP-11 have both been reported to be induced in human fibroblasts treated with conditioned medium from MDA-MB-231, a highly invasive mammary tumour cell line [ 23 ]. Thus, we investigated the induction of MT1-MMP and MMP-11 in NIH3T3 and MEFs co-cultured with a range of HBC (human breast cancer) cell lines of varied invasive and metastatic potential. This allowed direct in situ interaction of MMPs, growth factors and other potentially regulatory agents between the fibroblasts and HBC cells. Methods Cell lines NIH3T3 culture was obtained from the ATCC, while MEF cultures were constructed at the John Curtin School of Medical Research, Australian National University. Briefly, embryos were removed from 13-day pregnant mice. Following removal of the liver and head, the remaining tissue was placed in PBS (1× pH 7.4, Sigma-Aldrich). Tissue was homogenised using an 18 guage syringe (Terumo) in PBS, plated in 25 cm 2 tissue culture flasks (Costar) DMEM (Dulbecco's modified Eagles medium) with 10% FCS (Invitrogen) and allowed to attach for 6–12 hrs at 37°C, 5% CO 2 prior to use. HBC cell lines were originally obtained from the ATCC and maintained by The Lombardi Cancer Center. ML-20 cells were derived from The Lombardi Cancer Center MCF-7 cells by clonal selection after transfection with a CMV-driven expression plasmid encoding bacterial β-galactosidase. Collagen coats and fibronectin treatment Collagen coats and fibronectin treatments were prepared on non-coated tissue culture plates as follows: prior to seeding, triplicate sets of 2 ml 24 well plates (Costar) were overlaid with 150 μl of collagen I, IV and V (Sigma, 1 mg/ml) solutions. For the fibronectin treatments, 15 μg of fibronectin (Promega), equivalent to 5 μg/cm 2 of flask area were added to each well in 2 ml 24 well plates (Costar). The wells were then incubated at 37°C for 1 hour to allow the collagen coats and fibronectin to set in the wells. Excess solution was aspirated from all wells before seeding with 10 5 cells from either NIH3T3, or MEF cultures in 1 ml DMEM - 10% fetal calf serum (Invitrogen). Treated cultures were then incubated at two time-points, 24 hours and 48 hours at 37°C, 5% CO 2 , before undergoing RNA extraction. Con-A treatment Non-coated 2 ml 24 well tissue culture plates (Costar) were seeded with either NIH3T3 or MEF at 10 5 cells in 1 mL Dulbecco's modified Eagles medium (DMEM) - 10% fetal calf serum (Invitrogen). Treatments consisted of 25 μg Con-A added to the 1 ml of DMEM in each well. Treated cultures were then incubated at two time-points, 24 hours and 48 hours at 37°C, 5% CO 2 , before undergoing RNA extraction. Cytokine treatments 2 ml 24 well plates (Costar) were seeded with 10 5 cells from either NIH3T3 or primary mouse fibroblast cultures (passage 5) in 1 mL Dulbecco's modified Eagles medium (DMEM) - 10% fetal calf serum (Invitrogen). Cultures were incubated at 37°C, 5% CO 2 for 24 hours prior to cytokine treatment, thus allowing complete attachment. All treatments were separate and consisted of final solutions of 100 units/ml IL-1β, IL-2, IL-6, 0.1 units/ml of IL-8 and 1 ng/ml pan-TGF-β (Sigma-Aldrich) in cultures grown in triplicate. Co-cultures 30 mm 6 well plates (Costar) were seeded with 10 6 cells of either NIH3T3 or MEFs (passage 5), which were each allowed to attach overnight in 1 mL Dulbecco's modified Eagles medium (DMEM) - 10% fetal calf serum (Invitrogen). The DMEM was then aspirated and 24 mm Transwell Porous Cell Culture Inserts (Costar) were seeded with 10 5 cells of either MDA-MB-231, MDA-MB-435, MDA-MB-436, MCF-7s or ML-20 epithelial tumour cultures in 2 ml of RPMI medium (Trace Biosystems) - 10% fetal calf serum. Use of the porous inserts meant that co-cultures were not in direct physical contact, but shared growth media at all times. These HBC cultures exhibit different levels of invasive and metastatic potential (see Table 1 ). Each co-culture then underwent 48 hour incubation at 37°C, 5% CO 2 , before RNA extraction. RNA extraction and quantification Total RNA was extracted from the fibroblasts using an RNeasy Mini Kit (QIAGEN) as previously described [ 24 ]. Total RNA was then quantified at OD 260 using a QuantaGene spectrophotometer (The Australian Chromatography Company, Sydney, NSW, Australia). Oligonucleotide primers All 3' primers were labeled with the TET dye phosphoramidite and were commercially obtained (Applied Biosystems, Brisbane, QLD, Australia). Oligonucleotides for the target genes, MT1-MMP and MMP-11, were generated using Amplify 2.1 analysis software in conjunction with published cDNA sequences [ 6 , 25 ]. The oligonucleotide sequences for the housekeeping gene 18S rRNA were supplied by Ambion. All sequences utilised are outlined in Table 2 . Constructing cDNA competitive templates cDNA competitive templates for the target genes, MT1-MMP and MMP-11 were made by PCR amplification using purpose built 47-mer oligonucleotides as described [ 26 ]. The 47-mers consisted of a 23 bp 5' primer complementary to target sequence either 28 bp, for MT1-MMP, or 20 bp, for MMP-11, further 5' than the remaining 24 bp of the 47-mer, which consisted of the normal 5' primer sequence for PCR. Hence, a PCR product generated from these 5' 47-mers in combination with the normal 3' primers, produced fragments shorter than when using the normal 5' primer, as these deletions were left out during sequence extension complementary to the 47-mer. Both 5' and 3' ends of the competitive template remain complementary to normal primers and therefore undergo further complementary-copying during additional PCR in a manner identical to the target sequences, except that the missing nucleotide fragments remain absent. (Table 2 ) Constructing mRNA competitive templates The cDNA competitive templates generated by PCR with the 47-mers described above were each ligated into pGem-T Easy Vectors (Promega, Annandale, NSW, Australia). JM109 (Promega) competent cells were then transformed and grown overnight. Plasmid DNA was extracted using a Magic Miniprep kit (Promega) and linearised by nuclease digest. Mimic RNA was generated from the "mimic" plasmids by transcription of the insert with T7 RNA polymerase (Promega). The post-transcription mixture then underwent RQ1 DNase (Promega) digestion; the remaining RNA was then purified using QIAGEN RNeasy Minikit, RNA competitive template was then resuspended in DEPC-treated H 2 O and stored at -70°C. To ensure that amplification would not occur as a result of residual DNA contamination, control samples were run in triplicate without the reverse transcriptase thermal cycle. No amplification products were visualised on resultant agarose gels. RT-PCR Single tube, one step RT-PCR was performed as previously described [ 24 ] for all samples using an ABI 480 thermal cycler in 25 μl volume containing: 24 ng total RNA, 1–2 μl of RNA competitive template, being equivalent to 24 ng of control total RNA, 200 mM dNTPs (Promega), 5× RT-PCR buffer [300 mM Tris-HCl, pH 8.3; 2.5 mM DTT; 250 mM KCl; 0.5% Triton X-100 (Evergreen Scientific); 30 μm EDTA; 7.5 mM MgCl 2 ], 0.4 μM each primer, 1 unit Taq polymerase (Perkin Elmer ABI), 1.2 units AMV reverse transcriptase (Promega) and 900 ng tRNA (Boehringer Mannheim)]. Thermal cycling was as follows: 30 mins 50°C (reverse transcription), then 2 mins 95°C, followed by 35 cycles of 30 sec 95°C, 30 sec 65°C, 45 sec 72°C, followed by a final incubation of 5 mins at 72°C. GeneScan 3' primers for MT1-MMP, MMP-11 and 18S rRNA were labeled with TET fluorescent dye, thus allowing RT-PCR samples to be analysed by capillary electrophoresis using a 310 Genetic Analyzer with GeneScan software (ABI) as described previously [ 24 ]. Using laser technology to excite the fluorescently labeled primers, samples were sized to within one base pair and the amount of amplification product of any given size, determined by fluorescence peak area. Ratio calculations The GeneScan values for fluorescence peak area of the target gene transcripts and the control transcripts produced the target:control ratio. All graphed values represent averages from triplicate samples and were adjusted to produce a control value of 1.00. These target:control ratio values were used to plot time series graphs of all combined data for each gene. Data on the time series graphs was then normalised to 18S rRNA expression, as measured in the same respective cultures. The 0.5× control and 2.0× control are the resultant target:control ratios produced using half and double the amount of total RNA used in the control sample, respectively. Results Induction treatments MT1-MMP: 3, 6, 12 and 24 hour treatments Competitive quantitative RT-PCR using capillary electrophoresis and GeneScan analysis revealed no substantial changes in expression of MT1-MMP in either NIH3T3, or MEF cultures under any treatment conditions carried out over 3, 6, 12, and 24 hours (Tables 3 and 4 ). Although as indicated in Table 3 there are some minor initial variations, when the normalised to 18S rRNA no substantial changes are observed. The data presented are derived from GeneScan values for fluorescence peak area of the target gene transcripts and the control transcripts, thus producing the target:control ratio. All values represent averages from samples run in triplicate and have been adjusted to produce a control value of 1.00. Results for these time periods display only minor variations and generally fall between 0.5× control and 2.0× control expression levels. MT1-MMP: 48 hour treatments Induction Initial analyses of the 48 hour IL-1β, IL-2 and fibronectin treatments indicated consistently basal MT1-MMP expression levels (Table 5 ). Yet data for these treatments do indicate induction when normalized to the unexpectedly low level of 18S rRNA expression (Tables 7 and 8 ). However, 18S rRNA expression is monitored primarily to double check for substantial anomalies in mRNA quantification when positive results are found, rather than as a precise means of normalization for results which appear consistent. Furthermore, competitive RT-PCR of MT1-MMP may be considered more accurate than non-competitive differential RT-PCR of 18S rRNA. Initial competitive RT-PCR of MT1-MMP expression did produce consistently unchanged results for these treatments, while expression of 18S rRNA was found to display a relatively high level of variation in this treatment group. Specifically, the level of 18S rRNA expression exhibited in the control was high, therefore causing all other treatments to appear relatively low, which in turn caused an artificial elevation of all treatment data values during normalization. Hence, a consistent group result for competitive RT-PCR of MT1-MMP should not be discarded in favour of a single set of unusually efficient differential RT-PCRs of 18S rRNA control samples. In our opinion, MT1-MMP expression should be considered unaffected by IL-1β, IL-2 and fibronectin treatments during this time schedule. Inhibition Results indicated substantial inhibition of MT1-MMP expression by collagen I and collagen IV (Tables 5 and 6 ). Expression values for these treatments remained well below the 0.5× control expression level even after normalization (Tables 7 and 8 ), despite this procedure raising the values in a way that may have been to some degree artificial, as described above. MMP-11: 3, 6, 12 and 24 hour treatments Competitive quantitative RT-PCR using capillary electrophoresis and GeneScan analysis revealed no substantial changes in expression of MMP-11 in either fibroblast culture under any treatment conditions carried out over 3, 6, 12 and 24 hours (Tables 3 , 4 and 5 ). Results for these time periods display only minor variations and generally fall between 0.5× control and 2.0× control expression levels. The use of an internal competitor to obtain increased specificity of target:template detection demonstrate data for these time periods and is highly consistent. Such consistency within groups of treatments carried out simultaneously (i.e: over an individual time schedule) further suggests that the minor variations from control expression levels which are present, may be due to amplification anomalies rather than individual differences in gene expression caused by specific treatments. MMP-11: 48 hour treatment At the 48 hour time point, MMP-11 expression appeared to be stimulated at least 2 fold by IL-1β, TGF-β, fibronectin and collagen V. Results indicate induction of MMP-11 by these agents independently of normalization to 18S rRNA expression (Tables 7 and 8 ), which further enhanced these expression values. IL-2 and IL-6 appeared to produce a very mild induction of MMP-11 without normalization to 18S rRNA, and only appeared to produce a greater than 2 fold induction when undergoing such normalization. The data suggesting that MMP-11 expression is induced by IL-1β, TGF-β, fibronectin and collagen V is highly persuasive given that these are the only data that demonstrate considerable within-group variation. The standard deviation for the 48 hour MMP-11 treatments was 47.6 while the second greatest standard deviation value was 40.5 for the 24 hour MMP-11 treatments. The average for all other MMP-11 groups was 17.3. This increase over the average variation is suggestive of differences in initial transcript levels for individual treatments. Furthermore, the fact that MT1-MMP expression levels (non-normalised) remained consistently normal for this treatment, effectively acting as an additional unregulated control, lends further credit to the likelihood of specific regulation of MMP-11 after 48 hours of treatment. There are also mild trends toward up regulation for the IL-2 and IL-6 treated samples. Co-culture with HBC cell lines Initial results demonstrate no induction or inhibition of either MT1-MMP or MMP-11 in mouse fibroblasts by co-culture with any of the HBC cell lines (Table 9 ). Results are highly consistent and all treatment values fall between 0.5× control and 2.0× control. Moreover, minor variations from control expression values observed for each co-culture are almost identical, between the two target genes. Hence, the mild expression elevations seen with ML-20 and MDA-MB-436 for both MT1-MMP and MMP-11 are in our opinion, artifactual. Discussion This study investigated the effects of a number of possible regulatory agents on the expression of two members of the MMP gene family for their role in human breast cancer metastases. NIH3T3 and MEF cultures were treated with cytokines, collagens, fibronectin, Con-A or matrigel, and co-cultured with various HBC cell lines. The expression of MT1-MMP and MMP-11 was then determined by competitive RT-PCR using capillary electrophoresis and GeneScan technology. Only MMP-11 expression at 48 hours was affected by treatment with IL-1β, TGF-β, fibronectin and collagen V, in the tested fibroblasts cultures. During tumour progression, the release of exogenous cytokines by neoplastic cells and immunological cells may cause a stimulatory effect on adjacent stromal fibroblasts, resulting in matrix metalloproteinase (MMP) induction. Past studies indicate that some MMPs undergo regulatory changes under the influence of cytokines [ 18 , 27 ]. Another recent study reported the induction of MMP-11 expression in primary cultured human fibroblasts over a 48 hour period by IL-6, IGF-2, EGF and PDGR-BB, but not by IL-1β or TNF-β [ 28 ]. Similar to the MT1-MMP results presented in the present study, MT1-MMP gene expression was reportedly not affected by insulin-like growth factor (IGF)-2, epidermal growth factor (EGF), platelet derived growth factor (PDGR)-BB, IL-6, TNF-β or IL-1β [ 28 ]. Any influence exerted on MT1-MMP by cytokines may be of particular interest because of its specific interaction with MMP-2, an MMP with a broad substrate specificity. Hence, further analysis of candidate stimulatory agents is required to elucidate possible mechanisms by which MT1-MMP expression is stimulated. Our study did however, find evidence for an inhibitory effect from both collagen I and collagen IV on MT1-MMP expression. Curiously, there is evidence suggesting that collagen I actually stimulates MT1-MMP mRNA production in fibroblasts and activates MMP-2 in a variety of cells over a longer time period [ 8 , 29 ]. However, past studies on rat smooth-muscle cell culture support findings that collagen IV exhibits MT1-MMP inhibitory effects [ 12 ]. Despite reports of MDA-MB-231 conditioned media inducing MT1-MMP expression [ 23 ] in human fibroblasts, co-culture between MEFs and MDA-MB-231, did not produce induction of either MT1-MMP or MMP-11 in this study. This finding is likely to be due to species-specificity, since mouse mammary stromal cells are not capable of supporting normal human mammary reorganization in vivo [ 30 ]. Although the co-cultures carried out here did not allow cell-cell contact, the previous study also involved no cell-cell contact, but rather treatment with pre-conditioned media (application of isolated medium from one cell population following in vitro expansion to an another isolated cell population). Another possible explanation is that the basal level of expression of MT1-MMP and MMP-11 in cultured murine fibroblasts is elevated, thus rendering them refractory to exogenous regulatory agents. We are currently investigating this possibility. Conclusion This study examined the effects of a number of regulatory agents and treatments on the induction of MMP-11 and MT1-MMP gene expression, factors that may influence breast cancer tumour progression. Stimulation of MMP-11 was demonstrated after treatments by both cytokines (IL-1β, TGF-β) and by matrix constituents (collagen IV and fibronectin), whilst MT1-MMP was inhibited by matrix constituents (collagen I and fibronectin). The results presented suggest that several mechanisms may be involved in MMP-11 and MT1-MMP regulation as part of the complex epithelial-stromal interactions that occur within human breast carcinomas. Competing interests None declared. Authors contributions S.S. performed the molecular genetic and in vitro studies, designed PCR primers and drafted the manuscript. L.M.H. contributed to the manuscript design and finalisation. E.W.T. contributed toward the design of the study. K.I.M. contributed toward the design and performance of competitive PCRs. M.G.I. participated in the conception and design of the study. L.R.G. participated in the conception and design of the study and its coordination. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC503389.xml |
548149 | Comparison of nested case-control and survival analysis methodologies for analysis of time-dependent exposure | Background Epidemiological studies of exposures that vary with time require an additional level of methodological complexity to account for the time-dependence of exposure. This study compares a nested case-control approach for the study of time-dependent exposure with cohort analysis using Cox regression including time-dependent covariates. Methods A cohort of 1340 subjects with four fixed and seven time-dependent covariates was used for this study. Nested case-control analyses were repeated 100 times for each of 4, 8, 16, 32, and 64 controls per case, and point estimates were compared to those obtained using Cox regression on the full cohort. Computational efficiencies were evaluated by comparing central processing unit times required for analysis of the cohort at sizes 1, 2, 4, 8, 16, and 32 times its initial size. Results Nested case-control analyses yielded results that were similar to results of Cox regression on the full cohort. Cox regression was found to be 125 times slower than the nested case-control approach (using four controls per case). Conclusions The nested case-control approach is a useful alternative for cohort analysis when studying time-dependent exposures. Its superior computational efficiency may be particularly useful when studying rare outcomes in databases, where the ability to analyze larger sample sizes can improve the power of the study. | Background The nested case-control design employs a case-control approach within an established cohort [ 1 , 2 ] to obtain estimates from a sample of the cohort that are similar to estimates obtained from analysis of the entire cohort [ 3 , 4 ]. The nested case-control design is being increasingly used in large cohorts of patients from prospective studies and randomized clinical trials. This design has become popular because it allows for statistically efficient analysis of data from a cohort with substantial savings in cost and time [ 5 , 6 ]. When studying exposures that vary with time, an additional level of complexity is introduced by the need to account for time-dependent exposure in both the design and analysis [ 7 , 8 ]. This can be accomplished by including time-dependent covariates in a Cox proportional-hazards regression model [ 9 ]. Alternatively, a nested case-control approach can be used provided that the exposure and covariate information for controls reflects values corresponding to the time of selection of their respective case. This study compares nested case-control and survival analysis methodologies for evaluating time-dependent exposure. The risk of pacemaker insertion associated with dosage of amiodarone (an anti-arrhythmic medication used for the treatment of atrial fibrillation (AF)), represented by a time-dependent covariate, is evaluated in cohort of patients with AF using both methods for illustrative purposes. The comparability of results is evaluated and differences in computational efficiency are quantified for increasing cohort sizes. Advantages and limitations of the respective methodologies are discussed. Methods Study cohort A cohort of 11395 elderly (>65 years of age) Quebec residents with AF and a myocardial infarction (MI) between 1991 and 2000 was created by linking the provincial hospital discharge summary database with the provincial physician and drug claims database, using methods described previously [ 10 ]. Approval for the study was obtained from the McGill University Faculty of Medicine Institutional Review Board. In order to evaluate the effect of amiodarone dose on the previously demonstrated association between amiodarone therapy for AF and an increased risk of permanent pacemaker requirement [ 10 ], only patients newly started on amiodarone after their diagnosis of AF were included in the study cohort. Amiodarone dose was represented as a binary time-dependent variable comparing daily doses >200 mg to ≤200 mg. Covariate information included age, sex, calendar year of cohort entry, baseline sinus node or conduction disorder, ventricular arrhythmia, and time-dependent exposure to five categories of medications. The final study cohort included 1340 subjects followed from the date of their first prescription of amiodarone until the first of pacemaker implantation, death, or March 31, 2001. Statistical analysis The data for the entire cohort of 1340 subjects including all fixed and time-dependent variables was represented in counting process notation suitable for Cox regression of time-dependent exposure [ 11 ]. Multiple records (with consecutive start and end times) were created for each subject to account for every change in exposure to any of the time-dependent variables over the study period. The hazard ratio (HR) of pacemaker insertion associated with amiodarone doses >200 mg per day was estimated using a Cox proportional-hazards model including all fixed and time-dependent covariates. The timescale used in the model was time since first prescription of amiodarone. Non-significant variables (other than age and sex) were sequentially removed if the resultant model had no significant increase in Akaike Information Criteria (AIC) and no significant change in the HR for amiodarone dose. The nested case-control approach was also used to estimate the HR of pacemaker insertion associated with amiodarone doses >200 mg per day. Cases of pacemaker insertion were identified and controls were randomly selected from the risk-set of each case (i.e. subjects present in the cohort at the time the case is defined). After selecting all controls and recording their index dates (i.e. the time, in cohort time, at which the respective case is defined) the relevant time-dependent covariate information was retrieved by merging with the database configured in counting process notation. The relevant subject record was selected by requiring that the index date fall within the start and end time of the subject record for each control. The nested case-control approach was repeated using 4, 8, 16, 32, and 64 controls per case. For each number of controls per case, random sampling of controls for all cases and conditional logistic regression analysis was repeated 100 times using the OUTEST option in the PROC PHREG statement to create an output SAS data containing all the parameter estimates [ 11 ]. The mean and standard deviation (SD) of the parameter estimates for each number of controls per case was calculated. Computational times for regression models of time-dependent exposures using nested case-control and survival analysis methodologies were compared. The nested case-control samples with 4 and 32 controls per case were analyzed using conditional logistic regression with the PHREG procedure in SAS Release 8.2 [ 12 ]. The full cohort was analyzed using Cox regression adapted for analysis of time-dependent covariates with the PHREG procedure in SAS Release 8.2. Ties were handled using the TIES = EFRON option in the PHREG procedure [ 11 ]. All analyses were performed using an Intel Pentium 4 computer with a 1.80 GHz central processing unit (CPU) and 256 MB of random access memory (RAM). Relative computational efficiencies were evaluated by comparing the CPU times of the three regression models used to analyze the cohort. Relative increases in computational time as a function of sample size were quantified by repeating the analyses on progressively larger cohorts. This was done by progressive doubling of the original cohort to 2, 4, 8, and 32 times its original size. Given that the objective was to compare computational times, all fixed and time-dependent exposures were included in all models regardless of statistical significance. The computational efficiency of the Cox regression model was also compared to the conditional logistic regression model where the nested case-control sample included all possible controls for each case. This comparison was performed for the original cohort of 1340 subjects with 53 cases (including 2 ties). All analyses were performed using the PHREG procedure. Ties were handled using the TIES = EFRON option in the PHREG procedure, and subsequently using the TIES = DISCRETE option in the PHREG procedure for comparison. Results Comparative risk estimates Pacemaker implantation occurred in 53 of the 1340 subjects during the study period. In the final Cox regression model, amiodarone daily dose (>200 mg vs. ≤200 mg) was associated with an increased risk of pacemaker insertion (HR: 2.03; 95 percent confidence interval: 1.00, 4.14; p = 0.05; Parameter Estimate: 0.71; Standard Error of Parameter Estimate: 0.36), after adjusting for age, sex, as well as baseline sinus node or conduction disorder (the only covariate that was an independent predictor of outcome). The results of 500 nested case-control analyses (i.e. 100 for each of 4, 8, 16, 32, and 64 controls per case) are summarized in Table 1 . When using any number from 4 to 64 controls per case, the mean point estimate (of 100 analyses repeating the random sampling of controls) of the parameter estimate (and HR) was very similar to that obtained using Cox regression on the full cohort (i.e. HR: 2.03; Parameter Estimate: 0.71). The SDs of the parameter estimates decreased with increasing numbers of controls per case (Table 1 ). Table 1 Nested case-control analyses with repeated sampling for increasing numbers of controls per case: Hazard ratio of pacemaker insertion associated with amiodarone dose* in 1340 elderly Quebec residents with atrial fibrillation Controls per Case (n) Repeated Sampling (n) Mean HR† SD† of HR Min HR Max HR Mean Parameter Estimate SD of Parameter Estimates 4 100 2.14 0.51 1.27 3.59 0.73 0.23 8 100 2.19 0.41 1.39 3.86 0.77 0.18 16 100 2.02 0.22 1.55 2.51 0.70 0.11 32 100 2.07 0.15 1.78 2.55 0.73 0.07 64 100 2.02 0.11 1.82 2.33 0.70 0.05 * The effect of amiodarone dose, represented by a binary time-dependent covariate (>200 mg vs. ≤200 mg), was adjusted for age, sex, and baseline sinus node or conduction disorder in all models. † HR, hazard ratio; SD, standard deviation. Comparative computational efficiencies The computational time required to analyze the cohort of 1340 subjects with 53 events (cases) in models including four fixed variables and seven time-dependent variables is presented in Table 2 . CPU times are displayed for the Cox regression models and the nested case-control regression models (with 4 or 32 controls per case). For the cohort in its initial size, using 32 rather than 4 controls per case increased CPU time by a factor of 3, whereas using Cox regression increased CPU time by a factor of 42. As the cohort size was increased to 32 times the original (i.e. 42880 subjects), the increase in CPU time was greater for the Cox regression model than the nested case-control models. Figure 1 displays graphically the increase in CPU time with increasing sample size for nested case-control and Cox regression models. The relative computational efficiency was magnified as the sample size increased, such that the CPU time for Cox regression with a cohort of 42880 subjects was 125 times greater than the nested case-control model with 4 controls per case (Table 2 ). Table 2 Computational times for nested case-control and survival analyses of time-dependent data for cohorts of increasing sizes: Models* of the risk of pacemaker insertion in elderly Quebec residents with atrial fibrillation Cohort Size†: #Subjects (#Cases) 1340 (53) 2680 (106) 5360 (212) 10720 (424) 21440 (848) 42880 (1696) Cohort Size: multiple of original 1 2 4 8 16 32 CPU‡ Time (seconds): Nested: 4 Controls per Case 0.03 0.05 0.07 0.11 0.18 0.4 Nested: 32 Controls per Case 0.08 0.1 0.2 0.41 0.7 1.49 Survival Analysis 1.26 2.51 5.06 9.54 19.53 49.91 CPU Time (multiple of Nested 4): Nested: 4 Controls per Case 1 1 1 1 1 1 Nested: 32 Controls per Case 3 2 3 4 4 4 Survival Analysis 42 50 72 87 109 125 * All models included four fixed variables (age, sex, calendar year of cohort entry, and baseline sinus node or conduction disorder) and seven time-dependent variables (amiodarone dose, ventricular arrhythmia, and exposure to sotalol, class I antiarrhythmic agents, beta-blockers, calcium channel blockers, and digoxin). † The original cohort of 1340 subjects was increased by progressive duplication to 2, 4, 8, 16, and 32 times its original size. ‡ CPU, central processing unit. Figure 1 Increase in computational time with increasing sample size for nested case-control and survival analysis of cohort data with time-dependent covariates The computational time of the Cox regression model was also compared to the nested case-control model including all possible controls for each case. When ties were handled using the TIES = EFRON option, the CPU time for Cox regression was 1.06 times greater than the CPU time for nested case-control model (1.26 vs. 1.19 seconds). When ties were handled using the TIES = DISCRETE option, the CPU time for Cox regression was 3.91 times greater than the CPU time for nested case-control model (14.65 vs. 3.75 seconds). Discussion In this study we illustrate empirically that a nested case-control approach can be used to analyze a cohort with time-dependent covariates, with results that are similar to those obtained by Cox regression. Additionally, given that the nested case-control approach obviates the computationally intensive calculations involved in Cox regression when time-dependent covariates are used, the example also illustrates quantitatively the large reduction in CPU time required for analysis. The similarity between the two methodologies is expected given that conditional logistic regression used to analyze nested case-control studies (as well as other matched case-control studies) is based on inference procedures adapted from Cox regression; i.e. the conditional likelihood used in conditional logistic regression is exactly the same form as the partial likelihood used in Cox regression except that the denominator includes only a selected number of sampled controls as opposed to all subjects available in the risk set [ 13 ]. The inclusion of time-dependent covariates adds an additional level of complexity to the analysis but remains based on the same inference procedures. The statistical efficiency of the nested case-control approach for cohort analysis depends on the number of controls per case selected. Our example demonstrates the expected decrease in the SD of the parameter estimates as the number of controls per case increases. This decrease in variance is explained by the fact that as the number of controls per case increases (towards the total number of controls in a case's risk-set), the probability of choosing the same controls increases, as does the proportion of available controls selected (i.e. approximating the situation in Cox regression where every case is compared to all controls in its risk-set). In general, the use of 4 controls per case provides a relative statistical efficiency of 0.8 compared to the use of an infinite number of controls [ 14 ]. However, the relative efficiency also depends on the probability of exposure among the controls and on the magnitude of the estimated relative risk. Gains in statistical efficiency are possible by using greater than 4 controls per case particularly when the probability of exposure among the controls is <0.1 [ 4 , 15 ]. In addition to situations where exposure prevalence in controls is low, increasing the number of controls per case is beneficial when the number of case-control sets is small [ 16 ]. The major reason for the superior computational efficiency of the conditional logistic regression method for nested-case control analysis of time-dependent covariates is that only a sample of all possible controls are included in the risk set of each case (whereas all are included in Cox regression). As illustrated in Figure 1 and Table 2 , the impact on computational efficiency of sampling a fixed number of controls per case is greater for larger cohorts because the sample of controls represents a smaller proportion of the all the possible controls for each case. While this effect of sampling controls may be the main reason for the computational efficiency of the nested case-control approach is not the only reason. As demonstrated, even when all possible controls are included in the risk set of each case, the computational time of the conditional logistic regression increases significantly but remains faster than Cox regression. This is because the two analyses process time-dependent covariates differently. In Cox regression, risk sets and time-dependent covariates are calculated at the time of each case failure. In conditional logistic regression, the risk sets and time-dependent covariates are calculated in advance. The relative efficiency also depends on how ties are handled, with Cox regression relatively less efficient when ties are handled using the TIES = DISCRETE option compared to the TIES = EFRON option. The nested case-control approach for cohort analysis offers some advantages over analysis of an entire cohort that may be important regardless of the type of cohort used. A potential advantage with respect to design is the option to match controls to cases on the basis of possible confounding covariates for which estimation of effect is not of interest. Another advantage is that substantial savings in cost and time can be achieved by analyzing the cases and only a sample of the controls (as opposed to the entire cohort), particularly when the collection and/or processing of exposure information is very expensive and/or time consuming [ 6 ]. While cost is often a major factor in preferring a nested-case control approach over analyzing an entire cohort, there may be advantages even when differences in cost are not significant. In recent years, large administrative healthcare databases, such as the one from which the example cohort for this study was selected, have become particularly useful in studying outcomes that are very rare because they allow for adequate sample sizes [ 17 - 20 ]. Once a database with all exposure and outcome information is available, analyzing a sample of rather than the entire cohort does not necessarily decrease costs. However, depending on the size of the cohort (as well as the speed of the processor and amount of memory in the computer), it may not be possible to analyze an entire cohort when complex modeling of time-dependent covariates is needed. Such was the case in another study based on a cohort derived from the Quebec provincial healthcare database, where in one analysis the number of covariates was restricted to four and in another analysis only a sub-cohort (i.e. 15529 of 31062 subjects) could be included because the substantial computational resources required were prohibitive [ 21 ]. Depending on the rarity of the outcome under study, it may not be possible to analyze the required sample size when performing Cox regression on the whole cohort, whereas it may be possible to do so using a nested case-control approach. While it is recognized that issues of computational resources are overcome with time as computers and software become more efficient, limitations are likely to remain as the size of databases and complexity of time-dependent analyses will also increase. Both the nested case-control approach described (using conditional logistic regression) and the Cox proportional-hazards model with time-dependent covariates similarly account for the time-dependence of exposure when levels of exposure in subjects vary over time. A different and more complex issue is the possibility that the effect of a given exposure varies over time. This can be addressed by analyzing latency-weighted exposures using either Cox regression or a nested case-control approach, the latter being computationally faster [ 22 ]. Alternatively, Cox regression can accommodate changes in the hazard ratio over time with a flexible generalization of the Cox proportional hazards model using a regression spline technique [ 23 - 25 ]. Conclusions A nested case-control approach is a useful alternative for analysis of a cohort when time-dependent covariates are used. The expectedly similar risk estimates are obtainable with superior computational efficiency. Particularly when studying the effects of time-dependent exposures on rare outcomes in very large databases, study power can be improved by being able to run complex regression models on a larger number of affected subjects. Competing interests The author(s) declare that they have no competing interests. Authors' contributions All authors participated in the conception and design of the study. VE performed the statistical analysis and drafted the manuscript. All authors contributed to the interpretation of the study and revision of the manuscript. The final manuscript was read and approved by all authors. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC548149.xml |
516035 | The impact of surgery and mild hyperthermia on tumor response and angioneogenesis of malignant melanoma in a rat perfusion model | Background The aim of this experimental study was to determine the effect of mild hyperthermia on tumor response and angioneogenesis in an isolated limb perfusion model with a human melanoma xenograft. Methods A human melanoma xenograft was implanted into the hindlimbs of 30 athymic nude rats. The animals were randomized into five groups: group I: control, group II: sham group, group III: external hyperthermia with a tissue temperature of 41.5°C for 30 minutes without ILP, group IV: normothermic ILP (tissue temperature 37°C for 30 minutes, group V: hyperthermic ILP (tissue temperature 41.5°C for 30 minutes). Tumor response was evaluated by tumor size determination and immunohistochemical analysis 6 weeks postoperatively. Tissue sections were investigated for expression of CD34 and basic fibroblast growth factor (bFGF). Results Average tumor volumes of the controls (I) increased from 105 mm 3 to 1388 mm 3 . In the sham operated group (II) tumor volumes were significantly larger than in group I. Tumor volumes in group IV were significantly smaller than in group I and lowest in group V. There were no significant differences in size between group I and group III after six weeks. In group III and IV each, 5 animals showed tumor progression and one had a partial tumor response. In group V only 2 animals showed tumor progression. Immunhistochemical analysis of the tissue sections demonstrated that angioneogenesis was more pronounced in group II than in group I and less pronounced in group IV and V compared with group I. Conclusions Our results suggest that even a surgical manipulation such as a skin incision promotes tumor growth, probably by induction of growth factors like bFGF. External hyperthermia of 41.5°C tissue temperature for 30 minutes only has no impact on tumor growth and angioneogenesis in vivo. | Background Hyperthermic isolated limb perfusion is an established treatment for multiple, locoregional intransit metastases of malignant melanoma and soft tissue sarcomas of the extremities. It was introduced 1957 by Creech and Krementz [ 1 ]. The procedure allows a high locoregional concentration of cytostatics with few systemic side effects [ 2 ]. Isolated limb perfusion achieves local tumor control in a high percentage of cases. Complication rate of perfusion is acceptably low. A surviving model for isolated perfused rat hindlimbs was reported by Wu et al. [ 3 ] based on the perfusion model of Nagel et al.[ 4 ]. Angioneogenesis has been recognized for many years to play a central role in the growth of primary tumors and the formation of metastases in general [ 5 , 6 ]. Melanomas express basic fibroblast growth factor (bFGF) and fibroblast growth factor receptor-1 (FGFR-1) in their dermal nevocytes and in the stroma. bFGF promotes angiogenesis in vivo [ 7 ] and in vitro [ 8 ]. Antisense targeting of bFGF/FGFR-1 in malignant melanomas blocks intratumoral angioneogenesis [ 9 ]. Hyperthermia inhibits angioneogenesis. Some of the antineoplastic effects of hyperthermia are caused by ischemia due to obstruction or destruction of the tumor vessels. Eikesdal et al. described by hyperthermic temperatures disrupture of 25–50 % of the vasculature in malignant tumors. The heated tumors (44°C) had a blood flow reduction of 40–60 % after 24 hours [ 10 ]. In vitro data indicate that capillary endothelials cells of malignant neoplasms are more thermosensitive than endothelial cells of normal tissues. The vascularization of tumors can be significantly damaged at temperatures which may alter but do not damage the vasculature of normal tissue [ 11 ]. The extent of the proliferation of vessels is also inversely related to temperature [ 12 ]. The critical temperature at which a direct vascular damage occurs is between 42.7 and 43.7 degrees C [ 13 ]. Hyperthermia increases the effects of angioneogenesis inhibitors on tumor growth and tumor angioneogenesis [ 14 ]. The aim of this experimental study was to determine the impact of hyperthermia on tumor response and angioneogenesis in an isolated limb perfusion model. Methods Thirty inbred male nude athymic rats (Rowett rnu/rnu) weighting 200 to 280 g were used in this study. Animals were kept separately during the experiment with 12 hours of light per day. They were fed a standard laboratory diet and tap water ad libidum . Maintenance and care of all experimental animals were carried out according to the guidelines of the local responsible Animal Protection Commission and carried out in compliance with national guidelines (National Institute of Health for Use of Laboratory Animals). A solid sample (2 × 2 × 2 mm) of a human melanoma xenograft (SK-MEL-3), derived from a lymph node melanoma metastasis, was implanted into the hindlimbs of 30 nude rats. The largest width and the maximum tumor diameter perpendicular to the width were measured with a micrometer. Only tumors with a diameter greater than > 12 mm in the largest diameter were included in the study. The details of the perfusion system have been described previously [ 4 ]. The equipment consisted of a miniature oxygenator, a heat exchanger and a roller pump. Venous blood of the limb was oxygenated in the oxygenator (99.15% O 2 ; 0.85% CO 2 ) and warmed in the heat exchanger. The warmed arterialized perfusate is driven by a roller pump with two synchronously running pump-heads on a single axis for the arterial and venous lines (Masterflex ® ). The outer diameters of the tubes corresponded with the diameter of the femoral vessels of nude rats weighting about 250 g (arterial 0.7 mm; venous 1.0 mm). Rat hindlimbs were perfused with ringer's solution and sodium heparin (25 IU/100 μl saline) at 37 or 41.5 degrees C for 30 minutes with a flow rate of 4 ml/min. External hyperthermia was applied by an infrared lamp positioned in a fixed distance to the tumor. For temperature measurement during limb perfusion, a nickel-chrom-nickel thermocouple of 0.6 mm in diameter (Standard Integrated Thermocouple Thermocoax, Phillips, Hamburg, Germany) was placed at the macroscopic tumor margin. The thermocouple was calibrated before use in a high-precision water bath. Baseline temperature was recorded for 5 minutes before treatment. Temperature was continuously measured during application. In group I (control) and group II (sham operated) temperature measurement was not performed. The animals were randomized into a control and four study groups of six animals each: group I: control (no therapy), group II: sham group (skin incision without ILP), group III: external hyperthermia with a tissue temperature of 41.5°C for 30 minutes without ILP, group IV: normothermic ILP (Ringer's solution as perfusate, tissue temperature 37°C for 30 minutes, group V: hyperthermic ILP (tissue temperature 41.5°C for 30 minutes). ILP was subsequently performed with the above-mentioned parameters. Antineoplastic agents were not applied. The rats were anaesthetized with Ketamin (Ketanest ® , 80 mg/kg i.m.; Pearl-Davis, Berlin, Germany) and Xylazin (Rompun ® , 10 mg/kg i.m.; Bayer, Leverkusen, Germany), and fixed in a supine position. For perfusion, the animals underwent a 3-cm incision of the groin. The left femoral artery and vein were isolated using an operation microscope. A vascular clip was placed across the artery and a silicon catheter (outer diameter 0.7 mm) was introduced via a transverse arteriotomy. The femoral vein was cannulated with a silicon catheter of 1 mm outer diameter (Fig. 1 ). Tourniquets were applied around the left hindlimb to ensure isolation. Rat hindlimbs were perfused at a flow rate of 4 ml/h, for 30 min. After treatment, the silicon catheters were removed, and the vein and the artery were sutured (8/0 Prolene), to restitute normal blood flow. After the isolated limb perfusion the wound was closed in two layers. Total operation time rated about 90 min. During the first 5 days after operation the rats were weighed daily, and once a week there-after. The animals were kept under standardized conditions and were killed by an overdose of anesthetic and cervical dislocation 6 weeks after treatment. The greatest tumor width and the maximum diameter perpendicular to the width were macroscopically measured with a micrometer. Lesion volume was calculated using the formula v = 4 × π × a × b 2/3, in which a and b are the radii of the measured axes. The tumor was then completely sectioned. Each tumor was divided into two parts. One part was fixed in 4 per cent formalin solution for 5 days, embedded in paraffin and stained with hematoxylin and eosin (H&E). The other part was fixed in 0.9 % saline solution, frozen and stored at -80°C until use. Vascular density and tumor morphology were evaluated by immunohistochemistry. Tumor tissue was tested for expression of basic fibroblast growth factor (bFGF) and CD34. Tissue specimens were fixed in 10 % formaldehyde and embedded in paraffin according to routine protocols. Sections 4 μm thick were deparaffinized and rehydrated using graded ethanols following routine protocols. For antigen retrieval all sections were microwave treated for 20 minutes in target retrieval solution (for bFGF; DAKO, Glostrup, Denmark) or 10 mM citrate buffer (for CD34) at 700 W. The sections were incubated overnight with primary mouse anti-bFGF (1:50; Becton Dickinson, Heidelberg, Germany) and anti-CD34 antibody (1:100; Immunotech, Hamburg, Germany). After rinsing in Tris buffer detection was carried out with a biotinylated secondary rabbit anti-mouse antibody diluted 1:50 (DAKO, Glostrup, Denmark) for 30 minutes at room temperature. The sections were washed again as above in Tris buffer and then incubated with streptavidin-biotin-alkaline-phosphatase (10 μl each solution A+B (Sigma, Deisenhofen, Germany) in 10 ml Tris). Immunohistochemical staining was visualized with Fast Red (2 mg Naphtol-As-Mx-phosphate, 10 μl 1 M Levamisol and 10 mg Fast-Red (Sigma) in 0.2 ml N, N-dimethylformamide, (Merck, Darmstadt, Germany) and 9.8 ml 0.1 M Tris-HCl buffer at pH8.6). The sections were rinsed in H 2 0 and counterstained with hemalaun. Negative controls without primary antibody were run for each sample. Tumor response was graded according to the system described by de Wilt et al. [ 15 ]. Progressive disease (PD) was defined as a tumor size > 125 % of the size at the time of treatment, in tumor sizes of 100 ± 25 % no change (NC) was stated. Response was graded as partial (PR) if tumor sizes was between 10 and 75 % and complete (CR) for tumor size < 10 % of its original diameter. In areas of intense neovascularized spots microvessels were counted in a 100× field. A microvessel was defined as a lumen surrounded by a rim of endothelial cells highlighted by immunostaining with anti-CD34 antibodys or anti-bFGF antibodys. Five separate intense neovascularized areas were assessed, and the mean was calculated as microvessel density of each tumor. The score was assessed by two independent observers. Data were analyzed using SPSS/PC+ statistical software. The mean and range of tumor and lesion volumes were calculated for each group. For comparison of tumor volume and microvessel density between the diffent groups a non-parametric test was used (Kruskal-Wallis). A p value < 0.05 was considered to be significant. Results During intervention, the required limb tissue temperature was reached within 8–12 minutes in group IV and 10–15 minutes in group III. Stable temperatures were then maintained for a further 30 minutes with a mean of 41.4°C ± 0.5. To reach limb tissue temperature the perfusion fluid maintained 43°C ± 0.4. Tumor and lesion Volumes Before treatment, the mean (s.e.m) volume of the treated tumors in groups I, II, III, IV and V was 105 mm 3 (4.3), 98 mm 3 (5.7), 108 mm 3 (3.4), 95 mm 3 (5.0) and 107 (4.1) mm 3 respectively, with no significant difference between the groups. Average tumor volume of the controls (I) increased to 1388 mm 3 (101) during six weeks. In the sham operated group (II) tumor volume was significantly larger than in group I (2350 mm 3 (198.6), P = 0.021). Tumor volume in group IV was significantly smaller (1009 mm 3 (122.5)) than in group I and lowest in group V 405 mm 3 (103.6) ( P = 0.036 and P = 0.021, respectively.). There were no significant differences in size between group I and group III (1135 mm 3 (99)) after six weeks ( P > 0.05)(Fig. 2 ). Body weight After the perfusion, body weight decreased in all groups during the first 5 days with a maximum of 8 %. In the following weeks, body weight mounted up to 130 % of the preoperative value. In the last week of the experiment, the weight almost reached a plateau. No significant differences between the five experimental groups were found. Histology In the H&E staining the untreated melanoma (group I) presented as a subcutanous unilocular nodule of moderate differentiation. The tumor showed mainly tubular structures with vacuole like lumen formation and was infiltrated with fibrous septa. The spontaneous rate of necrosis was estimated to be 20 %. Numerous mitoses were detected. Tumor morphology did not change with ageing or tumor size. Histology was similar in group II. In group V the tumors showed signs of irreversible cell damage after treatment. Tumor cells displayed clear shrinkage and partial loss of cell contact. Thromboses of the larger adjacent vessels were found on the tumor-skin border. Macrophage infiltration was present in all groups, but was more numerous pronounced in group V. In group III and IV there were less signs of irreversible cell damage after treatment than in group V. Immunhistochemical analysis of the histological sections for bFGF demonstrated that microvessel density of group I (Fig 3 ) ranged from 18 to 23 with a mean value of 20. Angioneogenesis was more pronounced in group II (Fig 4 ) than in group I (p = 0.003) and less pronounced in group IV and V (Fig 5 ) compared with group I (p = 0.023, p = 0.001). There were no significant differences in the expression of angiogenetic markers between group I and group III (p >0.05). Immunhistochemical analysis of the histological sections for CD34 showed that microvessel density of group I ranged from 17 to 23 with a mean value of 19. Angioneogenesis was more pronounced in group II than in group I (p = 0,001) and less pronounced in group IV and V compared with group I (p = 0.023, p = 0.017). There were no significant differences in the expression of angiogenetic markers between group I and group III (p > 0.05). Strong expression of bFGF and CD34 was found in the cytoplasm of intimal endothelial and medial smooth muscle cells. Expression of bFGF and CD34 was stronger at the invasion front than in the centre of tumors. Summarized data of the expression intensity of bFGF and CD34 in the tumors are given in table 1 . Tumor remission In group I and II tumor progression was observed in all animals macro- and microscopically. In group III and IV 5 animals showed tumor progression and one had a partial tumor response. In group V, 2 animals showed tumor progression, 2 had a complete remission. One had a partial tumor response and one had no change in tumor response (table 2 ). Discussion A phenomenon that has been described by numerous authors is the rapid increase of tumor growth following surgical manipulation [ 16 , 17 ]. This was also observed in our sham-operated group of animals (group II) in which tumors grew more rapidly than in the control group (group I). This implies that resection is associated with an accelerated growth in residual tumors. It is assumed that the surgical manipulation induces liberation of growth factors that in addition to their effect on healing also have a stimulating effect on tumor proliferation. This hypothesis is supported by the results of our immunohistochemical investigations, which revealed an enhanced expression of fibroblast growth factor (bFGF) in group II. In an intraperitoneal tumor model, EGGERMONT et al. showed that the surgical intervention of laparotomy induces an increase of carcinomatosis in the entire peritoneal cavity. At the same time, the immunotherapeutic effect of interleukin 2 and natural killer cells is significantly reduced. It must therefore be speculated that surgical trauma induces a temporary immunosuppressive effect that influences tumor growth [ 16 , 18 , 19 ]. In our experiments, group V (hyperthermic perfusion) showed a significant slowing of tumor growth in comparison with the control group. This was also confirmed by the immunohistochemical investigations that showed a reduction in the expression of relevant markers of the vascular endothelium. Indeed, two of the six animals occurred a complete remission of their tumor. The observation that two animals in this group experienced tumor progression can probably be explained by the duration of the intervention, since factors affecting the kinetics of cell death during hyperthermia are not only the maximum temperature induced, but also the duration of the elevated temperature. In an overview report BHUYAN noted that, as a function of various tumor cell types, a temperature of 43°C must be applied between 30 and 150 minutes to irreversibly damage the tumor cells. At a temperature of 45°C, the application time required varied between 13 and 85 minutes [ 20 ]. In our experiments, the application time was only 30 minutes at a temperature of 41.5°C. Nevertheless, the relatively short application time also had an appreciable damaging effect on the metastatic vessels. Although the temperature in group III (external hyperthermia) was also 41.5°C applied for 30 minutes, no significant changes in the growth pattern of the tumor could be observed. Furthermore, the histological and immunohistochemical evaluations showed a mild hyperthermic damage at best. A comparison of treatment groups III and V revealed no differences in terms of duration of treatment and induced tissue temperatures. In contrast to group III, however, hyperthermia in group V has been mediated directly by the vascular system. In normal tissues subjected to warming, blood perfusion increases reactively via dilatation of the vessels together with an increase in vessel wall permeability. A decisive factor is that the tumor cells themselves, as well as the endothelial cells of tumor vessels are more thermosensitive than normal tissue [ 16 , 21 , 22 ]. Other authors have also shown that hyperthermia is capable of destroying vascular architecture, so that perfusion of the tumor is reduced. The sensitivity of various tumors, however, varies considerably [ 14 ]. ILP and other vasculotoxic effects like TNFalpha in combination with melphalan or doxorubicin increase the uptake of these drugs of three to six times [ 23 ]. Heat transfer between the vascular system and the tissue is determined in particular by the nature of the blood vessels. Important factors are the number, length, and diameter of the vessels, and also the velocity of blood flow [ 24 ]. CREZEE described the relationship between tumor perfusion and heating of the tumor tissue under systemic hyperthermia. He could show that the induction of heat in the tumor depends on the density of the blood vessels within the tumor itself. The more vessels in the tumor exist, the easier it is to heat the lesion via the vascular system. Heattransfer is particularly pronounced with large-calibre blood vessels or high flow velocity, and its rate is all the greater the higher the temperature gradient between the tumor tissue and the blood [ 25 , 26 ]. On the other hand, tumor vasculature carries away the externally applied heat (group III) which is then no longer available to destroy the tumor cells. In this case, the temperature gradient decreases from the tumor to the vascular system. Furthermore, in the presence of hyperthermia, microcirculation in the tumor tissue is more affected than in normal tissue [ 27 ]. In addition, manipulation to the vascular system, and thus oxygenation of the tumor, also appears to play an important role. Although the body's own oxygenated blood is added to the perfusate, the dilution effect together with the clamping time, leads to a relative reduction in perfusion. It is known that hypoxia, in conjunction with a decrease in the pH and energy status of the cell, enhances thermosensitivity [ 28 - 30 ], and this increase correlates with the size of the tumor [ 31 ]. Our results have also confirmed that surgical manipulation of the vascular system and perfusion have an influence on the tumor. Overall, the effect of the intravascular application of hyperthermia appears to be mediated by a combination of hypoxia and heat. Conclusions Our results suggest that even a surgical manipulation such as a skin incision promotes tumor growth, probably by induction of growth factors like bFGF. External hyperthermia of 41.5°C tissue temperature for 30 minutes has no impact on tumor growth and angioneogenesis in vivo. Hyperthermic isolated limb perfusion effectively suppresses tumor angioneogenesis respectively tumor growth. Competing interests None declared. Authors' contributions JP carried out the treatment and drafted the manuscript. MM carried out the treatment. CS carried out the immunohistochemical studies. JG participated in the design of the study. AD carried out the histological studies. WH participiated in the design of the study. TM participiated in the design and coordination of the study. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516035.xml |
518959 | Artificial neural network approach for selection of susceptible single nucleotide polymorphisms and construction of prediction model on childhood allergic asthma | Background Screening of various gene markers such as single nucleotide polymorphism (SNP) and correlation between these markers and development of multifactorial disease have previously been studied. Here, we propose a susceptible marker-selectable artificial neural network (ANN) for predicting development of allergic disease. Results To predict development of childhood allergic asthma (CAA) and select susceptible SNPs, we used an ANN with a parameter decreasing method (PDM) to analyze 25 SNPs of 17 genes in 344 Japanese people, and select 10 susceptible SNPs of CAA. The accuracy of the ANN model with 10 SNPs was 97.7% for learning data and 74.4% for evaluation data. Important combinations were determined by effective combination value (ECV) defined in the present paper. Effective 2-SNP or 3-SNP combinations were found to be concentrated among the 10 selected SNPs. Conclusion ANN can reliably select SNP combinations that are associated with CAA. Thus, the ANN can be used to characterize development of complex diseases caused by multiple factors. This is the first report of automatic selection of SNPs related to development of multifactorial disease from SNP data of more than 300 patients. | Background In recent years, the number of patients suffering from allergic asthma has increased [ 1 ], and allergic diseases including asthma have become a social problem affecting medical costs and quality of life. Allergic asthma is a complex disorder characterized by airway inflammation, bronchial hyperresponsiveness and reversible airway obstruction. Elevated numbers of activated Th2 cells, mast cells and eosinophils in the bronchial mucosa cause certain features of asthma, including increased serum IgE levels in allergic asthma. The available data suggest that there are many potential susceptible genes for allergic asthma, including genes for cytokines, receptors, transcription factors, immune recognition and regulation of lipid mediator generation. A few susceptible genes for allergic asthma have been identified that may be associated with the asthmatic phenotype [ 2 - 4 ], but definite susceptible genes have not been identified yet. Thus, large-scale analysis of gene markers is needed, along with identification of association between these genetic polymorphisms and the asthmatic phenotype, and its development mechanism. It has been reported that the human genome has 3 to 10 million single nucleotide polymorphisms (SNPs). A SNP in a coding region can cause amino acid substitution, resulting in functional modification of the protein; a SNP in a promoter region can affect transcriptional regulation; and a SNP in an intron region can affect splicing and expression of the gene. Thus, SNPs can be highly informative for identifying genetic factors of multifactorial disease such as allergic asthma. In the present study, we analyzed associations between SNPs and childhood allergic asthma (CAA), which is more strongly influenced by genetic factors than other types of allergic asthma. We performed this analysis using an artificial neural network (ANN), which is a computer-based algorithm that can be trained to recognize and categorize complex patterns [ 5 - 8 ]. ANNs have been used for discrimination between subtly different clinical disease lesions; e.g., premalignant lesion Barrett's versus esophageal cancer, based on microarray data [ 6 ]. In a previous study, we performed severity assessment of senile dementia of Alzheimer type using ANN modeling of electroencephalogram data. The average error of the ANN model for assessment scale (HDS-R) score was 2.64 points out of 30 [ 7 ]. We have also used an ANN for prediction of 4 allergic diseases using SNP data [ 8 ]; 82 subjects with data for 6 SNPs were analyzed, and the ANN model predicted diagnosis with accuracy of more than 78%. Thus, we have achieved sufficiently high accuracy with ANNs using relatively little SNP data. Here, we propose an ANN model (its structure is shown in Figure 1 ) suitable to diagnostic prediction of 172 subjects with CAA and 172 healthy subjects, using 25 SNPs in 17 genes shown in Table 1 . For comparison with ANN, we also used logistic regression (LR) analysis, which is currently used to analyze medical statistics and equivalent to ANN with a single hidden node [ 9 ]. In order to selectively identify susceptible SNPs, a susceptible marker-selectable ANN is proposed, in which a parameter decreasing method (PDM) is incorporated. Information on obtaining the execute code, example data and documentation of this software is available at . Associations between combinations of important SNPs and CAA pathogenesis were investigated. A χ 2 test was performed for all 2-SNP and 3-SNP combinations. Results SNPs selected for diagnostic prediction with ANN Several reports have suggested linkage between asthma and chromosomes. For example, genes in the 5q31-5q33 region code for Th2-type cytokines (IL-4, IL-13, which regulate B cell heavy-chain class-switching to IgE production) [ 10 ] and ADRB2 (which mediates airway smooth muscle relaxation and protects against bronchial hyperreactivity) [ 11 , 12 ]. IL-4 operates via the IL-4 receptor (IL-4R), which is encoded by a gene in chromosomal region 16q12. Mice deficient in the IL-4Rα chain lack IgE production and Th2 inflammatory reactions, and it has been shown that total IgE level is dependent on Ile50Val substitution [ 13 ]. In the present study, we analyzed 25 SNPs (Table 1 ) in 17 genes known to be associated with development of asthma. Association between these SNPs and CAA was assessed by P -value. As shown in Table 1 , 21 of these 25 SNPs had a P -value greater than 0.1. The P -values of CysLT2 (108 C/A), IL-4Rα (148 G/A), ADRB2 (265 A/G) and C5 (4266 G/A) were 0.0036, 0.0155, 0.0541 and 0.0581, respectively. When CysLT2 (108 C/A), which had the lowest P -value of 25 SNPs, was used for discrimination between case and control as a sole factor, prediction accuracy was 54.4%, and the sensitivity and specificity was 12.8% and 95.9%, respectively, compared with the number of case and control subjects to assess discrimination performance (genotype CC; case 150, control 165, genotype CA or AA; case 22, control 7). Thus, we constructed a susceptible marker-selectable ANN model, which can discriminate between cases and controls using the selected susceptible SNPs, and which can include the association between combinations of SNPs and development of CAA. Diagnostic prediction using 25 SNPs We used a three-layered ANN with input, hidden and output layers (Figure 1 ). An ANN model and LR model, with 25 SNPs as input variables, were constructed with learning data, and we performed diagnostic prediction with evaluation data. The results of diagnostic prediction are shown in Figure 2a,2b and Table 2a . The ANN had higher prediction accuracy than LR. Accordingly, sensitivity and specificity, with both evaluation data and learning data, were higher for the ANN than for LR (Table 2a ). In LR analysis, Monte Carlo study was performed to evaluate the effect of number of events per variable (EPV) [ 14 ]. It suggested that at least 10 events per variable analyzed were desirable to maintain the validity of the model. In the present study, we used 172 events per group and 51 variables (25 SNPs and 1 teacher value). LR would not have an enough power for parameter selection, because 172 events per group is small compared with that of variable. The construction of optimized LR model should be furthermore investigated. Selection of susceptible SNPs for CAA Ritchie et al . [ 5 ] reported the optimization of the architecture using genetic programming neural networks (GPNN) [ 5 ]. If important SNPs were previously determined, optimization of network architecture should be carried out. Genetic programming neural networks have contributed the construction of ANN model with high performance. In the present study, however, many candidate SNPs were used and the selection of SNPs is firstly desired. Therefore, in order to extract SNPs closely associated with CAA, we tried optimization of input variables by PDM in the ANN model, while the architecture of a neural network was not modified. Five PDM trials were performed. Figure 3 shows typical results for change of accuracy during PDM procedure. When input variables were excluded one by one to preserve prediction accuracy (as described in Methods), the accuracy began to decrease after the number of SNPs used for modeling reached 10. When the number of SNPs used for modeling decreased, coincidence of genotyping pattern between cases and controls inevitably occurred. When genotyping pattern of a case was coincident with that of controls, the learning for model construction did not progress well. We investigated the rate of case subjects whose genotype patterns were coincident with that of control subjects at each step of PDM (Figure 3 ). Rate of case subjects [%] in Figure 3 means N' case / N case . In this case, N' case is the number of cases whose genotype pattern is match to control's genotype pattern at least one control ( N case = 172 subjects). As shown in Figure 3 , there was little coincidence of genotype patterns when more than approximately 7 SNPs were used in ANN modeling. Therefore, the decrease in accuracy was considered to be due to omission of a highly important SNP. The remaining 10 SNPs were worth investigating as important factors. To investigate the important SNPs, we counted the number of SNPs that remained within the last 10 input variables in 5 trials. The significance order of remaining SNPs was listed, and a score of order ranging from 1 to 10 points was determined, based on the significance order. The remaining SNPs were reordered according to sums of scores, as shown in Table 3 . We believe that SNPs with higher scores are more important for development of CAA, because significance of SNPs correlated with the order of elimination via the PDM procedure described in methods section. ANN models were reconstructed using SNPs listed in Table 3 . The number of input SNPs varied from three ( IL-4Rα (148 G/A), CysLT2 (2534 A/G) and IL-10 (-571 C/A)) to 17 SNPs (all listed SNPs) according to the order of Table 3 . When more than 10 SNPs were used as input variables, average accuracy for learning and evaluation data was high (Figure 4 ), and was almost equal to that of the model using 25 SNPs. These results suggest that the 10 SNPs selected in Table 3 are very important for prediction of development of CAA. The results of diagnostic prediction using the 10 SNPs selected by PDM are shown in Figure 2c , and the accuracy, sensitivity and specificity are shown in Table 2b . In the ANN model, the accuracy, sensitivity and specificity with evaluation data were again sufficiently high, and were somewhat similar to the results from the analysis using 25 SNPs, although the number of input variables was markedly smaller than in the analysis using 25 SNPs. In particular, sensitivity was significantly high (77.9%), indicating that case subjects were more correctly diagnosed by this model. To compare with the LR model, LR model consisting of 10 SNPs selected by ANN was constructed (Figure 2d ). As shown in Table 2b , the LR model constructed showed low accuracy. This result indicates high performance of ANN modeling for CAA prediction although selected SNPs would not be suitable for LR analysis. We concluded that the ANN model constructed with 10 SNPs could discriminate between cases and controls as precisely as the model constructed with 25 SNPs. Interaction between SNP and another SNP for CAA To understand the importance of the 10 SNPs selected, we analyzed combinations of these 10 SNPs. We paid particular attention to SNP combinations associated with CAA, and assessed whether any combinations consisting of SNPs selected by ANN were associated with CAA. The relationships between 2-SNP or 3-SNP combinations and CAA development were examined by calculating P -value using the χ 2 test. In models using 10 SNPs selected by PDM or the other 15 SNPs, the total number of 2-SNP combinations and 3-SNP combinations ( N comb ) is 90 ( 10 P 2 ) or 210 ( 15 P 2 ), and 360 ( 10 P 3 /2) or 1365 ( 15 P 3 /2), respectively. With respect to 2-SNP combination between the SNP of interest and SNP A, P -value was calculated as follows. When patients were limited with certain pattern of another SNP, such as AA major homozygote of SNP A, patient distribution of the SNP of interest was investigated. With respect to 3-SNP combination, between the SNP of interest, and SNP A and B, P -value was calculated as follows. When patients were limited with certain pattern of two other SNPs, such as AA major homozygote of SNP A and BB major homozygote of SNP B, patient distribution of the SNP of interest was investigated. To evaluate P -value of the combination, the usual Bonferroni correction of P -values was first investigated. To select the 2-SNP combination accompanied with minimum false positive, the criterion was P < 0.05/300. Here 300 cases correspond to 25 C 2 . Under this severe condition, there were no significant SNPs. As the same as 2-SNP combination, any significant combination was not obtained on 3-SNP combination under the threshold of P < 0.05/2300. Next, to determine important combination, P -value without Bonferroni correction was used, that is P < 0.05. Results are shown in Table 4 . In 2-SNP combination, there were 13 combinations with P < 0.05 among total 90 combinations. In the case of 3-SNP, 72 combinations with P < 0.05 were existed in 360 exhaustive combinations. However, combinations possibly include several false positive significant combinations. Therefore, we paid attention to the SNP, of which P -value effectively decreases by combining with genotype or allele of other SNPs. We defined effective combination value (ECV). ECV2 or ECV3 is the ratio of 2 or 3-SNPs P -value to the product of each P -value. ECV is not indicator for avoiding false positives but for evaluation of interaction. For example, in 2 SNP combinations, when patients were limited with certain pattern of another SNP, such as AA major homozygote of SNP A, patient distribution of the SNP X of interest is investigated ( P = P ax ). If the 2-SNP combination is independent (no interaction) each other, P ax equals multiplication of P a and P x . ECV<1 means that the 2-SNP combination is not independent and two SNPs have any interaction each other. The effect of ECV on number of effective combinations is shown in Table 4 . About half number of 2-SNP combination satisfied the condition ECV2<1 ( N ECV 2<1 = 47). Among 13 combinations with P < 0.05 mentioned above, 11 combination also satisfied the same condition ( N ECV 2<1, P = 11). When ECV2<0.5, N ECV 2<0.5 decreased the number to 27 and 10 of N ECV 2<1, P = 11 still remained. When ECV2<0.1, N ECV 2<0.1 became small and it was thought that positive combination may be lost. In the case of 3-SNP combination, only 20% of the total combination satisfied the condition P < 0.05. The combinations of 12% among the total combinations (43 combinations) satisfied ECV3<1. All of these 43 combinations also satisfied the condition P < 0.05. From these results, it was concluded that P < 0.05 is not strict criterion for 3-SNP combination analysis. In the case of 2-SNP combination, ECV2<0.5 was adequate as a selection of effective combination, because 77% of the combination with P < 0.05 still remained. From these consideration, we selected these two evaluation bases ( P -value and ECV) in order to determine effective combinations and the combination with N ECV <0.5, P ( P < 0.05 and ECV<0.5) was picked up. The combinations were used for the following investigation. The number of combination which satisfies the condition, P < 0.05 and ECV2<0.5 in 2-SNP combination and P < 0.05 and ECV3<0.5 in 3-SNP combination was designated as N ef , the number of effective combination, respectively (Table 5 ). It is very important to clearly determine whether effective combinations frequently occur among groups of 10 SNPs selected by ANN modeling with PDM. It would be difficult to investigate the phenotypes associated with each of such a large number of combinations. Identifying effective 2-SNP combinations using the conditions described above is a useful method of identifying 2-SNP combinations that merit further investigation. Ten effective combinations were found among the 10 SNPs selected by ANN; 23 effective combinations were found between those 10 SNPs and the remaining 15 SNPs; and 3 effective combinations were found among the remaining 15 SNPs. It is likely that the former 10 combinations are more important than the latter 26 combinations, because the ANN model constructed using only the selected 10 SNPs exhibited sufficiently high accuracy to predict development of CAA. Susceptible genes for development of a multifactorial disease like CAA can correctly classify many subjects as cases or controls, and it is very important that those genes involve SNP combinations that have important interaction with high concentration ratio. We defined the concentration ratio as the ratio of effective rate to random selection rate . When the effective rate, N ef /Σ N ef , was calculated, it was found to be 0.28 (10/36) for the 10 SNPs selected by PDM, 0.64 (23/36) for combinations between the 10 selected SNPs and the remaining 15 SNPs, and 0.08 (3/36) for combinations of the remaining 15 SNPs (Table 5 ). The random selection rate, N com / 15 P 2 shown in Table 5 , represents the rate which the combination is selected from all 2-SNP combinations independently, 0.15 (90/ 15 P 2 ) for the 10 SNPs selected by PDM, 0.5 (300/ 15 P 2 ) for combinations between the 10 selected SNPs and the remaining 15 SNPs, and 0.35 (210/ 15 P 2 ) for combinations of the remaining 15 SNPs (Table 5 ). The concentration ratio was found to be 1.85 for the 10 SNPs selected by PDM, 1.28 for combinations between the 10 selected SNPs and the remaining 15 SNPs, and 0.24 for combinations of the remaining 15 SNPs (Table 5 ). The concentration ratio was higher for combinations among the 10 selected SNPs than for other combinations, so we can select 2-SNP combinations associated with CAA with high rate. The results are shown in Table 6 . In the next step, 3-SNP combinations were analyzed. The effective rate, the random selection rate, and the concentration ratio were calculated as well as the case of 2-SNP combination (Table 5 ). It was found to be 2.28 for each of the 10 selected SNPs alone (3:0 in Table 5 ), 1.33 for 2 of the 10 selected SNPs and 1 of the remaining 15 SNPs (2:1 in Table 5 ), 0.67 for 1 of the 10 selected SNPs and 2 of the remaining 15 SNPs (1:2 in Table 5 ), and 0.94 for the remaining 15 SNPs alone (0:3 in Table 5 ). The concentration ratio was higher for combinations among the 10 selected SNPs than for other combinations, so we can select 3-SNP combinations associated with CAA with high rate. The combination with the lowest ECV3 consisted of the genes IL-4Rα , and C3 (0.03526). This is about 3% of the value multiplied each P -value of 2-SNP combination (0.5060). For patients with genotype GA of IL-4Rα (148 G/A: Val50Ile) and genotype CT of C3 (4896 C/T), patient frequency against genotype of C3 (1692 G/A) had a P -value of 0.01784. For C3 (1692 G/A) alone, a P -value of 0.6993 was obtained, which was 40 times greater than the P -value of the 3-SNP combination. Thus the rate of correct identification of effective combinations evaluated by adjusted P -value and ECV selected based on PDM trials was higher than the corresponding randomized rate, implying that the ANN can reliably select SNP combinations that are associated with CAA. The 2-SNP combinations with the conditions described above among selected 10 SNPs are shown in Table 6 . For example, in Table 6 , for combinations between CysLT2 (2534 A/G) and IL-4Rα (148 G/A: Val50Ile), among subjects with a CysLT2 (2534 A/G) genotype of AG or GG (CAA, 107 subjects; healthy controls, 103 subjects), there was an important correlation with IL-4Rα (148 G/A: Val50Ile) genotype of GG, GA, AA ( P = 0.00030). We examined the distributions of important combinations among subjects. A total of 52 CAA subjects and 24 healthy controls had genotype AG or GG at CysLT2 and genotype GG at IL-4R α (148 G/A) (Figure 5a ). The present findings also indicate that the 3-SNP combination consisting of IL-10 (-571 C/A), IL-4 (-590 C/T) and C3 (1692 G/A) is a susceptible factor of CAA ( P = 0.00426). No association with CAA was found for any of these 3 SNPs alone ( P = 0.1074, 0.9085, 0.6993, respectively; Table 1 ) or for any 2-SNP combinations of them ( P = 0.1851, and 0.3002, respectively). Subjects with genotype CA at IL-10 (-571 C/A), genotype CT at IL-4 (-590 C/T) (CAA, 34 subjects; healthy controls, 38 subjects) and genotype GG at C3 (1692 G/A) (CAA, 12 subjects; healthy controls, 6 subjects) were estimated to be at high risk for pathogenesis of CAA. Furthermore, among the subjects with the same genotype pattern, the number of subjects with genotype AA at C3 (1692 G/A) were CAA, 3 and healthy controls, 13, respectively (Figure 5b ). Other remarkable combinations shown in Table 6 were also found among the 10 selected SNPs. For example, the number of cases with GG genotype at IL-4Rα ( 148G/A) and TT genotype at C3 (4896C/T) was 4 times the number of controls with that genotype combination (CAA, 20 subjects; healthy controls, 5 subjects) ( P = 0.00271). There are no previous reports of association between these genotype combinations and CAA. The combination of IL-4Rα (148 G/A: Val50Ile) and IL-4 (-590 C/T) was also associated with CAA ( P = 0.00689); association between allergic asthma and this combination has previously been reported [ 15 , 16 ]. Discussion To characterize the development mechanism, we investigated several relationships between SNPs and development of CAA, referring to previous papers, as described below. IL-4 is produced by Th2 cells, and exerts its activity by interacting with the receptor IL-4Rα, located on the surface of B cells. It has been reported that the V50 (148G)/R551(1827G) combination of IL-4Rα polymorphisms may be associated with enhancement of IL-4Rα function [ 16 ]. As concerns the polymorphisms on IL-4 , it was reported that the -590T allele increases the strength of the IL-4 promoter compared with the -590C allele [ 15 ]. C3 is a proinflammatory mediator that binds to specific cell surface receptors and causes leukocyte activation, smooth muscle contraction and vascular permeability [ 17 ]. C3 -deficient mice challenged with allergen show diminished airway hyperresponsiveness and lung eosinophilia, with dramatic reduction of the number of IL-4-producing cells and attenuation of IgE responses [ 18 ]. In the present study, we found that interaction between genotype TT at C3 (4896 C/T) and genotype GG at IL-4Rα (148 G/A) may be associated with CAA, but details of interaction between these polymorphisms combinations and development mechanisms have not been clarified. The present findings indicate that, among subjects with an IL-10 (-571 C/A) genotype of CA and an IL-4 (-590 C/T) genotype of CT, there is important correlation with a C3 (1692 G/A) genotype of GG or AA (Figure 5b ). CysLTs, which are produced by inflammatory cells including eosinophils, are mediators of leukotrienes, and have been implicated in the pathogenesis of allergic diseases. Recently, it has been reported that CysLTs can act as autocrine or paracrine mediators to stimulate rapid, nonexocytotic release of IL-4 [ 19 ]. These findings are consistent with the present results, in which subjects with CT or TT genotype at IL-4 (-590 C/T), AG or GG genotype at CysLT2 (2534 A/G) and GG genotype at IL-4Rα (148 G/A) were estimated to be at high risk for pathogenesis of CAA ( P = 0.00022). However, 2-SNP interaction between CysLT2 (2534 A/G) and IL-4Rα (148 G/A) ( P = 0.00030) markedly affected the 3-SNP interaction. In the present study, we examined correlation between CAA and 25 SNPs in 17 genes using an ANN model. We think that there are not a few main effects and interactions which can explain development of multifactorial disease like CAA, because it is thought that interactions of genetic risk factors might be different individually among CAA patients in spite of same disease. So it is very important to select multiple genetic factor models associated with multifactorial disease like CAA with high concentration ratio. We found that 10 of these SNPs are important factors in development of CAA. Important combinations among these 10 SNPs were also extracted. As described above, several of these combinations (listed in Table 6 etc.) have been found to be important factors in allergic disease, in previous biological and epidemiological studies. We also found several novel important combinations. The present data about important combinations suggests multiple patterns of CAA development. It should be noted that these findings were obtained automatically using an ANN model constructed without priori knowledge. Using an ANN model with 10 SNPs, we were able to discriminate between cases and controls with more than 70% accuracy. We concluded that the ANN is an effective tool for predicting development of CAA, using SNP data. However, further investigation of other genetic and environmental factors associated with CAA is needed. We previously constructed an advanced modeling method, the fuzzy neural network [ 20 , 21 ], which is an ANN model. When this model is applied to analysis, the susceptibility rules of interaction can be explicitly and linguistically described. Also, it can be used to describe susceptible interaction between genetic factors such as SNPs and environmental factors such as favorite foods and life style. Using the rules obtained with this model, we can plan protocols for preventive treatment of subjects with high-risk genetic profiles. Network analysis tools such as ANNs can be applied to analysis of multifactorial disease using SNP data such as selection of important SNPs or description of interactions between SNPs. Conclusions Relationships between CAA and 25 SNPs in 17 candidate genes were analyzed using an ANN. In diagnostic prediction, ANN discriminated cases from controls more precisely than LR. From among the 25 original SNPs analyzed, we selected 10 SNPs that were closely associated with CAA. Calculating P -value using the χ 2 test, we found that 2-SNP and 3-SNP combinations of these 10 SNPs were associated with CAA. The ANN was able to represent associations between CAA and these 2-SNP or 3-SNP combinations using complicated nonlinear relations. Thus, the ANN can be used to characterize development of complex diseases caused by multiple factors. Methods Subjects and SNP data SNP data were kindly provided by the ethics committees of Tohoku University and RIKEN. We analyzed the SNP data for 25 polymorphisms in the 17 genetic regions listed in Table 1 . Each SNP was detected using the established method based on TaqMan PCR [ 22 ]. The study population comprised 172 subjects with childhood allergic asthma (CAA) who were under 17 years of age and 172 healthy subjects with no signs or symptoms of atopy-related diseases selected from general population, all of whom gave written informed consent for SNP analysis. The subjects were diagnosed by experienced doctors, as "positive" (with allergic asthmatic symptoms) or "negative" (without allergic asthmatic symptoms). In the present paper, the subjects with CAA are referred to as "cases" and the healthy subjects are referred to as "controls". Genotype patterns of the 25 SNPs were compared between cases and controls. None of the cases had genotype patterns coinciding with those of controls. Data preprocessing To use SNP data as input data for the ANN, we converted the genotyping data into 2-numeral data. In ANN modeling, input and output variables are normalized into 0.1–0.9 [ 8 ]. In SNP data, there are 3 genotypes per locus. Therefore, we provided 2 inputs per SNP: (0.1, 0.1) for homozygote of the major allele, (0.1, 0.9) for heterozygote, and (0.9, 0.9) for homozygote of the minor allele. Since from the genetic point of view it may be difficult to estimate that heterozygote affects a disease by half the extent that homozygote affects it, the coding of (0.1), (0.5) and (0.9) was never used. The diagnosis data were also converted into numerical data, referred to hereafter as "teacher" values: 0.9 for "positive (case)", and 0.1 for "negative (control)". For LR, we converted SNP data into numerical input data as follows: (0.1, 0.1) for homozygote of the major allele, (0.1, 0.9) for heterozygote, and (0.9, 0.9) for homozygote of the minor allele. Positive and negative diagnoses were also converted into numerical data: 0.9 and 0.1, respectively. ANN model and model construction For SNP analysis, we used a three-layered ANN with input, hidden and output layers (Figure 1 ). For model construction, the performance index of the ANN was assessed using a method we previously proposed [ 7 , 8 ], with slight modifications. N error (number of missed points) and Er (sum of squared error) were defined and calculated for learning data and evaluation data as follows: N error = N error , l + N error , e (3) where Y and T represent the predicted value and the teacher value, respectively. N l and N e represent the number of subjects as learning and evaluation data, respectively. N error is the number of output data with an error of >0.4 between the predicted value and teacher value as shown above. Er is calculated with the square of error as shown above. For ANN learning, the connection weights were initially randomly set from 0 to 1, and were altered using the back propagation methods [ 23 ] with learning data so as to minimize the value of Er l . Learning rates of 0.1, 0.2, 0.3, 0.4 and 0.5 were examined. The maximum learning time was 2000 iterations. The best ANN model (selected for SNP analysis) was that in which N error reached minimum within the maximum learning time. When minimum N error was equal to that of other models within the maximum learning time, the model with minimum Er value was selected. Prediction accuracy of the constructed model was defined as follows. Threshold was set at 0.5. If the teacher value was 0.9, and the predicted value was greater than 0.5, the prediction was true (true positive; TP); for predicted values lower than 0.5, the prediction was false (false negative; FN). If the teacher value was 0.1, and the predicted value was lower than 0.5, the prediction was true (true negative; TN); for predicted values greater than 0.5, the prediction was false (false positive; FP). We calculated the prediction accuracy ( Ac ) as follows: The sensitivity ( Se ) and specificity ( Sp ) of predicted values were defined as follows: where N TP , N FN , N TN and N FP are the number of TN, FN, TN and FP subjects, respectively. N case and N control are the number of case and control subjects, respectively. Parameter Decreasing Method (PDM) In order to extract SNPs closely associated with CAA, we selected the input variables by parameter decreasing method (PDM) after the ANN model with 25 SNPs was constructed. In PDM, 1 SNP was excluded from input variables in turn, and ANN models were constructed with the remaining 24 SNPs by performing the cross-validation described below. From among the 25 models thus constructed, the model with minimum N error averaged in the cross-validation step was selected. When minimum N error was equal to that of other models within the maximum learning time, the model with minimum Er value was selected as described above. The PDM step was repeated until 1 SNP remained as input variable. The PDM procedure was performed 5 times with unifying learning rates of 0.1 and learning time of 2000, and the rank of importance of selected SNPs was determined as described in Results section. We performed 5 PDM trials so that the effects of randomized initial connection weights might be minimized. In 5 PDM trials, data set for cross-validation mentioned below was reconstructed every time. Cross-validation Cross-validation allows estimation of the prediction error of a model by leaving out a portion of the data as an evaluation data [ 24 ]. In the present study, to investigate the flexibility of the ANN, learning and evaluation were performed using the ANN and 5-fold cross-validation. With 5-fold cross-validation, the data set for the 172 cases and 172 controls was divided into 5 groups with randomizing and alternating the data. In each group, the number of cases was equal to that of controls. Four groups were assigned as learning data, and 1 group was assigned as evaluation data; this learning and evaluation process was repeated 5 times, so that each group was assessed once as evaluation data. Then, the prediction accuracy of evaluation data across all 5 trials was calculated and averaged for the overall prediction accuracy of the ANN model shown in Table 2 . Sensitivity and specificity were also calculated. Logistic Regression (LR) Model An LR model was constructed using SPSS 11.5J statistic software for Windows (SPSS Japan Inc., Tokyo), for comparison with the ANN model. All 25 SNPs were used as input variables of LR. For LR analysis, we used 50 main effects plus an intercept but not any interaction terms. As with the ANN model, the data set was divided into 5 groups and the cross-validation was performed. Prediction accuracy, sensitivity and specificity were calculated. Determination of differences in frequency of alleles and genotypes We also examined association between CAA and combinations of SNPs by calculating P -value using a χ 2 test. The χ 2 test was used to evaluate the differences in frequencies of alleles or genotypes between cases and controls. The P -values shown in Table 1 were calculated using 172 cases and 172 controls. Degree of freedom (D.F.) (shown in Table 1 ) was 2 for 3 types of subjects; e.g., homozygote of the major allele, heterozygote, and homozygote of the minor allele. In the test with one SNP, when the expectancy for subjects homozygous for the minor allele (calculated from the frequency of the genotype) was less than 5 subjects for both case and control, we regarded the homozygote of the minor allele and the heterozygote as identical and defined degree of freedom as 1. In the tests with 2-SNP and 3-SNP combinations, we used the D.F. shown in Table 1 to find the change of differences in frequency under the same condition of SNP alone. If, in more than 5 subjects, all expectancies for subjects satisfied the test conditions, we calculated P -value with χ 2 test. In order to determine important combinations, we use two evaluation bases ( P -value and effective combination value (ECV)) mentioned in Results section. Authors' contributions YT carried out ANN modeling of SNP data including PDM and calculating P -value using a χ 2 test. ST and YH carried out the basic analysis using ANN and data preprocessing. YS and TS participated in providing of SNP data and the design of LR analysis. TK participated in the design of the study. HH conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC518959.xml |
546038 | The Genomes of Oryza sativa: A History of Duplications | We report improved whole-genome shotgun sequences for the genomes of indica and japonica rice, both with multimegabase contiguity, or almost 1,000-fold improvement over the drafts of 2002. Tested against a nonredundant collection of 19,079 full-length cDNAs, 97.7% of the genes are aligned, without fragmentation, to the mapped super-scaffolds of one or the other genome. We introduce a gene identification procedure for plants that does not rely on similarity to known genes to remove erroneous predictions resulting from transposable elements. Using the available EST data to adjust for residual errors in the predictions, the estimated gene count is at least 38,000–40,000. Only 2%–3% of the genes are unique to any one subspecies, comparable to the amount of sequence that might still be missing. Despite this lack of variation in gene content, there is enormous variation in the intergenic regions. At least a quarter of the two sequences could not be aligned, and where they could be aligned, single nucleotide polymorphism (SNP) rates varied from as little as 3.0 SNP/kb in the coding regions to 27.6 SNP/kb in the transposable elements. A more inclusive new approach for analyzing duplication history is introduced here. It reveals an ancient whole-genome duplication, a recent segmental duplication on Chromosomes 11 and 12, and massive ongoing individual gene duplications. We find 18 distinct pairs of duplicated segments that cover 65.7% of the genome; 17 of these pairs date back to a common time before the divergence of the grasses. More important, ongoing individual gene duplications provide a never-ending source of raw material for gene genesis and are major contributors to the differences between members of the grass family. | Introduction The importance of the rice genome is reflected in the fact that rice was the first crop plant to have its genome sequenced; astonishingly, it was sequenced by four independent research teams at Beijing Institute of Genomics [ 1 ], Syngenta [ 2 ], International Rice Genome Sequencing Project (IRGSP) [ 3 , 4 , 5 ], and Monsanto. Beijing analyzed the two parental strains, 93–11 and PA64s, for a popular land race of super-hybrid rice, LYP9, and released a 4.2x draft for 93–11, a cultivar of the indica subspecies. This draft was acquired by a whole-genome shotgun (WGS) method [ 6 ]. Syngenta and IRGSP worked on Nipponbare, a cultivar of the japonica subspecies. Syngenta also used a WGS method and published a 6x draft. IRGSP used the clone-by-clone method [ 7 ] and released a 10x draft that incorporates the Syngenta data. Their publications include the finished version of Chromosomes 1, 4, and 10. These efforts have been widely hailed not only because rice feeds much of the world's population but also because rice is expected, through comparative analyses, to play a major role in understanding the grass family of crop plants [ 8 , 9 , 10 , 11 , 12 , 13 ]. We will report on an improved version of Beijing indica, which brings the coverage of the 93–11 dataset up to 6.28x. In addition, we improved Syngenta japonica by reassembling their sequence from the raw traces (National Center for Biotechnology Information Trace Archive; http://www.ncbi.nlm.nih.gov/Traces/trace.cgi? ) and combining that information with our 93–11 assembly. We achieved almost three orders of magnitude of improvement in long-range contiguity, and put essentially all the genes on the map, by combining the two WGS assemblies in a manner that preserves the single nucleotide polymorphism (SNP) information for indica – japonica differences. Both of these WGS assemblies were constructed independent of the information in IRGSP japonica . Hence, the two japonica assemblies allow us to compare the WGS and clone-by-clone methods objectively. By taking the clone-by-clone assembly as a “gold standard,” we can estimate the intrinsic misassembly rates for our two WGS assemblies—not just the japonica WGS but also the indica WGS, as identical assembly procedures are used and both contain 6x coverage. If we compare IRGSP japonica to Beijing indica, any increases in the discrepancy rate beyond this intrinsic misassembly rate can be attributed to indica – japonica differences. In the same spirit, genes are identified for all three assemblies using the same annotation procedures, to assess gene content differences without the methodological inconsistencies that have plagued previous comparisons. Finally, we introduce a simple method for analyzing gene duplications that resolves the contradictory claims that rice is an ancient aneuploid [ 14 ] and an ancient polyploid [ 15 ]. In the process, we demonstrate that duplication of individual genes plays a major role in the continuing evolution of the grass genomes. Both WGS sequences, and details of our analyses, are available from our own Web site (Beijing Genomics Institute Rice Information System; http://rise.genomics.org.cn ) [ 16 ]. The version of IRGSP japonica that we use was downloaded October 5, 2003, from GenBank and DNA Data Bank of Japan according to the guidelines at http://www.genome.arizona.edu/shotgun/rice/status and the physical map at http://rgp.dna.affrc.go.jp/IRGSP/download . Results WGS Assembly of indica and japonica Many legitimate concerns have been raised about the differing qualities of the rice sequences that have been published [ 17 , 18 ] and on the idea that they must be “finished” [ 19 , 20 ]. Higher quality is of course a good thing, but it does come at a cost, and lost in the discussion is the reality that cost–benefit factors have always been important in sequencing. Most notably, all genome projects to date have focused primarily on the euchromatic regions that can be cloned and sequenced, even though important genes are missed as a result. For example, an essential 5.1-Mb fertility gene [ 21 ] resides in the heterochromatic Y chromosome of the Drosophila genome. In plant genomes, costs are primarily driven by the intergenic retrotransposon clusters [ 22 ] that account for about half of the rice genome, and even more of the larger maize (6x) and wheat (38x) genomes. Hence, our objective is merely to have all the genes assembled in one piece, without fragmentation, and anchored to the maps. A similar objective has been proposed [ 23 , 24 ] for crop genomes in general. Our benchmark is the set of full-length japonica cDNAs from the Knowledge-Based Oryza Molecular-Biological Encyclopedia [ 25 ] that contains 19,079 nonredundant cDNAs (nr-KOME). We begin with a few definitions. At the end of any WGS, a substantial fraction of the reads (specifically, those whose sequences are highly repeated across the genome) are invariably left unassembled. The usable reads are assembled into contigs, scaffolds, and super-scaffolds. In a contig, the identity of every base is defined. In contrast, scaffolds and super-scaffolds have gaps (regions of known length but otherwise unknown base content). The difference is that one refers to the sequence before any linking information from indica and japonica sources are combined (scaffold) and the other refers to the sequence after they are combined (super-scaffold). All of the raw data that went into these WGS assemblies are listed in Table S1 , and the assembly procedure itself is outlined in Figure 1 . Figure 1 Basic Algorithm for Construction of Scaffolds and Super-Scaffolds We start with the smallest plasmids and progressively work our way up to the largest BACs. Only links with two or more pieces of supporting evidence are made. These include 34,190 “anchor points” constructed from a comparison of indica and japonica . Each anchor is a series of high-quality BlastN hits (typically 98.5% identity) put together by a dynamic programming algorithm that allows for small gaps to accommodate the polymorphic intergenic repeats. Typical anchor points contain four BlastN hits at a total size of 9 kb (including gaps). Notice how in the beginning indica and japonica are processed separately, to construct what we called scaffolds. Only at the end do we use data from one subspecies to link scaffolds in the other subspecies, and these are what we called super-scaffolds. Compared with our previous 4.2x assembly of indica, more shotgun reads and a few directed finishing reads were added to increase the coverage to 6.28x. We did not use the older assembly at all. Instead, we went back to the raw reads and reassembled them, with an updated version of RePS [ 26 , 27 ] that incorporates some recent concepts from Phusion [ 28 ]. Increasing coverage is essential for reducing single-base error rates. Based on the estimates from RePS, 97.2% and 94.6% of our new assembly has an error rate of better than 10 −3 and 10 −4 , respectively. For the older assembly, the percentages were only 90.8% and 83.5%, respectively. Equally important, and as expected from Poisson sampling statistics [ 29 ], increasing coverage improves the scaffold size to a point where, even without additional finishing effort, most of the nr-KOME cDNAs can be aligned in one piece, without fragmentation. All we had to do was find a way to link these scaffolds together to create larger super-scaffolds, which could then be anchored to the physical [ 30 ] and genetic [ 31 ] maps. Mapped super-scaffolds for Beijing indica have a N50 size (the size above which half of the total length of a sequence dataset is found) of 8.3 Mb, which is a thousand times better than our previous draft, as shown in Table 1 . We used an unorthodox method to construct super-scaffolds of megabase size from initial scaffolds of 30-kb size. Most of the increase in long-range contiguity came from combining the two WGS assemblies, not from the bacterial artificial chromosome (BAC) end pairs, which were of limited utility because their insert sizes were too large. Notice that in combining indica and japonica data, we use the alternate subspecies only for order and orientation information, not to fill missing bases. In other words, every base in the indica assembly is from indica . Not one single base is from japonica . Another key point is that Syngenta japonica is our reassembly of their raw data, not the published assembly. By using RePS for both WGS assemblies, we obtain error estimates for every base, which will later be essential for use in polymorphism detection. We would concede that if genes are ordered differently in indica and japonica, there is a small probability that by forcing the two subspecies together, we lose this information. However, there is no evidence of a major reordering of the genes because, if there were, it would have been seen in all these years of genetic mapping. The benefits thus outweigh the risks. Table 1 Summary of Assembled Contigs, Scaffolds, and Super-Scaffolds Each piece can be further subdivided on the basis of whether or not it is mapped and, if not, on the basis of its size. N50 refers to the size above which half of the total length of the sequence set can be found. An equivalent size for the unassembled reads is computed by dividing the number of high-quality Q20 bases (estimated single-base error rate of 10 −2 ) by the effective shotgun coverage The total genome size, including the unassembled reads and the unmapped pieces of all sizes, is 466.3 Mb for Beijing indica and 433.2 Mb for Syngenta japonica . For this estimate, we added up all the pieces at the scaffold level (as opposed to the super-scaffold level, where the gap size estimates are taken from the alternate subspecies and may not be representative of the underlying genome). We believe this difference is real, because the two genome sizes are based on the same procedures and similar WGS datasets. Although many smaller pieces fall between the cracks in the maps, these unmapped pieces turn out to be extremely gene poor. Hence, in our submission to DNA Data Bank of Japan/European Molecular Biology Library/GenBank, we omit unassembled reads and unmapped pieces smaller than 2 kb, which has the advantage of also filtering out nonrice contaminants from inevitable mix-ups in the lab. Physical distance is defined along a pseudo-chromosome where gaps of estimated size larger than 200 kb (a typical BAC) are collapsed to 200 kb. Between adjacent super-scaffolds, where by definition we do not have an estimated gap size, we insert a 5-kb gap. To validate the long-range accuracy of our assemblies, we compared physical and genetic distances, as shown in Figures S1 and S2 . We use only those 1,519 markers that can be found in all three rice assemblies by Blastn at E-values of 10 −100 . There are two classes of discrepancies. First, the marker is on different chromosomes. All three rice assemblies agree with each other but not with the genetic map in 135 of 152 such markers. In the second class, the disagreement is on positions within a chromosome, and all three rice assemblies agree with each other but not with the genetic map in 41 of 60 such markers. Only a small handful of discrepancies are unique to any one assembly. It is highly unlikely that all three rice assemblies will make the same mistake, so we conclude that on the scale of hundreds of kilobases, our WGS data are better than the genetic map. Computed over every five markers, the mean (median) recombination rate is 4.5 (4.2) cM/Mb. We do expect smaller-scale misassemblies in the WGS data, as, for example, in Beijing indica, 98.1%, 71.0%, and 39.3% of the unassembled, assembled-but-unmapped, and mapped pieces, respectively, contain 20-mer repeats that are estimated to occur at least twice in the genome. About half of these 20-mer repeats are recognizable transposable elements (TEs) in RepeatMasker ( http://www.repeatmasker.org , and TE compositions in different categories of assembled data are summarized in Table S2 . The most problematic misassemblies are those that occur within genes, as these affect our ability to annotate the genome. Hence, we compared the WGS data to gene sequences defined by nr-KOME and excised from IRGSP japonica . We searched for alignment discrepancies of at least 500 bp, consistent with misassembled reads, and interpreted any increase in the discrepancy rate from Syngenta japonica to Beijing indica as being due to polymorphic differences. There are remarkably few genes with discrepancies in coding exons, only 0.23% in Syngenta japonica and 1.44% in Beijing indica . If we include UTR exons and introns, the rates are 0.84% in Syngenta japonica and 5.65% in Beijing indica . Hence, the implication is that WGS misassemblies occur less frequently than polymorphic differences. Table 2 shows the number of nr-KOME cDNAs that are found in each of the three rice assemblies, using the criterion that 95% of the coding region must be alignable in BLAT [ 32 ]. Some cDNAs align to multiple pieces of the assembly, but most align to one single piece. Even if we consider only the latter case, all three rice assemblies are at least 91.2% complete. Regardless of the assembly, the gaps seem to be random, as genes that are fragmented in one assembly are often intact in another. Of the cDNAs, 98.1% can be found in one piece in either Beijing indica or Syngenta japonica (if we also insist that they be anchored to the map, this number becomes 97.7%). Combining all three rice assemblies results in 98.6% completeness. Strikingly, only 0.7% of the genes align to the unmapped Beijing indica sequence, despite the fact these unmapped data were 12.3% of the searched sequence. This is the first of many examples that we will provide to support the idea that the unmapped pieces are extremely gene poor. Table 2 Summary of nr-KOME cDNAs with Complete Alignments (Not Including UTRs) in Each of the Three Rice Assemblies We require that 95% of the gene be aligned, but there are two ways to count. “Found in genome” will accept fragmented genes that are aligned in multiple pieces, whereas “aligned in one piece” will not Gene Identification and Classification We used an unorthodox method for gene identification. The conventional method, epitomized by Ensembl [ 33 ], uses sequence similarity to known genes and proteins to remove erroneous predictions, which are a serious problem for vertebrates because of the preponderance of large, multiexon genes, some of which can be megabases in size. However, plant genes are only a few kilobases in size, and given that Arabidopsis is still the only other sequenced plant, the Ensembl approach would remove many valid genes in a misguided effort to control a less serious problem. We removed erroneous predictions by relying instead on the fact most of them are actually TEs that are mistakenly called genes. Ultimately, our method is vindicated by whole-genome microarray experiments using 70-mer oligos that are hybridized to mRNA from five different tissue types. One finds that 82% of predicted rice genes with no homologs in Arabidopsis can be detected in this manner, as opposed to 88% of predicted rice genes with homologs (L. Ma, J. Wang, C. Chen, X. Liu, N. Su, et al., unpublished data). For the purpose of discussion, we will classify rice genes as WH (with homolog) or NH (no homolog), based on sequence similarity to Arabidopsis, with the stringency set to a level that is typically found in the literature. Nucleotide sequences are translated into protein sequences, and the Arabidopsis genome is searched in all six reading frames using TBlastN at E-values of 10 −7 . Putative exons are chained together, and success is declared if we can account for either 50% of the protein or 100 residues. We are not concerned that more sensitive search algorithms might identify homologies that we missed. Even the best algorithms are limited in their ability to identify structural homology by sequence similarity [ 34 ]. The main objective is to show how genes that are highly homologous or nonhomologous are sufficiently different as to merit special attention in data analysis, and the simplest way to emphasize this is to draw a dividing line. For methodological consistency, we annotated all three rice assemblies using the same procedures. We use FGENESH [ 35 ] for gene prediction because it has been shown to be the best of the available ab initio algorithms for rice [ 1 ]. An updated performance assessment is shown in Figure S3 . The challenge in removing erroneous predictions resulting from TEs lies in how we compensate for the fact that the database used by RepeatMasker is incomplete. Figure 2 demonstrates how grass genomes are organized as gene islands of low copy number separated by intergenic repeat clusters of high copy number. We set a dividing line at copy number 10, not because there are no TEs below it but because there are few genes above it. Specifically, for genes defined by nr-KOME, 99.4% of the exons and 98.1% of the introns are attributed to 20-mers of copy number under 10. Using the finished sequence of Chromosomes 1 and 10, we show in Figure S4 that the mean (median) sizes are 23.7 kb (9.6 kb) for gene islands and 5.6 kb (3.5 kb) for intergenic repeat clusters. Applying RepeatMasker to these intergenic repeat clusters only identifies 47.6% as TEs, overwhelmingly gypsy and copia . We therefore propose to filter the predictions by removing genes for which 50% of their coding region is attributable to any combination of RepeatMasker TEs or 20-mers of copy number over 10. Figure 2 A Region on Beijing indica Chromosome 2, Showing Three Gene Islands Separated by Two Intergenic Repeat Clusters of High 20-mer Copy Number Transposable elements identified by RepeatMasker are classified based on the nomenclature of Table S2 . Depicted genes include both nr-KOME cDNAs and FGENESH predictions. Although this filter might remove some real genes, it removes only a small fraction of them, as demonstrated by the nr-KOME cDNAs, where it eliminates 0.9% of these genes. In contrast, applying this same filter to the FGENESH predictions eliminates 19%–22% of the gene set, as indicated in Table 3 . We believe that most of the removed predictions are TEs and that the benefits of removing these artifacts outweigh the risks of removing real genes. After this procedure, the gene counts range from 49,088 (Beijing indica ) to 45,824 (Syngenta japonica ) to 43,635 (IRGSP japonica ). Previous estimates for Chromosomes 1, 4, and 10 made no such correction and found slightly larger numbers. About 45%–47% of predicted genes are NH, in contrast to 34.3% of nr-KOME cDNAs. This discrepancy is due to a combination of prediction errors and the fact that NH genes are difficult to clone because they are poorly expressed (data not shown). Radically different numbers have been given for mean gene size, from 2.6 kb in Chromosome 10 to 4.5 kb in our previous article. As we show in Table 4 , much of this discrepancy can be explained by differences in definition. Predicted genes have a mean (median) size of 2.5 kb (1.8 kb). We get the same result for nr-KOME if we exclude UTRs, but we get a size of 3.6 kb (2.9 kb) if we include UTRs. If we restrict the genes to WH genes, this raises the gene size to 4.0 kb (3.4 kb). Table 3 Number of FGENESH Predictions in All Three Rice Assemblies Filtering refers to the process in which we remove predictions where 50% of the coding region is attributable to any combination of RepeatMasker TEs or 20-mers of copy number over ten. EST confirmation requires 100 bp of exact match Table 4 Characteristics of FGENESH Predictions and nr-KOME cDNAs Predicted genes do not included UTRs. Mean (median) are both given Even after removing likely TEs, two particular subclasses warrant caution, as they contain a higher than normal rate of erroneous predictions, which is reflected in a reduced rate of confirmation by ESTs. Overall, we used 200,648 ESTs from indica, japonica, and other rice subspecies. The confirmation rule is exact match over 100 bp. Genes predicted in unmapped sequences are confirmed at much lower rates than genes predicted in mapped sequences—about 11 times lower, even after removing 3.4 times as many unmapped genes as likely TEs. Genes unique to only one assembly also show lower confirmation rates, by a factor of roughly nine, when compared with the 35,052–36,940 genes that are shared by all three assemblies, as summarized in Figure 3 . A more detailed analysis is given in Table S3 . What is important is that few of these genes are likely to be real. We can use the ratio of the EST confirmation rates to correct our gene count estimates. Beijing indica is computed as [(36,940 × 39.6) + (1967 × 28.1) + (1586 × 20.4) + (8595 × 4.9)]/39.6 = 40,216. Similarly, we get 37,794 for Syngenta japonica and 37,581 for IRGSP japonica . If unique genes are truly expressed at lower levels than shared genes, this procedure might underestimate the gene count. One should thus interpret these numbers as lower bounds. Figure 3 Overlapping FGENESH Predictions in All Three Rice Assemblies Two predictions are shared when 50% of their coding regions can be aligned. Because of imprecision in the predictions and overlap criteria, we get slightly different numbers for each assembly, and these are encoded through multiple color-coded numbers in the Venn diagram. EST confirmation requires 100 bp of exact match. Unlike the genes, we do not bother to show a different number for each assembly, because they are very similar. Using the same EST adjustments, the number of predicted genes in Beijing indica that are not found in either japonica assembly is 1,064. Conversely, Syngenta japonica has 1,517 predicted genes that are not in indica (the number for IRGSP japonica is 1,479). As a fraction of the totals, 2.2% and 3.3% of indica and japonica genes, respectively, are unique to the subspecies, which is plausibly comparable to the amount of sequence that might still be missing. There is little difference in gene content between indica and japonica, but major differences are seen in the intergenic regions. Only 260 Mb (72%) of the mapped sequences can be aligned. This remains true no matter how much we relax the alignment parameters, and despite the fact that we had 34,190 “anchor points” (see Figure 1 ), which ensure that the indica–japonica comparisons are always made between the same regions of the chromosomes from the two subspecies. This unalignable fraction would be even larger if unmapped and unassembled sequences were included. Notice also that 20-mer repeat content is 59.2% in mapped-but-unaligned regions, as compared to 31.8% in mapped-and-aligned regions. Everything that we see is consistent with the fact that plant intergenic regions are rapidly evolving [ 36 ]. As further proof of this fact, Table 5 shows the SNP rates in these alignable regions. The rates vary from as little as 3.0 SNP/kb in coding regions to as much as 27.6 SNP/kb in identifiable TEs. Table 5 Variation between indica and japonica Defined by SNP and Insertion–Deletion (Indel) Rates Variation rates for 5′ UTR, coding, intron, and 3′ UTR refer to gene regions defined by nr-KOME. To demonstrate where the high SNP rates come from, we consider regions of 20-mer copy number under ten and RepeatMasker TEs Biological functions are inferred by and displayed within the Bioverse framework [ 37 , 38 ] by combining more than seven of the latest computational techniques, including profile–profile comparison to well-curated protein families, motif discovery, and structural assignment/prediction. Note that we do not use transitive annotations, as their error propagation rates are too high. We present these results in Gene Ontology (GO) [ 39 ] and InterPro [ 40 ] formats. Functions are assigned to 60.2% of WH genes and even to 17.5% of NH genes, reflecting the fact that Bioverse uses highly sensitive techniques. Figure 4 shows a couple of our GO comparisons, focused on plant-specific categories in Gramene [ 41 ]. From the fraction of the gene set in each category, rice and Arabidopsis are remarkably similar. FGENESH-predicted genes and nr-KOME cDNAs exhibit very similar patterns too, confirming the unbiased nature of these cDNAs. InterPro domain categories tell much the same story, and these data are summarized in Table S4 . Figure 4 Functional Classifications from GO, Focused on Plant-Specific Categories Outlined by Gramene (A) compares predicted genes from Arabidopsis and Beijing indica . (B) compares predicted genes from Beijing indica with nr-KOME cDNAs. We ignore categories with less than 0.1% of the genes. Bioverse is distinguished from other annotation pipelines in that it also determines protein–protein interactions. Two proteins are predicted to interact if they are both similar in sequence to proteins involved in known interactions. The known interactions are taken from numerous sources, including Protein Data Bank [ 42 ] and the Database of Interacting Proteins (which stores yeast two-hybrid studies, affinity column studies, and literature searches) [ 43 ]. The resultant network has 1,879 proteins/nodes with 8,902 unique interactions. Figure 5 highlights a small portion of this network, for defense proteins (i.e., classified as “defense related” under GO molecular function or “defense response” under GO biological process) and their direct neighbors in the network. Many occupy central positions, meaning the network would fall apart if they were removed. Such genes are essential for cell survival [ 44 ]. More details can be found at http://bioverse.compbio.washington.edu . Figure 5 A Sample Bioverse-Predicted Interaction Network for Defense Proteins and Their Direct Neighbors The symbols are colored to indicate some of the major GO categories under “molecular function.” We draw a cross over the symbol for an NH gene. Rectangles indicate proteins that are manually classified as being R-genes. They appear on genes that are not colored as defense, because some genes have multiple functions, not because of an annotation error. The white circles with green outline are unannotated genes that might also belong to this network, at a lower confidence. Figure S5 shows that, near the centromeres, there is an increase in TE density (especially for large class I TEs like gypsy and copia ) and a decrease in gene density. A more detailed view is given by the pullout figures of Figure S6 , right down to the level of individual genes and TEs, to emphasize the excellent level of concordance between the two different WGS assemblies: Beijing indica and Syngenta japonica . Evidence of Whole-Genome Duplication Duplication of individual genes, chromosomal segments, or even entire genomes is an important source of raw materials for gene genesis [ 45 ]. In the extreme case of a whole-genome duplication (WGD), convincing examples are difficult to find because of the expected rapid loss of duplicated genes and because the rate of individual gene duplication is high enough to mask any remnants of an ancient WGD [ 46 ]. Yeast was the first genome in which a WGD was detected [ 47 ]. In plants, the existence issue is not disputed, as polyploidy is common [ 48 , 49 , 50 , 51 , 52 , 53 ], but even with complete genome sequence, many details remain obscure. For Arabidopsis, the number and timing of these duplication events is still unknown [ 54 , 55 , 56 , 57 , 58 , 59 ]. For rice, segmental duplications were known [ 60 , 61 , 62 ] before the rice genome sequence was published. However, detailed analysis of this sequence has resulted in the contradictory assertions that rice is an ancient aneuploid [ 14 ] and an ancient polyploid [ 15 ]. Here, we resolve this conflict by showing that every conceivable class of duplication that could have happened did in fact happen, including a WGD. We accept that every class of duplication is present in the same genome, and we thus explicitly assign, to every homolog pair, a status as to the class of duplication from which it came. For the sake of discussion, we define three classes: segmental duplication of multiple genes along a chromosome, tandem duplication of individual genes, and a category called background duplications to encompass everything else that cannot be so easily classified. In this conception, a WGD is a collection of segmental duplications that cover a majority of the genome, all of which date back to a common time in evolutionary history. All three rice assemblies give the same result, so we show only Beijing indica . Unlike previous analyses, we avoid predicted genes. Instead, we define a homolog pair to be a single nr-KOME cDNA and one of its potentially many homologs within rice. These homologs are defined by translating the cDNA's coding sequence into protein and searching the rice genome in all six reading frames for putative exons, with TBlastN at E-values of 10 −7 . Exons in the same order and orientation are linked together, and success is declared if these linked exons can account for 50% of the original protein sequence. This technique has the advantage that the homolog need not be a cDNA or a predicted gene (as neither dataset is likely to be complete). In fact, the homolog might even be a remnant of an ancient duplication that is no longer a functional gene. Complications are found at two extremes. Many cDNAs have no homologs, but many others have too many homologs. In particular, 24.5% of WH genes have no homologs in rice, whereas 64.4% of NH genes have no homologs in rice. Because NH genes are dispersed throughout the genome, sandwiched between WH genes, we cannot adopt a strict colinearity rule in our search for duplicated segments. There would be too many exceptions. Conversely, when there is at least one homolog in rice, the mean (median) number of homologs per cDNA is 40 (5). Rather than deal with the complexities of this situation, we focus first on the cDNAs with one and only one homolog. This reduces the background duplication noise and allows us to identify trend lines indicative of segmental and tandem gene duplications. We can then add back those cDNAs with more than one homolog that we had rejected earlier by using our newly defined trend lines to constrain the choices. The above procedure leaves us with 2,271 homolog pairs (or cDNAs). We adopt a graphical approach, because in the presence of massive background noise, trend lines are often easier to identify by eye than by software. Figure 6 depicts Chromosomes 2 and 6, and Figure S7 depicts all 12 chromosomes. There are 18 pairs of duplicated segments that together cover 65.7% of the length of all the mapped super-scaffolds. The mean (median) number of homolog pairs per segment is 34 (23). The segment sizes are 6.9 Mb (5.4 Mb), and they differ by 43% (42%) within a segment pair, which is not at all unexpected given the rapidly evolving nature of the rice intergenic regions. Instances of multiple duplicated segments on the same chromosomal region are extremely rare, covering only 0.9% of the total length. No additional multilevel duplications are detected if we use cDNAs with up to two homologs, as opposed to those with only one. Notice also that there are duplicated segments on all 12 rice chromosomes, as summarized in Figure 7 . Figure 6 Duplicated Segments in the Beijing indica Assembly Depicted here are the plots for Chromosomes 2 (A) and 6 (B). Each data point represents the coordinated genomic positions in a homolog pair, consisting of one nr-KOME cDNA and its one and only TBlastN homolog in rice. Shown on the x -axis is the position of a gene on the indicated chromosome, and shown on the y -axis is the position of its homolog on any of the rice chromosomes, with chromosome number encoded by the colors indicated on the legend at the right. Figure 7 Graphical View of All Duplicated Segments The 12 chromosomes are depicted along the perimeter of a circle, not in order but slightly rearranged so as to untangle the connections between segments. Overall, we cover 65.7% of the genome. One can date the duplications by computing the number of substitutions per silent site (Ks). Multiple substitution corrections are done within K-Estimator [ 63 ]. To improve our statistics, we now include the higher-order homologs (those cDNAs with more than one homolog that we had removed before). Table 6 shows that this doubles or triples the number of homolog pairs in every segment and brings the mean (median) to 74 (53). The resultant Ks distribution is shown in Figure 8 . One pair of segments on Chromosomes 11 and 12 is more recent in origin and has more homolog pairs per unit length than all the others. It was previously identified in many publications. If we ignore this segment pair, the mean Ks is 0.69, dating the duplication event to 53 million years ago (Mya), assuming a neutral evolutionary rate of 6.5 × 10 −9 substitutions per silent site per year [ 64 ]. Most of the uncertainties are due to the multiple-substitution corrections for Ks. Another popular algorithm for Ks [ 65 ] dates the duplication event to 94 Mya. Figure 8 Distribution of Substitutions per Silent Site (Ks) for Homolog Pairs in Segmental, Tandem, and Background Duplications In (A), contributions from the recent segmental duplication on Chromosomes 11 and 12 are colored in red. The tandem duplication data are shown on two different scales, one to emphasize the magnitude of the zero peak (B) and another to highlight the exponential decay (C). Background duplications are shown in (D). Table 6 Summary of Duplicated Segments in the Beijing indica Assembly We give start and stop positions on the pseudo-chromosome, segment sizes, number of homolog pairs, mean Ks rates, percentage of homolog pairs with Ks < 0.25, and flanking nr-KOME cDNAs. One set of numbers is for the initial analysis of those cDNAs with one and only one homolog. A second is for the analysis of additional cDNAs with higher-order homologs a Computed total and mean omit the recent segmental duplication on Chromosomes 11 and 12 Chr, Chromosome The molecular clock can also vary between genes and between taxa [ 66 , 67 ]. Evidence for the former is seen in the width of the distribution for Ks in Figure 8 , which has a standard deviation of 49.8% based on individual homolog pairs (as opposed to 14.5% when based on duplicated segment pairs). We believe that the variation between genes will cancel out, but we cannot remove the systematic error resulting from the multiple substitution corrections or the potential error in the 6.5 × 10 −9 evolutionary rate (which was derived from a small number of genes). However, all we really want to know is whether the duplication event occurred before or after the origin of the grasses, 55–70 Mya [ 68 ]. To this end, phylogenetic approaches can be used, albeit for a limited number of genes, because so few plants have been fully sequenced. A majority of these phylogenies indicate that the duplication event occurred before this pivotal point in evolution [ 14 ]. Almost certainly, the duplication event occurred after the divergence of monocots and eudicots, 170–235 Mya [ 69 ]. However, the best evidence for the statement that the duplication event must have predated the origin of the grasses is the fact that there is no other way to reconcile it with the widely observed synteny between different grass genomes [ 70 ]. In striking contrast, the Chromosome 11 to 12 duplication dates back to just 21 Mya, which postdates the origins of the grasses by a comfortable margin. If we accept that a WGD occurred before the divergence of maize–rice, and that a duplication in Chromosomes 11 and 12 occurred afterward, we might then expect to find two levels of duplication in this region of rice. We thus extended our analysis to consider cDNAs that map to as many as four loci. No indications of such a multilevel duplication could be found. Undaunted, we decided to try another approach and analyzed the maize–rice synteny, starting from the maize genetic map [ 71 ]. The results are given in Figures S8 and S9 . We found 35 pairs of syntenic segments covering 71.4% and 52.9% of the maize and rice genomes, respectively. All previously identified segments are confirmed, except for those on Chromosomes 11 and 12 of rice. No synteny is found in the vicinity of this recent duplication. There are many explanations, and they need not contradict our hypothesis, as only 65.7% of the rice genome is in identifiably duplicated segments, and the region from Chromosome 11 to 12 is a minuscule 3.0% of the genome. It is possible that any traces of the WGD had already been lost by the time this recent duplication occurred. The region is also sufficiently small that any synteny with maize would be difficult to detect. It is too early to draw conclusions, especially as maize–rice synteny appears to be much more complicated than previously thought [ 72 ]. Given how so much of the rice genome is covered by segmental duplications, and the fact that all but one of our 18 segment pairs date back to the same time, give or take a standard deviation of 14.5%, the simplest interpretation is that a WGD did occur and that it happened before the origin of the grasses. However, it is equally clear that other classes of duplications are also present, and these are worth investigating too. Ongoing Individual Gene Duplications Tandem duplications are represented by the trend along the diagonal, Y = X, that is observed in all chromosomes (see Figures 6 and Figure S7 ). Segmental duplications within the same chromosome are possible, but their trend would not be along the diagonal, and none were actually seen in our analysis. As an indicator of the prevalence of the three different duplication classes, we use the number of homolog pairs before and after the inclusion of higher-order homologs. Segmental duplications contain 609 and 1,340 pairs, whereas tandem duplications contain 311 and 957 pairs. We can increase the tandem numbers by relaxing our definitions to allow two TBlastN homologs of an nr-KOME cDNA to count as a homolog pair (instead of insisting that one always be a cDNA). This is what we use in the Ks distribution plot of Figure 8 , which contains 1,696 homolog pairs. Rather than a maximum in the distribution at some nonzero Ks, we find a big peak at zero Ks, followed afterward by an exponential decay. The implication is that tandem duplication is an ongoing evolutionary process that provides an endless source of raw materials for gene genesis. If we adopt the methods and parameters of the Arabidopsis genome paper, we find that 16.5% of the rice genome is tandemly duplicated, compared to 16.2% of the Arabidopsis genome. Note, however, that the Ks distribution for tandemly duplicated genes in Arabidopsis is highly unusual, in the sense that it does not exhibit the big peak at zero Ks that is seen in virtually every other plant genome [ 52 ]. In addition to segmental and tandem duplications, there is a third and last class of duplications that looks like background noise in our figures. The number of homolog pairs is 1,351 and 32,384 before and after higher-order homologs, respectively, although with no trend line to constrain the choice of homologs, that second number is almost certainly an overestimate, since only 4,212 cDNAs are involved. Surprisingly few of these higher-order homologs are the result of processed pseudogenes, as the number of cases in which a multiexon cDNA pairs with a single-exon TBlastN homolog is 9.8%. To demonstrate how overwhelmingly these higher-order homologs contribute to the background noise, Figure 9 depicts what Chromosome 2 would have looked like if we had included them. For simplicity of interpretation, Figure 8 is the Ks distribution of the cDNAs with one and only one homolog. This distribution has characteristics of the distribution for tandem duplications—large peak at zero Ks followed by exponential decay—except that the magnitudes of the Ks are much larger for background duplications. We believe that most of these background duplications were originally tandem duplications that, over time, migrated to other parts of the genome, but we cannot rule out the possibility of direct duplications to remote loci. Some older duplications may even be due to migration of genes from segmental duplications, but these are a small part of the overall picture. However we do the counting, it appears that this combination of recent tandem and background duplications, which we call individual gene duplications, would rival any contribution from the segmental duplications. Figure 9 A View of All Duplications Found on Rice Chromosome 2 In contrast to Figure 6 , where we featured those cDNAs with one and only one TBlastN homolog, here we show all detectable TBlastN homologs, up to a maximum of 1,000 per cDNA. Tandem and segmental duplications show markedly different Ka/Ks distributions, a popular test for evolutionary selection, where Ka and Ks refer to the fraction of nonsynonymous and synonymous sites, respectively, that are changed within a homolog pair [ 73 ]. Ka/Ks is one under neutrality, below one under purifying selection, and above one under adaptive selection. Tandem duplications tend to have larger Ka/Ks values, as we show in Figure 10 . The averages are 0.720 (tandem) and 0.365 (segmental), and more homolog pairs exhibit Ka/Ks > 1 in tandem duplications. This is consistent with the observation that more recent duplications tend to have larger Ka/Ks values [ 74 ] and with the idea that, immediately after duplication, one of the two genes undergoes a fast evolving phase [ 75 ]. Finally, let us consider again those nr-KOME cDNAs with one and only one homolog. Among the ones assigned to a tandem duplication, 65.3% are NH, but among the ones assigned to a segmental duplication, 23.8% are NH. Hence, there is a marked correlation between NH genes and tandem duplications. Figure 10 Ka/Ks Distribution for Homolog Pairs Ka and Ks are the fraction of the available nonsynonymous and synonymous sites that are changed in the homolog pairs. Ka/Ks > 1 is an indicator of positive selection. Shown is the Ka/Ks distribution for segmental duplications (A) and for tandem duplications (B). Our WGD is in good agreement with the results of Paterson et al. [ 15 ], but we can also explain the seemingly contradictory results of Vandepoele et al. [ 14 ] First, they did not have a complete genome; about two-thirds of their segmental duplications were interrupted by a break in the assembly. Second, their algorithms were very likely confounded by the many NH genes with no homologs in rice itself and by the many individual gene duplications that in aggregate masked the WGD. In fact, their segmental duplications had a Ks distribution similar to ours, but they only covered 15% of the genome. Then, when they examined the distribution of Ks for all duplicates, what they found was a big peak at zero Ks. This lead them to conclude there was no WGD, when, in fact, almost every class of duplication that had been hypothesized was present, and they needed only to allow for that. Discussion Until recently, Arabidopsis was the only sequenced plant genome. When two rice genomes were first published in draft format, the comparative analyses that could be done were hindered by a lack of long-range contiguity. Now, there are three plant genomes (indica rice, japonica rice, and Arabidopsis) with multimegabase contiguity. In our analyses, we strived to maintain methodological consistency. To assess the accuracy of our assemblies, we first compared IRGSP japonica to Syngenta japonica, so that polymorphic differences would not be a confounding factor. To compare gene content in the three rice assemblies, we annotated them all with the same procedures. Our conclusion is that, even if the WGS method does fall just slightly short of the clone-by-clone method in terms of accuracy and completeness, it comes remarkably close. This is why all the genome-sequencing projects now being funded by the National Human Genome Research Institute (in the United States) are being done with WGS methods ( http://www.genome.gov/11007951 ). Rice is also now one of the few organisms with the luxury of having a complete genome sequence for two important subspecies. Comparisons of indica and japonica reveal strikingly little difference in the gene content, but there are massive intergenic differences. This vindicates our strategy to focus on genic sequences, because if the intergenic sequences are so unstable even between indica and japonica, they are highly unlikely to be functional. Our analysis of the duplication history in rice resolves a simmering dispute and, at the same time, raises some intriguing questions. We find evidence for an ancient WGD, a recent segmental duplication, and massive ongoing individual gene duplications. This last phenomenon can explain certain unexpected findings. Sequencing of orthologous loci between grass genomes has identified many smaller-scale rearrangements that were not seen in the original map-based studies. Many of these exceptions to synteny are due to tandem duplications [ 76 , 77 , 78 ], which makes sense, given how these duplications are a frequent and ongoing event for grass genome evolution. In addition, the massive ongoing individual gene duplications provide a never-ending source of raw material for gene genesis. We believe that the large number of rice NH genes is a transient effect of this ongoing process. The contrary argument is that any such transients cannot be long-lived, as one of the two genes must decay rapidly to avoid the dosage-doubling problem [ 79 , 80 ]. We believe this is irrelevant when there is a continual injection of new gene duplicates. Additional details must, however, be deferred to a future article, in which we can better address other important issues, such as the critical need to confirm NH genes in proteomics and conservation in the maize genome sequence. Looking toward the future, we would point out that the Chinese Superhybrid Rice Genome Project was designed to include not only a major subspecies of rice, namely, the indica variety represented by 93–11, but also the maternal strain of the LYP9 superhybrid, PA64s, which has a complex breeding history incorporating genetic material from indica, japonica, and javanica —all of the major subspecies of cultivated rice. Work on PA64s is continuing at our Beijing center. For the research community, we will be providing DNA microarrays to facilitate the systematic studies of gene expression in different tissues and developmental stages, and under different physiological and environmental conditions. We will develop molecular markers for mapping causative genes in mutant lines and marker-assisted breeding. This publication, and the associated data release, is also a fitting way to celebrate the end of 2004, which the General Assembly of the United Nations declared to be the International Year of Rice ( http://www.fao.org/rice2004 ). Materials and Methods Construction of reference cDNAs: nr-KOME The initial Knowledge-Based Oryza Molecular-Biological Encyclopedia dataset [ 25 ] had 28,444 japonica cDNAs with complete open reading frames. These cDNAs were aligned to Syngenta japonica, and when two alignments overlapped by at least 100 bp, the smaller cDNA was removed. A small number of clones could not be aligned—not even partially—to any of our three rice assemblies (Beijing indica, Syngenta japonica, and IRGSP japonica ). Removing these as nonrice contaminants gave a set of 19,079 nonredundant cDNAs that we call nr-KOME. Because the sequence quality is so high, we could use the longest open reading frame for the overwhelming majority of these cDNAs, without having to correct for sequencing errors. Minor corrections are applied to 2.5% of these cDNAs, following the methods first developed for GenScan [ 81 ]. Repeats and their effects on WGS misassembly The basic procedure for converting sequence reads into contigs and scaffolds was described in our original publication on RePS [ 26 ], our WGS assembler. A common source of confusion is the distinction between mathematically defined repeats (MDRs) and biologically defined repeats. What we focus on are MDRs, which refer to 20-mer sequences that are exactly repeated in the genome, without regard to their underlying biological context. In our nomenclature, “depth” refers to the number of times that a 20-mer appears in the unassembled sequence reads and “copy number” refers to the number of times that it appears in the (correctly assembled) genome. “Coverage” is the number of times that the genome is redundantly sampled, and therefore depth = copy number × coverage. Special procedures are used to compute depths efficiently [ 27 ]. In a WGS assembly, the problems arise from the MDRs, which are not equivalent to the biologically defined repeats. For example, TEs qualify as biologically defined repeats, and they can be recognized, even after many millions of years of degradation, by specialized programs like RepeatMasker ( http://www.repeatmasker.org ). However, the degradation makes it trivial to distinguish between two copies of an ancient TE, so these do not cause assembly problems. It is also relatively easy to distinguish between gene duplicates, because their introns and flanking intergenic regions are under fewer evolutionary constraints than their exons. Even for recent TEs and gene duplicates, assembly problems can be avoided, because RePS computes the copy number for every 20-mer in the WGS assembly, and it will refuse to join anything that might be ambiguous. Indeed, the only way a misassembly can occur is if there is a low copy MDR and its copy number is underestimated by RePS. All of our tests show that, although this can happen, it is a rare event. On the usefulness (or not) of BAC end pairs The fundamental challenge was that we had to create super-scaffolds of megabase size from scaffolds of 30-kb size. It is generally thought that BAC end pairs are useful for this purpose, but this is not true when the BAC inserts, typically 122–187 kb, are much bigger than the scaffold sizes. Instead of linking adjacent scaffolds, they link every fourth to sixth scaffold. The fact that the density of BAC ends is 2.3 kb does not help, because there is no way to determine the order and orientation of the overlapping BACs. Fingerprint maps do provide some ordering information, but nothing like 2.3-kb resolution, and orientation information is still missing. The danger in using the BACs at this point is that you end up with a morass of interleaving super-scaffolds [ 26 ], with no way to untangle them. We actually did an assembly with only the BACs, and the result was that the super-scaffolds were 87% larger than they should have been. In the mouse project [ 82 ], the solution was to use fosmid end pairs, because these inserts are constrained to an almost ideal size of 40 kb. In the case of rice, we did not need to sequence fosmid end pairs, because by combining the indica and japonica WGS assemblies, it is possible to get linking information at the requisite length scales. We did of course use all available BAC end pairs [ 83 ] ( http://rgp.dna.affrc.go.jp/blast/runblast.html , but they were only useful after the intermediate-range linking that came from combining WGS assemblies. Misassemblies versus polymorphic differences To verify our WGS assemblies on the smaller-length scales that are more characteristic of genes, we compare them with IRGSP japonica, taking the latter as the “gold standard” not because it is perfect but because it more likely to be correct. We focus on gene regions by aligning nr-KOME cDNAs to IRGSP japonica and excising the sequences from the 5′ to 3′ UTRs, including introns and an additional 500 bp at both ends. What we search for are potential misassemblies due to misplaced reads. Given that a typical read is 500 bp, these should appear as segments of 500 bp or more in which the excised gene sequence cannot be aligned with the WGS assembly. Such discrepancies are noted based on where they occur in the context of the gene. Although it is possible to detect more than one discrepancy per gene, we only count the most serious discrepancy in each gene based on the likelihood of it being functional. The prioritization is from coding exon, to UTR exon, to intron. Notice that discrepancies of this nature are not always from misassemblies. In the Beijing indica comparison, they can also be due to polymorphic differences. Although there is no way to tell what any particular discrepancy is, we know the misassembly rate from the Syngenta japonica comparison. Therefore, any increase in the discrepancy rate in the Beijing indica comparison can be attributed to polymorphic differences. Ab initio predictions in WH versus NH genes FGENESH [ 35 ] behaves very differently for WH and NH genes, as defined by nr-KOME. Following the methods of our recent review [ 84 ], we compute false positive (FP) and false negative (FN) rates. Error rates are given on a per amino acid basis. This means that in addition to correctly identifying the coding bases, we require the reading frame to be correctly determined. WH genes show very low error rates (FP = 0.10 and FN = 0.05). Although NH genes show higher error rates (FP = 0.35 and FN = 0.25), these are not that much worse than human genes (FP = 0.30 and FN = 0.12), and like it or not, error rates like these are the state of the art in ab initio prediction. On closer examination, it is clear that most of the problems in rice are caused by single-exon genes with small coding regions, which are more prevalent among NH genes and form a category that all ab initio algorithms handle poorly. This category of genes does not affect the gene count because FP and FN cancel each other out. We therefore focus on removing TEs that are mistakenly called genes. Comparison of indica - japonica to identify SNPs The sequence alignments for indica and japonica are straightforward, with almost no chance of paralog confusion, because of our 34,190 unique “anchor points” (see Figure 1 ). We partition the sequence into four nonoverlapping categories called unassembled, assembled-but-unmapped, mapped-but-unaligned, and aligned. The last category is where almost all of the genes are, and where we can get polymorphism data. Detailed sequence alignments are computed with CrossMatch, a Smith-Waterman algorithm that is included in Phrap ( http://www.phrap.org ). This is preferred to any of the BLAST alignment tools, which, although they are faster, occasionally miss subtle details. To discriminate between polymorphisms and sequencing errors, we use the error probability p attached to every base, and given as Q = −10 × log( p ). Following the rules established in the early days of large-scale polymorphism discovery [ 85 ], we use thresholds of Q > 23 at the SNP site and Q > 15 for the two flanking 5-bp regions. Experience has taught us that higher thresholds (30 and 22, respectively) are required for the indels. For comparison, an independent analysis [ 86 ] reported mean rates of 7.1 SNP/kb and 2.0 indel/kb, with 98% of these SNPs experimentally confirmed. Our SNP rates are two times higher because we aligned more of the intergenic sequence. If we eliminate this factor, say, by restricting our rates to the introns of the genic regions defined by nr-KOME, our rates are 6.1 SNP/kb and 1.3 indel/kb, which are actually lower than the rates from that independent analysis. On the reliability of the p – p interaction data Bioverse annotations in this article are dated July 2003 (FGENESH) and November 2002 (nr-KOME). Two proteins are said to interact if they are similar to two other proteins that are known to interact. Our criterion is that the product of the similarity measures (percentage identity) must exceed 0.15. For example, two proteins with 45% and 30% identity to two other proteins that are experimentally determined to interact would be rejected, as their score is 0.45 × 0.30 = 0.135. The reliability of this approach, especially for transfer of interaction data between organisms, has been demonstrated in Saccharomyces cerevisiae, Caenorhabditis elegans, Drosophila melanogaster, and Helicobacter pylori analyses [ 87 ]. As an example of a predicted interaction for rice that has been independently confirmed, Bioverse identification numbers 21736 and 8526 (score 0.21) show an interaction between CDK-activating kinase and H-type cyclins [ 88 ]. A general way to verify the predicted interactions is to compare them against known protein complexes in the Protein Data Bank. Unfortunately, there are few Protein Data Bank structures from rice, and even fewer are of protein complexes. Given this dearth of experimentally determined interactions for rice, Bioverse is almost the only source of large-scale interaction data. Details of the duplication and synteny analysis We defined a homolog pair as a single nr-KOME cDNA and its TBlastN homolog, but occasionally that TBlastN homolog will overlap with another cDNA. To avoid double counting, we keep only the larger of these two cDNAs. Segmental duplications identified by visual inspection must have at least five homolog pairs, with no more than 5 Mb between adjacent homolog pairs. We approximate the trend line with a second- or third-order polynomial, and to capture what our eyes indicate should be captured, we accept homolog pairs within a 500-kb radius of this polynomial. Slightly different definitions are used for tandem duplications, depending on application. For Ks, we allow two TBlastN homologs to count as a homolog pair and accept homolog pairs within a 50-kb radius of the diagonal, although the mean (median) center-to-center distance is 6.8 kb (4.7 kb). To compare tandem duplications in rice and Arabidopsis, we use the methods described in the Arabidopsis genome paper and analyze predicted genes with BlastP at E-values of 10 −20 . To determine the maize–rice synteny, we began with 1,063 maize genetic markers [ 71 ] and searched for BlastN alignments to rice of at least 100-bp size and 80% identity. Given the segmental allotetraploid origins of maize [ 89 ], many markers are associated with two loci in maize. Each marker aligns to a mean (median) of 1.9 (1) loci in rice. We used only the longest of these alignments and verified in retrospect that using all of them would not have mattered. In the end, there are 35 pairs of syntenic segments, which cover 71.4% and 52.9% of the maize and rice genomes, respectively, and the mean (median) number of markers per syntenic segment is 18 (12). Supporting Information Figure S1 Genetic Versus Physical Map Distance for All 12 Rice Chromosomes, Based on Beijing indica Similar results are seen with the other two assemblies, Syngenta japonica and IRGSP japonica . (1 MB EPS). Click here for additional data file. Figure S2 Number of Discrepant Markers in Comparisons of Genetic and Physical Maps for 1,519 Markers Found in All Three Rice Assemblies We count discrepancies where the markers are found (A) on different chromosomes and (B) in different locations on the same chromosome. (458 KB ZIP). Click here for additional data file. Figure S3 Gene Prediction by FGENESH, Tested against nr-KOME cDNAs Genomic size refers to the unspliced transcript, with introns, but constrained to the region from the start to stop codons. CDS size refers to the spliced transcript, without introns. Predictions are assessed with FP and FN rates, where per-aa (per amino acid) refers to the fact that we check whether the reading frame is correct. (351 KB ZIP). Click here for additional data file. Figure S4 Distribution of Sizes for Gene Islands and Intergenic Repeat Clusters, Based on Complete Sequence of Chromosomes 1 and 10 from IRGSP japonica Intergenic repeat clusters are regions of size larger than 1.5 kb (i.e., between a MITE and a gypsy / copia TE), where most of the 20-mer copy numbers exceed ten. Lower copy number regions are tolerated up to a “maximum gap size,” which defaults to 150 bp. Regions lying between two adjacent intergenic repeat clusters are taken to be gene islands. (233 KB ZIP). Click here for additional data file. Figure S5 Gene and TE Densities for Beijing indica Chromosome 7, as a Percentage of Sequence Length Near the centromeres, there is an increase in TE density (especially for the large, class I TEs such as gypsy and copia ) and a decrease in gene density. This is not an artifact of the fact that WGS assemblies underrepresent larger TEs, as much the same effect is observed when we use IRGSP japonica instead (data not shown). (362 KB ZIP). Click here for additional data file. Figure S6 Coordinated Annotation of the Individual Chromosomes for Beijing indica and Syngenta japonica We depict all the genetic markers, nr-KOME cDNAs, FGENESH gene predictions, and transposable elements identified by RepeatMasker. Genes are depicted as WH (colored blue) or NH (colored red) based on their similarity to Arabidopsis . TEs are decomposed into classes I, II, and III. Correspondence between indica and japonica is indicated by drawing a connecting line between the 5′ ends of the nr-KOME cDNAs that clearly align to both assemblies. (9.6 MB ZIP). Click here for additional data file. Figure S7 Duplicated Segments in the Beijing indica Assembly for All 12 Chromosomes, Plotted in the Manner of Figure 6 , and with a Total of 12 Panels (507 KB ZIP). Click here for additional data file. Figure S8 Complete Synteny between Maize and Rice I Each point indicates the genomic positions for a maize genetic marker and its highest confidence match in rice. The x -axis shows a specific chromosome for one genome, and the y -axis shows all chromosomes for a second genome, with the chromosome numbers color-coded as per the legend. We show here 12 panels for rice. (311 KB ZIP). Click here for additional data file. Figure S9 Complete Synteny between Maize and Rice II Each point indicates the genomic positions for a maize genetic marker and its highest confidence match in rice. The x -axis shows a specific chromosome for one genome, and the y -axis shows all chromosomes for a second genome, with the chromosome numbers color-coded as per the legend. We show here ten panels for maize. (288 KB ZIP). Click here for additional data file. Table S1 Raw Data for Beijing indica and Syngenta japonica Assemblies Read length is the number of Q20 bases with an error rate of 10 −2 or better. Effective coverage is based on the depth of reads in contigs over 5 kb in size, ignoring regions with 20-mer repeats. Clone insert sizes are specified in terms of tenth and 90th percentiles. (16 KB XLS). Click here for additional data file. Table S2 Transposable Elements Identified with RepeatMasker Are Put into Classes I, II, and III As a result of our efforts to identify indica – japonica polymorphisms, the sequence is divided into four nonoverlapping categories: unassembled, assembled-but-unmapped, mapped-but-unaligned, and aligned (with all the SNPs). (28 KB XLS). Click here for additional data file. Table S3 Detailed Analysis of Gene Overlaps from Figure 3 For each region of the Venn diagram, we use BLAT to align the predicted gene to the other assembly (or assemblies) where the gene is supposedly missing. The objective is to determine whether it is the sequence that is missing, or whether the discrepancy is due to the errors in the ab initio predictions. What we find is a bit of both. However, fragmented sequence assemblies are not a problem. If the gene is found at all, it is usually found in one piece. What is striking is that predicted genes that are unique to the two WGS assemblies do tend to be genuinely missing from IRGSP japonica sequence. This supports the idea that the WGS method can sometimes identify genes that are not well represented in the BAC clone libraries. (17 KB XLS). Click here for additional data file. Table S4 Table of InterPro Domain Rankings One table compares predicted genes from Arabidopsis and Beijing indica . The second table compares predicted genes from Beijing indica with nr-KOME cDNAs. (169 KB XLS). Click here for additional data file. Accession Numbers The DNA Data Bank of Japan/European Molecular Biology Laboratory/GenBank (BGI-RIS http://rise.genomics.org.cn [ 16 ]) project accession numbers for the WGS sequences discussed in this article are Beijing indica (
AAAA00000000, version
AAAA02000000) and Syngenta japonica (AACV00000000, version AACV01000000). Note Added in Proof The idea that TEs are often mistakenly annotated as genes was also suggested in a recent paper by Bennetzen et al. [ 90 ]. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC546038.xml |
549531 | Plasma antibodies against heat shock protein 70 correlate with the incidence and severity of asthma in a Chinese population | Background The heat shock proteins (Hsps) are induced by stresses such as allergic factors and inflammatory responses in bronchi epithelial cells and therefore may be detectable in patients with asthma. However, the etiologic link between anti-Hsps and asthma (its severity and related inflammatory responses such as interleukin-4 and immunoglobulin E) has not been established. We determined whether antibodies against Hsp60 and Hsp70 were present in patients with asthma and evaluated their associations with risk and severity of asthma. Methods We determined the levels of anti-Hsp60 and anti-Hsp70 by immunoblot and their associations with risk and symptom severity of asthma in 95 patients with asthma and 99 matched non-symptomatic controls using multivariate logistic regression analysis. Results Compared to the controls, asthma patients were more likely to have detectable anti-Hsp60 (17.2% vs 5.1%) and anti-Hsp70 (33.7% vs 8.1%) (p ≤ 0.001). In particular, the presence of anti-Hsp70 was associated with a greater than 2 fold risk for asthma (adjusted OR = 2.21; 95% CI = 1.35~3.59). Furthermore, both anti-Hsp60 and anti-Hsp70 levels were positively correlated with symptom severity (p < 0.05) as well as interleukin-4 and immunoglobulin E (p < 0.05). Individuals with antibodies against anti-Hsp60 and anti-Hsp70 were more likely to have a family history of asthma (p < 0.001) and higher plasma concentrations of total immunoglobulin E (p = 0.001) and interleukin-4 (p < 0.05) than those without antibodies. Conclusions These data suggest that anti-Hsp60 and especially anti-Hsp70 correlate with the attacks and severity of asthma. The underlying molecular mechanisms linking antibodies to heat shock proteins and asthma remain to be investigated. | Background Heat shock proteins (Hsps) are highly conserved proteins inducible in response to a wide variety of stresses (such as exposure to heat) and pathological (viral, bacterial or parasitic infections, and inflammation) or physiological (growth factors, cell differentiation, and hormonal stimulation) stimuli [ 1 , 2 ]. There are six main Hsp families (i.e., Hsp110, Hsp90, Hsp/Hsc70, Hsp60, Hsp40, and Hsp10-30) categorized on the basis of their apparent molecular masses detected by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). Hsps are involved in various biological functions including 1) intracellular chaperones of naive, aberrantly folded or mutated proteins, 2) cytokines of signal transduction cascades involved in inflammatory response, and 3) cytoprotective agents in response to the aforementioned stress stimuli [ 1 , 3 , 4 ]. In addition, Hsps are also involved in transport of proteins and peptides through cellular compartments, and can bind to endogenous antigenic peptides and transport them to the major histocompatibility complexes [ 5 , 6 ]. This suggests that Hsps may modulate immune and inflammatory responses and may be involved in the pathogenesis and/or be markers for risk and prognosis of certain diseases including asthma [ 7 - 13 ], given that many of the stress stimuli mentioned above are factors that can induce attacks of asthma. Asthma is a multifactorial and likely multigenic immune inflammatory disease of the upper airways, arising from complex interactions among environmental and genetic factors [ 14 , 15 ]. These factors may induce Hsp60 and Hsp70 in bronchi epithelial cells during the development of asthma [ 16 ]. Some Hsps present as self-antigens to the immune system, resulting in the production of autoantibodies in patients with inflammatory diseases and immune disorders after infections by bacteria, mycobacteria and Chlamydia [ 17 - 19 ]. Studies have demonstrated that these autoantibodies against Hsps were involved in the pathogenesis and/or prognosis of some diseases [ 8 , 20 - 23 ]. Up to now, few studies investigated possible associations of autoantibodies to human Hsps with the severity of asthma. In the present study, we determined the presence of autoantibodies to human Hsp60 and Hsp70 in 193 subjects with (n = 95) and without (n = 99) asthma by immunoblot analysis, and evaluated the associations of these autoantibodies with asthma severity and their correlation with interleukin-4 (IL-4) and immunoglobulin E (IgE) both involved in the development of asthma, by using multivariate logistic regression analyses. Methods Subjects and groups This 95 patients with asthma (54 males and 41 females) and 99 healthy, age-matched non-asthmatic controls (64 males and 35 females) were residents living in the same geographic area. Patients and controls were from Wuchang, one of the three cities of Wuhan and were all of Han nationality. Their age ranged from 10 to 45 years old (Table 1 ). All 95 patients were diagnosed according to diagnostic criteria and principles of management of asthma proposed by the American Thoracic Society [ 24 ] and did not have other pulmonary, cardiovascular and gastro-duodenal diseases. A standardized questionnaire was completed for each individual by physicians with extensive experience in allergic and immune diseases to obtain demographic information and known risk factors for asthma including personal and family history of asthma and frequency of attacks. Selection criteria for the controls included the absence of any personal history of asthma. Health examination and physical sign findings such as wheezing and forced expiratory volume (FEV%Pre) were also recorded. Neither patients nor controls had any history of chronic diseases such as cancer, diabetes, cardiovascular diseases and gastro-duodenal diseases. Patients with asthma were grouped by symptom severity and medication use according to the 2002 Global Initiative for Asthma Guidance [ 25 ] as intermittent, mild persistent, moderate persistent and severe persistent. Venous blood was collected into heparinized tubes to separate plasma for the detection of anti-Hsp60 and anti-Hsp70 as well as IgE and IL-4. Plasma samples from patients and controls were stored in aliquots at -80°C and thawed only once immediately before the tests were performed. Written informed consent was obtained from patients and controls, and the study was approved by the Tongji Medical College Ethics Committee. Table 1 Comparison of selected variables between patients with asthma and healthy controls Patients with asthma (n = 95) Control subjects (n = 99) P value Sex (M/F) 54/41 64/35 0.266 Age (years, mean ± SD) 28.2 ± 15.8 29.1 ± 13.9 0.680 Attack once a week or a day (yes/no) 95/95 0/99 <0.000 Sign (wheezing) (yes/no) 95/95 0/99 <0.000 FEV 1 %Pre 58.4 ± 18.9 96.5 ± 9.8 <0.001 IgE (IU/ml, mean ± SD) 486.9 ± 595.5 75.8 ± 124.9 <0.001 IL-4 (ng/L, mean ± SD) 31.7 ± 17.1 5.1 ± 3.8 <0.001 Determination of anti-Hsp60 and anti-Hsp70 Recombinant human Hsp60 and inducible Hsp70 were obtained through the expression of corresponding cDNA in NaCl-induced E. coli GJ1168 cells using pET30 (Novegen) as the expression vector [ 26 ]. Approximately 10–15 μg of recombinant human Hsp60 or Hsp70 was loaded on each SDS-PAGE gel without combs, separated, and transferred by electrophoresis to nitrocellulose membranes. The band containing Hsp60 or Hsp70 was cut into 2 mm × 3 mm pieces and marked with a small red dot on the protein side of the membrane. These membrane pieces were placed in individual wells of an ELISA plate, rinsed with PBS, saturated with 100 μl of blocking buffer (PBS containing 5% skim milk powder) for 1 h at 37°C with gentle agitation and washed with PBS-0.05% Tween 80 for 5 min. The plasma diluted 1:10, 1:20, 1:40 and 1:80 in 100 μl PBS containing 5% skim milk powder was incubated with the membrane pieces at 37°C for 2 h with gentle agitation. After washing the membrane pieces six times (10 min each) with 200 μl PBS-0.05% Tween 80, 100 μl of HRP labelled goat anti-human IgG (Sigma) in blocking buffer (1:2500) was added and the incubation continued at 37°C for 1 h. The membrane pieces were washed again six times (10 min) with 200 μl PBS-0.05% Tween 80. The presence of anti-Hsp60 or anti-Hsp70 was then revealed with DAB (3,3-diaminobenzidine tetra hydrochloride) for 3–5 min. A visible brown band on the membrane piece was regarded as a positive test and a colourless membrane as a negative test [ 21 , 22 ]. An example of the microblot technique is shown in Figure 1 . Samples were scored in a double blind manner by three different investigators. Figure 1 Purified recombinant Hsp70 was electrophoresed in SDS-PAGE, transferred to nitrocellulose membranes and cut into 2 – 3 mm wide strips. These were incubated with the plasma and the presence of antibodies to Hsp70 detected as described in Methods. Lane 1 – 3: positive; Lane 4: negative. Detection of plasma IgE and IL-4 Total IgE was measured in plasma by using a fluorescence enzyme immunoassay kit from Bayer Company (Leverkusen, Germany). IL-4 was determined using a commercial enzyme-linked immunosorbent assay kit from OptEIA (Pharmingen, California, U.S.A). Each sample was tested in duplicate by a series of dilutions using a standard provided with the kit. Statistical analyses All the continuous data (e.g., age, FEV1, IGE, IL4) were presented as the mean ± standard deviation (SD) and analysed by the Student's t test. Frequency data (e.g., sex) were analysed by the Chi-square test. The associations were estimated by fitting univariate and multivariate logistic regression models. Statistical inferences were based on a significance level of 0.05. All analyses were two-sided and performed by using the Statistical Package for Social Sciences (SPSS) software (Version, Chicago). Results The patient and control groups comprised 56.8% and 64.6% of males, respectively, and had mean ages of 28.2 and 29.1 years, respectively (Table 1 ). All patients had regular asthma attacks and sign of wheezing, while none of the controls showed any of these signs. Asthma patients had a significantly lower FEV1%Pre than the controls (58.4 vs 96.5, p < 0.001). In addition, the asthma patients had significantly higher concentrations of total IgE and IL-4 than the controls (p < 0.001 for all comparisons). Presence of anti-Hsp60 and anti-Hsp70 in plasma We first looked for the presence of anti-Hsp60 and anti-Hsp70 in plasma at dilutions of 1:10 to 1:80 in the patients with asthma and in the matched controls. At a dilution of 1:10, asthma patients had a significantly higher positive rate of anti-Hsp60 than the controls (17.9% vs 5.1%, p = 0.001). In the case of Hsp70, antibodies were observed in 33.7% of patients as compared to 8.1% in the controls at a dilution of 1:10 (p < 0.001). At plasma dilutions between 1:20 to 1:80, the difference in the detection rates of both anti-Hsp60 and anti-Hsp70 between the patients and controls remained highly significant (Table 2 ). The combined detection rate of both anti-Hsp60 and anti-Hsp70) at the lower plasma dilution (1:10) was also globally much higher in the asthma patients (38.9%) than in the controls (9.1%). Table 2 Comparison of positive rates of different titers for ant-Hsp60 and anti-Hsp70 in plasma of patients with asthma and healthy controls Titers Patients with asthma (n = 95) Control subjects (n = 99) P value n (%) n (%) Anti-Hsp60 1:10 17 (17.9) 5 (5.1) 0.001 1:20 8 (8.4) 1 (1.0) 0.014 1:40 5 (5.3) 0 (0.0) 0.027 1:80 5 (5.3) 0 (0.0) 0.027 Anti-Hsp70 1:10 32 (33.7) 8 (8.1) <0.001 1:20 19 (20.0) 4 (4.0) <0.001 1:40 13 (13.7) 1 (1.0) <0.001 1:80 9 (9.5) 0 (0.0) 0.001 Anti-Hsps* 1:10 37 (38.9) 9 (9.1) <0.001 * Combined positive rate of anti-Hsp60 and /or anti-Hsp70 for titers 1:10–1:80. Association between anti-Hsp60 and anti-Hsp70 with risk for asthma Further analysis for asthma risk factors (sex, age, family history) and anti-Hsp60 and anti-Hsp70 was carried out by using a multivariate logistic regression model built with a forward stepwise selection procedure (p values for entry and removal, 0.10) and also based on clinical experience. The results in Table 3 show a statistically significant positive association between the presence of anti-Hsp70 and risk for asthma (p = 0.001), representing a greater than 2-fold increased risk for asthma (adjusted OR = 2.21; 95% CI = 1.35~3.59) (Table 3 ). However, no significant association of anti-Hsp60 with risk for asthma was found (p = 0.161). Table 3 Multivariate logistic regression analysis of the association between anti-Hsp60 and anti-Hsp70 with risk for asthma Variables* Adjusted Regression coefficient Standard error χ 2 Value P value OR (95% CI)** Constant -0.977 0.509 3.676 0.055 Sex 0.392 0.318 1.519 0.218 1.48 (0.79~2.76) Age 0.017 0.109 0.026 0.872 1.02 (0.82~1.26) Family history 0.602 0.219 8.215 0.085 2.01 (1.25~3.25) Anti-Hsp60 0.459 0.328 1.965 0.161 1.58 (0.83~3.01) Anti-Hsp70 0.794 0.248 10.224 0.001 2.21 (1.35~3.59) *The dependent variable is the status of asthmas patient or control; the independent variables included Sex: 0 = male and 1 = female; Age: continuous variable in years; Family history: 0 = no and 1 = yes; Anti-Hsp60: 0 = negative and 1 = positive; Anti-Hsp70: 0 = negative and 1 = positive. **OR, odds ratio, and CI, confidence interval. Correlation of anti-Hsp60 and anti-Hsp70 with the severity of asthma To understand the possible significance of the anti-Hsp60 and anti-Hsp70 in asthma, we analyzed the correlation of anti-Hsp60 and anti-Hsp70 with the severity of asthma. Table 4 shows that there was a significant increase of positive rates and dilutions of anti-Hsp60 and anti-Hsp70 as the severity of asthma increased. This table also shows that there were significantly positive correlations of anti-Hsp60 and anti-Hsp70 with the numerical categories of symptom severity (p < 0.05) Table 4 Correlation of anti-Hsp70 and anti-Hsp60 with symptom severities Symptom Severity No. Anti-Hsp70 No. (%) Anti-Hsp60 No. (%) 1:10 1:20 1:40 1:80 1:10 1:20 1:40 1:80 Step1: intermittent 30 2 (6.7) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) Step2: mild persistent 36 10 (27.8) 5 (13.9) 2 (5.6) 1 (2.8) 3 (8.3) 1 (2.8) 1 (2.8) 1 (2.8) Step3&4 moderate & severe persistent** 29 20 (69.0) 14 (48.3) 11 (37.9) 8 27.6) 14 (48.3) 5 (17.2) 4 (14.0) 4 (14.0) R value* 0.809 0.958 0.968 0.959 0.954 0.864 0.947 0.947 P value* 0.000 0.000 0.000 0.001 0.000 0.000 0.016 0.016 * The analyses of correlation of symptom severities with different dilutions of anti-Hsp70 and anti-Hsp60 **: there are two severe persistent patients with asthma Differences in the levels of IgE and IL-4 between asthma patients with positive and negative anti-Hsps We finally compared the levels of IgE and IL-4, two important known risk factors for asthma, in the 95 asthma patients who were either positive (37 patients) or negative (58 patients) for the presence of anti-Hsp60 and anti-Hsp70. The patients positive for anti-Hsps were more likely than the antibody-negative group of patients to report a family history of asthma (48.6% Vs 13.8%, p < 0.001) and had higher concentrations of total IgE (758.2 Vs 313.9, P = 0.001) and IL-4 (36.9 Vs 28.5, p = 0.019) (Table 5 ). Further analysis showed that the presence of either anti-Hsp60 or anti-Hsp70 or both was significantly correlated with the levels of IgE and IL-4 in asthma patients (Table 6 ) (p < 0.05). These preliminary data also indicated a positive correlation between the presence of these autoantibodies and the severity of the disease (r = 0.461, p < 0.001 for anti-hsp60 and r = 0.538, p < 0.001 for anti-Hsp70) (Table 6 ) as well as a statistically significant correlation between anti-Hsp70 and anti-Hsp60 (r = 0.485, p < 0.001) by using the rank correlation analysis. Table 5 Differences in selected risk factors, IgE, and IL-4 between asthma patients with positive and negative anti-Hsps Anti-Hsps(+) (n = 37) Anti-Hsps(-) (n = 58) P value Sex (M/F) 23/14 31/27 0.403 Age (years, mean ± SD) 25.6 ± 15.6 29.3 ± 15.7 0.249 Family history (yes/no) 18/19 8/50 <0.001 IgE (IU/ml, mean ± SD) 758.2 ± 685.3 313.9 ± 458.2 0.001 IL-4 (ng/L, mean ± SD) 36.9 ± 17.2 28.5 ± 16.3 0.019 Table 6 Correlation between anti-Hsps, IgE, and IL-4 in 95 asthma patients IgE IL-4 Anti-Hsp70 Anti-Hsp60 Anti-Hsps r P r P r P r P r P IL-4 0.701 <0.001 Anti-Hsp70 0.369 <0.001 0.222 0.010 Anti-Hsp60 0.262 0.010 0.259 0.011 0.485 <0.001 Anti-Hsps 0.366 <0.001 0.241 0.019 0.534 <0.001 0.814 <0.001 Asthma severity 0.330 0.001 0.236 0.022 0.461 <0.001 0.538 <0.001 0.417 <0.001 Discussion The patients included in the present study had frequent asthmatic attacks, with signs of wheezing and higher levels of IgE and IL4 and low levels of FEV 1 %Pre that are characteristics of asthma. We found that these asthma patients also had a significantly higher incidence of autoantibodies against combined Hsp60 and Hsp70 than the matched non-asthmatic controls and that, in particular, the presence of anti-Hsp70 was associated with asthma. Furthermore, there was a significant positive correlation between anti-Hsp60 and anti-Hsp70 and symptom severity of asthma. Thus among asthma patients, those who had positive anti-Hsp60 and anti-Hsp70 were more likely to report a family history of asthma and had higher levels of IgE and IL-4 than those without such antibodies. These findings provide evidence to support the hypothesis that the presence of anti-Hsp60 and especially anti-Hsp70 in asthma patients is strongly associated with asthma and the presence of these antibodies may predict symptom severity of asthma and provide new strategies for diagnosis and perhaps treatment of this disease. Asthma is an immune and inflammatory disease, arising from complex interactions among genetic and environmental factors including bacterial or viral infection [ 14 , 15 ]. The production of autoantibodies against Hsps may result from genetic factors, infection, denaturation and release of Hsps as a result of cell damage, and the presence of antigen-specific lymphocytes [ 22 , 23 ]. Hsps are often the target of humoral and T cell-mediated immune responses to infection and may provide a link between the immune response to infection and autoimmunity caused by T lymphocyte cross-reactivity among Hsps of different origins [ 8 , 27 , 28 ]. It remains to be determined whether there is a relationship between the induction of Hsp70 and production of plasma autoantibodies against this Hsp and whether there is a cross-response of induced Hsps and autoantibodies against Hsps before and during the development of asthma. However, there are several lines of evidence that support an association between anti-Hsp60 and anti-Hsp70 and symptom severity in asthma patients. Firstly, as molecular chaperones, Hsps facilitate the synthesis, folding, assembly and intracellular trafficking of many functional proteins [ 3 , 29 ] and protect cells and organs against different types of damages [ 30 , 31 ] as observed in transient protection from ischemic injury in whole organs such as heart, brain and kidney [ 31 - 34 ]. Hsp70 has also been suggested to play an autoprotective role in asthma and lung injury [ 35 - 37 ]. Secondly, autoantibodies against Hsps may have significant roles in the pathogenesis and prognosis of diseases. For example, Shinghai et al reported the presence of antibodies against Hsps in patients with autoimmune liver diseases and suggested that the presence of anti-Hsp70 was an indicator for the disease activity of primary biliary cirrhosis [ 8 ]. Earlier results from our lab also suggested that the presence of such antibodies might help assess if workers are experiencing abnormal stress within their living and working environment [ 21 - 23 ]. Xu et al and Schett et al have shown that mycobacterial Hsp65 may serve as an antigen to instigate chronic immune responses characteristic of human atherosclerosis. These antibodies were sustained among patients with the most severe degree of underlying atherosclerosis and were demonstrated to predict 5-year mortality [ 11 , 12 , 20 ]. Thirdly, enhanced expression of Hsp70 has been detected in bronchi and alveolar macrophages of patients with asthma and correlated with intrapulmonary eosinophilia, airway inflammation, hyperresponsiveness of bronchi [ 38 ], and severity of the disease [ 14 , 35 , 39 ]. A cross-response of induced plasma and cellular Hsps and autoantibodies against Hsps in human, has been suggested to play a role in the development and prognosis of atherosclerosis [ 40 , 41 ]. However, it is still unknown whether there is a cross-response between the induction of Hsps in bronchi of patients with asthma and the presence of anti-Hsps and its biological effects. The development of most immune diseases depends on the cytokines interleukin-2 and interferon-γ produced by type 1 helper T cells (Th1), whereas the development of allergic diseases requires IL-4 and IL-5, both of which are produced by type 2 helper T cells (Th2). The reciprocal down-regulation of Th1 cells by Th2 cytokines raises the possibility that these cytokines are involved in allergy or immunity [ 42 ]. IL-4 is one of the first signals for a switch to the synthesis of IgE and IL-4 binds to receptors on B cells to induce and amplify the synthesis of IgE [ 43 ]. There is a cross-linking of IgE with allergens to activate a series of response seen in asthma [ 44 ]. Epidemiological and clinical observations have linked IgE antibodies to the severity of asthma and the initial and sustained responses of the airways to allergens [ 45 , 46 ]. At this time, the molecular events that link antibody to Hsp70 to the production of IL-4 and IgE and the interaction among these factors in patients with asthma remain to be investigated. Conclusions The present study showed that there was a significant increase in positive rates of antibodies against Hsp60 and Hsp70 in patients with asthma and that the presence of autoantibodies against Hsp70 was associated with the severity of asthma. The presence of anti-Hsp70 associated with a high risk of asthma was also correlated with the family history of asthma and higher levels of total IgE and IL-4 in the patients. These results suggest that anti-Hsp70 correlate with the pathogenesis of asthma, but the precise underlying molecular mechanisms for these interactions remain to be established. Authors' contributions MY and LC performed the immunoblot assays, the acquisition of data and wrote the first draft of the manuscript. FW carried out the collection and statistical analysis of data. QW was responsible for the analysis and interpretation of data, and critical revision of the manuscript. TW and RMT initiated the project, designed the experiments, and wrote the manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549531.xml |
340941 | Lefty Blocks a Subset of TGFβ Signals by Antagonizing EGF-CFC Coreceptors | Members of the EGF-CFC family play essential roles in embryonic development and have been implicated in tumorigenesis. The TGFβ signals Nodal and Vg1/GDF1, but not Activin, require EGF-CFC coreceptors to activate Activin receptors. We report that the TGFβ signaling antagonist Lefty also acts through an EGF-CFC-dependent mechanism. Lefty inhibits Nodal and Vg1 signaling, but not Activin signaling. Lefty genetically interacts with EGF-CFC proteins and competes with Nodal for binding to these coreceptors. Chimeras between Activin and Nodal or Vg1 identify a 14 amino acid region that confers independence from EGF-CFC coreceptors and resistance to Lefty. These results indicate that coreceptors are targets for both TGFβ agonists and antagonists and suggest that subtle sequence variations in TGFβ signals result in greater ligand diversity. | Introduction The analysis of whole-genome sequences has revealed that most signaling systems consist of multiple ligands that converge on a relatively small set of receptors and pathway-specific transcription factors. In the case of human transforming growth factor-β (TGFβ) signaling, 42 TGFβs converge on seven type I receptors, five type II receptors, and two classes of Smad signal transducers (reviewed in Shi and Massague 2003 ). This convergence has raised the question of how ligand diversity and signaling specificity among different signals can be achieved. If different TGFβs activate the same receptors, it is unclear how these ligands can vary in their function (diversity) or how a given signal can have a unique role (specificity). Biochemical studies have suggested that ligand diversity can be attained by differential stability and receptor affinity, leading to differences in signaling strength (reviewed in Piek et al. 1999 ; Shi and Massague 2003 ). An additional source of ligand variability stems from differential ligand movement through a field of cells, rendering related signals either short- or long-range ( Chen and Schier 2001 ). Finally, specificity and diversity can also be determined by ligand-specific cofactors or inhibitors ( Piek et al. 1999 ; Shi and Massague 2003 ). A prominent example involves epidermal growth factor–Cripto/FRL-1/Cryptic (EGF-CFC) coreceptors and the TGFβs Activin, Nodal, and Vg1/GDF1 (growth and differentiation factor-1). In this case, differential dependence on a coreceptor leads to ligand diversity and signaling specificity (reviewed in Schier 2003 ). Members of the Nodal, Activin, and Vg1/GDF1 subfamilies display similar activities and are potent mesendoderm inducers in vertebrates (reviewed in Schier and Shen 2000 ). Genetic and biochemical studies have shown that EGF-CFC proteins are essential for signaling by Nodal and Vg1/GDF1 ( Gritsman et al. 1999 ; Reissmann et al. 2001 ; Yeo and Whitman 2001 ; Bianco et al. 2002 ; Sakuma et al. 2002 ; Yan et al. 2002 ; Cheng et al. 2003 ). EGF-CFC proteins are extracellular glycosylphosphatidylinositol (GPI)-linked factors and include One-eyed pinhead (Oep) in zebrafish and mammalian Cripto and Cryptic (reviewed in Shen and Schier 2000 ; Minchiotti et al. 2002 ; Schier 2003 ). Genetic studies in zebrafish and mouse have shown that EGF-CFC proteins and Nodal are required for mesoderm and endoderm induction ( Conlon et al. 1991 , 1994; Zhou et al. 1993 ; Ding et al. 1998 ; Feldman et al. 1998 ; Gritsman et al. 1999 ). For example, zebrafish embryos lacking both the maternal and zygotic contribution of Oep (MZ oep ) lack all endoderm and most mesoderm, similar to the double mutants for the zebrafish Nodal-related genes cyclops and squint ( sqt ) ( Feldman et al. 1998 ; Gritsman et al. 1999 ). Moreover, Nodal and Vg1/GDF1 are inactive in MZ oep mutants ( Gritsman et al. 1999 ; Cheng et al. 2003 ). During later stages of development, Oep, Cryptic, Nodal, and GDF1 are required for proper left–right axis formation ( Gaio et al. 1999 ; Yan et al. 1999 ; Bamford et al. 2000 ; Rankin et al. 2000 ; Brennan et al. 2002 ; Long et al. 2003 ). The EGF-CFC protein Cripto is highly overexpressed in human epithelial cancers, such as breast and colon carcinomas (reviewed in Salomon et al. 2000 ), and has been implicated in tumorigenesis ( Ciardiello et al. 1991 , 1994; Baldassarre et al. 1996 ; De Luca et al. 2000 ; Salomon et al. 2000 ; Adkins et al. 2003 ). The mechanism by which Cripto mediates tumorigenesis is not well understood. Several possibilities include mediating Nodal/GDF1 signaling ( Gritsman et al. 1999 ; Reissmann et al. 2001 ; Yeo and Whitman 2001 ; Bianco et al. 2002 ; Sakuma et al. 2002 ; Yan et al. 2002 ; Cheng et al. 2003 ), antagonizing Activin signaling ( Adkins et al. 2003 ; Gray et al. 2003 ), or activating Akt and mitogen-activated protein kinase (MAPK) pathways independently of the TGFβ signals and Activin receptors ( Ebert et al. 1999; Bianco et al. 2002 , 2003). Whatever the molecular mechanism of Cripto activity, inhibition of Cripto by antisense or antibody blockade can inhibit tumor cell proliferation in vitro and in vivo ( Ciardiello et al. 1994 ; Baldassarre et al. 1996 ; De Luca et al. 2000 ; Adkins et al. 2003 ). Biochemically, EGF-CFC proteins can act as coreceptors for Nodal and Vg1/GDF1 to bind and activate the type I Activin receptor Alk4 and the type II Activin receptor ActRIIB ( Reissmann et al. 2001 ; Yeo and Whitman 2001 ; Sakuma et al. 2002 ; Yan et al. 2002 ; Bianco et al. 2002 ; Cheng et al. 2003 ). In the absence of EGF-CFC proteins, these TGFβs cannot form a complex with Activin receptors. Strikingly, Activin utilizes the same receptors as Nodal and Vg1/GDF1, but does not require EGF-CFC coreceptors ( Mathews and Vale 1991 ; Attisano et al. 1992 , 1996; Hemmati-Brivanlou and Melton 1992 ; Mathews et al. 1992 ; ten Dijke et al. 1994 ; Chang et al. 1997 ). For instance, Activin can signal in MZ oep mutants ( Gritsman et al. 1999 ). This ligand diversity between Activin and Nodal or Vg1/GDF1 raises the question of which sequences confer coreceptor dependence or independence. Activin, Nodal, and Vg1/GDF1 are highly related and are thought to acquire very similar folds. Like other TGFβ ligands, Activin has four major structural features: a β-stranded Finger 1, an α-helical Heel, a β-stranded Finger 2, and three conserved disulfide bonds that form a cysteine knot ( Shi and Massague 2003 ; Thompson et al. 2003 ). Sequence comparisons indicate that the highest divergence among Activin, Nodal, and Vg1/GDF1 is in the N-terminal segment of Finger 1, the central α-helix, and the loop of Finger 2 with approximately 10%, approximately 15%, and approximately 25% sequence identity, respectively. These regions are potential candidates to determine the specificity of receptor–coreceptor–ligand interactions. In addition to coreceptors, antagonists represent another class of extracellular factors that control ligand access to receptors (reviewed in Piek et al. 1999 ; Freeman 2000 ; De Robertis et al. 2001 ; Shi and Massague 2003 ). For example, the divergent TGFβ class of Lefty proteins antagonizes Nodal signaling (reviewed in Hamada et al. 2002 ; Schier 2003 ). Unlike other TGFβs, Lefty proteins may function as monomers due to the lack of a cysteine residue involved in dimerization ( Meno et al. 1996 ; Thisse and Thisse 1999 ; Sakuma et al. 2002 ). Lefty overexpression in zebrafish induces phenotypes identical to cyclops;sqt double mutants and MZ oep mutants ( Bisgrove et al. 1999 ; Meno et al. 1999 ; Thisse and Thisse 1999 ). Furthermore, the loss of Lefty activity leads to enhanced Nodal signaling during mesoderm induction and left–right axis determination ( Meno et al. 1999 , 2001; Agathon et al. 2001 ; Branford and Yost 2002 ; Chen and Schier 2002 ; Feldman et al. 2002 ). Although it has not been determined whether Lefty directly blocks Vg1/GDF1 signaling ( Branford et al. 2000 ), it has been proposed that Lefty inhibits signaling by Activin. Misexpression of Activin or ActRIIB can overcome the inhibitory effects of Lefty ( Meno et al. 1999 ; Thisse and Thisse 1999 ; Cheng et al. 2000 ; Tanegashima et al. 2000 ; Sakuma et al. 2002 ). Hence, some members of the Lefty family have been called Antivins for their anti-Activin properties ( Thisse and Thisse 1999 ; Cheng et al. 2000 ; Ishimaru et al. 2000 ; Tanegashima et al. 2000 ). However, it has been elusive how Lefty functions at the molecular level. Here we present genetic and biochemical studies in zebrafish and Xenopus that indicate that Lefty is an in vivo antagonist of EGF-CFC coreceptors. We find that Lefty can antagonize signaling by the coreceptor-dependent ligands Nodal and Vg1/GDF1, but not Activin. Lefty genetically interacts with Cripto and Oep and competes with Nodal for binding to Cripto, representing a novel mechanism for antagonizing TGFβ signaling. We identify a short region in Finger 2 of Activin, Nodal, and Vg1 that determines EGF-CFC coreceptor-dependent or coreceptor-independent signaling and governs susceptibility to Lefty. These results indicate that subtle sequence variations in TGFβ ligands can dramatically expand signaling diversity by determining interactions with coreceptors and their antagonists. Results Lefty Antagonizes Nodal and Vg1 Signaling, but Not Activin Signaling TGFβ ligands that activate Activin receptors can be categorized into two classes. The Activin class activates Activin receptors in an EGF-CFC coreceptor-independent manner, whereas the Nodal and Vg1/GDF1 classes require EGF-CFC proteins for receptor activation ( Gritsman et al. 1999 ; Cheng et al. 2003 ). To determine whether these classes are also differentially susceptible to inhibition by the TGFβ antagonist Lefty, we coexpressed zebrafish Lefty1 with Xenopus ActivinβB, Xenopus ActivinβA, Sqt (a zebrafish Nodal-related protein), or zebrafish Vg1 (a chimeric molecule containing the Xenopus ActivinβB prodomain fused to the mature domain of zebrafish Vg1) in zebrafish embryos ( Smith et al. 1990 ; Thomsen et al. 1990 ; van den Eijnden-Van Raaij et al. 1990 ; Helde and Grunwald 1993 ; Erter et al. 1998; Feldman et al. 1998 ; Thisse and Thisse 1999 ). As a readout for active signaling, we analyzed the ectopic induction of the Nodal downstream gene goosecoid ( gsc ). ActivinβB, ActivinβA, Sqt, and Vg1-induced ectopic gsc expression in wild-type embryos ( Figure 1 D, 1 G, 1 J, and 1 M; Gritsman et al. 1999 ; Cheng et al. 2003 ). Coexpression of Lefty1 efficiently inhibited gsc induction by Sqt ( Figure 1 K and 1 L; Bisgrove et al. 1999 ; Meno et al. 1999 ; Thisse et al. 2000 ) and Vg1 ( Figure 1 N and 1 O), but not ActivinβB or ActivinβA ( Figure 1 E, 1 F, 1 H, and 1 I). To examine whether Lefty1 can antagonize the induction of a gene that responds to very low levels of Activin signaling, we titrated ActivinβB levels so that no tail ( ntl ; also known as brachyury/T ) expression was only weakly induced (see arrowhead in Figure 1 A). The coexpression of Lefty1 did not inhibit ntl induction by ActivinβB ( Figure 1 B and 1 C), but inhibited the dorsal margin expression of ntl that is dependent on endogenous Nodal signaling (see asterisks in Figure 1 B and 1 C; Feldman et al. 1998 ). In a more quantitative assay, we overexpressed Lefty1, ActivinβB, Sqt, and Vg1 in zebrafish embryos in the presence of the luciferase reporter A3-luc, which contains FoxH1/P-Smad2 response elements ( Chen et al. 1996 ). Consistent with the gsc and ntl induction assays, Sqt and Vg1 signaling, but not ActivinβB signaling, is inhibited by Lefty1 ( Figure 1 P). These results indicate that Lefty1 efficiently antagonizes Nodal and Vg1/GDF1 signaling, but not Activin signaling. Figure 1 Lefty Antagonizes Nodal and Vg1 Signaling, but Not Activin Signaling, in Zebrafish ntl mRNA expression (A–C) and gsc mRNA expression (D–O) in wild-type zebrafish embryos at shield stage, animal pole view. Embryos were injected with low levels (1 pg) of activin βB mRNA (A–C), high levels (10 pg) of activin βB mRNA (D–F), 200 pg of activin βA mRNA (G–I), 75 pg of sqt mRNA (J–L), or 200 pg of Vg1 (M–O). Embryos were further double-injected with either 500 pg of LacZ mRNA (A, D, G, J, and M), 100 pg of lefty1 and 400 pg of LacZ mRNAs (B, E, H, K, and N), or 500 pg of lefty1 mRNA (C, F, I, L, and O). Ectopic ntl expression (arrowheads) in activin βB mRNA-injected embryos was not inhibited by Lefty1 (B and C) when compared with LacZ mRNA-coinjected controls (A). Note the dorsal expression of ntl (asterisks)—that is, dependent on endogenous Nodal signaling—is inhibited by Lefty1 in these embryos (B and C). Ectopic gsc expression in activin βB and activin βA mRNA-injected embryos was not inhibited by Lefty1 (E and F and H and I, respectively). In contrast, ectopic gsc expression in sqt and Vg1 mRNA-injected embryos was inhibited by both levels of Lefty expression (K and L and N and O, respectively). Wild-type embryos (P) were injected with 10 pg (low) and 20 pg (high) of activin βB /HA, 75 pg of sqt , or 200 pg of Vg1 mRNA. Embryos were further double-injected with 500 pg of LacZ mRNA, 100 pg of lefty1 , and 400 pg of LacZ mRNAs, or 500 pg of lefty1 mRNA. Smad2 pathway activation was measured by an Activin response element luciferase reporter, A3-luc. Values are folds over wild-type control injected with 500 pg of LacZ mRNA and A3-luc reporter. An asterisk indicates a significant difference from the level of activation with ligand and LacZ expression alone (Student's t -test, p < 0.05). EGF-CFC Proteins Genetically Interact with Lefty The molecular mechanism of Lefty action has been unresolved. Lefty seems to act upstream of the Activin receptors, as Lefty cannot block signaling from ligand-independent constitutively activated receptors ( Thisse and Thisse 1999 ). Our finding that Lefty blocks Nodal and Vg1 signaling, but not Activin signaling, suggests that Lefty blocks extracellular components specific to the Nodal and Vg1 pathways. The only such factors identified to date are the EGF-CFC coreceptors. We therefore examined whether the EGF-CFC genes zebrafish oep and mouse Cripto genetically interact with Lefty1. Overexpression of Lefty1 in wild-type zebrafish resulted in embryos lacking head and trunk mesendoderm due to inhibition of endogenous Nodal signaling ( Figure 2 A1 and 2 A′; Bisgrove et al. 1999 ; Meno et al. 1999 ; Thisse and Thisse 1999 ). Coexpression of Cripto or Oep partially suppressed Lefty-induced defects ( Figure 2 B1– 2 B3; data not shown), restoring trunk and head mesoderm, including the notochord, and resulting in the separation of the eye field into two eyes. These results indicate an antagonistic relationship between EGF-CFC coreceptors and Leftys. Figure 2 EGF-CFC Coreceptors Genetically Interact with Lefty Live wild-type zebrafish embryos at 30 h postfertilization (hpf). (A1, B1, and C1) Ventral views of the head. (A1′, B2, B3, and C1′) Lateral views, with anterior to the left, dorsal up. (A1, A1′, B1, B2, and B3) Wild-type embryos were injected with 20 pg of lefty1 mRNA. Embryos were further double-injected with either 200 pg of LacZ mRNA (A1 and A1′) or 200 pg of Cripto mRNA (B1, B2, and B3). (C1 and C1′) Wild-type embryos injected with 200 pg of Cripto mRNA and 20 pg of LacZ mRNA. Misexpression of Lefty1 results in cyclopia and other head and trunk mesoderm defects ([A1 and A1′] 32 of 32 embryos had the phenotype shown; arrow shows cyclopia). Coexpression of Cripto with Lefty in embryos leads to rescue of two eyes ([B1] four of 50; arrows show two eyes), notochord ([B2] 20 of 50; inset shows trunk somites and notochord, red bar delineates notochord), and trunk somites ([B3] 50 of 50). Embryos injected with Cripto mRNA only show normal wild-type phenotype ([C1 and C1′] 30 of 30; arrow in [C1] shows two normal eyes, and inset in [C1′] shows normal notochord and trunk somites, red bar delineates notochord). Lefty Binds to Cripto, but Not to ActRIIB or Alk4 Because Lefty and EGF-CFC proteins interact genetically, we examined whether Lefty interacts biochemically with Cripto and/or ActRIIB/Alk4 receptor complexes. We expressed and immunoprecipitated epitope-tagged ligands (zebrafish Lefty1/HA [hemagglutinin], zebrafish Lefty1/Glu, or zebrafish Sqt/HA), receptors (mouse ActRIIB[KR]/Myc and human Alk4[KR]/Flag), and a coreceptor (mouse Cripto/Flag) in Xenopus embryos ( Yeo and Whitman 2001 ; Cheng et al. 2003 ). Similar to other Nodal-related proteins ( Reissmann et al. 2001 ; Yeo and Whitman 2001 ; Bianco et al. 2002 ; Sakuma et al. 2002 ; Yan et al. 2002 ), Sqt formed a complex with the type II receptor ActRIIB, type I receptor Alk4, and Cripto ( Figure 3 A). In contrast, Lefty1 coimmunoprecipitated Cripto, but not ActRIIB or Alk4 ( Figure 3 A). Since Cripto is bound to Alk4 even in the absence of ligand ( Reissmann et al. 2001 ; Yeo and Whitman 2001 ; Bianco et al. 2002 ; Yan et al. 2002 ), Lefty seemed to disrupt the Cripto–Alk4 interaction. In reverse experiments, Cripto efficiently coimmunoprecipitated Lefty1 ( Figure 3 B). Since Sqt and Lefty1 can both bind to Cripto ( Figure 3 C; Reissmann et al. 2001 ; Yeo and Whitman 2001 ; Bianco et al. 2002 ; Sakuma et al. 2002 ; Yan et al. 2002 ), these two ligands might compete for binding to Cripto. Indeed, the coexpression of Lefty1 led to decreased interactions between Cripto and Sqt ( Figure 3 C). To determine whether Cripto can directly interact with Lefty, we immunoprecipitated purified mouse Lefty1 protein (mLefty1) in the presence of either purified mouse Cripto protein or a control cysteine-rich protein, mouse vascular endothelial growth factor-D (VEGF-D). mLefty1 protein directly interacted with Cripto, but not with VEGF-D. Together, these results suggest that Lefty inhibits Nodal signaling by associating with Cripto and blocking it from interacting with Nodal. Figure 3 Lefty Binds to Cripto, but Not to the Activin Receptors ActRIIB and Alk4 (A and B) Lefty1 interacts with Cripto. RNAs (1 ng each) encoding ActRIIB(KR)/Myc, Alk4(KR)/Flag, Cripto/Flag, Lefty1/HA, or Sqt/HA were injected into Xenopus embryos. After chemical cross-linking, lysates were immunoprecipitated for either Lefty1/HA or Sqt/HA (A) with anti-HA antibody, or ActRIIB(KR)/Myc, Alk4(KR)/HA, Cripto/Flag (B) with, respectively, anti-Myc, anti-HA, or anti-Flag antibodies. Note that Lefty1 specifically interacts with Cripto (A and B), and these Lefty/Cripto complexes do not contain Alk4 (A). Moreover, processed Lefty1 binds much more efficiently to Cripto than full-length Lefty1 precursor (B). In contrast, Sqt can bind to ActRIIB, Alk4, and Cripto (A). The 55 kDa protein marker in (B) is estimated based on molecular weight markers. (C) Lefty1 competes with Nodal for binding to Cripto. RNAs encoding Sqt/HA (1 ng), Cripto/Flag (100 pg), or Lefty1 (2 ng) were injected and anti-Flag antibody was used to immunopreciptate Cripto/Flag. (D) mLefty1 binds directly to Cripto. Purified mouse Lefty1 protein (mLefty1; 10 μg/ml) was mixed with either soluble purified Cripto/His protein (5 μg/ml) or purified control VEGF-D/His protein (5 μg/ml). After chemical cross-linking, mLefty1 was immunoprecipitated with anti-mLefty1 antibody. mLefty1 associates with Cripto, but not with control VEGF-D. Proteins in the coimmunoprecipitates and total extracts were probed in Western blot analysis with the indicated antibodies: ActRIIB(KR)/Myc (kinase-defective receptor, approximately 120 kDa; anti-Myc), Alk4(KR)/Flag (kinase-defective receptor, approximately 70 kDa; anti-Flag), Cripto/Flag (approximately 30 kDa; anti-Flag), Lefty1/HA (mature ligand, approximately 36–40 kDa; anti-HA; Sakuma et al. 2002 ), Sqt/HA (unprocessed precursor, approximately 55 kDa; mature ligand, approximately 22 kDa; anti-HA), Lefty1/Glu (unprocessed precursor, approximately 55 kDa; mature ligand, approximately 38 kDa; anti-Glu; Sakuma et al. 2002 ), mLefty1 (mature ligand, approximately 36 kDa, anti-mLefty1; Sakuma et al. 2002 ), Cripto/His (soluble form, approximately 22–25 kDa; anti-His), and VEGF-D/His (mature ligand, approximately 15–20 kDa; anti-His). Activin Loop-β8 Region Confers EGF-CFC Coreceptor Independence to Sqt The finding that TGFβ ligands that activate Activin receptors can be grouped into a EGF-CFC coreceptor-dependent class that is susceptible to inhibition by Lefty (Nodal and Vg1) and a class that is independent of EGF-CFC proteins and resistant to Lefty (Activin) prompted us to examine which sequences underlie this ligand diversity. We therefore generated chimeras between Activins (EGF-CFC-independent) and Sqt or Vg1 (EGF-CFC-dependent) (Figures 4 and 5 ). As a readout for active signaling, we injected mRNAs encoding these chimeric ligands into wild-type and MZo ep zebrafish embryos and analyzed the ectopic induction of the downstream genes ntl and gsc . Sqt, Vg1, and Activins induced these genes in wild-type embryos, allowing us to determine which chimeric ligands were active. Activins, but not Sqt or Vg1, were active in MZ oep mutants, allowing us to test which sequences conferred EGF-CFC coreceptor dependence or independence. Figure 4 Chimera Analysis to Identify TGFβ Sequence Determinants Conferring EGF-CFC Coreceptor Dependence or Independence Schematic depiction of chimeras of mature ligand domains, Finger 1 (F1), Heel (H), and Finger 2 (F2), between Xenopus ActivinβB and zebrafish Sqt. HA indicates an hemagglutinin epitope tag. Schematic is not drawn to scale. The letters in these three-lettered (XXX) chimeras represent the Finger 1, Heel, and Finger 2, respectively. S denotes Squint; A denotes ActivinβB. Synthetic mRNAs (200 pg) encoding chimeras were injected into wild-type and MZ oep embryos. gsc and ntl mRNA expression is shown at shield stage; animal pole views are dorsal to the right. gsc is expressed in the dorsal organizer (shield) in wild-type embryos, but is absent in MZ oep mutants. ntl is expressed around the entire margin in wild-type embryos, but the dorsal margin expression is lost in MZ oep mutants. The presence of the ActivinβB prodomain and epitope tag does not alter the specificity or functionality of wild-type ActivinβB (AAA) or Sqt (SSS). AAA can induce ectopic gsc and ntl expression in both wild-type and MZ oep embryos. In contrast, SSS can induce ectopic gsc and ntl expression in only wild-type embryos. Similar to ActivinβB, chimeras SSA, SAS, ASA, and SAA can induce ectopic gsc and ntl expression in both wild-type and MZ oep embryos. Chimeras ASS and AAS are inactive in both wild-type and MZ oep embryos. Western blot analysis indicated that all chimeric constructs produce stable ligands (data not shown). Figure 5 Sequence Determinants Conferring Independence from EGF-CFC Coreceptors (A) Sequence alignment of Finger 2 region of EGF-CFC-dependent and EGF-CFC-independent TGFβ ligands. Location of secondary structure elements, β-sheets (β6–β9) and loop, are shown ( Kirsch et al. 2000 ). Residue numbering is from mouse ActivinβA. (B–E) Synthetic mRNAs (200 pg) encoding chimeras of Finger 2 subregions between Xenopus ActivinβB or ActivinβA and zebrafish Sqt or Vg1 were injected into wild-type and MZ oep embryos. Schematic is not drawn to scale. gsc and ntl mRNA expression is at shield stage; animal pole views are dorsal to the right. (B) Sqt Act β B[loop β 8 β 9] and Sqt Act β B[loop β 8] can induce gsc and ntl expression in both wild-type and MZ oep embryos. (C) Sqt Act β B[ β 8] can weakly expand ntl expression in MZ oep mutants. ntl mRNA expression in MZ oep mutants is at shield stage; lateral view. (D) Other TGFβs conform to loop-β8 EGF-CFC-independent determinant. Note that Xenopus ActivinβA can induce ectopic gsc in both wild-type and MZ oep embryos. In contrast, Vg1 can only induce gsc in wild-type embryos. Similar to Activins, chimeric Sqt Act β A[loop β 8] and Vg1 Act β B[loop β 8] can induce ectopic gsc in both wild-type and MZ oep embryos. (E) Wild-type and MZoep embryos were injected with 5 pg of activin βB , 100 pg of sqt , 100 pg of Vg1 , 125 pg of Sqt ActβB[loopβ8] , 250 pg of Sqt ActβA[loopβ8] , or 100 pg of Vg1 ActβB[loopβ8] mRNA. Smad2 pathway activation was measured by an Activin response element luciferase reporter, A3-luc. Luciferase units are relative to wild-type or MZ oep control injected with the A3-luc reporter alone. (F) Sqt Act β B[loop β 8] can bind to ActRIIB and Alk4 in the absence of EGF-CFC coreceptors. RNAs (1 ng each) encoding ActRIIB(KR)/Myc, Alk4(KR)/Flag, Cripto/Flag, ActivinβB/HA, Sqt/HA, or Sqt Act β B[loop β 8] /HA were injected into Xenopus embryos. Proteins in the coimmunoprecipitates and total extracts were probed in Western blot analysis with the indicated antibodies: ActRIIB(KR)/Myc (approximately 120 kDa; anti-Myc), Alk4(KR)/Flag (approximately 70 kDa; anti-Flag), Cripto/Flag (approximately 30 kDa; anti-Flag), ActivinβB/HA (mature ligand, approximately 16 kDa; anti-HA), Sqt/HA (mature ligand, approximately 22 kDa; anti-HA), and Sqt Act β B[loop β 8] /HA (mature ligand, approximately 22 kDa; anti-HA). Initially, swaps of the Finger 1, Heel, or Finger 2 domains of Sqt and ActivinβB were generated. As shown in Figure 4 , the Finger 2 region of ActivinβB contains sequence determinants that conferred EGF-CFC-independent activity on chimeric ligands. Chimeric SSA, ASA, and SAA that contain the Finger 2 region of ActivinβB were active in both wild-type and MZ oep embryos ( Figure 4 ). To further delineate this region, we generated additional chimeras ( Figure 5 B). Short stretches of full-length Sqt were replaced by the corresponding region of ActivinβB, including the β6β7, loop, β8, or β9 subregions ( Figure 5 A and 5 B; data not shown). Analysis of these chimeras revealed that the 14 amino acids encoding the loop and β8 region of ActivinβB (Sqt Act β B[loop β 8] ; the bracketed region in superscript denotes substituted domains) were sufficient to confer EGF-CFC independence. Further dissection of this region into loop alone (Sqt Act β B[loop] ) or β8 alone (Sqt Act β B[ β 8] ) yielded no or much weaker activity in MZ oep mutants as compared with wild-type embryos ( Figure 5 B and 5 C). These results suggest that a 14 amino acid region in Activin is sufficient to confer EGF-CFC independence when placed into Sqt. Activin Loop-β8 Region Confers EGF-CFC Coreceptor Independence to Vg1 To determine whether the loop-β8 region has a wider role in conferring coreceptor independence, we generated additional chimeras using ActivinβA (another EGF-CFC-independent ligand) and Vg1. Sqt Act β A[loop β 8] (full-length Sqt with an ActivinβA loop-β8 region) and Vg1 Act β B[loop β 8] (Vg1 with an ActivinβB loop-β8 region) both induced gsc expression in MZ oep mutants with similar efficiencies as in wild-type embryos ( Figure 5 D). These results were also corroborated using the A3-luc reporter assay ( Figure 5 E) and suggest that the loop-β8 region has a general role in conferring EGF-CFC coreceptor independence. Activin Loop-β8 Region Confers Binding to Activin Receptors in the Absence of EGF-CFC Coreceptors Sqt Act β B[loop β 8] can signal in an EGF-CFC-independent manner in vivo, suggesting that this chimeric protein might bind to ActRIIB and Alk4 receptors in the absence of EGF-CFC coreceptors. To test this idea, we coexpressed and immunoprecipitated epitope-tagged ligands (ActivinβB/HA, Sqt/HA, Sqt Act β B[loop β 8] /HA), receptors (ActRIIB[KR]/Myc and Alk4[KR]/Flag), and a coreceptor (Cripto/Flag) in Xenopus embryos ( Yeo and Whitman 2001 ; Cheng et al. 2003 ). We found that Sqt binding to the ActRIIB and Alk4 receptor complex required Cripto ( Figure 5 F). In contrast, ActivinβB and Sqt Act β B[loop β 8] can bind to Activin receptors in the absence of Cripto. Moreover, Cripto did not significantly increase Sqt Act β B[loop β 8] ligand–receptor complex formation. These results indicate that the loop-β8 region is a determinant of TGFβ ligand binding to Activin receptors independent of EGF-CFC coreceptors. Multiple Residues in the Loop-β8 Region Contribute to Coreceptor Independence An alignment of the loop-β8 region of EGF-CFC-dependent and EGF-CFC-independent TGFβs ( Figure 5 A) reveals the presence of several residues unique to Activins. These include (i) a Lys102–X–Asp104 motif (numbering from ActivinβA) that forms a significant binding interface with the type II receptor ActRII ( Wuytens et al. 1999 ; Greenwald et al. 2003 ; Thompson et al. 2003 ); (ii) Gln/Pro106 and Asn107, which contribute to the dimerization interface responsible for conformational arrangement ( Thompson et al. 2003 ); and (iii) an Asn insertion at position 99. We therefore mutated the corresponding residues in Sqt, individually or in combination, to the ActivinβB sequence and tested them in the gsc / ntl induction assay ( Figure 6 ). All constructs were active in wild-type embryos. The Sqt3 and Sqt5 constructs containing the Lys102–X–Asp104 motif and Asn99 insertion showed weak expansion of ntl expression animally and dorsally in MZ oep mutants. The incorporation of Pro106–Asn107 (Sqt2, Sqt4, and Sqt5) in Sqt did not enhance activity in MZ oep mutants. These results suggest that multiple residues contribute to coreceptor independence, with the type II receptor-binding interface being an essential determinant. Figure 6 Conserved Residues in Activin Loop-β8 Region Confer Independence from EGF-CFC Coreceptors Synthetic mRNAs (200 pg) encoding Sqt harboring multiple mutations from ActivinβB (shown in red) were injected into wild-type and MZ oep embryos. gsc and ntl mRNA expression is shown at shield stage; animal pole views are dorsal to the right. Schematic is not drawn to scale. Note that the Sqt3 and Sqt5 constructs containing the Lys102–X–Asp104 motif and Asn99 insertion show weak expansion of ntl expression animally and dorsally in MZ oep mutants. The Loop-β8 Region in Sqt Is Inhibitory The results described above identified the loop-β8 region of Activin as a region that confers coreceptor-independent signaling to ligands that are normally EGF-CFC-dependent. In a reverse set of experiments, we asked which regions confer dependence on EGF-CFC coreceptors. To identify domains that confer EGF-CFC dependence, chimeras between ActivinβB and Sqt (see Figures 4 and 7 ) were analyzed for their inability to signal in MZ oep mutants. Chimeras containing the Sqt Finger 2 domain (AAS and ASS; see Figure 4 ) or only the Sqt loop-β8 region (Act Sqt[loop β 8] ; Figure 7 ) were inactive in both wild-type and MZ oep embryos. Western blot analysis demonstrated that these chimeras generate stable ligands (data not shown). The addition of Finger 1 in SAS or Act Sqt[Finger1-loop β 8] relieved the inhibitory effect of the loop-β8 region of Sqt in wild-type embryos ( Figure 7 ). These chimeras were inactive in MZ oep mutants. These results indicate that the loop-β8 region in Sqt acts as an inhibitory domain and that Finger 1 relieves this inhibition by conferring dependence on EGF-CFC coreceptors. Figure 7 Sequence Determinants Conferring EGF-CFC Dependence Synthetic mRNAs (200 pg) encoding ActivinβB with single or double region substitutions from Sqt were injected into wild-type and MZ oep embryos. gsc and ntl mRNA expression is shown at shield stage; animal pole views are dorsal to the right. Schematic is not drawn to scale. HA indicates a hemagglutinin epitope tag. Note that Act Sqt[loop β 8] containing the loop-β8 region of Sqt is inactive in both wild-type and MZ oep embryos. In Act Sqt[Finger1-loop β 8] , the additional substitution of Sqt Finger 1 region relieves the inhibitory presence of the Sqt loop-β8 region. Similar to Sqt, Act Sqt[Finger1-loop β 8] can induce ectopic gsc and ntl in wild-type, but not in MZ oep embryos. Western blot analysis indicates that these chimeric constructs produce stable ligands (data not shown). Specificity of Antagonism by Lefty Is Determined by EGF-CFC Coreceptor Dependence Our genetic and biochemical studies suggested that Lefty blocks Nodal and Vg1 signaling via EGF-CFC coreceptors. In contrast, the coreceptor-independent signaling by Activins cannot be blocked by Lefty. This finding predicts that the EGF-CFC-independent chimeric ligands Sqt Act β B[loop β 8] , Sqt Act β A[loop β 8] , and Vg1 Act β B[loop β 8] should also be resistant to Lefty. Conversely, the coreceptor-dependent chimera Act Sqt[Finger1-loop β 8] should be suspectible to inhibition by Lefty. To test this hypothesis, we coexpressed chimeric ligands and Lefty1 and analyzed gsc expression and A3-luc reporter induction ( Figure 8 A– 8 M). As predicted, Lefty1 did not inhibit signaling by Sqt Act β B[loop β 8] ( Figure 8 B and 8 C), Sqt Act β A[loop β 8] ( Figure 8 E and 8 F), or Vg1 Act β B[loop β 8] ( Figure 8 H and 8 I), but antagonized Act Sqt[Finger1-loop β 8] ( Figure 8 K and 8 L). These results indicate that the incorporation of the Activin loop-β8 region into Nodal and Vg1 can render these ligands EGF-CFC-independent and therefore resistant to Lefty. Figure 8 EGF-CFC Coreceptor Depen-dence Determines Susceptibility to Antagonism by Lefty (A–L) Embryos were injected with 75 pg of Sqt ActβB[loopβ8] mRNA (A–C), 75 pg of Sqt ActβA[ loop β 8 ] mRNA (D–F), 200 pg of Vg1 ActβB[loopβ8] mRNA (G–I), or 200 pg Act Sqt[Finger1-loopβ8] mRNA (J–L). Embryos were further double-injected with either 500 pg of LacZ mRNA (A, D, G, and J), 100 pg of lefty1 and 400 pg LacZ mRNAs (B, E, H, and K), or 500 pg of lefty1 mRNA (C, F, I, and L). gsc mRNA expression in wild-type zebrafish embryos is shown at shield stage, animal pole view. Note that both levels of Lefty1 cannot inhibit the ectopic gsc expression induced by Sqt Act β B[loop β 8] (B and C), Sqt Act β A[loop β 8] (E and F), and Vg1 Act β B[loop β 8] (H and I). In contrast, Lefty1 can inhibit Act Sqt[Finger1-loop β 8] (K and L). (M) Wild-type embryos were injected with 75 pg of either Sqt Act β B[loop β 8] , Sqt Act β A[loop β 8] , Vg1 Act β B[loop β 8] , or 200 pg of Act Sqt[Finger1-loop β 8] mRNA. Embryos were further double-injected with 500 pg of LacZ mRNA, 100 pg of lefty1 , and 400 pg of LacZ mRNAs, or 500 pg of lefty1 mRNA. Smad2 pathway activation was measured by an Activin response element luciferase reporter, A3-luc. Values are folds over wild-type control injected with 500 pg of LacZ mRNA and A3-luc reporter. An asterisk indicates a significant difference from the level of activation with ligand and LacZ expression alone (Student's t -test, p < 0.05). Discussion Lefty Antagonizes EGF-CFC Coreceptors Lefty molecules are key regulators of mesendoderm development and left–right axis determination, but the molecular basis of Lefty-mediated antagonism of Activin-like pathways has been elusive ( Hamada et al. 2002 ; Schier 2003 ). Our genetic and biochemical studies provide three lines of evidence that Lefty blocks EGF-CFC coreceptors. First, Lefty only inhibits EGF-CFC-dependent TGFβ ligands such as Nodal and Vg1, but not EGF-CFC-independent ligands such as Activins. A striking example of this coreceptor-specific interaction is the finding that changing only 14 amino acids in Nodal or Vg1 to the corresponding residues in Activins renders the resulting TGFβ ligands independent of EGF-CFC coreceptors and resistant to Lefty. Second, the EGF-CFC proteins mouse Cripto and zebrafish Oep can partially suppress the effects of Lefty overexpression in zebrafish. Third, Leftys can bind to EGF-CFC coreceptors and block the coreceptors from interacting with Nodal. Furthermore, Lefty/EGF-CFC complexes seem to exclude interactions with type I and type II Activin receptors. Taken together, these results indicate that Lefty blocks a subset of TGFβ signals by the novel mechanism of blocking pathway-specific coreceptors ( Figure 9 A– 9 D). Figure 9 Model for EGF-CFC, Activin Receptors, Lefty, and TGFβ Interactions (A) In the absence of ligands, the EGF-CFC coreceptor (solid pink) is constitutively bound to the type I receptor Alk4 (solid green). (B) Nodal (solid blue) binds to receptor complexes consisting of EGF-CFC/Alk4 and ActRIIB (solid green). (C) Lefty (solid yellow) sequesters the EGF-CFC coreceptor, thereby preventing Nodal binding to the receptor complexes. (D) Subtle sequence differences determine the interaction with the EGF-CFC coreceptor and the Lefty inhibitor. Nodal and Vg1/GDF1 (solid blue) require the EGF-CFC coreceptor for signaling through ActRIIB and Alk4, while Activin (solid red) does not. Sqt Act β B[loop β 8] and Vg1 Act β B[loop β 8] (solid blue with red strip) containing the loop-β8 region of ActivinβB can bind to ActRIIB and Alk4 without the EGF-CFC coreceptor and therefore cannot be blocked by Lefty. Act Sqt[Finger1-loop β 8] (solid red with two blue strips) requires the coreceptor for receptor complex binding and can be inhibited by Lefty. The observation that Lefty does not block signaling by Activin seems in apparent contrast to previous studies that led to naming some Lefty family members Antivins, for their anti-Activin properties ( Thisse and Thisse 1999 ; Cheng et al. 2000 ; Ishimaru et al. 2000 ; Tanegashima et al. 2000 ). In particular, it has been found that misexpression of Activin can suppress the defects caused by Lefty misexpression in vivo ( Thisse and Thisse 1999 ). Our results do not undermine this finding, but suggest an alternative explanation. Previous studies have shown that Activin can suppress the loss of EGF-CFC activity in MZ oep mutants ( Gritsman et al. 1999 ; Cheng et al. 2003 ). Analogously, we suggest that the blocking of EGF-CFC activity by Lefty can be bypassed by Activin, because this ligand can activate Activin receptors independent of co-receptors. A similar scenario can also account for the suppression of Lefty gain-of-function phenotypes by misexpression and activation of Activin receptors ( Meno et al. 1999 ; Thisse and Thisse 1999 ; Sakuma et al. 2002 ). Hence, Activin and Activin receptors bypass the loss of EGF-CFC coreceptor function that is caused either by mutations in oep or by overexpression of Lefty. Conversely, Lefty cannot block Activin signals and Activin receptors because of their independence from EGF-CFC coreceptors. Is the block of EGF-CFC coreceptors by Lefty a general and conserved mechanism? Although we have only analyzed a representative subset of these protein families (zebrafish and mouse Lefty1; zebrafish Oep; mouse Cripto; zebrafish Sqt), previous studies have suggested that heterologous Nodal, Lefty, and EGF-CFC proteins have similar activities in zebrafish ( Schier 2003 ). For example, mouse Nodal, mouse Lefty2, and mouse Cripto are active in zebrafish, despite less than 30% overall sequence conservation with their zebrafish counterparts ( Toyama et al. 1995 ; Meno et al. 1999 ; Gritsman et al. 1999 ). These studies suggest that the molecular mechanisms described here apply to most, if not all, Nodal/Lefty/EGF-CFC interactions. This does not exclude the possibility that Lefty has additional means of blocking TGFβ signaling. First, Leftys might block the processing of Nodals. However, Sqt is processed normally at levels of Lefty that block Nodal signaling (unpublished data). Second, Leftys might bind Nodal signals. This could result in blocking receptor interactions or antagonizing TGFβ dimerization. However, Sqt is not bound by Lefty at Lefty levels that block Nodal signaling and lead to complex formation with Cripto (unpublished data). Moreover, a Sqt protein containing the loop-β8 region of Activin is resistant to Lefty, whereas changing only the dimerization residues in this region does not confer resistance. Third, Leftys might interact with additional extracellular factors. Indeed, the overexpression of the extracellular domain of the type II receptor ActRIIB has been shown to suppress Lefty activity ( Meno et al. 1999 ). Although zebrafish Lefty1 does not appear to bind to ActRIIB, it is conceivable that overexpression of soluble ActRIIB might protect EGF-CFC coreceptors or another yet-to-be identified protein from antagonism by Lefty. In addition, overexpression of EGF-CFC proteins in zebrafish does not induce dominant phenotypes ( Gritsman et al. 1999 ). It is therefore possible that an additional factor would be required to completely block Lefty in these experiments. Alternatively, overexpression levels of EGF-CFC coreceptors might not be high enough to block Lefty at blastula stages. It is also possible that coreceptor overexpression might block Nodal signals, because it has been shown that EGF-CFC proteins and Nodals can directly interact. The complex feedback interactions between Lefty and Nodal might also overcome an initial reduction of Lefty activity by increasing Lefty transcription. These considerations and the data presented here therefore suggest that a major, but perhaps not exclusive, role of Leftys is to block a subset of TGFβ signals by interaction with EGF-CFC coreceptors. Implications for the Role of Lefty during Development Our finding that Leftys can block Vg1 signaling also has important implications for the developmental control of TGFβ signaling. Based on previous studies revealing that Lefty proteins inhibit Nodal signaling, the mouse Lefty mutant phenotypes have been interpreted as a consequence of increased or sustained Nodal signaling ( Hamada et al. 2002 ; Schier 2003 ). For example, loss of mouse Lefty2 has been thought to increase Nodal signaling, resulting in an enlarged primitive streak ( Meno et al. 1999 ). Similarly, the left–right defects observed in mouse Lefty1 and left-side specific Lefty2 ( Lefty2 ΔASE ) mutants have been thought to be caused by inappropriate spread of Nodal signaling ( Meno et al. 1998 , 2001). Our finding that Vg1/GDF1 signaling can also be blocked by Lefty suggests a more complex scenario. In particular, GDF1 (the mouse homologue of Vg1) is required for proper left–right axis determination ( Rankin et al. 2000 ). GDF1 mutants appear to have the opposite phenotypes as Lefty1 and Lefty2 ΔASE mutants. While GDF1 promotes the expression of left-side-specific genes such as Pitx2 on the left, Leftys appear to block Pitx2 expression on the right ( Meno et al. 1998 , 2001; Rankin et al. 2000 ). In light of our findings, we suggest that during left–right axis formation, Leftys act as antagonists of not only Nodal, but also GDF1. In this scenario, loss of Lefty1 or Lefty2 would lead to ectopic and sustained GDF1 signaling. This model is particularly attractive when one considers the expression patterns of Lefty1 , Lefty2 , the EGF-CFC gene Cryptic , Nodal , and GDF1. Lefty 1 and GDF1 are coexpressed in the developing midline ( Meno et al. 1996 , 1997; Rankin et al. 2000 ), whereas Lefty2 and Nodal are coexpressed in left-lateral plate mesoderm ( Conlon et al. 1994 ; Meno et al. 1997 ). Cryptic is expressed in both the lateral plate and midline ( Shen et al. 1997 ). It is therefore conceivable that GDF1 signaling is restricted by Lefty1-mediated inhibition of Cryptic in the midline and its progenitors, whereas Nodal signaling is antagonized by Lefty2-mediated block of Cryptic in the lateral plate. Our results might also have implications for the role of Cripto in tumorigenesis. Cripto is highly overexpressed in human epithelial cancers, such as breast and colon carcinomas ( Salomon et al. 2000 ), and has been implicated in tumor formation ( Ciardiello et al. 1991 , 1994; Baldassarre et al. 1996 ; De Luca et al. 2000 ; Salomon et al. 2000 ; Adkins et al. 2003 ). Although the mechanisms by which Cripto acts in these circumstances are unclear, inhibition of Cripto by antisense or antibody blockade can inhibit tumor cell proliferation ( Ciardiello et al. 1994 ; Baldassarre et al. 1996 ; De Luca et al. 2000 ; Adkins et al. 2003 ; reviewed by Shen 2003 ). Since Lefty is an in vivo antagonist of EGF-CFC activity, it might also serve as a therapeutic agent to block Cripto. Subtle Sequence Differences Determine the Interaction with Coreceptors and Inhibitors The finding that the highly related ligands Activin, Nodal, and Vg1/GDF1 activate the same receptors but differ in their dependence on coreceptors allowed us to determine how ligand diversity and signaling specificity can be achieved. We have identified the loop-β8 region as a 14 amino acid domain, a mere 4% of the entire TGFβ signal, that contributes to coreceptor dependence or independence. Sqt and Vg1 incorporating the loop-β8 region of Activin can bind to the Activin receptors in the absence of EGF-CFC proteins. Conversely, Activin incorporating the loop-β8 region of Sqt is inactive, suggesting that the Nodal/Vg1 loop-β8 region might be inhibitory. This inhibition can be relieved by the Finger 1 domain of Sqt, which results in the dependence on EGF-CFC coreceptors ( Figure 9 D). These results indicate that rather subtle sequence variations can lead to striking changes in ligand diversity. Structural considerations suggest that the loop-β8 region determines coreceptor independence or dependence at least in part by its interactions with type II receptors. The conserved Lys102–X–Asp104 motif in the Activin loop-β8 region has been shown to be important for high-affinity binding to the ActRII receptor ( Wuytens et al. 1999 ; Greenwald et al. 2003 ; Thompson et al. 2003 ). In the crystal structure of the ActivinβA-ActRIIB complex, Lys102–X–Asp104 forms an intramolecular salt bridge that interacts with a hydrophobic interface on ActRIIB ( Thompson et al. 2003 ). Mutational analysis has shown that substituting Lys102 with a neutral charge (Ala) significantly reduces receptor binding affinity and signaling ( Wuytens et al. 1999 ). In contrast to Activin, EGF-CFC-dependent ligands such as Nodals and Vg1/GDF1 have the differentially charged residues Met/Leu102 and His104 at the corresponding positions (numbering according to ActivinβA). Similarly, in BMP7 the corresponding residues are Leu102 and Lys104. It has been shown that modeling Lys onto the aligned 102 residue in BMP7 positions it within hydrogen-bonding distance to Glu29 of ActRII and may allow for greater hydrophobic packing at the interface ( Greenwald et al. 2003 ). Analogously, we propose that Sqt Act β A[loop β 8] , Sqt Act β B[loop β 8] ,Vg1 Act β B[loop β 8] , Sqt3, and Sqt5 are coreceptor-independent because of their favorable binding to ActRIIB receptors. Conversely, the corresponding region in Sqt and Vg1 might be inhibitory because of inefficient interaction with ActRIIB receptors. Detailed structural studies should reveal whether EGF-CFC proteins overcome this inhibition by changing the conformation of Nodal and Vg1 or by providing an additional interaction surface that allows the assembly of receptor complexes. In summary, our results lead to two major conclusions. First, Lefty inhibits a subset of TGFβ signals by using the novel mechanism of blocking pathway-specific coreceptors belonging to the EGF-CFC family. Second, subtle sequence changes in TGFβs determine their signaling specificity and dependence on coreceptors. Although Drosophila has an Activin signaling pathway, Nodals, Leftys, and EGF-CFC proteins seem to be restricted to chordates ( Brummel et al. 1999 ; Schier 2003 ). The evolution of Activin-like signaling pathways therefore represents a remarkable example of how a simple signaling pathway consisting of ligand and receptors can be diversified by subtle sequence changes that modulate the interaction with coreceptors and their inhibitors. Materials and Methods Strains and embryos. Adult homozygous fish for oep tz57 were generated as described previously ( Zhang et al. 1998 ; Gritsman et al. 1999 ). Xenopus embryos were obtained as described in Hemmati-Brivanlou et al. (1992 ). Generation of constructs. Epitope-tagged and chimeric constructs were made using PCR-based methods and confirmed by sequencing. pCS2-zebrafish Lefty1/HA and Lefty1/Glu constructs were generated by inserting three tandem copies of HA-epitope or Glu-epitope, respectively, after Val145. The initial three-lettered (XXX) Sqt/ActivinβB chimeras were generated by subcloning the prodomain of Xenopus ActivinβB (codons Met1 to Gly256) fused to an HA-epitope/XhoI fragment (YPYDVPDYALE) and followed by the mature chimeric ligand into pcDNA3 vector. S denotes Sqt; A denotes ActivinβB. The boundaries for Sqt mature ligand domains are as indicated: Finger 1 (Asn263 to Cys325), Heel (Pro326 to Cys358), and Finger 2 (Val359 to His392). The boundaries for Xenopus ActivinβB mature ligand domains are as indicated: Finger 1 (Cys215 to Cys299), Heel (Pro300 to Cys335), and Finger 2 (Ile336 to Ala370). Full-length chimeras were generated by incorporating the indicated regions into Sqt, ActivinβB, or zebrafish Vg1, which were then subcloned into the pT 7 T S vector ( Ekker et al. 1995 ). The boundaries for Sqt Finger 2 structural subregions are as indicated: β6β7 (Val359 to Try370), loop (Tyr371 to Met376), β8 (Val377 to Gly383), and β9 (Met384 to His392). The boundaries for Xenopus ActivinβB Finger 2 structural subregions are as indicated: β6β7 (Ile336 to Try347), loop (Phe348 to Ile354), β8 (Val355 to Asn356), and β9 (Met357 to Ala370). The Xenopus ActivinβA loop-β8 region sequence is FDRNNNVLKTDIAD (also identical in Xenopus ActivinβD). The zebrafish Vg1 loop-β8 region is from Try332 to Asp345. pcDNA3-zebrafish Vg1/HA, pcDNA3-Squint/HA, pCS2-Alk4(KR)/Flag (a kinase-defective mutant of human Alk4 with Lys234 to Arg234 substitution), pCS2-ActRIIB(KR)/Myc (a kinase-defective mutant of mouse ActRIIB with Lys217 to Arg217 substitution), and pCS2-Cripto/Flag have been described elsewhere ( Yeo and Whitman 2001 ; Cheng et al. 2003 ). Embryo microinjection. Plasmids were linearized and sense strand-capped mRNA was synthesized using the mMESSAGE mMACHINE system (Ambion, Austin, Texas, United States). Zebrafish embryos were dechorinated by pronase treatment and injected between the one- and four-cell stage. Xenopus embryos at the one- to two-cell stage were used for injections into the animal pole. Phenotypic analysis. Zebrafish embryos at 24 h were mounted in 2% methylcellulose and photographed using a Zeiss (Oberkochen, Germany) M2Bio dissecting microscope. In situ hybrization was performed as described previously ( Thisse et al. 1993 ), using RNA probes to gsc and ntl ( Stachel et al. 1993 ; Schulte-Merker et al. 1994 ). Luciferase reporter assay. Luciferase assays were performed with three to six samples and five embryos in each sample. Results are representative of three independent experiments. The injection mixtures were equalized with respect to total mRNA amount with LacZ mRNA. The A3-luc reporter DNA construct (25 pg) ( Chen et al. 1996 ) was also coinjected. Whole zebrafish embryos were harvested at shield stage. Luciferase activity was analyzed using the Luciferase Reporter Assay system (Promega, Madison, Wisconsin, United States) according to the manufacturer's instruction in a Lumat LB9501 (Berthold Technologies, Bad Wildbad, Germany). Owing to the technical aspects of microinjections, in rare circumstances, a single outlier was statistically removed from a population using Grubbs' test/extreme studentized deviate method. Inclusion of outliers into the populations does not change the statistical significance of the p values; that is, p remains <0.05, where indicated. Coimmunoprecipitation analysis. Xenopus embryos were harvested at stage 10. For chemical cross-linking of proteins, animal halves were incubated in PBS with 10 mM 3,3′-dithiobis(sulfo-succinimidyl propionate) (DTSSP) (Pierce Biotechnology, Rockford, Illinois, United States) and incubated for 2 h on ice. Coimmunoprecipitation was performed as described previously ( Yeo and Whitman 2001 ). Purified processed mouse Lefty1, soluble mouse Cripto, and mouse VEGF-D proteins were obtained from R&D Systems (Minneapolis, Minnesota, United States). Activity assays were performed by R&D Systems. The proteins were incubated in PBS with 1 mM DTSSP for 1 h on ice. Coimmunoprecipitation was performed as described previously ( Yeo and Whitman 2001 ). Samples were treated with 100 mM DTT to cleave DTSSP prior to SDS-PAGE analysis. The following antibodies were used for immunoprecipitation and Western blot analysis: anti-Flag mouse monoclonal antibody (clone M2; Sigma, St. Louis, Missouri, United States), anti-HA mouse monoclonal antibody (clone 16B12; Covance, Princeton, New Jersey, United States), anti-HA rabbit polyclonal antibody (Y-11; Santa Cruz Biotechnology, Santa Cruz, California, United States), anti-c-Myc rabbit polyclonal antibody (A-14; Santa Cruz Biotechnology), anti-c-Myc mouse monoclonal antibody (clone Ab-1; Oncogene Science, Tarrytown, New York, United States), anti-His mouse monoclonal antibody (clone 6-His; Covance), anti-mLefty1 goat polyclonal antibody (R&D Systems), and anti-Glu mouse monoclonal antibody (clone Glu-Glu; Covance). Proteins were visualized using the Super Signal West Pico/Femto Chemiluminescent Substrate system (Pierce). Supporting Information Accession Numbers The GenBank ( http://www.ncbi.nlm.nih.gov/entrez/query.fc-gi?db=Nucleotide ) accession numbers for the sequences discussed in this paper are ActivinβA (Q9W6I6), ActivinβB (Q91350), human ActRIIB (P08476), mouse ActRIIB (P27040), rat ActRIIB (P38445), Alk4 (Z22536), BMP7 (P23359), Cripto (P51865), Cryptic (P97766), cyclops (P87358), GDF1 (P20863), goosecoid (P53544), Lefty1 (Q9W6I6), Lefty2 (P57785), mLefty1 (Q64280), Nodal (P43021), no tail (Q07998), one-eyed pinhead (O57516), Pitx2 (P97474), squint (O13144), VEGF-D (P97946), and Vg1 (P09534). | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC340941.xml |
535808 | Extractor for ESI quadrupole TOF tandem MS data enabled for high throughput batch processing | Background Mass spectrometry based proteomics result in huge amounts of data that has to be processed in real time in order to efficiently feed identification algorithms and to easily integrate in automated environments. We present wiff2dta, a tool created to convert MS/MS data obtained using Applied Biosystem's QStar and QTrap 2000 and 4000 series. Results Comparing the performance of wiff2dta with the standard tools, we find wiff2dta being the fastest solution for extracting spectrum data from ABIs raw file format. wiff2dta is at least 10% faster than the standard tools. It is also capable of batch processing and can be easily integrated in high throughput environments. The program is freely available via , and is also available from Applied Biosystems. Conclusions wiff2dta offers the possibility to run as stand-alone application or within a batch process as command-line tool integrated in automation and high-throughput environments. It is more efficient than the state-of-the-art tools provided. | Background In tandem mass spectrometry proteins are identified by matching the measured fragment ion spectra derived from peptides with theoretical spectra calculated from known DNA or protein sequences, for example the NCBI sequence database [ 1 ]. Algorithms used for this purpose usually have their own input formats and are not able to read the proprietary binary file formats of the mass spectrometer manufacturers. Nevertheless, they are able to read a common format, the DTA format introduced by the Sequest™ algorithm [ 2 ]. Thus, needs exist for converting mass spectra into this common format in order to feed the different identification algorithms such as Sequest™ or Mascot™ [ 3 ]. The conversion must be accomplished efficiently, requiring as few user interaction as possible. Integrated in high-throughput environments, mass data processing must be realized. Applied Biosystems mass spectrometers are controlled by a software called Analyst™. This software is used for data evaluation purposes, too. It offers a possibility to integrate extensions called "scripts". One of these scripts available from the manufacturer [ 4 ] is "Export IDA Spectra.dll", the only known possibility besides the mzStar from SASHIMI Project [ 5 ] to export DTA files from Applied Biosystems ESI data. Using the tools provided by SASHIMI results in two steps: first mzStar must be used to create an XML [ 6 ] document (mzXML Schema) as intermediate step, then mzXML2Other must be applied for creating DTA or other formats from the mzXML document, and thus conversion consumes a lot of time and computational power. mzStar is not designed for batch processing nor for converting more than one wiff file in a single run. The Analyst™ script itself requires each chromatogram being opened in Analyst™ per conversion, resulting in a lot of user interaction for each single export. This leads to the effect that batch processing is impossible in both cases and only one binary file can be converted at once. A schematic diagram of the conversion method workflows is shown in figure 1 . Another script named mascot.dll provides support for invocating Mascot™ as protein identification algorithm using Applied Biosystems Analyst™. Such a script does not exist for Sequest™. In most proteomics labs support for Mascot™ as well as for Sequest™ is needed, because these two algorithms are most commonly used in this research field. Although the additional information that can be stored in mzXML is needed in the case of quantitative proteomics experiments based on isotopic labelling of peptides (ICAT [ 7 ] or SILAC [ 8 ]), this format can be read neither by Mascot™ nor by Sequest™. We decided to develop a tool for converting data obtained from Applied Biosystems QStar™, providing features like batch processing in an operatorless high throughput environment. If no ER, NL or Prec scans are used, data acquired using a QTrap™ 2000/4000 can be converted, too. This tool is named wiff2dta. Implementation The implementation was done according to the Analyst™ Cookbook, a documentation available from Applied Biosystems upon request. wiff2dta is implemented in Visual Basic™ (Microsoft Corp.) because ActiveX™ is provided as the one and only application programming interface (API) by Applied Biosystem's Analyst™ software. Therefore, this is needed for accessing the binary wiff files. Thus, this tool is operating system dependant and only runs on Windows™ (Microsoft Corp.) systems. We use the code provided by the Analyst™ software API in order to benefit from new releases and maintain coherence. The program has two modes of user interaction: one provides a graphical user interface (GUI) and requires user interaction (GUI-mode); the other uses command-line parameters and suppresses the GUI as no user interaction is required (batch-mode). In batch mode, automation of conversion processes can be achieved. The GUI is shown in figure 2 . Conversion can be done in two modes. On one hand only a single binary file can be selected for conversion (file-mode). On the other hand, a whole directory tree can be traversed and all binary ESI MS/MS files in all (or only selected) folders can be converted in one run (directory-mode). For example this mode can be used to convert a folder full of MS/MS data at once. In file-mode distinct samples of one data file can be marked for conversion, if desired. In directory-mode, each sample of each ESI MS/MS file is processed. Used in directory-mode, wiff2dta can be forced to save all resulting DTA files in one single folder by checking "all in one folder". Otherwise, the converted files are stored in a single folder with the name derived from the source ESI MS/MS data file. This folder is placed in the same directory where the corresponding binary file was found. The conversion itself can be controlled by entering appropriate values in the text fields displayed under the title "Parameters", shown in figure 2 . Parameters are "Mass tolerance for combining MS/MS spectra", "MS/MS export threshold", "Minimum number of MS/MS ions for export", "Centroid height percentage", "Centroid merge distance", "Minimum charge of exported spectra" and "Maximum charge of exported spectra". These are parameters of identical function as used by the export of DTA provided by Applied Biosystems' script. wiff2dta produces the same values as this tool, as shown in table 1 . Support for other formats, like mascot generic format (MGF) [ 9 ] and mzXML [ 10 ] will be added. We first focussed on high throughput for conversion into DTA in order to be able of feeding our search programs efficiently. wiff2dta is able to be integrated in automation and high throughput environments. This can be achieved making use of the command line options. All parameters and modes can be controlled by command-line parameters. These are shown in figure 3 . Every GUI parameter has a corresponding command line option. Batch-mode is entered by providing the parameter /auto at the command-line. If this is not present, the values provided override the defaults in the GUI and the form will be displayed. Results The program can be started in multiple instances, resulting in parallel processing. Using this feature, it is possible to use several processors on one computer. Additional to this, wiff2dta is about 10% faster than the original tool provided by Applied Biosystems and about 20 times faster than mzStar of the Sashimi project. See table 2 . During a 24 hour conversion, the 10% performance gain in savings of about 2.5 hours using the tool original tool provided by Applied Biosystems. Conclusions wiff2dta demonstrates improvements in reducing computation time by exploiting a range of optimizations in coding and using the COM interfaces to Analyst™. Useful features like the capability of being integrated in batch processes and mass data processing lead to immense time savings, too. Availability and requirements wiff2dta has to be installed in the BIN directory of an installed Analyst™ version 1.3 or higher. The installation consists just of copying the file wiff2dta.exe into this directory. If desired, a link to the program file can be created that can be placed onto the desktop or into the start menu. The program is freely available from Applied Biosystems (UK) upon request and freely available via and for download. List of abbreviations used API: application programming interface DNA: desoxyribonuclein acid DTA: file extension ms spectra data in Sequest™ format ER: enhanced resolution ESI: electron spray ionization GUI: graphical user interface MS: mass spectrometry, mass spectrometer MGF: mascot generic format, file extension used for this format NL: neutral loss Prec: precursor ion TOF: time-of-flight WIFF: file extension of Applied Biosystems raw data files Authors' contributions AB implemented the program and made a draft of the manuscript. RPG and AS contributed with ideas and proofread the manuscript. RPG supervised the final testing. All authors have read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC535808.xml |
553970 | Acute ST-segment elevation myocardial infarction after amoxycillin-induced anaphylactic shock in a young adult with normal coronary arteries: a case report | Background Acute myocardial infarction (MI) following anaphylaxis is rare, especially in subjects with normal coronary arteries. The exact pathogenetic mechanism of MI in anaphylaxis remains unclear. Case presentation The case of a 32-year-old asthmatic male with systemic anaphylaxis, due to oral intake of 500 mg amoxycillin, complicated by acute ST-elevation MI is the subject of this report. Following admission to the local Health Center and almost simultaneously with the second dose of subcutaneous epinephrine (0.2 mg), the patient developed acute myocardial injury. Coronary arteriography, performed before discharge, showed no evidence of obstructive coronary artery disease. In vivo allergological evaluation disclosed strong sensitivity to amoxycillin and the minor (allergenic) determinants of penicillin. Conclusion Acute ST-elevation MI is a rare but potential complication of anaphylactic reactions, even in young adults with normal coronary arteries. Coronary artery spasm appears to be the main causative mechanism of MI in the setting of "cardiac anaphylaxis". However, on top of the vasoactive reaction, a thrombotic occlusion, induced by mast cell-derived mediators and facilitated by prolonged hypotension, cannot be excluded as a possible contributory factor. | Background Acute myocardial infarction (MI) complicating anaphylaxis induced by drugs or other chemicals is uncommon and only sporadic cases have been reported [ 1 - 9 ]. The underlying pathogenetic mechanisms have not been fully elucidated. The case of a 32-year-old man with normal coronary arteries, who developed acute MI following amoxycillin-induced anaphylaxis, is the subject of this report. The underlying pathogenetic mechanisms are also discussed and the relative literature is reviewed. Case presentation A 32-year-old man was admitted to the emergency room of the local Health Center because of anaphylaxis, which developed 2 hours and 15 minutes after the ingestion of amoxycillin (500 mg), prescribed by his dentist. Prodromal signs of anaphylaxis (flushing, pruritus, warmth, urticaria) reportedly occurred about 15 minutes before the onset of symptoms from other systems. Ten days earlier the patient, an asthmatic since childhood, had completed a 4-day course of amoxycillin (500 mg TID) without any side effects. Apart from obesity, there were no other risk factors for coronary artery disease. On admission the patient was in acute distress. He was complaining of dizziness, blurred vision, dyspnea and abdominal pain. Initial examination revealed an obese man (Weight = 130 kg, Body Mass Index = 38 kg/m 2 ) with generalized erythema, angioedema, cyanosis and diffuse wheezing; the systolic blood pressure was 70 mmHg and the pulse rate 120 bpm. The patient was then connected to a cardiac monitor, which showed sinus tachycardia (approximately 140 bpm) without ST-segment and T wave abnormalities (Fig. 1A ). Pulse oxymetry demonstrated an oxygen saturation (SpO 2 ) of 90%. Figure 1 Lead II monitor strip recorded shortly after admission (A) and at the chest pain onset (B). Epinephrine (0.3 mg or 0.3 ml of a dilution 1:1000) was injected subcutaneously (SC); dimethindene (4 mg) and hydrocortizone (500 mg) were administered intravenously (IV) in slow infusion. Nebulized salbutamol and supplemental oxygen were given as well. Normal saline with 50 mg ranitidine hydrochloride and Ringer's solution were infused through separate intravenous lines. Due to unsatisfactory clinical response (persistence of hypotension and tachycardia, despite improvement of the pulmonary signs) a second dose of epinephrine (0.2 mg) was given SC 20 minutes later. Almost simultaneously with the administration of the second dose of epinephrine, ST-segment elevation appeared on the monitor (Fig. 1B ) and the patient complained of substernal chest pain. A 12-lead electrocardiogram (ECG) showed ST segment elevation in leads II, III, aVF, and V 3 to V 6 (Fig. 2 ). A single dose of 325 mg acetylsalicylic acid was given per os; heparin (5000 UI) and pethidine hydrochloride (25 mg) were administered IV. An hour later the patient was hemodynamically stable. The arterial blood pressure was 125/90 mmHg and SpO2 rose to 96%. However, the chest pain persisted and nitroglycerin titrated at a dose of 25 μg/min was infused IV, without complete symptom relief. At this point the patient was transferred to the Coronary Care Unit (CCU) of the nearest general hospital. Figure 2 ECG recorded during chest pain. Treatment in the CCU included thrombolysis with reteplase – administered according to the standard protocol – 2 hours after the onset of chest discomfort, which according to the established clinical and electrocardiographic criteria was considered successful [ 10 ]. The peak levels of serum troponin I and creatine phosphokinase were 45.5 ng/ml (normal = 0–2 ng/ml) and 575 U/L (normal = 25–195 U/L), respectively; while the MB fraction was 77 U/L (normal = 0–24 U/L). The rest of the laboratory results, including serum cholesterol, LDL, HDL, Lpa and triglycerides, were within normal limits. An ECG performed before discharge showed complete loss of potentials in leads III and aVF and partial loss of potentials with a small q wave in lead II (Fig. 3 ). An echocardiographic study performed on the 5 th hospital day showed preserved systolic function (ejection fraction 60%) without wall motion abnormalities. Left (Fig. 4 ) and right (Fig. 5 ) coronary angiography showed no evidence of obstructive coronary artery disease; left ventriculography was normal as well. The patient recovered completely and was discharged a week after admission. He was referred to the adult Allergology Center of our area performing diagnostic work up for drug allergy. Figure 3 ECG recorded on hospital day 4 th . Figure 4 Left coronary angiogram. Figure 5 Right coronary angiogram. Prick skin tests (PST) were performed using penicilloyl polylysine (PPL), and minor determinant mixute (MDM) [supplied as Allergopen by Allergopharma (Reinbeck, Germany)]; amoxycillin, ampicillin and cefamandole solutions were also used (concentration 20 mg/ml) for PST. PST were strongly positive [4+ reaction (on a scale of 1–4) i.e. wheal >5 mm in diameter with pseudopodes, with no reaction at all to the diluent control] to amoxycillin and ampicillin. Intradermal tests were performed only to PPL, MDM (dilution 1/10) and cefamandole (0.2 mg/ml and 2 mg/ml) and resulted in strongly positive reaction (4+) only to the MDM. Circulating specific IgE to penicillin V, penicillin G, amoxycillin, ampicillin and cefaclor was not demonstrable by CAP (Pharmacia, Sweden). Discussion Anaphylactic reactions may trigger cardiovascular events, including MI [ 1 - 9 ], primarily in subjects with underlying ischemic heart disease [ 1 , 2 , 4 , 7 , 9 ] and rarely in individuals with normal coronary arteries [ 3 , 5 , 6 , 8 ]. Cardiovascular complications in the setting of anaphylaxis rarely occur in subjects less than 35 years old [ 11 ]. Acute coronary syndromes following antibiotic-induced anaphylaxis is uncommon and only few cases have been reported so far [ 1 , 4 , 12 - 14 ]. To our knowledge, only one case of amoxycillin-induced anaphylaxis complicated with acute MI has been documented in a 62-year-old man with underlying coronary artery disease [ 4 ]. Evidence that the human heart is the target as well as the site of anaphylaxis is constantly growing. A number of studies by Marone et al [ 15 , 16 ] and others [ 17 ] have demonstrated mast cells within the human heart; they are strategically located perivascularly, in close proximity to myocytes, in the arterial intima and the shoulder region of atheromas. Cardiac mast cells express on their surface high affinity receptors for IgE (FcεRI) and also C5a receptors [ 18 ]. Mast cells activation through immunological and non-specific (non-immunological) stimuli leads to the release of a variety of vasoactive and pro-inflammatory mediators, preformed or de novo synthesized (histamine, prostaglandins, leukotrienes, enzymes, cytokines and others) [ 15 , 19 ]. Coronary spasm, which is proposed as the main underlying mechanism of allergy-induced coronary syndromes [ 5 , 11 , 13 , 14 , 20 , 21 ], can be caused, directly or indirectly, by potent vasoactive mast cell-derived mediators such as histamine, prostaglandin D2, thromboxane, cysteinyl leukotrienes [ 15 , 18 , 22 - 24 ]. In addition, mast cells are likely to affect coagulation and fibrinolysis at different levels through enzymes and mediators they secrete [ 15 ]. Therefore, mediators from perivascular and interstitial cardiac mast cells – as well as those reaching the heart from the pulmonary circulation – might affect coagulation, favoring platelet aggregation and thrombus formation [ 15 , 25 ]. In the present case, the temporal sequence of events suggests that cardiac anaphylaxis was the triggering factor of MI. We propose that prolonged coronary vasospasm induced by vasoactive and inflammatory mediators, released during anaphylaxis, was the main causative mechanism. However, in view of the persistence of chest pain following IV nitroglycerin infusion, and its complete resolution promptly after IV thrombolysis, a thrombotic vascular occlusion, on top of the vasospastic reaction, cannot be excluded. The latter is also supported by the prolonged systemic hypotension, which, as it has been emphasized in previously reported cases [ 26 ], probably caused further reduction of the myocardial perfusion, thus, favoring in situ thrombus formation and subsequent coronary artery occlusion. In acute MI cases, presenting to centers with cardiac catheterization facilities, urgent coronary arteriography appears as the management strategy of choice for the final diagnosis. However, in the setting of acute anaphylaxis, this might not be the wisest choice, particularly in cases, like the present one, in which the offending allergen had been administered orally. Moreover, continuous or delayed allergen absorption could further aggravate anaphylaxis, and a late phase reaction represents a potential risk in all anaphylactic reactions. Finally, no bibliographical data are available concerning the tolerability of IV administered contrast agents in patients who have suffered a recent episode of severe systemic anaphylaxis. The administration of epinephrine – a life saving agent in cases of anaphylaxis – has been implicated as a cause of acute MI in a limited number of reports [ 27 , 28 ]. In the present case, however, it appears unlikely that exogenous epinephrine was the initiating event for the following reasons: a) the first epinephrine dose (0.3 mg) was rather low to induce significant vasoconstriction in a subject with a body weight of 130 kg, b) the mode of administration (SC) is considered the safest in this regard, c) the second dose (0.2 mg), administered at a safe interval (20 min) after the first one, does not seem to be involved since its injection coincided with the onset of chest pain, before any anticipated drug absorption. On the contrary, in the above quoted cases [ 27 , 28 ] the epinephrine-induced MI developed 5–15 minutes post injection. Moreover, the possibility of inadvertent intravenous administration of epinephrine appears improbable, since no blood was withdrawn in the syringe before drug injection. However, one might argue that the exogenous epinephrine could have aggravated a pre-existing coronary spasm, induced by mast cell-derived mediators. Hence, the precise effect of epinephrine in this clinical setting remains a matter of speculation. The allergological evaluation performed in our patient showed strong sensitivity to amoxycillin and to the minor determinants, which are the allergenic epitopes associated with systemic anaphylaxis. The negative CAP results do not negate the in vivo findings, since it is well established that the in vitro techniques are not as sensitive as skin tests. Furthermore, CAP detects antibodies against the major determinants of penicillins (involved pathogenetically in penicillin-induced urticaria) and not to the minor ones, which are implicated in systemic anaphylaxis. For technical reasons, relating to both the Allergology Center and the patient's professional obligations as well as residence distance, the allergological work up was completed 6 months later; such a delay might also have contributed to the negative in vitro findings. Conclusion Acute ST-elevation MI is a rare but potential complication of anaphylactic reactions, even in young adults with normal coronary arteries. Physicians should be alert for such a complication in order to diagnose it early and treat properly. It is fairly well established that that human heart can be both, the site and the target of severe anaphylaxis; in this setting cardiac mast cells – activated and releasing multiple vasoactive mediators – play an important role in pathogenesis of cardiac complications [ 16 - 18 ]. In the above case, mediator-induced coronary artery spasm was the main, but probably not the exclusive causative mechanism of anaphylaxis-related MI. The thrombotic vascular occlusion, induced by inflammatory mediators and facilitated by prolonged hypotension, cannot be excluded as a possible contributory factor. Abbreviations ECG = electrocardiogram; IV = intravenously; MDM = minor determinant mixute; MI = myocardial infarction; NV = normal values; PPL = penicilloyl polylysine; PST = prick skin tests; SC = subcutaneously. Competing interests The author(s) declare that they have no competing interests. Authors' contributions AG was responsible for the initial evaluation and management of the patient, GL was involved in the cardiological evaluation and management and, KK-F performed the allergological evaluation. All authors have equally contributed in the preparation and revision of the manuscript. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC553970.xml |
549519 | Age dependent association of endometrial polyps with increased risk of cancer involvement | Background Endometrial polyps (EMPs) are commonly encountered in routine surgical pathology practice, but opinions differ on whether they are intrinsically a marker for concurrent or subsequent malignancy. The objectives of the present study are 1) to investigate the age-group in which EMP are most commonly encountered 2) to document the age-group in which EMP are most commonly associated with malignancies 3) To investigate whether the age of diagnosis of the various carcinoma subtypes in EMPs is congruent with published data on similar malignancies arising in non-polypoid endometrium and 4) To investigate whether the histologic subtype distribution of malignancies associated with EMPs are similar or different from the distribution of malignancies arising from non-polypoid endometrium based on published data. Patients and methods All cases of EMPs were retrieved from the files of Yale-New Haven Hospital for the period 1986–1995. The patients were divided into 5 age groups: Each group was further subclassified based on an association (or lack thereof) of EMPs with endometrial carcinoma. Chi-square test was used to compare the proportion of malignancy associated EMPs between the age groups. Results We identified 513 EMPs, of which 209 (41%) were from biopsy specimens and 304 (59%) from hysterectomy specimens. Sixty six (13%) of all EMPs were malignant. The 66 malignant EMPs included 58 endometrioid, 6 serous, 1 carcinosarcoma, and 1 clear cell carcinoma. In age group >35, only 1(2.5%) of 40 EMPs was associated with endometrial malignancy. In contrast, 37(32%) of 115 EMPs were associated with malignancy in the age group > 65. The frequency of malignant EMPs increased with age and reached statistical significance in the age group >65 (p < 0.001). The most common histologic type of malignancy was endometrioid adenocarcinoma. Conclusions EMPs show statistically significant age dependent association with malignant tumor involvement. Careful search for malignancy, particularly in women with multiple risk factors is advised in daily practice. Additional studies are needed to address the histological features and immunohistochemical profiles in the context of association between endometrioid and high-grade endometrial carcinoma and endometrial polyps. | Background Endometrial polyps (EMPs) are generally considered benign proliferative lesions and are commonly encountered in routine surgical pathology practice. The usual histological pattern of endometrial polyps is characterized by irregular proliferative glands, with a fibrotic stroma containing thick-walled blood vessels [ 1 ]. The morphologic diversity of endometrial polyps is reflective of the morphologic spectrum of the background endometrium from which EMPs arise. As such, EMPs may range from atrophic to hyperplastic to carcinomatous. However, opinions differ on whether EMPs are intrinsically a marker for concurrent or subsequent malignancy. Endometrial polyps were identified in 12–34% of uteri containing endometrial carcinoma in two earlier studies [ 2 , 3 ]. In another case-control study examining previous pathology in women diagnosed with endometrial carcinoma, endometrial polyps were twice as likely to be detected than in the control group [ 4 ]. Rarely, serous endometrial intraepithelial carcinoma (EIC), the presumptive early form of uterine papillary serous carcinomas, may be identified as very minute foci in EMPs [ 5 , 6 ]. This finding may be interpreted as the EMP homologue of similar changes that are occasionally identified in non-polypoid atrophic endometrium. However, given that nonrandom chromosomal aberrations and monoclonality that have been demonstrated in EMPs [ 7 , 8 ], an alternate interpretation is that molecular and/or cytogenetic alterations inherent to EMPs facilitate a neoplastic transformation. The latter interpretation would imply that endometrial polyps are a risk factor for the development of endometrial tumors. And indeed a possible association between endometrial polyps and endometrial malignancy in postmenopausal women has been suggested couple of decades ago [ 4 ]. However, there is no direct evidence for a greater propensity of polypoid endometrium to undergo malignant change as compared to the adjacent normal endometrium, and EMP may simply represent am embodiment of the greater propensity of the host endometrium to develop proliferative/neoplastic changes in general [ 9 ]. A recent study designed and conducted to investigate the pathological significance of EMPs and their association with pre-malignant and malignant conditions failed to supply evidence of such association. That study involved a large cohort of patients seen in outpatient hysteroscopy clinic for abnormal uterine bleeding. To determine the magnitude of malignant potential among polyps, the authors compared the pathological findings in polyps with non-polypoid specimens. The comparative analysis established that endometrial hyperplasia was more frequent in endometrial specimens with polyps, but the incidence of frank carcinoma in polypoid and non-polypoid endometrium remained the same. Although not age stratified, the study showed that in abnormal uterine bleeding, hyperplasia presented more frequently in women with EMPs compared to those without polyps, but cancer involvement regardless of the histological pattern was not significantly different [ 9 ]. Similar results and failure to establish any association of endometrial polyps and carcinoma were demonstrated in another recent study dealing with endometrial polyp characteristics in menopausal women on hormonal replacement therapy [ 10 ]. Most standard pathology texts list endometrial polyps as being most prevalent in perimenopausal women and suggest possible association between polyps and malignant involvement [ 9 , 11 ]. However, there has been no detailed age-based analysis of the incidence and malignant involvement of EMPs. In this report, age-related differences in the incidence of EMP at the time of diagnosis in the practice of a busy academic center is examined, with a detailed analysis of the incidence and histologic subtypes of malignancies associated with EMPs. The objectives of the study are 1) to investigate the age-group in which EMP are most commonly encountered in routine surgical pathology practice 2) to document the age-group in which EMPs are most commonly associated with malignancies 3) To investigate whether the age of incidence of the various carcinoma subtypes in EMPs at the time of diagnosis is congruent with published data on similar malignancies arising in non-polypoid endometrim and 4) To investigate whether the histologic subtype distribution of malignancies associated with EMPs is similar or significantly different from the distribution of malignancies arising from non-polypoid endometrium. Patients and methods Case retrieval and pathologic classifications All cases with a diagnosis of EMP were retrieved from the computerized database of the Pathology Department at Yale-New Haven Hospital for the 10-year-period from 1986 to1995. All cases were further investigated for involvement of endometrial cancers including malignancies without myometrial invasion; histologic subtypes of all malignant tumors were catalogued. Histologic types of endometrial malignancies were characterized according to the WHO classification [ 12 ]. All cases were reviewed microscopically and confirmed by a second pathologist. For endometrial cancer with mixed histologic type, the presence of a second component was considered if it involved more than 10% of all available sections containing tumor. The cases of endometrial malignancy involving both EMPs and non-polyp endometrium, were classified into the category of EMPs with cancer involvement. Patients groups For comparative purposes, the patients were divided into 5 age groups: 25–35; 36–45; 46–55; 56–65; and >65 years; and each group was further classified based on an association (or lack thereof) with endometrial carcinoma. A starting point of 25 years of age was selected due to the very low incidence of endometrial polyps in patients below this age. Two patients (ages 18 and 19) were excluded from the study as they did not represent a sufficiently large for statistical analysis group. The proportion of both groups (polyps associated with malignancies (malignant polyps) and polyps not associated with malignancies (benign polyps) were statistically compared for each of the aforementioned age-groups. Subsequently, we merged the younger age groups and preserved the >65 year group, which we referred to as "postmenopausal", since significant differences were observed in this specific subset of patients. Larger age groups were arbitrarily designated as reproductive years (25–45), perimenopause (shortly before or after menopause, 46–65) and postmenopause (>65) and statistically analyzed. The postmenopausal status of all women above the age of 65 was verified and the term "postmenopausal" was occasionally used when referring to this particular age group. Our use of this term, although arbitrary, was important in order to put the emphasis on the fact that any pathomorphological findings in this age group are unlikely to be attributed to changes characteristic of the cycling endometrium. Statistical analysis Chi-square test was used to compare the proportion of malignancy associated EMPs between the age groups and in regards to particular histological type of malignancies. Results Out of all diagnostic and therapeutic procedures performed over this period, a total of 513 EMP were identified. The latter included 304 (59%) endometrial biopsies/curetting samples, and 209 (41%) hysterectomy specimens. In cases in which endometrial biopsies and hysterectomies both showed presence of EMP, only the hysterectomy specimen was considered. The age of patients ranged from 18–91 years with a median of 54 years. Sixty-six (13%) of 513 EMPs were malignant. The histological subtype distribution of those 66 malignancies included 58 endometrioid (87%), 6 serous (9%), 1 carcinosarcoma, and 1 clear cell carcinoma. No mixed histological type of endometrial cancer was found in our series. The incidence of EMP peaked at age group 46–55 years, which was similar to previous reports. In age group 25–35, only 1 (2.5%) of 40 EMPs was associated with endometrial malignancy. In contrast, 37 (32%) of 115 EMPs were associated with malignancy in the age group >65 years (Figure 1 ; Figure 2 ). The frequency of EMPs with malignancy involvement increased with age and reached statistical significance ( p < 0.001) in the age group >65 years (Figure 3 ). The most common histological type of malignancy was endometrioid carcinoma, followed by serous carcinoma. The same statistically significant difference for age group >65 years (p < 0.05) remained when larger age groups, including reproductive (25–45), perimenopausal (46–65) and postmenopausal (>65) patients were compared (Figure 4 ). Figure 1 Frequency of occurrence of benign endometrial polyps by age group . The frequency of occurrence of EMPs at the time of diagnosis peaked in the age group 46–55 years (29%), followed by 36–45 (27%), 56–65 (18%) and >65 years (17%). The incidence of EMPs in the age group 25–35 years was significantly lower (9%). Figure 2 Frequency of occurrence of malignant endometrial polyps by age group . In age group 25–35 years, only 2.5% of the EMPs were associated with endometrial malignancy. In contrast, in the age group >65 years, 32% of the EMPs were associated with malignancy. Figure 3 Distribution of benign and malignant endometrial polyps by age group Although the incidence of EMPs at the time of diagnosis in the age group > 65 years was among the lowest, the incidence of malignancy associated EMPs was the highest. Figure 4 Benign and malignant endometrial polyps by age group Discussion Our results indicated a strong age dependent association of endometrial polyps and endometrial carcinoma. A linear relationship in the association rate of EMPs with malignancies and increasing age was observed, with the highest association rate identified in the >65 years age group, where 32% of the EMPs were associated with malignancy. Histological evaluation and characterization of the morphological types of carcinoma demonstrated that the vast majority (87%) of endometrial carcinomas associated with EMPs were of endometrioid, followed by the serous type (9%). These relative proportions of both major histological subtypes are in accordance with the well-known distribution of similar subtypes of endometrial carcinoma in non-polypoid endometrium [ 12 ] and thus demonstrates that neither histologic subtype is more likely than the other to develop in a polyp as compared to the adjacent endometrium. The majority of the serous carcinomas developed in the oldest age group (>65 years), whereas the majority of the endometrioid carcinomas occurred in the 46–55 age group, followed by the 36–45 age group. These age distributions are in accordance with the general concept of Type I and Type II endometrial carcinogenesis [ 13 ] and provide some evidence suggesting that carcinomas developing in EMPs do not necessarily have clinicopathologic differences from carcinoma arising in the background endometrium. It is well established that serous carcinoma may exist as a minute foci in the endometrium devoid of myometrial invasion and still show extrauterine involvement [ 14 - 16 ]. In a study of EMPs with serous carcinoma involvement with no or minimal invasion, Silva et al ., [ 17 ] reached similar findings: in 6 (37.5%) of 16 cases in that study, there was evidence of extra uterine involvement at presentation. The similarities between the patients who presented with advanced disease and the patients who presented with initial stage disease, suggested that serous carcinoma involving endometrial polyps may represent one aspect of a multicentric disease in which, the entire female genital tract and the abdominal peritoneal surfaces would be at high risk for concurrent or subsequent involvement by serous carcinoma even in the absence of myometrial invasion [ 17 ] or the extrauterine disease may represent transtubal metastasis [ 18 , 19 ]. This study also confirms previous findings that EMPs are most prevalent in the perimenopausal age group. The reason(s) for this age-segregation, which has remained remarkably consistent across various studies since the mid-fifties, is unclear. Chavez et al ., [ 20 ] speculated that with the introduction of new minimally invasive technologies (such as office hysteroscopy and sonohysterograms), the demographics of patients with EMPs will change over time as younger women undergoing evaluation for infertility will have "latent" EMPs discovered. However, when the authors compared the mean ages of women with EMPs in 1990 and 1996, there was no statistically significant difference. In addition, multiple EMPs are more prevalent in the postmenopausal women (26%) as compared with their premenopausal (15%) counterparts with EMPs. These findings suggest that the factor, or the constellation of factors responsible for the above mentioned observation is intrinsic to the endometrial polyps and surrounding endometrium condition depending on the age group. Lower incidence of endometrial polyps in the younger age group may be attributed to a possible spontaneous regression mechanism, which is characteristic of the cycling endometrium in young reproductive age women. Despite the supportive evidence of no difference in the clinico-pathological features and overall distribution of carcinomas arising from EMPs with those arising from non-polypoid endometrium, our data suggest a strong age dependent association between the presence of EMPs and involvement by endometrial carcinoma. The pathogenesis and mechanisms underlying such association are complex and not well established. Recently published data, however, provided some clue of significant differences in receptor expression, response to stimuli, and apoptosis regulation in EMPs compared to benign non-polypoid endometrium which could potentially elucidate some aspects of the possible malignant potential of EMPs. Estrogen and progesterone act as modulators of endometrial proliferation and differentiation through their receptors. Glandular epithelial expression of estrogen and progesterone receptors in polyps is not significantly different from that of the normal cycling endometrium. However, fewer stromal cells express estrogen and progesterone receptors in polyps which suggests that EMPs may result from a decrease in estrogen and progesterone receptors in the stromal cells [ 21 ]. In addition, although EMPs depend partially on estrogen receptors and grow in response to estrogen stimulation, their growth is not entirely dependent on them, this is especially so in postmenopausal women. The presence of c-erbB2 over-expression in endometrial polyps, in association with higher proliferation rates were established in a recent study [ 22 ]. This finding could explain the presence of polyps showing signs of proliferation even when the adjacent endometrium is atrophic. Thus, C-erbB2 over-expression in endometrial polyps and not in the adjacent atrophic mucosa may render polyps more sensitive to the combination of high gonadotropins and low estrogen levels, which is characteristic in the postmenopausal women. Another significant histological finding is the glandular epithelia hyperplasia in C-erbB2 -positive polyps as opposed to rather atrophic architecture in C-erbB2 -negative polyps [ 22 ]. These findings indicate that the relationship between the expression of estrogen receptors and cell proliferation in normal endometrium and EMPs differ significantly. The balance between mitotic activity and apoptosis, which regulates normal endometrial development in EMPs also shows significant alterations. Bcl-2 is a proto-oncogene, which prolongs the cell survival by inhibiting apoptosis. Bcl-2 expression has been characterized in normal cycling endometrium. Recent studies have also observed that Bcl-2 is strongly expressed in hyperplastic and malignant endometrium [ 23 ]. A localized increase in Bcl-2 expression and consequential decline or cessation of apoptosis may be another mechanism underlying the pathogenesis of endometrial polyps [ 24 ]. Elevated Bcl-2 expression results in failure of the polyp tissue to undergo normal cycle dependent sequence of proliferation, differentiation and shedding. These data imply that the relationship between receptor expression, cell proliferation and apoptosis in normal and polypoid endometrium differ significantly. Such differences combined with the nonrandom chromosomal aberrations and monoclonality, suggest that EMPs may provide a suitable microenvironment for the development of malignancy, particularly epithelial cancers. In this aspect, the molecular and/or cytogenetic alterations inherent to EMPs in a postmenopausal background could be viewed as factors facilitating and contributing to the process of malignant transformation. Our results showed a strong association of EMPs in postmenopausal patients with endometrial cancer. It raised the possibility that EMPs in postmenopausal women could represent some intermediate stage in the development of carcinoma. A similar suggestion was proposed in a study evaluating the spectrum of pathological findings in Tamoxifen treated breast cancer patients whom develop polyps and carcinoma significantly more frequently than the general population [ 25 ]. Also in favor of this hypothesis were the results provided by Silva et al ., who found that 10 (76%) of 13 Tamoxifen-related endometrial carcinomas were associated with EMPs [ 26 ]. One potential limitation of our study is our lack of consideration of the impact of variables such as hypertension, obesity and family history. However, since data regarding such possible confounders were not available to us, we set the goals of our investigation to be primarily focused on age related distribution of coinciding morphologic findings. Although we are aware of the limitations of our study and the introduced analytical bias, drawbacks that certainly pertain to any similar retrospective pathomorphologic study, we feel that we have adequately addressed the proposed investigative tasks according to the initially set scope of the study. By using the database of Yale-New Haven Hospital, we collected and analyzed a significant number of cases over an extensive period of time and thereby our study population constituted an adequate representation of the general population in respect to the morphological parameters we investigated. In summary, the age distribution, histological subtype distribution, and peak incidence of EMPs was similar to previous reports. In contrast, EMPs in postmenopausal women showed a significantly higher association with malignant tumor involvement. Careful microscopic search for malignancy in patients with multiple risk factors, particularly in postmenopausal women is advised in daily surgical pathology practice. Competing interests The author(s) declare that they have no competing interests. Authors' contributions DH wrote the original version of the manuscript. OF made substantial contributions to the content of the manuscript and participated in manuscript preparation. MM collected clinical and pathological data and revised the final version. WZ analyzed and interpreted the data and supervised the entire project. All authors have read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549519.xml |
516021 | Human cytokine-induced killer cells have enhanced in vitro cytolytic activity via non-viral interleukin-2 gene transfer | Modulation of the immune system by genetically modified immunological effector cells is of potential therapeutic value in the treatment of malignancies. Interleukin-2 (IL-2) is a crucial cytokine which induces potent antitumor response. Cytokine-induced killer cells (CIK) have been described as highly efficient cytotoxic effector cells capable of lysing tumor cell targets and are capable of recognizing these cells in a non-MHC restricted fashion. Dendritic cells (DC) are the major antigen presenting cells. This study evaluated the antitumor effect of CIK cells which were non-virally transfected with IL-2 and co-cultured with pulsed and unpulsed DC. Human CIK cells generated from peripheral blood were transfected in vitro with plasmid encoding for the human IL-2. Transfection involved a combination of electrical parameters and a specific solution to deliver plasmid directly to the cell nucleus by using the Nucleofector ® electroporation system. Nucleofection resulted in the production of IL-2 with a mean of 478.5 pg/10 6 cells (range of 107.6–1079.3 pg /10 6 cells/24 h) compared to mock transfected CIK cells (31 pg/10 6 cells) ( P = 0.05). After co-culturing with DC their functional ability was assessed in vitro by a cytotoxicity assay. On comparison with non-transfected CIK cells co-cultured with DCs (36.5 ± 5.3 %), transfected CIK cells co-cultured with DC had a significantly higher lytic activity of 58.5 ± 3.2% ( P = 0.03) against Dan G cells, a human pancreatic carcinoma cell line. | Introduction Advances in the characterization of cytokines and tumor antigens, coupled with our increasing ability to manipulate gene expression, have fostered a new era of tumor immunotherapy [ 1 ]. Interleukin-2 (IL-2) affects a variety of components of the cellular immune system, including B cells and macrophages by inducing the secretion of tumor necrosis factors (TNF) α and β and interferon-γ. Mainly, IL-2 is responsible for the proliferation of T cells. In animal and some human studies, systemic administration of IL-2 has antitumor effects, mediated by cytotoxic effector cells (such as lymphokine-activated killer – LAK cells and cytotoxic T lymphocytes) [ 2 ]. Such systemic administration often induces high toxicity and is shown to be inferior to local continuous production of cytokine, for recruitment of T cells [ 3 ]. Cytokine-induced killer cells (CIK) are non-major histocompatibility complex-restricted cytotoxic lymphocytes generated by incubation of peripheral blood lymphocytes with anti-CD3 monoclonal antibody, interleukin (IL)-2, IL-1 and interferon gamma (IFN-γ). CIK represent cells with high antitumor cytotoxicity in vitro and in vivo [ 4 ]. CIK cells possess enhanced cytotoxic activity as compared to standard lymphokine activated killer (LAK) cells [ 5 , 6 ]. CIK cells, express CD4 (45.4+/-3.2) % and CD8(47.7+/-11.0%) markers. It has been shown that NKT cells co-expressing CD3 and CD56 markers on their surface represent the major cytotoxic subset of CIK cells [ 7 ]. These NKT cells are derived from T cells [ 5 ]. Because of the increase in cytotoxicity and high proliferative response, CIK cells have a 73-fold increase in total lytic units per culture as compared to IL-2-stimulated LAK cells. Gene transfers of cytokine genes to CIK and tumor cells have been extensively studied. CIK cells transfected with cytokine genes have shown to induce antitumor effects [ 8 ]. Dendritic cells (DC) are specialized antigen-presenting cells located throughout the human body. They represent heterogeneous cell population, residing in most peripheral tissues where they represent 1–2% of the cell numbers. In the absence of ongoing inflammatory and immune responses, dendritic cells constitutively patrol through the blood, peripheral tissues, lymph and secondary lymphoid organs [ 9 ]. Morphologically, mature DC are large cells with elongated and stellated processes. They express high levels of MHC I and II, CD11 a, b, c, CD40, CD54, CD58, CD80, CD83, CD86. The most typical markers at present are MHC I, II and co-stimulatory markers such as CD80, CD86, [ 10 ] which present signals to CD4 and CD8 positive T cells. Once T cells are activated after interaction with a DC exhibiting the appropriate tumor-associated peptide antigen and class I molecule, they kill other cells that express these molecules such as tumor cells. Exocrine pancreatic carcinomas have a very poor prognosis and a resistance to conventional therapy. This is mainly induced by lack of immuno-competent cells. Therefore, it might be beneficial for patients with pancreatic cancer to induce an immune attack against the tumor by inserting a cytokine gene into the immunological effector cells. The aim of this study was to evaluate the antitumor immune responses of a cytokine immunotherapy using gene transfer to provide continuous and local cytokine production and therefore showing an improved cytotoxic effect against pancreatic cancer cells. Material and methods Generation of dendritic cells DC were generated as described before [ 4 , 7 ]. Blood was drawn according to our protocol accepted by the local ethics committee from healthy volunteers. Briefly, peripheral blood lymphocytes were isolated from buffy coats by Ficoll density gradient centrifugation (Lymphoprep, Nycomed, Oslo Norway). These cells were allowed to adhere in six-well-plates at a density of 5 × 10 6 cells/ml for one hour at 37°C in complete RPMI 1640 with 10% heat-inactivated fetal calf serum, 100 U/ml penicillin and 100 μg/ml streptomycin. The non-adherent cells were collected for generating CIK cells. The adherent cells were cultured in 2 ml RPMI 1640 with autologous, heat-inactivated serum, 750 IU GM-CSF and 500 IU IL-4 (Essex Pharma, Nürnberg, Germany), 100 U/ml penicillin and 100 μg/ml streptomycin per well for seven days for generating DC. The media along with the necessary cytokines were changed every third day. Generation of CIK CIK cells were generated as described previously [ 11 ]. In brief, non-adherent Ficoll separated human peripheral blood mononuclear cells derived from healthy individuals were prepared and grown in RPMI 1640 medium (Gibco BRL, Berlin, Germany), containing 10% fetal calf serum (Gibco BRL), 25 mM Hepes, 100 U/ml penicillin and 100 μg/ml streptomycin. One thousand IU/ml human recombinant interferon γ (Boehringer Mannheim, Germany) was added on day 0. After 24 hrs of incubation, 50 ng/ml of an anti-CD3 (Orthoclone OKT 3, Cilag GmbH, Sulzbach, Germany), 100 U/ml interleukin-1β and 300 U/ml interleukin-2 (R and D Systems, Wiesbaden, Germany) were added. Cells were incubated at 37°C in a humidified atmosphere of 5% CO 2 and sub-cultured every third day in fresh complete medium with 300 U/ml IL-2 at 3 × 10 6 cells/ml. CIK cells were harvested on day +7 and were co-cultured for seven days with autologous DC at a stimulator (DC) to responder (CIK) ratio of 1:5. Cell lines The human pancreatic carcinoma cell line DAN-G was purchased from DSMZ (Deutsche Sammlung für Zellkultur, Braunschweig, Germany). The cells were maintained in RPMI 1640 supplemented with 10% fetal calf serum (FCS, PAA) 100 U/ml penicillin and 100 μg/ml streptomycin (Seromed, Jülich, Germany) and grown at 37°C in a humidified atmosphere of 5% CO 2 . Preparation of IL-2 plasmid cDNA of human IL-2 was cloned in the plasmid pMTV.05 (Invitrogen, Karlsruhe, Germany). The recombinant pMTv-hIL-2 (referred to as pIL-2) was transformed and the plasmid was eluted using a mini-prep column (Qiagen GmbH, Hilden Germany) according to the manufacturer's protocol. Pulsing of DC DCs were pulsed with tumor lysate of Dan-G cells on day +5 [ 12 ]. Gene transfer by nucleofection CIK cells were subjected to a combination of electrical parameters and specific solution to deliver the DNA directly to the cell nucleus under mild conditions by using a commercially available nucleofection system on day 10 according to manufacturer's protocol (Nucleofector Amaxa Biosystems GmbH, Cologne, Germany). Five times 10 6 CIK cells were nucleofected in an electroporation cuvette along with pre-warmed nucleofector solution and 3 μg of pMTV-hIL-2 using the programme U-14. Once nucleofected, CIK cells were transferred into fresh pre-warmed media with the necessary cytokines and serum. IL-2 measurement Cell culture supernatants from the nucleofected and non-nucleofected CIK cells were sampled at 24 hrs and 48 hrs, respectively. An enzyme linked immunosorbent assay for IL-2 with matched antibody pairs was performed according to the manufacturer's instructions (R and D Systems, Wiesbaden, Germany). Cytotoxicity assay A DELFIA EuTDA ® non-radioactive cytotoxicity assay was used as a fluorometric alternative to the 51 Cr release assay (Perkin Elmer Wallac Life Sciences, Brussels, Belgium). The assay is based on loading target cells with a fluorescence enhancing ligand. After cytolysis the ligand is released and introduced to the DELFIA ® Europium solution. The measured signal correlates directly with the amount of lysed cells [ 13 ]. Each experiment was performed in triplicates and the mean value was calculated. After incubation, 20-microl aliquots from all wells are transferred to a fresh 96-well plate. To each well of the plate, 180 microl of the Europium solution mix is added and incubated at room temperature for 15 min on a shaker. Fluorescence data are collected using a 96-well plate in a time-resolved fluorometer (PerkinElmer, Brussels, Belgium). Maximum release was obtained by incubating Dan-G cells with 1% lysis Buffer (Perkin Elmer Wallac Life Sciences, Brussels, Belgium). Target cells without effector cells are used as negative control (spontaneous release). Specific releases are calculated as percentage cytotoxicity = experimental release (counts) minus spontaneous release (counts) divided through maximum release (counts) minus spontaneous release (counts) of target cells. Students't' test was applied. A P value < 0.05 was considered significant. Results In vitro generation of DC and CIK DC were generated from CD14 + monocytes using GM-CSF and IL-4. Adherent cells showed cytoplasmic processes typical for DC. After co-culture with CIK cells they formed typical cluster. Flow cytometry showed CD14 negative populations, expressing markers typical for DC (CD80 + , CD83 + , CD86 + and HLA-DR expressing cells). The CIK cells were phenotyped with antibodies against CD3, CD8, CD16, CD40L, CD56, HLA-ABC and HLA-DR. Data was similar to other studies by our group (data not shown) [ 5 , 11 , 14 ]. Transfection efficiency Transfection efficiency was determined by eGFP expression analysis using a fluorescence activated cell sorter. Viable cells were determined by propidium iodide staining. Nucleofection efficiency for eGFP gene transfer into the stimulated CIK cells resulted in a transient expression of 43 +/- 3.8% of the cells after 24 hours. 17% of the cells were not viable after transfection. The amount of IL-2 was the maximum after 24 hrs. An irrelevant plasmid containing eGFP was nucleofected and compared to the plasmid containing the IL-2 insert in various samples (n = 8). Nucleofection resulted in the production of IL-2 with a mean of 478.5 pg /10 6 cells (range of 107.6–1079.3 pg /10 6 cells/24 h) compared to irrelevantly transfected (containing eGFP) CIK cells (31 pg/10 6 cells) ( P = 0.05). CIK cells secreting IL-2 were co-cultured from days +7 to +14 with DC, 10 days of age. Cytotoxicity of effector cells was analyzed. Co-culture of effector cells with DC led to an increase in cytotoxic activity as measured in a Eu-release assay using Dan-G. Eu-release in co-cultured CIK cells transfected with pIL-2 and DC was 58.5 ± 3.2% at an effector:target ratio of 1:40 (Fig. 1 ) compared to non-transfected CIK cells co-cultured with DC (36.5 ± 5.3%, P = 0.03). In order to further enhance cytotoxic activity DC were pulsed with tumor lysate of Dan-G cells on day +5. However, lytic activity (50.3%) was not significantly enhanced ( P = 0.33) when compared to non-transfected cells (Fig. 1 ). Lytic activity of DCs pulsed with tumor lysate and co-cultured with non-transfected CIK cells was 48.9% where as DC pulsed with tumor lysate and co-cultured with CIK cells transfected without the plasmid was 46.3% and CIK cells alone 40.3% (Fig. 1 ). Figure 1 Cytotoxic activity of immunological effector cells that had been co-cultured with DC against Dan-G pancreatic carcinoma cells. Immunological effector cells from a donor were co-cultured from days +10 to +14 with autologous DC cultures seven days of age, as described in materials and methods. DC were pulsed at day +7 with 200 ng/ml of tumor lysate. Cytotoxic activity at various effector to target cell ratios was measured by Europium release assay. Dan-G cells were used as targets. Data represent results of three separate experiments and are shown as mean. CIKs = CIKS cells only DC+CIKS = naive DC co-cultured with CIK cells DC+CIKSpIL-2 = naive DC co-cultured with CIK cells nucleofected with pIL-2 DC-Tu+CIKS = DC pulsed with tumor lysate and co-cultured with CIK cells DC-Tu+CIKSpIL-2 = DC pulsed with tumor lysate and co-cultured with CIK cells nucleofected with pIL-2 DC-Tu+CIKS (nucleofected) = DC pulsed with tumor lysate and co-cultured with CIK cells nucleofected without plasmid Discussion In this report, transfection of CIK cells with IL-2 demonstrated a prominent augmentation of antitumor immunity in vitro against pancreatic carcinoma cell lines via secreting significant amounts of IL-2. Ductal pancreatic adenocarcinoma is the fourth leading cause of cancer death in the Western world. Unfortunately, recent advances in diagnostics, staging, and therapy have not resulted in significant improvements. Thus, new approaches are necessary to improve the outcome of patients with exocrine pancreatic cancer. CIK cells are the most potent mediators of the lyses of autologous and allogeneic cancer cells in vitro in a non MHC restricted fashion [ 15 ], have a higher antitumor toxicity as compared to standard lymphokine activated cells [ 5 , 6 , 15 ] and may be suitable to remove tumor cells resistant to chemotherapy [ 8 ]. Therefore, they are ideal candidates for further enhancing cytotoxic activity. CIK cells have been shown to upregulate DC specific markers [ 16 ]. Transgene candidates to potentially activate systemic immune response include genes encoding for co-stimulatory molecules, lymphotactic chemokines, allogeneic MHC molecules, or cytokines like IL-2. Because of the serious toxicity of systemically administered IL-2 observed in clinical practice [ 17 ] it can be expected that local expression of IL-2 is less harmful to the patient than systemic administration to trigger the immune system. In this regard, adenoviral-mediated expression of IL-2 cytokine gene in several tumor models has been found to induce strong and specific antitumor responses [ 18 ] by stimulating immune cells including T and natural killer cells. But adenoviral transfection may raise safety questions in human gene therapy. Therefore, we were interested in evaluating the potential of IL-2 non-viral transfected CIK cells for their ability to stimulate and activate immunologic effector cells. The use of gene transfected lymphocytes were hampered by a poor efficiency of gene transfer in lymphocytes and a down regulation of cytokine expression [ 19 ]. In contrast, nucleofection is a fast and cost-effective method for transfection of large amounts of cells. Here, nucleofection resulted in a significant higher production of IL-2 compared to mock transfected CIK cells ( P = 0.05). This results matches perfect to a previously reported result by our group [ 8 ]. To the best of our knowledge no further report about nucleofection of CIK cells were available. We then showed, that transfection of CIK cells with IL-2 enhances cytotoxic activity (Fig. 1 ) compared to non-transfected CIK cells co-cultured with DC. No significant cytotoxic activity was seen when DC were pulsed with tumor lysate of Dan-G cells were used with IL-2 transfected CIK cells. This may be due to inhibitory factors in the tumor lysate which may contribute to a decrease in lytic activity. This effect is due to increased amounts of CIK cells during co-cultivation of IL-2 secreting CIK cells with DC. IL-2 secretion by the CIK cells enhances the NK cell antitumor activity [ 20 ]. NK cells proliferate in the presence of IL-2 [ 21 ]. This led to a higher amount of effector cells resulting in a higher cytotoxic activity. This effect of inducing proliferation of tumoricidal lymphocytes is well known and the most important biologic effect of IL-2 on immune cells. The reproducible observation that virtually all malignant cells can be lysed by IL-2 stimulated lymphocytes in a manner directly related to the intensity of IL-2 administration encouraged the pursuit of aggressive, intensive clinical trials, especially in renal cell carcinoma and melanoma. Several authors have shown the efficacy of transfecting primary tumor cells and tumor cell lines with plasmid DNA/lipid complexes [ 22 ]. Local production of high concentrations of IL-2 and IFN-alpha at the tumor site was more effective in preventing tumor growth than systemic administration in patients with metastatic renal cell carcinoma [ 22 ]. There are several reports introducing IL-2 producing genes into pancreatic cancer, but there are no reports about IL-2 secreting lymphocytes functioning as immune enhancer cells. Therefore, our report is the first describing CIK cells to have enhanced in vitro cytolytic activity via non-viral interleukin-2 gene transfer against pancreatic cancer cell lines. Direct delivery of plasmid IL-2 gene to the established tumors in mice showed an increase in both early and long term survival [ 3 ]. Preclinical efficacy studies in a renal cell carcinoma, murine model also showed that direct intra-tumoral administration of an IL-2 plasmid DNA/DMRIE/DOPE complex resulted in the generation of tumor specific lymphocytes and complete tumor regression [ 23 ]. It is reasonable too, that these effector cells given in a pancreatic carcinoma model should enhance cytotoxic activity. These investigations are ongoing. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516021.xml |
554979 | CXCR3/CXCL10 interactions in the development of hypersensitivity pneumonitis | Background Hypersensitivity pneumonitis (HP) is an interstitial lung disease caused by repeated inhalations of finely dispersed organic particles or low molecular weight chemicals. The disease is characterized by an alveolitis sustained by CD8(+) cytotoxic T lymphocytes, granuloma formation, and, whenever antigenic exposition continues, fibrosis. Although it is known that T-cell migration into the lungs is crucial in HP reaction, mechanisms implicated in this process remain undefined. Methods Using flow cytometry, immunohistochemistry, confocal microscopy analysis and chemotaxis assays we evaluated whether CXCL10 and its receptor CXCR3 regulate the trafficking of CD8(+) T cells in HP lung. Results Our data demonstrated that lymphocytes infiltrating lung biopsies are CD8 T cells which strongly stain for CXCR3. However, T cells accumulating in the BAL of HP were CXCR3(+)/IFNγ(+) Tc1 cells exhibiting a strong in vitro migratory capability in response to CXCL10. Alveolar macrophages expressed and secreted, in response to IFN-γ, definite levels of CXCL10 capable of inducing chemotaxis of the CXCR3(+) T-cell line. Interestingly, striking levels of CXCR3 ligands could be demonstrated in the fluid component of the BAL in individuals with HP. Conclusion These data indicate that IFN-γ mediates the recruitment of lymphocytes into the lung via production of the chemokine CXCL10, resulting in Tc1-cell alveolitis and granuloma formation. | Background Hypersensitivity pneumonitis (HP) is an interstitial lung disease (ILD) caused by the inhalation of and sensitization to a variety of environmental organic antigens. The immune mediated nature of the disorder is testified to by the characteristic sequel of events taking place in the lung after antigenic inhalation: an acute pulmonary neutrophilia occurs early followed by an interstitial T-cell infiltration of CD8 T-cell showing a limited expression of the T-cell receptor [ 1 ]. A number of data point to chemokines as orchestrators of inflammatory disorders which are characterized by a massive accumulation of immunocompetent cells within affected organs, including the lung [ 2 ]. Chemokines, which can be divided into four groups based on the positioning of the cysteine residues in the mature protein [ 3 - 6 ], induce directional migration of immune cells through their interactions with G-protein coupled receptors. Three chemokines induced by IFN-γ, IFN-γ-inducible protein-10 (IP-10, CXCL10), monokine induced by IFN- (Mig/CXCL10), interferon-inducible T-cell α-chemoattractant (I-TAC/CXCL11) bind to the CXCR3 receptor molecule which is expressed by activated T lymphocytes and natural killer cells [ 7 , 8 ]. We have recently found that CXCR3 is expressed in vivo by CD4+ Th1 infiltrating the lung of patients with sarcoidosis and by T cells accumulating in the pulmonary parenchyma of lung-transplant recipients with rejection episodes [ 9 , 10 ], providing evidence that CXCR3 expression constitutes an important mechanism in the regulation of T-cell migration to the lung. Furthermore, recent data in the animal model suggest that CXCR3/CXCL9, CXCL10, CXCL11 interactions are central in the pathogenesis of hypersensitivity reactions to Saccharopolyspora rectivirgula (SR) and successive granuloma formation [ 11 ]. Using immunohistochemical studies of tissue sections and a flow cytometry evaluation of cells recovered from the bronchoalveolar lavage (BAL), we studied the role of CXCR3/CXCL10 interactions in the regulation of T-cell migration into the lung of patients with hypersensitivity pneumonitis. We have shown that CXCR3 is expressed by T cells accumulating in the lower respiratory tract of patients with this hypersensitivity disorder. In addition, we found that signalling of CXCR3 with CXCL10 induces the in vitro migration of CXCR3(+)T cells. The ligand CXCL10 can be detected in pulmonary macrophages and is released by these cells. Materials and Methods Study population 12 HP patients were included in the study (9 males and 3 females; mean age 38.3 ± 6.4 yr). The majority of the patients had farmer's lung disease (10 patients); 1 patient had bird fancier's lung, 1 patient had mushroom worker's lung. The following criteria for HP diagnosis were used: a) history of exposure to HP antigens, b) a symptomatic acute episode with chills, fever, cough, breathlessness 4 to 8 hours after exposure to specific antigens, c) radiological features (mainly diffuse reticular pattern) and/or a functional pattern of interstitial lung disease, and d) evidence of antibodies against S. rectivirgula in all except one case (bird fancier's lung). Each patient underwent bronchoscopy for transbronchial biopsy (TBB) and BAL analysis. BAL was performed according to the technical recommendations and guidelines for the standardization of BAL procedures [ 12 ]. Briefly, a total of 200 ml of saline solution was injected in 25-ml aliquots via fiberoptic bronchoscopy, with immediate vacuum aspiration after each aliquot. Immediately after the BAL, the fluid was filtered through gauze and the volume measured. A volume of 100-200 ml of BAL recovery and a sample of 50% of the instilled volume with a minimum of 50 ml was considered acceptable. The percentage of BAL recovery was 54.9% ± 4.2. Cells recovered from the BAL were washed 3 times with PBS, resuspended in endotoxin tested RPMI 1640 (Sigma Chemical Co., St. Louis, MO) supplemented with 20 mM HEPES and L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% FCS (ICN Flow, Costa Mesa, CA) and then counted. A standard morphological and immunologic analysis of BAL cellular components was performed and included cell recovery, differential count of macrophages, lymphocytes, neutrophils, and eosinophils, and flow cytometry analysis of the CD4/CD8 BAL T-cell ratio. Five healthy controls were selected (3 men and 2 women; average age 37.3 ± 4.3 yr; 2 non-smoking healthy adults and 3 non-smoking subjects evaluated for complaints of cough without lung disease). They showed normal physical examinations, chest X-rays, lung function tests and BAL cell numbers. Purification of alveolar macrophages and T cells Alveolar macrophages (AMs) and T cells were enriched from the BAL cell suspensions by rosetting with neuraminidase-treated SRBC followed by F/H gradient separations and residual CD3 + lymphocytes were removed using high-gradient magnetic separation columns (Mini MACS, Miltenyi Biotec, Germany) [ 13 ]. Following this multistep selection procedure more than 95% of the above cells were viable, as judged by the trypan blue exclusion test. Staining with mAb showed that more than 99% of purified lymphocytes were CD3+ T cells. Monoclonal antibodies and cytokines The commercially available conjugated or unconjugated mAbs used belonged to the Becton Dickinson and PharMingen series and included: CD3, CD4, CD8, isotype matched controls. Anti-IL-4 and anti-IFNγ mAbs were purchased from PharMingen (San Diego, CA). Purified rabbit anti-human CXCL10 polyclonal antibody (R&D Systems Inc, Minneapolis, MN) and anti-hCXCR3 mAb (R&D Systems Inc) were also used. Immunohistochemical analysis of CXCR3+ cells and CXCL10 producing cells Open lung biopsies from 8 patients with clinical and histological diagnosis of hypersensitivity pneumonitis were studied by immunohistochemistry for the immunophenotype characterization of inflammatory cells and for CXCR3 and CXCL10 expression. Immunohistochemistry for the characterization of inflammatory infiltrate was carried out using the following antibodies (Dako Glostrup, Denmark): CD45 (1:20), CD43 (1:40), CD45RO (1:100), CD20 (1:100), CD3 (1:50), CD68 (1:50), CD4 (1:100), and CD8 (1:100). The immunoreaction products were developed using the avidin-biotin-peroxidase complex method. Immunostaining for CXCR3 was performed as previously described. Briefly, after the microwave antigen retrieval procedure and neutralization of endogenous peroxidase activity, the slides were incubated with primary antibody for 1 hr in a humidified chamber at 37°C (anti-hCXCR3 mAb 1:100). Immunoreactivity was detected using biotinylated secondary antibodies incubated for 45 min followed by a 30 min incubation with avidin-peroxidase and visualized by a 7 min incubation with the use of 0.1% 3,3'-diaminobenzidene tetrahydrochloride as the chromogen. Parallel control slides were prepared either lacking primary antibody or lacking primary and secondary antibodies, or stained with normal sera to control for background reactivity. The intensity of antibody staining was classified in three groups: strong, weak, negative. Confocal microscopy for the identification of CXCR3+ cells Paraffin sections were prepared for immunofluorescent labelling. Briefly, primary antibodies against CD3 and CXCR3 (1:100 diluted and 1:100 diluted in phosphate-buffered saline with 5 g/L bovine serum albumin and 1 g/L gelatine, respectively) and secondary antibodies (goat anti-mouse IgG and donkey anti-goat IgG) conjugated with TEXAS red or ALEXA 488 (Sigma) were used. Double labelling using both antibodies on the same section was performed. Primary antibodies and secondary antibodies were incubated for 1 h at room temperature. Nuclear staining was carried out with DAPI (Sigma) in PBS. Slides were stored at 4°C and analysed within 24 h. As a control, the primary antibody was omitted. Immunofluorescence was evaluated with a confocal microscopy (Biorad 2100 Multiphoton; Hercules, CA), We used an argon laser at 488 nm in combination with a helium neon laser at 543 nm to excite the green (CD3) and red (CXCR3) fluorochromes simultaneously. Emitted fluorescence was detected with a 505–530 nm band pass filter for the green signal and a 560 nm long pass filter for the red signal. Images were analyzed using the Adobe Photoshop 7.0 program. Phenotypic evaluation of BAL cells The frequency of BAL cells positive for the above reagents was determined by overlaying the flow cytometry histograms of the samples stained with the different reagents as previously reported [ 12 ]. Cells were scored using a FACScan ® analyzer (Becton Dickinson), and data were processed using the Macintosh CELLQuest software program (Becton Dickinson). The expression of cytoplasmic cytokine was evaluated following permeabilization of cell membranes using 1:2 diluted PermeaFix (Ortho, Raritan, NJ) for 40 min. After permeabilization procedures anti-IL-4, anti-IFN-γ and anti-CXCL10 antibodies were added. Since pulmonary cells bear cytoplasmic cytokines in a unimodal expression pattern, indicating that the entire cell population exhibits relatively homogeneous fluorescence, the percentage of positive cells does not represent the most accurate way of enumerating positive cells. Mean fluorescence intensity (MFI) was used to compare the positivity of these specific antigens on different cell populations. To evaluate whether the shift of the positive cell peak was statistically significant, the Kolmogorov-Smirnov test for analysis of histograms was used according to the Macintosh CELLQuest software user's guide (Becton Dickinson). For immunofluorescence analysis, control IgG1 and IgG2a and IgG2b were obtained from Becton-Dickinson; control rat antiserum consisted of ascites containing an irrelevant rat IgG2b; control rabbit antiserum consisted of rabbit IgG (purified protein) purchased from Serotec (Serotec, U.K.); goat-anti-rabbit IgG and goat F(ab')2 anti-rat IgG were obtained from Immunotech (Marseille, France). Determination of IP-10/CXCL10 and Mig/CXCL9 mRNA levels Each PCR product was analysed and quantitated by Bio-Rad's Image Analysis System Gel Doc using Quantity One software (Bio-Rad, Hercules, CA). Briefly, the images of the gels were acquired from the Gel Doc system densitometer and saved in digitised forms to perform volume analysis. The intensity of each band was differentiated by the intensity of the background, whose value was subtracted from each individual band and the resulting PCR product value was expressed in mm*mm*intensity of the pixels of the specific band in the gel. Generation of macrophage supernatants To verify the ability of AMs to release CXCL10, AMs (1 × 10 6 /ml) were isolated from the BAL of HP patients, resuspended in RPMI medium and cultured for 24 hr in 24-well plates at 37°C in 5% CO 2 . In separate experiments AMs were stimulated with IFN-γ (100 U/ml), PMA (10 ng/ml) and LPS (10 μg/ml; Difco Lab., Detroit, MI). Following the incubation period, supernatants were harvested, filtered through a 0.45 μm Millipore filter and immediately stored at -80°C. At the end of the culture time AM viability was always greater than 95%. Chemotactic activity of supernatants was determined as reported below. Migration activity of pulmonary T cells in response to CXCLIO T-cell migration was measured in a 48-well modified Boyden chamber (AC48 Neuro Probe Inc., USA). The chamber is made of two sections: different chemotactic stimuli were loaded in the bottom section while cells were added in the top compartment. Polyvinylpyrrolidone-free polycarbonate membranes with 3 to 5 μm pores (for lung T cells obtained from HP patients and the CXCR3+ and CXCR3- T-cell lines, respectively) (Osmonics, Livermore, CA) and coated with fibronectin were placed between the two chamber parts. Only the bottom face of filters was pretreated with fibronectin; the fibronectin pretreatment maximizes attachment of migrating cells to filters, avoiding the possibility that they may not adhere. Using this procedure in preliminary experiments we demonstrated that only a trivial number of cells may be recovered in the bottoms of the wells. To avoid the shedding of fibronectin, fibronectin-treated filters were extensively washed. In preliminary experiments, fibronectin-treated filters did not induce spontaneous chemotaxis in absence of chemokines. To evaluate the migratory properties of pulmonary T lymphocytes rhIP-10/CXCL10 (200 ng/ml) were used. The CXCR3- and CXCR3+ cell lines (300-19, kindly provided by Dr. B. Moser, Theodor-Kocher Institute, University of Bern, Switzerland) were used as negative and positive controls. 30 μl of chemokines or control medium were added to the bottom wells, and 50 μl of 5.0 × 10 6 cells/ml T cells or CXCR3-/+ cells resuspended in RPMI 1640 were added to the top wells. The chamber was incubated at 37°C with 5% CO 2 for 2 hrs. The membranes were then removed, washed with PBS on the upper side, fixed and stained with DiffQuik (Dade AG, Düdingen, Switzerland). Cells were counted in three fields per well at 800× magnification. All assays were performed in triplicate. In blocking experiments, cell suspensions were preincubated before chemotaxis assay for 30 min at 4°C with anti-human CXCR3 mAb at a concentration of 20 μg/ml. Chemotactic activity of the fluid component of BAL and macrophage supernatants The CXCR3(-) and CXCR3(+) cell lines were also used to evaluate both the chemotactic activities of macrophage supernatants and the fluid component of BAL samples. Supernatants from cell cultures and the fluid components of BAL were obtained as reported above and used undiluted; different concentrations of CXCL10 were utilized as a positive control. Chemotactic assays were performed as reported above. In blocking experiments, anti-CXCL10 was added to the cell supernatants before chemotaxis assay at a concentration of 20 μg/ml. Statistical analysis Data were analysed with the assistance of the Statistical Analysis System. Data are expressed as mean ± SD. Mean values were compared using the ANOVA test. A P value <0.05 was considered as significant. Results Immunohistochemical analysis of CXCR3 expression in lung biopsies In all cases typical pathological examination showed features with poorly formed non-necrotizing granulomas and widespread thickening of the alveolar walls by a diffuse lymphocytic infiltrate. Pleural lymphoid aggregates were seen in a few cases and pleural lymphoid aggregates were seen in a few samples. Diffuse interstitial lymphocytic infiltrates were characterized by an accumulation of T cells and a few B-lymphocytes. Sub-pleural and peri-bronchiolar nodules consisted mostly of T lymphocytes mainly represented by CD8 cytotoxic T lymphocytes which strongly stained for CXCR3 in all cases (Figure 1A and 1B ). Marked CXCR3 immunostaining was also seen in peribronchial lymphocytic infiltrate and in the interstitial non-necrotising granuloma (Figure 2A and 2B ) Both interstitial and intra-alveolar macrophages (CD 68 positive) showed weak or negative CXCR3 staining and multinucleated giant cells always stained negatively (Fig. 2C inset). Endothelial and epithelial cells close to more intense lymphocytic infiltrate were sometimes positively marked. Figure 1 Immunohistochemistry for CD8 and CXCR3 in lung biopsy from HP patient. Most lymphocytes positive for CD8 (panel a) and CXCR3 (panel b) were clearly visible in a subpleural focus. Original Magnification × 25. Figure 2 Immunohistochemistry for CXCR3 in lung biopsy from HP patient: positive marked lymphocytes while negative or weak staining macrophages were also seen in peribronchial space (panel A) and in the setting of non-necrotizing interstitial granuloma (panel B) (original magnification × 50). Note negative or weak staining of macrophages and giant cells forming the central core of the granuloma (inset panel C, original magnification × 100). Confocal microscopy analysis of lung biopsies confirmed that lymphocyte infiltrates were formed by T cells coexpressing CXCR3 (Figure 3 panels A, B and C). Figure 3 Immunofluorescence confocal laser scanning microscopy analysis shows the presence of CXCR3 (panel B, red) on CD3+ T cells (panel A, green). Panel C shows the overlay image of A and B in yellow (original magnification × 1000). Morphological and phenotypical features of cells obtained from the BAL Morphological and phenotypical features of cells obtained from the BAL of 12 patients with HP and 5 controls are reported in Table 1 . All HP subjects showed a high intensity lymphocytic alveolitis sustained by CD8(+) Tc1 cells (Table 1 and Figure 4A and 4B ). These cells were CXCR3(+) and bore IFN-γ but not IL-4 receptor (Figure 4C ). Furthermore, pulmonary T cells expressed activatory molecules such as CD103 and IL-12β2 receptor (Figure 4C ). The percentage and absolute number of BAL CXCR3(+) was significantly higher in HP patients with respect to control subjects (Table 2 ). Table 1 Broncholaveolar findings in patients with hypersensitivity pneumonitis and controls Study population Cell Recovery Lymphocytes CD4 T cells CD8 T cells cells × 10 3 /ml % % cells × 10 3 /ml % cells × 10 3 /ml HP (n. 12) 351.9* ± 62.3 44.4** ± 8.1 25.6** ± 5.3 38.6*** ± 9.3 53.7** ± 6.3 83.4*** ± 8.5 Controls (n. 5) 138.6 ± 12.7 8.2 ± 2.2 48.3 ± 3.2 5.4 ± 0.9 23.7 ± 2.2 2.5 ± 0.3 Significance as follows: *p < 0.05; **p < 0.01; ***p < 0.001 Figure 4 The flow cytometry profile of BAL T cells recovered from 2 representative patients with hypersensitivity pneumonitis and a control subject. BAL T cells were gated as shown in panel A. In patients with hypersensitivity pneumonitis the majority of lymphocytes were CD8(+) T cells (panel B). Panel C shows that BAL T cells from patients with hypersensitivity pneumonitis are CD45RO(+) T cells which express CXCR3, IFN-γ but not IL-4, or other activation markers including CD103 and IL12Rβ2. Table 2 Expression of CXCR3 by CD8+ T cells and expression of IP-10/CXCL10 and Mig/CXCL9 mRNAs by alveolar macrophages from patients with hypersensitivity pneumonitis and controls* Study population CXCR3+/CD8+ T cells IP-10/CXCL10 Mig/CXCL9 % cells × 10 3 /ml mRNA levels* mRNA levels HP (n. 6) 51.7** ± 5.9 80.1*** ± 7.7 2.55** ± 0.14 2.25** ± 0.25 Controls (n. 4) 23.7 ± 2.2 2.5 ± 0.3 0.70 ± 0.05 0.48 ± 0.06 * Band intensity calculated as follows: mm × mm × pixel Significance as follows: **p < 0.01; ***p < 0.001 CXCR3 mediates pulmonary T cell chemotaxis To define the biological activities of CXCR3, highly purified T cells obtained from the BALs of patients with HP were assessed for their migratory capabilities in response to different concentrations of CXCL10. The evaluation of the migratory potential of T lymphocytes obtained from the BAL of the controls was prevented by the low number of cells recovered. For this reason, the 300-19 T-cell lines expressing high levels of CXCR3 or not expressing CXCR3 were used as positive and negative controls respectively for the in vitro chemotaxis assay (Figure 5 , panel B and C respectively). Figure 5 Chemotactic activity of CXCL10 on BAL CD8(+)/CXCR3(+)T cells highly purified from 2 representative patients with hypersensitivity pneumonitis. The assays were performed using a modified Boyden chamber in triplicate and data are given as mean ± SD. CXL10 shows significant chemotactic activity on BAL T cells (panels A) and the CXCR3(+) T-cell clone (panel B) but not on CXCR3(-) T-cell clone. As shown in panel A of Figure 4 , the migratory capability of T lymphocytes of patients with HP is regulated by CXCR3. In fact, CXCR3+ lung T cells exhibited a strong, definite migration in response to CXCL10. To further verify the functional role of the CXCL10 receptor, CXCR3+ pulmonary T cells were preincubated with anti-CXCR3 neutralizing antibody: the blocking of the receptor determined a marked inhibition of CXCL10-induced chemotaxis (panel A). These data suggest that pulmonary T lymphocytes that sustain T-cell alveolitis in patients with HP express a functional CXCR3 receptor and actively migrate in response to CXCR3 ligands. Lung macrophages express CXCR3 ligands and release ligands showing chemotactic activity on CXCR3(+) cells In order to analyse whether CXCL10 is expressed in vivo by lung cells of patients with hypersensitivity pneumonitis, BAL cells were stained with a anti-CXCL10 antibody as described above. Flow cytrometric analysis (Figure 6 , panels A and B) revealed that AMs of patients with HP express CXCL10; macrophages retrieved from control subjects lacked the CXCR3 ligand (panel C). Figure 6 Flow cytometric analysis of CXCL10 expression by AMs infiltrating the lung of patients with hypersensitivity pneumonitis. Panels A-C shows the flow cytometry profile of AMs recovered from the BAL of 2 representative patients and a control subject. AMs from patients with hypersensitivity pneumonitis (panels A and B) but not control subject (panel C) express CXCL10. Measurement of mRNA levels of the CXCR3 ligands demonstrated that unstimulated alveolar macrophages isolated from the BAL of HP subjects expressed increased mRNA levels of CXCL9 and CXCL10 with respect to macrophages obtained from control subjects (Table 2 and figure 7 ). Spearman Rank correlation coefficients between BAL T CD8(+)/CXCR3(+) T cell number and levels of CXCR3 ligands were also calculated. Interestingly, a positive correlation was demonstrated between mRNA levels of CXCL10 and CXCL9 and the absolute numbers of lung CD8(+)/CXCR3(+) T cells (r 0.815, p < 0.001 and r 0.825, p < 0.001, respectively). Figure 7 IP-10/CXCL10 and Mig/CXCL9 mRNA levels expressed by AMs recovered from the BAL of 2 representative patients and 2 control subject. AMs from patients with hypersensitivity pneumonitis express higher amounts of IP-10/CXCL10 and Mig/CXCL9 mRNA levels than control AMs. Cell-free supernatants were obtained from 24-hour cultured AMs in the presence of IFN-γ and tested for their ability to induce T-cell migration. Supernatants obtained from AMs of patients with HP exerted chemotactic activity on the CXCR3(+) cell line; the CXCR3(-) cell line did not migrate in the presence of supernatants (data not shown). The addition of an anti-CXCL10 neutralizing antibody inhibited chemotactic activities of supernatants. The inhibitory activity shown by the neutralizing antibody was not complete, suggesting that other CXCR3 ligands (CXCL9 and CXCL11) are likely to be present in supernatants. CXCR3 ligands may be demonstrated in the fluid component of BAL To assess whether CXCR3 ligands are released in vivo in the lung microenvironment, the fluid component of BAL obtained from 10 HP patients was evaluated for chemotactic activity on CXCR3(+) cell lines (Figure 8 ). Measurable biological activity was demonstrated in 7 out of 10 patients with HP; this migration was partially abrogated by an anti-CXCL10 neutralizing antibody. Figure 8 The fluid components of the BAL obtained from 4 patients with hypersensitivity patients exert chemotactic activity on the CXCR3(+) 300-19 T-cell line, indicating the presence of CXCR3 ligand(s). The migration was partially abrogated by an anti-CXCL10 neutralizing antibody (white bars). The assays were performed in triplicate, and data are given as mean ± SD. Discussion We have herein shown that CXCR3 expression represents a crucial mechanism which is involved in the recruitment of activated Tc1 cells in the pulmonary microenvironment of patients with HP. The continuous recruitment of CXCR3(+) T cells might play a role not only in the pathogenesis of T-cell alveolitis but also in favouring granuloma formation since T cells surrounding the macrophagic core of the HP granuloma expressed this chemokine receptor. This mechanism is likely to be shared by various ILDs since we and others have previously demonstrated the presence of a significant infiltrate of CXCR3(+) Th1 cells in other ILDs characterized by T cell alveolitis and granuloma formation, such as sarcoidosis and tuberculosis [ 13 , 14 ]. Our data provide definitive confirmation of the recent findings obtained in an animal model of IFN-γ-knockout (GKO) mice exposed to the particulate antigen Saccharopolyspora rectivirgula (SR) (i.e., the etiologic agent involved in the immunopathogenesis of HP reaction in the majority of our patients) [ 11 ]. While WT mice show the production of IP-10/CXCL10, Mig/CXCL9, and I-TAC/CXCL11 during the development of the classic HP reaction, GKO mice have reduced or no levels of IP-10/CXCL10, Mig/CXCL9 and I-TAC/CXCL11 in the lungs and reduced T-cell alveolitis in response to SR exposure. The present study suggests the role of CXCL10/CXCR3 and CXCL9/CXCR3 interactions in driving local CD8 immune responses to SR (Figure 9 ). A logical question is whether our data may have therapeutic implications. Because of the role of CXCR3 expression in the migration of T cells, strategies to block CXCL10 could in theory be proposed to prevent the development of HP reactions, particularly in subjects continuously exposed to inhaled antigens and thus at risk for the development of lung fibrosis. Further data are required to evaluate the in vivo role of IP-10/CXL10 in preventing or favouring pulmonary fibrosis in HP before proposing this strategy. Figure 9 Putative scheme of the effect of the CXCR3/CXL10 interactions in the lung of hypersensitivity patients. As a consequence of the antigenic stimulation APC release IFN-γ. This cytokine stimulates AM to release CXCL10, a chemokine which favours the recruitments of activated CD8(+) /CXCR3(+) Tc1 lymphocytes. These cells, which release IFN-γ, in turn contribute to maintain the activation state of APC at sites of hypersensitivity reaction. The relationship between CXCL10 release and the local production of other chemokines involved in HP immune response is another important aspect that deserves further investigation. In a murine model it has recently been shown that monocyte chemotactic protein-1 (MCP-1/CCL2) is increased in the BAL of mice challenged with SR [ 15 ]. Furthermore, Pardo et al. [ 16 ] have recently examined the expression of dendritic cell (DC)-derived CC chemokine 1 (CK1)/CCL18 in the lungs of patients with HP. CCL18 expression is significantly increased in lungs affected by HP, with higher levels in the subacute rather than in the chronic phase of the disease. Macrophages, dendritic cells, and alveolar epithelial cells are the main sources of CCL18 whose expression is induced by various profibrogenic cytokines including IL-4, IL-10, and IL-13. Interestingly, a direct correlation between the levels of tissue CCL18 and the number of lymphocytes has been demonstrated in the bronchoalveolar lavage fluids. Thus, our findings and the data of Pardo et al. [ 16 ] suggest that chemokines ordinarily induced by profibrogenic cytokines (CCL18) and chemokines induced by antifibrotic agents (IFN-γ and CXCL10) can be demonstrated in the lung of HP patients with T cell alveolitis. Whether the presence of the two chemokines is concomitant and there are common molecular mechanisms involved in the CC and CXC chemokine release is unknown. Given the heterogeneous pattern of pulmonary infiltrate during different phases of the disease, a full understanding of the interdependence of the local hyperproduction of chemoattractant molecules may help to clarify the pathogenesis of HP. In this context, it is also mandatory to investigate the production of CC chemokines in relationship with CXCL10 release in individuals with acute, subacute and chronic HP. For instance, it is known that the combination of RANTES/CCL5 and CXCL10 but not other chemokines (MIP-1α/CCL3, MIP-1β/CCL4, Mig/CXCL9, and ITAC/CXCL11) markedly increases T-cell recruitment [ 17 ]. Since there are data indicating that circulating, antigen-reactive, memory T cells, generated by previous sensitization to organic antigens, migrate into lung parenchyma in response to chemokines such as RANTES [18], it is possible that the interplay of CXCL10 with CCL5 may serve to finely tune inflammatory responses in vivo in HP lungs. Conclusion Our findings clearly indicate the effects of CXCR3/CXCL10 interactions on hypersensitivity reaction to SR antigens. Considering the importance of CD8 T cells in mediating granuloma formation and lung damage, further studies are needed in animal models to explore the therapeutic potential of CXCR3 antagonists with the ultimate goal of offering new clues for immune intervention in subjects continuously exposed to inhaled antigens and thus at risk of developing HP-related lung fibrosis. List of abbreviations AM: alveolar macrophage; BAL: bronchoalveolar lavage; CXCR3: Receptor 3 for CXC chemokines; HP: Hypersensitivity pneumonitis; IP10/CXCL10: IFN-γ-inducible protein-10; I-TAC/CXCL11: interferon-inducible T-cell αlfa-chemoattractant; Mig/CXCL10: monokine induced by IFN-γ; TBB: transbronchial biopsy. Authors' contributions CA: planned the experimental design, coordinated the research group and drafted the manuscript FC: participated in the study design and in immuohiostochemical analysis VP: participated in the study design and in the bronchoalveolar lavage execution GM: participated in the study design and in the clinical evaluation of the patients MF: participated in the study design and performed migration assays studies MM: participated in the study design and performed cellular studies AC: participated in the study design and performed flow cytometry studies IB: participated in the study design and performed flow cytometry studies RZ: participated in the study design and helped the draft of the manuscript LT: participated in the study design and helped the draft of the manuscript GS: participated in the study design and coordination of the research group | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554979.xml |
549082 | Review of "Therapy, Sensory Integration and the Autistic Child" by DS. Berger | null | Innumerable animal species produce noises of different kind and many of them sound musical to us – human beings – eliciting a variety of emotional feelings, as the frogs' chants in a rainy night, or the wild birds' songs in a sunny morning perhaps mixed with our finch's joyful trills from its small cage, or the deep thrilling whales' hissing who very few have had the chance to listen (as I once did in Colombia). Most likely, they all are just communication ways, perhaps to let the other guy know that the male is there, waiting for the female to come and meet it, because usually only males sing. Is all that really music? I do not think so, for music is unique to man and woman and only we, human beings, get together either in small or large groups just to listen to music or to make music. Music conveys a message, which only the human being can decipher in his/her own fashion; as in animals, it is no doubt a communication means. The isolated man or woman, the Robinson Crusoe, is essentially non-existent. Music is a natural human need, almost at the level of the basic biological ones (self and species preservation). The current literature on music and the human being is enormous, from artistic to technical to psychological to physiological aspects, or what have you. Music is so much embedded in us that never many studies are or will be enough. This little book, little in size but not in scope, authored by Dorita S. Berger (an accomplished concert pianist and certified music therapist, with superb academic background and solid experience), seems to cover a paucity because it falls within the reach of those who, without the requisite of being specialists are not fully outsiders; it also serves as an introduction to those who may like to proceed further in any of the several paths that converge and diverge to and from this fascinating area of human adaptation. The book consists of a foreword, a preface, 11 chapters and 2 appendices. Its foreword is unique and deep in content, indeed. Written by Donna Williams, herself a diagnosed autistic, captivating and exemplary woman, author of several publications (among them a book entitled Autism: An Inside-Out Approach , 1996, Jessica Kingsley Publishers), its sentences must be read over and over until one really apprehends its profound meaning, as her initial paragraph, where she states: "Mind and consciousness are like a gradual sunrise that takes time to come up. In the meantime, we are not stagnant, waiting for life to happen." She refers to "the music of life, felt with our bodies, long before we identify mind with self ... ... some of us stay here longer than others." And a little further down, "music in its broader sense is everywhere, and it is our first language ... it has the most important of all places in the lives of those who find the realm of the mind a place of rusty cogs and heavy effort." Donna declares that music was an exceptionally important part of her life and almost naïvely and charmingly confesses: "I could not understand three sentences in a row until I was nine". These deep thoughts and experiences remind of James Joyce's agenbite of inwit , or the wound of in-looking, the hurt of self-analysis, the painful knowing of oneself, introduced in his famous, conflicting and monumental Ulysses . Man at large is the prey of inner suffering. The preface is already in the hands of the author, where the book is introduced along with some general concepts, such as: "Entering the brain through the auditory system is a key to obtaining automatic brain attendance and response" and "physiologic changes resulting from music intervention (or stimulation) can be a major factor in bringing about a feeling of well-being". Dorita aims the text at music therapists, parents, teachers and health clinicians. However, in my opinion and as anticipated above, its audience should not be restricted to such relatively small group but rather can be much broader, as there are, for example, students or even researchers developing specific projects in the area. Already ending the preface, the author underlines quite sensibly that the book is not a recipe since music therapy is an approach to solving a physiologic or psychological problem. Chapter 1 (Introduction: Who Defines "Appropriate"?, pp17–25) is a dareful good beginning that encourages the reader to proceed. For the human animal, the expectation is to "follow the rules" of the pack and behave accordingly, and Dorita very wisely asks who defines "appropriateness" and for whose benefit the rules are designed. What is normal and typical? Are the two words interchangeable? What determines "right" and "wrong" responses? Are these not "labels" somehow imposed by society? The concepts of "right" and "wrong" are basically determined by socio-cultural standards rather than scientific givens. All this set of questions go far beyond the intended objective of the book and scrape deep in the human self, both as an individual and as a member of society. Allow me to bring about the painful, senseless, dire and cruel Irak War (April 2003): Is the Bush administration not imposing its own pack rules, denying rights as illegal for other countries (other societies or packs) while retaining those very same rights as legal for itself? Is it bad behavior for the others to have weapons while it is good behavior for the USA to have them? No doubt, Dorita touches, perhaps without realizing, a huge and essentially unsolvable problem that only true generosity and unabashed love, mostly and so far unsuccessfully called for by the core of our unfulfilled religions, might one distant and dreamt of day bring forth. Yes, excellent and well-posed question: Who defines "appropriate"? In Chapter 2 (Aspects of Autism, pp26–34) the author says that autism is a neurologically atypical manner of function, with genetic and sensory-motor implications; what appears in childhood, remains throughout life. She introduces the concept of a different world the child is immersed in and, as such, cannot possibly get connected to this "real" one, with other channels and incomprehensible rules. His/her desire for sameness is broken at worst or disturbed at best. Another good point: By and large, most of the people avoid anything bringing change, anxiety, confusion, insecurity and the like. Sameness, conservation of what we know, appears as safer. A trip to a foreign country or place carries with its sole organization an element of tension and we often think "oh, we had better call it all off". And children are particularly susceptible and sensitive to these changes (such as a new school or neighborhood). Chapters 3 (Aspects of Sensory Integration, pp35–48), 4 (Functional Adaptation Defined, pp 49–60), 5 (Understanding Basic Sensory Systems, pp 61–77) and 6 (Are You Listening? Part One: About Hearing and Listening, pp78–90) deal mainly with physiological aspects, perhaps a little in excess because it tends to overload the reader removing him/her from the central subject, which is music, as stimulus or intervention, akin in a sense to electrical or mechanical or chemical or thermic or magnetic stimulation. However, it is fine and useful. As an interested reader and somewhat also as an actor within the discipline, the comment may be taken simply as an expression of desire. One paragraph called my attention almost at the end of chapter 3: "Music does not require semantic interpretation. It simply provides an environment – a sound blanket – wrapping itself around the body and providing a sense of safety and security." Good attractive concept, but I would emphasize what I say previously: Music carries a message, it may not be semantic or absolute, but the listener somehow decodes it to suit his/her own cenesthetic state. The basic driving forces of man and woman are clearly stated in Chapter 4, two are purely biological (as anticipated above); other two pertain only to the human being, self-determination or freedom of choice and spiritual fulfillment . Let me add that one essential tool for their satisfaction is communication , which is a need, and music appears as a phenomenal way to accomplish it. The messages left by Johann Sebastian Bach, Wolfgang Amadeus Mozart or Ludwig van Beethoven are there and will be there forever, hence demonstrating unparalleled self-determination and spiritual fulfillment, even after death. Finally, accommodation, as discussed here, is a good classical physiology concept that fits very well this new frame of music and human adaptation. An appealing idea advanced in Chapter 5 refers to a possible release of chemical neurotransmitters (such as dopamine, endorphines, encephalines or others) elicited by music stimulation. The whole thing may mark a difficult but attractive research avenue. Chapter 6 is in my modest opinion and with due respects the weakest of the book: auditory scanning is definitely obscure, the section titled "sound coding tango" (what does the word "tango" do here?) does not show a relationship with the text that follows and the remaining sections are at best confusing. The last paragraph in page 89 sounds very appealing, at least to me, and I tend to agree with it ... but it lacks so far solid scientific support. In Chapter 7 (Are You Listening? Part Two: Dimensional Hearing and Erroneous Assumptions, pp 91–111) Dorita starts with the case of a four year old boy, Jason. No musical instrument really moved him, until one day he spotted a gong. It captured his love and reacted as fully unfazed by its intense volume. Quite an amazing response considering that most people (the pack) get annoyed by a gong sound, especially if it is played within a small room. Thereafter, she briefly discusses some common erroneous assumptions, as for example, "music is always fun" and "the content of a recognizable tune is being perceived by another in the same format in which it is being presented by the therapist". The latter agrees with the previous contention: each listener to suit his/her cenesthetic needs decodes the musical message. Such assumptions deserve to be taken into consideration in other environments, too. Example: A person complains that his neighbor plays piano every day after midnight and, thus, disturbs his sleep. The player argues back saying that he plays beautiful classic music loved by everybody. Who is right? The author brings also forward an interesting and novel possibility: Is there such a thing as auditory dyslexia ? There seems to be no term for a phenomenon in which sounds are retrieved in incorrect order (as in reading or writing dyslexia). Researchers may have in this another subject to investigate. After finishing this chapter and reviewing the variety of concepts and subjects touched in it, I could not but wondering that the human being enjoys with the eyes the landscape he or she moves about, simultaneously the complex auditory tract perceives the soundscape that surely accompanies the former, probably also seasoned by an odorscape , tastescape and even a touchscape ; but everything is slowly evolving in the timescape , as a multimensional function of time. The whole configures the complex scenery we live in creating what Donna calls the music of life. Chapter 8 (Elements of Music for Sensory Adaptation, pp 112–129) goes, with some tardiness in the development of the book (at least to my particular taste), into music itself, and I would like to use the term music stimulation , characterized by six properties, as Dorita explains: rhythm , melody , harmony , dynamics , timbre , and form . This is not the place to give details about these elements for the author clearly fathoms in them solidly and without a single doubt. My only comment, and just as a curiosity recalling my days as cardiovascular physiologist, refers to one teacher actually using pulsed hand-clapping (a pacemaker) to call her entire class of children to order, to subdue chaotic impulses (page 114). A fibrillating heart has its fibers in a dynamic state of dyssynchrony (as the children shouting in the classroom). No effective contraction to eject blood takes place. One or may be two electric shocks usually stop the disorder and the myocardial mass resumes its rhythmic synchronic activity. This is called a defibrillating maneuver. Are the similarities of nature not nice? As Albert Einstein used to say, the most incomprehensible thing of the world is that the world is comprehensible. Chapter 9 (Music Therapy in the Realm of Sensory Integration, pp 130–151) is more practical as it discusses the different types of musical instruments from the viewpoint of their suitability to specific therapeutic objectives. In Chapter 10 (Formulating Music Therapy Treatment for Sensory Adaptation Goals, pp 152–165), instead, several aspects are discussed, some of them somewhat superficially and not meeting the expectations developed by the subtitles, as for example "adaptive responses to environment of auditory and visual stimuli" or "auditory integration and differentiation". The last Chapter 11 (Conclusions, pp 166–175) is some kind of summary. Let me quote a wise paragraph, a caveat emptor or let the buyer beware: "Music therapy does not lend itself readily to rigorous scientific explanation. Findings are anecdotal at best, and based on observations, insights and informed assumptions. Music therapy does not replace other interventions, nor does it need to copy the goals of other therapies in order to be effective". This is why the field is fully open for serious research. I think the book meets the intent of its author (page 173): [It] "is an introduction to physiologic information and perspectives that can help the music therapist assess more astutely the behaviors and presenting problems of an atypically functioning patient." Now, and to finish this review by being a little prickly but honestly (and perhaps naïvely) constructive, let me point out a few minor aspects that might improve another edition of the book: The title does not properly reflect the content of the book. It should be named The Autistic Child, Sensory Integration and Music Therapy , for this is really the order followed in it. The physiology section could be reduced. There are also some unnecessary repetitions. The bibliography is given in three groups: specific references to each chapter, recommended references, and a general list at the end of the book. It is not practical. I would rather have a single list at the end properly referred to in the course of the text. Appendix A is the reproduction of an article by Schneck and Berger. It adds something to the book but not to its core material. Thus, it might be deleted in the future. Appendix B is a collection of four case histories. By far is too long containing unnecessary information. It should be greatly reduced and limited to the pertinent information. The music portion could be enhanced. Having said more than enough, I recommend the book and congratulate the author for her dedication, effort and beautiful activity by combining art with benefit to the many time forgotten children. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549082.xml |
524520 | The connection domain in reverse transcriptase facilitates the in vivo annealing of tRNALys3 to HIV-1 genomic RNA | The primer tRNA for reverse transcription in HIV-1, tRNA Lys3 , is selectively packaged into the virus during its assembly, and annealed to the viral genomic RNA. The ribonucleoprotein complex that is involved in the packaging and annealing of tRNA Lys into HIV-1 consists of Gag, GagPol, tRNA Lys , lysyl-tRNA synthetase (LysRS), and viral genomic RNA. Gag targets tRNA Lys for viral packaging through Gag's interaction with LysRS, a tRNA Lys -binding protein, while reverse transcriptase (RT) sequences within GagPol (the thumb domain) bind to tRNA Lys . The further annealing of tRNA Lys3 to viral RNA requires nucleocapsid (NC) sequences in Gag, but not the NC sequences GagPol. In this report, we further show that while the RT connection domain in GagPol is not required for tRNA Lys3 packaging into the virus, it is required for tRNA Lys3 annealing to the viral RNA genome. | Background During assembly of HIV-1, the major tRNA Lys isoacceptors in mammalian cells, tRNA Lys1,2 and tRNA Lys3 , are selectively incorporated into the virus [ 1 ]. tRNA Lys3 is the primer for initiating minus-strand cDNA synthesis, and its annealing to the 18 nucleotide primer binding site (PBS) region in the 5' part of the viral genome via the 3' 18 nucleotides in tRNA Lys3 complementary to the PBS, is a key step in viral replication [ 2 ]. Other regions upstream and downstream of the PBS may also anneal with additional sequences in the tRNA [ 3 , 4 ]. Both tRNA Lys3 and sites of annealing in viral RNA contain double stranded regions which may require denaturation for annealing to proceed efficiently. Nucleocapsid protein (NC) has been shown to facilitate tRNA Lys3 annealing both in vitro [ 5 , 6 ] and in vivo [ 7 ], primarily through basic amino acids flanking the first zinc finger. While NC may destabilize viral RNA secondary structure, it has been demonstrated by several groups that nucleocapsid protein does not unwind the secondary structure of tRNA in vitro , and that the protein only has very subtle tertiary structural and helix destabilization effects on tRNA Lys3 alone [ 8 - 11 ]. Although processed nucleocapsid proteins have been shown to facilitate tRNA Lys3 annealing to genomic RNA in vitro , the annealing of primer tRNA onto the genomic RNA within HIV-1, murine leukemia virus, and avian retrovirus occurs independently of precursor protein processing [ 12 - 14 ]. However, while, tRNA Lys3 is annealed efficiently in protease-negative HIV-1 (about 80% that found in wild-type virions), optimal placement on the viral genome to achieve efficient initiation of reverse transcription requires exposure of the viral genome to mature nucleocapsid protein [ 15 ]. In these protease-negative viruses, mutations in NC sequences within Gag inhibit tRNA Lys3 annealing, while mutations in NC sequences within GagPol do not, indicating the importance of Gag NC sequences in the annealing [ 16 ]. In vitro , Gag has been reported to facilitate tRNA Lys3 annealing to viral RNA as efficiently as mature NC [ 17 ]. Nevertheless, we will present evidence in this report that GagPol still plays an important role in tRNA Lys3 annealing onto the viral RNA, independent of its role in the packaging of tRNA Lys3 into the virion. We present data herein indicating that the RT connection domain, while non-essential for tRNA Lys3 incorporation into virions, is required for tRNA Lys3 annealing to the viral RNA genome Results The RT connection domain within GagPol is not required for tRNA Lys incorporation into virions, but is required for the annealing of tRNA Lys3 to the viral genome . 293T cells were transfected with protease-negative HIV-1 proviral DNA coding for either full length, protease-negative, GagPol (BH10.P-) or C-terminally deleted GagPol species. The different constructs are shown in Figure 1A , and are named according to the number of amino acids deleted from the C terminus of GagPol. Figure 1B shows Western blots of lysates of the viruses produced from the different transfections, probed with anti-CA, and shows that all forms of GagPol deletion mutants tested here are incorporated into the virion. Total viral RNA was isolated from these virions, and dot blots of this RNA were annealed with probes specific for either viral genomic RNA or tRNA Lys3 , to determine the tRNA Lys3 /genomic RNA in each viral variant. These results are shown graphically in Figure 1C , and support our previous results using COS7 cells [ 18 ], which indicate that tRNA Lys incorporation into virions is not dramatically affected until GagPol sequences including the thumb domain of RT are deleted (Δ581 and Δ715). Figure 1 The incorporation of GagPol and tRNA Lys3 into wild-type and mutant HIV-1. A . Schematic showing the deletions made in the Pol region of GagPol. Δ# designates the number of amino acid residues deleted from the C terminus of GagPol, and solid black lines represent the sequences not deleted. The RT sequence is divided into its known structural domains. The mutation D25G inactivates the viral protease. B. Western blots of viral lysates, probed with both anti-CA and anti-RT as previously described [18]. C. Incorporation of tRNA Lys3 into wild-type and mutant virions. Dot blots of viral RNA were hybridized with probes specific for tRNA Lys3 or genomic RNA, and the tRNA Lys3 :genomic RNA ratios, normalized to BH10.P- were determined by phosphorimaging. The values are the means +/- standard deviations of experiments performed three or more times. To measure the amount of tRNA Lys3 annealed in vivo to the viral RNA genome, total viral RNA was used as the source of primer/template in an in vitro reverse transcription reaction, using exogenous HIV-1 RT, dCTP, dTTP, α- 32 P-dGTP, and ddATP. This assay measures the amount of extendable tRNA Lys3 placed onto the viral genome. It is not known if all annealed tRNA Lys3 is extendable. Since the sequence of the first six dNTP's incorporated is CTGCTA, annealed primer tRNA Lys3 will be extended by 6 bases, and the extended tRNA Lys3 can be resolved and detected by one dimensional polyacrylamide gel electrophoresis (1D PAGE). These results are shown in Figure 2A , and presented graphically in Figure 2B . The left side of panel A shows that there is a linear increase in the reverse transcription signal over an almost 10 fold change in the amount of BH10.P- viral genomic RNA used in the reaction. The data in the right side of panel A indicate that C-terminal deletions of GagPol extending into the connection domain result in an 85% or greater decrease in the initiation of reverse transcription. Thus, the data in Figures 1 and 2 indicate that deletions extending into the RT connection domain do not significantly effect tRNA Lys incorporation, but do severely reduce the ability of tRNA Lys3 to be functionally annealed to the viral RNA genome. Figure 2 tRNA Lys3 annealing to viral genomic RNA. A. Total viral RNA was used as the source of primer tRNA Lys3 /viral RNA template in an in vitro reverse transcription reaction as described in Methods. Six base extended tRNA Lys3 was resolved by 1D PAGE and quantitated by phosphorimaging. Each reaction used an equal amount of viral genomic RNA, as determined by hybridization with a genomic RNA-specific probe. B. Graphic presentation of 6 base-extended tRNA Lys3 :genomic RNA ratios, normalized to BH10P-. The values are the means +/- standard deviations of experiments performed three or more times. Rescue of tRNA Lys3 annealing by GagPol As shown in Figure 3 , this annealing defect can be rescued by coexpression of full-length GagPol. 293T cells were transfected with plasmids coding for BH10P-, Δ467, or Δ486, or cotransfected with either Δ467 or Δ486 and a plasmid coding for full-length GagPol. Western blots of cell lysates probed with anti-RT or anti-β-actin are shown in panel A, while Western blots of lysates of virus produced from these cells and probed with anti-RT and anti-CA are shown in panel B. These data indicate that both full length GagPol and the truncated GagPol are incorporated into the viruses with similar efficiencies. As previously indicated in Figure 1C , the mutant virions incorporate approximately 80–85% of the tRNA Lys3 as BH10P-, but cotransfection of mutant DNA with DNA coding for GagPol gives a small increase in tRNA Lys3 packaged to over 90% of BH10P- (Figure 3C ). Figure 3 Rescue by GagPol of tRNA Lys3 annealing in mutant virions. COS7 cells were transfected with either BH10P-, Δ467.P-, or Δ486.P-, and were also cotransfected with one of these plasmids and a plasmid coding for full-length GagPol (hGagPolΔFSΔPR). A. Western blots of cell lysates, probed with anti-RT or anti-β-actin. B. Western blots of viral lysates, probed with anti-RT and anti-CA. C. Incorporation of tRNA Lys3 into wild-type and mutant virions. Dot blots of viral RNA were hybridized with probes specific for tRNA Lys3 or genomic RNA, and the tRNA Lys3 :genomic RNA ratios were determined by phosphorimaging. The values are the means +/- standard deviations of experiments performed three or more times. D,E . tRNA Lys3 annealing in wild-type and mutant virions. tRNA Lys3 annealing was measured as described in the Figure 2 legend. The values shown in E are the means +/- standard deviations of experiments performed three or more times. As shown in panels D and E, cotransfection with GagPol also moderately rescues tRNA Lys3 annealing in these mutant virions. Using equal amounts of total viral RNA as the source of primer/template in the in vitro RT assay, the ability of primer tRNA Lys3 to be extended 6 deoxynucleotides is shown in panel D, which shows the extended 6 base product resolved by 1D PAGE. Quantitation of these bands by phosphorimaging is presented graphically in panel E. As previously shown (Figure 2 ), tRNA Lys3 annealing is reduced to 12–15% that of BH10P-, but can be increased 4–5 fold by the additional presence of full-length GagPol. The fact that tRNA Lys3 annealing is only rescued by GagPol to approximately 50–55% the level of that obtained when only wild-type GagPol is present may reflect the fact that in these rescue experiments, the viral population contains approximately equal amounts of wild-type and mutant GagPol (Figure 3B ). Attempts were also made to rescue tRNA Lys3 annealing using mature RT fused to Vpr [ 19 ], but unlike full-length GagPol, the Vpr-RT was unable to rescue tRNA Lys3 annealing in the mutant virions (data not shown). Discussion In vitro studies of the interaction between purified RT and tRNA Lys3 have indicated an interaction between the RT thumb domain and the tRNA [ 20 - 22 ]. In vivo studies also indicate an important role of the RT thumb domain in GagPol in tRNA Lys3 viral packaging. tRNA Lys3 incorporation into HIV-1 is not affected by deletion of the IN domain in GagPol, nor by further deletion of the RNaseH and connection domains in RT, but is severely inhibited by further deletion of the thumb domain as well [ 18 ]. Thus tRNA Lys3 interacts with the RT thumb domain during incorporation into virions, and Gag nucleocapsid plays a role in promoting tRNA Lys3 annealing to viral RNA [ 5 - 7 ], presumably through a denaturation of annealing RNA sequences. What then is the role the RT connection domain sequence in GagPol in facilitating tRNA Lys3 annealing? One possibility, suggested by in vitro studies, is that RT plays a direct role in tRNA Lys3 annealing. Early work indicated that the in vitro annealing of primer tRNA Trp to AMV genomic RNA was promoted by the addition of AMV reverse transcriptase [ 23 ]. In a later work, in which it was demonstrated that HIV-1 RT interacted with the D arm and TΨC loop of tRNA Lys3 , HIV-1 RT was also shown facilitate the in vitro annealing of tRNA Lys3 to the PBS sequence [ 24 ]. These in vitro works suggest that RT alone can directly promote tRNA Lys3 annealing to viral RNA. Whether the RT sequences in GagPol can function similarly in vivo is not known. Alternatively, the RT connection domain may undergo interactions with Gag that may result in placing the tRNA Lys3 bound to the thumb domain in RT closer to either NC in Gag or to the genomic RNA that is bound to Gag NC. Recent work has indicated that that Pol sequences alone can bind to Gag p6 through the RT sequences in Pol [ 25 ]. Pol protein alone is sufficient for obtaining both tRNA Lys incorporation into the virus and tRNA Lys3 annealing to the viral genome at levels approximately 35% those achieved using full-length GagPol. Thus, in addition to the interactions which probably occur between Gag and homologous sequences in the Gag part of GagPol, the interaction of RT sequences in GagPol with Gag p6 could place the RT-bound tRNA Lys3 closer to Gag NC sequences and viral RNA in the packaging complex. It remains to be determined which sequences within RT bind to Gag p6, but if it were those of the connection domain, this could explain how these sequences could promote tRNA Lys3 annealing through altering the configuration of GagPol. Thus, two separate RT domains (thumb and connection) appear to be involved, respectively, in the viral incorporation of tRNA Lys3 , and its annealing to HIV-1 RNA. One also finds two separate domains in Gag involved in these same processes. Evidence has been presented supporting the role of lysyl-tRNA synthetase (LysRS) in targeting tRNA Lys for viral incorporation, through a specific interaction of Gag capsid sequence with LysRS in a tRNA Lys /LysRS complex [ 26 ], while other evidence shows that Gag nucleocapsid sequence is involved in tRNA Lys3 annealing [ 6 , 16 , 17 ]. It is not known if LysRS plays any direct role in tRNA Lys3 annealing, and LysRS may be required to dissociate from tRNA Lys3 so as to free this tRNA for annealing to the viral RNA. Methods Plasmid construction BH10 and BH10P- are protease-positive and protease-negative strains of HIV-1, respectively [ 18 ]. All deletions mutants used here were derived from BH10.P-, and their construction has been previously described [ 18 ]. hGagPolΔFSΔPR was a gift from Y. Huang and G. Nabel [ 27 ]. It was constructed by deleting 5 thymidines in the frame shift site, and codes for GagPol. The codons have optimized for mammalian cell codon usage, which results in more efficient translation and protein production, and also makes nuclear export of these mRNAs Rev-independent through modification of the INS [ 27 , 28 ]. hGag-PolΔFSΔPR contain an inactive protease due to an R42G mutation in the active site. Production of wild type and mutant HIV-1 virus Transfection of COS7 cells with wild type and proviral DNA was performed using the calcium phosphate method as previously described [ 29 ]. Briefly, virus were isolated from the cell culture medium 63 hours post-transfection. The supernatant was first centrifuged in a Beckman GS-6R rotor at 3000 rpm for 30 minutes, and the virus were then pelleted from the resulting supernatant by centrifuging in a Beckman Ti45 rotor at 35,000 rpm for one hour. The viral pellet was then purified by centrifugation at 26,500 rpm for 1 hour through 15% sucrose onto a 65% sucrose cushion, using a Beckman SW41 rotor. Protein Analysis Viral particles were washed with 1X TNE and cellular or viral proteins were extracted with 1X RIPA buffer (10 mM Tris pH 7.4; 100 mM NaCI; 1% DOC; 0.1% SDS; 1%NP40; 2 mg/ml Aprotinin; 2 mg/ml Leupeptin; 1 mg/mlPepstatin A; 100 mg/ml PMSF). Western analysis was performed using 300 mg cellular protein or 10 μg viral protein, as determined by the Bradford assay [ 30 ]. The cellular and viral lysates were resolved by SDS-1D PAGE, followed by blotting onto nitrocellulose membranes (Gelman Sciences). Detection of protein on Western blots utilized monoclonal antibodies or antisera specifically reactive with viral capsid (mouse antibody, Intracel), viral reverse transcriptase (rabbit antibody), or β-actin (mouse antibody, Sigma Aldrich). Western blots were analyzed by enhanced chemiluminescence (ECL kit, Amersham Life Sciences) using goat anti-mouse or donkey anti-rabbit (Amersham Life Sciences) as a secondary antibody, and quantitated using UN-SCAN-IT gelTM automated digitizing system. The sizes of the detected protein bands were estimated using pre-stained high molecular weight protein markers (GIBCO/BRL). RNA Isolation and Analysis Total viral RNA was extracted from viral pellets by the guanidinium isothiocyanate procedure [ 31 ], and dissolved in 5 mM Tris buffer, pH 7.5. To measure the incorporation of tRNA Lys3 into virions, hybridization to dot-blots of viral RNA was carried out with DNA probes complementary to tRNA Lys3 [ 1 ] or to genomic RNA [ 16 ]. To measure the amount of tRNA Lys3 annealed to genomic RNA, tRNA Lys3 -primed initiation of reverse transcription was measured using total viral RNA as the source of primer tRNA/template in an in vitro HIV-1 reverse transcription reaction, as previously described [ 32 ]. The sequence of the first 6 deoxynucleoside triphosphates incorporated is CTGCTA, and in the presence of dCTP, dGTP, dTTP, and ddATP, tRNA Lys3 is extended by 6 bases, and this product can be resolved by 1D PAGE, and quantitated by phosphorimaging, as previously described [ 15 ]. Authors' contributions SC carried out the molecular genetic studies, assisted by MJ. LK conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC524520.xml |
549525 | In vitro and in vivo evaluation of NCX 4040 cytotoxic activity in human colon cancer cell lines | Background Nitric oxide-releasing nonsteroidal antiinflammatory drugs (NO-NSAIDs) are reported to be safer than NSAIDs because of their lower gastric toxicity. We compared the effect of a novel NO-releasing derivate, NCX 4040, with that of aspirin and its denitrated analog, NCX 4042, in in vitro and in vivo human colon cancer models and investigated the mechanisms of action underlying its antitumor activity. Methods In vitro cytotoxicity was evaluated on a panel of colon cancer lines (LoVo, LoVo Dx, WiDr and LRWZ) by sulforhodamine B assay. Cell cycle perturbations and apoptosis were evaluated by flow cytometry. Protein expression was detected by Western blot. In the in vivo experiments, tumor-bearing mice were treated with NCX 4040, five times a week, for six consecutive weeks. Results In the in vitro studies, aspirin and NCX 4042 did not induce an effect on any of the cell lines, whereas NCX 4040 produced a marked cytostatic dose-related effect, indicating a pivotal role of the -NO 2 group. Furthermore, in LoVo and LRWZ cell lines, we observed caspase-9 and -3-mediated apoptosis, whereas no apoptotic effect was observed after drug exposure in WiDr or LoVo Dx cell lines. In in vivo studies, both NCX 4040 and its parental compound were administered per os . NCX 4040 induced a 40% reduction in tumor weight. Conversely, aspirin did not influence tumor growth at all. Conclusions NCX 4040, but not its parental compound, aspirin, showed an in vitro and in vivo antiproliferative activity, indicating its potential usefulness to treat colon cancer. | Background One of the most important approaches for reducing cancer incidence is chemoprevention. This is especially relevant for colon cancer, which represents the third leading cause of cancer mortality in developing countries and for which diagnostic tests and clinical treatments are not satisfactory [ 1 ]. Recently published reviews [ 2 - 4 ] underline the growing role of nonsteroidal antiinflammatory drugs (NSAIDs) in preventing colon cancer. Epidemiologic studies [ 5 ] have found that long-term users of aspirin or other NSAIDs have a lower risk of colorectal adenomatous polyps and colorectal cancer compared to nonusers. Randomized clinical trials have confirmed that two NSAIDs, sulindac [ 6 - 9 ] and the selective cyclooxygenase (COX)-2 inhibitor, celecoxib [ 10 ], effectively inhibit the growth of adenomatous polyps and cause regression of existing polyps in patients with familial adenomatous polyposis (FAP), although a recent study showed very clear benefits only for the first six months of treatment with sulindac [ 11 ]. Despite a multiplicity of studies conducted on this compound family of drugs, little is known about the molecular targets that are responsible for their tumor-preventing properties and which are important, not only from a mechanistic point of view, but also because of their clinical implications in identifying individuals or subsets of patients who are sensitive to these drugs. Originally, it was believed that the regular use of these compounds acted exclusively through the inhibition of COX-1 and COX-2 activity involved in the production of prostaglandins [ 12 ]. There is now mounting evidence that the chemopreventive action of NSAIDs may involve COX-independent mechanisms [ 13 , 14 ]. Furthermore, despite initial enthusiasm about the potential relevance of NSAIDs, especially the selective COX-2 inhibitors, their use as chemopreventive or anticancer agents has been greatly limited by side-effects on gastrointestinal and renal systems. In an attempt to reduce gastrointestinal toxicity, conventional NSAIDs have been coupled with a nitric oxide (NO)-releasing moiety [ 15 , 16 ]. The rationale for this approach is that NO compensates for the functions of prostaglandins, which are inhibited by conventional NSAIDs, in the gastrointestinal tract. Both prostaglandins and NO are capable of enhancing mucosal blood flow, mucus release and repair of mucosal injury in humans [ 17 - 19 ] and of reducing the severity of gastric injury in experimental models [ 16 ]. Interestingly, the activity of some NO-NSAIDs is considerably higher than that of the parental drugs. For example, NCX 4016 has been found to be much more effective than the parent aspirin in reducing aberrant crypt foci, a precancerous lesion, in a rat model of colorectal cancer [ 20 ] and in inducing cell perturbations and growth inhibition in a panel of colorectal cancer cell lines in vitro [ 21 ]. Similarly, Williams et al. [ 22 , 23 ], in another panel of human colon cancer lines, reported a higher antiproliferative activity of NO-NSAID compounds compared to the parent aspirin. In the present work, we compared the effect of a novel NO-releasing derivate, NCX 4040, with that of aspirin and its denitrated analog in in vitro and in vivo human colon cancer models and investigated the mechanisms of action underlying its antitumor activity. Materials and methods In vitro studies Cell lines The studies were performed on colon adenocarcinoma cell lines: 2 (LoVo and WiDr) were obtained from the American Type Culture Collection (Rockville, MD), 1 (LRWZ) was isolated in our laboratory and derived from a patient with a confirmed diagnosis of colon adenocarcinoma, and 1 was doxorubicin-resistant (LoVo Dx) derived from the above-mentioned LoVo cells. Cell lines were maintained as a monolayer at 37°C and subcultured weekly. Culture medium was composed of DMEM/HAM F12 (1:1) supplemented with fetal calf serum (10%), glutamine (2 mM), non-essential aminoacids (1%) (Mascia Brunelli s.p.a., Milan, Italy), and insulin (10 μg/ml) (Sigma Aldrich, Milan, Italy). Cells were used in the exponential growth phase in all the experiments. Drugs Aspirin, NCX 4040 (NO-releasing aspirin), NCX 4042 (denitrated analog of NCX 4040) (Fig. 1 ), (all supplied by NicOx S.A., Sophia Antipolis, France), Z-LEHD-FMK (caspase-9 inhibitor) (BD Biosciences Pharmingen, Milan, Italy), sodium nitroprusside dihydrate and S-nitroso-N-acetylpenicillamine (NO donors) (Sigma Aldrich) were solubilized in dimethylsulfoxide (DMSO) (Sigma Aldrich) and freshly diluted in culture medium before each experiment. The final DMSO concentration never exceeded 1% and this condition was used as control in each experiment. Figure 1 Chemical structure and molecular weight (m.w.) of traditional aspirin, its NO-derivative, NCX 4040, and the NCX 4040 denitrated analog, NCX 4042. Chemosensitivity assay Sulforhodamine B (SRB) assay was used according to the method by Skehan et al [ 24 ]. Briefly, cells were collected by trypsinization, counted and plated at a density of 10,000 cells/well in 96-well flat-bottomed microtiter plates (100 μl of cell suspension/well). In the chemosensitivity assay, NCX 4040, NCX 4042 and aspirin were tested at scalar concentrations ranging from 1 to 100 μM for 24 h and 48 h or for 24 h and 48 h followed by a 24-h culture in drug-free medium. Experiments were run in octuplet, and each experiment was repeated three times. The optical density of treated cells was determined at a wavelength of 540 nm by means of a fluorescence plate reader. Growth inhibition and cytocidal effect of drugs were calculated according to the formula reported by Monks et al. [ 25 ]: [(ODtreated - ODzero)/(ODcontrol - ODzero)] × 100%, when ODtreated is > to ODzero. If ODtreated is below ODzero, cell killing has occurred. The ODzero depicts the cell number at the moment of drug addition, the ODcontrol reflects the cell number in untreated wells and the ODtreated reflects the cell number in treated wells on the day of the assay. Western blot The cells were treated according to the previously described Western blot procedure. [ 26 ] COX-1 antibody (polyclonal C-20, dilution 1:100) was purchased from Santa Cruz-Biotechnology (Santa Cruz, CA), COX-2 (monoclonal C22420, dilution 1:250) from Transduction Laboratories (Lexington, KY), and caspases-3 and -9 (polyclonal antibodies, dilution 1:500) from Cell Signaling Technology, Inc. (Beverly, MA). RT-PCR Total RNA was isolated from cells by direct lysis and quantified spectrophotometrically. After quantification, RT reaction was performed with 1 μg of each sample using Gene Amp Gold RNA PCR Core Kit (Perkin Elmer Biosystems, Milan, Italy). The same cDNA was used for RT-PCR amplification of COX-1 and COX-2. PCR reactions were carried out in a final volume of 25 μl containing 2 μl of cDNA template, 1 unit of Klen Taq, 0.2 mM dNTP and 0.4-μM amounts of each forward and reverse primer, using a thermal cycler (PTC 200, Genenco, Florence, Italy). 5-μl aliquots of the amplified DNA fragments were separated on an ethidium bromide-stained 2% agarose gel. Cell cycle distribution After a 24-h exposure to 10 μM of NCX 4040, cells were harvested and stained in a solution containing RNase (10 Kunits/ml; Sigma Aldrich) and NP40 (0.01%; Sigma Aldrich). After 30–60 min, samples were analyzed by flow cytometry using a FACS Vantage flow cytometer. Data acquisition (10,000 events were collected for each sample) was performed using CELLQuest software. Data were elaborated using Modfit (DNA Modelling System) software and expressed as fractions of cells in the different cycle phases. Samples were run in triplicate, and each experiment was repeated three times. Apoptosis Apoptosis was evaluated by flow cytometric analysis according to the previously described TUNEL assay procedure [ 26 ]. Briefly, after a 24- and 48-h exposure to 5, 10 and 50 μM of NCX 4040, cells were trypsinized, fixed, exposed to TUNEL reaction mixture and counterstained with propidium iodide before FACS analysis. In LoVo cells, positivity to TUNEL assay was also evaluated by fluorescence photomicroscope (Zeiss, Axioscope 40) according to the manufacturer's instructions ( In situ cell death detection kit, fluorescein; Roche Diagnostic GmbH, Mannheim, Germany). Finally, the cell-permeable DNA dye 4',6-DAPI and a fluorescence photomicroscope (Zeiss, Axioscope 40) were used to visualize chromatin condensation and/or fragmentation typical of apoptotic cells. In vivo studies Animals Antitumor efficacy was evaluated on 6–8-week old CD-1 male nude (nu/nu) mice weighing 22–24 g (Charles River Laboratories, Calco, Italy). All procedures involving animals and their care were conducted in conformity with institutional guidelines, which are in compliance with national (D.L. No. 116, G.U., Suppl. 40, Feb. 18, 1992; Circolare No. 8, G.U., July 1994) and international laws (EEC Council Directive 86/609, OJ L 358. 1, Dec 12, 1987; Guide for the Care and Use of Laboratory Animals, United States National Research Council, 1996). Drugs NCX 4040 and aspirin were homogeneously suspended in 0.5% carboxymethyl cellulose (CMC) containing 10% of DMSO for antitumor efficacy studies. All the drugs were administered orally. Antitumor efficacy Since LRWZ line showed a low tumorigenicity when cells were injected into nude mice, in vivo experiments were performed on WiDr, LoVo and LoVo Dx lines. Tumor cells were resuspended (5 × 10 6 viable cells) in 0.2 ml of serum-free medium and injected into the hind leg muscles of mice. Each experimental group included at least 6 mice. 10 mg/kg of aspirin and NCX 4040 were administered orally five days a week for 6 consecutive weeks starting on the 6 th day after the tumor cell implant, when a tumor mass of about 300 mg was evident in all the animals. The dose of aspirin and NCX 4040 was chosen on the basis of previous observations obtained using the parental NCX 4016 compound [ 27 ]. Toxicity of treatments was evaluated in terms of body weight loss and drug deaths. The tumor weight was calculated from caliper measurements according to the method of Geran et al. as previously reported [ 28 ]. The antitumor efficacy of treatments was assessed by the following endpoints: a) percent tumor weight inhibition (TWI%), calculated as [1-(mean tumor weight of treated mice/mean tumor weight of controls)] × 100; b) tumor growth delay, evaluated as T – C , where T and C are the median times for treated and control tumors, respectively, to achieve equivalent size. The significance of results was analyzed by the Mann-Whitney non parametric test. Differences were considered significant at P values < 0.05 (two-sided). Results In vitro studies The effect of aspirin, NCX 4040, and its structural analog, NCX 4042, on cell growth was determined after different exposure schemes and using various concentrations. Aspirin and NCX 4042 did not exhibit an effect on any of the cell lines (Fig. 2 ). In contrast, NCX 4040 showed both cytostatic and cytocidal effects, as evaluated by Monk's model (Fig. 3 ). In particular, the highest cytostatic effect was observed after a 24-h drug exposure followed by a 24-h washout, with a 50% growth inhibition (GI 50 ) ranging from 5.4 μM in LoVo cells to 24 μM in LoVo Dx. The same treatment scheme also induced the highest cytocidal effect, with a modulation for the different cell lines and a 50% lethal concentration (LC 50 ) in LoVo cells (9.6 μM) about four- and threefold lower than that observed in WiDr and LRWZ, respectively. In LoVo Dx, no cytocidal effect was observed at any of the concentrations tested. Figure 2 Antiproliferative and cytocidal activity of aspirin and NCX 4042 at concentrations of 1, 5, 10, 50 and 100 μM. Exposure time to drugs:□, 24 h; Ж, 24 h + 24-h washout; ○, 48 h; ▲, 48 h + 24-h washout. Each point indicates the mean of at least three experiments and SD never exceeded 5%. Figure 3 Cytotoxic activity of NCX 4040 after 24-h exposure followed by 24-h washout. Ж, 24 h + 24 h-washout. The two drugs were used at concentrations of 1, 5, 10 and 50 μM. Each point indicates the mean of at least three experiments; SD never exceeded 5%. Flow cytometric analysis of apoptosis performed after exposure to various concentrations of NCX 4040 showed 90% of apoptotic cells at 24 h starting from a 10-μM concentration in LoVo line, whereas apoptosis was not detected at the same time in any other cell line at any of the concentrations tested (Table 1 ). After a 48-h exposure, apoptosis was also induced in LRWZ cells, albeit to a lesser degree, at 10-μM (20%) and 50-μM (60%) concentrations. Table 1 Apoptotic cells after a 24- or 48-h exposure to various concentrations of NCX 4040 Cell Line 24 h 48 h 5 μM 10 μM 50 μM 5 μM 10 μM 50 μM LoVo Dx 0.7 ± 0.01 0.5 ± 0.01 0.5 ± 0.02 0.5 ± 0.0 0.02 ± 0.01 0.2 ± 0.0 WiDr 1.0 ± 0.01 4.0 ± 0.1 1.0 ± 0.05 2.0 ± 0.03 5.0 ± 0.02 5.0 ± 0.03 LRWZ 3.0 ± 0.03 5.0 ± 0.02 4.0 ± 0.1 4.0 ± 0.01 21.0 ± 0.3 60.0 ± 1.3 LoVo 2.0 ± 0.05 90.0 ± 2.3 n.e.* 3.0 ± 0.02 n.e. n.e. * n.e.: not evaluated In LoVo cells, which proved to be the biological system most sensitive to NO-aspirin, the induction of apoptosis produced by a 10-μM concentration was analyzed as a function of time. The results showed that the production of apoptotic cells is time-dependent, reaching as much as 80% after a 16-h drug exposure (Fig. 4A ). Apoptotic elements were clearly visible using the in situ TUNEL analysis (Fig. 4B ) and at fluorescent microscopic examination of cells stained with DAPI nuclear dye (Fig. 4C ). Furthermore, cell death was strongly inhibited by simultaneous exposure of LoVo cells to a 10-μM concentration of NCX 4040 and a 100-μM concentration of caspase-9 inhibitor (Fig. 4D ). Similar findings were obtained after exposure to caspase-3 inhibitor (data not shown). Figure 4 (A) Percentage of apoptosis in LoVo cells after exposure times of 4, 8 12, 16, 20 and 24 h to 10 μM of NCX 4040. (B) Apoptosis in LoVo cells after a 24-h exposure as evidenced by in situ TUNEL assay. (C) Apoptosis in LoVo cells after a 24-h exposure as evidenced by DAPI staining. (D) Inhibition of NCX 4040-induced apoptosis in LoVo cells after a 24-h simultaneous exposure to the NO-NSAID and the caspase-9 inhibitor, as evidenced by TUNEL assay. To understand the mechanisms responsible for the induction or lack of apoptosis in the different cell lines, we explored the modulation of COX-2 expression and the activation of caspases-3 and -9, which are involved in the mitochondrial pathway of apoptosis. All the cell lines used were positive for the presence of the isoenzymatic form COX-1, as evaluated by mRNA and protein expression (Fig. 5A ). Furthermore, after a 24-h exposure to NCX 4040, the COX-1 expression level remained unchanged in all the cell lines (Fig. 5B ). COX-2 mRNA and protein expression were both observed in WiDr and LoVo Dx cell lines, whereas neither was detected in LRWZ, and only mRNA was observed in LoVo cells. Furthermore, in WiDr and LoVo Dx cells, no change was observed in the COX-2 protein expression level after a 24-h exposure to NCX 4040 or a 24-h exposure followed by a 24-h washout (Fig. 5C ). Figure 5 (A) COX-1 and COX-2 basal mRNA and protein expression in the different cell lines. Lane a, LoVo cells; lane b, WiDr cells; lane c, LRWZ cells; lane d, LoVo Dx. (B) COX-1 protein expression in the different cell lines. LoVo : lane 1, untreated cells; lane 2, cells exposed to 10 μM of NCX 4040 (24 h); LoVo Dx : lane 3, untreated cells; lane 4, cells exposed to 50 μM of NCX 4040 (24 h); WiDr : lane 5, untreated cells; lane 6, cells exposed to 50 μM of NCX 4040 (24 h); LRWZ : lane 7, untreated cells; lane 8, cells exposed to 50 μM of NCX 4040 (24 h). (C) COX-2 protein expression in WiDr and LoVo Dx cell lines after different exposure schemes to 10 μM of NCX 4040. Lane 1, untreated cells (24 h); lane 2, NCX 4040 (24 h); lane 3, untreated cells (24 h) + washout (24 h); lane 4, NCX 4040 (24 h) + washout (24 h). Although all 4 lines basally expressed the 2 caspases (data not shown), these proteases were activated only in LoVo and in LRWZ cells, where apoptosis was found. In particular, in LoVo cells caspase-3 and -9 activation was already evident after 3 h, whereas in LWRZ, activation took a longer time to occur (at least 10 h for caspase-9 and 12 h for caspase-3). The temporal activation in these 2 cell lines corresponds to the early or late induction of apoptosis (Fig. 6A,6B ). Figure 6 (A) Activation of caspases-9 and -3 after exposure times of 1, 2, 3, 4, 5 and 6 h to 10 μM of NCX 4040 in LoVo cells. (B) Activation of caspases-9 and -3 after exposure times of 8, 10, 12 and 14 h to 50 μM of NCX 4040 in LRWZ cells. An antibody for actin was used as loading control. In vivo studies The administration of aspirin in mice bearing all 3 cell lines did not influence the growth of tumors (Fig. 7A,7B,7C ). In contrast, treatment with NCX 4040 was effective in the different cell lines employed, with a reduction of about 40% in the tumor mass (P < 0.001), evaluated at the nadir of the effect, compared to untreated and aspirin-treated groups. A difference in NCX 4040 sensitivity was also evident between the lines. Tumor regrowth was observed after 8 and 11 days for LoVo Dx and WiDr lines, respectively. Conversely, inhibition of the tumor mass for LoVo tumors was maintained from day 15 to day 30 following the injection of tumor cells and a tumor growth delay of 19 days was observed (P < 0.001). These results are in agreement with the in vitro data, which demonstrated that LoVo line is the most sensitive to NCX 4040 exposure. Figure 7 Antitumor efficacy in mice implanted with LoVo Dx (A), WiDr (B) and LoVo (C) cell lines. Each experimental group included 8 mice, each experiment was repeated at least three times, and representative independent experiments are reported. Experimental points represent means of 24 experiments (bars, SD). Arrows indicate the start of treatment. Discussion Our study evaluated the antitumor activity of the novel NO-aspirin NCX 4040 in vitro on a panel of human colon cancer lines with different genetic profiles representing clinical tumor heterogeneity, and in vivo on xenografted immunosuppressed mice. Neither aspirin nor the denitrated analog, NCX 4042, influenced the growth of any of the cell lines studied, whereas NCX 4040 produced an important cytocidal effect. These findings suggest a pivotal role of the -NO 2 group in the mechanism of action underlying its in vitro activity. Extensive evidence indicates that nitric oxide is capable of initiating apoptotic cell death in some cell types as a consequence of DNA-induced damage [ 29 , 30 ], but the pathways involved are still not fully understood. In the present study, it was seen that NCX 4040 induced apoptosis in two (LoVo and LRWZ) of the four cell lines investigated, albeit with a modulated activity. In particular, we observed early apoptosis in LoVo cells starting from a 6-h drug exposure, with a peak after 20 h, as shown by the TUNEL assay. We found that apoptotic death was induced by the concomitant activation of caspases-9 and -3 a few hours after NCX 4040 exposure. In LRWZ cells, a relevant apoptosis was also observed but after a longer exposure time and at higher NCX 4040 concentrations, after which the irreversibly damaged cells underwent apoptotic death through the activation of caspases-9 and -3. Moreover, two NO donors were used to better clarify the role of the aspirin component. These compounds, at concentrations similar to those utilized for NCX 4040, showed a very modest cytotoxic activity (data not shown), confirming the importance of the aspirin component for the antineoplastic effectiveness of the NCX 4040 molelcule. In addition, WiDr and LoVo Dx cell lines, which proved to be resistant to NCX 4040-induced apoptosis, were characterized by baseline expression of COX-2 protein. This finding, which is consistent with data recently published for breast and colorectal cancer cells, would seem to support the hypothesis that a high level of COX-2 protein protects cells from apoptosis, especially that induced by NO-compounds [ 31 , 32 ]. With the exception of LRWZ, all the lines evaluated in in vitro studies showed a high tumorigenic capacity, making it possible to evaluate in vivo the antitumor efficacy of NCX 4040. The long-term administration of this drug was well tolerated by the mice and reduced the tumor mass of LoVo, LoVo Dx and WiDr lines compared to untreated mice. Analysis of growth curves highlighted differences in the degree of drug sensitivity between cell lines. NCX 4040 was very effective in delaying tumor growth in LoVo tumors, whereas WiDr, and especially LoVo Dx lines, regrew after the end of treatment. These results confirm the in vitro data and seem to indicate that, as observed in vitro , the ability of NCX 4040 to induce apoptosis has a key role in the antitumor efficacy of this drug. Conclusions In conclusion, NCX 4040 was effective in reducing the growth of several human colon cancer cell lines both in vitro and in vivo , whereas the parental compound aspirin showed no activity, which suggests that the NO-releasing derivate could prove useful in the clinical management of colon cancer, possibly in combination with conventional antineoplastic drugs. Abbreviations COX, cyclooxygenase; NO, nitric oxide; NO-NSAID, nitric oxide-releasing nonsteroidal antiinflammatory drug releasing; NSAID, nonsteroidal antiinflammatory drug; SRB, sulforhodamine B; TUNEL, terminal uridine nick end labeling. Competing interests Manlio Bolla is an employee of NicOx SA. Authors' contributions AT was responsible for study design, data analysis, and drafting the manuscript. WZ, DA and MB participated in the study design and acted as scientific advisors. AT, PU, FF and MR performed the in vitro experiments. CL, MS and GZ also participated in the study design and carried out the in vivo experiments. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549525.xml |
449900 | Nanotubes Make Big Science | Tiny protrusions on the surface of cells might be a more common mechanism for cell communication than previously expected | During development, cells need to communicate with each other to establish properly organised and functional tissues. Cells communicate with each other in various ways, such as by secreting and receiving diffusible molecules (morphogens, hormones, and neurotransmitters) or by establishing intercellular connections (gap junctions and cell protrusions) to allow a direct exchange of instructive factors. A recent paper has shown that communication via tiny cell protrusions might be a more common mechanism than previously expected ( Rustom et al. 2004 ). Many different types of cell extensions have been described in a variety of developmental processes and organisms ( Miller et al. 1995 ; Bryant 1999 ; Chou and Chien 2002 ; Rorth 2003 ), and for most of them a role in cell-to-cell communication has been hypothesized. For example, in the mouse, Salas-Vidal and Lomeli (2004) have described long processes (filopodia) that connect tissues in early embryos. Because these protrusions contain receptors for some well-known signalling molecules, it is thought that they might be responsible for receiving signals from neighbouring cells. Similarly, it has been proposed in the development of the Drosophila wing and eye imaginal discs (precursors of adult structures) that signals modulating the growth and patterning of one epithelial layer of cells are received through microtubule-based cell extensions arising from the apposing epithelium ( Cho et al. 2000 ; Gibson and Schubiger 2000 ). Furthermore, in the wing imaginal disc, planar extensions called cytonemes arise from the periphery of the epithelium and grow towards a central area in the wing disc that produces the signalling molecule Decapentaplegic. This directionality of growth, and the observation of vesicles inside cytonemes, led Ramirez-Weber and Kornberg (1999) to propose that cells meant to receive a signal were searching actively for it, extending long cell protrusions towards the region from which signals were emanating. Although cell protrusions have been described in different developmental processes, tissues, and organisms, their potential role in cell signalling has been difficult to pin down. Most cell processes are very fragile, and their study is mainly limited to live tissues; in these conditions it is technically challenging to define how the signalling is mediated via protrusions. Possibly, it could occur through the release of free molecules, in a similar manner to synaptic neurotransmission, or shedding of vesicles as exosomes followed by endocytosis by the recipient cell. Alternatively, membranetethered ligands (such as Delta) on the protrusion could bind and activate receptors displayed on the surface of the receiving cell ( De Joussineau et al. 2003 ). The paper by Rustom and colleagues has provided a new outlook on the role of cell protrusions, by reporting a novel mechanism employed to transmit signals between cells connected by a protrusion. Surprisingly, they did not observe any of the mechanisms described above. Rather, transfer of molecules and organelles occurred directly from the cytoplasm of one cell to the other, passing through a protrusion that established membrane continuity between the connected cells. Using rat PC12 cells, Rustom and colleagues observed ultrafine protrusions (with a diameter of only 50–200 nm and a length spanning several cell diameters) connecting sparse cells in culture ( Figure 1 ). Similar to other cell protrusions, these structures, termed tunnelling nanotubes (TNTs), displayed a pronounced sensitivity to both mechanical stress and chemical fixation and even to prolonged light excitation, resulting in the rupture of many of them. TNTs are actin-based and devoid of microtubules: interestingly most other types of cell protrusions also contain actin ( Condeelis 1993 ; Rorth 2003 ). The researchers also confirmed the existence of TNTs in a human cell line (human embryonic kidney cells) and rat primary cells (normal rat kidney cells), suggesting that TNTs are not a peculiarity of PC12 cells. Figure 1 A TNT Connecting Two Neighbouring Cells Immunofluorescence analysis of wheat germ agglutinin–stained PC12 cells that shows a TNT, a novel type of cell protrusion that establishes membrane continuity between two neighbouring cells. Transfer of molecules and organelles can occur directly from the cytoplasm of one cell to that of the other. (Image courtesy of Hans-Hermann Gerdes.) Aiming to investigate how TNTs were linking cells, the authors performed scanning and transmission electron microscopy of TNTs. They observed a seamless transition between TNTs and the cells they were connected to, suggesting that indeed there was continuity between the membranes of the two connected cells. Rustom and colleagues then went on to test whether TNTs could be used to transmit signals between cells. The experimental approach used was to mark two populations of cells in a distinct way, either by introducing genes that encoded proteins tagged with green fluorescent proteins or by using dyes. The two different cell populations were then mixed, cocultured, and analysed for transfer of marked proteins or dye-stained organelles from one cell to another, between cells that were differently marked and connected by a TNT. Strikingly, soluble cytoplasmic molecules could not pass freely along the TNTs (with actin tagged with green fluorescent protein being the only exception), whereas membrane-bound proteins were transferred along TNTs and detected in the receiving cells, further supporting the likelihood of membrane continuity between connected cells. Rustom and colleagues also observed transport of vesicles, which seemed to be unidirectional. Finally, in transfer experiments performed at close to 0 °C, where endo-, exo-, and phagocytosis would be blocked, vesicle exchange still occurred, suggesting that these events are not required for vesicle transfer and further supporting the idea that membrane continuity exists between connected cells. By contrast, interfering with actin polymerization, using the drug latrunculin-B, led to protrusion removal and arrest in organelle transfer, indicating that actin is required both for protrusion biogenesis and organelle transport. Taken together, the experiments performed by Rustom and colleagues strongly suggest a role for cell protrusions in cell-to-cell communication. They also provide evidence, in culture, for a novel mechanism used by cell protrusions to transport molecules and organelles. It will be interesting to test whether TNTs also exist in living tissues and, if so, what molecules they transport. TNTs could be distinct from the protrusions known so far and could be responsible for establishing another type of connection between cells. They could connect all cells in a tissue, directly or indirectly, establishing a global interaction network potentially important in exchanging basic survival information as well as positional cues ( Milan et al. 2001 ). Another interesting question is how connections such as TNTs are established. Rustom et al. have shown that, initially, many filopodial extensions arise from one cell and are directed toward a neighbour. As soon as one of them reaches the target, it is stabilised, while the others degenerate. It is possible that membrane fusion occurs between the tip of the protrusion and the planar plasma membrane of the target cell. However, membrane fusion can be more easily achieved if the tips of two cell protrusions fuse with each other, thus suggesting the participation, in the process of membrane fusion, of microvilli or other tiny protrusions belonging to the target cell. Fusion between two protrusions is reported to rely on the cylindrical shape and narrow diameter of cell protrusions and also on the localised concentration of adhesion/fusion molecules at the tips of the cell protrusions, such as microvilli, that display particular tip-specific membrane microdomains ( Monck and Fernandez 1996 ; Wilson and Snell 1998 ; Roper et al. 2000 ). The work performed by Rustom and colleagues suggests that cell protrusions are a general mechanism for cell-to-cell communication and that information exchange is occurring through the direct membrane continuity of connected cells, independently of exo- and endocytosis. It is important to determine whether events similar to these seen in cell culture are occurring in tissues and what functions cell protrusions perform during tissue morphogenesis. In my work as a graduate student, I am trying to address this question. We need to identify the types of cell protrusions that are present in tissues and the molecular complexes localizing on them as well as their functions. To then prove that cell protrusions are important in cell-to-cell communication in tissues, we would need to remove the protrusions and see how this affects tissue architecture and function. However, the necessary tools are still missing, given the lack of knowledge of the specific molecules important for the biogenesis of these protrusions. Thus far, the function of cell protrusions has been hypothesized mainly on the basis of their location in tissues and on crude attempts to remove them, for example by altering the actin cytoskeleton or even by removing the entire epithelium they belong to. The paper by Rustom et al. has shed some new light on these still mysterious cellular arms and has further boosted my interest in this emerging field of cell and developmental biology. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC449900.xml |
547911 | Enterococcal meningitis caused by Enterococcus casseliflavus. First case report | Background Enterococcal meningitis is an uncommon disease usually caused by Enterococcus faecalis and Enterococcus faecium and is associated with a high mortality rate. Enterococcus casseliflavus has been implicated in a wide variety of infections in humans, but never in meningitis. Case presentation A 77-year-old Italian female presented for evaluation of fever, stupor, diarrhea and vomiting of 3 days duration. There was no history of head injury nor of previous surgical procedures. She had been suffering from rheumatoid arthritis for 30 years, for which she was being treated with steroids and methotrexate. On admission, she was febrile, alert but not oriented to time and place. Her neck was stiff, and she had a positive Kernig's sign. The patient's cerebrospinal fluid was opalescent with a glucose concentration of 14 mg/dl, a protein level of 472 mg/dl, and a white cell count of 200/μL with 95% polymorphonuclear leukocytes and 5% lymphocytes. Gram staining of CSF revealed no organisms, culture yielded E. casseliflavus . The patient was successfully treated with meropenem and ampicillin-sulbactam. Conclusions E. casseliflavus can be inserted among the etiologic agents of meningitis. Awareness of infection of central nervous system with Enterococcus species that possess an intrinsic vancomycin resistance should be increased. | Background Enterococcal meningitis is an uncommon disease accounting for only 0.3% to 4% of cases of bacterial meningitis which is nevertheless associated with a high mortality rate. It has been described most frequently in patients with neurosurgical conditions (i.e. head trauma, shunt devices, or cerebrospinal fluid leakage), although it can also occur as a "spontaneous" infection complicating remote enterococcal infections such as endocarditis or pyelonephritis [ 1 ]. Enterococcus faecalis and Enterococcus faecium are the two species most frequently isolated during the course of meningitis (76%–90% and 9–22% respectively). Enterococcus casseliflavus , first considered as a subspecies of E. faecium , is a motile enterococcus that produces a yellow pigment in agar and often has a VanC phenotype determining an intrinsic low level resistance to vancomycin. It has been implicated in a wide variety of infections in humans, especially immunocompromised hosts, but to the best of our knowledge it has never been associated to meningitis [ 2 - 5 ]. We describe here a case of enterococcal meningitis caused by E. casseliflavus that was believed to originate from the gut in an old patient with bowel erosions. Case presentation A 77-year-old Italian female presented for evaluation of fever, stupor, diarrhea and vomiting of 3 days duration. She had no urinary symptoms. There was no history of head injury nor of previous surgical procedures. She had been suffering from rheumatoid arthritis for 30 years, for which she was being treated with steroids and methotrexate; other medical problems were insulin-dependent diabetes and moderate renal failure. On admission, she was febrile (temperature, 38.0°C), alert but not oriented to time and place. Her neck was stiff, and she had a positive Kernig's sign. A CT brain scan showed an increase in subarachnoid space and in the volume of the ventricular system. Laboratory examinations revealed a white blood cell count of 15,100/μL with 70% neutrophils and 23% lymphocytes. Results of urinalysis were unremarkable. Cultures of blood and urine were drawn and subsequently resulted negative. The patient's cerebrospinal fluid (CSF) was opalescent with a glucose concentration of 14 mg/dl, a protein level of 472 mg/dl, and a white cell count of 200/μL with 95% polymorphonuclear leukocytes and 5% lymphocytes. Gram staining of CSF revealed no organisms. Pending the culture results the patient was empirically treated with intravenous meropenem, cotrimoxazole, acyclovir and dexamethasone. Culture of CSF yielded E. casseliflavus that was identified using the Vitek-2 system (bioMérieux-Vitek) on the basis of 6.5% NaCl tolerance, bile-esculin hydrolysis, and growth rate at 45°C, arginine hydrolysis, methyl-a-D-glucopyranoside testing, and acid production from ribose, motility testing, and yellow pigmentation testing. The isolate was sensitive to penicillin, ampicillin, ampicillin-sulbactam, imipenem, teicoplanin, tetracyclines and linezolid; it exhibited intermediate sensitivity to vancomycin (MIC, ≥8 μg/mL), trimethoprim-sulfamethoxazole (MIC, ≥10 μg/mL), levofloxacin (MIC, 4 μg/mL), norfloxacin (MIC, 8 μg/mL), ciprofloxacin (MIC, 2 μg/mL) and quinupristin-dalfopristin (MIC, 2 μg/mL); it was resistant to clindamycin (MIC, 4 μg/mL) and showed high resistance to gentamicin, streptomycin and kanamycin (MIC, ≥2000 μg/mL). The patient became afebrile 48 hours after the beginning of antibiotic therapy with rapid improvement of her mental status and disappearance of meningeal signs (within 36 hours). Once the organism was identified (4 days later), trimethoprim-sulfamethoxazole and acyclovir were discontinued and ampicillin-sulbactam (3 g every 6 hours) was added. After 2 weeks of antibiotic therapy the patient was discharged in good health with sterilization of the CSF culture. An echocardiogram revealed no vegetations whereas a colonoscopy examination showed two ulcerative lesions associated with two polyps, oedema and multiple punctuate erosions. Enterococcal infections of the central nervous system are quite rare and according to a MEDLINE search of the English literature only three cases of CNS infection by a motile Enterococcus identified as E. gallinarum have been previously documented; they occurred in patients with ventriculoperitoneal shunts for hydrocephalus [ 6 , 7 ]. E. casseliflavus and E. gallinarum are responsible for 1–2% of all enterococcal infections and are characterized by the fact that they possess intrinsic low-level vancomycin resistance [ 8 ]. The VanC-1 ligase is specific for E. gallinarum , and the VanC-2/3 ligase is specific for E. casseliflavus [ 9 ]. Organisms with resistance to VanC remain susceptible to teicoplanin. This naturally occurring vancomycin resistance has not been shown to be transferable, and the related genes are chromosomally encoded in the members of these species [ 4 , 9 , 10 ]. Despite the intrinsic low-level vancomycin resistance exhibited by E. casseliflavus it is important to remember that most strains are susceptible to penicillin and ampicillin. Combination therapy of ampicillin with an aminoglycoside such as gentamicin or streptomycin is considered the standard therapy of enterococcal meningitis due to ampicillin-susceptible strains [ 1 ]. Meropenem is not superior to ampicillin for therapy of enterococcal infections and most species of E. casseliflavus are beta-lactamase negative. There was not therefore a clear indication for the use of combination therapy with meropenem and a beta-lactamase inhibitor in the case reported. Our patient had gastrointestinal signs and symptoms, and colonoscopy revealed multiple erosions, which probably were the portal of entry of E. casseliflavus . In fact, enterococcal meningitis may appear as a complication of diverse gastrointestinal diseases such as enterocolitis, peritonitis, abdominal surgery, or bowel carcinoma. A case of bowel erosions as the portal of entry of enterococcal meningitis has been previously reported [ 11 ]. Several studies have demonstrated that E. gallinarum and E. casseliflavus colonize the gastrointestinal tracts of both hospitalized individuals and nonhospitalized healthy ones [ 8 , 12 , 13 ]. Therapy with various antimicrobial agents, including cephalosporins and vancomycin, may play a role in increasing colonization with these organisms. Edlund et al. reported a significant increase in the emergence of E. gallinarum and E. casseliflavus in healthy subjects who were administered oral vancomycin [ 14 ]. Our patient was not taking any antimicrobials before hospital admission. E. gallinarum and E. casseliflavus/flavescens are part of the normal stool flora of the general population; this has perhaps impacted the ability of researchers to detect specific risk factors [ 8 ]. Although our patient had significant underlying conditions this case of meningitis was mild and had most of the typical features of "spontaneous" enterococcal meningitis (community-acquired infection, severe underlying diseases, and immunodepression). The clinical significance of the enterococci that are intrinsically resistant to vancomycin has not been fully established yet. Infection with E. gallinarum and E. casseliflavus has been associated with high mortality, but it is difficult to attribute their mortality directly to infection or to the underlying conditions of the patient. Conclusions This case is the first report of E. casseliflavus meningitis and it is the fourth documented so far with a motile Enterococcus species. Awareness of infection of central nervous system with Enterococcus species that possess an intrinsic vancomycin resistance should be increased. Competing interests The author(s) declare that they have no competing interests. Authors' contributions IC and RDL carried out the clinical study of the patient and conceived of the study. GS and GBC carried out the microbiologic studies. AT and AC carried out the clinical study of the patient, conceived of the study and drafted the manuscript. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC547911.xml |
509317 | Retrovirus Integration into the Human Genome | null | When gene therapy was introduced nearly fifteen years ago, it was widely hailed as a panacea. Since many diseases have a genetic component, the hope was that gene therapy could replace compromised genes with healthy versions to treat everything from inherited disorders like cystic fibrosis to cancer and HIV. That great promise was quashed when a teenager suffering from a rare hereditary liver disorder died after participating in an experimental gene therapy trial in 1999: four days after being injected with millions of viruses engineered to deliver healthy genes to his liver, Jesse Gelsinger died. It seems the virus, derived from an adenovirus, targeted his immune cells rather than his liver cells, which triggered an immune response against the virus, resulting in massive organ failure. In another case, two young boys who received gene therapy for the severe immunodeficiency disorder known as “bubble boy disease” developed leukemia-like symptoms 30 months after treatment. In this case, the viral vector inserted itself near a promoter region—a site that initiates gene transcription—of a proto-oncogene, a gene that can initiate cancer. Since viral vectors can integrate at various genomic locations, the safety and effectiveness of gene therapy ultimately depends on being able to predict a virus's particular bias. Comparing retroviral vectors derived from three viruses, including two common gene therapy vectors, Rick Mitchell et al. report 3,127 sites where these viruses typically integrate into the human genome. The different vectors, they found, show different target preferences. Retroviruses use viral enzymes to copy their own genome, which is stored in an RNA transcript, into DNA. Now recognizable by the host's genome, the virus can integrate into one of the host's chromosomes. In this study, Mitchell et al. studied vectors derived from the human immunodeficiency virus (HIV), avian sarcoma-leukosis virus (ASLV), and murine leukemia virus (MLV). Introducing the viral vectors into human cells, the authors analyzed the gene expression profiles of the cells to determine where vectors integrate into human chromosomes and which, if any, genes they activate. Mitchell et al. then compared the integration sites with the transcription profiles. Each retrovirus, they discovered, showed distinct preferences for genome integration. HIV vectors tend to integrate into sites of active transcription, favoring chromosomal regions rich in expressed genes. MLV vectors tend to integrate near transcription initiation sites, confirming the results of a previous study, with a weak bias toward active genes. In contrast, the authors report, the ASLV vector “does not favor integration near transcription sites, nor does it strongly favor active genes.” Early efforts to understand how chromosomes may influence where viruses insinuate themselves into a chromosome focused on factors governing accessibility. Viruses are more likely to be integrated into chromosomal regions that are more accessible, which tend to be transcriptionally active sites. But since each of the three viruses studied here routinely targeted different sequences, the authors note, accessibility is probably just one factor. Specific chromosomal proteins, for example, might interact with the viral integration machinery and facilitate integration at nearby sites. Another possibility, the authors propose, is that DNA-binding proteins that bind to specific DNA sequences assist integration of one virus while impeding another. This could explain why ASLV behaved as it did in the human cells studied here. The virus might have more refined integration preferences during normal infection of chicken cells, the authors note, but its integration machinery can't interact properly with human cells. The leukemia-like effects of the bubble boy gene therapy stemmed from integration of a mammalian retrovirus—the MLV vector—near an oncogene promoter region. Since ASLV tends to avoid both transcription initiation sites and active gene sites, it could be a more promising candidate for human gene therapy. With the draft chicken genome sequence now complete, researchers can investigate whether that proves true. But for now, Mitchell et al. make the case that scientists can gain more control over where viral vectors integrate into the human genome by selecting different retroviral integration systems. Only time will tell whether more control translates into safer gene therapy protocols. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC509317.xml |
546400 | Epidemiology of neural tube defects and folic acid | This review article combines four disparate observations about Neural Tube Defects (NTDs). They are the worldwide decline in the birth incidence that began prior to prenatal diagnosis; family recurrence risks; the effect of prenatal diagnosis and termination of affected pregnancies; and the effect of folic acid. | Discussion Variations in birth incidence NTDs are due to many different causes [ 1 ]. The incidence varies and some etiologies are more common at different prenatal ages and birth [ 2 - 5 ]. Epidemiological data from the period before in-utero diagnosis and termination of affected pregnancies established two trends. The first illustrated epidemics of the incidence of NTDs that cycled over years, and the second established that there has been a general decline in the birth incidence of NTDs over several decades, which has been noted worldwide. The best example of epidemics of the occurrence of NTDs was reported from Birmingham, England [ 6 ]. The prevalence rose from 2.0 per 1000 births in 1936–7, to 2.8 in 1940, fell to 1.54 in 1948 and peaked again in 1956 at 2.8 per 1000. Elwood and Elwood [ 7 ] have documented both "epidemics" and the decline in incidence worldwide. Other "epidemics" have been described in Boston, Providence and New York in the USA, Canada, Berlin and in Western Germany, and in a number of other areas [ 7 ]. The general decline in birth incidence has been reported in most countries studied but it has occurred at different times. A variation within countries was also reported, with a Northeast to Southwest gradient from higher to lower in the UK and from East to West in Canada and the USA. For example, the highest occurrence risk for the general population on the North American continent is 2.5 – 3.5 per 1000 amongst citizens of Nova Scotia [ 8 ]. These data include findings of elective terminations, spontaneous miscarriages and births. The lowest general population occurrence is 0.21 – 0.03 per 1000 live births in the Pacific Northwest of the USA (Figure 1 ). These geographic variations may relate to migrations of ethnic groups. The highest birth incidences for NTDs are reported amongst descendants of the Celts in the UK, Canada and the USA. These racial data suggest the importance of a genetic component that has been borne out in family studies [ 7 ]. Figure 1 These data were obtained from the Birth Defects Clinic attendance records. Patients with myelomeningocele seen in this clinic represent 98% of individuals with myelomeningocele born in Washington State, USA, during the period that the state maintained a Birth Defects Registry. Preliminary data for the year 2003 suggest a decline in incidence to near the lowest level observed prior to 2002. Family recurrence risks Family studies suggest the recurrence risk for first-degree relatives of affected individuals is approximately 1 in 30. For second-degree relatives (the children of the mother's sisters and brothers) the risk is approximately 1 in 220 [ 9 ]. However, the authors of the same study state there is not yet agreement upon the accurate recurrence risk data for family members. Others report a recurrence risk among the mother's or an affected child's first-degree relatives is as high as 1 in 70 to 140 [ 10 , 11 ]. McManus [ 12 ] reports recurrence risks in first and second-degree relatives of 1 in 40 for sisters of the mother of an affected child, and 1 in 90 for the offspring of those sisters. She reported the mother's brothers and the father's sisters and brothers to have lower recurrence risks of 1 in 140 to 1 in 190. Arata et al [ 13 ] report that affected mothers appear to have only a 0.5 to 1% chance of having a child with an NTD. This last observation concurs with our finding that only one mother with myelomeningocele has an affected child. That child is one of 106 otherwise unaffected by an NTD. Prenatal diagnosis and termination The third aspect, prenatal diagnosis and termination of affected pregnancies, is one that should be discussed with all women in the reproductive age range and, more importantly, with patients who have a family history of an NTD. Folic acid taken orally on a daily basis is shown to lower the occurrence and recurrence of NTDs in their own offspring and in their relatives. The Medical Research Council [ 14 ] was the first to prove conclusively that when women who had had a previous child affected by an NTD took 4.0 mg of folic acid daily, beginning three months prior to conception, there was a 70% reduction in the recurrence in subsequent offspring. Wald et al [ 15 ] have recommended 5 mg daily. Because of the higher occurrence in first- and second-degree relatives, we recommend 4.0 mg daily, beginning three months prior to a planned conception. We suggest the effect in the United States may be nearer to a 40 – 50% reduction for two reasons. First, Berry et al [ 16 ] demonstrated a 79% reduction in occurrence in north China (an area of high incidence) but only 40% in south China (where there is a low incidence) when the women took 0.4 mg of folic acid periconceptually. The incidence of NTDs in the United States is closer to that seen in south China rather than that in north China. Secondly, recent data from Canada and Mexico are the first to indicate that lower incidence communities on the North American continent can achieve a 45 – 55% reduction in occurrence with a regimen of 5 mg of folic acid per week [ 17 ] and dietary fortification added to recommendations for supplementation periconceptually with 0.4 mg dose of folic acid daily [ 8 , 18 ]. Folic acid supplementation When recent trends in the birth incidence of NTDs are reported, they focus additionally on the effect that folic acid has on the early second trimester prevalence of affected fetuses. As for the epidemiological studies noted above, these reports include varying types of cases; some report only "spina bifida", others "spina bifida" and anencephaly, and still others mention these two types and encephalocele with or without hydrocephalus. Some studies report only deaths due to complications in these groups of patients as stillborns, or deaths in the neonatal time period; other reports study all affected newborns, and still others cover selective or spontaneously aborted fetuses. Those that include time intervals after the introduction of intrauterine diagnosis and selective termination do not take into consideration the variations in incidence at different gestational ages and at birth, whether stillborn or live [ 5 ]. Creasy and Alberman [ 2 ] reported that 3% of 1216 (30 per 1000) spontaneously aborted fetuses had central nervous system malformations. The majority of the malformations were NTDs. Forty percent also had chromosomal aberrations. The prevalence varied from about 21 per 1000 during each three-week period of gestational age between 8 and 19 weeks, to 105 per 1000 amongst fetuses greater than 27 weeks gestational age. The live born birth incidence at that time was 1.5 per 1000. Nishimura et al [ 4 ], reported 13 per 1000 spina bifida embryos amongst 3402 induced abortion fetuses for social reasons at a gestational age between 3 and 10 weeks old. The live born incidence at the time was 2 per 1000. Adams et al [ 19 ] reported 10 of 34 fertilized ova up to the age of 17 days were malformed (an incidence of 294 per 1000). Studies after the initiation of prenatal diagnosis and before folic acid supplementation and fortification clearly demonstrate a remarkable decrease in live born birth incidence ([ 1 , 20 , 21 ] and Figure 1 ). In Washington State, USA, prenatal diagnosis and termination of affected pregnancies began in 1980. Beginning in 1991, concerned specialists and the media advised supplementation with 400 mcg (0.4 mg) of folic acid daily for women in the childbearing age group. Fortification with an estimated 140 mcg (0.14 mg) per serving of flour-containing foods was added to the recommendations, beginning in 1996, and implemented over the next three years. The marked increase of 7 fold from 0.03 per 1000 in 2001 to 0.21 in 2002 coincided with the completion of fortification (Figure 1 ). Conclusions Our data and this review clearly demonstrate the effects of intrauterine diagnosis and selective termination prior to the recommendation for supplementation and fortification of foodstuffs with folic acid. Because the reason for termination of a pregnancy is not reportable in our state and the USA, we cannot determine the effect of folic acid on the prevalence of myelomeningocele and anencephaly in first and early second trimester fetuses. Studies of the effect of folic acid in reducing the birth incidence in communities with a low incidence, and active prenatal diagnosis associated with termination of affected fetuses, require longer-term studies than published to date. The differences in data discussed above need to be considered if one is to evaluate the effect of prenatal diagnosis and elective termination as well as the effects of fortification or supplementation with folic acid. We recommend that these variables be discussed with women of reproductive age, particularly if they are relatives of a patient with an NTD. Regardless of the uncertainties, we recommend supplementation of the diet of women, beginning three months prior to an anticipated pregnancy. We recommend all women of childbearing age take at least 400 mcg of folic acid daily when they begin sexual activity. Relatives of a patient with an NTD should take 4.0 mg daily, beginning three months prior to conception. Competing Interests The authors declares that he has no competing interests. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC546400.xml |
520746 | Integrated analysis of metabolic phenotypes in Saccharomyces cerevisiae | Background The yeast Saccharomyces cerevisiae is an important microorganism for both industrial processes and scientific research. Consequently, there have been extensive efforts to characterize its cellular processes. In order to fully understand the relationship between yeast's genome and its physiology, the stockpiles of diverse biological data sets that describe its cellular components and phenotypic behavior must be integrated at the genome-scale. Genome-scale metabolic networks have been reconstructed for several microorganisms, including S. cerevisiae , and the properties of these networks have been successfully analyzed using a variety of constraint-based methods. Phenotypic phase plane analysis is a constraint-based method which provides a global view of how optimal growth rates are affected by changes in two environmental variables such as a carbon and an oxygen uptake rate. Some applications of phenotypic phase plane analysis include the study of optimal growth rates and of network capacity and function. Results In this study, the Saccharomyces cerevisiae genome-scale metabolic network was used to formulate a phenotypic phase plane that displays the maximum allowable growth rate and distinct patterns of metabolic pathway utilization for all combinations of glucose and oxygen uptake rates. In silico predictions of growth rate and secretion rates and in vivo data for three separate growth conditions (aerobic glucose-limited, oxidative-fermentative, and microaerobic) were concordant. Conclusions Taken together, this study examines the function and capacity of yeast's metabolic machinery and shows that the phenotypic phase plane can be used to accurately predict metabolic phenotypes and to interpret experimental data in the context of a genome-scale model. | Background The development of numerous high-throughput experimental techniques such as DNA microarrays, genome sequencing, and protein chips has revolutionized the analysis of biological systems and generated a catalog of information about a cell's components [ 1 - 3 ]. Efforts are now focused on the integration of this data to enable the systemic understanding of cellular functions [ 4 - 6 ]. This integration is typically in the form of a mathematical model that can be used to simulate complex cellular behaviors based on a limited amount of biological data. Several modeling approaches have been implemented in the study of Saccharomyces cerevisiae . Flux-balance models of yeast have appeared for small-scale network reconstructions. Most of these studies are specific for growth conditions, such as anaerobic, glucose-limited metabolism [ 7 ], aerobic growth on galactose [ 8 ] or growth on mixtures of glucose and ethanol [ 9 ]. In addition to flux-balance models, many dynamic models of simplified central metabolic networks in yeast also have been developed [ 10 , 11 ], along with full-scale kinetic models specific to pathways such as glycolysis [ 12 , 13 ] and the pentose phosphate pathway [ 14 ]. These models have been useful to study detailed metabolic events such as concentration changes of individual metabolites and key flux splits. Small-scale reconstructions can be limited in their prediction of cellular functions since these processes are typically dependent on the interaction of components at a whole-cell level. This has motivated the development of genome-scale models, several of which have already appeared for bacterial cells, including Escherichia coli [ 15 , 16 ], Haemophilus influenzae [ 17 ], and Helicobacter pylori [ 18 ]. We have previously reconstructed a genome-scale metabolic network of Saccharomyces cerevisiae based on its annotated genome sequence and a thorough examination of online pathway databases, biochemistry textbooks, and journal publications [ 19 ]. A total of 708 open reading frames, 1175 metabolic reactions, and 733 metabolites are accounted for in this stoichiometric network, which includes both cytosolic and mitochondrial compartments. This genome-scale reconstruction is the most comprehensive model of yeast metabolism to date and has been previously validated through in silico gene deletion studies [ 20 ] and the calculation of key physiological parameters [ 21 ]. The reconstruction and analysis of genome-scale microbial networks have advanced significantly in recent years [ 22 , 23 ], as has the development of a variety of constraint-based modeling methods that allow for the deduction a cell's phenotype based on its genotype and environmental conditions [ 24 - 28 ]. Phenotypic phase plane (PhPP) analysis is a constraint-based method used to obtain a global perspective of genotype-phenotype relationships in genome-scale metabolic networks. In PhPP analysis, flux balance analysis and linear programming are used to map all of the cellular growth conditions represented by two environmental variables onto a two-dimensional plane and identify phases with distinct metabolic pathway utilization patterns. Some applications of PhPP analysis include the study of optimal growth rates [ 29 ], adaptability of microorganisms [ 30 , 31 ], metabolic network functions and capacities [ 15 ], and the impact of gene regulations [ 32 ]. Thus, PhPP analysis provides a way to guide experiments and analyze phenotypic functions based on genome-scale metabolic networks. The constraint-based modelling approach is based on the assumption that organisms have developed control structures to ensure optimal growth in response to environmental constraints [ 35 ]. Numerous experimental observations have been made in support of this hypothesis [ 36 ]. The mathematical descriptions for the PhPP have relevant metabolic meaning for the biological systems being studied. For example, any point in the PhPP corresponds to a single solution of the linear programming problem, which metabolically represents a possible growth behavior. A phase or region in the PhPP (where the shadow price is constant) represents a metabolic phenotype with specific pathway utilization. The shadow prices change continuously at the boundary from one phase to the next. Accordingly, the metabolic phenotype will vary. Metabolically, this is interpreted as a different optimal utilization of the metabolic pathways since each basis solution corresponds to a different flux distribution. In this study, we formulate a glucose-oxygen phenotypic phase plane for yeast based on its recent genome-scale metabolic reconstruction [ 19 ]. The growth states predicted by the PhPP are then characterized using shadow price analysis, in silico gene deletion simulations, and in vivo growth experiments. Finally, we evaluate the network's predictions for these growth states by comparing in silico biomass formation and by-product secretion rates to in vivo measurements. Results S. cerevisiae phenotypic phase plane (PhPP) The S. cerevisiae genome-scale metabolic network constructed by Forster et al. [ 19 ] was used to generate a PhPP [ 33 ] that describes yeast's metabolic states at various levels of glucose and oxygen availability (Fig. 1 ). The surface of the three-dimensional PhPP corresponds to the maximum growth rate allowable for each pair of glucose and oxygen uptake rates in the x-y plane (Fig. 1a ). All feasible metabolic flux distributions lie on or below this surface. The two-dimensional projection of the PhPP (Fig. 1b ) has been divided into seven regions, or "phases," to allow for qualitative comparisons (P 1 - P 7 ). The seven phases represent areas of the PhPP that have distinct metabolic phenotypes as defined by shadow price analysis, which identifies how changes in metabolite levels affect biomass formation [ 33 ]. There also are two regions of the PhPP with infeasible steady-state flux distributions: the area along the y -axis and the small square near the origin. Growth is infeasible in the region between the ordinate and P1 since yeast cannot use more than six oxygen molecules per glucose molecule. The two red lines in Fig. 1b are lines of optimality (LO). LO growth represents optimal aerobic glucose-limited growth of S. cerevisiae in which substrates are completely oxidized to produce biomass and LO ethanol corresponds to maximum ethanol production under microaerobic conditions while growth is maximized. Figure 1 The yeast glucose-oxygen phenotypic phase plane (PhPP). (a) The three-dimensional S. cerevisiae PhPP drawn with Statistica™ (Statsoft, Tulsa, OK). The x and y axes represent the glucose uptake rate and oxygen uptake rate, respectively. The third dimension is the cellular growth rate. (b) A two-dimensional projection of the 3-D polytope in panel (a). The two lines of optimality are shown in red. LO growth represents optimal aerobic glucose-limited growth and LO ethanol corresponds to maximum ethanol production under microaerobic conditions. P 1 - P 7 represent phases with various metabolic phenotypes. The hatched regions correspond to infeasible growth conditions. The orange line (glucose uptake flux = 5 mmol/gDCW/hr) represents the conditions which were used for the simulations in Figure 2. Simulation of optimal metabolic phenotypes Computer simulations (Fig. 1b ) were used to illustrate the change of metabolic phenotypes described by the yeast phase plane. For the simulations, we arbitrarily set glucose uptake rate to 5 mmol/gDCW/hr and varied the oxygen uptake rate from 0 to 20 mmol/gDCW/hr. This allowed us to study the influence of a single environmental variable on cellular metabolism. Small amounts of NH 3 , sulfate and phosphate were introduced for the biomass synthesis. During anaerobic conditions (OUR = 0, on the x -axis), the growth rate was low and the respiratory quotient (RQ) was infinite by definition (Fig. 2a ). As the oxygen uptake rate increased to 13 mmol/gDCW/hr to reach LO growth , the growth rate increased to its maximum value and the respiratory quotient approached 1.06. Further increasing the oxygen uptake rate caused both the growth rate and respiratory quotient to decrease due to futile cycles in which a combination of two or more biochemical reactions resulted only in the hydrolysis of ATP or other high-energy compounds [ 33 ]. Figure 2 Simulation of metabolic behavior for optimal cellular growth as a function of oxygen availability, ranging from completely anaerobic fermentation to completely aerobic growth in S. cerevisiae . The range of oxygen uptake rates used in the simulations (orange line, Fig. 1) allows for the characterization of the PhPP's seven phases (P 1 - P 7 ) and two lines of optimality (LO growth , LO ethanol ). (a) Growth rate and respiratory quotient (RQ). (b) Secretion profile for acetate, succinate, ethanol, and glycerol. Metabolic by-product secretion profiles also were calculated with increasing oxygen uptake rates. Since alternative optimal solutions exist in the genome-scale metabolic flux models [ 34 ], a range of secretion rates can be found amongst all of the equivalent optimal solutions for a fixed point in the PhPP. Remarkably, there was less than 1% difference between the maximum and minimum allowable secretion rates for a fixed maximal growth rate; thus, only the maximum predicted secretion fluxes for ethanol, succinate, glycerol, and acetate are shown (Fig. 2b ). During anaerobic fermentation, ethanol, glycerol, and succinate were produced. Maximum ethanol production occurred at an oxygen uptake rate of 0.5 mmol/gDCW/hr, a condition defining LO ethanol . Glycerol production ceased at this point. With a slight increase in oxygen uptake rate above LO ethanol , acetate began to be secreted but succinate secretion decreased to zero. Ethanol and acetate were no longer secreted once the oxygen uptake rate was equal to or greater than 13 mmol/gDCW/hr, a point on LO growth where the metabolic pathway utilization enables complete aerobic growth. Further characterization of oxidative-fermentative phases (P2 - P6) Linear programming simulations generate parameters called shadow prices that can be used to evaluate how changes in metabolite availability affect the biomass formation [ 33 ]. Shadow price analysis was used to further characterize the oxidative-fermentative phases. A positive shadow price indicated that a metabolite was available in excess, meaning that a decrease in its availability would increase biomass synthesis, and a negative shadow price indicated that a metabolite was limiting such that increasing its availability would increase the biomass synthesis. In silico gene deletions were also performed in order to determine which reactions were essential in each phase. Essential reactions were defined as those whose deletion resulted in no predicted growth (growth rate equal to zero). This approach was especially useful for interpreting the physiological differences between growth states in phases 2 – 6 since their phenotypes were indistinguishable in terms of their secretion profiles. Phase 2 In phase 2, the ratio of oxygen uptake rate and glucose uptake rate (GUR) is lower than that on the line of optimality. As a result, the cell is oxygen limited and begins to ferment. Mitochondrial NAD+ is available in excess, meaning that the biomass synthesis would improve if its availability decreased. In order to maintain the cell's redox balance, the excess mitochondrial NAD+ must be reduced. This is done through the production of acetate and ethanol, which begin to be secreted in this phase. Thus it is the production of acetate and ethanol that makes the optimal growth rate less than that defined on the line of optimality. Phase 3 As the ratio of oxygen and glucose uptake rates is further decreased, three lower glycolysis reactions (fructose bis-phosphate aldolase, triose phosphate dehydrogenase, and phosphoglycerate kinase) become essential for growth in phase 3. Although these deletions severely hinder growth in phase 2 (reducing the growth rate by 55%, 19%, and 19%, respectively), the simulated growth rate is non-zero so these reactions were not considered to be essential. However, these reactions are essential in subsequent phases as the oxygen uptake rate is further decreased. Due to the limited oxygen, more carbons "overflows" into the fermentation pathway while at the same time oxidative metabolism becomes less effective. Phase 4 Shifting from phase 3 to phase 4, the pentose phosphate pathway is utilized to generate NADPH because not enough NADPH is produced through respiration at the lower oxygen uptake rate. The NADPH is then converted to NADH which is subsequently used for ATP production. Phase 5 Further lowering the ratio of oxygen and glucose uptake rates restricts the cell's ability to produce pyruvate in phase 5. Yeast can no longer utilize the oxidative pathways because an insufficient amount of cytosolic NAD+ is produced. When comparing phases 4 and 5, all of the metabolites with shadow price sign changes were folate intermediates. These are important energy carriers that are directly linked to the availability of both cytosolic and mitochondrial NAD+ and NADP+. Phase 6 As you enter phase 6, the acetate production is completely ceased. Ethanol is secreted as the only metabolic by-product to balance the redox potential of the cell. Growth experiments Three groups of experiments were conducted under different growth conditions in the PhPP (Fig. 3a ). These three conditions were: Figure 3 Growth experiments shown on the PhPP. (a) The three groups of experimental data displayed on the S. cerevisiae PhPP were used as an index for the time course profiles in panels (b), (c) and (d). (b) Aerobic glucose-limited growth controlled by fed-batch operation. (c) Oxidative-fermentative growth with unlimited glucose and oxygen availability. (d) Microaerobic growth with unlimited glucose and very low oxygen availability. The AGL (b) and MA (d) data sets are located on lines of optimality and as a result are stable metabolic states with only one degree of freedom (glucose for AGL and oxygen for MA). OF (c) is an unstable metabolic state with two degrees of freedom (glucose and oxygen), making it more difficult to control this type of growth condition. By perturbing the environmental conditions, cells in OF can be shifted to either AGL or MA (unpublished results). • Aerobic, glucose-limited (AGL) growth experiments were conducted with a data acquisition and process control system. The dissolved oxygen (DO) level was maintained above 30% by sparging the compressed air into the CelliGen ® Plus bioreactor during the cell cultivation. The system was controlled at the respiratory quotient of 1.06 for optimal S. cerevisiae growth by RQ-stat feeding strategy to maintain the glucose concentration at a low, stable level (Fig. 3b ). The cell concentration rose steadily with almost no acetate or ethanol production and most of the carbon was incorporated into the yeast biomass. • Oxidative-fermentative (OF) batch growth experiments were carried out by allowing an essentially unlimited supply of oxygen and glucose. The DO level was maintained above 30% by sparging the compressed air into the shaker's flasks during the cell cultivation. Ethanol and acetate were accumulated in the aerated processes (Fig. 3c ). • Microaerobic (MA) batch cultivations with S. cerevisiae were performed at a low dissolved oxygen level. The experiments were performed in side-arm flasks (Fig. 3d ) in which a small amount of air was allowed to diffuse into the flasks via the cotton filter on the side arms. The DO level was measured to be less than 5%. The initial glucose concentration and the limited oxygen supply resulted in high levels of ethanol and low levels of acetate. Integration of experimental data and in silico predictions The S. cerevisiae PhPP is a genome-scale model-based visualization platform which allows experimental data and simulation results to be displayed and compared. The three groups of batch and fed-batch experimental data are projected on Fig. 3a using the experimentally measured OUR and glucose uptake rates. These rates were then used as constraints in the computer simulations. Table 1 shows that the experimental observations and the in silico predictions are in good agreement. Table 1 Comparison of In Silico Predictions and Experimental Measurements. Microaerobic fermentation Oxidative-fermentative growth Aerobic, glucose-limited growth OUR = 1, GUR = 14 OUR = 9, GUR = 12 OUR = 8, GUR = 2.5 In silico Experimental In silico Experimental In silico Experimental Growth rate 0.33 0.31 0.53 0.51 0.22 0.20 Ethanol 21.29 20.08 11.98 11.07 0 0.16 Acetate 0.26 0.22 2.62 2.57 0 0.31 Abbreviations: oxygen uptake rate (OUR), glucose uptake rate (GUR) Units: growth rate (1/hr), substrate uptake rates and metabolite production rates (mmol/gDCW/hr) Discussion In this study, the S. cerevisiae genome-scale metabolic network constructed by Forster et al. [ 13 ] was used to generate a PhPP [ 9 ] that describes yeast's metabolic states at various levels of glucose and oxygen availability (Fig. 1 ). Examination of the S. cerevisiae PhPP has led to clear interpretation and prediction of its metabolic capabilities. First, only a few distinct optimal S. cerevisiae growth phenotypes were found in silico , and these phenotypes correspond to well-defined phases of the PhPP. Second, two lines of optimality were identified in yeast's PhPP: LO growth , which represents optimal biomass production during aerobic, glucose-limited growth, and LO ethanol , which corresponds to both maximal ethanol production and optimal growth during microaerobic conditions. The predictions of S. cerevisiae 's PhPP and genome-scale model were compared to independent experimental data. The results showed that the agreement between the computed and observed growth rates, uptake rates, and secretion rates was close to the measured values or within the experimental error, and qualitatively the predictions agreed with published literature. Analysis of experimental data within the PhPP formalism suggests that yeast has only a few primary phenotypes, designated by the various phases. In P 1 , the oxygen supply is sufficient for growth by aerobic respiration, resulting in carbon dioxide as the sole by-product. Phases P 2 -P 6 correspond to states of oxidative-fermentative growth, which is characterized by secretion of oxidative and fermentative metabolic by-products, i.e. , acetate and ethanol, respectively. These states are highly similar since the phases are essentially co-planar in the 3-dimensional PhPP (Fig. 1a ). The secretion profile (Fig. 2b ) does not show any phenotypic differences between phases P 2 - P 6 . However, through the use of shadow price analysis and in silico gene deletions, distinct pathway utilization patterns could be found for each phase. Finally, P 7 represents microaerobic conditions. In this environment, yeast grows primarily by fermentation and secretes ethanol, glycerol, and succinate (Fig. 2b ). This limited range of metabolic states is strikingly different from that found for E. coli , whose glucose-oxygen PhPP has five distinct optimal in silico phenotypes [ 32 ]. Comparison of the simulation results generated by the E. coli [ 35 ] and S. cerevisiae [ 19 ] models indicates that E. coli 's metabolic by-product secretion patterns are more sensitive to the OUR variation than those of S. cerevisiae . Moreover, computer simulations show that when the OUR is lower than 7 mmol/gDCW/hr at a glucose uptake rate of 5 mmol/gDCW/hr, the TCA cycle in E. coli is broken into two branched pathways, one operating as a reductive pathway reversing the usual sequence from succinate to oxaloacetate and the other continuing to operate oxidatively to convert oxaloacetate to α-ketoglutarate. For S. cerevisiae, the pathway still functions as a cycle even when the OUR is as low as 1 mmol/gDCW/hr. Thus, it can be concluded by the comparison of E. coli and S. cerevisiae metabolic networks that yeast appears to be more robust to environmental perturbations. Furthermore, we predict that yeast secretes fewer metabolic by-products under these conditions, suggesting that its metabolism is more efficient than that of E. coli . This may represent a universal difference in how prokaryotes and eukaryotes respond to shifts in environmental parameters. Another feature that distinguishes the S. cerevisiae PhPP from the E. coli PhPP is the existence of two definable lines of optimality. The conditions that define LO growth are similar to those that define E. coli 's sole line of optimality, i.e. they both represent the relationship between the glucose and oxygen uptake rates that results in optimal growth rate. Analysis of yeast's PhPP suggests that at a specific ratio of glucose and oxygen uptake rates glycerol production is halted and NADH is re-oxidized by maximal ethanol formation. This phenomenon, defined by LO ethanol , has been supported by many research reports in the literature [ 3 , 17 , 25 ]. For example, Cysewski and Wilke [ 37 ] found a sharp stimulation of the specific ethanol productivity at a very low but non-zero level of dissolved oxygen. Later studies showed that a value of 10 ppb of dissolved oxygen maximized ethanol production in yeast chemostat cultures [ 25 ]. Thus, LO ethanol , the second line of optimality predicted by the genome-scale model, is consistent with the experimental observations. A useful application of the S. cerevisiae PhPP is to qualitatively classify yeast's metabolic state based on phenotypic observations made in vivo (Fig. 3 ). The metabolite concentration profile obtained from all of the experiments qualitatively agrees with the corresponding metabolic state predicted by the PhPP. For example, in growth conditions near LO ethanol , cells are expected to grow almost entirely by fermentation, with significant production of ethanol and lesser amounts of glycerol, acetate and succinate secretion. This phenotype is qualitatively similar to experimental observation, in which more ethanol is produced than acetate as shown in Fig. 3d . Points in the PhPP representative of the three data sets also were used to quantitatively predict yeast's metabolic phenotype (Table 1 ). Overall, the predictions are in good agreement with the experimental measurements. However, the predicted growth rates are slightly higher than the measured values. This difference may result from the model's prediction of optimal performance not reflecting suboptimal growth in vivo . Conclusions The genome-scale metabolic networks developed for other microorganisms, namely Escherichia coli , Haemophilus influenzae , and Helicobacter pylori , have led to useful insights into substrate preferences, the effects of gene deletions, optimal growth patterns, outcomes of adaptive evolution, and shifts in expression profiles [ 22 ]. With the recent reconstruction of S. cerevisiae 's genome-scale metabolic network [ 19 ], these analytical techniques can now be applied to the first genome-scale model of an eukaryotic cell. By developing methods such as the PhPP to explore in silico the metabolic capabilities of microorganisms, we can generate new hypotheses as to how these organisms operate, and, more importantly, we can gain insight into the impact of individual cellular components on the organism as a whole. Methods Experimental methods Strains and media All cultures were grown at 30°C in SD medium [ 38 ] and supplemented with glucose (Sigma Chemical Co., St. Louis, MO) as appropriate for each phase of the experiment conducted. The S. cerevisiae strain FY4 MATα [ 39 ] was used in this study. Growth and fermentation system For experiments, 5 ml of overnight culture inoculated from single colonies grown on YPD agar was used to seed 50 ml of SD media pre-warmed to 30°C in a 250-ml Erlenmeyer flask, which was placed in a 30°C shaking incubator at 225 rpm for approximately 12 hours. This secondary seed was then used to inoculate either a 1.5-L Erlenmeyer flask with side arms for parallel batch fermentations or a 1.0-L bioreactor (CelliGen ® Plus, New Brunswick Scientific Co., Inc., Edison, NJ, USA). Cultures for aerobic, glucose-limited, fed-batch growth were initially grown in a batch mode, and a specific substrate limited after the culture reached particular biomass concentrations in each respective experimental condition. All batch culture experiments were performed in our multiple fermentation system which consists of acrylic enclosures filled with de-ionized water that can accommodate 32 cultures in parallel in batch operation mode at volumes ranging from 100 mL to 1500 mL capacity. We used this setup with either shaker's flasks as reactor vessels. A magnetic agitator (Bellco Glass, Inc., Vineland, NJ, USA) was used to continually mix flask contents at a speed of 225 rpm, and each flask was sealed with a rubber stopper containing apertures for probes, nutrient inlets/outlets, and sample harvesting. Temperature is strictly and uniformly controlled using a water circulator (model C10, Thermo Haake, Portsmouth, NH) with a temperature control module that drives a closed circuit of water to and from the controller to inlet and outlet drains on the water bath. Dissolved oxygen is measured and controlled using a polarographic electrode connected to DO meters/controllers. The fed-batch S. cerevisiae cultivations were automatically controlled in the 1-L bioreactor (CelliGen ® Plus, New Brunswick Scientific Co., Inc., Edison, NJ, USA). It has its own controllers for temperature, pH and dissolved oxygen (DO). A Pentium II computer (233 MHz processor, Microsoft Windows 98) equipped with an AT-MIO-16E-10 Analog Input computer interface board (National Instruments Corp., Austin, TX, USA) was used for data acquisition and process control for both the multiple fermentation system and CelliGen ® Plus bioreactor. Data from the batch and fed-batch cell cultures, including pH, temperature, and dissolved oxygen concentration were acquired through the interface board. The real-time graphical data acquisition and process control programs was written in LabVIEW 6.0 (National Instrument Corp., Austin TX). Media fed to the bioreactor was controlled by a feeding pump (Masterflex Computerized Drive 7550-90, Cole-Parmer Instrument Co., Chicago, IL, USA), with a RS 232 serial link accepting control signal from the computer, for fed-batch cell cultivation process. Acquisition of dissolved oxygen (DO) data was obtained with the aid of a respirometer [ 40 ] using a dissolved oxygen probe (Cole-Parmer Instrument Co., Chicago, IL, USA). For all the experiments, temperature was controlled at 30°C. Determination of respiratory quotient (RQ) The ratio of carbon dioxide evolution rate (CER) to oxygen uptake rate (OUR) has previously proven useful in inferring a lack of substrates in the growth medium and in the calculation of feeding rates [ 41 ]. For the fed-batch experiment, compressed air was fed into the bioreactor through a gas flowmeter (Manostat 125, New York, NY, USA), which was manually adjusted to a flow rate of 100 mL/min. The composition of exhaust gas from the bioreactor was measured using a gas analyzer (1440C Gas Analyzer, Servomex Co., Inc., Norwood, MA, USA) connected to the interface board to gauge exiting O 2 and CO 2 levels. Calculations for CER, OUR, and RQ were performed using the equations: where O 2, in and CO 2, in and O 2, out and CO 2, out are the oxygen and carbon dioxide fractions in % v/v in the inlet air and exiting gas measurements, respectively, Q in and Q out are the air flow rates, and V m is the working volume of the bioreactor. When the estimated RQ reached 1.06, a peristaltic pump (Cole-Parmer) was utilized to begin feeding 10X concentrated growth medium into the bioreactor, and this quotient was maintained by an RQ-stat control strategy to limit the production of by-products or consumption of these by-products as an alternative energy source. Sampling procedures During cultivations, two separate 1-ml aliquot samples were taken at early, mid, and late log-phase from both the bioreactor and flasks. The first aliquot was used to determine cell density by measuring the optical density A 600 , A 420 , using a spectrophotometer (Beckman DU640, Beckman Coulter, Inc., Fullerton, CA, USA), and cell counts (Coulter Electronics Inc., Hialeah, FL, USA). The second aliquot was then filtered on a manifold containing a dry, pre-weighed, 0.2 μm pore-size filter to isolate a cell pellet, and was washed three times with 250 ml of sterile deionized water to ensure all salts were removed. After washing, the filters are placed in aluminum foil inside a 65°C incubator for 24 hours and subsequently weighed in an analytical balance to measure cell dry weight. The second aliquot was filtered through a 0.45 μm acrodisc syringe filter to separate cells from supernatant. The concentrations of metabolites in the supernatant such as glucose, acetate, ethanol, and glycerol were determined by using enzyme-based assay kits (glucose and acetate assay kit, Sigma Chemical Co., St. Louis, MO, USA; ethanol and glycerol assay kit, R-Biopharm, Inc., Marshall, MI, USA). Growth rate, specific uptake/production rates and OUR flask All specific growth rate curves were obtained by a linear regression of all data points within the exponential growth phase using the following formula X = X o e μt , where X is the cell concentration (gL -1 ), X o is the initial inoculum cell concentration (gL -1 ), t is the time of inoculation, and μ is the specific growth rate (1 hr -1 ). A minimum of ten optical density measurements were needed for the growth rate determination for both batch and fed-batch cultures. The specific glucose uptake rate (GUR), ethanol and acetate formation rates, and OUR flask (OUR for the batch culture using flasks) were determined by fitting the dynamic mass balance equations for glucose, ethanol, acetate and DO measurements to the data points spanning the time period of the exponential cell growth phase. The specific uptake and production rates were then calculated by solving the dynamic mass balance equation within the culture medium using the following equation: where V (L) is the culture volume, [S] (mM) is the substrate/product or DO concentration in the flask, q (mmole/g-dry weight/hr) is the substrate uptake rate or by-product formation rate or OUR flask , and X(t) (g-dry weight/L) is the biomass concentration at time = t (hr). + is for the by-product formation and - is for the substrate consumption. The solution to this equation was fitted to the experimental data points using the nonlinear estimation routine in Statistica (StatSoft Inc, Tulsa OK) or the solver in Microsoft Excel. All data, to be considered valid and included in the analysis, must have correlation coefficients of 0.95 or greater. The data that were generated in this way represented the "pseudo-steady-state" [ 42 ] of the batch or fed-batch cell culture, and thus suitable for the calculation of growth rate, specific uptake and production rates, and OUR flask . In silico calculations Flux balance analysis and linear programming A genome-scale S. cerevisiae metabolic network has been reconstructed [ 19 ]. The network includes 733 metabolites and 1175 metabolic reactions, which are compartmentalized between the cytosol and the mitochondria. In metabolic network analysis, the relationship between metabolite concentrations, x , and reaction activities, v , is described by the dynamic mass balance equation [ 43 , 44 ]: where S is an m × n matrix of stoichiometric coefficients, x is an m × 1 vector of metabolite concentrations, and v is and n × 1 vector of reaction activities. Thus, the rows of S correspond to the internal metabolites and the columns represent the reactions in the network. Under steady-state conditions, the dynamic mass balance equation simplifies to: S • v = 0 (Eq. 6) Since the number of reactions is often greater than the number of metabolites, Eq. 6 is underdetermined and contains multiple solutions. One approach to solving Eq. 6 for microbial networks is to define a set of inputs and outputs that correspond to the growth conditions and use linear optimization to maximize the cell growth [ 35 ]. This approach has been successful in capturing the phenotypic behavior of S. cerevisiae for various growth conditions [ 21 ]. Phenotypic phase plane (PhPP) formulation The S. cerevisiae PhPP displays optimal growth rates for all possible variations in two constraining environmental variables, such as the carbon substrate and oxygen uptake rates. In this study, the glucose uptake rate (x-axis) was allowed to vary from 0 to 20 mmol/gDCW/hr and the oxygen uptake rate (y-axis) ranged from 0.1 to 20 mmol/gDCW/hr. The oxygen uptake rate was not allowed to reach zero because anaerobic simulations required additional supplements to maintain cell growth (ergosterol and zymosterol). Linear programming was used to calculate the optimal growth rate for all points in the x-y plane. Growth rate values were then plotted as the z-axis to form the surface of a three-dimensional PhPP (Fig. 1a ). A two-dimensional PhPP was formed by projecting the 3-D PhPP onto the x-y axis (Fig. 1b ). The phases of the PhPP were determined by the calculation of shadow prices [46], which describe the sensitivity of the objective function (Z) to changes in the availability of each metabolite: where b i is the i th metabolite and γ i is the i th shadow price. Shadow prices were calculated for each point in the x-y plane during the linear programming simulations. By definition, phases were identified as regions of the PhPP in which all of the points have the same shadow prices. Lines of optimality, which represent the optimal ratio of glucose and oxygen uptake rates for maximal biomass synthesis, were also identified using shadow price analysis [ 45 ]. Shadow price analysis and in silico gene deletions To obtain a physiological interpretation of the differences between the oxidative fermentative phases (phases 2–6), we analyzed how the shadow prices of key metabolites changed across the phase boundaries. Simulations were run at a fixed glucose uptake rate of 5 mmol/gDCW/hr and an oxygen uptake rate ranging from 1.5 to 15 mmol/gDCW/hr. The sign of the shadow price was used to identify whether a small change in the metabolite's availability would positively or negatively affect the objective value. According to the convention defined in [ 45 ], a negative shadow price indicates that a metabolite is limiting, e.g. the value of the objective function increases if the metabolite's net production increases or its net consumption decreases. Similarly, a positive shadow price indicates that a metabolite is available in excess and a shadow price equal to zero indicates that a change in the availability of the metabolite does not affect the objective value. Phases 2 – 6 were also characterized by performing gene deletions in silico (as described in [ 20 ]). Single genes were deleted at a representative point within each phase to determine which reactions were essential for viability in that region. Secretion profile calculations The first step in generating the secretion profile was to calculate the optimal growth rate for a given glucose and oxygen uptake rate. For the simulations in Figures 2a and 2b , the glucose uptake rate was fixed at 5 mmol/gDCW/hr and the oxygen uptake rate varied from 0 to 16 mmol/gDCW/hr (ergosterol and zymosterol uptake rates of 5.92 × 10 -5 and 1.27 × 10 -4 mmol/gDCW/hr, respectively, were used for the calculation at OUR = 0). The simulations were then re-run with a fixed glucose uptake rate, oxygen uptake rate, and growth rate to determine the maximum and minimum secretion rates of each metabolite with a shadow price equal to zero. Authors' contributions NCD calculated the phenotypic phase plane (Fig. 1 ), carried out the secretion profile simulations (Fig. 2 ), characterized the phases, calculated the flux predictions for Table 1 , and drafted the manuscript. BOP conceived the study, participated in its design and coordination, and assisted with manuscript preparation. PF designed the study and conducted the growth experiments (Fig. 3 , Table 1 ) and revised the manuscript. All authors have read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC520746.xml |
524252 | The Risk of a Mosquito-Borne Infectionin a Heterogeneous Environment | A common assumption about malaria, dengue, and other mosquito-borne infections is that the two main components of the risk of human infection—the rate at which people are bitten (human biting rate) and the proportion of mosquitoes that are infectious—are positively correlated. In fact, these two risk factors are generated by different processes and may be negatively correlated across space and time in heterogeneous environments. Uneven distribution of blood-meal hosts and larval habitat creates a spatial mosaic of demograPhic sources and sinks. Moreover, mosquito populations fluctuate temporally, forced by environmental variables such as rainfall, temperature, and humidity. These sources of spatial and temporal heterogeneity in the distribution of mosquito populations generate variability in the human biting rate, in the proportion of mosquitoes that are infectious, and in the risk of human infection. To understand how heterogeneity affects the epidemiology of mosquito-borne infections, we developed a set of simple models that incorporate heterogeneity in a stepwise fashion. These models predict that the human biting rate is highest shortly after the mosquito densities peak, near breeding sites where adult mosquitoes emerge, and around the edges of areas where humans are aggregated. In contrast, the proportion of mosquitoes that are infectious reflects the age structure of mosquito populations; it peaks where old mosquitoes are found, far from mosquito breeding habitat, and when mosquito population density is declining. Finally, we show that estimates for the average risk of infection that are based on the average entomological inoculation rate are strongly biased in heterogeneous environments. | Introduction Understanding the spatiotemporal distribution of risk for mosquito-borne infections is an important step in planning and implementing effective infection control measures ( Greenwood 1989 ; Charlwood et al. 1998 ; Chadee and Kitron 1999 ; Focks et al. 1999 ; Mendis et al. 2000 ; Carter 2002 ; Killeen et al. 2003 ). Remote sensing, geographical information systems, and predictive algorithms have made it possible to develop coarse-grained maps of vector habitat ( Pope et al. 1994 ; Beck et al. 1997 ; Kitron 1998 ; Rogers et al. 2002 ), and epidemiological studies have identified statistical risk factors for human infection or disease ( Snow et al. 1998 ; Ghebreyesus et al. 2000 ; Snow and Gilles 2002 ). Mathematical models can bridge the gaps between landscape ecology, vector biology, and human epidemiology, linking large-scale maps to individual risk in local human populations at spatial scales ranging from 10 m up to 10 km. At these spatial scales, transmission dynamics for vector-borne infections are linked to the seasonal dynamics, demography, and behavior of adult female mosquitoes, as well as the spatial distribution of larval habitat and blood hosts ( Bidlingmyer 1985 ). Given a map of potential or actual mosquito sources and human habitation, what factors determine where and when the risk of a mosquito-borne infection is highest? The risk of a mosquito-borne infection is estimated by the entomological inoculation rate (EIR): the number of bites by infectious mosquitoes per person per day ( Macdonald 1957 ). EIR is the product of the human biting rate (HBR)—the number of bites by vector mosquitoes per person per day—and the proportion of mosquitoes that are infectious (PIM) (e.g., for malaria transmission, the sporozoite rate) ( Birley and Charlwood 1987 ). We focus on the processes that generate patterns in the two components of EIR in temporally and spatially heterogeneous mosquito populations. We show that HBR and PIM peak at different times and places. We also show that estimates of average EIR in variable environments generate biased estimates of the relationship between EIR and the proportion of humans that are infected ( Dye and Hasibeder 1986 ). We develop theory to illustrate simple patterns in the EIR in heterogeneous environments, focusing on EIR's separate components. We follow the a priori approach of Ross, developing simple mathematical models as tools for qualitative and quantitative reasoning ( McKenzie 2000 ). We develop a basis for micro-epidemiological models for mosquito-borne infections that can be combined with surveys of larval habitat to map the local risk of mosquito-borne infections ( Greenwood 1989 ). Effective use and refinement of such maps depend on an understanding of the dynamics and behavior of specific mosquito populations and the transmission of specific infectious agents ( Focks et al. 1999 ). Results Temporal Heterogeneity Fluctuating mosquito density affects EIR through changes in HBR: transmission increases as mosquito density increases. Following an increase in the rate at which adult mosquitoes emerge, mosquito density and HBR peak (illustrated in Figure 1 ). The peak in EIR and the density of infected mosquitoes follow the peak in total mosquito density because it takes time for an infectious agent to spread through the human and mosquito populations. Increased HBR leads to secondary increases in the proportion of infected humans, and thus to increases in PIM. As the density of infected mosquitoes declines, decreasing transmission is followed by a decline in the prevalence of infection in humans. Figure 1 Dynamics with Temporal Heterogeneity The components of EIR follow different trends when mosquito populations vary temporally. Mosquito density (solid black) forms the dominant component of HBR. The density of infected mosquitoes (solid gray) peaks shortly after the density of mosquitoes (dotted vertical lines align the peaks). In contrast, the proportion of infectious mosquitoes (dashed) peaks while the mosquito population is declining. Seasonal mosquito emergence was modeled to have an long-term average M/H ≈ 2 ( K = 2 and H = 1). The ticks on the x-axis mark the peaks of the wet and dry seasons. In contrast, larger fluctuations in PIM are generated by the shifting age distribution in fluctuating mosquito populations. Adults emerge uninfected, but they become infected some time after biting infectious humans. Growing populations are dominated by young, uninfected mosquitoes, while shrinking populations are dominated by older mosquitoes. Since the proportion of mosquitoes that are infected and infectious increases with the age of the mosquito, PIM is a proxy for the age distribution of mosquito populations. As populations decline, surviving mosquitoes continue to bite and oviposit but few young mosquitoes emerge, so declining populations have a larger fraction of old mosquitoes. Thus, PIM increases during the dry season as mosquito populations, HBR, and EIR decline. Spatial Heterogeneity The distribution of adults is determined by the distribution of larval habitat, the distribution of blood hosts, and the alternating activities of blood-meal-seeking and oviposition. When mosquito emergence rates and human population distributions are constant over time, the distribution of mosquitoes reaches a static spatial distribution. We focus on the patterns that form along a transect. In Figure 2 , we assume a single point source for mosquitoes and a homogeneous distribution of humans. In Figure 3 , the same number of adult mosquitoes emerges, but the spatial distribution of larval emergence is uniform along the transect and the distribution of humans varies: human density is low at one end, high at intermediate locations, and intermediate at the opposite end, approximating a small town with fewer dwellings on the edge nearest a swampy area. In Figure 4 we combine the two kinds of spatial heterogeneity. Figure 2 Statics with Homogeneous Humans and Heterogeneous Mosquitoes The components of EIR follow different trends when larval habitat is distributed at a single point and humans are uniformly distributed (the gray background illustrates the human distribution). (A) Mosquito density (solid) declines monotonically, but PIM (dashed) increases monotonically. The density of infected mosquitoes ( Z, dotted) also declines monotonically. (B) HBR (solid) and EIR (dashed) both decline monotonically away from the source, reflecting the steep gradient in mosquito density. (C) The density of infected humans (dashed) and prevalence of infection in humans (solid) also decline monotonically (the curves coincide). Figure 3 Statics with Heterogeneous Humans and Homogeneous Mosquitoes HBR and EIR reflect mosquito movement and human distribution patterns when larval habitat is evenly distributed but humans have a low–high–medium distribution, such as a town with rural and suburban populations on either side (the gray background illustrates the human distribution). (A) Mosquito density (solid) is highest in town, peaks at the edges of town, and dips just outside of town. PIM (dashed) and the density of infected mosquitoes ( Z, dotted) follow similar patterns. (B) HBR (solid) and EIR (dashed) are both high on the low-density side of town and lowest on the medium-density side of town, with peaks just inside town and troughs just outside of town. (C) The density of infected humans (dashed) and the prevalence of infection in humans (solid) peak at the edge of town, but prevalence of infection in humans is less variable than HBR or EIR. Figure 4 Statics with Heterogeneous Humans and Mosquitoes When human density increases smoothly away from a larval habitat (the gray background illustrates the human distribution), the patterns of EIR components reflect heterogeneity in the distribution of larval habitat and human populations. (A) Mosquito density (solid) peaks an intermediate distance away from the source. The peak density of infected mosquitoes ( Z, dotted) is further from the source because PIM (dashed) increases monotonically away from the source. (B) HBR (solid) decreases monotonically away from the source, reflecting mosquito density, but EIR (dashed) has a minor peak away from the source. (C) The density of infected humans (dashed) peaks away from the source, but the prevalence of infection in humans (solid) remains relatively constant near the source, dropping off sharply further away. Gradients in EIR Away from Larval Habitat When mosquitoes emerge from a point source, the density of mosquitoes tends to decline with distance from larval habitat, such as a gradient along a transect away from a swamp or river ( Figure 2 A). The shape of the gradient is determined by the emergence rate of adult mosquitoes, the mortality of existing mosquitoes, and random drift away from the source. In contrast, PIM increases monotonically away from the source because of a shift in the age distribution and parity of mosquitoes ( Figure 2 B). Young mosquitoes tend to be close to their birthplace because they have moved less; older mosquitoes have moved more and so are dispersed further from the source, on average. The spatial distribution of HBR and EIR reflect the gradients in mosquito density, not the gradient in PIM ( Figure 2 A and 2 B). The prevalence of infection in humans declines monotonically with distance from the mosquito source ( Figure 2 C). Heterogeneous Distributions of Humans When human populations are distributed heterogeneously, but the larval habitat of mosquitoes is distributed uniformly, adult mosquito distributions become heterogeneous because mosquitoes tend to aggregate around humans. Whether this leads to an increase in HBR depends on whether mosquito distributions become more aggregated than the distribution of their human hosts. HBR tends to increase when searching mosquitoes move rapidly through sparse human populations and linger in areas with dense human populations. Thus, mosquito distributions tend to become more aggregated than human distributions when the mosquito species is long-lived with long daily flight distances (see below). We illustrate this principle for one particular set of parameters that leads to increased mosquito aggregation. The human population is distributed heterogeneously in blocks of low, high, and medium density, approximating a town with a rural population on one side and an intermediate-density population on the other. The distribution of adult mosquitoes is influenced by the distribution of humans ( Figure 3 A). Aggregations of mosquitoes form spontaneously at the edges of human settlements simply because mosquitoes tend to move until they find a host. We note that the major peaks in HBR are away from town, where human population density is lowest, and at the edge of town, where human population density is highest ( Figure 3 B). EIR also peaks at the edge of town, but it is lowest on the low-human-density side of town. With these movement rules, the mosquitoes found on the side of town with low human density tend to be younger, hence PIM is low ( Figure 3 A). The prevalence of infection in humans is lowest overall in the patches with low human density ( Figure 3 C). This model also makes the surprising prediction that the risk of infection is lowest just outside the edge of town: the sharp difference in human density at the edge leads to a strong tendency for mosquitoes to be drawn into, rather than away from, town, decreasing HBR and PIM ( Figure 3 B and 3 C). Heterogeneous Larval Habitat and Human Population When mosquitoes and humans are distributed unevenly, the distribution of mosquitoes and risk may be dominated either by proximity to larval habitats and gradients away from them or by the tendency of mosquitoes to aggregate around humans. The realized pattern depends on the relative distribution of larval habitat and humans, and whether mosquito aggregation around humans increases HBR. We illustrate one kind of pattern for parameters that lead to increased HBR. In this case, human density increases away from larval habitat. The density of mosquitoes peaks a short distance from the source, and the density of infected mosquitoes peaks slightly further away ( Figure 4 A). HBR declines monotonically away from the source, but EIR peaks at an intermediate distance ( Figure 4 B). The density of infected humans peaks well away from the source, but the fraction of infected humans remains relatively constant near the source, declining abruptly at distances beyond the peak in infected humans ( Figure 4 C). Despite the sharp peaks in risk, PIM displays a robust monotonic increase with distance away from the source ( Figure 4 A). If the gradient is reversed, so that human density decreases with the distance away from larval habitat, mosquitoes remain close to the source and mosquito aggregation is exaggerated, compared with Figure 2 (data not shown). Measuring EIR in Heterogeneous Environments Variability in EIR across a landscape can lead to systematic bias in the estimation of risk. In Figure 5 , we plot local EIR and its components against the local prevalence of infection in humans for the individual patches in Figures 2 – 4 . We also plot the average EIR for each transect. In addition, we overlay the temporal patterns from Figure 1 as a phase diagram. We note that local EIR and local prevalence of infection in humans at equilibrium,
x¯, have a clear nonlinear relationship given by the following formula: Figure 5 The Relationship between EIR and Human Prevalence, with Heterogeneity (A) EIR and the prevalence of infection in humans have a tidy relationship among patches; each small symbol is from a single patch in Figures 2 – 4 . The relationship, given by equation 1 , is plotted in gray. The phase plane of the dynamic relationship over time from Figure 1 is plotted with dashed lines. Average EIR is plotted against the average prevalence (large symbols). Predicting the average prevalence of human infection from average EIR leads to underestimates. (B–D) The density of infectious mosquitoes (Z) (B), HBR (HBR = aM / H ) (C), and PIM (PIM = Z / M ) (D) are plotted against the proportion of humans who are infected and infectious. PIM is a particularly bad measure of the risk of infection; in heterogeneous habitats, it peaks far from larval habitat, where mosquito density and prevalence of infection in humans is lowest. This accounts for the large number of points where PIM is high, but the proportion of infectious humans is low. In contrast, the relationship between average EIR and average prevalence of infection in humans is biased, such that average prevalence always falls below the true relationship ( Figure 5 A). The bias is due to an inherent mathematical property of nonlinear relationships known as Jensen's inequality ( Krantz 1999 ). Since the relationship between EIR and the prevalence of infection in humans is concave down, aggregating estimates of EIR in variable habitat will always underestimate the true relationship, sometimes spectacularly ( Ruel and Ayres 1999 ). The local density of infectious mosquitoes ( Figure 5 B) and the local HBR ( Figure 5 C) provide reasonably good estimates of risk. In both cases, spatial heterogeneity in human density or PIM is a substantial source of variability in measures of average risk. In contrast, PIM displays no clear pattern along the transect ( Figure 5 C). The patches in which PIM is high but the prevalence of infection in humans is low are all far from larval habitat. Sensitivity Analysis The patterns illustrated in Figures 2 – 4 are based on a single set of entomological parameters in order to facilitate comparisons among situations in which only the distributions of mosquitoes and hosts vary. The distribution of risk will change for different values of the parameters. We explored the effects of mosquito movement and the duration of the incubation period on the distribution of risk (below and Protocol S1 ). The tendency of mosquitoes to aggregate at the edges of a town or away from larval habitat depends on mosquito searching behavior and demography. Three important parameters that affect these patterns are the maximum daily flight distance of a mosquito, mosquito longevity, and mosquito searching efficiency. The distribution of a mosquito cohort initially reflects the distribution of larval habitat. As mosquitoes search for hosts, the distribution of the cohort shifts to reflect the distribution of human hosts. These tendencies are also reflected in the static spatial distributions of mosquitoes. The distribution of long-lived mosquitoes with long daily flight distances will tend to reflect the underlying distribution of humans, while the distribution of short-lived mosquitoes with short flight distances will tend to reflect the distribution of larval habitat. Mosquito searching efficiency determines the relative rates of movement through habitats that vary in human density. A strong tendency for mosquitoes to aggregate at the edges of dense human populations occurs when mosquitoes move quickly through areas that are sparsely populated by humans and linger in areas that are heavily populated. In other words, mosquitoes tend to become more aggregated than their hosts, increasing HBR, when mosquito searching is relatively inefficient at low human densities. The distribution of relative risk also changes with the time required for incubation of the infectious agent, with mechanically transmitted agents at one extreme. When all else is equal, HBR is higher in areas in which human density is low, since human population density is in the denominator of HBR. On the other hand, HBR may decline in low-human-density areas because mosquitoes tend to move up a gradient of human population density in search of a blood-meal host. Such migration will tend to lower the average age of mosquitoes in low-human-density patches, especially near the edge of a town. This will tend to lower PIM for infectious agents with a long incubation period. In contrast, PIM for mechanically transmitted infectious agents will not be as strongly affected, so in comparison, the relative risk may be higher at that same edge of town. Discussion EIR is generally considered to be the best estimate of the risk of mosquito-borne infections, but EIR varies over space and time. EIR varies spatially because larval habitat and blood-meal hosts are heterogeneously distributed across a landscape. Temporal variability is generally driven by weather, especially rainfall, temperature, and humidity. To compound the problem, heterogeneity in human feeding over short distances can be caused by vector preferences for individual humans based on odor or other cues ( Takken and Knols 1999 ; Kelly 2001 ). Heterogeneous biting has important implications for the dynamics and control of mosquito-borne infections ( Dietz 1980 ; Dye and Hasibeder 1986 ; Woolhouse et al. 1997 ). Heterogeneous biting also has important implications for the measurement of EIR. Estimates of EIR may vary substantially over short distances depending on the place and time at which the measurement is made. Depending on the method used, EIR may also vary with the relative attractiveness of the human bait. Our mathematical models have shown that average EIR in heterogeneous environments gives a strongly biased estimate of average risk, even when local estimates of EIR provide a perfect measure of local risk. The bias is unavoidable because the relationship between EIR and the proportion of humans who are infected is nonlinear, which leads to a bias due to Jensen's inequality ( Krantz 1999 ). A similar bias is likely to arise when estimating risk for other infectious diseases, a problem that is pervasive and generally underappreciated in epidemiology and public health ( Ruel and Ayres 1999 ). Therefore, mathematical models are an indispensable tool for the design and interpretation of field studies ( Becker 1989 ). Mathematical models provide a sound approach to understanding risk and planning for control in heterogeneous environments, especially when the models are based on the ecology of the local vector populations and a sound understanding of the entomological parameters relevant for transmission ( Killeen et al. 2000a , 2000b ). Creating micro-epidemiological maps for the distribution of risk would involve mapping larval habitat and humans, and combining these maps with an understanding of the temporal dynamics, blood-meal-seeking behavior, and oviposition habits of mosquitoes. A critical assumption of the models described here is that mosquitoes are able to oviposit everywhere. Vector species may be very selective about where they oviposit, forcing a return to larval habitat to oviposit between successive bites. Thus, the heterogeneous distribution of oviposition sites may also affect the distribution of risk. A dominant component of EIR is the density of mosquito vectors relative to human density. Our models show that mosquito densities and the proportion of humans who carry a mosquito-borne infection decline with distance away from larval habitat because of random movement and mosquito mortality. Such patterns have been documented by numerous field studies ( Trape et al. 1992 ; Hii et al. 1997 ; Charlwood et al. 1998 ; Clarke et al. 2002 ; Minakawa et al. 2002 ; Keating et al. 2003 ; Staedke et al. 2003 ; Konradsen et al. 2003 ; van der Hoek et al. 2003 ). The steepness of the gradient in EIR varies, depending on the ecology of the vector ( Hii et al. 1997 ). Unlike the patterns in human biting, the proportion of mosquitoes that are infectious depends on the age structure of the mosquito population. The likelihood of infection in mosquitoes shows a strong association with the age or parity of mosquitoes ( Lines et al. 1991 ). The average age differs in growing, stable, and declining populations ( Aron and May 1982 ). Our models also predict that the proportion of infectious mosquitoes increases monotonically with the distance away from sources of emerging adults; young, pre-gravid mosquitoes are found more frequently near larval habitat, while older mosquitoes are found further away. Such patterns have also been observed in the field ( Charlwood et al. 1998 ). Mosquito aggregation around dense human populations depends on the details of mosquito searching behavior. Long-lived mosquitoes with long flight distances tend to become more aggregated than their human hosts over intermediate distances. For example, EIR may peak at the edges of a village, as has been documented by one field study ( Ribeiro et al. 1996 ). At larger spatial scales, increasing human density may decrease EIR; one field study concluded that human density was protective against disease ( Snow et al. 1998 ). We have emphasized heterogeneous biting that arises from proximity to larval habitat and from mosquito aggregation due to blood-meal-seeking behavior. In our models, aggregation is generated by the tendency of mosquitoes to migrate more slowly when blood-meal hosts are readily available. Aggregation in human biting may be enhanced if mosquitoes fly toward humans that are more attractive at medium and long distances ( Ansell et al. 2002 ; Mukabana et al. 2002 ). It remains to be seen how these factors interact; for instance, at what distances are preferred hosts more attractive to mosquitoes ( Ansell et al. 2002 )? The use of remote sensing and GIS provides a potentially powerful tool for understanding the distribution of mosquito-borne infections at large spatial scales, but dynamics and control of mosquitoes and mosquito-borne infections occur locally. These technologies will be most effective if they are coupled with micro-epidemiological models of malaria, dengue, and other mosquito-borne infections ( Greenwood 1989 ). Such models can predict variability in local risk based on the distribution of larval habitat, the distribution of humans, and the demography and behavior of the local vectors. To generate realistic predictions for the distribution of risk, it is necessary to understand where and when adult mosquitoes will emerge, and how blood-meal-seeking and the distribution of humans will affect the distribution of HBR. It follows that a knowledge of mosquito demography and behavior should play a central role in the surveillance and control of mosquito-borne infections. Materials and Methods We use mathematical models strategically, to illustrate general principles that may apply to many mosquito-borne infections, not to make predictions about the distribution of a particular infectious agent or the incidence of disease. The models we present and analyze are based on the models for malaria infection developed by Ross (1911) . We generate a suite of complex models by elaboration, adding a realistic incubation period, temporal heterogeneity, mosquito movement, patchy space, and spatial heterogeneity ( Black and Singer 1987 ). By comparing models, we associate an effect with a factor. First, we allow mosquito birth rates to vary temporally, and focus on the temporal changes in the components of EIR ( Aron and May 1982 ). Next, we illustrate how spatial variability in the distribution of larval habitat generates source–sink relationships in landscapes and leads to variability in the spatial distribution of HBR and PIM. Then, we explore the consequences of heterogeneous human distributions. Host-seeking behavior by mosquitoes can produce mosquito distributions that are more (or less) aggregated than the distribution of humans, generating an uneven distribution in risk. Thus, we develop conceptual models to illustrate which components of the vector biology determine the distribution of risk. The model Let x denote the proportion of humans who are infected and infectious and H denote the population density of humans. We assume that the human infectious period is exponentially distributed with average duration of infection 1/ r . Thus, we are following Ross in developing a model for infection ignoring superinfection, immunity, and clinical disease ( Fine 1975 ; Aron and May 1982 ; Cohen 1988 ; Dietz 1988 ). We extend the Ross model by adding temporal variability in mosquito density. Let ɛ ( t ) denote the rate adult female mosquitoes emerge from larval habitat; we do not assume that the emergence of adults is explicitly linked to the density of adult mosquitoes. Let M denote the population density of mosquitoes, Z the density of infectious mosquitoes, and z = Z / M the proportion of mosquitoes that are infectious. We assume that the mosquito lifespan is exponentially distributed with a mean lifetime of 1/ g d. We incorporate a realistic incubation period by subdividing the incubation period into n stages of equal duration; the proportion of mosquitoes that are infected and incubating in stage k is denoted y k , and the density of mosquitoes in that stage is Y k . We assume the incubation period has mean of 1/ q d. The probability of surviving the incubation period is ( qn /( qn + g )) n (approximately e − g / q for large n ), and the duration of the incubation period (for surviving mosquitoes) has a Gamma distribution with mean 1/ q and variance 1/( q 2 n ); for the numerical simulations, we use n = 64 ( Protocol S1 ). The larger n is, the smaller the variance is. In the limit as n approaches infinity, the dynamics approach a fixed time delay. Let a denote the human feeding rate, the number of human bites per mosquito per day, b denote the probability an uninfected human becomes infected from a single bite from an infectious mosquito, and c denote the probability that a mosquito becomes infected from biting an infectious human host. The transmission dynamics of mosquito-borne infections are complex, and it is easy to lose sight of what terms such as EIR and HBR actually mean. HBR is the number of bites received by a human each day. Thus, it is the product of the human feeding rate, a —the number of human blood meals per mosquito per day—and the number of mosquitoes per human (i.e., HBR = aM / H ). Therefore, when mosquito density changes, HBR changes proportionally. In contrast, EIR is the number of infectious bites per human per day. Thus, it is the product of PIM and HBR (i.e., EIR = z HBR = azM / H ). Table 1 lists variable and parameter names and other important terms for the models. Table 1 Variables and Parameters Used in the Model The dynamic process is embedded into a spatial context by subdividing a landscape into a set of patches linked by the movement of mosquitoes. The subscript i is added to variable names to denote the value in the i th patch. Thus, H i denotes local human population density and x i the local prevalence of infection in humans. Similarly, M i denotes local mosquito population density, and Z i denotes the density of infectious mosquitoes. The density of infected mosquitoes in patch i and incubation stage k is denoted Y i,k . This deterministic approach to incorporating space has some limitations ( Mollison 1984 , 1986 ; Durrett and Levin 1994a , 1994b ). Larval habitat and human distributions form a template that determines mosquito distributions and the distribution of risk. The emergence rate of adults in the i th patch is ɛ i ( t ); the emergence of adult female mosquitoes depends predictably on time and location. Following emergence, female mosquitoes spread into surrounding areas seeking blood hosts; they feed, oviposit, and then repeat the cycle. We assume that heterogeneity in larval habitat takes the form of differences in quality of larval development rather than availability of places to oviposit. In other words, we assume that suitable sites for oviposition are distributed homogeneously throughout the habitat, but that patches may vary in the successful development of adults. Some patches may produce no adults. Heterogeneity in the availability of oviposition habitat would affect the distribution of risk because mosquitoes would alternate between finding a place to oviposit and finding a blood meal. If oviposition were not possible in most patches, those that allowed oviposition would become focal points for mosquito aggregation. Thus, these results apply mainly to mosquito species for which heterogeneous availability of oviposition sites is relatively unimportant for the distribution of risk. We assume that humans do not move among patches. The density of humans and the productivity of the larval habitat may vary over space. As we change the distribution of humans and larval habitat to explore the effects of spatial heterogeneity, we hold the total emergence rate of adult mosquitoes per human constant; only the distribution of humans and adult mosquito emergence changes. We assume that mosquitoes are more likely to stay in a patch if they encounter a human, and that they are more likely to find humans where humans are more abundant. Let Φ( H i ) denote the per capita emigration rate of mosquitoes away from patch i regardless of infection status. We assume that Φ( H ) is a decreasing function of H; the more humans, the less likely mosquitoes are to leave a patch in search of another blood-meal host. Thus, mosquitoes move more rapidly through patches with low human densities. A parameter, κ i,j , describes the fraction of mosquitoes leaving patch i that fly to patch j, and Σ j κ i,j = 1. Thus, the rate that mosquitoes move from patch i to patch j is Φ( H i ) κ i,j M i . The transmission dynamics are described by the following set of equations: This patch-based modeling framework is suitable for modeling an array or grid of contiguous habitat or an arbitrary network of patches. Numerical solutions Our intent is to focus on the effects of temporal and spatial heterogeneity. Consequently, we have used a single set of mosquito life-history parameters and a single duration of infection in humans. The parameters are roughly consistent with Anopheles gambiae and the infectious period for malaria ( a = 0.3; b = c = 0.5; 1/ g = 1/ q = 10 d). The human infectious period for this case is 100 d ( r = 0.01), roughly consistent with malaria. Constant mosquito populations were modeled using a constant birthrate, ɛ i ( t ) = K i g, while temporal heterogeneity was modeled using the seasonal forcing function ɛ i ( t ) = K i g (1 sin(2 πt /365)). In a homogeneous landscape, K i is the long-term average density per patch, often called the carrying capacity. Throughout, K was chosen such that the average number of mosquitoes per human across all patches was 2, i.e., Σ i M i /Σ i H i = 2. Figure 1 was generated using a single patch. Initial conditions were x = 0.01 and Y i,k = Z i = 0. We generated numerical solutions for 4 y and plotted the last three. For Figures 2 – 4 , we focused on the relatively simple patterns that form along a spatial transect, a linear array of seventeen patches that can be thought of as a long, rectangular island. We have assumed that κ i,j = 0 unless two patches are adjacent, and we plot the values at equilibrium. We assume that no humans live in the patches at the extreme ends of the transect, and that all of the mosquitoes leaving one of these edges return to the adjacent patch; thus κ 1,2 = κ 17,16 = 1, a reflective boundary. Otherwise, we assume that mosquitoes move in either direction at random; thus, κ i,j = 0.5 for i = 2…16 and j = i ± 1. Mosquito migration was described by the function
Φ(H i ) = ζ e −θH i . In Figures 2 – 4 , we used ξ = 10 and θ = 4. These correspond to a maximum daily flight distance (i.e., without humans) of about ten patches per day. Adult mosquito emergence for Figures 2 and 4 was gK ( P − 2) in patch 1 ( K = 2 and P = 17); no adults emerged within other patches. The adult emergence rate for Figure 3 was gK in each patch with humans ( K = 2 in patches 2–16). For Figure 2 , human density was 1.0 in patches 2–16. For Figure 3 , human density was 0.2 in patches 2–6, 1.8 in patches 7–11, and 1.0 in patches 12–16. For Figure 4 human density was (0,1,2,3,…,15,0)/120. Otherwise, the parameters were the same as in Figure 1 . Supporting Information Protocol S1 Additional Methods (285 KB PDF). Click here for additional data file. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC524252.xml |
546399 | Lifestyle in adults aged 35 years who were born with open spina bifida: prospective cohort study | Background and Methods From 1963 to 1971, 117 babies with open spina bifida were treated non-selectively from birth. In 2002 we reviewed all the survivors by postal questionnaire and telephone call. The aims were to find out how many were living independently in the community or were in open employment or drove a car. In addition to these achievements we recorded health, medication and admissions to hospital and asked how much daily help they needed. Results Ascertainment was 100%. There had been 63 deaths, mainly of the most severely affected. The mean age of the 54 survivors was 35 years. The outcome in terms of disability ranged from apparent normality to total dependency. It reflected both the neurological deficit, which had been recorded in infancy in terms of sensory level, and events in the CSF shunt history. Overall about 2 in 5 of the survivors lived independently in the community, 2 in 5 drove a car, 1 in 5 was in competitive employment and 1 in 5 could walk 50 metres. Conclusion Although those who survived to age 35 years tended to be less disabled, 2 in 5 continued to need daily care. | Background Neurosurgical intervention in babies with open spina bifida had dramatic results in terms of survival. However, the disability and the complications of the survivors were often severe [ 1 - 5 ]. Many efforts were made to enable them to walk, to control their urinary incontinence while safeguarding renal function, and to overcome problems associated with the shunt treatment of hydrocephalus. Promising new methods of management, such as the psoas transplant, urinary diversion and artificial urinary sphincters, which seemed highly successful in the short term, lost favour after 10 or 15 years because of disappointing long-term results. In this unsteady course of progress it is helpful to have a long term follow up of a complete cohort of patients with open spina bifida as a realistic basis for helping parents facing the difficult decisions about termination of an affected pregnancy or treatment after birth. Methods Patients In 1963 the Regional Neurosurgical Unit at Addenbrooke's Hospital, Cambridge, England offered treatment to all cases of open spina bifida, without any attempt at selection. Between 1963 and 1971, after a detailed neurological examination, 117 babies (50 male, 67 female) had their open spinal defects closed within 48 hours of birth. A ventriculo-atrial cerebrospinal fluid (CSF) shunt was inserted for hydrocephalus when required. Data collection In 2002 all survivors were surveyed by confidential questionnaire and telephone interview. They were asked about health, disability and achievements in terms of living independently, driving a car and working in open employment. Causes of death for those who had died were obtained from medical records and from the Office of National Statistics. The study was approved by the Cambridge Local Research Ethics Committee. Statistical analysis When first surveyed at the mean age 4 years, the cohort had been classified into four groups according to sensory level to pin prick recorded in infancy [ 4 ]. Those with intact sensation right down to the knee (sensory level below L3) had a better short-term outcome than those with no sensation below the umbilicus (sensory level above T11). Mortality and measures of disability and achievement were compared in those with different sensory levels and CSF shunt histories using χ 2 . Results Ascertainment was 100%. Twenty (37%) of the 54 survivors responded to the questionnaire and all survivors or a carer or relative were interviewed by telephone. Mortality Figure 1 summarises the outcome for the complete cohort. Sixty-three cases had died, 25 before their first birthday and a further 15 before their fifth. Thereafter the death rate remained constant with an average of 1% of the remainder dying each year [ 6 ]. Figure 1 Outcome in open spina bifida at the mean age of 35 years Survivors Of the original 117 babies there were 54 survivors (46%) age range 31–38 years. Of these 24 were male and 29 female and there was one who had undergone a gender change from male to female. The disability of the survivors ranged from blindness with paraplegia and double incontinence to apparent normality. Table 1 shows that sensory level in infancy was a predictor of overall disability, the need for a CSF shunt, IQ, and need for a wheelchair or daily care at age 35 years. Table 1 Sensory level in infancy related to disability at the mean age of 35 years in 54 survivors with spina bifida Sensory level n = 54 (%) Below L 3 n = 24 L 3 -T 11 n = 15 Above T 11 n = 12 Asymmetrical n = 3 χ 2 for trend (1) Severe disability (2) n = 20 (37) 2 6 12 0 p < 0.0001 CSF shunt n = 46 (85) 17 15 11 3 p < 0.05 IQ < 80 n = 15 (28) 3 4 6 2 p = 0.05 Wheelchair n = 38 (70) 9 14 12 3 p < 0.0001 Daily care needed n = 20 (37) 5 6 9 0 p < 0.05 Lack of achievement (3) n = 27 (50) 7 9 10 2 p < 0.01 Notes: (1) Patients with lower sensory levels have less disability. Asymmetrical sensory level excluded from the analysis. (2) Severe disability defined as very poor mobility and incontinent with additional handicaps including low IQ, epilepsy, visual defects (2 blind), severe spinal deformity and pressure sores. (3) Lack of achievement in terms of living independently, driving a car or working in open employment. Mobility Only 16 (30%) remained community walkers defined as being able to walk ≥50 metres with or without aids. Ten of the 16 could walk at least a kilometre. Table 2 shows the deterioration in walking since childhood and its relationship to sensory level. By the age of 35 there was only one community walker with a sensory level as high as L3. But of those with a low sensory level of L5 and below, 88% (14/16) remained walkers. In terms of motor function, 30 had been recorded as having bilateral quadriceps activity in infancy, but only 53% of them (16/30) remained walkers at the age of 35 years. Table 2 Influence of sensory level and age on walking in 54 survivors with spina bifida Sensory level in infancy n = 54 Walkers 1 at age 9 n = 31 (57%) Walkers at age 35 n = 16 (30%) Above T11 n = 12 0 0 T11-L3 n = 15 5 1 L4 n = 8 8 1 L5-S2 n = 6 6 5 No sensory loss n = 10 10 9 Asymmetrical loss n = 3 2 0 1 Walkers defined as able to walk ≥50 metres using aids if required. Survivors with lower sensory levels more likely to be walkers. (χ 2 for trend p < 0.0001 for both age 9 and age 35. Asymmetrical sensory loss excluded from the analysis.) Cerebrospinal fluid (CSF) shunts Of the 54 survivors, eight (15%) never had a shunt, seven of whom had little or no disability and a sensory level below L3. The remaining 46 had had a ventriculo-atrial shunt inserted. In 16 the shunt had never been revised. The other 30 had had a total of 104 revisions: for shunt insufficiency (65), infection (15), detachment (14), extrusion or leaking wound of back (5), unknown (5). In 9 patients revisions were done only before the age of 2 (mean 1.3 revisions, range 1–3), and in 21 between the ages of 2 and 35 (mean 3.1 revisions, range 1–14). Elective revisions were not performed; and shunts were inserted or revised only in response to definite clinical need. Of those who had revisions, 75% had had symptoms of raised intracranial pressure. Health Nearly half of the survivors had been in hospital during the previous 5 years. The main reasons were urological (7 patients), neurosurgical (3), and sepsis (7). Pressure sores were responsible for four of the admissions for sepsis, and 12 patients were currently being treated at home for pressure sores. Only 11 patients (20%) were fully continent of bladder and bowel without the use of catheters or appliances. Two patients needed nocturnal respiratory support, two were totally blind following shunt dysfunction and four others had severe visual defects. Endocrine conditions were common: two patients had diabetes mellitus, one had adrenal hyperplasia, one had primary azoospermia and six had had precocious puberty. Twenty-four patients were on long-term therapy: antihypertensives (12), anticonvulsants (10), antibacterials (10) and antidepressants (4). Eight patients needed regular analgesics for musculo-skeletal pain, mainly backache. Parenthood Seven women and two men had become parents. One man had minimal disability and no detectable sensory loss; the other had undergone percutaneous epididymal sperm aspiration followed by intracytoplasmic sperm injection. None of the 13 children had visible spina bifida. Residence and dependency Twenty-two individuals (41%) lived independently in the community, 11 of them used wheelchairs. A further 12 (22%) were personally independent but had supervision and help when required. The remaining 20 (37%) needed help daily for dressing, shaving, toilet or nursing care (mainly pressure sores). Ten of these still lived with a parent now aged 52–77, two women were in the care of their partners, five were in residential establishments and three lived in the community with help from social services (Table 3 ). Table 3 Where are the 54 survivors living? Residence and Dependency Number of individuals Percentage Independent living 22 41 Sheltered environment with help available 12 22 Dependent on daily help 20 27 Car drivers Twenty-nine (54%) of the survivors had passed the driving test, but 9 had discontinued driving. Eleven others were unfit to drive on account of poor sight (3), epilepsy (3) or severe cognitive or perceptual defects (5). Employment Nine men and four women were in open employment. All had an IQ ≥80 and five used wheelchairs. Three did clerical work, three were teachers, two were unskilled manual workers and the remainder were a business executive, accountant, engineer, van driver and builder. Three were studying in addition to working full time. Three men and two women were in sheltered employment. Lifestyle and achievements Twenty-seven survivors (50%) had one or more achievements in terms of living completely independently in the community (22), driving a car (20) or working in open employment (13). Achievements were related to sensory level in infancy and to shunt history (Tables 1 and 4 ). All but one of the 8 patients without a shunt and 75% (12/16) of those in whom the shunt was never revised were classified as achievers. They lived independently or drove a car or worked in open employment compared with 40% (4/10) of those needing revision at age <2 and 20% (4/20) of those revised after age 2 (p < 0.01). Table 5 shows that late revisions of shunt after the age of 2 were also associated with a birth head circumference ≥90th centile relative to birthweight, a history of symptoms of raised intracranial pressure, visual defects and the need for daily care. Table 4 Lifestyle related to history of CSF shunt in 54 survivors at the mean age of 35 years Lifestyle No shunt n = 8 Shunt not revised 1 n = 16 Shunt revised age <2 n = 10 Shunt revised age 2–35 n = 20 Living independently n = 22 (41%) 7 9 4 2*** Driving a car n = 20 (37%) 5 8 4 3* In open employment n = 13 (24%) 3 6 3 1* Any achievement (2) n = 27 (50%) 7 12 4 4** χ 2 comparing those with shunt revisions aged 2–35 with those never revised or revised at age <2. *p < 0.05 **p < 0.01, ***p < 0.001. Those with no shunts excluded from the analysis. (1) Shunts were only inserted or revised in response to definite clinical need such as symptoms or signs of raised intracranial pressure. (2) Any achievement: in terms of living independently, driving a car or working in open employment. Table 5 Features related to CSF shunt history in 54 patients with open spina bifida at the mean age of 35 years Features No shunt n = 8 Shunt not revised n = 16 Shunt revised age <2 n = 10 Shunt revised age ≥2 n = 20 χ 2 for trend Birth head circumference ≥90 centile n = 10 0 1 1 8 p = 0.05 History of symptoms of raised intracranial pressure n = 23 0 1 5 17 p < 0.0001 Visual defects (mainly squint) n = 33 3 7 6 17 p < 0.05 Daily care needed n = 20 1 2 2 15 p < 0.001 Discussion By the mean age of 35 years, over half the cohort had died, mainly the most disabled. About 40% of the survivors lived independently, 20% needed some support and 40% needed daily care. Lifestyle and achievements depended on the degree of disability, which could have been forecast from sensory level recorded in infancy, indicating the extent and severity of the neural deficit. Data from this cohort show that babies with sensation below the knee (L3) are unlikely to be seriously disabled and could be achievers in adulthood. Babies who cry during a routine heel prick have a sensory level of S1 or below and are likely to remain community walkers in adulthood. Those with sensation to pin prick in the saddle area (S2, 3, 4) are likely to have bladder and bowel control. Events in the history of the CSF shunt also had a profound effect on outcome and achievement. Revisions of shunt were associated with poor achievement particularly when the revisions were needed after the age of 2. The cranial sutures have usually fused by the age of 2 after which the skull is less expansile rendering the brain more susceptible to pressure [ 7 ]. Table 5 shows that most of those who had revisions after the age of 2 had had symptoms or signs of raised intracranial pressure. By contrast an uneventful shunt history was sometimes associated with remarkable achievement despite severe disability. This may imply that it is the raised intracranial pressure which has the adverse long term effect on achievement and enterprise [ 8 ]. The main strength of the study is the community basis, which provides social as well as clinical data enabling the realities of adulthood to be seen against the optimistic forecasts of the early years [ 9 ]. Less than half of the survivors were still attending hospital. Thus a hospital-based study would have given an incomplete picture. As the patients grew older, the reduction in support, rehabilitation and encouragement from dedicated physiotherapists, parents and other carers revealed an outcome which was related to the patient's own motivation and enterprise as well as to the basic neurological deficit [ 8 ]. The main limitation of the study is that the very long follow up relates to some treatments, which have been superseded. Improvements in the diagnosis and management of renal and neurological problems have halved the mortality by the age of 5 [ 10 , 11 ], but have less influence on long term disability. Although outcome in childhood of early operated spina bifida has been widely reported [ 1 - 5 , 11 ], there are few studies of long term outcome. McLone has argued strongly that prognosis is improving due to advances in treatment [ 12 ]. Our results may not predict outcome using today's standards of care. However, a recent survey from McLone's group of a cohort of 118 adults aged 20–25 with 16% loss to follow up, found continuing deterioration and a formidable number of neurosurgical and spinal operations [ 13 ]. Ours is the only 35-year prospective study of open spina bifida with 100% ascertainment by the same independent observer. Conclusions These data may help health professionals counselling parents of a child with spina bifida. They show a range of possible outcomes in adulthood when parents may no longer be able or willing to look after their child [ 14 - 16 ]. Two out of 5 survivors continue to need daily care. Advances in treatment may have improved prognosis, but the most important predictor remains the basic neurological deficit. Those looking after patients with spina bifida need to know both their long term potential and the limitations of treatment in order to focus on realistic goals [ 17 ]. Competing interests The authors declare that they have no competing interests. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC546399.xml |
547905 | Common variants of the beta and gamma subunits of the epithelial sodium channel and their relation to plasma renin and aldosterone levels in essential hypertension | Background Rare mutations of the epithelial sodium channel (ENaC) result in the monogenic hypertension form of Liddle's syndrome. We decided to screen for common variants in the ENaC βand γ subunits in patients with essential hypertension and to relate their occurrence to the activity of circulating renin-angiotensin-aldosterone system. Methods Initially, DNA samples from 27 patients with low renin/low aldosterone hypertension were examined. The DNA variants were subsequently screened for in 347 patients with treatment-resistant hypertension, 175 male subjects with documented long-lasting normotension and 301 healthy
Plasma renin and aldosterone levels were measured under baseline conditions and during postural and captopril challenge tests. Results Two commonly occurring βENaC variants (G589S and a novel intronic i12-17CT substitution) and one novel γENaC variant (V546I) were detected. One of these variants occurred in a heterozygous form in 32 patients, a prevalence (9.2%) significantly higher than that in normotensive males (2.9%, p = 0.007) and blood donors (3.0%, p = 0.001). βENaC i12-17CT was significantly more prevalent in the hypertension group than in the two control groups combined (4.6% vs. 1.1%, p = 0.001). When expressed in Xenopus oocytes, neither of the two ENaC amino acid-changing variants showed a significant difference in activity compared with ENaC wild-type. No direct evidence for a mRNA splicing defect could be obtained for the βENaC intronic variant. The ratio of daily urinary potassium excretion to upright and mean (of supine and upright values) plasma renin activity was higher in variant allele carriers than in non-carriers (p = 0.034 and p = 0.048). Conclusions At least 9% of Finnish patients with hypertension admitted to a specialized center carry genetic variants of β and γENaC, a three times higher prevalence than in the normotensive individuals or in random healthy controls. Patients with the variant alleles showed an increased urinary potassium excretion rate in relation to their renin levels. | Background Epidemiological studies have shown a significant correlation of blood pressure levels in close relatives and higher concordance values for occurrence of hypertension in monozygotic vs. dizygotic twins, and thus support the idea that genetic factors influence susceptibility to essential hypertension [ 1 ]. While recent molecular genetic studies have provided compelling evidence for mutations in at least seven different genes underlying rare forms of monogenic hypertension [ 1 , 2 ], progress in the understanding of the molecular basis of human essential hypertension has been much slower. Hundreds of case-control studies have suggested hypertension-related genetic variants of which only a few if any have tolerated replication analyses; it is possible that common variants of angiotensinogen [ 3 ], α-adducin [ 4 ] and the G-protein β subunit [ 5 ] confer susceptibility to elevated blood pressure in at least some populations. Since 1999, a number of genome-wide linkage studies in families with multiple affected hypertensive members have been published with highly varying results (for review, see [ 6 ]). Recent large-scale searches for genes predisposing to hypertension, published as a recent series of articles [ 7 - 11 ], failed to identify definite linkage of hypertension to any chromosomal locus, although some DNA regions showing suggestive linkage were disclosed. Reasons for these disappointing data were put on the account of the unsuitability of using a single-locus linkage strategy for a multifactorial genetic disease, inherent genetic heterogeneity of essential hypertension, and complex interplay of genetic and environmental factors underlying regulation of blood pressure variation [ 12 ]. Disappointments in the previous strategies justify alternative approaches in which a better phenotyping of the study individuals is connected to their targeted molecular genetic characterization. There are several features that collectively make the genes encoding the beta (βENaC) and gamma (γENaC) subunits of the kidney tubular epithelial sodium channel as serious candidates for susceptibility genes of low-renin human essential hypertension. First, gain of function mutations in β and γ ENaC subunits cause Liddle's syndrome, a well-known monogenic form of human hypertension associated with low renin activity and low plasma aldosterone level [ 13 - 15 ]. Second, common βENaC variants occur in increased frequency in hypertensive black individuals [ 16 - 18 ]. Third, an extensive locus-targeted study on hypertensive family members demonstrated a significant linkage of hypertension to chromosome 16q region harboring both the βENaC and γENaC genes [ 19 ]. These data prompted us to carry out a search for common variants of these two genes in Finnish hypertensive patients who were admitted to a special center because of treatment-resistant hypertension and whose renin-aldosterone system was systematically examined. These circumstances provided a group of hypertensive patients, in which secondary forms of hypertension were effectively excluded and who originated from a genetic isolate. Our data suggest that common variants of the ENaC subunits confer susceptibility to human essential hypertension. Methods Patients with hypertension The clinical records of all consecutive patients with hypertension (n = 615) referred to the Hypertension Outpatient Ward, Helsinki University Central Hospital, between 1992–96 were reviewed. Moderate-to-severe hypertension, suspicion of secondary forms of hypertension, or hypertension resistant to drug treatment were causes to the admittance. A letter with request to donate a blood sample for genetic studies on hypertension was sent to those 598 individuals whose address became available in 1998. A total of 399 individuals (67%) of these responded and were subsequently examined at the Hypertension Outpatient Ward in 1998 to 1999. Clinical and family histories were recorded, and venous blood samples taken for DNA analysis. Based on the previous documents and current examinations, altogether 52 individuals were excluded from the present study: clinical records were missing or insufficient in four cases, 22 subjects were considered as normotensive, while 26 were judged to have a secondary form of hypertension. The latter group consisted of the following cases: renal artery stenosis (n = 12), adrenal cortical adenoma (n = 3), hydronephrosis (n = 2), pheochromocytoma (n = 1), IgA glomerulonephritis (n = 1), non-specific chronic glomerulonephritis (n = 1), LED nephritis (n = 1), diabetic nephropathia (n = 1), chronic pyelonephritis (n = 1), hypernephroma (n = 1), fibromuscular dysplasia (n = 1), unspecified renal failure (n = 1). The remaining 347 patients (186 females and 161 males, mean age 49.3 years, SD ± 10.0) comprised our final cohort of patients with moderate-to-severe essential hypertension. Antihypertensive drug treatment was in use in 283 (82%) of the patients (diuretics, 19%; beta-blocking agents, 35%; calcium-channel blockers, 21%; ACE-inhibitors, 33%; angiotensin receptor antagonists, 1%). At least two concomitant drugs were used by 24% of the patients. A flow-chart of the study design is illustrated in Fig. 1 . Control individuals Blood donors DNA was extracted from 301 randomly selected healthy blood donors aged 40–50 years (mean, 45 years) visiting the Finnish Red Cross Blood Transfusion Service. Their residences represent the same capital area from which the hypertensive patients originated. Normotensive controls These individuals were selected from the participants in the Alpha-Tocopherol, Beta Carotene (ATBC) study [ 20 ] using the criteria described previously [ 21 ]. In brief, a total of 27271 male smokers (aged 50 to 69 years) with no previous history of myocardial infarction were initially recruited for a cancer prevention trial. DNA samples were available from 70% of the original participants. We picked up all the available blood samples from those fulfilling the following criteria: no known hypertensive disorder, no antihypertensive drugs ever in use, systolic and diastolic blood pressure values ≤ 128 and ≤ 84 mmHg, respectively, at each blood pressure measurement, repeated five times at one-year intervals during a five-year follow-up. We ended up with 175 normotensive men whose mean systolic and diastolic blood pressures were 114.9 (SD ± 5.4) and 73.7 (SD ± 4.3) mmHg, respectively, during this five-year follow-up. The Ethics Review Committee of the Helsinki University Central Hospital approved this study, and all patients and controls gave their informed consent. Laboratory measurements in the hypertensive patients The patients were advised to stop using estrogens and spironolactone at least 4 weeks before the tests, diuretics and prostaglandin inhibitors at least two weeks before the tests, and β-adrenergic antagonists and ACE inhibitors at least one week before the test. The only antihypertensive agents permitted at the time of the test were calcium channel blockers. Some of the patients were on oral potassium supplementation because of hypokalemia. The mean baseline blood pressure level at the time of captopril test was 139 ± 16/94 ± 10 mmHg in those without any drugs (n = 79), and 142 ± 16/95 ± 11 mmHg in those with calcium channel blockers (n = 234). Fasting blood samples were taken for determination of serum creatinine, uric acid, cholesterol, potassium, sodium and blood glucose concentrations. Urine samples for determination of the daily (24 h) excretion of potassium and sodium were collected. Most hypertensive patients (n = 298) underwent a test for the responsiveness of serum aldosterone level and plasma renin activity to postural change. To this end, the first blood sample was taken after at least 60 minutes of rest in supine position. After 2 hours of standing and moderate walking, a second blood sample was taken. This test was carried out at the inpatient ward in 220 cases and at the outpatient ward in 78 cases. Urinary electrolyte excretion rates were analyzed in 262 patients (26 ENaC variant carriers and 236 non-carriers) who did not use potassium supplementation. One to three days later, a captopril challenge test (CCT) was carried out as described earlier [ 22 ]. This test was carried out in a total of 313 patients, and was performed at the inpatient ward in 229 cases and at the outpatient ward in 84 cases. CCT was started by sitting for at least 30 minutes, followed by oral administration of 50 mg captopril. Blood pressure in the non-dominant arm was measured at 15-minute intervals. Blood samples for the determination of plasma renin activity and serum aldosterone concentration were drawn immediately before and 60 minutes after captopril administration. DNA analysis Genomic DNA was extracted from peripheral venous blood using standard techniques. For targeted search for ENaC variants postulated to be associated with increased channel activity, we chose to sequence the exons 13 coding for the carboxyterminal domains of βENaC (amino acids 515–640) and γENaC (amino acids 524–649), as well as the 5'-flanking intronic regions, using oligonucleotide primers, PCR (polymerase chain reaction) conditions and sequencing instruments described previously by us [ 21 ]. DNA samples of 27 patients of those 399 initially visiting the Hypertension Outpatient Ward showing the lowest plasma renin activities (median 0.7 μg/L/h at 0 minutes and 0.9 at 60 minutes) and serum aldosterone concentrations (median 236 pmol/L at 0 minutes and 212 at 60 minutes) during CCT were selected for this initial step. Specific PCR-based methods were set up for assaying the three ENaC variants detected during the present study. After PCR of the βENaC fragment, the βENaC -i12 -17CT and βENaC G589S variants could be assayed simultaneously. An aliquot (8 μl) of the PCR product was digested with 3.0 U of Alu I (New England Biolabs, Beverly, Massachusetts, USA), followed by analysis of the cleavage products on a 12% polyacrylamide gel. The wild-type (wt) allele results in longest fragments of 266 and 137 bp, while the variant allele produces fragments of 266 and 147 bp for βENaC -i12 -17CT, and 240 and 137 bp for βENaC G589S. For the γENaC V546I variant, 2.0 U of Sfa NI (New England Biolabs) was used and the cleavage products were analyzed on a 2% agarose gel. The resulting fragment sizes were 357 and 77 bp for the wild-type allele and 279, 78 and 77 bp for the variant allele. For studies on the possible splicing errors brought about by the βENaC i12-17 variant, we collected lymphocytes from two subjects heterozygous for this variant and one control subject. Total lymphocytic RNA was isolated using Qiagen RNeasy kit (Qiagen, Valencia, California, USA), and first strand synthesis was performed using Superscript system for RT-PCR (Invitrogen Corporation, Carlsbad, California, USA). For gene-specific PCR, we used two sets of primers amplifying a fragment extending from exon 12, either to exon 13 (180 bp) or 141 bp downstream of exon 13 (551 bp). The amplified products were run on a 12 % polyacrylamide gel and visualized by ethidium bromide. The amplified fragments were also sequenced to exclude presence of any splicing defects. Additionally, possible splicing differences between the wild-type and i12-17 variant of βENaC were studied in silico using GrailEXP v3.3 (Perceval) exon prediction program [ 23 ]. Site-directed mutagenesis and functional characterization of the ENaC variants The human βENaC cDNA and γENaC cDNA cloned into the pBSK-SP6-globin vector were used in construction of the βENaC G589S and γENaC V546I mutations, respectively. Site-directed mutagenesis was performed using the Transformer site-directed mutagenesis kit (Clontech Laboratories, East Meadow Circle, California, USA). Mutagenic primers used were 5'-cacaccaacttt A gcttccagcctg-3' and 5'-gctgctctgttgtctgc A tcatcgagatcatcgagg-3' for the G589S and V546I mutations, respectively. The primer 5'-ccctcgctcg T gtgatctggt-3', which mutates the Xho I restriction enzyme site in the pBSK-SP6-globin vector, was used as the selection primer in the mutagenesis reactions. The mutagenic clones were sequenced to confirm the presence of the mutations and to exclude undesired errors during mutagenesis. Healthy stage V and VI Xenopus oocytes were injected with mRNAs encoding the β human (h)ENaC wild-type or βG589S hENaC mutant, the γhENaC wt or γV546I hENaC mutant together with the mRNA encoding the αhENaC wt. The total amount of mRNA encoding the three αβγ ENaC subunits was 10 ng. Electrophysiological measurements were taken at 16–24 hours after injection. ENaC activity was assessed by measurement of the amiloride-sensitive current (I Na in μA) recorded at -100 mV with a two-electrode voltage clamp amplifier (TEV-200, Dagan Corp.) in a standard solution containing 110 mmol/L NaCl, 1.8 mmol/L CaCl 2 , 10 mmol/L HEPES-NaOH, pH 7.35. The amiloride concentration used was 5 μmol/l in the bath solution. Four batches of oocytes were obtained from different Xenopus frogs in which 5 to 7 oocytes were tested for each αβγ ENaC wt and ENaC variants. Statistical analysis The renin and aldosterone values were nonnormally distributed, as analyzed using skewness, kurtosis and Kolmogorov-Smirnov tests. Therefore, nonparametric tests (Mann-Whitney's U) were used in the statistical analyses, and interquartile (IQ) range and median are used to describe the distributions of target variables. When covariates were included in the analyses, ANCOVA with ranks or logarithm-transformed values of the variables was used. Chi square test, or Fisher's exact test if observed frequency in any cell was less than five, were used for the frequency analysis of the variants. Logistic regression was used to obtain age and gender adjusted odds ratios for hypertension in ENaC variant carriers vs. non-carriers. All data was analyzed using statistical SPSS program (version 11.0). Because of relatively small variant group sizes, the primary analyses were performed with all variant groups combined. Secondarily, the variant groups were also compared separately with the wild-type ENaC group. Results Identification of three common ENaC variants and screening for their presence in the three different study groups DNA samples of 27 hypertensive patients with lowest renin activities and aldosterone concentrations were initially selected for targeted search for ENaC variants. The sequencing strategy chosen permits detection of mutations and polymorphisms in the entire coding parts of exons 13 of β and γENaC genes, as well as 26 or 43 nucleotides at the 3'-ends of introns 12. Three different single-nucleotide substitutions were detected, two in the βENaC and one in the γENaC subunit (Fig. 2 ). Four out of the 27 samples showed a previously unreported substitution of T for C in intron 12 of the βENaC gene (i12-17CT), located 17 nucleotides upstream of the 5'-end of exon 13. In one DNA sample a single G to A substitution changed the codon 589 of βENaC from GGC to AGC, predicted to result in a substitution of serine for glycine (G589S). This variant has been described previously [ 24 , 25 ]. Upon screening of exon 13 of the γENaC gene for mutations, one sample was detected with a novel point mutation changing codon 546 from GTC to ATC, which results in a substitution of isoleucine for valine (V546I). We next conducted a search for these three ENaC variants in our whole material of patients with essential hypertension (n = 347), normotensive males (n = 175) and randomly chosen blood donors (n = 301) (Table 1 ). Altogether, we identified 46 heterozygous carriers of these variant alleles, but no homozygous or compound heterozygous individuals. Their prevalence was significantly different in the three study groups (χ 2 = 15.0, p = 0.0006). Subanalysis of the three groups indicated that the variant allele frequency was higher among the hypertensive patients (9.2%) than in normotensive males (2.9%; p = 0.007) or blood donors (3.0%; p = 0.001), while in the latter two groups it was similar (Table 1 ). When frequencies of the individual gene variants in the hypertensive patients were compared to those in the two other groups (normotensive males and blood donors) combined, the βENaC i12-17CT variant was found to occur significantly more often among the hypertensive patients than in other groups (p = 0.001) whereas the differences in the prevalences of βENaC G589S (p = 0.15) and γENaC V546I (p = 0.14) did not reach statistical significance. Clinical characteristics of the variant allele carriers and non-carriers Clinical and laboratory data of the hypertensive patients grouped according to their carrier status of the three ENaC variants detected are summarized in Table 2 . There were no significant differences, associated with carrying a variant allele, in the sex, age or BMI of the hypertensive patients, nor their serum creatinine, lipid, potassium or sodium levels. Variant alleles did not seem to associate with cerebrovascular events or diabetes among the hypertensive patients, but small numbers prevent definitive conclusions (Table 2 ). Our original study protocol was not designed to disclose health information of the two reference groups (normotensive males and healthy blood donors), and a similar comparison of variant allele carriers and non-carriers in these groups is therefore not feasible. Relation of the variant ENaC alleles to the activity of the renin-aldosterone system The dynamics of the circulating renin and aldosterone levels in most hypertensive individuals were studied during two challenge tests: during a supine-upright postural test and in response to captopril administration. Baseline plasma renin activity was very similar in the patients with and without variant alleles, whether investigated during the postural test or captopril administration (Table 3 ). Plasma renin levels after attainment of upright posture (p = 0.11) and captopril administration (p = 0.12) were not significantly different among carriers and non-carriers of the ENaC variants (Table 3 , Fig. 3 ). Plasma aldosterone levels did not significantly vary according to the ENaC variant carrier status (Table 3 ). We also analyzed renin responses (stimulated value minus baseline value) in the two tests according to the ENaC variant carrier status. We found some evidence of a blunted renin response to both postural (p = 0.21) and captopril (p = 0.087) challenge tests in carriers of variant alleles compared to non-carriers, but there was wide interindividual variation in the test results (Fig. 4 ). Use of covariates (urinary sodium excretion, age and BMI) did not cause significant changes in the results of these analyses. We next related the activity of circulating renin-aldosterone system to sodium-potassium homeostasis in ENaC variant carriers and non-carriers. Serum sodium and potassium concentrations in these two groups of hypertensive patients were similar (Table 2 ). Urinary sodium excretion rate was not associated with the ENaC polymorphisms studied (data not shown). Urinary potassium excretion rates were not statistically significantly different in patients with (median, 83 mmol/day) and without (median, 79 mmol/day, p = 0.23) ENaC variants (Table 4 ). However, when daily potassium excretion (dU-K, in mmol/day) was related to plasma renin activity (in μg/L/h), as mirrored by the renin levels during the postural challenge test, a significant difference was noticed: the median dU-K/renin ratios in the variant carriers vs. non-carriers were 114 vs. 92 when supine (p = 0.29) and 56 vs. 38 when upright (p = 0.034) (Table 4 ); the corresponding values for the average (mean of supine and upright) dU-K/renin ratios were 74 and 51, respectively (p = 0.048) (Table 4 ). A similar analysis of dU-K/plasma aldosterone ratios demonstrated higher ratios in female variant carriers vs. non-carriers for supine (p = 0.16), upright (p = 0.014) and the average values (p = 0.012), while no significant differences were seen in males. Collectively, these data suggests that hypertensive individuals carrying the ENaC variants tend to excrete increased amounts of potassium in relation to prevailing plasma renin and aldosterone levels. Molecular characterization of the ENaC variants We tested whether the βG589S or γV546I have any functional impact on ENaC expressed in Xenopus oocytes, the most commonly used expression system for ENaC functional studies. When αβγ ENaC subunits were co-expressed to obtain maximal channel activity, neither the βG589S nor γV546I affected ENaC activity as measured by the amiloride-sensitive Na + currents. In other words, these data indicate that the current carried by Na + ions through ENaC channels present at the cell surface is similar for ENaC wild-type and mutant channels, indicating that the βG589S and γV546I mutations have no detectable functional consequences on ENaC activity, at least when expressed in Xenopus oocytes. In order to clarify whether the C-T substitution at position -17 of the intron 12 of the βENaC could affect mRNA splicing, cDNA was synthesized from an RNA fraction prepared from lymphocytes of two hypertensive patients heterozygous for the βENaC i12-17CT mutation and a control individual without this gene variant. Primer pairs for reverse transcription were designed in a way permitting identification of a possible failure to splice intron 12 properly. Regardless of the primer pairs used, similar DNA fragments were generated from the samples of the βENaC i12-17CT carriers and control subject (data not shown). Furthermore, sequence analysis of the amplified DNA fragments revealed the presence of only normally spliced DNA sequence in the βENaC i12-17CT carriers. Furthermore, in silico analysis of the βENaC wild-type and variant DNA sequences suggested no differences in exon splicing. However, this analysis is only of predictive value and does not exclude a splicing defect introduced by the variant nucleotide in renal tissue. Discussion The present study indicates that three common variants of the kidney epithelial sodium channel ENaC occur approximately three times more often in patients with moderate-to-severe essential hypertension compared to normotensive males and
While direct in vitro studies have failed to demonstrate a gain-of-function for these ENaC variants, their association with an increased urinary potassium loss in relation to existing plasma renin activity suggests that in the long run in vivo they may result in sodium retention, suppression of renin and aldosterone levels and hypertension. A large number of common and rare polymorphisms of the α-, β-and γENaC have been described in different populations (reviewed in [ 26 ] and [ 27 ]), but their pathophysiologic role, if any, has remained obscure, at least in the White populations. A systematic search in approximately 500 hypertensive probands, mostly of Caucasian origin, disclosed seven variants of the βENaC and six variants of the γENaC subunit, but no variant, with the possible exception of the βENaC G589S substitution, showed an increased ENaC activity in vitro , nor showed cosegregation with hypertension [ 24 , 28 ]. The G589S variant was also identified in a Swedish hypertensive patient [ 25 ]. Two amino acid variants, βENaC G589S and γENaC V546I, both occurred with a frequency of about 2% in the hypertensive patients but in only 1% of the background population or normotensive males. The G589 is located in the poorly conserved cytoplasmic carboxyterminal portion of βENaC, 27 amino acids upstream of the functionally important PY motif. Persu et al. [ 24 ] identified the same substitution in a hypertensive female with mild hypokalemia and suppressed plasma renin activity. Using measurements of sodium channel activity and amiloride-sensitive sodium flux in Xenopus oocytes, these investigators were able to show a borderline 1.3 to 1.5-fold increase in activity for the G589S variant compared with the wild-type subunit. A similar trend was noticed in our experiments (Fig. 5 ). It remains possible that the functional expression of ENaC in Xenopus oocytes is not sensitive enough to detect subtle increases in ENaC activity, as it could well be the case for βG589S ENaC variant, and only mutations leading to large changes in ENaC activity are liable to be detected. On the other hand, even minute changes in ENaC may result in significant in vivo effects when operating for decades under the influence of unfavorable living habits or variants of other modifier genes promoting salt reabsorption. Accordingly, the βENaC G589S could confer some susceptibility to low-renin hypertension, but more data on untreated patients and families are needed. The γENaC V546I substitution is located in the second transmembrane domain of the ENaC subunit, and it has not been described previously. Seven out of the eight carriers were females, and their renin and aldosterone levels were very similar to those in non-carriers. When expressed in vitro in Xenopus oocytes, this substitution did not result in an increase in sodium current (Fig. 5 ). It is not possible at present to deduce whether the V546I variant constitutes a pathophysiologically significant allele by itself or merely a genetic marker conferring susceptibility to hypertension. The C→T variant of the nucleotide -17 of intron 12 of βENaC is a novel one and, interestingly, it was present in 4.6% of the hypertensive patients but in only 1% of the 301 random blood donors (p = 0.009) and 175 normotensive males (p = 0.043). Patients with this variant allele displayed the lowest plasma renin levels and responses of all the subgroups examined (Table 3 , Fig. 4 ), but due to large interindividual variation the differences were not statistically significant. This βENaC variant may have remained undetected in earlier studies as they have mostly employed 5'-PCR primers annealing at the region containing this substitution. Theoretically, a mutation at this site of an intron could affect RNA splicing. We explored this possibility by reverse transcription-PCR experiments of RNA samples from two variant carriers and a control individual, prepared from peripheral lymphocytes known to express βENaC [ 29 ]. We could not demonstrate a splicing error, but since homozygous individuals were not available for studies, we may have missed subtle changes. Furthermore, it is not known how well βENaC mRNA splicing in lymphocytes reflects the mechanism in kidney epithelial cells. Another possibility is that the DNA region around the variant nucleotide -17 of intron 12 contains interaction site for regulatory factors affecting transcription of βENaC in tubular cells, or the i12-17CT variant may be in linkage disequilibrium with some yet unidentified mutation present elsewhere in the βENaC or in the closely linked γENaC gene. The fact that we did not find hypokalemia or statistically significant suppression of renin levels in our patients with variant ENaC alleles does not abandon the hypothesis that they act as subtle genes conferring liability to sodium retention and hypertension during lifetime. In fact, even in cases with unequivocal Liddle's syndrome due to activating ENaC mutations the penetrance of disease phenotype is variable, with inconstant occurrence of hypertension, hypokalemia and suppressed renin levels from patient to patient [ 13 , 30 - 32 ]. This suggests that Liddle's syndrome may represent an intermediate between single-gene and complex genetic diseases, necessitating the effect of extrinsic factors, such as substantial salt intake or other modifier genes, to complete the spectrum of syndrome manifestations. It is of particular note that molecular variants resulting in increased ENaC activity may occur outside the cytoplasmic PY motif that long was considered as a critical domain to be affected in Liddle patients [ 21 , 33 , 34 ]. Our present data are supported by findings of Rayner et al. [ 18 ] who recently discovered another βENaC variant (R563Q), which is located in the cytoplasmic domain just adjacent of the cell membrane and was found to be strongly associated with low-renin, low-aldosterone hypertension in a South African black population. Unfortunately, functional characterization of the R563Q variant was not carried out. Previously, another βENaC variant (T594M) was identified in the African Americans [ 35 ]. Although initially not linked to elevated blood pressure in the Blacks [ 35 ], subsequent studies in a London black population suggested a positive association with hypertension [ 16 , 36 ]. The T594M substitution was reported to result in an increased responsiveness to a cAMP analog due to loss of protein kinase C inhibition of the ENaC [ 35 , 37 ], but other studies have failed to show increased sodium currents in transfected cells [ 24 ]. An additional βENaC variant (G442V) present almost exclusively in Blacks has also been suggested to be associated with biochemical alterations compatible with increased ENaC activity in vivo [ 17 ]. Our present results and the previous data summarized above suggest that subtle β and γENaC variants do exist in the population that may variably result in elevated ENaC activity, suppression of plasma renin and aldosterone levels, urinary loss of potassium, and elevated blood pressure levels. Individual patients may variably manifest either only one or several of these features, and in some of the variant carriers these parameters may be entirely normal. It will be of interest to test the effectiveness of amiloride in our patients with the βG589S, i12-17CT and γV546I variants as an antihypertensive drug as this specific ENaC antagonist was shown to control blood pressure as well as increase plasma renin, aldosterone and potassium levels in black hypertensive individuals carrying the βT594M allele [ 38 ]. There are certain limitations in our study. First, our hypertensive patients represent a highly selected type of patients, since they were recruited by admittance to a specific center focusing on problems in conventional treatment. The clinical study protocol was initially designed for studies on screening for renovascular hypertension in a population, which explains the use of captopril test in the test panel. Unfortunately, urinary aldosterone levels, integrating aldosterone secretion rate over a longer observation period and serving as a valuable marker of Liddle's syndrome [ 13 , 31 ], in particular when related to urinary potassium excretion levels [ 30 ],were not studied systematically. Our single-point plasma renin and aldosterone measurements may have been liable to incidental variations in their plasma levels, and prevent direct comparison to previous studies relying on urinary aldosterone assays. Second, our normotensive reference population comprised of male patients only. However, we had the advantage of picking up the extreme lowest end, as regards systolic and diastolic blood pressure levels in the absence of any antihypertensive drugs, from a very large material of more than 27000 individuals [ 20 ]. Third, due to ethical limitations of the study design, we did not have access to the clinical data of the subjects in the two reference groups (normotensive males and healthy blood donors); it would have been of interest to review the health data of ENaC variant carriers in these two groups. Fourth, our study had limited statistical power for several of the questions asked, particularly when either genetic variant was analyzed alone in carriers versus non-carriers. Therefore, for several of the questions asked in this study these variants may well have only modest effects, too small to be detected using the parameters of the present study. Fifth, pooling of the three genetic variants may represent an oversimplification, as it is uncertain whether these three variants exert similar effects on the various endpoints studied. Finally, our statistical analyses were not corrected for multiple comparisons and therefore some of the results observed in this study could represent chance findings rather than real phenomena. However, this is probably not the case for the observed increased frequency of ENaC genetic variants in hypertensive patients versus normotensive males and blood donors, because this was the primary hypothesis tested and the p values for these comparisons were 0.007 and 0.001, respectively. Conclusions We have demonstrated that almost 9% of Finnish patients with hypertension admitted to a specialized center carry genetic variants of β and γ subunits of the kidney epithelial sodium channel ENaC, a percentage three times higher than that in the normotensive individuals or random healthy controls. Patients with the variant alleles tended to have suppressed renin levels and renin responsiveness to challenging stimuli, and they showed a significantly increased urinary potassium excretion in relation to their renin levels. It will be important to study whether carriers of ENaC variants respond favorably to ENaC blockers (amiloride and triamterene). Competing interests The authors declare that they have no competing interests. Authors' contributions TH-H collected the clinical material, and participated in the DNA analyses and drafting of the manuscript. KK and TPH designed the study and drafted the manuscript. IT, TT, FF, KH designed the clinical chemical and hormonal assays, and participated in collection of the patient material. HF, HEM and KP participated in DNA analyses and bioinformatics. JV and TK collected the control populations and designed their studies. SS consulted in the statistical analyses and assisted in data handling. IG and LS carried out the electrophysiological studies and participated in drafting of the manuscript. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC547905.xml |
520752 | Multi-line split DNA synthesis: a novel combinatorial method to make high quality peptide libraries | Background We developed a method to make a various high quality random peptide libraries for evolutionary protein engineering based on a combinatorial DNA synthesis. Results A split synthesis in codon units was performed with mixtures of bases optimally designed by using a Genetic Algorithm program. It required only standard DNA synthetic reagents and standard DNA synthesizers in three lines. This multi-line split DNA synthesis (MLSDS) is simply realized by adding a mix-and-split process to normal DNA synthesis protocol. Superiority of MLSDS method over other methods was shown. We demonstrated the synthesis of oligonucleotide libraries with 10 16 diversity, and the construction of a library with random sequence coding 120 amino acids containing few stop codons. Conclusions Owing to the flexibility of the MLSDS method, it will be able to design various "rational" libraries by using bioinformatics databases. | Background The combinatorial synthesis method has been demonstrating its effectiveness in discovering novel functional molecules. Examples of this method in the field of evolutionary protein engineering are selections of a novel functional peptide from a random library on solid support [ 1 ], phage display [ 2 ] or in vitro virus (synonym for RNA-peptide fusion or mRNA-display) [ 3 - 5 ]. The efficiency of the methods depends on the screening technique employed and the library quality. In the display methods, a library of polynucleotide templates must be prepared in order to obtain a random peptide library. A primitive random library of such templates is (NNN) n (N = equimolar mixture of A, T, G and C). This library leads to premature short peptides and a particular bias of the amino acid composition, which makes the effective searchable sequence space biased. A slightly improved library NNK or NNS (K / S = equimolar mixture of T and C / G and C) has been conventionally used. Several methods have been developed for a more improved library. Various "rational" libraries in which the nucleotide mixtures were optimized for a target amino acid composition by using a computer calculation have been developed [ 6 - 8 ]. Removal of stop codons to obtain long ORFs is important for the evolutionary design of a novel protein starting from a random library. Several methods based on random block-ligation were reported [ 9 , 10 ]. Two high quality libraries that lead to the successful evolutionary protein design were as follows: the trinucleotide phosphoramidites (3NPs) method using twenty pre-synthesized trimers of nucleotide phosphoramidites [ 11 - 14 ], and the pre-selecting method using an mRNA display with a C-terminus affinity tag in order to remove stop codons [ 15 ]. We report in this article on a convenient method for the construction of a high quality library based on combinatorial DNA synthesis. This library has few stop codon and has an optimized amino acid composition for various purposes. A random library based on the split synthesis [ 1 ] is made routinely in combinatorial chemistry, but a few methods [ 16 , 17 ] and a few applications [ 18 , 19 ] have been reported for oligonucleotides synthesis. They were used for mutagenesis and the products did not have high quality for evolutionary protein engineering. We applied the split synthesis to oligodeoxyribonucleotide synthesis and developed a new procedure, based on the synthesis of designed codon mixtures using multi-line DNA synthesizers. Our method, Multi-Line Split DNA Synthesis (MLSDS), requires only standard reagents and three or four synthesizers for DNA synthesis. MLSDS can make various "rational" libraries of huge diversity with few stop codons. Results and Discussions Adaptive design to the target amino acid composition Scheme of the MLSDS method is shown in Fig. 1 and Table 1 , and described in detail in Methods section. MLSDS is able to remove not only stop codons but also particular codons. It is able to design the codon composition. We incorporated the effect of the single nucleotide deletion during a general oligonucleotide synthesis [ 20 ] into the design. Designed biased libraries are useful for creating various novel proteins such as a functional peptide without Cys [ 21 ] or an engineered protein without Met [ 22 ]. Unnatural codons and unnatural amino acid [ 23 ] will be also incorporated in desired composition. It will be able to incorporate various results of analysis of bioinformatics databases in order to make an initial library with higher evolvability in experimental protein evolution. The optimum amino acid composition in the library may be different for each target protein. For example, when we want to explore the global protein sequence space exhaustively, the uniform amino acid composition may be the best. When we want to explore only a proven region in the protein sequence space, the use of the average amino acid composition among natural proteins [ 24 ] might be better for many aspects. When we want to design a protein with some specific properties, a library with increased or decreased fraction of specific amino acid should be constructed for each segmental region on the polypeptide chain. Among these wide spectra of requirements, we designed DNA libraries that code peptide libraries having various characteristics and have no stop codons. Examples are: a library with the average amino acid composition of natural proteins [ 24 ], which is named "Natural" library in this article, the uniform amino acid composition; and the uniform composition except [Cys] = 0. A library encoding only four kinds of amino acid (a c-Fos mutant library [ 26 ]) was also designed. Designed molar mixing ratios of A:T:G:C for some of these libraries are shown in Table 1 . Another interesting example was obtained when the target composition was "Uniform except [Met] = 0 and [Term] = 0". The designed molar mixing ratio of A:T:G:C gave the high fitness F value (0.96 on three lines splits) and gave no stop codon even if the effect of a point deletion was included in the GA calculation. A Met-less random library may be the best starting library for global search of the protein sequence space. This speculation is supported by the report [ 22 ] stating that a mutant dihydrofolate reductase generated by the replacement of all Met had much higher enzymatic activity than the wild type. Internal deletion problem in the oligonucleotides synthesis process is important. It destroys the codon-based design, leading to stop-codon generation and undesirable amino acid composition. Our program incorporated deletion effects into the GA calculation and succeeded to minimize the deletion problem. Moreover it was reported that contamination of deletion products could be decreased on a denaturing PAGE for DNA of this length [ 15 ]. We also investigated the practical number of DNA synthesizers. For this purpose, we calculated the final correlation coefficient between the designed and the various target compositions with up to 6-line DNA synthesizers. As shown in Fig. 2 , the final correlation coefficient (= the final fitness) became saturated at about 3- or 4-lines on this program. Our GA program is not the best for obtain best F value but suitable for designing actual synthesizing operations. These results showed MLSDS method gave a high quality library even with three DNA synthesizers. When we took the natural abundance as the target amino acid composition, we got a highest fitness value F = 0.99 (on three lines) in the GA calculations. This is reasonable, because the average amino acid composition among natural proteins highly correlates to the number of synonymous codons in the standard genetic code table [ 25 ]. Synthesis of MLSDS libraries We synthesized a "Natural" library and a "Uniform except [Cys] = 0" library mentioned in the previous section. In Table 2 the compositions of the actually synthesized DNA libraries are listed in comparison with the target compositions. They were high quality libraries ( F = 0.85 and 0.66, respectively) without stop codons in full-length DNAs. The deletion rate was about 0.3% per coupling. For the total DNAs including deletants, F = 0.90 and 0.60, respectively. We also synthesized MLSDS products composed of limited kinds of amino acid. It has been regarded that such a peptide can be synthesized only by 3NP method. A mutant c-Fos library that contained only four kinds of amino acid was synthesized, which was equivalent to a library synthesized by 3NP method [ 26 ]. It was a high quality library ( F = 1.00) (Table 2 ). So far, fifteen libraries with various amino acid compositions were successfully synthesized. In order to make long ORFs, we assembled 8 units of the oligomers. Stem sequences of them did not have any stop codons. A DNA library encoding 120 amino acids plus nine 5'- and 3'-flanking semi-random di-peptides (thus, total 138 amino acids) was constructed (Fig. 3 ). The diversity of the synthesized library is about 10 16 judging from the mass (data of A 260 ) and purity (data of PAGE) of synthesized DNA. With an in vivo selection, there is a diversity limit by the transformation step. But with an in vitro selection, there is no such limitation. Thus exploration of huge sequence space by in vitro virus [ 3 - 5 ] or related techniques [ 28 , 29 ] will become possible, depending on the experimental cost. Comparison of MLSDS with other methods So far, a really random library has been generated by four methods. Other methods do not give a really random library, because they can not provide a library in which all the 20 amino acids are encoded at all sites. A comparison of library quality for three methods is shown in Table 3 . An application of 3NPs method to mutagenesis of antibodies [ 27 ] or coiled-coils [ 30 ] gave good results. Twenty kinds of 3NPs mean one codon per one amino acid, but the codons are degenerate. Thus 3NPs method makes many tRNAs useless. The translation efficiency was calculated based on the codon usage, giving maximum 4-fold decrease in Triticum aestivum . It was reported that the reaction efficiency of 3NPs was far from uniform. The sequence data of synthesized DNA using an equimolar mixture of 19 kinds of 3NPs (without Cys) showed 12-fold (maximum) difference in composition [ 27 ] or more [ 12 ]. The coupling yield was affected by the mixing ratio of 3NPs and by the context, showing 8-fold (maximum) difference for the same 3NP [ 27 ]. Thus it will be difficult to correct reaction efficiencies by adjusting the mixing ratio. The correlation coefficient between the target composition and the actual composition was about 0.4 (for uniform 19 kinds of amino acids) [ 27 ] (Table 3 ). Dimer-phosphoramidites [ 17 ] method is a variation of 3NPs method, using pre-synthesized amidites, and had the same problems. In fact, the bias was observed [ 17 ]. A pre-selecting method using an mRNA display [ 15 ] was fruitful in evolutionary protein design. Novel peptide aptamers were evolved starting from a long ORF random library [ 31 , 32 ]. But this method could not remove all the stop codons. It gave limited library diversity. This method has low flexibility in amino acid composition. For example it is difficult to generate a "Uniform except [Met] = 0" library. The correlation coefficient between the target composition and the actual composition were not so high (Table 3 ). The Y-Ligation Block Shuffling (YLBS) method [ 9 ] has high potentiality in the evolutionary design of peptides. It has problems on deletion and reaction bias of RNA ligase. MLSDS produced libraries with high quality as shown in Table 3 . Above-mentioned problems are not so severe for MLSDS method, because it uses only standard phosphoramidites and is free from any biochemical bias such as in mRNA display and in YLBS. It was reported that the difference in the reaction efficiency of equimolar mixture of four kinds of mono-phosphoramidites was only about 1–5 % [ 33 , 34 ]. MLSDS can create any specific amino acid composition as same as 3NP method, and a MLSDS library is made at lower cost than that made with other methods. Conclusions We applied the split synthesis to oligodeoxyribonucleotide synthesis and developed a new procedure, Multi-Line Split DNA Synthesis (MLSDS), based on the synthesis of designed codon mixtures using three-line DNA synthesizers. MLSDS can make various "rational" libraries of huge diversity with few stop codons by using bioinformatics databases. Combination of an MLSDS library with a screening method for huge diversity will accelerate the protein evolution in vitro . Methods A random MLSDS library was synthesized as follows. A standard DNA synthesis method was used in three lines of DNA synthesizer running in parallel. The randomized regions were combinatorialy synthesized in codon units. Triplet codons were synthesized separately in the three synthesizers as an elongation reaction of oligonucleotides on beads made of controlled pore glass (CPG). CPG beads were mixed together manually, and then splitted again into three reaction tubes manually and the next triplet codons were synthesized (Fig. 1 ). The sequence of a 87 mer library was 5'-GAT GAG GCG AAG ACG N AC TG S ( 123/456/789 ) 15 N AC TG S GAG GCT GGC TGC CAC-3', where N and S denote A/T/G/C and G/C, respectively. The A:T:G:C mixing ratio in each letter of three codon groups 123 , 456 , and 789 was shown in Table 1 . These values were calculated as described below. Both flanking regions contain the recognition sequences of type-IIs restriction enzymes Bbs I and Bbv I, respectively. In order to make longer sequences, we ligated 2 to 8 units of oligomers at the cohesive ends (the underlined sequences shown above) generated by the restriction enzyme treatment. The assembly method was as described in Ref. [ 16 ]. The italicized sequence shown above represents the assembly unit (random region of 45 bp and flanking semi-random linking region of (6+6)/2 bp). The synthesized DNA libraries were amplified by PCR using KOD Dash polymerase (TOYOBO), inserted into pCR2.1TOPO vector (Invitrogen) and cloned, avoiding cloning bias. The clones were sequenced. Computer calculations to determine the optimum molar mixing ratio of four bases in the codon synthesis step were performed by using Mathematica (Wolfram Research). We made a GA program for this purpose. Firstly, the target amino acid composition, p T = ( p T 1 , p T 2 , p T 3 ,..., p T 21 ), was established, where normally p T 21 = 0 for stop codons. Secondly, we calculated an expected amino acid (plus stop codons) composition, p = ( p 1 , p 2 , p 3 ,.., p 21 ), from the molar mixing ratio of the bases, x = ( x 1 , x 2 , x 3 ,....., x 12 L ), where 12 L is equal to 4(number of bases) × 3(number of codon letters) × L (number of synthesizer-line). For example, the mixtures for the first letter and the second letter of the first DNA synthesizer have the molar mixing ratio [A]: [T]: [G]: [C] = x 1 : x 2 : x 3 : x 4 and x 5 : x 6 : x 7 : x 8 , respectively. And for example, when L = 3, the expected alanine composition p 1 is given by: for the full-length sequence without deletion. We solved an integer-programming problem (6-valued 12 L -dimensional optimization problem) having the solution x i as integer (0,1,2,3,4,5). The reason for 6-digits "integer" was to simplify the DNA synthesizer handling and also to simplify the calculation. As the fitness F of x in the GA, we took a correlation coefficient between the expected (or designed) amino acid composition and the target amino acid composition: where N = 21 for our normal case. The optimum x , which gave the maximum fitness F , was calculated using a simple GA program. It was reported the deletion rate during a general oligonucleotide synthesis is about 0.5% per coupling [ 20 ], and our data (about 0.3% per coupling) were compatible with this value. We incorporated the effect of the single nucleotide deletion into the GA calculation. We considered only the affect of a point deletion in a synthesized oligonucleotide because the deletion rate is low enough. When a point deletion occurs in the 5' constant region, all the amino acids in the random region are the frame shifted ones. When the event occurs at the i-th site of the random region, it affects the composition in the all downstream from the i-th site, and so on. We incorporated all these effects into the calculation of the composition. Details are described in Additional file 1 . Authors' contributions IT conceived of this specific study and participated in its design and coordination. SS carried out GA calculation. SU carried out the sequence analysis and the making of the assembled longer sequence. YH conceived of general background and mathematical detail. Abbreviations MLSDS, multi-line split DNA synthesis; ORF, open reading frame; CPG, controlled pore glass; PAGE, polyacrylamide gel electrophoresis; GA, genetic algorithm; 3NPs, trinucleotide phosphoramidites. Supplementary Material Additional File 1 In the additional WORD file (MLSDS22AdditionalFile.doc), the detail of calculation method is described for the the expected amino acids composition p = ( p Ala , p Arg , ....., p term ) considering the single nucleotide deletion. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC520752.xml |
553964 | GANN: Genetic algorithm neural networks for the detection of conserved combinations of features in DNA | Background The multitude of motif detection algorithms developed to date have largely focused on the detection of patterns in primary sequence. Since sequence-dependent DNA structure and flexibility may also play a role in protein-DNA interactions, the simultaneous exploration of sequence- and structure-based hypotheses about the composition of binding sites and the ordering of features in a regulatory region should be considered as well. The consideration of structural features requires the development of new detection tools that can deal with data types other than primary sequence. Results GANN (available at ) is a machine learning tool for the detection of conserved features in DNA. The software suite contains programs to extract different regions of genomic DNA from flat files and convert these sequences to indices that reflect sequence and structural composition or the presence of specific protein binding sites. The machine learning component allows the classification of different types of sequences based on subsamples of these indices, and can identify the best combinations of indices and machine learning architecture for sequence discrimination. Another key feature of GANN is the replicated splitting of data into training and test sets, and the implementation of negative controls. In validation experiments, GANN successfully merged important sequence and structural features to yield good predictive models for synthetic and real regulatory regions. Conclusion GANN is a flexible tool that can search through large sets of sequence and structural feature combinations to identify those that best characterize a set of sequences. | Background The minimal requirement for transcriptional activation is recruitment of an RNA polymerase complex to a promoter sequence of DNA upstream of an open reading frame (ORF). Most genes are also potentially under the control of DNA-binding regulatory proteins or transcription factors that can activate or silence transcription. In bacteria, activator and repressor proteins bind to operator sequences that are typically found near the promoter, and promoter specificity is typically conferred through the sigma subunit of RNA polymerase, which binds the promoter directly [ 1 ]. Eukaryotic transcription factors interact with DNA within the promoter, and are responsible for recruitment of the RNA polymerase complex [ 2 ]. Regulatory proteins also bind to conserved sites near the promoter region, as well as to enhancers that can be far (> 10 000 nucleotides) upstream or downstream of the promoter. In all domains of life, transcription factors that bind near the promoter are typically involved in either stabilizing or disrupting the initiation of transcription, while distal enhancer sequences are needed to destabilize the nucleosomes that usually prevent the initiation of transcription in eukaryotes [ 3 ]. The identity and spacing of these protein binding sites are key contributors to the responsiveness of a gene to changing cellular conditions, and sets of genes or operons that are expressed under similar conditions often have similar sets of regulatory elements in their 5' upstream regions. Recent programs such as ClusterBuster [ 4 ] and Promoter2.0 [ 5 ] recognize the need to detect combinations of binding sites in order to characterize promoters or whole regulatory regions. An additional challenge is the possibility of multiple cluster types within a class of genes that respond to the same regulatory stimulus. Artificial neural networks (ANNs) are suited to the task of discovering complex interactions within a set of features, and identifying multiple alternative solutions that yield the same type of response. This type of problem is not linearly separable [ 6 ], and would apply to classification problems such as the regulatory cascade of genes that is induced by lipopolysaccharide in mammals, where the entire set of genes is upregulated in response to a stimulus, but time or subresponse specificity is conferred by one of a set of regulatory modules [ 7 , 8 ]. Another issue in the detection and modelling of regulatory regions is the assumption of additivity in DNA-protein interactions when position-specific scoring matrices (PSSMs) are used to model binding sites. In fact, binding affinity has been shown in some cases to depend on interactions between sites [ 9 - 11 ], which suggests that more-sophisticated modelling schemes may be necessary to build accurate models of binding site affinity. PSSMs are a useful and versatile tool and may be adequate for binding site modelling in many cases (see for instance [ 12 ]), but the additional flexibility of ANNs may be useful in representing non-additive relationships among components of a binding site. The genetic algorithm (GA) is a powerful tool for combinatorial problems of model optimization and feature selection when the 'model space' is complex and has many local optima. Genetic algorithms carry out a number of simultaneous searches in model space, with one or more recombination operators to periodically combine the results of two or more searches, permitting large scale 'jumps' out of locally optimal regions. GAs have been applied to several tasks in computational biology, including sequence alignment (SAGA: [ 13 ]) and phylogenetic inference (MetaPIGA: [ 14 ]). ModuleSearcher [ 15 ] is a recent application of genetic algorithms to the problem of cis -regulatory module detection, with the stochastic GA approach shown to yield similar outcomes to an exact search method in substantially less time. ClusterScan [ 16 ] is another recent approach that uses genetic algorithms to detect optimal combinations of binding sites from the TRANSFAC database [ 17 ]. Another important question in modelling regulatory regions involves the representation of binding sites. While position-specific scoring matrices (PSSMs) are a popular and effective way of representing conserved sites [ 18 ], other strategies such as consensus sequences, sequence composition and structural features [ 19 ] can be considered as well. Structural features should be of particular interest, since DNA deformability appears to play a role in at least some regulatory interactions such as the binding of Escherichia coli integration host factor (IHF) to its target sequence [ 20 ], the correct orientation of both halves of bacterial promoter sequences [ 21 ] and the dynamics of histone/DNA interactions [ 22 ]. DNA structural properties have been derived from crystal and nuclear magnetic resonance experiments and from theoretical simulations, and different oligonucleotides have different propensities toward unwinding, wrapping around other molecules, and deformation in response to ligand binding (reviewed in [ 23 ]). While there is still considerable controversy in the structural field about issues such as A-tract curvature [ 24 ], and there have been questions about the role of experimental conditions in determining results [ 25 - 27 ], carefully selected parameters can permit the testing of specific structural hypotheses pertaining to regulatory protein-DNA interactions. We have developed GANN, a software suite that uses machine learning methods to identify combinations of the features listed above that best distinguish between a positive set (containing, for instance, a set of regulatory regions from co-expressed genes) and a negative set. GANN implements all of the binding site representations described above, allowing examination of models of different complexity as warranted by the type of binding site modelled and the amount of training data that is available. Implementation GANN contains a set of programs for sequence extraction, retrieval and grouping of requested patterns, neural network analysis of these patterns and collection of results. Each program in the suite can accept either the output of the previous program, or an appropriate set of data generated from an external method. The components of GANN and the flow of data through the system are shown in Figure 1 . 'Indices' and the core machine learning component are implemented in C++, while the other programs that read, interpret and combine text files are written in Perl. Sequence extraction The first program (GetSeq) reads in either a GenBank file of annotated genome sequence, or raw sequence and a list of open reading frames (ORFs) of interest generated with the NeuroGadgets Inc. Bioinformatics Web Service [ 28 ]. GetSeq identifies and extracts upstream intergenic regions of a specified length and labels them as the positive set, and can also extract negative set sequences from intergenic regions that are not immediately upstream of ORFs, or directly from the protein-coding regions. Generation of sequence and structure indices The Indices program takes as input a set of positive and negative set sequences (such as those generated by GetSeq), and can compute various properties of the sequences. The input sequences can be subdivided into overlapping windows of any size prior to the calculation of index values. The following indices can be calculated: - Oligonucleotide frequencies are computed by counting the number of instances of a given k -mer within a window, then dividing by the length of that window. The program can determine the frequency of all k -mers of a specific length, or can assess any user-specified set of k -mers, which may include IUPAC notation to represent degenerate nucleotides. - User-specified PSSMs can be counted for each sequence window. The user provides a set of scores for each type of nucleotide at each position within a PSSM, and a threshold score. The program will then count and record the number of sequence instances within each window that yield a PSSM score greater than the specified threshold. - Structure and flexibility rules are implemented via a text file, by assigning floating-point values to each k -mer of a given length. The average score for a given sequence window is then computed by adding the scores for each overlapping k -mer within the sequence, and dividing by the total number of k -mers considered. Any numeric encoding of a complete set of k -mers can be specified: features sampled from publications such as [ 29 - 31 ] are available at the GANN website (see below). After the extraction of indices, the Combine program merges the different index files into a single large file that is used as the input for the machine-learning software. Combine also allows the computation of Z-scores, thus representing each index value in terms of the number of standard deviations from the mean, and can identify peak values for a given index across a set of windows. Combine randomly subdivides the positive and negative index sets into training and test sets, and can also generate a negative control by randomly reassigning some positive and negative set members to the opposite category, yielding a disruption of patterns that were previously consistent within a single set. This type of control sets a 'baseline' for classification accuracy that can be compared to real experimental results. Pattern classification The core of GANN is the neural network classification system. The indices generated from 'Indices' and 'Combine' are presented as input to an artificial neural network, which is trained with either backpropagation or a genetic algorithm to maximize the discrimination between the positive and negative sets. Since the number of indices associated with each sequence is potentially very large, the Outer Genetic Algorithm (OGA) presents random subsets from the pool of indices to a series of neural networks. The unit of selection for the OGA is a 'Chromosome' that contains a predetermined number of indices sampled from the larger pool, and a set of parameters that define the architecture and connectivity of the ANN. The constitution of a population of OGA Chromosomes is determined randomly in the first generation, with random sampling of indices from the pool and ANN parameters sampled randomly from within a set of ranges specified by the user. Each OGA Chromosome is used to construct an ANN, which is then trained to yield optimal predictive accuracy on the training set defined by the Combine program above. At the end of training, performance on the test set is evaluated, and the fitness of the OGA Chromosome is equal to its predictive accuracy on the test set samples. The predictive accuracy is defined as follows: Where TP and TN are the number of correctly classified positive and negative test set examples, and || Pos || and || Neg || the size of the positive and negative test sets, respectively. This formula assigns equal weight to the positive and negative sets regardless of their size, so the ANN cannot achieve an artificially high score by predicting every case as a member of the larger (training or test) set. The OGA Chromosomes with highest fitness are then permitted to 'recombine', yielding new subsets of indices that are trained in the same manner, while less successful indices are gradually lost from the population. The classification potential of indices can be evaluated by examining the scores of neural networks that include these indices in their input set, and through a 'population genetics' approach that traces the frequency of indices through several rounds of OGA recombination and selection. Training of the neural networks is performed using either backpropagation of errors [ 32 ] or with an 'inner' genetic algorithm (IGA). An IGA Chromosome consists of a set of floating-point values, each representing a connection weight within the ANN that is being trained. The fitness of each IGA Chromosome within a population is equal to the predictive accuracy of the specified ANN as defined in the equation above, but on the training set. IGA Chromosomes with relatively high fitness are then subjected to stochastic recombination and mutation of parameters to yield a new population of Chromosomes that are used in the next round of training. While gradient-descent training methods for ANNs (such as backpropagation) can easily get trapped in local optima of the solution space, the recombination option of genetic algorithms permits a search to 'jump' through the solution space and escape local optima. The optimisation of network architecture and connection weights is similar to the 'structure evolution' method of [ 33 ], which has been applied to problems of biological pattern detection [ 34 , 35 ], but our method differs in the partitioning of the connection weight and architecture components of the optimisation. There are many variables within GANN whose values can be specified by the user. While most default settings will be adequate in most situations, parameters such as the number of feature combinations generated by the OGA and the number of features in each combination should be chosen carefully. The 'DefineVars' program provides a set of menus that allow the user to set these parameters and write them to a configuration file that is input to GANN. In addition to reporting the scores of trained artificial neural networks, GANN will save information about the topology, connection weights and constituent indices of each neural network instance that achieves a generalization score above a specified minimum threshold. If GANN is invoked with any of these saved neural networks as input in addition to a table of indices, then it will use the input neural network to classify the new table. This process can yield functional predictions for sequences whose true classification is unknown. Results and discussion Two sets of detection experiments, both based on DNA sequences extracted from the Escherichia coli K12 genome, are presented to illustrate the performance of GANN. Both of these experiments included a set of 250 nucleotide sequences, each 100 nucleotides (nt) in length, which were extracted from between convergently transcribed genes in the E. coli genome using the GetSeq program. These sequences were chosen because they are intergenic like upstream regulatory regions and not subject to the evolutionary constraints of protein-coding sequences, but are not expected to contain functional transcriptional regulatory features since they are exclusively 'downstream' of one gene in each direction. In the first experiment, we created an artificial positive set by inserting conserved sequences into a subset of the 250 sequences, with the remainder constituting the negative set. The entire set of 250 sequences was used as the negative set in the second experiment, while the positive set consisted of 212 upstream regulatory regions containing experimentally validated binding sites for the σ 70 protein of E. coli . Several run parameters were consistent across both experiments. Each OGA Chromosome contained a total of 8 indices, to allow the simultaneous representation of several sequence and structure properties. The population size (= number of OGA Chromosomes) was determined by multiplying the total number of indices by 10, then dividing by the number of indices (8) per OGA Chromosome. This formula ensured that indices would be represented 10 times each on average in the initial randomly generated population, and 99.9% of all indices should occur at least 3 times in the population according to the Poisson distribution. Thirty rounds of OGA Chromosome evaluation and selection were performed in each run. While GANN can evolve ANN architecture and learning parameters as well as combinations of indices, we chose reasonable ANN parameters (available at the GANN website) and fixed them for the entire run. The performance of different sets of indices was expressed in terms of the predictive accuracy (= score on the test set of sequences) described above. Differences in predictive accuracy are expected across replicates, because random partitioning of sequences into training and test sets is likely to yield variation in the frequency of some features that do not define the whole set. However, average predictive accuracy can be estimated by taking the mean across replicates. Indices that are retained in every replicate of an experimental run are more likely to reflect true characteristics of the sequences under consideration, though the redundancy of many indices (different window sizes, different percentiles for PSSM scores, and correlated frequencies of some k -mers) may yield multiple alternative solutions that are equally good. A final indicator of index performance is the composition of OGA Chromosomes that yield high predictive accuracy: if a set of features is important for characterization of a set of sequences, then each of those features should be represented by at least one index in the best OGA Chromosomes. Experiment 1 – Synthetic positive set In the first experiment, the positive set was constructed by adding conserved features to 76 of the 250 sequences (~30%) described above. Each member of the positive set was modified by adding a nucleotide decamer with high conformational mobility (CM). Five thousand unique decamers were generated randomly, and each of these was assigned a CM score based on the dinucleotide table in [ 30 ]. Decamers from this set that scored in the top 5% of all CM values were selected at random to be added to members of the positive set. One of two types of conserved binding site was also added to each member of the positive set. The set of experimentally validated binding sites for cAMP receptor protein (CRP) and leucine-responsive regulatory protein (Lrp) were extracted from RegulonDB [ 36 ], and each positive set sequence gained a binding site randomly selected from one list or the other. The construction of synthetic conserved regions is summarized in Figure 2 . 'Type A' positive set sequences consist of a high CM decamer beginning anywhere between positions 10 and 20, and a randomly chosen CRP binding site of length 19 that starts between positions 65 and 70. The order of patterns is inverted in the 'Type B' sequences, with the 12 nt Lrp binding site beginning between positions 25 and 30, and the high CM decamer starting anywhere between positions 75 and 80. A total of 34 Type A and 42 Type B sequences were generated. Once the positive and negative set sequences were obtained and assembled, the Indices program was used to extract several different types of information from them, with varying window sizes depending on the features being examined. The overlap between adjacent windows was chosen to be 50% of the window length, so if a window of size 10 covered sites 1–10 in a sequence, the next window would cover sites 6–15. - PSSMs for Lrp and CRP binding sites were constructed from the set of binding sites in RegulonDB. Frequency matrices for each site were constructed by dividing the number of occurrences of each residue at each site by the total number of sites (72 Lrp, 128 CRP). These frequencies were then divided by the 'background' frequency of each corresponding nucleotide in the set of intergenic sequences. Since the background frequency of each nucleotide was within the range 0.250 ± 0.005, background frequencies of 0.25 were assigned to each nucleotide. The PSSM was then obtained by taking the natural logarithm of each value in the corrected frequency matrix. Threshold values for PSSM predictions were determined by scoring 50 000 random sequences of the appropriate length against the PSSM, and identifying the scores that corresponded to the 99 th , 95 th , 90 th , and 80 th percentiles. The number of sequence matches above each PSSM threshold was computed for windows of size 20 and 40. - The conformational mobility of sequence windows of size 10 and 20 was computed according to the dinucleotide values in [ 30 ]. - Counts of all k -mers of size 1 (mononucleotides) and size 2 (dinucleotides) were computed for windows of size 10 and 20. The computations described above yielded a total of 450 indices: 360 describing k -mer counts, 64 describing the counts of PSSM 'hits' at different thresholds, and 36 describing the conformational mobility of different windows of sequence. These indices were then combined in four different ways to yield separate tests of different subsets. The k -mer frequencies were included in every set as a 'background' measure of predictive power with no explicit hypothesis. Set 1.1 included only the k -mer frequencies, while set 1.2 added the indices of conformational mobility, set 1.3 included the PSSM scores, and set 1.4 included all three types of index. We initially performed runs where index values were not standardized, but found that indices with values that were not close to zero, particularly the CM indices which ranged between 40 and 70, did not perform well and were consistently eliminated from the population of OGA Chromosomes. In response to this, we standardized all indices for the experiments described below and in Experiment 2. Set 1.4, with the full set of 450 indices, was used to define the number of OGA Chromosomes. The formula at the beginning of this section yielded a recommendation of 562.5 OGA Chromosomes per generation, which was rounded up to 600 and applied to all four sets. 'Combine' was used to randomly subdivide the positive and negative set sequences into training and test sets with a ratio of 2:1. This random reassignment was repeated five times, and five corresponding negative control sets were generated as described in Generation of Sequence and Structure Indices above. The mean of the best generalization scores achieved in each replicate over 30 rounds of OGA evaluation and selection is shown in Figure 3 . The 4 groups of negative control runs corresponding to the four data sets all yielded a mean best score between 0.78 and 0.79, and the range of scores in each case did not overlap with the range of the corresponding five experimental replicates. However, the average generalization score of experimental set 1.1 was only 5–6 % higher than the corresponding negative control runs. Set 1.2, which included CM as well as k -mer counts, yielded a mean generalization score of 0.880, a substantial improvement over set 1.1 with no overlap in the range of maximum scores between the two sets. Set 1.3, which considered k -mer counts and PSSM scores at several thresholds, yielded a mean best generalization score of 0.883, which was substantially better than set 1.1 and indistinguishable from set 1.2. Finally, set 1.4 yielded a small improvement over sets 1.2 and 1.3 in generalization score, with a mean of 0.900. These results suggest that the inclusion of PSSMs and flexibility indices yielded a substantial increase in predictive accuracy over the background of k -mer counts, with the combination of the two possibly producing a further slight increase. Figure 4 shows the change in the mean, maximum and minimum generalization scores for the 600 OGA Chromosomes in each of 30 training rounds for set 1.4. The mean over all five replicated runs is shown for both the experimental and negative control runs. There is an upward trend with all six values, which shows that improvements in the mean performance are due to both the creation of new, advantageous combinations of indices by the OGA as evidenced by the increase in the maximum score, and through the elimination of bad indices, shown with the increase in the minimum score. The difference in mean generalization score between the experimental and negative control runs is very low (< 0.025) in the first OGA generation, but increases rapidly to 0.11 – 0.12 within the first ten generations of optimisation. This trend is consistent with the idea that many indices in the experimental runs are not good at distinguishing between the positive and negative sequence sets, and their replacement with more copies of good indices yields better predictive accuracy. However, poor indices are expected to persist to some degree through the population, since they can 'hitchhike' with good indices through many rounds of OGA training and may even increase in frequency if they are associated with an otherwise good combination of indices. The low (< 0.7) predictive accuracy of some experimental OGA Chromosomes in the last round of training may be due to recombination events that merge sets of hitchhiking indices. If sets of indices that yield the best predictive accuracy are preferentially selected for recombination by the OGA, then good indices should increase in frequency with successive rounds of testing and recombination. Only five indices were present in the final population of all five replicated experimental runs of set 1.4: one measuring the count of sites scoring in the 99 th percentile of the CRP matrix between positions 60 and 99, another measuring the same quantity for Lrp sites between positions 20 and 59, an index describing the conformational mobility between sites 80 and 90, and two indices representing the count of GC dinucleotides and C mononucleotides between positions 29 and 39. The first three indices are easy to understand, as they correspond directly to features (binding sites and a high CM region) that were deliberately inserted into the positive sequence set. However, the third index is less clear until the Lrp sites from RegulonDB are examined in detail: these sites are G+C poor in general, and of the 72 × 11 = 792 dinucleotides contained in the full set of Lrp binding sites, only 20 of these are GC steps. If all 16 possible dinucleotides were present with equal frequencies, then the pair would be present 49 or 50 times, so GC is strongly underrepresented in this data set. Thus, it appears that the GC content in this region is included in many OGA Chromosomes because it is another indicator of the presence or absence of the Lrp binding site in this region. Experiment 2 – σ 70 positive set The second experiment tested the ability of GANN to distinguish between the 250 unmodified intergenic sequences described earlier, and a set of sequences containing binding sites that are recognized by the 'housekeeping' σ 70 subunit of RNA polymerase in E. coli . The 212 promoter-containing sequences in the positive set were extracted from a larger set defined in [ 37 ]. Where multiple promoter sequences were identified in a single upstream region, one of these sequences was chosen at random for inclusion in the positive set. All sequences in this experiment were 100 nt in length, and the positive set sequences were aligned at the transcription start sites of the relevant promoters. The -35 box (consensus 'TTGACA') was typically contained between positions 40 and 49 in the sequence, while the Pribnow box (consensus 'TATAAT') was located approximately between positions 65 and 70 in the positive sequences. Indices were constructed in a similar manner as in Experiment 1. Separate PSSMs for the -10 and -35 boxes recognized by σ 70 were constructed from 250 promoter sequences in the data set, again with background frequencies of 25% for each nucleotide. As with CRP and Lrp above, the 99 th , 95 th , 90 th and 80 th percentile scores were generated for the -10 and -35 boxes from a set of random sequences, though only 1000 random sequences were generated for each case. Since the two halves of the σ 70 consensus sequence are only six bases in length, indices based on PSSM matches for a window size of 10 nt as well as 20 nt and 40 nt were calculated. Indices of conformational mobility and k -mer frequency were calculated as in Experiment 1 above. A total of 618 standardized indices were generated in this experiment: 360 describing k -mer counts, 222 describing the counts of matches to the -10 and -35 PSSMs at different thresholds, and 36 describing the conformational mobility of different windows of sequence. Five sets of experiments were performed, with five experimental and five negative control runs in each. All sets (2.1 to 2.5) included the k -mer count indices, and these were the only indices considered in set 2.1. Sets 2.2 and 2.3 also included the -35 and -10 PSSMs respectively, while set 2.4 included both. All 618 indices were considered by set 2.5. Runs were performed as in Experiment 1 above, with the exception of the OGA Chromosome population size. With 618 indices in total, the formula described at the beginning of this section yielded a recommendation of 772.5 OGA Chromosomes per generation, which was rounded up to 800 and applied to all five sets. The mean of the best generalization scores achieved in each replicate over 30 rounds of OGA evaluation and selection is shown in Figure 5 . Remarkably, there was no substantial difference between any of the experimental sets, which yielded mean predictive accuracies between 0.828 for set 2.2 and 0.843 for set 2.5. This range was smaller than that of the predictive accuracies of the five negative control treatments, which ranged from 0.701 for set 2.2 to 0.737 for set 2.4. No clear trend exists for either the experimental or negative control sets, suggesting that the CM and PSSM indices did not yield any improvement in predictive accuracy over the k -mer counts alone, and that the inclusion of the σ 70 PSSMs did not yield a more-precise model of the promoter sequence. While the predictive accuracy did not change across multiple sets of experiments, the type of indices that were selected by the OGA varied from set to set. The most successful index overall described the frequency of the dinucleotide 'TA' between positions 65 and 74, which corresponded to the position of the Pribnow box in the positive set sequences. This index was the only one present in all five replicates of sets 2.1, 2.3 and 2.4 after 30 rounds of OGA training, and was one of only three such indices for set 2.2. The other two indices that were retained by all five replicates of set 2.2 described the frequency of TA between positions 60 and 69, and the frequency of CA between positions 55 and 64. No single index was retained in all five replicate runs of set 2.5. In all sets that included PSSM representations, the appropriate PSSM component(s) were always retained in four out of five replicated runs. While PSSMs with a very high score threshold (99 th percentile) were favoured in Experiment 1, the PSSMs most commonly retained in the σ 70 experiments favoured lower percentile thresholds, with a roughly even distribution of indices representing the 95 th , 90 th and 80 th percentiles. This effect may be due to greater degeneracy of σ 70 sites relative to the CRP and Lrp sites modelled earlier. Finally, four out of five replicate experimental runs of set 2.5 retained a CM index that covered positions 30–50 in the sequence. While no single index was retained in all five replicate runs of set 2.5, the OGA Chromosomes with the highest generalization scores all contained at least one instance of a PSSM for each of the two halves of the σ 70 consensus and a flexibility feature. Thus, more-selective indices such as those based on PSSMs were included in the majority of OGA Chromosomes when they were present in the experimental set, even if the gain in predictive accuracy was marginal. Conclusion We have explored several features of GANN, most notably the ability to build classification rules for positive set members that form natural subsets, and the capacity to search through large sets of DNA sequence and structural indices to find combinations that yield optimal predictive accuracy. Our generalization accuracy of ~84% on the σ 70 promoter set is similar to the sensitivity of 86% and specificity of 85% reported by [ 38 ], though these results may not be directly comparable due to differences in the size of the data set and the definition of 'negative' examples. While our use of PSSMs in these experiments implies acceptance of the statistical mechanical theory of binding sites [ 39 ], GANN could also be used to build models that take into account interactions between individual residues within a binding site. The model thus constructed could then be compared against a traditional PSSM to see if better predictive accuracy is obtained on a test set of sequences. In focusing on the generation and testing of combinations of indices, we have not examined the performance of GANN when ANN architectural parameters are optimised alongside index combinations. One approach that avoids dealing with too many interactions at once can be to first use GANN to screen a large set of indices and generate a smaller list of indices with predictive power, and to perform a subsequent run where this smaller set of indices are examined in combination with variable ANN parameters. In Experiment 1, we found that indices with mean values that are not close to zero should be standardized. A disadvantage of this approach is that standardization of a column of values is entirely sample dependent, since different finite samples from the same population of values will typically have different means and standard deviations, which may limit the accuracy of an ANN trained on one sample that is used to classify other subsamples from the same population. If standardization is to be applied, then there should be sufficient values in each index to yield a stable estimate of the population mean, as indicated with a low standard error. This is of particular concern since biological sequence samples often do not represent a random sample of all possible sequences, leading to biased estimates of sample mean and standard deviation. The two primary goals of GANN are to allow multiple alternative representations of DNA features, and to permit the discovery of important combinations of these features through the hybrid genetic algorithm / neural network approach. A long-term goal is to use GANN to identify important combinations of motifs predicted from programs such as PatSer [ 40 ], which would permit the application of GANN to the task of identifying complex regulatory features in multicellular eukaryotes. The software packages are released under the GNU GPL and have been successfully tested and run on Win32 systems and on several flavours of UNIX. Complete documentation for GANN and the key files used in the experiments described in this manuscript are available at the project Web site. Availability and requirements - Project name: GANN - Project home page: - Operating systems: Win32, UNIX - Programming language: C++, Perl - Other requirements: none - License: GNU GPL - Any restrictions to use by non-academics: none Abbreviations CM – Conformational mobility GANN – Genetic Algorithm Neural Networks OGA – Outer Genetic Algorithm PSSM – Position-specific scoring matrix Authors' contributions RGB was responsible for software design and implementation, and manuscript writing. RLC contributed to the design, and to the implementation of the genetic algorithm classes in C++. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC553964.xml |
534096 | Conservative treatment in patients with an acute lumbosacral radicular syndrome: design of a randomised clinical trial [ISRCTN68857256] | Background The objective is to present the design of randomised clinical trial (RCT) on the effectiveness of physical therapy added to general practitioners management compared to general practitioners management only in patients with an acute lumbosacral radicular syndrome (also called sciatica). Methods/Design Patients in general practice diagnosed with an acute (less than 6 weeks) lumbosacral radicular syndrome and an age above 18 years are eligible for participation. The general practitioners treatment follows their clinical guideline. The physical therapy treatment will consist of patient education and exercise therapy. The primary outcome measure is patients reported global perceived effect. Secondary outcome measures are severity of complaints, functional status, health status, fear of movement, medical consumption, sickness absence, costs and treatment preference. The follow-up is 52 weeks. Discussion Treatment by general practitioners and physical therapists in this study will be transparent and not a complete "black box". The results of this trial will contribute to the decision of the general practitioner regarding referral to physical therapy in patients with an acute lumbosacral radicular syndrome. | Background Why a design article? Publishing the design of the trial has several advantages. It may prevent publication bias [ 1 ]. A study producing positive results seems more likely to be published than a study that reports no or negative results [ 2 , 3 ]. Also, the study can be included in systematic reviews because data can be retrieved from the researcher [ 2 ]. Publishing the design of a study before the results are available provides an opportunity to reflect critically on the design of the study, irrespective of the results. Also, a design article provides detailed information about the intervention within the trial to care givers. The lumbosacral radicular syndrome (LRS) is a complex of symptoms related to the lumbosacral nerve roots. The LRS is a disorder with radiating pain in the leg below the knee in one or more lumbar or sacral dermatomes, and can be accompanied by phenomena associated with nerve root tension or neurological deficits (i.e. sensory deficits in the leg, decreased muscle strength in the leg, decreased reflexes, urinary problems) [ 4 - 7 ]. A prolapsed disc is a frequent cause of LRS, but other causes include spinal or lateral recess stenosis, tumours and radiculitis [ 4 , 5 , 7 , 8 ]. The incidence of LRS in the Netherlands is estimated at 5 per 1000 persons a year [ 8 ]. Most patients seeking medical care in the Netherlands will first visit a general practitioner (GP), who is regarded as the 'gatekeeper' of the health care system. The majority of health problems presented to GPs are treated by the GPs themselves and they are responsible for most referrals to (para) medical specialists. In 1996 the Dutch College of General Practitioners published their clinical guideline for LRS [ 5 ]. There is consensus that treatment of LRS in the first six to eight weeks should be conservative. The exact content of the conservative treatment is yet not clear [ 9 ]. Since the study of Vroomen et al. [ 10 ] bed rest is not regarded a treatment option for LRS anymore. Primarily, treatment consists of adequate pain medication, giving information about the natural course of LRS, which in general is favourable, and stimulating to continue the normal daily activities of the patient. GPs in the Netherlands largely comply with the recommendations stated in the clinical guideline regarding management in patients with LRS [ 11 ]. However, they deviated regarding the referral to physical therapy (PT), almost half of patients with LRS were referred, whereas this was not recommended in the guideline. No specific patients characteristics could be found for the prescription of physical therapy. So, in general practice referral to PT in patients with LRS is common. However, there is a lack of knowledge of the effectiveness of PT in LRS. Therefore, the aim of this article is to present the design of a randomised clinical trial of conservative treatment (general practitioners and physical therapy) in patients with acute LRS. Methods/Design Aim The LRS trial aims to assess the effectiveness of PTs management added to GPs management compared to GPs management only in patients with acute LRS. We will use a multicentre, randomised clinical trial as study design. Figure 1 shows the flow chart of the proposed design of the LRS trial. The procedures and design of this study are approved by the Erasmus Medical Center Ethics Committee. Study population Participating GPs in and around Rotterdam, the Netherlands, will invite patients with suspected acute LRS to participate in the trial. GPs will invite patients from May 2003 till November 2004 if they have radiating (pain) complaints in the leg below the knee; duration of the (pain)complaints is less than 6 weeks, the age is above 18 years and they present one of the following symptoms: more pain on coughing, sneezing or straining, decreased muscle strength in the leg, sensory deficits in the leg, decreased reflex activity in the leg or a positive straight leg raising test. Patients will receive a letter of information about the LRS trial from their GP. Patients' name and telephone number will be faxed to the research institute. Subsequently, a researcher (PL) will screen eligible patients by telephone and make an appointment to check inclusion and exclusion criteria, to complete the informed consent procedure and to perform the baseline measurement. Figure 2 shows the criteria that must be fulfilled to participate in the LRS trial. A research assistant will check these criteria during patients first visit. The informed consent procedure is completed when patients meet the criteria, are willing to participate and give their written consent. Hereafter, the baseline measurement will take place. Randomisation Randomisation will take place after baseline measurement by the research assistant. We use a concealed randomisation procedure using a computer generated randomisation list developed by an independent person. Patients' specific and unique trial number will be typed in a special developed database (i.e. not editable for research assistant and a second randomisation action using the same trial number is not possible) and the random allocation will appear on screen. In order to prevent unequal treatment group sizes, block randomisation will be used with blocks of 10 patients [ 12 ]. This means that after every 10th patient the number of patients allocated to both treatment groups is equal. Towards every randomised patient will be explained that the management of their complaint by his or her GP will be continued. Patients who are allocated to physical therapy will be shown a list of participating physical therapists of which he or she can make a choice. The research assistant makes the first appointment with the physical therapist most easily accessible by the patient. Blinding For obvious reasons GPs and PTs are not blinded for treatment allocation. But they are not involved with treatment effect measurements. The patients cannot be blinded because of the ethical reasons as stated by the Medical Ethical Committee. The researcher is involved in the statistical analysis, but the analysis and interpretation of the findings will be audited and verified by an independent and not involved statistician. In this trial the primary outcome measurement and most of the secondary outcome measurements will be scored by the patients. Studies from Ostelo et al [ 13 ] and Scholten-Peeters et al [ 14 ] mentioned that in this type of study patients are blinded to a certain extent because they are unaware of the exact content of both treatments or may be called naive to the content of the treatment not received. Other more or less similar designed trials from Vroomen et al [ 10 ] and Hofstee et al [ 15 ] reported that it is not possible to blind participating patients for allocated treatment. Therefore, we think it is important to know any treatment preference of the patients at baseline. Supplementary analysis may show to what extent this effects the scores on outcome measurements of the patients. GP intervention All patients will be treated by the GP according to their clinical guideline (see Figure 3 ). GPs will give information and advice about LRS. If necessary they prescribe adequate pain medication. We asked the GPs not to refer patients to paramedical specialists (i.e. manual therapist, physical therapist, exercise therapist, etc.). Referral to PT is based on randomisation and performed by the research assistant. PT intervention PT treatment will imply information and advice about LRS and exercise therapy. Passive modalities such as massage, manipulation techniques or applying applications (e.g. ultra sound or current waves) are not allowed in the PT treatment. This PT treatment protocol was accomplished in a consensus meeting with participating PTs. The PT will report what kind of information/ advice and what type of exercise the patient receives in each session. Both GP and PT intervention will be restricted to a maximum of 9 treatments/ consultations in the first 6 weeks after randomisation. Theoretical background In the Netherlands, PTs are mainly taught the biomechanic model [ 13 ]. This model focuses on somatic issues; it assumes a causal relation between tissue damage and pain. PT could be of additional value in the management of patients with LRS because PTs are 'the experts' in treating musculoskeletal disorders with exercises and advice/ information. The pain reported by a patient is used as guidance to determine the intensity of the exercises and the advice about resuming normal daily activities and work. This study assumes that focussing on (pain) complaints with exercises and advice is the optimal PT treatment in the acute phase (0 to 6 weeks) of LRS. It is possible that patients may suffer from a fear of movement because of pain [ 16 ]. Good advice/ information will reassure these patients and exercises will show them that movement is possible. So, the secondary treatment goal of the PT is to decrease the possibly present fear of movement in these patients. Sample size This trial attempts to enrol 182 patients with LRS, 91 patients in both treatment groups. This sample size is regarded sufficient to detect a difference of 20% (with a α of 0.05 and a power of 80%) in the primary outcome (GPE) between the two treatment groups. A difference of 20% is considered to be clinically relevant [ 17 ]. Measurements Figure 4 shows the outcome measures and the points of time they are collected. At baseline we will collect patients characteristics such as gender, date of birth, height and body weight. In standardised history taking there will be established whether patients are familiar with LRS in the past, report more pain in the leg on coughing/ sneezing or straining, on sitting, standing, walking and lying down, and if patients report a decreased muscle strength and sensory deficits in the leg. The physical examination consists of the straight leg raising test, the crossed straight leg raising test, test of Bragard, finger-floor distance, standing on toes and heels, knee tendon reflex, ankle tendon reflex, strength of m. extensor hallicus longus, sensory tests (touch, sharp and blunt) in the dermatomes L5/ S1 in the feet. Primary outcome measure The primary outcome measure is the Global Perceived Effect (GPE), measured on a 7 points scale ranging from 1 = completely recovered to 7 = vastly worsened. It is regarded a clinical relevant outcome measure and is regarded valid and responsive to measure the patients' perceived benefit [ 18 - 20 ]. Secondary outcome measures Pain severity of the leg and the back will be scored on a 11 points Visual Analogue Scale (VAS) ranging from 0 = no pain to 10 = unbearable pain. Reliability, validity and responsiveness of the VAS have been shown [ 21 - 23 ]. The functional status will be measured with the Roland Morris Disability Questionnaire (RDQ) for sciatica [ 24 ]. The scoring of the RDQ is achieved by counting the number of positive responses: a patient individual score can vary from 0 (no disability) to 24 (severe disability). The RDQ has proved to be a valid instrument and appears to be responsive for clinical relevant changes [ 20 , 25 - 28 ]. Health status will be measured by the 36-item short form (SF-36) [ 29 ] and the Euroqol (EQ-5D) instrument [ 30 , 31 ]. Validity and responsiveness on both SF-36 [ 32 - 34 ] and EQ-5D [ 35 - 37 ] proved to be sufficient. Fear of movement will be measured by the Tampa scale for kinesiophobia (TSK) [ 38 , 39 ]. The TSK consists of 17 items; each rated on a 4-point likert scale. The TSK has been shown to be a valid and responsive instrument [ 40 , 41 ]. Costs will be calculated and include LRS related sickness absence from work, medical consumption (i.e. medication use, additional therapies, visits to health care providers), out-of-pocket expenses and paid help. Patients' treatment preference will be evaluated at baseline and at 4 follow-up measurements. Statistical analysis Baseline comparability will be investigated by descriptive statistics to examine whether randomisation was successful. If necessary, adjustments for baseline variables will be performed in the analysis. Group differences and 95% confidence intervals will be calculated for all outcome measures. The statistical analysis will be performed according tot the intention-to-treat principle, analysing the patients in the treatment group to which they were randomly allocated. Between group differences will be calculated using the Student t-test for continuous variables or Chi-Square for dichotomous variables. In addition a per-protocol analysis will be performed, analysing only those patients with no serious protocol deviations. Comparing the results of the intention-to-treat and the per-protocol analysis will indicate if and to what extent protocol deviations might have biased the results. Multivariate regression analysis will be conducted to examine the influence of baseline variables on outcome. Discussion This article introduces a design of a RCT to evaluate the additional effectiveness of PTs management added to GPs management in patients with LRS. The study is designed in a way that GP and PT treatment is transparent (according a guideline and a consensus meeting) and not a complete "black box". The results of this trial will contribute to the decision of the GP regarding referral of patients with LRS to PT. The inclusion of patients will run until the end of the year 2004. The follow-up measurements will be completed in the end of the year 2005. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC534096.xml |
516237 | Regulatory role of E-NTPase/E-NTPDase in Ca2+/Mg2+ transport via gated channel | Background E-NTPase/E-NTPDase is activated by millimolar concentrations of Ca 2+ or Mg 2+ with a pH optimum of 7.5 for the hydrolysis of extracellular NTP and NDP. It has been generally accepted that E-NTPase/E-NTPDase plays regulatory role in purinergic signalling, but other functions may yet be discovered. Results In this article it is proposed on the basis of published data that E-NTPase/E-NTPDase could play a role in the influx and efflux of Ca 2+ and Mg 2+ in vivo. Conclusions Attenuation of extracellular Ca2+ influx by rat cardiac sarcoplasmic anti-E-NTPase antibodies and oligomerization studies on mammalian CD39 conclusively point towards the existence of a new channel in the membrane. Further studies on these properties of the E-NTPase/E-NTPDase may provide detailed mechanisms and identify the potential patho-physiological significance. | Background The mechanism by which [Ca 2+ ] i is increased in excitable cells differs from that obtaining in non-excitable cells. Excitable cells exhibit an action potential, a substantial general depolarization of the plasma membrane, in response to depolarizing stimuli; influx of Ca 2+ occurs via plasma membrane Ca 2+ channels and/or release from sarco (endo) plasmic reticulum via ryanodine-receptor Ca 2+ channels which regulate the excitation – contraction coupling [ 1 , 2 ]. The factors that determine the extent of Ca 2+ entry are (i) magnitude of the membrane potential and ( ii) magnitude of the transmembrane Ca 2+ gradient. These two factors also determine whether Ca 2+ or Mg 2+ enters and the time (probably milliseconds) that elapses between channel opening and termination of Ca 2+ or Mg 2+ transport [ 3 ]. In non-excitable cells, the increase in [Ca 2+ ] i results from influx of Ca 2+ across the plasma membrane and Ca 2+ release from the endoplasmic reticulum. Ca 2+ release from the SER depends on the binding of inositol 1,4,5-triphosphate (InsP 3 ) to its receptor Ca 2+ channels, and also on Ca 2+ binding to ryanodine receptor – Ca 2+ channels. Ca 2+ is removed from the cell by the following means. i: the sarco (endo) plasmic reticular Ca 2+ pump ATPase (SERCA), which transports Ca 2+ from the cytoplasm into the SER lumen (~70% of the activator Ca 2+ ); ii: The plasma membrane Ca 2+ pump ATPase (PMCA), which exports Ca 2+ across the plasma membrane (~1% of the activator Ca 2+ ); iii: Mitochondrial Ca 2+ Uniporters (mCa 2+ uniporters), which transport Ca 2+ into mitochondria (~1% of the activator Ca 2+ ); iv: the Na + /Ca 2+ exchanger (28% of the activator Ca 2+ ). This last transport system is reversible but under normal physiological conditions, in the Ca 2+ extrusion mode, it exhibits a stoichiometry of 3 Na + influx/1 Ca 2+ efflux [ 4 ]. Ca 2+ enters animal cells via (i) voltage-operated Ca 2+ channels (VOCC), ( ii) ligand gated non-specific cation channels (LGCCS), and ( iii) stretch/receptor activated non-specific Ca 2+ channels (RACC) [ 4 , 5 ]. A "receptor operated Ca 2+ channel" (ROCC) is defined as a plasma membrane Ca 2+ channel other than VOCC or RACC. VOCC opening depends on membrane depolarization, whereas RACC opening depends on both direct and indirect activation of membrane bound receptors. In contrast, ROCC opening depends solely on agonist-receptor interaction. It has also been suggested that mobile intracellular messengers such as elevated [Ca2+]i play a role in ROCC opening [ 5 , 6 ]. Different types of ROCC are activated (opened) by diverse cell signaling mechanisms such as ligand specificity, increase in [Ca 2+ ] I , increase in [cAMP] i [ 7 ] and activation/inactivation of specific trimeric G proteins [ 8 ]. Opening of Ca 2+ channels must be a highly regulated event involving physical movement of channel components inclusive of the alteration in channel protein conformation; Also, an extracellular source of free energy (ΔG) could be of critical importance. This might be supplied by E-NTPase/E-NTPDase mediated hydrolysis of NTP/NDP. Co-ordination of this process might play a role in the opening of Ca 2+ channels, independently of membrane depolarization or other factors. The biochemical, structural, and functional properties of E-type nucleotidases have been covered in several excellent reviews: i . Extracellular metabolism [ 9 ]; ii . purine signalling [ 10 , 11 ]; iii . adhesion [ 12 ]; iv . transporter functions [ 13 ]; v . pathophysiology [ 14 , 15 ]. Rationale for the proposed hypothesis: E-NTPase/E-NTPDase mediated Ca 2+ /Mg 2+ transport It has been suggested that Ca 2+ entry during the slow inward current in normal myocardium involves membrane-bound channels potentially controlled and/or regulated by metabolic energy transfer from unknown sources, though Ca 2+ enters the cell down its concentration gradient [ 16 ]. Electrical stimulation and membrane phosphorylation by cAMP-dependent protein kinase have been shown to increase E-NTPase/E-NTPDase activity. Metal ions such as Mn 2+ , Co 2+ , Ni 2+ and La 2+ that attenuate Ca 2+ influx also inhibit the E-NTPase. In the late stages of heart failure the E-NTPase is down regulated. Activation of E-NTPase by various concentrations of Ca 2+ has been shown to correlate linearly with cardiac contractile force development [ 17 ]. "Calcium paradox" is defined as irreversible functional and structural protein loss in the isolated heart that is first perfused with Ca 2+ -free buffer and then reperfused with Ca 2+ -containing buffer [ 18 ]. E-NTPase activity is highest during the initial phases of reperfusion, which might favour the initial Ca 2+ influx that causes Ca 2+ overload. During the later stages of reperfusion with Ca 2+ -containing buffer there is a loss of E-NTPase activity. During mild stages of Ca 2+ paradox, E-NTPase retains its function and continues to favour Ca 2+ influx, resulting in the development of intracellular Ca 2+ overloads. However, during severe stages of calcium paradox, impaired E-NTPase activity may contribute to irreversible failure of contractile force recovery [ 19 ]. To date there is no report describing the detailed mechanism of E-NTPase/E-NTPDase-mediated channel gating and its role in Ca 2+ /Mg 2+ transport. In this article an attempt is made to delineate the molecular mechanism of Ca 2+ /Mg 2+ transport, identifying the source of energy and the activation and termination of the process. The central issues are: a . How the metabolic energy from nucleotide hydrolysis is effectively utilized in channel opening; b . What stage of the opening/closing cycle requires energy; c . By what (probable) mechanism the proposed scheme is completed; d . How, if at all, homeostasis is affected The current hypothetical proposal is set out in three sections with appropriate illustrations. Phase I: Activation identifies the evidence that leads to the current proposal and describes how the metabolic energy from nucleotide triphosphate hydrolysis is utilised to assemble a functional homo-oligomer of the E-NTPase/E-NTPDase, forming a channel that is subsequently opened. Phase II: Suggested: Ca 2+ /Mg 2+ Transport Describes, with supporting evidence, how the energy released from [NTP] o/ [NDP] o hydrolysis might be utilized for opening the channel formed by the homo-oligomeric ENTPase/E-NTPDase. Phase III: Termination of the transport processes outlines the intracellular and extracellular factors that would influence the termination of the Ca 2+ /Mg 2+ transport processes, and the experimental evidence obtained in favor of the whole proposal. Phase I: Activation of E-NTPase/E-NTPDase and channel formation Membrane depolarization could locally alter protein conformation. This in turn could potentially induce post-translational modification in the (intracellular) monomer subunits of the E-NTPase/E-NTPDase, followed by translocation to the membrane (depending on the tissue type(s) and functional requirement(s)) (Fig. 1 ). Fig. 2 shows the proposed functional state of the E-NTPase/E-NTPDase after oligomerization and assembly in the membrane to form a gated Ca 2+ /Mg 2+ channel. Fig. 3 , indicates that the oligomerized E-NTPase/E-NTPDase is likely to possess sensors to control the opening and closing of the Ca 2+ /Mg 2+ channel gate. Fig. 4 , represents an interior view of the E-NTPase/E-NTPDase in the functional state after oligomerization and assembly in the membrane. Figure 1 Phase I: Activation. Based on direct experimental evidence, suppose that in response to electrical stimuli, an increased phosphatidylinositol turnover leads to elevated intracellular phospholipid. This in turn could induce post-translational modification of the monomer subunits of E-NTPase/E-NTPDase in the intracellular milieu. Subsequently, the monomers are translocated to the membrane, depending on the tissue type(s) and functional requirement(s). Figure 2 Phase I: Activation. Proposed model for E-NTPase/E-NTPDase in a functional state after oligomerization and assembly in the membrane, functioning as a gated channel. Figure 3 Phase I: Activation. The oligomerized E-NTPase/E-NTPDase would probably possess hypothetical sensors acting to open/close the gates. Figure 4 Phase I: Activation. Interior view of E-NTPase/E-NTPDase in a functional state in the membrane. Probable energy sources and other significant factors are as follows. The source of extracellular nucleotides could be spontaneous release from dead cells or exocytosis from live/damaged cells [ 20 ]. In ocular ciliary epithelial cells, ATP is released in hypotonic conditions, and this release is inhibited by NPPB (5-nitro-2-(3-phenyl propylamine benzoic acid), a potent inhibitor of CFTR (cystic fibrosis transmembrane receptor) and p-glycoprotein mediated ATP release [ 21 ]. On the other hand, the endogenous CD39 of oocytes transforms under hypertonic conditions to a conformation mediating ATP transport to the extracellular environment, either by exocytosis or by acting as an ion channel [ 22 , 23 ]. However, under what conditions (hyper-or hypotonic) might CD39 assume an extracellular nucleotide hydrolyzing activity; and under those conditions, can this property be coupled to ion influx? This question remains unanswered. At normal physiological temperature in presence of divalent succinyl CoA, Con A mediates the oligomerization of E-NTPase monomers/dimers to form a holoenzyme with enhanced activity. Eosin iodoacetamide (EIAA), a fluorescein iodoacetamide that forms thioester bonds with cysteine at neutral pH, enhances chicken gizzard ecto-ATPase activity [ 24 ]. There are ten conserved cysteine residues in E-NTPase (with additional cysteine residues in the N-terminal region that are known to mediate disulfide bond formation, essential in oligomerization). CD39, an ecto-Ca 2+ /Mg 2+ apyrase that hydrolyses ATP and ADP [ 25 ], forms tetramers and might act as a bivalent cation channel. However, the precise mechanism and functional properties are not known at present. CD39 expression is associated with ATP release; it was speculated that ATP release (along with drugs) into the extracellular milieu is followed by the hydrolysis of the extracellular nucleotides by CD39 [ 26 ]. Furthermore, native CD39 (ecto-ATP/Dase/ apyrase) forms tetramers upon oligomerization. Loss of either of the two transmembrane domains of rat CD39 ecto-ATP/Dase impairs enzyme activity. It has been suggested that the functional (holoenzyme) E-NTPase/E-NTPDase is a homotrimer in mammals. Differences in enzyme activity among different species have been attributed to variations in the interaction among the monomers resulting in homotrimeric holoenzyme formation (66 kDa-ATPase) [ 27 ]. It seems clear that changes in the conformation of the E-NTPase/E-NTPDase could mediate changes in the channel transport function. Phase II: Ca 2+ /Mg 2+ Transport Fig. 5a , illustrates the possible utilization of the energy released from [NTP] o /[NDP] o hydrolysis (-7.3 kcal mol -1 or by formation of AMP, -10.9 kcal/mol -1 ) for opening the channel formed by the homo-oligomeric E-NTPase/E-NTPDase. This channel is postulated to open and close in response to energy availability (Fig. 5b ). Fig. 6A , is an artist's impression of the three-dimensional configuration of the E-NTPase/E-NTPDase in vivo. Ca 2+ might enter the cell and excess Mg 2+ might leave by the influx and efflux mechanisms depicted in Fig 6b . Figure 5 Phase II: Ca 2+ /Mg 2+ Transport. (A) Free energy released from ATP hydrolysis by E-NTPase on the outer membrane surface would yield -7.3 kcal mol -1 or by formation of AMP by E-NTPDase would yield -10.9 kcal mol -1 . (B) The energy is utilized for opening the channel formed by the E-NTPase/E-NTPDase, by altering the conformation of the sensors. This altered conformation has an inherent channel-opening effect; loss of the energy source causes the sensors to revert to the resting state, which corresponds to channel closing. Figure 6 Phase II: Ca 2+ /Mg 2+ Transport. (A) Three-dimensional impression of the E-NTPase/E-NTPDase in vivo. (B) It is possible that Ca 2+ can enter the cell and excess Mg 2+ can leave via the influx/efflux mechanisms depicted in the figure. The opening of the slow inward Ca 2+ current channel in cardiac sarcolemma during the plateau phase of the action potential requires ATP [ 28 ]. Furthermore, protein kinase-A (PKA) dependent phosphorylation appears to mediate the increase in Ca 2+ influx in hormonal modulation of that process [ 29 ]. A similar model has been proposed for sodium channels in nerve membranes, in which a cycle of phosphorylation and dephosphorylation is proposed for opening and closing [ 30 ]. Other corroborating evidence implicating E-NTPase in Ca2+/Mg2+ transport via the gated channel is briefly summarised. Rat cardiac sarcolemmal E-NTPase has considerable sequence homology with the human platelet thrombospondin receptor CD36 [ 31 ]. An antibody directed against the purified E-NTPase blocked the increase in intracellular calcium concentration, implying that the E-NTPase plays an unknown but significant role in the delayed Ca 2+ influx or Mg 2+ efflux during the plateau phase of the action potential (Unpublished observation). Activation of E-NTPase by millimolar concentrations of Ca 2+ and electrical stimulation is linearly related to the contractile force developed in the myocardium [ 32 ]. Gramicidin S inhibits the E-NTPase activity and it attenuates the slow channel efflux in perfused frog left ventricles. Based on these observations, we propose that E-NTPase might be involved in providing energy for Ca 2+ /Mg 2+ influx-efflux in the cardiac sarcolemma, opening the channel formed by the E-NTPase/E-NTPDase protein by altering the conformation of the sensors. The altered channel sensor conformation opens the channel; loss of the energy source allows the sensors to revert to the resting state, which corresponds to channel closing. There are at least two Mg 2+ transport systems: (a) rapid transport down the concentration gradient and (b) efflux in low Ca 2+ Ringer during ventricular perfusion in vitro. In rat liver mitochondria, 50 nM cAMP or 250 μM ADP induced rapid loss of 6 mmol of Mg 2+ /mg protein coupled with the stimulation of ATP efflux. This effect was specific and was blocked by adenosine nucleotide translocase inhibitors. Evidently cAMP acts as a mobilizer of Mg 2+ in isolated rat liver mitochondria. Adenine nucleotide translocase is the cAMP target [ 33 ]. Myocardial Mg 2+ content is maintained at physiological level by the sarcolemmal transport system, which pumps Mg 2+ across the plasma membrane when the extracellular [Mg 2+ ] o concentration is <1 mM and restores [Mg 2+ ] i when the heart is perfused with Ringer buffer containing 5 × 10 -7 M Mg 2+ . Failure of either of these two transport mechanisms may result in a rise in [Mg 2+ ] i , impairing the contractile machinery of the myocardium [ 34 ]. Gramicidin S inhibits total Mg 2+ efflux in the myocardium, while epinephrine restores Mg 2+ efflux and contractile force development in the frog ventricle perfused with 10 mM Mg 2+ . It should be pointed out that both E-NTPase activity and myocardial contraction and relaxation are inhibited by gramicidin S [ 35 ]. In the light of the evidence surveyed here, there would appear to be a significant functional role for activated E-NTPase in Ca 2+ influx and Mg 2+ efflux (or vice versa) in the myocardium. Phase III: Termination of the transport process Fig. 7 summarizes the possible means by which the transport process is terminated. There are several potential contributing factors that can be grouped into two categories, extracelluar and intracellular. Additional experimental evidence is indicated. Based on the heterologous expression of ecto-apyrase in COS cells in the presence of tunicamycin, glycosylation might be required for homo-oligomerization and nuclotidase activity. Conversely, deglycosylation might impair the E-type nucleotidase activity by weakening the monomer-monomer interaction and altering the tertiary and quaternary structures, result in the loss of holoenzyme. Essentially, glycosylation and deglycosylation of the ecto apyrase (HB6) monomer and the consequences for homodimer formation have been regarded as an on-off switch for ecto nucleotidase activity [ 36 ]. Figure 7 Phase III: Termination of the transport processes. (A) Several factors might contribute to the termination of Ca 2+ /Mg 2+ transport via channel gating by E-NTPase/E-NTPDase: extracelluar and Intracellular. Additional experimental evidence is mentioned. Decreased flow of Ca2+/Mg2+ due to closing of the channel gate. Fig. 8a is a three-dimensional impression of the ecto-ATPase in vivo at the termination of ion transport. Fig. 8b illustrates how biochemical modifications such as deglycosylation of the E-NTPase/E-NTPDase oligomers might cause dissociation of the homo-oligomers to individual monomers This is a potential mechanism for the disassembly of the functional channel and closure of Ca 2+ influx and Mg 2+ efflux. Also, an increase in membrane fluidity induced by cholesterol oxidation might cause defective association or disassociation due to weak interaction among the E-NTPase monomers, whereas increased membrane cholesterol might sustain higher E-NTPase activity. Oligomerization of E-NTPase and associated increase of activity could also be responsible for the rapid termination of the purinergic response mediated by extracellular ATP [ 37 ]. Figure 8 Phase III: Termination of the transport processes. (A) Three-dimensional impression of the E-NTPase/E-NTPDase in vivo when termination of the ion transport function commences. (B) Biochemical modifications of the E-NTPase/E-NTPDase oligomers such as deglycosylation would probably cause instability, leading to dissociation of the homo-oligomers. Disassembly of the functional molecule would ensue, closing the Ca 2+ influx and Mg 2+ efflux processes, as portrayed in the figure. The extracellular nucleotide mediated activation of channel gating could be terminated by ecto (extracellular)-adenylate kinase, which catalyzes trans-phosphorylase activity (ADP+ADP→ ATP+AMP). This enzyme has a higher affinity for extracellular nucleotides than the dephosphorylating enzyme (E-NTPase/E-NTPDase) or ecto-nucleotide pyrophosphatase/phospho-diesterase (ATP→ AMP +ppi) [ 38 ]. As the transport process winds down, ecto-adenylate kinase mediated ATP generation might maintain the extracellular nucleotide level. However, the precise biochemical kinetic process by which this process is completed remains to be elucidated [ 39 ]. Pathophysiological Significance of E-type nucleotidase mediated Ca 2+ /Mg 2+ transport Impairment of E-Type nucleotidases during Ca 2+ paradox in isolated rat heart model warrants investigation of the molecular mechanism(s) involved. Knowledge obtained from these studies will elucidate the observed protective effects of anti-rat cardiac Ca 2+ /Mg 2+ -ecto-ATPase antibodies in ischemia reperfusion induced damage, which is a corollary of organ transplantation. Furthermore, the antiproliferative effect(s) of these antibodies in left anterior descending coronary artery smooth muscle cell(s) emphasize the need to explore more fully the hypothesis proposed in this article. Authors' contributions HMS participated and provided the hypothetical scheme of the gating mechanism with appropriate literature. SK conceived and carried out experimental part of the investigation and formulating the hypotheses. Abbreviations E-NTPase = Ecto or Extracellular Nucleotide triphosphatase; E-NTPDase = Ecto or Extracellular Nucleotide triphosphate diphosphohydrolase; [Ca 2+ ] i = Intracellular Ca 2+ ; [NTP] 0 = Extracellular Nucleotide triphosphate; [NDP] 0 = Extracellular Nucleotide diphosphate; ROCC = Receptor Operated Ca 2+ channel; SER = Sarco (Endo) plasmic reticulum; [cAMP] i = Cytoplasmic or intracellular cAMP; PMCA = Plasma membrane Ca 2+ pump ATPase. VOCC = voltage-operated Ca 2+ channel; LGCCS = Ligand gated non-specific cation channels. RACC = Stretch/Receptor activated non-specific Ca 2+ channels. SUR = Sulfonylurea Receptor Proteins. CD36 = Thrombospondin receptor on platelets. CD39 = Ecto Ca 2+ /Mg 2+ apyrase. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516237.xml |
524508 | Lack of association between vascular dementia and Chlamydia pneumoniae infection: a case-control study | Background Chronic inflammation appears to play a role in the pathogenesis of vascular dementia. Given the association between Chlamydia pneumoniae and stroke, the possibility exists that previous exposure to C. pneumoniae may play a role in vascular dementia. The objective of this study was to determine if there was an association between serological evidence of C. pneumoniae infection or inflammatory markers with vascular dementia. Methods 28 case-patients with vascular dementia at a geriatric clinic and 24 caregiver-controls were tested for C. pneumoniae IgG and IgA antibodies. The association between vascular dementia and C. pneumoniae titres as well as inflammatory markers was estimated by using both conditional logistic regression and stratified logistic regression. Results When matched cases were compared to controls, there was no significant difference in elevated C. pneumoniae specific IgG antibodies (titre ≥ 1:32), odds ratio [OR] 1.3 (95% confidence intervals [CI] 0.3 to 6.0), p = 0.71, or in elevated C. pneumoniae specific IgA antibodies (titre ≥ 1:16), OR 2.0 (95%CI 0.5 to 8.0), p = 0.33 indicative of past or persistent C. pneumoniae infection. Similarly, no difference in high IgG or IgA antibody levels (IgG titre ≥ 1:512 or IgA titre ≥ 1:64) between the two groups, indicative of recent C. pneumoniae infection, was found, OR 0.4 (95%CI 0.1 to 2.1), p = 0.27. For C-reactive protein (CRP), the mean difference between 18 matched pairs (case – control) was – 3.33 mg/L. There was no significant difference between cases and controls when comparing log transformed values, OR 0.03 (95%CI 0.00 to 2.89), p = 0.13 or comparing CRP values above or below the median, OR 0.8 (95%CI 0.2 to 3.4), p = 0.71. For fibrinogen, the mean difference between pairs (case – control) was -0.07 g/L. There was no statistical difference between cases and controls when comparing log transformed values, OR 0.6 (95%CI 0.0 to 31.2), p = 0.79 or between fibrinogen values above and below the median, OR = 0.5 (95%CI 0.1 to 2.0), p = 0.50. Conclusion We found no evidence for a significant association between C. pneumoniae infection, inflammatory markers such as CRP and fibrinogen, and vascular dementia. | Background Vascular dementia is characterized by a loss of cognitive function and social adaptive functions in individuals with cerebrovascular disease [ 1 , 2 ]. Vascular dementia is the second most common cause of dementia and accounts for 10% to 15% of all cases [ 3 ]. The clinical presentation of this illness is variable, depending on the site and extent of the lesion or infarct [ 2 ]. The pathogenesis of vascular dementia has not been well defined [ 1 , 3 ]. Chronic inflammation and cytokine dysregulation may play a role [ 4 ] similar to that seen in Alzheimer's disease [ 5 ]. Recent data from serological and PCR studies support an association between Chlamydia pneumoniae and cerebrovascular disease. C. pneumoniae has been associated with stroke, transient cerebral ischemia, and atherosclerosis in the middle cerebral artery in both prospective and case-control studies [ 6 - 12 ]. Since stroke is an important precursor to vascular dementia, these data raise the possibility that C. pneumoniae infection may also be a risk factor for vascular dementia. To our knowledge, this potential relationship has not previously been assessed. We conducted a pilot case-control study to determine an association between serological evidence of C. pneumoniae infection and vascular dementia. We also sought to determine if the inflammatory markers, C-reactive protein (CRP) and fibrinogen were associated with this illness. Methods Study design Patients with vascular dementia were enrolled from the Geriatric Clinic at Henderson Hospital, an outpatient clinic affiliated with a tertiary hospital in Hamilton, Ontario. The diagnosis of vascular dementia for participants enrolled was determined in accordance with criteria established by the Neuroepidemiology Branch of the National Institute of Neurological Disorders and Stroke and Association Internationale pour la Recherche et l'Enseignement en Neurosciences (NINDS-AIREN) International workshop [ 13 ]. This includes both physical and imaging evidence of strokes, and a temporal relationship between stroke and dementia [ 3 ]. Case-patients meeting any of the following criteria were excluded: 1) cognitive impairment due to acute cerebral trauma, hypoxic cerebral damage post cardiac arrest, vitamin deficiency states, central nervous system infection, cerebral neoplasia, significant endocrine or metabolic disease, mental retardation; 2) stroke within the last 6 weeks; 3) patients known, in the past 3 months, to have taken a 7 day or more course of antibiotics with activity against C. pneumoniae (erythromycin, clarithromycin, azithromycin, levofloxacin, trovafloxacin, doxycycline, or tetracycline). The controls for this study were chosen from a list of all caregivers who attended the geriatric clinic at the time of the study, regardless of the diagnosis of their spouse or family member. For each case, one caregiver matched for age (within five years) and sex was selected. Caregivers were excluded if they had a diagnosis that included any of the following: dementia, stroke, or cognitive impairment as determined by a Standardized Mini-Mental Status Examination score of < 27 [ 14 ]. Enrolment was from July 1999 to October 2001. All eligible cases and controls who attended the clinic during the study period were approached for consent to participate in the study. Demographic data (age, sex), medical history, and smoking history were collected as well as blood samples for C. pneumoniae IgG and IgA antibodies, CRP, and fibrinogen. This study was approved by the research ethics board at McMaster University. Signed consent was obtained for all participants (proxy consent was utilized for participants considered decisionally impaired). Laboratory methods For C. pneumoniae IgG and IgA antibody detection, all sera were titrated at two-fold dilutions from 1:16 to endpoint. Samples were analyzed by microimmunofluorescence (MIF), using a 16 hour incubation of serum and substrate at 4–8°C with the same batch of C. pneumoniae IgG/IgM MIF slides (LabSystems OY, Helsinki) (23). To prevent IgG interference, sera used for IgA detection were first treated with goat anti-human IgG antibodies (GullSorb; Gull Laboratories, Salt Lake City, UT, USA). CRP was measured using a high sensitivity automated rate nephelometric immunoassay (Dade Behring high-sensitivity CRP, BNII Nephelometer System, Marburc, DE). Fibrinogen was assayed using an automated STA fibrinogen assay (von Clauss method) on a Roche/Stago (Diagnostica Stago SA). Analysis The presence of elevated antibody levels, indicative of past or persistent C. pneumoniae infection, was defined as an IgG titre of 1:32 or greater and IgA of 1:16 or greater [ 15 , 16 ]. High antibody titres to C. pneumoniae , suggesting a more recent infection, was defined by IgG titres of 1:512 or greater or IgA titres of 1:64 or greater [ 16 ]. Because a skewed distribution was anticipated and a linear relationship with risk was not expected, CRP and fibrinogen were analyzed in two ways: using a log transformation of the values and dichotomizing at the median. The association between vascular dementia and C. pneumoniae titres and inflammatory markers was estimated using a matched analysis. Conditional logistic regression analyses were performed for antibody levels, dichotomized and log transformed CRP, as well as dichotomized and log transformed fibrinogen. A stratified logistic regression analysis was also conducted for C. pneumoniae titres and inflammatory markers (both log transformed and dichotomized at the median), stratifying by age and sex (the following age strata were used: ≤ 70 years, 71–80 years, 81–90 years). All analyses for C. pneumoniae titres and inflammatory markers were also performed with adjustment for current smoking status. Data analyses were performed with SPSS version 10 or Egret for Windows version 2.0.3. The original protocol involved two concurrent case-control studies: one including 30 vascular dementia patients and 30 controls and the other with 30 Alzheimer disease patients and 30 controls. The analysis was to include the additional 30 Alzheimer disease controls in the vascular dementia analysis (giving a 1:2 case:control ratio). Assuming that one third of controls would have elevated C. pneumoniae titres, for an alpha of 0.05 and 80% power, matching 30 cases to 60 controls would allow for detection of an odds ratio of 3.8 or higher. As the study proceeded, it became apparent that enrolling the Alzheimer's patients was not feasible. We decided then to limit the analysis to a 1:1 case:control ratio focusing on 30 patients with vascular dementia. Results Participants A total of 28 case-patients were enrolled: mean age 76.2 years (minimum to maximum: 56 to 90 years); 18 (64%) were male. Nine of the 28 cases had at least one comorbidity (including angina, coronary heart disease, vascular disease, liver disease and renal disease); 1 case had 3 or more comorbidities. Thirteen of the cases were current smokers. Twenty of the 28 cases could be matched to caregiver-controls, for a total of 20 case-control pairs. Of these 20 caregiver-controls, 16 were unrelated to a case, and four were spouses of a case. However, none of these four were matched to their spouse. Four additional caregiver-controls were selected for the unmatched analyses, so that data on a total of 24 caregiver-controls was obtained. Where there was incomplete data on antibody or inflammatory marker levels, those pairs were excluded from matched analyses. Incomplete information on cases and controls occurred when individuals consented to participate in the study and provided medical information but did not attend the out-patient clinic for the required blood collection. All individuals with complete data were included in the unmatched analyses. C. pneumoniae serology Univariate analysis of C. pneumoniae specific IgG antibodies showed no statistically significant difference in elevated antibody levels between matched case-patients and controls, odds ratio [OR] = 1.3 (95% confidence intervals [CI] 0.3 to 6.0), p = 0.71. When pairs were broken and stratified analyses were performed the difference was not statistically significant, OR = 1.8 (95%CI 0.4 to 8.3), p = 0.46. The analysis was also performed with adjustment for participants' current smoking status, OR = 1.8 (95%CI 0.2 to 12.2), p = 0.57 (see Table 1 ). Table 1 Comparison of analyses used to assess for associations of C. pneumoniae specific serology and vascular dementia Variable Conditional Stratified* Adjusted** OR p n 1 OR p n 2 OR p n 2 IgG response 1.5 0.66 15 1.8 0.46 48 1.7 0.50 48 IgA response 1.7 0.48 15 2.8 0.13 48 3.0 0.11 48 High titre response 0.4 0.27 15 0.6 0.40 48 0.5 0.34 48 1 Number of pairs included in the analysis 2 Number of individuals included in the analysis * Stratified on age (≤ 70 years, 71–80 years, 81–90 years) and gender **Stratified analysis with adjustment for current smoking status Similarly, no statistically significant difference between elevated C. pneumoniae specific IgA antibodies was found between matched pairs, OR = 2.0 (95%CI 0.5 to 8.0), p = 0.33. The stratified analysis produced slightly higher odds ratio estimates, although not statistically significant. For the unadjusted analysis, OR = 2.8 (95%CI 0.7 to 10.4), p = 0.13 and for the adjusted analysis, OR = 2.7 (95%CI 0.5 to 14.2), p = 0.24 (see Table 1 ). There was also no statistical difference in high antibody levels between matched pairs, OR = 0.4 (95%CI 0.1 to 2.1), p = 0.27 or in the stratified analysis, OR = 0.6 (95%CI 0.2 to 2.0), p = 0.40 for the unadjusted analysis and OR = 0.5 (95%CI 0.1 to 2.5), p = 0.41 for the adjusted analysis (see Table 1 ). Inflammatory markers A matched analysis of CRP performed on 18 of the 20 matched pairs (all pairs with complete data) revealed no significant difference between log-transformed values, OR of 0.03 (95%CI 0.00 to 2.89), p = 0.13. Similarly, there was no difference comparing matched cases and controls with CRP values above or below the median, OR = 0.8 (95%CI 0.2 to 3.4), p = 0.71. In the stratified analysis (Table 2 ), the log transformed CRP variable and the CRP variable dichotomized at the median were not statistically significant in both the unadjusted (OR = 0.5 (95%CI 0.1 to 3.6), p = 0.50 and OR = 2.2 (95%CI 0.7 to 7.2), p = 0.20, respectively) and the adjusted analysis (OR = 0.4 (95%CI 0.04 to 4.7), p = 0.49 and OR = 1.4 (95%CI 0.3 to 6.3), p = 0.64, respectively). Table 2 Comparison of analyses used to assess for associations of inflammatory markers and vascular dementia Variable Conditional Stratified* Adjusted** OR p n 1 OR p n 2 OR p n 2 LogCRP 0.0 0.13 18 0.5 0.50 49 0.5 0.52 49 CRP 0.8 0.71 18 2.2 0.20 49 2.2 0.19 49 LogFibrinogen 0.6 0.79 18 0.8 0.92 49 0.6 0.83 49 Fibrinogen 0.5 0.33 18 0.6 0.38 49 0.5 0.32 49 1 Number of pairs included in the analysis 2 Number of individuals included in the analysis * Stratified on age (≤ 70 years, 71–80 years, 81–90 years) and gender **Stratified analysis with adjustment for current smoking status A matched analysis of fibrinogen performed on the same 18 pairs revealed no significant difference between the log-transformed values of the two groups, OR = 0.6 (95%CI 0.0 to 31.2), p = 0.79. Similarly, there was no difference when comparing the pairs on fibrinogen values above and below the median, OR = 0.5 (95%CI 0.1 to 2.0), p = 0.33. In the stratified analysis (Table 2 ), the log transformed fibrinogen variable and the fibrinogen variable dichotomized at the median were not statistically significant in both the unadjusted (OR = 0.8 (95%CI 0.0 to 71.6), p = 0.92 and OR = 0.6 (95%CI 0.2 to 2.0), p = 0.38, respectively) and the adjusted analysis (OR = 0.1 (95%CI 0.0 to 73.9), p = 0.43 and OR = 0.3 (95%CI 0.1 to 1.6), p = 0.17, respectively). Discussion In this case-control study, we found no significant association between elevated or high C. pneumoniae specific IgG or IgA antibodies and vascular dementia. To our knowledge, this is the first epidemiologic study to test for an association between vascular dementia and infection with C. pneumoniae . We conducted this study on the basis of evidence linking C. pneumoniae to cardiovascular disease and stroke. There is an extensive literature supporting an association between C. pneumoniae and atherosclerosis [ 17 - 19 ]. Although the majority of these studies initially focused on coronary heart disease more recent evidence also supports an association with stroke [ 6 - 12 , 20 ]. However, the clinical importance of this association is uncertain. Although no significant associations were noted, the relatively small sample size and the odds ratio estimates for elevated IgA and IgG antibodies do not definitively rule out an association. In fact, we powered this study to detect a minimally important association between antibodies and vascular dementia of 3.8. Given that the odds ratio 95% confidence interval of IgG is from 0.3 to 6.0, and 0.5 to 8.0 for IgA, our data do not rule out clinically important associations. The point estimates for elevated IgA and IgG antibodies (2.0 and 1.3, respectively) are similar to the recent meta-analysis odds ratio estimates for coronary heart disease of 1.25 (95% CI 1.03 to 1.53) and 1.15 (95% CI 0.97 to 1.36), respectively [ 21 , 22 ]. In both cases the odds ratio estimate for IgA titres is slightly higher than IgG titres, but not statistically different. The meaning of this difference is uncertain. Danesh et al [ 21 ] suggest that these differences are likely due to chance, selection biases, or selective emphasis on particular reports. In contrast, other studies have suggested that IgA titres are more strongly associated with disease outcomes because they are a better indicator of chronic C. pneumoniae infection [ 23 , 15 , 10 ]. Vascular dementia is the second most common cause of dementia, second only to Alzheimer's disease. It was previously believed that most cases of dementia were the outcome of one of these two distinct diseases. However, the clear division between them has recently been challenged. It is now widely believed that vascular risk factors are also associated with Alzheimer's disease and Alzheimer's and vascular dementia may share many common clinical and pathological characteristics [ 3 , 24 - 26 ]. A number of studies have examined the association between Alzheimer's disease and C. pneumoniae infection. In 1998 Balin et al [ 27 ] found an extremely high association between the presence of C. pneumoniae in post-mortem brain samples and late-onset Alzheimer's disease. However, more recent studies have not repeated these findings [ 28 - 31 ]. A recent randomized controlled clinical trial [ 16 ], based on the hypothesis that chronic C. pneumoniae infection contributes to Alzheimer's disease, found an improved long-term cognitive state in patients with mild to moderate Alzheimer's disease who had been treated with doxycycline and rifampin. However, the serological data did not suggest that this clinical effect was due to treatment of chronic C. pneumoniae infection. One study has looked for C. pneumoniae in brain samples of vascular dementia patients. This study, like the later AD studies, did not identify C. pneumoniae in any of the brain samples [ 32 ]. These results suggest that the presence of C. pneumoniae in the brains is not strongly associated with late-onset Alzheimer's disease or vascular dementia. Inflammatory responses are also known to be associated with cardiovascular disease and have recently been implicated in dementia [ 33 ]. Elevated levels of serum C-reactive protein (CRP), a non-specific marker of inflammation, predict cardiovascular disease [ 34 ] and dementia [ 33 ], and have been associated with stroke patients [ 35 ]. Recently, an association between inflammatory markers alpha 1-antichymotrypsin, interleukin 6, and, to a lesser extent, C-reactive protein were associated with an increased risk of dementia [ 36 ]. In this study we did not find a significant difference in CRP levels between the cases and controls. This most likely was due to the limited power in the study and the limitations of measuring serum CRP. Although CRP was originally thought to be produced almost exclusively by hepatocytes, CRP is now known to be synthesized in brain cells and upregulated in Alzheimer tissue [ 37 , 38 ]. Consequently, localized increases in CRP may be associated with vascular dementia but not detected with serum measurements. We found no significant association between increased fibrinogen levels and vascular dementia. Abnormalities of haemostasis are thought to be important in the pathogenesis of cardiovascular disease, ischaemic stroke, and vascular dementia. Within the pathways of coagulation and fibrinolysis, fibrinogen represents an important marker. Elevated levels of fibrinogen are associated with increased risks of cardiovascular disease and ischaemic stroke [ 39 , 40 ] but the results are less conclusive for vascular dementia [ 41 , 42 ]. Lowe and Haverkate [ 43 ] believe that because vascular dementia is only one phenotype of the systemic atherothrombosis disease, associations between haemostatic variables and any given phenotype should be interpreted with caution. To show a specific association with a single phenotype, a study would need an extremely large sample size to overcome the overlap in phenotypes and risk factors seen in atherothrombosis. We acknowledge several limitations of this study. Because of the relatively small sample size, the analyses were adjusted for only a small number of potentially important covariates and the analysis of CRP and fibrinogen was restricted to above and below the median (while quartiles would have been more sensitive). To adjust for variables that were not used as matching criteria and to maximize the data collected a stratified analysis was also done. The additional stratified analysis adjusted for current smoking status. We adjusted for smoking status because there is a known strong association between smoking and C. pneumoniae titres; and between smoking and vascular dementia [ 44 ]. However, it may also be important to adjust for additional factors that may affect inflammatory markers. We also acknowledge that C. pneumoniae serology is an imperfect test of C. pneumoniae exposure and chronic infection. First, the high prevalence of C. pneumoniae exposure makes it difficult to detect true serological differences between cases and controls. Second, it is unclear what the appropriate serological cut-offs should be for identifying exposure versus chronic infection or recent infections. As a result, different groups have used different criteria making comparisons across studies more difficult. However, the importance of this inconsistency is unclear. In the meta-analysis reported by Danesh et al [ 21 ] no significant heterogeneity was found among the studies even though four different cut-off titres were used to determine seropositivity in the microimmunofluorescence assays. An alternative test, that may prove to be more reliable, involves the detection of C. pneumoniae DNA in peripheral blood mononuclear cells [ 45 ]. Another potential limitation is the choice of controls; because C. pneumoniae is infectious an increased exposure in the caregivers could potentially mask a statistically significant association between the patients and controls. There is also evidence that caregivers, because of stress, may have altered immune systems [ 46 ] which could interfere with their generation of antibodies and inflammatory markers [ 47 - 49 ]. Conclusions In summary, a case-control study of vascular dementia patients suggests that there is no significant association between C. pneumoniae antibodies and vascular dementia. We found no evidence for a significant association between systemic inflammatory markers and vascular dementia. While this study can rule out a strong association, larger studies are necessary to determine if a weak association exists. Competing interests The authors declare that they have no competing interests. Authors' contributions ML, MS, WM, CG, JM, and MC conceived and designed the original study. SCC conducted the analysis of data and drafted the manuscript. SS and TS coordinated the study and collected data. JG conducted the serological testing. All authors offered critical input into the manuscript and all have read and approved the final version. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC524508.xml |
532395 | Norovirus Cultured for the First Time | null | Norwalk virus and related noroviruses cause a short but unpleasant illness known variously as stomach flu, viral gastroenteritis, and winter-vomiting disease. They are the major causative agent of epidemic gastroenteritis worldwide. Norwalk virus was first described in 1968 when teachers and pupils at a school in Norwalk, Ohio, succumbed to acute gastroenteritis. The causative agent was identified in 1972, and since then, many other Norwalk-like viruses have been recognized. Noroviruses are usually picked up from contaminated food or water but can also be spread by person-to-person contact. There are no cures for norovirus infections and no vaccines. Worse still, noroviruses survive freezing, heating to 60 °C, and the amounts of chlorine added to public water supplies. Little surprise, then, that around 23 million Americans get a norovirus infection annually. To reduce this disease burden, better prevention and control strategies for noroviruses are urgently needed. However, the development of such strategies has been hampered by the inability of scientists to find a way to grow noroviruses in cultured cells. Like all viruses, noroviruses replicate inside host cells, and they are choosy about which cells they will grow in. The discovery by Herbert Virgin and colleagues that a mouse norovirus can be grown inside dendritic cells and macrophages, two types of immune system cells, opens the door to a better understanding of norovirus biology and better disease control strategies. The first norovirus cell culture system In 2003, Virgin's team discovered MNV-1, a mouse norovirus, and reported that the innate arm of the immune system (as opposed to the adaptive arm) was important in combating MNV-1 infection. The adaptive immune system involves cells that respond to a disease-causing bacterium or virus by making “adaptations” to their genomes that result in specific anti-pathogen responses. The innate immune system, on the other hand, contains cells that recognize general features of pathogens and rapidly attack them when they enter our bodies. It is our first line of defense against bacteria and viruses, and Virgin and coworkers found that mice with defective innate immune systems were particularly sensitive to MNV-1. Now, by examining tissues taken from mice infected with MNV-1 infection, the researchers show that in live animals the virus infects macrophages (cells that engulf and kill pathogens) and dendritic cells (cells that display pathogen proteins to the adaptive immune system). This observation provided the clue needed to grow a norovirus in cultured cells for the first time: when the researchers took uninfected dendritic cells or macrophages out of animals with defective innate immune systems and grew them in the laboratory, they found that MNV-1 could replicate within these cells. The researchers then used physical and biochemical techniques to verify that what they were growing in culture was indeed MNV-1 and also determined the cellular factors that control MNV-1 growth in culture, thereby confirming that the innate immune system is important for combating norovirus infection. Analysis of the sequence of a virus attenuated by growth in vitro identified an important part of the viral capsid that plays a role in MNV-1 virulence, potentially opening up an avenue to vaccination with attenuated viruses. The researchers speculate that dendritic cells in the gut may provide the route by which noroviruses gain access to cells deep within the lining of our guts and thus cause disease. The next step will be to see whether human noroviruses can also grow in macrophages and dendritic cells. If they do, researchers should at last be able to elucidate the lifecycle of human noroviruses and identify the cellular factors needed for their replication. Hopefully, this new knowledge will swiftly suggest ways to prevent and control the human diseases caused by noroviruses. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC532395.xml |
554762 | A cluster of genes located in 1p36 are down-regulated in neuroblastomas with poor prognosis, but not due to CpG island methylation | Background A common feature of neuroblastoma tumours are partial deletions of the short arm of chromosome 1 (1p-deletions). This is indicative of a neuroblastoma tumour suppressor gene being located in the region. Several groups including our have been studying candidate neuroblastoma genes in the region, but no gene/genes have yet been found that fulfil the criteria for being a neuroblastoma tumour suppressor. Since frequent mutations have not been detected, we have now analyzed the expression and promoter CpG island methylation status of the genes UBE4B , KIF1B , PGD , APITD1 , DFFA and PEX14 in the 1p36.22 region in order to find an explanation for a possible down-regulation of this region. Results The current study shows that gene transcripts in high stage neuroblastoma tumours are significantly down-regulated compared to those in low stage tumours in the 1p36.22 region. CpG island methylation does not seem to be the mechanism of down-regulation for most of the genes tested, since no methylation was detected in the fragments analyzed. One exception is the CpG island of APITD1 . Methylation of this gene is also seen in blood from control individuals and is therefore not believed to participate in tumour development. Conclusion The genes UBE4B , KIF1B , PGD , APITD1 , DFFA and PEX14 are down-regulated in high stage NB tumours, a feature that can not be explained by CpG island methylation. | Background Neuroblastoma (NB) is the most common paediatric solid tumour, responsible for 15% of cancer deaths of childhood. It is a tumour of the postganglionic sympathetic nervous system that develops from immature or dedifferentiated neural crest-derived cells [ 1 ]. The distal part of chromosome 1p shows loss of heterozygosity (LOH) in 20–40% of NB tumours and has therefore been alleged to contain one or more tumour suppressor genes. We and others have previously analyzed the chromosomal region 1p36.2-3 [ 2 - 12 ] and we have recently focused on the gene region involving the genes: UBE4B-KIF1B-PGD-APITD1-CORT-DFFA-PEX14. These genes have been analyzed for mutations and a few have been found in rare tumours [ 13 - 17 ]. In search of tumour suppressor genes, the focus has in the last years moved towards epigenetics and methylation of promoter regions. Methylation of cytosines in CpG-dinucleotides is a common modification in mammalian genomes. Methylated cytosines are more susceptible to deamination, which have lead to an erosion of the number of CpG-dinucleotides. The vast majority of CpGs resides within repetitive elements and is methylated. There are also stretches of DNA rich in CpG that are gene associated, i.e. CpG islands, which are normally unmethylated [ 18 ]. Methylation is generally associated with repression of transcription. Gene regulation by methylation includes tissue-specific regulation during development and processes as X-chromosome inactivation and genomic imprinting, reviewed by Herman and co-workers. [ 19 ]. Cancer is associated with a genome-wide hypomethylation and a more gene-specific hypermethylation. Hypermethylation of CpG islands has been shown to be a common mechanism for the inactivation of tumour suppressor genes and is found in a wide range of tumour types [ 20 - 23 ]. According to the Knudson two-hit hypothesis, two successive mutations are required to inactivate a tumour suppressor gene and turn a normal cell into a malignant one. Inactivation could be due to deletions or mutations. Epigenetic events, such as hypermethylation of promoter-associated CpG islands have come in focus during the last decade as a route to inactivation [ 19 ]. The most common way to analyze methylation status is based on bisulphite modification of DNA [ 24 ]. In this reaction, unmethylated cytosines are deaminated to uracil, while methylated cytosines remain unconverted. The region of interest can be amplified using non-discriminating primers, amplifying both methylated and unmethylated templates in one reaction, or with methylation-specific PCR (MSP) in which methylated and unmethylated templates are amplified in separate reactions [ 25 ]. Some genes have been analyzed in NB tumours focusing on methylation status. For example, CASP8 , on 2q33, was one of the first genes found to be methylated in neuroblastoma, with a frequency of about 40 % in primary NB tumours [ 26 ]. On 3p21.3, RASSF1A and BLU have been shown to be frequently methylated [ 27 - 29 ]. In a study by van Noesel and co-workers, 34 genes in 12 different chromosomal regions were analyzed in neuroblastoma cell lines [ 30 ]. A total of six genes that were methylated in at least three of the 22 cell lines were found. CASP8 , as already known, was one of these genes, also FLIP at 2q33 was methylated as well as four genes in the chromosome region 8p21; DR4 , DR5 , DCR1 and DCR2 . Genes on 1p36 were also included in this study, none of which were found to be methylated. No genes in the NB tumour suppressor candidate region on 1p36.22 were included. Alaminos and co-workers have in a study screened 45 genes in NB cell lines [ 31 ]. 34 genes were found to be methylated in at least one of the ten cell lines. The analysis also demonstrated that the percentage of methylated genes was higher in cell lines with MYCN-amplification than in those without. UBE4B , located in 1p36.22, was included in this study and found to be unmethylated. The genes in the 1p36 consensus deleted region are all, except for CORT , associated with a CpG island in their respective promoter regions. The fact that this region is deleted in a subset of NB tumours and that frequent mutations are not found, makes it a possible candidate region for hypermethylation of genes. In the current study, we wanted to analyze the expression pattern of these genes in the 1p36.22 region using TaqMan technology and the methylation status of promoter regions associated with CpG islands, in a panel of NB cell lines and primary NB tumours. Results Expression analysis GUSB was selected as an endogenous control for real-time PCR quantification, and further used as an internal reference for normalization. Real time-PCR studies of all the genes in the region on cDNA-samples showed a significant reduction of mRNA levels in high stage NB tumours compared to low stage tumour mRNA levels, ranging from 49% for PEX14 to 79% for APITD1 (Table 4 ) [ 16 ]. Table 4 Average relative gene expression (gene/GUSB). Gene Average relative expression in NB cell lines Average relative expression in low stage NB Average relative expression in high stage NB % less expression in high stage NB compared to low stage NB UBE4B 1.1 1.2 0.33 73% KIF1B 0.51 3.6 0.99 72% PGD 0.95 1.1 0.38 65% APITD1 1.3 0.35 0.075 79% DFFA 1.5 1.2 0.51 57% PEX14 1.1 1.9 0.96 49% Data analysed by TaqMan, and grouped in neuroblastoma cell lines, low stage primary tumours and high stage NB primary tumours. The average expression in NB cell lines compared to primary tumours varied between the different genes in the region. Cell lines showed a higher expression of APITD1 and DFFA compared to primary tumours. Lower expression in cell lines was seen in KIF1B and equal levels of cell lines and stage 2 primary tumours in UBE4B , PGD and PEX14 (Table 4 ; Fig. 1 ). The 1p-deleted cell line SK-N-AS generally showed lower expression of the genes in the region compared to the other NB cell lines tested (Fig. 1 ). Figure 1 Expression studies using the TaqMan-technique for detecting mRNA in NB cell lines and primary tumours . The results are grouped into three groups indicated at the top; also the 1p-deletion status for each sample is indicated at the top. The genes tested are indicated to the left. The Y-axis indicates relative expression level compared to the housekeeping gene GUSB (gene-mRNA concentration / GUSB-mRNA concentration). The identity of each sample is indicated at the bottom. Methylation analysis The analyzed fragments of the genes UBE4B , KIF1B , PGD , DFFA and PEX14 were not methylated in the panel of NB primary tumours and cell lines. For APITD1 , the region -393 to -222 relative to initiation codon showed methylation. This region was therefore amplified and cloned with a T/A cloning kit in order to give a value of the percentage of clones being methylated (Fig. 2 ). The region contains 16 CpG sites and these sites show various degree of methylation, in different clones as well as in different samples. Methylation of cytosine in front of guanine is concentrated to the beginning of this "methylated region"; the three first cytosine bases showing the highest degree of methylated clones (Fig 3 ). No methylation was seen up-stream of -393 in the fragments analyzed. At some CpG sites in the region, methylation was never detected. Only one sample had no clones with methylation for any of the sites; the stage 2 tumour 20S9. Blood from control individuals also showed various degree of methylation at these sites. The cell lines did not show methylation down-stream of CpG -212 in the analyzed fragments, clones from tumours were not methylated down-stream of CpG -311. Healthy controls showed a smaller region of methylation, no methylated clones were detected down-stream of CpG -280. Figure 2 Methylation status of APITD1 . Black boxes indicate methylation, white boxes no methylation, position relative to start codon is indicated at the top. Figure 3 Organization of the 5' region of APITD1 . The CpG island, predicted promoter and analyzed region including the region of partial methylation are indicated. Discussion Deletion of parts of chromosome 1p is a common feature of neuroblastoma tumours. We have previously narrowed down the shortest region of overlap of deletions (SRO) to 25 cM in our tumour material [ 6 , 7 ]. By including a germ cell tumour with 1p deletion we could confine the SRO to 5 cM [ 11 ]. This SRO was further confirmed when a cell line with a homozygous deletion of 500 kb was found within this region [ 32 ]. We have previously screened the 500 kb region for mutations and only rare have been detected. The aim of this study was to investigate the expression of the genes in the region and to correlate this with the methylation status. TaqMan expression data showed a clear down-regulation in gene expression of genes in the region tested when comparing high stage NB tumours (stage 3 and 4) and low stage tumours (stage 2). This down-regulation was seen regardless if the high stage tumour was 1p-deleted or not (Fig. 1 ). This finding further supports that this region might be involved in NB development and progression. We can speculate about some central mechanism for the expression of the genes in the entire region that is affected in tumours with poor prognosis, either due to the deletion of one allele and/or by some other mechanism. An attractive explanation for this would be methylation of CpG islands that can lead to inactivation of the respective gene. This does not seem to be the case though, since none of the genes, APITD1 excepted, showed methylation. Furthermore, DNA from healthy controls is also methylated in the same region of the APITD1 gene, making methylation a less likely explanation for the difference in APITD1 gene expression between advanced and favourable primary NB tumours. CpG methylation is not the only possible explanation for the down-regulation in gene expression shown in high stage tumours. Analysis of chromatin modifications could be the next step in a further analysis of this region, in search of a mechanism that could lead to down-regulation of the genes. Deficiencies in transcription factors and upstream elements could also account for the decrease in gene transcripts. Another possibility is that the variations are due to differences in the treatment of the tumours before and after surgery. Radiation and chemotherapy could probably have a vast affect on the expression patterns of a substantial amount of genes in the tumour cell. It is however unlikely that all the genes in the analyzed region would be affected by the treatment alone, rather this might explain the differences in the level of down-regulation between different genes. The difference in expression between NB cell lines and primary NB could be explained by the difference in cell environment between cells in culture and in vivo . In primary tumours, there are always some non-NB cells in the RNA preparations that might affect the gene expression results. On the other hand, genetic events happen to the cells when they get immortalized in cell culture, some genes tend to be down-regulated and others up-regulated. We can also speculate that some of these genes, for example APITD1 , might even be so crucial that only cells with expression of this gene are capable of surviving in culture. Conclusion In conclusion, in our aim to find a mechanism that could inactivate the genes in the candidate gene region we analyzed the expression status and the methylation profile of six genes. The techniques used were TaqMan real-time-RT-PCR technology and bisulphite DNA sequencing. Of the genes analyzed all were down-regulated in high staged NB tumours as compared to low stage tumours. Promoter methylation was not detected in the genes analyzed, except for the CpG island of APITD1. This methylation does not seem to be tumour specific, since methylation was also detected in healthy blood controls. Hence, the six genes UBE4B , KIF1B , PGD , APITD1 , DFFA and PEX14 are down-regulated in high stage NB tumours, a feature that can not be explained by methylation, rather by a mechanism still remaining to be discovered. Methods Cell lines and patients A panel of 10 tumours from primary NBs (4 Stage 2, all from patients with no evidence of decease at last follow-up, and 6 Stage 3 or 4 tumours, all from patients with adverse outcome) and 7 NB cell lines (IMR-32, SK-N-AS, SK-N-BE (2), SK-N-SH, SK-N-DZ, SK-N-F1 and SH-SY-5Y) were analyzed (Table 1 ). Table 1 Clinical data for the primary tumours used in this study. Patients/cell lines NB stage 1p-del Ploidi Outcome 18F8 2A neg NED 20S9 2 neg NED 23S4 2 neg 3n NED 25S9 2 neg NED 4F1 4 neg DOD 10S2A 4 pos DOD 13S0 4 pos DOD 13S1 3 pos DOD 15S3 4 neg/pos DOD 17S2 4 neg DOD Column 3: 1p-del, 1p-deletion; neg, negative; pos, positive; neg/pos; ambiguous results based on microsatellite marker analysis (according to Martinsson et al. [6]) and FISH. Column 5: NED, no evidence of disease; DOD, dead of disease. Expression analysis cDNA preparation Total RNA was extracted from frozen (-70°C) NB tumour tissue using RNeasy RNA extraction kit (Qiagen, Hilden, Germany). 2.4 μg total-RNA of each sample was reversed transcribed to cDNA using Superscript II (Amersham, Buckinghamshire, UK) and random hexamer primers, all according to supplier's protocol. All cDNAs were quality tested by amplification of the housekeeping genes UNPH and GAPDH . Real time PCR -Endogenous control To select the most appropriate endogenous control for the real-time PCR quantification analysis, we tested eight different primary NB samples of different stages for their expression levels of ten commonly used housekeeping genes with TaqMan Human Endogenous Control Plate, (Applied Biosystems, Foster City, CA). Analysis was performed according to supplier's protocol. GUSB (β-glucuronidase) and B2M (β 2 -microglobulin) showed least variations in ΔC T levels, and were expressed at constant levels in all samples regardless of NB-stage. GUSB was selected, and further used as an internal reference for normalization in the real-time PCR quantification analysis (Abel et al., submitted). Real time PCR- TaqMan TaqMan primers and probes were derived from the commercially available "TaqMan ® Assays-on-Demand™ Gene Expression Products" (URL: ). Real-time PCR was performed in 384-well plates using ABI PRISM ® 7900HT Sequence Detection System (Applied Biosystems). Amplification reactions (10 μl) were carried out in duplicate with 0.1 μl template cDNA according to manufacturers protocol (Applied Biosystems). In each assay, a standard curve with six cDNA dilutions was recorded and two non-template controls were included. Quantification was performed by the standard-curve method. The mean C T -value for duplicates were calculated, and the gene concentration (or gene copy numbers) of test samples was interpolated based on standard curves. All samples were normalized by dividing the concentration of the test gene with the concentration of the housekeeping gene β-glucuronidase ( GUSB ) in the same cDNA sample. The logarithms of the expression levels were compared with Student's two-sided t-test on each group of tumours; low stage and high stage tumours. Methylation analysis Bisulphite modification DNA was phenol extracted with the use of phase lock gel (Eppendorf AG, Hamburg, Germany) according to standard procedure and was, with some minor changes, modified according to previously published papers [ 24 , 33 ]. Briefly, 1 μg of genomic DNA was digested with restriction endonucleases that cut close but outside the region of interest. The DNA was then denaturated in 0.3 M freshly prepared NaOH at 40°C for 15 minutes. Sodium metabisulphite (Sigma-Aldrich CO, St Louis, MO) and urea, at a final concentration of 1.73 M and 5.36 M respectively, were added in order to sulphonate the unmethylated cytosines, along with hydroquinone (0.5 mM). Conversion was carried out at 55°C for 16 hours, with a temperature rise to 95°C for 30 seconds every third hour. DNA was purified with Wizard DNA clean up system (Promega Corporation, Madison, WI) according to the manufactures instructions and desulphonated in 0.3 M NaOH at 37°C for 15 minutes. DNA was then precipitated in ethanol, resuspended in distilled H 2 O and stored at -20°C. Promoter analysis and DNA amplification The putative promoter regions of the genes were predicted using Genomatix Promoter Inspector software (URL: ) and CpG islands with MethPrimer software (URL: ; Table 2 ) [ 34 ]. These regions, or parts of them, were amplified with one primer pair, or if needed, with semi-nested primers (Table 3 ). The methylation status was analyzed using bisulphite sequencing. Conditions for PCR amplification were 1× PCR Gold Buffer (Applied Biosystems), 0.5 mM dNTPs, 2.0–3.0 mM MgCl 2 , 0.4 μM of forward and reverse primers respectively and 1 unit of AmpliTaq Gold, in a total volume of 50 μl. Reactions were denatured at 95°C for 10 min followed by 5 cycles of 95°C for 1 min, 49–55°C for 2 min, 72°C for 3 min and 30 cycles of 95°C for 30 sec, 49–55°C for 2 min, 72°C for 1 min 30 sec and ending with 10 min extension at 72°C. The PCR products were immediately sequenced or cloned into a sequencing vector using the TOPO T/A cloning kit (Life Technologies Invitrogen, Carlsbad, CA), where after 10–30 clones were picked. PCR products were purified with ExoSAP-IT™ (USB Corporation, Cleveland, Ohio) and sequencing was carried out using forward or reverse primer with ABI Prism BigDye™ cycle sequencing Ready Reaction Kit (Applied Biosystems). The samples were analyzed in an ABI 3100 or an ABI 3730 Genetic Analyzer. Table 2 Putative promoter regions and CpG island predictions. Gene Promoter region CpG island Amplified region UBE4B -1212 to -492 -1157 to -150 -959 to -494 KIF1B -22392 to -21141 -22444 to -20935 -22517 to -21874 PGD -185 to +7 -680 to +728 -157 to +144 APITD1 a -250 to +30 -389 to +643 -418 to +321 DFFA +65 to +264 -164 to +256 -266 to +142 PEX14 - -2806 to -2362 -2752 to -2481 a For APITD1 , the methylation status was analyzed for three different fragments. Column 2: Putative promoter regions according to Genomatix Promoter Inspector software. No promoter region was detected in PEX14 . Column 3: CpG island prediction by MethPrimer (for CpG island criteria, see materials and methods). Column 4: Region amplified with the primers used in this study. Table 3 PCR primers for amplification; all primers are designed for the sense strand. Gene Primer Sequence Length of fragment (bp) Accession number UBE4B FP 5'-TTGTTAGTTTATTTGGTTTAGGTT-3' 466 NM_006048 RP 5'-TAACAAAACCCAACACTATAAAAAAAACCCCT-3' KIF FP 5'-TTTTTAAGGGTATTTTTTAGAAGGG-3' 644 NM_015074 RP 5'-ACTATAACCAATCACAACACAAAACTC-3' PGD FP-A 5'-GTGAGTTGTTATGGTTATAGTTG-3' 301 NM_002631 FP-B 5'-ATGGTGTGGTTTTATGGTTTTATTT-3' RP 5'-CAAAATCACAAAACCCCAAATAA-3' APITD1 1FP-A 5'-GATTTTGTAAGATATATTTGAGGTAT-3' 231 chr1_29_927.b 1FP-B 5'-ATGGAGTTTTTGATAATGTGTATTG-3' 1RP 5'-AACCCCCTACTCAACTTACTCTAC-3' 2FP-A 5'-ATTAGGTTTTGGGGTGTAGTAGTGAT-3' 199 2FP-B 5'-GTAGAGTAAGTTGAGTAGGGGGTTG-3' 2RP 5'-ACCCTAAACAAAAACAAAAAAAC-3' 3FP-A 5'-GTAGAGTAAGTTGAGTAGGGGGTTG-3' 350 3FP-B 5'-TTGTTTTTGTTTAGGGTCGGTT-3' 3RP 5'-CAAAACCAAAAAATAACCTCTC-3' DFFA FP 5'-AAGTTAAAAATAATTTTTAGGTTGAAT-3' 407 NM_004401 RP 5'-ACCAACCCTTACTCCTCAAATCT-3' PEX14 FP 5'-TGATTAGTTAGGTTTTAGAAAGATGG-3' 333 NM_004565 RP 5'-CAAATAAAACCAAAAATACTAACAAAC-3' Column 2: FP, forward primer; RP, reverse primer; FP-A and FP-B, primers used for semi-nested PCR together with the same reverse primer. Column 5: UCSC Genome Browser August 2001 draft sequence was used as reference sequence (URL: ). Authors' contributions HC participated in the design of the study, carried out the methylation analysis and drafted the manuscript. SF, LH and FL participated in parts of the methylation study. KE and CK carried out the TaqMan runs. KE also contributed to drafting the manuscript. RMS participated in the cloning. TM coordinated the study. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554762.xml |
509303 | Mouse Models of Human Autoimmune Diseases: Essential Tools That Require the Proper Controls | What can we learn about human autoimmune disorders that have a genetic component -- such as systemic lupus erythematosus -- from mouse models? | Autoimmune diseases afflict a large segment of the population in Western countries. Many of them have been described, and rheumatoid arthritis, type I diabetes (also called insulin-dependent diabetes mellitus), multiple sclerosis, and systemic lupus erythematosus (SLE) are among the most common. Although tremendous progress has been made in disease management over the last decade, cures for these diseases have not yet been found. Consequently, a large research effort is sustained in this field. In addition, autoimmunity has intrigued basic immunologists since the early realization that the ability to discriminate self from non-self was at the core of the immune system's ability to protect an organism from pathogens while avoiding self-destruction. A failure of this mechanism results in autoimmune reactions that often lead to clinical disease. In spite of massive research efforts, the mechanisms by which autoimmune diseases develop are not clearly understood. Genetic predisposition as well as environmental triggers plays a role, but the identity of these factors has been largely elusive. The identification of the most common genetic and environmental factors that set off autoimmunity may lead to a better understanding of the ensuing pathogenesis, and offers the best hope for improved therapies, and ultimately, cures. So far, animal models have proved the best way to probe the mechanisms of disease in general, and autoimmune diseases in particular. In the past few decades, the mouse has become the model of choice for experimental medicine, and the rat is following close behind. Starting in the early twentieth century at the Jackson Laboratory (Bar Harbor, Maine, United States), the production of inbred strains of mice and the systematic collection and characterization of naturally occurring mutants have created the building blocks on which much of the research using animal models is now based. Inbred strains are collections of genetically identical animals obtained through selective breeding. These strains have provided homogenous experimental groups, with interindividual variability reduced to environmental (and stochastic) factors. In addition, inbred strains have an assortment of distinct phenotypes that have then been exploited as models of human diseases. Since the 1980s, techniques have been developed to manipulate the mouse genome. Specific genes now can be routinely over-expressed as transgenes, or eliminated by gene targeting, which creates a “knockout” ( Smithies 1993 ; Wassarman and DePamphilis 1993 ). This approach, known as reverse genetics, has shown itself to be a powerful tool with which to evaluate the role of individual genes in various biological processes. The use of genetically engineered mouse models will likely play a major role in deciphering the function of a multitude of new genes revealed in the recently completed sequence of the mouse genome ( Mouse Genome Sequencing Consortium 2002 ). Interestingly, the percentage of mouse genes without any homolog currently detectable in the human genome (and vice versa) has been estimated to be less than 1%, a fact that strengthens the validity of mouse models of human diseases. However, one has to be careful in directly applying data obtained from animal models to human diseases. Most human autoimmune diseases show an extremely heterogeneous clinical presentation, which animal models present as simplified versions. A mouse model, as in any reductionist approach, is both inconvenient, because it provides only a partial representation of the real biological complexity underlying the human disease, and advantageous, because it is a more tractable tool with which to probe mechanistic issues. In addition, a number of differences exist between the human and rodent immune systems ( Mestas and Hughes 2004 ). Since immune dysfunctions are at the root of autoimmune diseases, such differences may limit extrapolations from animal models to autoimmune patients. Nonetheless, animal models are at the core of autoimmune research, and a large body of literature reflects the many advances brought by these models in terms of deciphering disease mechanisms. The relative lack of progress in certain human autoimmune diseases for which an animal model does not exist, such as neuropsychiatric lupus, corroborates the indispensable role played by animal models. Three basic types of animal models have been used in autoimmune research: spontaneous models, induced models, and genetically engineered models. The latter category has been further subdivided into transgenic and knockout strains ( Table 1 ). Table 1 Examples of Common Mouse Models of Autoimmune Diseases Spontaneous models were produced through fortuitous observations of clinical symptoms reminiscent of a given human autoimmune disease developing in a given mouse strain, or in crosses between mouse strains. This happened, for example, with the nonobese diabetic (NOD) mouse, which developed type 1 diabetes, and the hybrid between the New Zealand Black (NZB) and the New Zealand White (NZW) mouse, (NZB × NZW)F1, which developed a lupus-like disease. Unfortunately, spontaneous models are not available for all human autoimmune diseases. Therefore, scientists have created induced models, often by exposing the animals to high doses of a suspected autoantigen at the same time as stimulating the immune system. Interestingly, marked differences exist between strains in their responses to these autoimmune inductions that reflect genetic variation associated with susceptibility to autoimmune diseases. For this reason, induced models have been limited to a small number of strains. For example, experimental autoimmune encephalomyelitis has been widely used as a model of multiple sclerosis. In this model, spinal cord homogenate or a protein derivative such as myelin basic protein is injected with a mixture of potent immunostimulants, most commonly in mice from the SJL strain. Another example of an induced model is collagen-induced arthritis, which has been used to study rheumatoid arthritis. In this model, type II collagen, a joint component, is injected also with immunostimulants, most commonly into mice from the DBA/1 strain. Finally, the ability to turn specific genes on or off, in specific cell types and/or at specific times, has created a plethora of mouse models limited only by the immunologist's imagination. Genes suspected to play a role in the pathogenesis of various autoimmune diseases have been evaluated this way, and those studies have been very informative in mapping out functional pathways that are targeted in these diseases. However, these models have intrinsic problems that have become more apparent in the past few years, and require careful controls to avoid possible misinterpretation. These problems are a result of the fundamental way in which transgenic and knockout strains are produced. When a piece of DNA carrying the gene of interest is injected as a “transgene” into fertilized eggs, it integrates randomly into the genome, and in doing so, potentially modifies the expression of the gene it integrates into. Since all genetic studies have recognized that a large number of genes are involved in autoimmune disease susceptibility, the potential for a transgene to hit one of those susceptibility genes is not negligible. This potentially confounding factor is usually controlled by producing and comparing several independent transgenic lines. More problematic is the interpretation of results obtained with knockout models. A gene is “knocked out” (KO) by homologous recombination of a disrupting piece of DNA within that gene. This genetic manipulation takes place in embryonic stem (ES) cells, which once mutated, are introduced into the inner cavity of a blastocyst (very early mouse embryo), creating chimeric embryos that are put back into female mice that carry the pregnancy to term. After multiple trials (and errors), ES cell lines from the 129 strain and blastocysts from the C57BL/6 (B6) strain have shown themselves to be superior in terms of efficiency and reliability. Consequently, this strain combination is at the origin of the overwhelming majority of knockout strains. Although the chimeric mice are usually backcrossed to B6 to dilute the contribution of the 129 genome, the knockout strains are always a mixture of the two genomes ( Figure 1 ). Most importantly, a large region flanking the KO gene remains of 129 origin, unless extreme measures are taken to select for recombination between tightly linked markers. There have been sporadic reports of phenotypes initially attributed to deficiency in the expression of a given gene that disappeared with additional backcrosses to B6. The only possible interpretation was that these phenotypes were in fact due to 129 alleles that were replaced by B6 alleles with further backcrossing. Figure 1 Breeding Strategy Usually Performed to Transfer a KO Allele from the 129 Genome to the C57BL/6 (B6) Genome For clarity, only 4 of the 20 pairs of the mouse chromosomes are represented by black (B6) or red (129) bars. The KO allele is shown by a white box. Chimeric males are obtained from the integration of ES cells from the 129 strain that have been engineered to carry the KO allele to B6 blastocysts. These males are then bred to normal B6 females, resulting in an N1 progeny that is made up of 50% B6 and 50% 129 genome. N1 mice are subsequently “backcrossed” to B6, and their N2 progeny is selected for the presence of the KO allele. The contribution of 129 genome among the N2 progeny is normally distributed around a mean of 25%. This process can be repeated (shown here to N4), resulting in an average reduction of the 129 genome to one half of what it was in the previous generation. At any point in the process, homozygosity for the KO allele, which is necessary to prevent expression of that gene, can be obtained by intercrossing heterozygous mice, shown here at N4. The production of autoantibodies and mild antibody-related renal pathology, highly relevant to autoimmune diseases, especially SLE, has been reported independently in four mouse models using the (129 × B6) genetic background ( Obara et al. 1979 ; Botto et al 1998 ; Bickerstaff et al. 1999 ; Santiago-Raber et al. 2001 ). It is generally accepted that genetic susceptibility to autoimmune diseases is conferred by multiple highly interactive genes that have small individual effects. In this context, the autoimmune phenotypes resulting from the combination of the 129 and B6 genomes may might therefore provide a primed background upon which the effects of deficiency in the target gene can be amplified. On the other hand, the autoimmune phenotype may be overwhelmingly contributed by the (129 × B6) genomic combination, with little if any effect of the deficiency of the KO gene. In this issue of PLoS Biology , Marina Botto and her colleagues have tested this hypothesis by taking one of their knockout models for SLE, the serum amyloid P component deficient mouse ( Apcs −/− ) ( Bygrave et al. 2004 ). This group has published that Apcs −/− mice on a (129 × B6) genetic background develop a lupus-like disease, even after repeated backcrosses to B6 ( Bickerstaff et al. 1999 ). Serum amyloid P component binds to debris generated from dying cells. Efficient removal of this debris has been shown to be critical to the prevention of the production of autoantibodies against intracellular material. Apcs was therefore an interesting candidate gene to evaluate. Apcs , however, is located in a region near the tip of mouse Chromosome 1 that is rich in SLE susceptibility loci and in genes that have been directly associated with SLE in humans ( Wakeland et al. 1999 ). As mentioned above, the Apcs −/− mouse, although on a mostly B6 genomic background, has its entire Apcs flanking region replaced with 129 alleles. In a critical experiment, Botto and colleagues compared the autoimmune phenotypes of Apcs −/− mice to congenic mice (i.e., genetically identical, except for the gene of interest), carrying the same 129 region on Chromosome 1, but expressing Apcs . Amazingly, no difference was found between the two strains regarding the production of autoantibodies, clearly eliminating Apcs deficiency as a mechanism for this autoimmune process. Apcs deficiency was however associated with markedly increased renal damage, suggesting that this gene may be involved in preventing pathological consequences of autoantibody production. It has been reported anecdotally that the most common outcome of a genetic knockout is a lupus-like disease. Botto and her colleagues may have identified the reason behind this somewhat surprising observation as a spurious consequence of the gene-targeting process. Gene targeting has been an invaluable tool in understanding the mechanisms of immunological diseases, and has still a very important role to play with increasingly sophisticated techniques of selective targeting. The immediate consequence of this work should be an increased scrutiny for appropriate controls, which may include congenic mice carrying the same 129 flanking region, but expressing the targeted gene ( Figure 2 ). Figure 2 Congenic Strains (Left and Right) of a Lupus-Prone Mouse (Middle). Image courtesy of Jessica Merritto The past few years have shown that genetic susceptibility to autoimmune diseases involves a large number of genes with small individual contributions. In spite of this great complexity, advances have been made, and a small but growing number of susceptibility genes have been identified ( Morahan and Morel 2002 ). A common trait shared by successful studies has been the use of mouse models, either directly or indirectly. As the pace of genetic analysis increases in autoimmune diseases, and powerful tools have been created to navigate between the mouse and human genomes, the use of mouse models has been reaffirmed at multiple levels. Mouse models are used to discover new susceptibility genes that can then be assessed in patient populations, as well as to validate genes that have been directly identified in human genetic studies. Mouse models are also used to perform detailed functional and physiological analyses that cannot be conducted in humans. Finally, mouse models have been invaluable to screen disease-specific therapeutic agents. Using mouse models has its pitfalls; many differences, both obvious and subtle, exist between mice and humans. Those differences are, however, outweighed by the power of the experimental system offered by the mouse. What the new study of Botto and colleagues reminds us is that the appropriate control is still crucial to meaningful data interpretation. Keeping that in mind, one can predict that many of the keys to human autoimmune diseases are still in the mouse room. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC509303.xml |
544565 | Organisms isolated from adults with Cystic Fibrosis | Background Patients with cystic fibrosis [CF] have frequent pulmonary exacerbations associated with the isolation of bacterial organisms from sputum samples. It is not clear however, if there are differences in the types of additional organisms isolated from patients who are infected with Burkholderia cepacia complex [BCC] or Pseudomonas aerugionsa [PA] in comparison to those who are not infected with either of these organisms [NI]. Methods Adult patients attending the regional CF unit were followed over a two year period and patients were assigned to three groups depending on whether they were known to be chronically infected with BCC, PA or NI. We compared the numbers and types of organisms which were isolated in each of these groups. Results Information was available on a total of 79 patients; BCC 23, PA 30 and NI 26. Total numbers of organisms isolated, expressed as median and IQR for each group, [P = 0.045] and numbers of co-infecting organisms [P = 0.003] were significantly higher in the BCC group compared to PA, and in the PA group [P < 0.001, p = 0.007 respectively] compared to NI patients. The pattern of co-infecting organisms was similar in all three groups. Conclusions Total numbers of organisms isolated and numbers of co-infecting organisms were significantly higher in the BCC group compared to PA, and in the PA group compared to NI patients. Types of co-infecting organisms are similar in all groups of patients. | Introduction Patients with CF experience frequent exacerbations of their symptoms; contributing factors include infections which may be bacterial or viral in nature [ 1 - 3 ]. Exacerbations which are associated with the identification of an infecting organism are associated with a more rapid decline in lung function, admission to hospital and earlier acquisition of P. aeruginosa [PA] [ 4 - 6 ]. It is also known that those patients who are infected with B. cepacia complex [BCC] have a worse prognosis and often a more rapid decline in lung function with increased mortality [ 7 ]. Previous investigators examining pathogens infecting the CF lung have identified P. aeruginosa , S. aureus , H. influenzae , S. maltophilia , A. xylosoxidans and Aspergillus species as being important pathogens [ 1 , 8 ]. This study examines the bacterial organisms cultured from an adult CF population, subdividing them into three groups; BCC, PA and those who were not infected with either of these organisms [NI] and to correlate with several clinical parameters. We hypothesized that those patients with BCC have more co-infection with other bacteria and this contributes to the greater morbidity in these patients. Materials and Methods Study Design and Eligibility Data were collected on all patients [>18 years old] attending the regional adult CF unit at the Belfast City Hospital for two consecutive years. Sputum specimens were obtained for each patient at regular clinic attendances [patients were reviewed at 3 monthly intervals] as well as twice during hospital admissions. BCC was cultured from sputum employing selective agar (Columbian Agar Base cat no: CM331, Oxoid Ltd., Hampshire + 5% Defibrinated Horse Blood E & O Laboratories, Bonnybridge, Scotland). All isolates were grouped into cultural phenotypes displaying similar visual characteristics and one colony from each phenotype identified by API. Sputum samples were well taken samples following physiotherapy. Sputum culture results were retrieved from the Bacteriology database for all samples submitted over the study time period. Patients were then divided into three groups using the following criteria; BCC if there was ever any sputum culture positive for these organisms. PA, if this bacterium was identified on two or more occasions over a twelve month period, and NI for all other patients who were not chronically infected with either of these organisms. It was noted as to whether the positive culture consisted of one organism or if there were additional organisms identified. In the BCC and PA groups, when more than one organism was identified from a sputum culture the additional organism [s] where recorded as 'co-infecting', as where all organisms in cultures of two or more infecting organisms in the NI group. Co-infecting organisms were divided into groups; P. aeruginosa , S. aureus , H. influenza , S. pneumonia , S. maltophilia and all other organisms including fungi. The facility to detect respiratory viral infection was not available at the time of this study. Spirometry [Vitalograph α ,Buckingham, UK] and oxygen saturation measurements were noted at the start and end of the study as well as the best measurement for each year. Weight was recorded at the beginning and end of the two year period. Other measurements included; smoking history, use of prophylactic and number of courses of intravenous antibiotics and a history of diabetes mellitus. Statistical Analysis Statistical analyses were performed using the SPSS version 11 package. The Chi-square test was used to analyse differences in the sex, smoking status and incidence of diabetes mellitus between groups. The Kruskal-Wallis and Mann Whitney test were used to compare antibiotic use and organisms isolated across groups. Lung Function and Oxygen saturation means were compared by one-way analysis of variance followed by the Student-Newman-Keuls method. A probability (P) value of less than 0.05 was considered statistically significant. Results Information was available on a total of 79 patients, mean [Standard Deviation] age 26 [7.9] years. Numbers in each group were as follows; BCC 23 [2 Burkholderia multivorans , 21 Burkholderia cenocepacia , all ET -12], PA 30 and NI 26. Those patients in the BCC group had a mean duration of infection of 67 months. There was no difference in the sex distribution between the groups, P = 0.18. There were a total of 10 [12.7%] patients with diabetes, but they had no relationship with the groupings [P = 0.19]. There were significantly more smokers in the BCC group [P = 0.005] in comparison to PA and NI. Prophylactic antibiotic use was similar throughout all groups, P = 0.14. Intravenous antibiotic use was similar in the BCC and PA patient groups but significantly higher compared to NI [P < 0.001], table 1 . There was no difference in oral antibiotic use between the groups. Those patients in the BCC group had a significant reduction in their weight compared to the PA and NI groups (P < 0.05); there was an average weight loss of 1.1 Kg. Patients in the PA and NI groups gained weight, 0.69 ± 2.5 and 1.82 ± 2.7 kilograms respectively. Table 1 Use of oral and intravenous antibiotics; median [Interquartile Range]. Antibiotic Courses Patient Group BCC PA NI No. of Patients 23 30 26 Oral 0 [0–2.0] 1 [0–2.0] 1 [0–2.3] Intravenous 3 [1.0–11.0] 3 [2.0–5.0] 0 [0–2.0] The changes between the best FEV 1 measured each year, the change in FEV 1 between the start and end were not significant between the groups. Oxygen saturation measured at the end of the study tended to be lower in the BCC and PA groups when compared to the NI group; however there were no significant differences for the best SaO 2 values for each year. Median and interquartile ranges of organism numbers [total numbers and numbers of co-infecting organisms] and specimen numbers in each group are shown in table 2 along with details of the types of co-infecting organisms. Total numbers of organisms isolated, that is the number of isolates of different organisms, [P = 0.045] and numbers of co-infecting organisms [P = 0.003] were significantly higher in the BCC group compared to PA and in the PA group [P < 0.001, P = 0.007] compared to NI. Total numbers of positive sputum cultures were greater in both the BCC and PA groups in comparison to NI [P < 0.0001]. The pattern of co-infecting organisms, Staphylococcus aureus , Haemophilus influenzae , Streptococcus pneumoniae , Stenotrophomonas maltophilia , was similar in all three groups except for P. aeruginosa being more frequently detected in the BCC group in comparison to NI [p < 0.0001]. Table 2 Median [Interquartile Range] of numbers of sputum specimens, total numbers of organisms and numbers of co-infecting organisms [per patient] in each group. Mean ± SD numbers of individual co-infecting organisms isolated per patient in each group. Measurement Patient Group BCC PA NI No. of Patients 23 30 26 Total No. of Specimens / patient 24 [9.0–59.0] 19.5 [5.5–34.5] 0 [0–9.8] Total No. of Organisms / patient † 32 [16.0–79.0] 24 [7.8–46.3] 0 [0–11.3] No. of Co-infecting Organisms / patient ‡ 15 [8.0–20.0] 3.5 [0.8–11.3] 0 [0–4.3] P. aeruginosa co-infection 14.4 ± 19.7 - 0.1 ± 0.3 S. aureus co-infection 3.6 ± 4.8 2.0 ± 4.1 3.7 ± 7.5 H. influenza co-infection 1.8 ± 2.9 0.7 ± 1.6 1.1 ± 3.1 S. pneumonia co-infection 0.6 ± 1.2 0.2 ± 0.5 0.4 ± 1.0 S. maltophilia co-infection 0.9 ± 1.5 2.6 ± 6.6 1.7 ± 6.3 Other organism co-infection 1.4 ± 1.9 3.1 ± 5.9 1.7 ± 3.2 † Total number of different bacterial organisms isolated in that group of patients during the study period. ‡ Number different bacterial organisms isolated excluding those isolates for BCC or PA in those patients who are colonized with these bacteria. Discussion This study identified individuals with BCC and PA more frequently require courses of IV antibiotic therapy in comparison to CF adults in whom neither pathogen was identified [NI]. This is in keeping with previous findings in which patients colonized with BCC [ 9 , 10 ] and PA [ 11 , 12 ] were noted to have a more rapid decline in health status with an associated increased number of hospital admissions for treatment of infective exacerbations. Prophylactic antibiotic use was found to be similar in all groups, however their role and the optimal duration of treatment of CF patients remains controversial [ 13 ]. The BCC group were also noted to have significant weight loss over the study period, again this has been shown to act as an independent predictor of survival in this group [ 14 ]. No significant differences were demonstrated between weight change in the PA and NI groups; it may be that a longer period of study is required to detect significant differences. On comparison of PA and BCC patients to those in the NI group, investigators have previously demonstrated higher morbidity [ 10 , 12 ]. In this study we have found higher total numbers of positive sputum specimens, total numbers of organisms and increased numbers of co-infecting organisms in the BCC and PA groups in comparison to NI [P < 0.01]. To our knowledge this is the first paper to relate bacterial colonization to the numbers and types of coinfecting organisms isolated during exacerbations. Patients infected with BCC are known to have elevated lung inflammation[ 15 ] with an accelerated rate of lung function decline[ 9 ], this in part may be due to BCC lipopolysaccharides mediated neutrophil recruitment and subsequent respiratory burst response [ 16 ]. This pro-inflammatory status may predispose individuals to increased morbidity and co-infection. BCC infection has also been linked to a compromised host response [ 17 ] and combined with a poor nutritional status [ 14 ] increases the likelihood of respiratory tract infection in these patients. In this study the types of organisms isolated did not differ significantly between the groups, organisms corresponding to those found on previous studies [ 1 ]. It should be borne in mind that these findings relate to the adult CF population in Northern Ireland which is unique in that the majority of B. cepacia complex positive patients are colonized with a single strain (B. cenocepacia ET12). It is well documented that CF patients experience a more rapid decline in their lung function [ 18 ]. However, no significant changes in lung function were seen between the start and end of the study and on comparison between groups. It is possible that this is due to the relatively small group numbers or that a longer follow-up period is required. The prevalence of diabetes mellitus amongst the groups was similar; its presence has recently been associated with a more rapid decline in lung function in CF patients [ 6 ]. Surprisingly, the incidence of smoking was highest amongst those with more severe disease in the BCC group. Investigators have shown evidence of this type of risk taking behaviour in these patients, however it is actually lower when compared than their peer group [ 19 ]. This emphasizes the importance of addressing relevant risk factors and behaviour amongst these patients. Conclusion Patients with BCC and PA infection have a similar use of antibiotics. Total numbers of organisms isolated and numbers of co-infecting organisms are significantly higher in the BCC group compared to PA, and in the PA group compared to NI patients. Types of co-infecting organisms are similar in all groups of patients. Abbreviations BCC Burkholderia cepacia complex CF Cystic fibrosis FEV 1 Forced expiratory volume in one second NI 'Neither Isolated' group No. Number P Probability PA Pseudomonas aeruginosa SAO 2 Oxygen Saturation SD Standard Deviation SE Standard Error SPSS Statistical Package for the Social Sciences SPECIAL CHARACTERS: < Less than α Alpha Competing Interests The authors' declare that they have no competing interests. Authors' Contributions TMcM Collected data relating to the study and drafted the paper. AMcD Participated in the analysis of clinical specimens and contributed to the manuscript content. JEM Participated in the study design and coordination. JSE Conceived of the study and participated in its design and coordination. All authors read and approved the final manuscript | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544565.xml |
543578 | Effects of thyroxine and 1-methyl, 2-mercaptoimidazol on phosphoinositides synthesis in rat liver | Background Phosphoinositides mediate one of the intracellular signal transduction pathways and produce a class of second messengers that are involved in the action of hormones and neurotransmitters on target cells. Thyroid hormones are well known regulators of lipid metabolism and modulators of signal transduction in cells. However, little is known about phosphoinositides cycle regulation by thyroid hormones. The present paper deals with phosphoinositides synthesis de novo and acylation in liver at different thyroid status of rats. Results The experiments were performed in either the rat liver or hepatocytes of 90- and 720-day-old rats. Myo -[ 3 H]inositol, [ 14 C]CH 3 COONa, [ 14 C]oleic and [ 3 H]arachidonic acids were used to investigate the phosphatidylinositol (PtdIns), phosphatidylinositol 4-phosphate and phosphatidylinositol 4,5-bisphosphate (PtdInsP 2 ) synthesis. 1-methyl, 2-mercaptoimidazol-induced hypothyroidism was associated with the decrease of myo -[ 3 H]inositol and [ 3 H]arachidonic acids incorporation into liver phosphoinositides and total phospholipids, respectively. The thyroxine (L-T 4 ) injection to hypothyroid animals increased the hormones contents in blood serum and PtdInsP 2 synthesis de novo as well as [ 3 H]arachidonic acids incorporation into the PtdIns and PtdInsP 2 . Under the hormone action, the [ 14 C]oleic acid incorporation into PtdIns reduced in the liver of hypothyroid animals. A single injection of L-T 4 to the euthyroid [ 14 C]CH 3 COONa-pre-treated animals or addition of the hormone to a culture medium of hepatocytes was accompanied by the rapid prominent increase in the levels of the newly synthesized PtdIns and PtdInsP 2 and in the mass of phosphatidic acid in the liver or the cells. Conclusions The data obtained have demonstrated that thyroid hormones are of vital importance in the regulation of arachidonate-containing phosphoinositides metabolism in the liver. The drug-induced malfunction of thyroid gland noticeably changed the phosphoinositides synthesis de novo. The L-T 4 injection to the animals was followed by the time-dependent increase of polyphosphoinositide synthesis in the liver. The both long-term and short-term hormone effects on the newly synthesized PtdInsP 2 have been determined. | Background The phosphoinositides is a family of lipids which members play an essential role in the receptor-mediated intracellular signaling cascades, vesicle trafficking and cytoskeletal rearrangements [for review see [ 1 , 2 ]] and, therefore, are crucial for the adaptation and survival of cells. Because of their importance in numerous signaling events, the phosphoinositides require an absolutely tight temporal and spatial regulation of synthesis and degradation, enabling the cell to maintain organelle identity and housekeeping functions. A rapid agonist-dependent burst of phosphoinositides biosynthesis was the first feature of the polyphosphoinositide signaling pathway to be discovered, illustrating that synthesis is tightly coupled to degradation. The two pools of phosphoinositides are supposed to exist in the cells [ 3 , 4 ]. One of these is sensitive to the hormone-induced hydrolysis, and the other is hydrolysis-insensitive. The biosynthesis of phosphatidylinositol (PtdIns) is the two-component process composed of cytidyltransferase followed by a synthase [for review see [ 5 , 6 ]]. The first detailed study of the occurrence and localization of the PtdIns synthase activity in various tissues from guinea pig was carried out by Benjamins and Agranoff [ 7 ]. They detected the enzyme activity in all the tissues tested, including brain, liver, kidney, heart, lung and spleen. The PtdIns synthase activity was also detected in the endoplasmic reticulum, plasma membranes, Golgi apparatus and nuclei. The hormone-sensitive pool of phosphatidylinositol 4,5-bisphosphate (PtdInsP 2 ) in the plasma [ 8 ] and nuclear [ 9 ] membranes re-synthesized in these membrane fractions. The PtdIns synthesized de novo in the endoplasmic reticulum could be converted to the glycosylPtdIns or transferred by transfer proteins to other cellular compartments where the lipid was used for the phosphatidylinositol 4-phosphate (PtdInsP), PtdInsP 2 and other polyphosphoinositides production [ 10 , 11 ]. The regulation of the key enzymes of the PtdIns synthesis pathway: the CDP-diacylglycerol (DAG) synthase and PtdIns synthase were examined [ 12 ]. Expression of the PtdIns synthase gene caused the overproduction of the both of the PtdIns synthase and PtdIns:inositol exchange reactions, indicating that the gene encode the both enzymes. However, the overexpression of PtdIns synthase or CDP-DAG synthase alone or in combination in the COS-7 cells did not enhance the rate of the PtdIns biosynthesis and did not result in a significant proportional increase in the CDP-DAG and PtdIns cellular levels. The CDP-DAG synthase activity was inhibited by polyphosphoinositides in vitro [ 13 ]. This makes possible to suggest that these end products of the pathway may function as the feedback inhibitors of PtdIns biosynthesis in vivo. The PtdIns synthase activity has been shown to be upregulated after the hormone-induced phospholipase C (PLC) mediated hydrolysis of phosphatidylinositol-polyphosphates [ 14 , 15 ]. The efficiency of the PtdIns synthesis is dependent on the CDP-DAG fatty acids composition [for review see [ 5 , 6 ]]. In the bovine brain, the CDP-DAG has been shown to present predominantly as the 1-steroyl, 2-arachidonyl which are also the main PtdIns components. Thyroid hormones are of vital importance in maintaining the initial level of phospholipids in cell membranes and fatty acids composition of the lipids. On the other hand, there is a little literature regarding the hormone regulation of phosphoinositides exchange. The levels of [ 32 P]phosphoinositides and inositol 1,4,5-trisphosphate were found to be significantly lower in the hypothyroid rat hearts [ 16 ]. The effect of hypothyroidism on the insulin- and epinephrine-stimulated phosphoinositide metabolism has been investigated in the rat adipocytes [ 17 ]. The hypothyroidism enhanced the insuline-mediated phosphoinositides synthesis. The hypothyroidism caused a significant increase in both the basal and ouabain-stimulated accumulation of [ 3 H]inositol phosphate in the hypothalamic slices, whereas the thyroxine (L-T 4 ) completely restored the hypothalamic [ 3 H]inositol phosphate formation [ 18 ]. The results indicate that the negative feedback action of the thyroid hormone may occur at a post-receptor site in the hypothalamus. The thyroid hormones might participate in regulating the muscarinic cholinergic neurotransmission in the adult rats striatum via the stimulatory action on the inositol phosphate formation in the [ 3 H]inositol pre-labeled tissue [ 19 ]. Thus, it becomes evident that the thyroid gland malfunction leads to the prominent disturbances of signal transduction in the adipocytes and nervous cells via the changes of phosphoinositides metabolism. In the paper, we examined the influence of the drug- and L-T 4 -altered thyroid status of animals on the phosphoinositides synthesis in the liver. Thus, it was determined that the hypothyroidism decreased considerably the levels of the newly synthesized lipids in the liver but thyroid hormones increased the synthesis of the polyphosphoinositides de novo and the arachidonic acid incorporation into the PtdIns and PtdInsP 2 . A single injection of the L-T 4 to the euthyroid rats lead to the rapid and transient decrease of the newly synthesized PtdInsP 2 followed by the increase of the PtdIns and PtdInsP 2 levels in the liver. Results and Discussion The present paper considers the influence of thyroid functional status on the phosphoinositides synthesis in the liver. To determine the role of thyroid hormones in the regulation of phosphoinositides synthesis de novo and lipids fatty acid remodeling, the euthyroid, MMI (1-methyl, 2-mercaptoimidazol)- and MMI+L-T 4 -treated rats and intact animals after the single hormone injection has been studied. It has been reported that the L-T 4 induced hydrolysis of PtdInsP 2 and inositol phosphates and DAG formation at the early stages of hormone action on the [ 14 C]oleic and [ 14 C]linoleic acid pre-labeled hepatocytes of adult 90-day-old rats [ 20 , 21 ]. The L-T 4 -mediated PLC activation was accompanied by the protein kinase C (PKC) translocation to membranes [ 20 ] and PKC dependent stimulation of mitogen-activated protein kinase [ 22 ] and acylation of phospholipids and triacylglycerol synthesis [ 23 ]. The PtdInsP 2 -specific PLC activation in the L-T 4 -stimulated cells was a short-lived event. The hormone-stimulated rise in inositol 1,4,5-trisphosphate (Ins(1,4,5)P 3 ) was followed by its conversion into the biologically inactive inositol 1,4 -bisphosphate and inositol 1-phosphate [ 21 ]. It is known that after the receptor-triggered hydrolysis of phosphatidylinositol-polyphosphates the phospholipid must be resynthesized in order to maintain a constant level of phosphoinositides in the membranes. A single injection of L-T 4 to the 90-day-old rats leads to the rapid and sustained increase in the L-T 4 and triiodothyronine (L-T 3 ) levels in blood serum (Table 1 ), the content of the newly synthesized PtdIns and transient decrease the PtdInsP 2 level in the [ 14 C]CH 3 COONa-pre-labeled liver, which was followed by the polyphosphoinositide level increase (Figure 1A ). Taking into account that the PtdInsP 2 suppresses the key enzymes of PtdIns synthesis [ 13 ], the transient hormone-stimulated and phospholipase C-mediated [ 20 , 21 ] drop of the newly synthesized polyphosphoinositide in the liver cells was supposed to be the stimulus for the PtdIns synthesis. The L-T 4 administration to the old 720-day-old rats did not change the newly synthesized PtdIns and PtdInsP 2 levels in the liver (Figure 1B ) although increased thyroid hormones contents in blood serum (Table 1 ). It has been shown that the thyroid hormones are unable to stimulate rapidly the phospholipase C in the [ 14 C]CH 3 COONa-pre-labeled liver slices and hepatocytes of old animals [ 24 ]. These observations, together with the earlier data [ 20 , 21 ], suggest that the L-T 4 stimulates rapidly and nongenomicaly the phosphoinositides degradation and resynthesis in the liver of adult rats and does not act on the phosphoinositides synthesis in the cells with the disability of polyphosphoinositide signaling at old age. Table 1 Thyroxine and triiodothyronine contents in the blood serum of the rats of different thyroid states and age. Animals T 4 T 3 Adult control 128 ± 3.80 1.21 ± 0.20 Adult MMl-treated 29.5 ± 5.00** 0.83 ± 0.01* Adult MMl+T4-treated 57.3 ± 7.30*** 1.61 ± 0.10*** Adult T4-treated 1 654 ± 50.0** 5.78 ± 0.20** 2 675 ± 98.5** 15.9 ± 4.10** Old control 70.7 ± 11.6 1.37 ± 0.07 Old T4-treated 1 730 ± 15** 5.8 ± 1.5** 2 537 ± 128** 4.9 ± 0.0** The T 3 and T 4 contents in the blood serum were determined by radioimmunoassays kits. The amount of thyroid hormones in serum was represented as nmol per liter. Treatment of the rats by MMl was performed as described in "Materials and Methods". T 4 (200 mg/100 g weight) was injected to the MMl-treated rats 48 h prior to killing or normal rats 15 (1) and 60 (2) min prior to killing. Results are mean ± S.E. of 6–8 individual experiments performing in duplicate. In the Tables 1–2 and Figures 1–3 one experiment is equivalent to measurement of the parameters studied in a sample of liver of single animal. * P < 0.05 vs. control, ** P < 0.001 vs. control, *** P < 0.05 vs. MMl-treated rats. Figure 1 Short-term effects of L-thyroxine on the newly synthesized phosphoinositides levels in liver. Panel A – 90-day-old rats; panel B – 720-day-old rats. The liver lipids were labeled by [ 14 C]CH 3 COONa as described in "Materials and Methods". The L-T 4 (200 μg/100 g weight) was injected to the normal rats 15 and 60 min prior to killing. The lipids were extracted and separated as described in "Materials and Methods" and the radioactivity was determined by a liquid scintillation counter. Results are mean ± S.E. of six experiments performed in duplicate. In addition to its established role as a precursor for the signaling molecules (Ins(1,4,5)P 3 and inositol 3,4,5-trisphosphate), the PtdInsP 2 is now recognized as an important plasma membrane signal that activates the PLD [for review see [ 25 ]] and thus establishes the sites for vesicular trafficking, membrane movement and cytoskeletal assembly. It has been demonstrated in the mammalian cells that the PtdInsP 2 is a membrane-associated cofactor of PLD [ 26 , 27 ]. The rapid stimulation of the PtdInsP 2 -specific PLC by the L-T 4 in the rat hepatocytes of adult 90-day-old rats was followed by the prominent PLD activation [ 20 ]. The PLD response in the stimulated cells was reduced by the both PKC inhibitor and a high affinity ligand of PtdInsP 2 – neomycin. Neomycin does not directly interact with PLC and PLD but interacts with the endogenous membrane PtdInsP 2 . Inclusion of the PtdInsP 2 in mixed phosphatidylcholine/phosphatidylethanolamine liposomes [ 26 ] and stimulation of the phosphoinositide 5-kinase by an addition of the MgATP [ 27 ] greatly potentiates the PLD activation. Direct evidence that phosphoinositide kinase is involved in the PLD activation comes with the use of an inhibitory antibody for this enzyme [ 26 ]. The results strongly demonstrate that the PLD activation requires the enhanced PtdInsP 2 synthesis and the resting cellular levels of PtdInsP 2 are insufficient for enzyme stimulation. The role of PKC has been determined in the regulation of PtdInsP 2 synthesis in the liver [ 28 ] and other cell types [ 29 ]. The PKC translocation to membranes and enzyme activation are initial steps in the PLD [for review see [ 25 ]] and phosphoinositide 5-kinase [ 29 ] stimulation. The lack of PKC/PLD response [ 24 ] correlates with the suppressed ability of the L-T 4 to stimulate the PtdInsP 2 synthesis in the liver cells of the 720-day-old rats (Fig. 1B ). However, in the liver cells of adult animals the L-T 4 -induced PKC activation and elevation of PtdInsP 2 synthesis might lead to the PLD activation. To determine whether the altered thyroid functional status influences the phosphoinositides synthesis, we studied the myo -[ 3 H]inositol incorporation into the PtdIns, PtdInsP and PtdInsP 2 in the liver of the drug- and L-T 4 -treated animals. The MMI administration to the rats was found to be accompanied by the decreased myo -[ 3 H]inositol incorporation into the PtdIns, PtdInsP and PtdInsP 2 in the liver (Figure 2A,2B ) and drop of thyroid hormones levels in the blood serum (Table 1 ). The data obtained are consistent with the observations [ 30 ] of the drug-induced thyroid gland malfunction. The L-T 4 injection to the hypothyroid rats increased the T 4 and T 3 contents in blood serum (Table 1 ), the myo -[ 3 H]inositol incorporation into the PtdInsP 2 (Figure 2B,2C ) and did not changed the levels of the other newly synthesized phosphoinositides in the hypothyroid liver. Figure 2 Phosphoinositides synthesis in the liver of rats of different thyroid status. Panel A shows myo -[ 3 H]inositol incorporation into the lipids of control rats. Panel B shows the lipid precursor incorporation into the lipids of liver of MMI-treated animals. Panel C shows effect of L-T 4 on phosphoinositides synthesis in liver of MMI-treated rats. Treatment of the 90-day-old rats by MMI was performed as described in ''Materials and Methods''. L-T 4 (200 μg/100 g weight) was injected to the MMI-treated rats 48 h prior to killing. Control rats received the same volume of 0.9% NaCl. Liver slices were incubated in the presence of the myo -[ 3 H]inositol and lipids were extracted and separated as described in ''Materials and Methods'' and the radioactivity was determined by liquid scintillation counter. Results are mean ± S.E. of 6 – 8 individual experiments performed in duplicate. * P < 0.05 vs. control, **P < 0.05 vs. MMI-treated rats. Thyroid hormones stimulate lipogenesis in the liver by inducing the enzymes in the lipogenic pathway. The acyl-CoA-glycerol-3-phosphate acyltransferase, which is known to catalyze a rate-limiting step for the synthesis of phosphatidic acid in the rat liver, is dependent on the thyroid gland function [ 31 ]. The L-T 4 administration to the euthyroid rats increased the incorporation of the [ 14 C]palmitic acid into the phosphatidic acid and PtdInsP 2 in the isolated hepatocytes [ 28 ]. The results obtained in the present work demonstrated that the L-T 4 addition to the culture medium significantly increased the mass of phosphatidic acid and did not change the content of PtdIns in the hepatocytes (Table 2 ). The hormone addition to the culture medium caused the prominent and rapid (within 60 min of cells incubation) increase in the phosphatidic acid synthesis de novo and [ 14 C]palmitic acid incorporation into the PtdInsP 2 and did not change the PtdIns labeling in the isolated hepatocytes [ 28 ]. It can be said that the L-T 4 stimulates the PtdIns precursor synthesis and accumulation in the liver cells in the both long- and short-term manner. The thyroid hormone activation of phosphoinositide synthesis in the liver cells can be supposed to go through an enhancement of the glycerol-3-phosphate acylation, phosphatidic acid accumulation and its conversion into phosphoinositides. Besides, the phosphatidic acid can activate the phosphoinositide 5-kinase [ 32 ] and thus stimulate the PtdInsP 2 synthesis in the hormone-treated liver cells. Considering that the hormone could rapidly (within 60 min) stimulate the phosphatidic acid and PtdInsP 2 accumulation and did not change the PtdIns content in the liver cells, it could be assumed that in such case the L-T 4 increases PtdInsP 2 synthesis via phosphoinositide 5-kinase activation rather than the lipid synthesis de novo. Table 2 Rapid effect of L-T 4 on phosphatidic acid and phosphatidylinositol contents in the isolated hepatocytes. Lipid Cells: Control L-T 4 -treated Phosphatidic acid 9,29 ± 0,71 17,7 ± 2,94 * Phosphatidylinositol 13,6 ± 0,62 13,7 ± 1,83 Hepatocytes were isolated from the liver of adult 90-day-old rats and lipids contents were analyzed as described in "Materials and Methods" and expressed as % of total phospholipids. Hepatocytes were treated with 100 nM NaOH (control) or 10 nM of L-T 4 for 30 min. Results are mean ± S.E. of six experiments. * p < 0,05 vs. control. It is well documented that the efficiency of polyphoshoinositide derived second messanger DAG in signaling pathways is closely dependent on the degree of its unsaturation. The unsaturated fatty acids are incorporated into the sn-2 position of the phospholipids by the deacylation-reacylation reactions. Some investigations demonstrated the changes of the fatty acid composition of membraneous lipids at the hypo- and hyperthyroid state of the rats [for review see [ 33 ]]. There were generally a reciprocal changes in membranes arachidonic acid contents, namely, a decrease in the hypothyroidism and its increase after the thyroid hormone injection. The L-T 3 administration to the euthyroid rats increases the saturated fatty acids and arachidonate/linoleate ratio of PtdIns in the liver cell mitochondria [ 34 ]. The hypothyroidism was associated with the decrease in the [ 3 H]arachidonic acid incorporation into the liver total phospholipids (Figure 3A ). The [ 14 C]oleic acid labeling of the liver phospholipids was not dependent on the thyroid status of the rats (Figure 3C ). The L-T 4 injection to the hypothyroid animals completely abolished the drug-induced changes of the [ 3 H]arachidonic acid incorporation into the liver total phospholipids (Figure 3A ). The hypothyroidism was associated with the increased arachidonic acid conversion into the prostaglandine E 2 in the adult rat liver [ 35 ]. The L-T 4 administration to the thyroidectomysed animals reduced the prostaglandine E 2 synthesis and its content in the liver. Conceivably, the thyroid hormones regulating the arachidonic acid metabolism could maintain the initial level of polyunsaturated phospholipids in the liver cells. Figure 3 Incorporation of [ 3 H]arachidonic and [ 14 C]oleic acid into liver total phospholipids and phosphoinositides of rats of different thyroid status. Panel A and B show [ 3 H]arachidonic acid incorporation into total phospholipids and phosphoinositides of rats of different thyroid status, respectively. Panel C and D show the [ 14 C]oleic acid incorporation into total phospholipids and phosphoinositides of rats different thyroid status, respectively. Treatment of the 90-day-old rats by MMI was performed as described in "Materials and Methods". L-T 4 (200 μg/100 g weight) was injected to the MMI-treated rats 48 h prior to killing. Control rats received the same volume of 0.9% NaCl. Liver slices were incubated in the presence of the [ 3 H]arachidonic or [ 14 C]oleic acid and lipids were extracted and separated as described in ''Materials and Methods'' and the radioactivity was determined by liquid scintillation counter. Results are mean ± S.E. of 6 – 8 individual experiments performed in duplicate. * P < 0.05 vs. control, **P < 0.05 vs. MMI-treated rats. The MMI did not change markedly the 14 C/ 3 H-labeling of PtdIns and PtdInsP 2 in the liver slices (Figure 3B,3D ). The L-T 4 administration to the rats increased the incorporation of [ 3 H]arachidonic acid into the PtdIns and PtdInsP 2 in the hypothyroid liver (Table 3B) and decreased the content of the oleate-labeled PtdIns in the liver slices (Table 3D). The incorporation of the [ 14 C]oleic acid into the PtdInsP 2 did not differ between the liver slices of control, drug- and hormone-treated animals. As can be seen from the Table 1 , the MMI reduced the T 4 and T 3 levels in the blood serum, but did not remove completely the hormones from organism. The T 3 content in the serum of the drug-treated animals was relatively high as compared with control animals. However, the MMI reduced the T 4 level by 77%. The T 4 administration to the MMI-treated rats increased significantly T 3 and T 4 contents in the serum. It seems possible that the 14 C/ 3 H fatty acids incorporation into the phosphoinositides was rather regulated by the T 3 than T 4 , although the both hormones are participated in the regulation of the other phospholipids acylation in the rat liver. Conclusions The present data have demonstrated that the phosphoinositides synthesis de novo and arachidonic acid incorporation into phospholipids are strongly dependent on the thyroid status of organism. The marked enrichment of animal cell phosphoinositides in arachidonate and the results obtained suggest an important role of the thyroid hormones in the regulation of polyunsaturated PtdIns and PtdInsP 2 synthesis, which are the predominant substrates of PLC in the numerous signaling pathways. The both long-term and short-term effects of hormone on lipid synthesis have been determined (Figure 4 ). The L-T 4 stimulates rapidly and nongenomicaly the PtdInsP 2 degradation and resynthesis in the liver of the adult rats and does not act on the phosphoinositides metabolism in the cells with the PLC/PKC signaling disability. Besides, the L-T 4 stimulates the polyphosphoinositides synthesis de novo via the long lag period of time essential to hormone interaction with nuclear receptors and stimulation of protein synthesis [ 36 , 37 ]. Considering that the PtdInsP 2 participates in the activation of different enzymes (the PLD, protein- and lipid-kinases) involved in the signal transduction in the stimulated cells, the abnormalities of the phosphoinositides synthesis at different pathological states of thyroid gland could disturb other hormones signaling in the target cells. Figure 4 Long- and short-term effects of thyroxine on phosphoinositides metabolism in rat liver. Hormone rapidly and non-genomicaly stimulates PtdInsP 2 degradation, DAG accumulation and PKC/PLD activation. Hormone-induced PLC activation is followed by PtdInsP 2 resynthesis, probably via phosphoinositide 5-kinase induction by newly synthesized- or PCh-derived PA. PA could further be converted to polyphosphoinositide precursor – arachidonate-contaning PtdIns under the long lag period of hormone action on the organism. PA – phosphatidic acid, CDP-DAG – cytidine diphosphate diacylglycerol, PAA – palmitic acid, AA – arachidonic acid, Ins – inositol, InsP – inositol 1-monophosphate, InsP 2 – inositol 1,4-bisphosphate, InsP 3 – inositol 1,4,5-trisphosphate, PCh – phosphatidylcholine. Materials and Methods Materials myo -[ 3 H]-inositol (58 mCi/mmol), [ 14 C]oleic acid (58 mCi/mmol) and [ 3 H]arachidonic acid (60 Ci/mmol) – Amersham Corp. and [ 14 C]CH 3 COONa (25 mCi/mmol) – BPO Isotop (Russia); silica gel from Woelm (Germany). Phosphatidylinositol, phosphatidylinositol 4-phosphate and phosphatidylinositol 4,5-bisphosphate lipid standards were obtained from Sigma (USA). T 4 and 1-methyl, 2-mercaptoimidazol were from Zdorov'e Trudyaschikhsya (Kharkov, Ukraine). T 4 and T 3 radioimmunoassay kits were from Minsk (Belarussia). Other chemicals used were of chemically pure grade. Animals 90- and 720-day-old male Wistar rats, which had a free access to food and water and were kept at 24°C on a cycle of 12 h light/12 h darkness were used for experiments. The MMI was injected intraperitoneally (1 mg/100 g weight) in 0.9% NaCl to the experimental animals every day during 16 days-experiment. In some cases, the MMI-treated rats were injected intraperitoneally by T 4 (200 μg/100 g weight) 48 h prior to killing. Besides, T 4 (200 μg/100 g weight) was injected to the normal rats 15 and 60 min prior to killing. Control rats received 0.9% NaCl of the same volume. The animals were starved overnight prior to experiment. The thyroid state of rats was monitored by radioimmunological determination of the T 4 and T 3 in blood serum. Experiments with Liver Homogenates and Slices The 1 mCi of [ 14 C]CH 3 COONa was intraperitoneally injected to rats four times every 30 minutes for 2 hours. The liver was perfused with 0.9% NaCl, then removed and washed in Krebs-Henseleit buffer, pH 7.4, containing 2 mM CaCl 2 and 0.2% BSA. The pre-labeled liver was used to obtain 10% homogenates and to analyze the 14 C-phosphoinositides. Besides, the slices of unlabeled liver were labeled by incubation in the Eagle medium containing 10% fetal calf serum, 100 units/l streptomycin, 100 units/l penicillin, 13 mg/ml gentamycin and 0.1 μCi/ml of mio- [ 3 H]inositol or 2.5 μCi/ml of [ 14 C]oleic acid or 2.5 μCi/ml [ 3 H]arachidonic acid for 1–2 h in 95% O 2 /5% CO 2 atmosphere at 37°C. The lipids were extracted and analyzed as described below. Cell Suspension Experiments The hepatocytes were isolated from the liver by the method described in [ 38 ]. Preparation of hepatocytes was started between 9:00 and 10:00 a.m. The cells (10 7 /ml) were incubated in the Eagle medium containing 10% fetal calf serum, 100 units/liter streptomycin, 100 units/liter penicillin, 13 mg/ml gentamycin and in the presence of 100 nM NaOH (control) or L-T 4 (10 nM) for 30 min in 95% O 2 /5% CO 2 atmosphere at 37°C. Before lipid extraction, the cells were washed twice with a Krebs-Henseleit buffer pH 7.4, containing 2 mM CaCl 2 , 25 mM HEPES, 0.1% BSA. The lipids were extracted and analyzed as described below. Extraction and Separation of Lipids The phospholipids were extracted according to Folch et al. [ 39 ], the phosphoinositides as described in [ 40 ]. The chloroform phase was collected and dried under N 2 at 37°C. The lipids were redissolved in chloroform/methanol (1:2, v/v) and applied on TLC plates. For a total phospholipids isolation, the solvent system: hexane/diethyl ether/acetic acid (80:20:2, v/v) was used, for PtdIns, PtdInsP and PtdInsP 2 – chloroform/methanol/NH 4 OH (50:40:10, v/v). For phosphatidic acid separation the two-dimensional TLC was used. The TLC plates were developed in chloroform/methanol/ NH 4 OH (60:35:5, v/v) (first direction) and after – in the second direction in chloroform/methanol/acetic acid/water (80:60:7.4:1.2, v/v). The phospholipids masses were determined as described in [ 41 ]. The gel spots containing [ 14 C/ 3 H]lipids were scraped and transferred to the scintillation vials. Radioactivity was measured by a scintillation counter. Authors' contributions NAB conceived of the study and participated in its design, coordination, and manuscript preparation. OAK participated in data collection and performed the statistical analysis. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC543578.xml |
552307 | Little evidence for an epidemic of myopia in Australian primary school children over the last 30 years | Background Recently reported prevalences of myopia in primary school children vary greatly in different regions of the world. This study aimed to estimate the prevalence of refractive errors in an unselected urban population of young primary school children in eastern Sydney, Australia, between 1998 and 2004, for comparison with our previously published data gathered using the same protocols and other Australian studies over the last 30 years. Methods Right eye refractive data from non-cycloplegic retinoscopy was analysed for 1,936 children aged 4 to 12 years who underwent a full eye examination whilst on a vision science excursion to the Vision Education Centre Clinic at the University of New South Wales. Myopia was defined as spherical equivalents equal to or less than -0.50 D, and hyperopia as spherical equivalents greater than +0.50 D. Results The mean spherical equivalent decreased significantly (p < 0.0001) with age from +0.73 ± 0.1D (SE) at age 4 to +0.21 ± 0.11D at age 12 years. The proportion of children across all ages with myopia of -0.50D or more was 8.4%, ranging from 2.3% of 4 year olds to 14.7% of 12 year olds. Hyperopia greater than +0.50D was present in 38.4%. A 3-way ANOVA for cohort, age and gender of both the current and our previous data showed a significant main effect for age (p < 0.0001) but not for cohort (p = 0.134) or gender (p = 0.61). Conclusions Comparison of our new data with our early 1990s data and that from studies of over 8,000 Australian non-clinical rural and urban children in the 1970's and 1980's provided no evidence for the rapidly increasing prevalence of myopia described elsewhere in the world. In fact, the prevalence of myopia in Australian children continues to be significantly lower than that reported in Asia and North America despite changing demographics. This raises the issue of whether these results are a reflection of Australia's stable educational system and lifestyle over the last 30 years. | Background The prevalence of myopia is currently receiving worldwide attention as many recent studies report dramatic increases over the last 20 years [ 1 , 2 ]. Myopia and its aetiology is an interesting example of the intertwining of 'nature and nurture' with both genetics and life-style environment as important issues [ 3 ]. There is strong evidence indicating that genetic inheritance is a major contributor, both from the examination of prevalences across different racial backgrounds [ 4 ], from family pedigrees [ 5 ] and from twin studies [ 6 ]. However, there is increasing evidence suggesting that high heritability does not preclude rapid environmentally-induced increases in prevalence [ 7 ], rather, inherited factors are likely to both drive the susceptibility and resistance to environmentally-induced myopia [ 6 , 8 ]. Despite much research interest over the last half century, there have been surprisingly few well-designed epidemiological studies of refractive error with large numbers of randomly selected younger school children to form the basis of valid world wide comparisons of the earliest stages of development of myopia [ 3 , 9 , 10 ]. However, a group sponsored by the World Health Organisation in 2001 has devised a protocol to be used during studies of refractive error across different cultural and ethnic settings: the 'Refractive Error Study in Children' (RESC) [ 11 ]. In general, estimates of the prevalence of myopia have shown less increase in the Western world than in Asia, and less increase in rural than in urban populations [ 1 , 10 , 12 - 16 ]. Five very large studies across two decades and involving over 10,000 children in Taiwan are very important for understanding the changing prevalence of myopia in young Asian children (1.8% in 1986 rising to 12% in 1995 for 6 year olds, 40% rising to 56% for 12 year olds) [ 2 ]. A similar change is also reflected in Singaporean studies of myopia in military conscripts aged 17 years (26% to 83% from the late 1970s to the late 1990s as reviewed by [ 1 ]), of whom notably 82% were Chinese [ 17 ]. It has often been suggested that myopia is more prevalent in ethnic Chinese (reviewed [ 18 ]), but only relatively recent studies compare the prevalence of myopia in young ethnic Chinese children living either in China and in other countries [ 1 , 12 , 15 , 18 - 22 ]. For younger Chinese children aged around 5–7 years, the prevalence of myopia was found to range from under 5% in rural China [ 14 , 23 ] to 24% in Chinese Malays [ 20 ] and 30% in urban Hong Kong [ 19 , 22 ]. For older Chinese children aged 11–12 years, the prevalence ranged from 23% of rural Chinese[ 14 , 23 ] to 40% in urban China[ 12 ], 47% of Chinese Malays [ 20 ] and 57% in urban Hong Kong [ 22 ]. Japan has a similarly high prevalence of myopia in young school children estimated in recent times to be 43.5% of 12 year olds [ 24 ]. By comparison, the epidemiology of refractive error for young Australian school children is relatively well documented and presents a very different profile. A number of studies were carried out in the early 1970s and the 1980s on relatively large groups of unselected primary school children from the socio-economic extremes (generally aged 5 to 12 years), and indicated a prevalence of myopia ranging from approximately 3% to 13% (see Table 1 ) [ 25 - 29 ]. Two of those early studies investigated children largely from underprivileged, rural, families [ 26 , 27 ], and the other was of children from several, middle to upper socio-economic class private schools [ 25 ]. One smaller study was carried out in the mid 1980s on children from a representative selection of government schools in Brisbane [ 29 ], and would therefore have investigated children from a broader range of backgrounds. Interestingly, this latter study was the only Australian study to have determined refractive error under cycloplegia, yet yielded the highest prevalence of myopia. Thus, it has been difficult to determine whether the prevalence of myopia has increased in young school children in Australia as reported elsewhere. The majority of Australian residents are of Caucasian extraction living a very western lifestyle, leading one to expect the prevalence of myopia to be similar to that found in US or Europe. Yet, studies suggest that the prevalence of myopia in Australian primary school children is low by world standards [ 10 ]. In 2003 we reported the relative proportions of refractive errors in a large unselected primary school population of 2,535 children drawn from a very broad range of socio-economic backgrounds in Sydney, the largest city in Australia, in the early 1990s [ 30 ]. The children attended fourteen primary schools and two preschools. As in the earlier studies, the proportion of children with myopia greater than -0.50 DS spherical equivalence, as determined by non-cycloplegic retinoscopy, was found to be low by world standards (1.0% of 4 year olds rising to 8.3% of 12 year olds). We have now analysed the prevalence of refractive error in a new similar group of 1,936 children unselected primary school children drawn generally from the same area as our first study. Methods The study design is a retrospective examination of records of the Vision Education Centre (VEC) [ 31 ] school vision screenings (so named because parents were not present to ratify history) conducted in the Clinic of the School of Optometry and Vision Science, UNSW. Approvals for the study and permission to approach schools were obtained from the Committee for Use of Humans in Research at the University of New South Wales (UNSW), Sydney, Australia. The protocols adhered to the tenets of the Declaration of Helsinki. Parents or guardians were provided with an information sheet and requested an outline of known symptoms. Signed consent was required prior to a child's participation. Sampling and recruitment Permission was obtained from the NSW Department of Education and the NSW Catholic Education Office to approach all schools in the eastern region of Sydney (some thirty coeducational primary schools) to send entire classes to the VEC. A flyer was sent describing the VEC science excursion and age-appropriate eye examination, inviting Years 1, 3 and 5 particularly to participate. The group of 1,936 children examined came from the eastern suburbs along the southern beaches of Sydney, and may be thought of as randomly selected with little likelihood of bias to the data as individual classes were free to respond. Children were drawn from twelve government and non-government primary schools and one pre-school and attended the clinic only once. During the 1996 Australian Bureau of Statistics census 14,785 children aged 4 to 12 years were recorded in this region (Randwick and Waverley precincts of Eastern Sydney) who came from a very broad range of ethnic and socio-economic backgrounds present, where 37 different languages might be spoken in the home [ 32 ]. This was reflected in the children attending VEC. Census data indicate approximately 9% of the children in the current study were likely to be of Asian origin [ 32 ], a figure supported by our interpretation of family name for each child [ 30 ]. Participation in the eye examinations was typically well over 90% for each class, with teachers reporting non-participation to be predominantly due to illness on the day. Less than 3% of parents intentionally prevented participation, even if eye care had previously been sought. This particularly high participation rate was largely due to the attraction of a an age-appropriate student-centred hands-on science lesson about eyes and vision [ 31 ] delivered alongside the eye examination. Clinical examination The comprehensive optometric examination by experienced paediatric practitioners included all age-appropriate tests meeting Australian Optometric Competency Standards, except that parents/guardians were not present to ratify history. Refractive error was determined by non-cycloplegic retinoscopy with optical fogging while the child maintained fixation on a distant non-accommodative (6 metre) target. In most cases refractive status was confirmed by subjective refraction. Other tests included letter visual acuity at 6 m and 33 cm, cover test for strabismus, motilities, saccades, pupil reactions, near point of convergence, heterophoria, stereopsis, accommodative facility, colour vision and ophthalmoscopy. Justification of choice of testing procedures Cycloplegic retinoscopy was not undertaken for many reasons including the fact that VEC studies started prior to the 2000 convention suggesting use of cycloplegic retinoscopy for studies of refractive error prevalence [ 11 ]. Secondly, the VEC visit was meant as an excursion and the children had to return to normal classes with near work demands after the morning outing. Thirdly, it was important for comparison purposes to use refractive data procured under the same conditions as that used for the earlier groups of children. Fourthly, an initial evaluation without cycloplegia is necessary in order to understand daily function. Fifthly, non-cycloplegic retinoscopy was only one component of the exam. Outcomes regarding a decision to refer would not alter for most children had a cycloplegic refraction been carried out, as several other near function tests that would also indicate the possible existence of latent hyperopia or pseudo-myopia were included. Lastly, the degree of refractive error may in fact be influenced by cycloplegia (see Discussion for elaboration [ 33 - 38 ]). Autorefractors were not employed as hand-held versions were unavailable when the first cohort was seen. Equally as important, there is no convincing evidence that the proportion of myopes identified in the sample would have changed [ 39 ]. Comparison with earlier data To compare the estimated prevalence of myopia in this urban population of 'Australian children' over the last decade, this more recent 2000s data set was analysed against data from an earlier cohort of 2,322 children with similar demographics seen in the early 1990's, using the same testing protocols and seen at the same venue [ 30 ]. The optometric results of that earlier cohort have previously been reported [ 40 ], and it was noted that 7.1% of those children were already wearing spectacles [ 30 ], indicating that our recruitment procedure did not preclude children already under the care elsewhere. The data for any child examined in both cohorts was deleted from the earlier data set to avoid bias in the analysis. The mean date of assessment for this last 2000s cohort was September 2000, and for the early 1990s cohort was June 1992. Thus, the average gap between assessments of children from the two cohorts was 8 years and 3 months. Statistical analyses Data was analysed by Analysis of Variance ANOVA (StatView software). Only refractive data from right eyes was used for the current refractive class analysis, as the correlation between right and left eye refractions was extremely high (p < 0.0005). The preferred criterion to define myopia in this study is that used clinically in Australia: a spherical equivalent equal to or more minus than -0.50 D. However, as myopia more minus than -0.50 D has occasionally been used to define myopia in epidemiological studies [ 13 , 19 , 41 ], analyses using the criterion 'myopia more minus than -0.50 D' were also performed for comparison. Hyperopia was defined as spherical equivalents greater than +0.50 D. Thus, emmetropia for this study was defined as refractions in the range -0.25 to +0.50 dioptres spherical equivalence inclusive. Means are quoted with the associated standard error. Results The records of 1,936 children aged 4 to 12 years from a non-clinical unselected population examined during the six years from March 1998 to May 2004 were analysed retrospectively to estimate the prevalence of different types of refractive error. Primary schools of their own choice sent more children from years 1, 3, and 5, which resulted in unequal numbers of children in each of the age groups. There were 925 boys and 951 girls, and the relative numbers for both males and females in each age group are shown in Table 2 . For 59 children, the gender was not indicated on the record card and could not be inferred with certainty from the given name. The data not associated with gender has only been included in analyses entitled 'All' as shown in Tables 2 and 3 . Mean age was 8.36 years. The relative proportions of the different classifications of refractive error for all children combined (including those of unknown gender) for each age group are shown in Table 2 . The mean spherical equivalent refraction of all 1,936 children was +0.45 ± 0.02 DS, however it should be noted that there is a preponderance of children aged 5–6, 9 and 11 years old corresponding with Years 1, 3, and 5 of primary school. Overall, there was no significant difference in spherical equivalent refractive error between girls and boys (p = 0.697). In general, mean refraction demonstrates a highly significant shift towards less hyperopia with increasing age (p < 0.0001) from 0.73 ± 0.1DS for 4 year olds to 0.21 ± 0.11 for 12 year olds, however this is more noticeable after the age of 9 years as seen in Fig. 1 . With increasing age, more children are found in the emmetropic category and fewer in the low hypermetropic category. A summary of the relative proportions of myopia and hyperopia for this cohort of children of all ages seen during the six years ('2000s' data) is given in Table 3 . The majority of children screened are emmetropic by our criteria: 53.0% averaged across all ages. The proportion of children manifesting moderate to high degrees of hypermetropia (≥+1.50 DS) is 6.2% across all ages. Only 6.9% of children of all ages had refractive errors more minus than -0.50 DS, ranging from 2.3% of 4 year olds to 13.3% of 12 year olds (Fig. 2 ). If the more liberal definition of myopia is applied (myopia equal to or more minus than -0.50), then 8.4% of all children were myopic (ranging from 2.3% of 2 year olds to 14.7% of 12 year olds). Only 0.8% of the 1,936 children were more than -4.00 DS myopic. Comparison with previous data The number of children per age group for the two cohorts is shown in Fig. 3 , and notably, the age profile differs slightly between cohorts, though the mean age of 8.37 years for the earlier cohort was similar to that of this later cohort. The mean and standard error for right eye spherical equivalent refractive error by age is shown for each cohort in Fig. 4 . A three-way analysis of variance was carried out for cohort, age and gender. There was a main effect for age (p < 0.0001), but no main effect for either cohort (p = 0.134) or gender (p = 0.61). However, there were age/gender/cohort interactions that indicate a trend towards an increasing shift away from the hyperopic refraction in the later cohort. Discussion An analysis of the prevalence of refractive errors in young school children in eastern Sydney during the last thirteen years has been presented. The latest data gathered from 1,936 unselected primary school-aged children in the last 6 years, indicates that the prevalence of myopia remains quite low compared to that reported for the western world and Asia, especially as refractive error was established by non-cycloplegic retinoscopy (as will be discussed later). These findings are not significantly different (p = 0.13) to our previous report [ 30 ] indicating that 6.5% of 2,535 unselected children aged 4 to 12 years seen in the early 1990s were myopic by at least 0.50 D. Notably, those children were of similar socio-economic and ethnic status drawn from the same region of Sydney and seen at the same Centre using the same testing protocol. Therefore, if we take the total 4,258 children seen since 1990, the relative frequency of refractive error across all is: 54.2% emmetropic by our criteria, 32.3% low to moderate hyperopes, 5.3% myopic greater than -0.50D spherical equivalence and 7.4% myopic by at least -0.50 DS. The number with myopia of at least -4.00 DS was an extremely small 0.6%. The prevalence of myopia in Sydney primary school children compared to the rest of the world As alluded to in the introduction, the proportion of Sydney children with myopia is dramatically less than in Asia. Indeed, the proportion appears significantly lower than in the USA [ 41 ] and Canada [ 42 ] (4% and 6% of 6 year olds respectively, or 20% of 12 year olds in USA), but higher than urban India with only 4.4% of all school children under 16 years myopic [ 13 ] and higher particularly than in other less developed countries [ 10 ]. In the past, a lack of internationally accepted definitions for 'myopia' has hampered valid comparisons across the various studies [ 10 ]. Commonly the criteria 'greater than -0.50 DS' or 'at least -0.50 DS' are employed. However, our separate analyses using both of these criteria only resulted in a difference of 1.5% of all children included as myopic, in keeping with other dual analyses [ 13 , 41 ], and is low either way when compared with Asia or North America. Comparison across studies is also difficult when only an 'overall' mean refraction is presented covering all children in a study, due to the well known increasing prevalence of myopia with age. Indeed, the comparison of data from our own two data sets is confounded to some extent by the slightly different age profiles for each cohort. However, in neither cohort was the age range nor mean significantly different, so the similar proportion of myopes is not unexpected. Comparison of refractive error with and without a cycloplegic agent The question of optimal ocular conditions for comparison of the prevalence of refractive errors remains controversial. A cycloplegic agent is typically proposed as the gold standard [ 3 , 43 , 44 ] in the belief that it will eliminate ciliary muscle action or spasm, and thus unmask latent hyperopia or pseudomyopia. Thus, the use of a cycloplegic would be firstly predicted to lead to a decrease in the prevalence of myopia, and an increase in the prevalence of hyperopia. However, as a cycloplegic also leads to associated mydriasis and the introduction of unpredictable spherical aberrations, it is arguable that cycloplegia will induce unpredictable errors. In fact, Gao et al [ 38 ] in 2002 reported significant changes in the refractive components of children's eyes under conditions of deep cycloplegia and mydriasis that were greatest in hyperopic eyes and smallest in myopic eyes, adding no definitive evidence as to the relative efficacy of cycloplegia. Thus there appears to be no scientific concurrence regarding the efficacy of cycloplegia for studies on the prevalence of myopia [ 35 - 37 ], with several major studies electing to use cycloplegia (see review in [ 10 , 9 , 11 ]) and others not [ 18 - 21 , 23 , 42 , 45 ]. Presumably this design variability exists because there is no decisive evidence indicating a difference between refractions determined with and without a cycloplegic agent in eyes that have a myopic refraction. In general, a more positive retinoscopic finding is reported under cycloplegia, though considerable individual variation is seen including a myopic shift in some [ 33 , 35 - 37 , 46 ]. Not surprisingly, the differences noted decreased both with age and with less positive refraction. As our refractive data was derived from non-cycloplegic retinoscopy we readily concede that mean refractive error may be less hyperopic than if a cycloplegic had been used. However, we suggest that as the influence of a cycloplegic is uncertain and is of least concern for myopes, the estimated prevalence of myopia will not be significantly altered by our decision to not use a cycloplegic. In support of this notion are new conference data from Rose et al [ 47 , 48 ] reporting refractive status ascertained by cycloplegic autorefraction in over 1,000 children aged 6–7 years from across the same city of Sydney. They reported values of 'around 3%' for the prevalence of myopia of at least 0.50D [ 47 ], and then the value of 1.5% for myopia of 'approximately 0.50D' [ 48 ] with a participation rate between 73 and 80%. From Table 2 it can be seen that 2.4% of our 6 year olds in the current study were at least 0.50D myopic – a value that is strikingly similar. Demographics versus lifestyle Worldwide patterns of the prevalence of myopia suggest significant differences are likely to be due to the different demographics and lifestyles [ 1 , 10 , 49 ]. Zadnik [ 41 ] concedes that the increase in numbers of myopic children in the US Orinda study may be due to changing ethnic demographics. The apparent slight increase in myopia in Australia reported in the current study may also be in part accounted for by our changing ethnic demographics in urban areas. However demographics and ethnic compositions are unlikely to be responsible for the large changes reported in Asian and some other western countries [ 1 , 50 ]. Whatever way it is argued, our results indicate little evidence for an epidemic of myopia although there is a developmental trend towards an earlier decrease in hyperopia to the point of myopia. Thus, the question of whether it is a matter of lifestyle, or perhaps familial environmental stress, or more, remains. Certainly, the education system and housing has changed little in Australia the last 30 years. By comparison, most Asian children participating in myopia epidemiological studies reportedly are more likely to live in high-rise residential blocks [ 17 ] and have strong demands at school to memorize along with parental and peer pressure to do well, and for some, a competitive entrance examination to enter school [ 19 , 51 ]. Conclusions It is concluded that despite some differences in methodology across earlier studies, the prevalence of myopia in young Australian school children does not appear to have increased significantly over the last 30 years if one allows for the change in ethnic demographics. It is also proposed that an explanation for the large increase in prevalence of myopia reported in other countries must include questions relating to lifestyle in addition to genetic propensity. Competing interests The author(s) declare that they have no competing interests. Authors' contributions SC conceived and designed the study, jointly analysed the results and worked the drafts through to the final version. BJ coordinated the study, collated the clinic records, jointly analysed the results, researched the background for the paper, prepared the draft manuscript and was responsible for presentation. Both authors read and approved the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC552307.xml |
544571 | Can gene expression profiling predict survival for patients with squamous cell carcinoma of the lung? | Background Lung cancer remains to be the leading cause of cancer death worldwide. Patients with similar lung cancer may experience quite different clinical outcomes. Reliable molecular prognostic markers are needed to characterize the disparity. In order to identify the genes responsible for the aggressiveness of squamous cell carcinoma of the lung, we applied DNA microarray technology to a case control study. Fifteen patients with surgically treated stage I squamous cell lung cancer were selected. Ten were one-to-one matched on tumour size and grade, age, gender, and smoking status; five died of lung cancer recurrence within 24 months (high-aggressive group), and five survived more than 54 months after surgery (low-aggressive group). Five additional tissues were included as test samples. Unsupervised and supervised approaches were used to explore the relationship among samples and identify differentially expressed genes. We also evaluated the gene markers' accuracy in segregating samples to their respective group. Functional gene networks for the significant genes were retrieved, and their association with survival was tested. Results Unsupervised clustering did not group tumours based on survival experience. At p < 0.05, 294 and 246 differentially expressed genes for matched and unmatched analysis respectively were identified between the low and high aggressive groups. Linear discriminant analysis was performed on all samples using the 27 top unique genes, and the results showed an overall accuracy rate of 80%. Tests on the association of 24 gene networks with study outcome showed that 7 were highly correlated with the survival time of the lung cancer patients. Conclusion The overall gene expression pattern between the high and low aggressive squamous cell carcinomas of the lung did not differ significantly with the control of confounding factors. A small subset of genes or genes in specific pathways may be responsible for the aggressive nature of a tumour and could potentially serve as panels of prognostic markers for stage I squamous cell lung cancer. | Background Lung cancer remains to be the leading cause of cancer death in many European and North American countries [ 1 , 2 ]. It accounts for 13% of all cancer diagnoses but is responsible for nearly 30% cancer deaths in the United States [ 2 ]. Substantial effort has been made to identify prognostic factors that can be used for better patient management and improved survival. As of 2001, as many as 169 prognostic factors were identified in Non-Small Cell Lung Cancer (NSCLC) [ 3 ]. However, only very few such as TNM stage or patient performance status are consistent predictors, but they still can not predict individuals' prognosis accurately within a stage. Indeed, why do some patients with stage I lung cancer progress very quickly while others survive for a long time cancer free? This puzzle naturally has prompted researchers to contemplate whether the aggressive nature of NSCLC is genetically predetermined and whether the difference in gene expression could be identified as a more reliable clinical outcome predictor. Searching for molecular prognostic markers is traditionally carried out by analyzing one or several gene expression products at a time, which can only touch a very small fraction of expressed genes in the genome. Fortunately, recently developed high-throughput technologies such as DNA microarray provide promising and efficient screening tools for this purpose. It has been used in lung cancer research to identify the subclasses associated with tumour differentiation and patient survival [ 4 , 5 ], to predict patient survival or potential metastasis of a tumour based on gene expression profiles [ 6 - 8 ], and to compare two predefined classes such as tumour vs. normal or smokers vs. non-smokers to reveal differentially expressed genes [ 9 - 13 ]. However, some of these findings are simply a reiteration of diagnoses that can be easily made by standard pathologic evaluation, and their added clinical values are limited. In addition, two major issues exist in most of those studies to search for prognostic markers: (1) Case selection criteria were not clearly defined. Different tumour type, grade, stage, treatment, and smoking history were often mixed together, making it difficult to assess whether gene expression profiling discriminated patient survival independent of other known predictors. Although tumour type and grade of differentiation are not consistently documented as prognostic factors, they are very important in determining a sample's class membership in gene expression profiling [ 4 - 6 ]. (2) A clustering approach has been used as a major analytical tool to characterize cancer phenotypes including histological type, metastatic potential or patient survival. However, clustering is more appropriate to visualize gene expression patterns, and its results are heavily affected by the distance matrix and linkage method selected [ 14 ]. The existing evidence supports the notion that a clustering algorithm mainly groups samples based on histology, a variable not yet proven as an independent factor in NSCLC prognosis. This reemphasizes a central question of whether a clustering approach can discern the aggressive nature of a tumour with the same histological type. In order to answer the question why do some patients with stage I squamous cell carcinoma progress rapidly after curative resection while others survive a long time without disease recurrence, we designed a case control study matching on important prognostic factors so that only the tumour genetic factor was assumed to be a major determinant in patients' prognosis. We explored whether the widely-used hierarchical clustering was applicable in our study and whether the differentially expressed genes or functionally related groups of genes had any predictive value in an independent group of similar patients. Results Clinical Characteristics of Selected Patients The clinical characteristics of the 15 stage I squamous cell carcinoma patients in our study is provided in the Additional file 1 . Since the first ten samples were matched and used for the initial marker selection, the two groups (sample# 1–5 vs. 6–10) were well balanced in terms of age, gender, tumour size, smoking history, and treatment. The characteristics of the additional five test samples were very similar to the group of low aggressive samples. Unsupervised Clustering When a subset of 2810 filtered genes was used to conduct hierarchical clustering for all 15 samples, two main clusters were formed (Figure 1 ). However, the clusters did not distinguish the two groups by survival outcome: high-aggressive and low-aggressive tumours were almost evenly distributed within each cluster. Three of the high-aggressive tumours were present in the left cluster and two in the right. For the ten low-aggressive tumours, five were in the left cluster and five were in the right. Figure 1 Hierarchical clustering for 15 samples. 2810 probe sets filtered by: standard deviation/mean across all samples > 0.06; and the expression level on the log2 scale ≥ 4.00 in ≥ 60% of the samples. H: indicates high aggressive tumors. Class Comparison and Top Candidate Gene Selection To identify a panel of genes that are differentially expressed between the high and low aggressive groups as potential prognostic biomarkers, we applied matched (pair of 1–6, 2–7, 3–8, 4–9, 5–10) and unmatched (group 1–5 vs. 6–10) t statistics to the ten well-matched samples. At p < 0.05, 294 and 246 genes were significant in matched and unmatched comparison, respectively, with 126 selected by both. The majority of significant genes were within a two-fold mean difference between the two comparison groups with p values ranging from 0.05 to 0.01 (Figure 2 , 3 ). Figure 2 Distribution of significant genes from matched analysis. 294 significant genes (P < 0.05) selected by matched analysis are plotted by fold difference (x-axis) vs. p value using t-test (y-axis) A y-axis greater than 1.3 is equivalent to a p value less than 0.05, and greater than 2 is equivalent to a p value less than 0.01. A positive or negative value at the x-axis indicates genes are up or down regulated in the high-aggressive group compared to the low aggressive group. Figure 3 Distribution of significant genes from unmatched analysis. 246 significant genes (p < 0.05) selected by unmatched analysis are plotted by fold difference (x-axis) vs. p value using t-test (y-axis) A y-axis greater than 1.3 is equivalent to a p value less than 0.05, and greater than 2 is equivalent to a p value less than 0.01. A positive or negative value at the x-axis indicates genes are up or down regulated in the high-aggressive group compared to the low aggressive group. From the list generated by matched analysis, we used 1–10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 250, and 294 genes each time and evaluated their discriminating capability for training samples by leave-one-out algorithm as illustrated in Figure 4 . As few as 10 genes were found sufficient to achieve 100% accuracy. The same procedure was performed for the gene list generated by unmatched analysis and similar results were obtained. The first 20 genes from each procedure, which had the highest signal-to-noise ratio and therefore accurately distinguished the training samples with contrasting outcome, were selected and combined. Table 1 lists the 27 unique genes between the two procedures. (There were 2 selected probe sets for each gene ATP1B1 and IGFBP3 , and they were counted once.) Figure 4 Leave-one-out prediction on training samples. The x-axis represents different numbers of significant genes from matched analysis that was used to predict a membership of a sample by the leave-one-out algorithm. The y-axis shows the correct prediction rate for the 10 training samples. Table 1 The top 27 unique genes with highest signal-to-noise ratios Gene symbol Gene name Unmatched Matched Up-regulated genes ATP1B1 ATPase, Na+/K+ transporting, beta 1 polypeptide TP53 tumor protein p53 (Li-Fraumeni syndrome) CYP26A1 cytochrome P450, family 26, subfamily A, polypeptide 1 SYCP2 synaptonemal complex protein 2 IGFBP3 insulin-like growth factor binding protein 3 CPOX coproporphyrinogen oxidase (coproporphyria, harderoporphyria) MAGEA1 melanoma antigen, family A, 1 (directs expression of antigen MZ2-E) H1F0 H1 histone family, member 0 MAGEA12 melanoma antigen, family A, 12 * Homo sapiens clone 23705 mRNA sequence * Homo sapiens cDNA: FLJ21672 fis, clone COL09025. FLJ20477 Homo sapiens cDNA FLJ39734 fis, clone SMINT2016146. Down-regulated genes P2RY5 purinergic receptor P2Y, G-protein coupled, 5 DKFZp586G0123 hypothetical protein DKFZp586G0123 EPB41L3 erythrocyte membrane protein band 4.1-like 3 DKFZP586A0522 DKFZP586A0522 protein CSAD cysteine sulfinic acid decarboxylase BRIP1 BRCA1-interacting protein 1 MYC v-myc myelocytomatosis viral oncogene homolog (avian) PDCD4 programmed cell death 4 (neoplastic transformation inhibitor) TAF6L TAF6-like RNA polymerase II, p300/CBP-associated factor (PCAF)-associated factor, 65 kDa ABCA12 ATP-binding cassette, sub-family A (ABC1), member 12 ZNF198 zinc finger protein 198 NOTCH2 Notch homolog 2 (Drosophila) TncRNA Human clone 137308 mRNA, partial cds. CLK1 CDC-like kinase 1 POGZ pogo transposable element with ZNF domain Significant in that particular analysis (matched or unmatched) * No gene symbol for these genes Linear Discrimination Analysis We applied linear discrimination analysis to the 27 genes selected from the previous step to assess the accuracy of class membership prediction for both the training and the test samples (Table 2 ). The overall error rate was 20% (3/15). Interestingly, the linear discrimination score of incorrectly classified samples was among the lowest (absolute value), suggesting a borderline expression pattern between the high and low aggressive tumours. Table 2 LDA classification using 27 top genes Sample LD score Class Prediction Probability Correct? 48521 -0.52 1 1 0.75 Yes 48536 -0.94 1 1 0.87 Yes 41923 -0.26 1 1 0.63 Yes 48549 -0.52 1 1 0.74 Yes 44680 -2.9 1 1 1 Yes 42613 1.0 2 2 0.89 Yes 76981 0.52 2 2 0.75 Yes 44661 2.08 2 2 0.99 Yes 86043 -0.19 2 1 0.59 No 86011 1.71 2 2 0.97 Yes 42616 0.12 ? 2 0.56 No 48556 0.05 ? 2 0.52 No 41932 -0.88 ? 1 0.86 Yes 42081 -0.52 ? 1 0.75 Yes 44656 -0.08 ? 1 0.54 Yes LD score: Linear discrimination function calculated value for a given sample; Class: a sample's membership to the low aggressive group (1) or high aggressive group (2), "?" is a test sample whose membership is not known for the procedure and needs to be predicted. Prediction: predicted sample membership. Probability: probability of a sample belonging to a given class based on the classifiers. "Correct?": Whether the prediction is correct compared to the true class of a sample. Gene Network Analysis The test statistics for all genes (22215) using R-package "global test" was not significant, indicating that the overall pattern was similar between high-aggressive and low aggressive groups in our study sample. Using the 126 overlapped genes between matched and unmatched comparisons, we found 24 gene networks from Ingenuity Pathways Knowledge Base. We performed an association test on each network and found that seven were strongly associated with survival (Table 3 ). We then used the genes in each of the seven networks to predict all 15 samples separately and detected an error rate ranging from 0 to 47%, with RAB6A network genes predicting all samples correctly. Table 3 Gene networks associated with survival Network genes † Score ‡ Association test p Prediction error APLP2 , ARL6IP , CASP3 , CCNG1 , CSF1 , DNMT1 , EPHA2 , ERCC3 , ERCC5 , F 2, F5 , FGF2 , FUBP1 , GPI , HAS2 , HMOX2 , IGFBP3 , LOC283120 , LOC91768 , MDM4 , MYC , P53AIP1 , PCNA , PEG3 , RARB , RPL21 , RPS6 , RRM2B , TAGLN2 , THBD , TMSB4X , TP53 , TP53I3 , TP73 , WT1 17 0.01* 7/15 AMSH , AR , ATF2 , BAG1 , BCL2L1 , CREBBP , CYBA , ENO1 , EP300 , FOXG1B , HOXA9 , HOXC8 , HSF2 , MADH1 , MADH2 , MADH3 , MADH4 , NCF1 , NCF2 , RBM14 , RNF14 , RTN1 , RUNX2 , SIAH1 , SP3 , TOB1 , TP53 , UBE2E3 , UBE2I , ZNF8 9 0.02* 3/15 SAC , TEC 2 0.01* 4/15 RIPK1 , TRIAD3 2 0.03* 4/15 MIR , TMEM4 2 0.01* 3/15 NSF , PIK3CG , RAB6A , RAB6KIFL 2 0.01* 0/15 ATP12A , ATP1A1 , ATP1A2 , ATP1A3 , ATP1B1 , FXYD7 1 0.004* 3/15 † Genes in bold face are focus genes (among 126 genes submitted to the Ingenuity knowledge base). ‡ Score indicates the probability that a collection of focus genes could be found in a given network by chance. It is the negative logarithm of the possibility. A score of 2 indicates that chance is only 1%. * Indicates a strong association between the expressions of genes in a network and survival. Discussion To address the critical clinical question of whether the aggressive nature of squamous cell carcinoma of the lung is genetically pre-programmed, we conducted a matched experiment using DNA microarray. The purpose of the design is to control for known confounding factors so that the true association between gene expression and patient survival can be determined. Our results have shown that microarray technology provides both opportunities and challenges in the identification of potential molecular prognostic markers. In our study, unsupervised clustering did not accurately separate patients based on their clinical outcome behavior. This is in contrast to some investigators [ 4 , 5 ] who, using a similar approach, have identified subclasses of tumours showing differing gene expression profiles correlated with varied clinical outcomes. Using the 19.2 K cDNA microarray chip, Wigle et al [ 7 ] successfully partitioned 39 mixed histological types and stages of non-small cell lung cancer into two distinct clusters, those with early recurrence and those without recurrence regardless of tumour types. There are several possible explanations for discrepant results between the studies. First, the formation of clusters is heavily affected by the number of genes used for clustering, the gene selection method, and the clustering algorithm. Highly varied genes generally dominate the clustering process and thus explain why highly different groups such as among subtypes of non-small cell lung cancer (squamous cell carcinoma vs. adenocarcinoma), primary vs. metastatic cancer, or cancer vs. normal tissue, can be reliably differentiated using this technique. However, for the same primary tumour where the clinical outcome is the only noticeable difference, as in our study, this approach might not be as useful. Second, it is not clear in Bhattacharjee et al and Garber et al studies [ 4 , 5 ] whether the gene expression profile was influenced by other prognostic factors such as stage, or whether it was truly a specific and an independent prognostic factor. Finally, differences in tumour series, microarray chip platforms, or data pre-processing could affect results across studies, even within a study [ 15 ]. In searching for genes responsible for tumour behavior and patient survival, a case control comparison between two different clinical outcomes (long survival vs. short survival or disease free vs. quick recurrence) or a survival cohort using Cox's proportional hazards model to find gene-outcome association are among the most common options [ 6 , 7 , 9 , 13 ]. However, a careful design and implementation for this type of study needs to be taken into consideration since a case-control or survival cohort design is prone to selection bias, i.e., patients enrolled in comparison groups are different other than the factors under study, which makes them incomparable [ 16 ]. Without taking any strategy such as randomization, matching, or stratification to deal with the potential biases, the study results should be reviewed with skepticism [ 16 ]. Specifically for microarray study of lung cancer outcome, there are many tumour, host, and environmental related factors that are associated with patient prognosis. The imbalance of these factors between the two comparison groups such as the extent of disease (stage), the presence of other diseases, and treatment makes it difficult to establish the true association. Although results have not been consistent in reporting tumour histology of NSCLC as an independent prognostic factor, available evidence has indicated that it could be important in gene profiling as major histological types could be easily separated by the clustering approach [ 4 - 6 ]. If we do not take histology into consideration in case selection and comparison, a distorted result might occur. In contrast, we focused on one subtype of NSCLC within the same stage and matched on all potential confounding factors of survival. The results showed little overall difference in the gene expression profile between the two outcome groups. Less than 300 probe sets were significant at p < 0.05 from over 22,000 probe sets and 20–30 of them were greater than 1.5 fold change between the two groups, which were hardly separable from random noise. It is a big challenge for microarray analysis to identify reliable genomic prognostic marker panels that can be generalized to independent samples. In our study, the markers based on signal-to-noise ratio did not perform very well on the independent samples although the small sample size could be partly responsible. The result may suggest: (1) survival of patients with squamous cell carcinoma could be the result of genetic and non-genetic factors acting together. Gene expression difference is only a partial explanation. (2) Gene expression among tumours is very heterogeneous, even for the same histological type. By examining our series of tumours, we noticed that even though cell type and grade were matched, there were still some other variables hard to control for, such as cancer cell growth patterns or the constitutions of cancer stroma. Different amounts of lymphocytes or fibroblasts may contribute to the heterogeneous gene expressions. (3) The aggressive nature of a tumour may be determined by a small portion of cells that acquire metastatic capacity through somatic mutation [ 17 ], and it is hard to capture these cells since microarray analysis can only examine a very limited portion of a tumour. (4) The genes responsible for tumour aggressiveness may be part of one or multiple pathways. The genes within specific pathways may not be the most differentially expressed and may be often overwhelmed by background noises across tumours; however, as a functional group, they could potentially determine the behavior of a tumour. We evaluated the pathway hypothesis by finding related genes in specific gene networks using our candidate genes and tested the correlation between the genes and prognosis. Using this strategy, we identified seven gene networks strongly associated with squamous cell carcinoma survival. Although the functional explanation of an entire gene network to survival is yet to be determined, the association of some individual genes such as p53, c-myc , and PDCD4 (programmed cell death 4) with lung cancer survival has been well-documented in the literature [ 18 - 22 ]. Rab6A and related genes, the network accurately separated tumours with 100% accuracy in our study, are involved in intracellular transport. Whether they are functionally relevant to cancer aggressiveness or just surrogate markers of the true underlying mechanism needs to be further clarified. Although using carefully-matched samples could potentially unveil a true association, the subjects eligible for inclusion are dramatically reduced, often leading to a relatively small sample size and insufficient power to detect a minor difference or overcome randomness [ 15 ]. Facing the reality of low reproducibility using microarray technology, it is important that an experiment starts with a good design to minimize various biases [ 15 ]. If results from a well-controlled study are promising, a larger scaled follow-up investigation will be warranted. Conclusions We found that the overall gene expression pattern between the high and low aggressive squamous cell carcinomas of the lung was similar after controlling for confounding factors. However, our results suggest a difference between high and low aggressive cancers may be due to a small number of functionally related genes; these are so-called pathway genes that are often overlooked by commonly used analytical approaches. Whether pathway genes work collectively as more reliable prognostic markers or not needs to be further investigated by more studies with a large number of samples. Methods Study Design and Sample Selection Cases were defined as the patients who survived less than 24 months after surgery (high-aggressive group) and controls were those who survived more than 54 months after surgery (low-aggressive group). The patient population, from which the cases and controls were drawn, was comprised of patients diagnosed with lung cancer from 1997 to 2001 who underwent curative resection at Mayo Clinic, Minnesota, USA. These patients were prospectively enrolled and had been actively followed since their initial surgery [ 23 ]. We restricted this study to stage I squamous cell carcinoma to gain more homogeneity in morphology and to be focused on a common type of lung cancer. Each case was matched to a control by tumour size and grade, age, gender, and smoking status so that the potential confounding factors could be minimized. For each potential patient, we carefully reviewed their medical records and follow-up data to confirm their clinical outcome and the cause of death if the patients were deceased. From a pool of 304 patients with stage I squamous cell carcinoma, five well-matched pairs were finally selected (See Additional file 1 ) and used for most of the analyses. Five additional patients who survived more than 52 months were included as a test group (See Additional file 1 ). All enrolled patients and use of their tissue samples in the study were approved by our Institutional Review Board. The resected tumour and adjacent lung tissues were fast frozen in -80°C within 30 minutes after the tissues were surgically removed. RNA Extraction and Microarray Hybridization All tissue specimens were reviewed by a pulmonary pathologist (MCA) to confirm their diagnosis and ensure that the tissue was appropriate for the experiments. Specifically, the percentage of total tumour, tumour necrosis, amount of inflammation associated with tumour, and cellularity of stroma were evaluated. In the frozen tissue blocks containing cancer, the non-neoplastic tissue was manually cut away from the block to assure at least 80% of the cancer component. Thirty mm 3 of each tissue were sectioned at 20 or 35 μm, collected in a buffer RLT (Qiagen, Valencia, CA) supplemented with β-mercaptoethanol, and homogenized using PT 1200C (Kinematica AG, Luzern, Switzerland) rotor/stator. The total RNA was isolated using the RNeasy kit (Qiagen, Valencia, CA) following the manufacturer's specifications. Microarray experiments were performed at the Mayo Clinic Microarray Core Facility. The quality and quantity of RNA samples were controlled by spectrophotometry and the Agilent 2100 Bioanalyzer. Hybridization washes and scanning were performed following the manufacturer's protocols (Affymetrix, Santa Clara, CA). The HG-U133A chip from Affymetrix was used and contains 22,283 probe sets, which we conveniently refer to as genes in this paper. Data Processing and Analysis The Affymetrix Microarray Analysis Suite version 5 (MAS5) was used to process the scanned chip images. This software generates a cell intensity file for each chip, which contains a single intensity value for each probe cell (cel file). Dchip 1.3 [ 24 , 25 ]( ) was used to calculate the Model Based Expression Index (MBEI). All chips were normalized against an array with a median overall intensity using the invariant set method, and their images were visually inspected for potential problems prior to any data processing and analysis. The MBEI was calculated using the Perfect Match (PM) only model with outlier detection and correction. The calculated expressions were log2 transformed. Control probe sets were excluded in the down-stream analyses. As a first step, we employed hierarchical clustering to evaluate the similarity and disparity in overall expression patterns among all 15 samples using a subset of 2810 genes, which were filtered by the following criteria: the standard deviation/mean across all samples >0.06 and the expression level on the log2 scale ≥ 4 in at least 60% of the samples. The distance matrix applied in the clustering was one minus the Pearson correlation coefficient (1-r), which measures the closeness between genes or samples, and the linkage method was centroid, which uses the centers of newly formed clusters (genes or samples) to calculate the distance between clusters [ 26 , 27 ]. In order to detect differentially expressed genes between the high and low aggressive groups, we conducted both matched and unmatched t statistics for the ten matched samples at the criteria of p < 0.05 and at least one present call in each comparison group. Next, we applied a feature selection process to isolate a subset of genes selected from the previous step that had a high discriminate power in separating the two distinct groups of tumours. In each step, one sample was withheld as a test sample, and a signal-to-noise ratio as described by Ramaswamy and colleagues [ 28 ] was calculated for each gene using the remaining nine samples in the two groups. Based on the number of genes (features) specified, the procedure chose the top genes with highest signal-to-noise ratios and created a linear model to predict the membership of the withheld sample using a weighted-voting algorithm [ 28 ]. This process was repeated 10 times (10 samples), and an error prediction rate was obtained for the specified number of genes. By trying out different numbers of genes, a zero error rate for training samples could be achieved. The minimum number of genes obtaining the zero error rate was chosen as the best candidates. Linear Discrimination Analysis [ 29 ] was applied using the subset of genes selected from the previous step to assess whether the genes can discriminate the high from the low aggressive nature of the training and test samples. This method utilizes all input genes (independent variables) to create a discriminant function that maximizes the ratio of between-group variance and within-group variance so that different classes (dependent variable, either low or high aggressive group in our study) can be better separated. Implicitly, each gene is assigned a weight in the function depending on how a gene separates in the two groups and how this gene correlates with other genes. After computation, each sample was given a discriminant score, predicted class membership, and probability for the assigned class. The prediction rate was calculated to evaluate the performance of the classifiers. Because of stringent matching criteria, we did not expect dramatic difference between the two comparison groups, as reported by other investigators who did not match comparison groups closely. We hypothesized that genes in certain pathways might play a role in squamous cell carcinoma prognosis. We submitted the significant genes selected by both matched and unmatched analysis to the Ingenuity Pathways Analysis application ( )and generated gene interaction networks. A test statistic on the association of gene members in a network with a clinical outcome was carried out by using the R package "global test" [ 30 ]. If a small p value (<0.05), particularly permutated when sample size is small, is obtained, there is a strong indication that the group of genes, no matter whether they are up or down regulated in the network, is associated with clinical outcome, i.e., long or short survival in our study. Authors' contributions ZS carried out the data analysis and participated in drafting the manuscript. PY designed the study, oversaw the analysis and interpretation, and participated in writing the manuscript. MC analyzed the tissue samples and contributed to the development of the manuscript. FK prepared the tissue samples for microarray assay. CE contributed to the development of the manuscript. JM contributed to the development of the manuscript. GM contributed to analysis of the data. Supplementary Material Additional File 1 Clinical characteristics of 15 cases of stage I Squamous Cell Carcinoma of the Lung. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544571.xml |
554776 | Rate variation in parasitic plants: correlated and uncorrelated patterns among plastid genes of different function | Background The analysis of synonymous and nonsynonymous rates of DNA change can help in the choice among competing explanations for rate variation, such as differences in constraint, mutation rate, or the strength of genetic drift. Nonphotosynthetic plants of the Orobanchaceae have increased rates of DNA change. In this study 38 taxa of Orobanchaceae and relatives were used and 3 plastid genes were sequenced for each taxon. Results Phylogenetic reconstructions of relative rates of sequence evolution for three plastid genes ( rbcL , matK and rps2 ) show significant rate heterogeneity among lineages and among genes. Many of the non-photosynthetic plants have increases in both synonymous and nonsynonymous rates, indicating that both (1) selection is relaxed, and (2) there has been a change in the rate at which mutations are entering the population in these species. However, rate increases are not always immediate upon loss of photosynthesis. Overall there is a poor correlation of synonymous and nonsynonymous rates. There is, however, a strong correlation of synonymous rates across the 3 genes studied and the lineage-speccific pattern for each gene is strikingly similar. This indicates that the causes of synonymous rate variation are affecting the whole plastid genome in a similar way. There is a weaker correlation across genes for nonsynonymous rates. Here the picture is more complex, as could be expected if there are many causes of variation, differing from taxon to taxon and gene to gene. Conclusions The distinctive pattern of rate increases in Orobanchaceae has at least two causes. It is clear that there is a relaxation of constraint in many (though not all) non-photosynthetic lineages. However, there is also some force affecting synonymous sites as well. At this point, it is not possible to tell whether it is generation time, speciation rate, mutation rate, DNA repair efficiency or some combination of these factors. | Background Rates of DNA sequence evolution vary among taxa and among genes, and the causes of this variation are many. In some cases, generation time has been shown to be correlated with rates in plants. For example, annual plants can sometimes have higher rates of DNA evolution than perennials [ 1 ]. In one study, it was shown that long-lived woody grasses exhibit slower rates than short-lived herbaceous ones [ 2 ]. However, a more extensive set of 33 phylogenetically independent comparisons failed to find a generation time effect for plants in general [ 3 ]. A useful method for distinguishing among the potential causes of rate variation is to separately examine nonsynonymous rates ( r N ) and synonymous rates ( r S ). For example, when r N increases relative to r S , relaxation of purifying selection is a possible explanation. However, when r S increases, but the r N / r S ratio is not greatly affected, then an increase in the mutation rate is a possibility. An example of this is Plantago mitochondrial DNA [ 4 ]. However, a decrease in DNA repair efficiency could also explain such a change. In addition, population processes, such as genetic drift could play a role. Reduced effective population size (N e ) can increase the fixation rate of neutral and slightly deleterious mutations. Thus, if slightly deleterious mutations are common, both r S and r N are expected to be higher when N e is low [ 5 , 6 ]. One expectation of this drift-based hypothesis is that r N will vary from protein to protein, as each protein will have different functional constraints and thus a different proportion of slightly deleterious mutations. In contrast, r S is expected to be similar among proteins when the cause is a change in mutation rate or repair efficiency[ 7 ]. During founder-effect speciation, genetic drift can be expected to increase the substitution rate, even at silent sites. This has been called the speciation-rate hypothesis [ 8 ]. For example, speciation rate has been used to explain the difference in non-coding DNA substitution rates between the sister genera Utricularia and Pinguicula in the plant family Lentibulariaceae [ 9 ]. Rates can also vary if the underlying mutation rate varies [ 10 , 11 ] or if DNA repair is impaired [ 12 , 13 ]. Although rates themselves are hard to measure, the number of synonymous ( d S ) and nonsynonymous ( d N ) substitutions can be measured and are used to compare rates and calculate rate ratios. Epifagus virginiana , a nonphotosynthetic plant, has an increased rate of sequence evolution for plastid DNA in general [ 14 - 17 ] and relative rates tests of the plastid rps2 gene indicate a significant increase for both d N and d S . This suggests that purifying selection is at least partially relaxed and that there has been an increase in the rate at which mutations are entering the population in this species, due to increased mutation rate or lax DNA repair. MatK , another plastid gene, is characterized by a partial relaxation of purifying selection in the clade containing Epifagus , Orobanche and Boschniakia [ 18 ]. In this paper, we explore rate variation in E. virginiana and 38 of its relatives for three plastid genes: rps2 , matK , and rbcL . Each of these genes is present in photosynthetic relatives of Epifagus, is accelerated (or even lost) in Epifagus or related parasitic plants. Although plastid encoded, the three genes encode proteins that participate in different processes in the plastid. rps2 encodes the ribosomal protein S2 in small subunit ribosome, matK is an intron maturase, and rbcL encodes the large subunit in the CO 2 -fixing enzyme RUBISCO. We ask several questions: When does the rate increase observed in Epifagus begin, relative to the evolutionary loss of photosynthesis? What are the causes? Relaxation of constraint? More mutations entering the population? Are these patterns consistent across multiple plastid genes? Results Phylogenies of the Orobanchaceae and relatives were constructed using maximum parsimony (MP) and maximum likelihood (ML). The MP analysis discovered four most parsimonious trees of 3817 steps, with CI = 0.6275, CI (excluding uninformative characters) = 0.5118, and RC = 0.3888. The strict consensus tree was unresolved as to the position of Schwalbea relative to the Alectra-Orobanche clade, the Bartsia-Melampyrum clade and the Castilleja-Pedicularis clade. It was also unresolved concerning the relationships among the outgroups Mimulus , Kigelia , Hemimeris , Verbascum , Antirrhinum and Veronica . The ML analysis found two trees, with -ln likelihood values of 24525.88663. The strict consensus of these trees was unresolved, but in a different place, regarding the position of the Cistanche-Epifagus clade. When the MP consensus and the ML consensus were combined into a semistrict consensus tree, a completely resolved tree resulted. This tree is shown in Fig. 1 . Figure 1 MP/ML consensus tree with MP bootstrap values All three gene trees exhibited statistically significant rate heterogeneity (p < 0.0005), as assessed by the Kishino-Hasegawa (K-H) test [ 19 ]. Synonymous and nonsynonymous branch lengths for each of the three genes are shown reconstructed in Fig. 2 . Figure 2 Nonsynonymous and synonymous rates of change in three genes. Nonphotosynthetic taxa are named in bold. (A) rps2 nonsynonymous branch lengths. (B) matK nonsynonymous branch lengths. (C) rbcL nonsynonymous branch lengths. (D) rps2 synonymous branch lengths. (E) matK synonymous branch lengths. (F) rbcL synonymous branch lengths. Taxa with rbcL pseudogenes are identified with Ψ. Uncertain pseudogene status is indicated by "Ψ?" [20, 26]. The correlation analyses show that there is a higher correlation of synonymous evolution across genes than nonsynonymous evolution (Fig. 3 ). They also show that within rps2 and rbcL synonymous and nonsynonymous evolution is poorly correlated, but in matK , the correlation is better (Fig. 4 ). Figure 3 Correlations across each possible pair of genes for synonymous and nonsynonymous rates. (A) rps2 versus matK . (B) rps2 versus rbcL . (C) matK versus rbcL . All nonsynonymous comparisons involving rbcL pseudogenes are indicated with a hollow triangle. Figure 4 Correlation plots of synonymous and nonsynonymous rates within each gene. (A) rps2 rates. (B) matK rates. (C) rbcL rates. In 4C, the Epifagus pseudogene has been excluded and the other pseudogenes are indicated with a hollow triangle. Some of the non-photosynthetic plants, ( Epifagus , Cistanche , and the Orobanche species) have increases in both synonymous and nonsynonymous rates. Rates are not, however, increased in Boschniakia , Harveya , Hyobanche , Lathaea , Alectra orobanchoides , and Striga gesnerioides . Synonymous rates vary markedly among taxa. For example, the branches leading to Epifagus are at least two or three times as long as those for most of the photosynthetic taxa. Despite this among-taxon variation, the pattern for each gene is strikingly similar (Fig. 2D, E, F ). For example, sister taxa (e.g. Veronica and Antirrhinum , Euphrasia and Panetucellia/Tozzia , Epifagus and Cistanche , O. fasciculata and O. corymbosa ) usually show an identical pattern of who is faster than whom. This indicates that the causes of synonymous rate variation are affecting the whole plastid genome in a similar way. There is even more extensive variation in nonsynonymous rates, both among taxa and among genes. This is not surprising because these genes have different functions and some of the taxa are photosynthetic while others are not. The scale bars in Figs. 2A, B , and 2C show the overall differences in nonsynonymous rates. MatK is much more rapidly evolving than the other two genes for nearly all taxa, but Epifagus' rbcL pseudogene has a similar rate. When looking at taxa across genes, there is much less consistency than with the synonymous rates. There are some big differences, such as the branch lengths for Striga , Cycnium and Schwalbea , and the Euphrasia species. Overall the picture is more complex, as could be expected if there are many causes of variation, differing from taxon to taxon and gene to gene. The pattern of matK rate variation is very similar in the synonymous and nonsynonymous figures. This fits well with the fact that it is less constrained overall, as can be seen by comparing the scale bars in Figs. 2A, B , and 2C . Rates were compared by using two categories at a time and testing for significant differences using likelihood ratio tests. These tests are summarized in Table 2 . In general, it was found that the nonphotosynthetic plants have higher synonymous and nonsynonymous rates of change, although when tested separately, it was found that Harveya , Hyobanche and Boschniakia do not have higher synonymous rates. Table 1 Specimens used for DNA sequencing, with GenBank accession numbers for rps2, matK, and rbcL sequences. GenBank accession numbers Species rps2 matK rbcL Alectra orobanchoides U48741 AF489960 AF026819 Alectra sessiliflora U48742 AF051977 AF026820 Antirrhinum majus U48766 AF051978 L11688 Bartsia alpina U48751 AY849600 AF190903 Boschniakia hookeri U48757 AF051979 AF026817 Boschniakia strobilacea U48758 AF051980 AF26818 Castilleja lineariifolia U48739 AF051981 AF026823 Cistanche phelypaea AY849597 AF056149 AY849862 Cycnium racemosum U48745 AY849601 AF026826 Epifagus virginiana EPFCPCG EPFCPCG, AF051982 EPFCPCG Euphrasia disjuncta AY849598 AY849602 AY849863 Euphrasia spectabilis U48752 AY849603 AY849864 Harveya capensis AF055142 AF489961 AF026829 Harveya purpurea U48749 AF051984 AF026830 Hemimeris sabulosa U48765 AF051985 AF123668 Hyobanche atropurpurea AY849599 AF051986 AF026831 Hyobanche sanguinea U48750 AF051987 AF026832 Kigelia africana U48764 AF051988 AF102648 Lathraea clandestina U48755 AF051989 AF026833 Lindenbergia phillipinensis AF055151 AF051990 AF123664 Melampyrum sylvaticum AF055148 AF051991 Melampyrum lineare AF026834 Melasma scabrum U48743 AY849604 AF190904 Mimulus aurantiacus AF055154 AY849605 AF026835 Nicotiana tabacum Z00044 Z00044 Z00044 Orobanche caryophyllacea AF055145 AF051992 AY582187 Orobanche cernua AF055147 AF056147 U73968 Orobanche corymbosa U48760 AF051993 U73969 Orobanche fasciculata AF055143 AF051994 U73970 Orobanche hederae AF055146 AF051995 AF078682 Orobanche ramosa U48761 AF056148 U73971 Parentucellia viscosa U48753 AY849606 AY849865 Paulownia tomentosa AF055255 AF051997 L36447 Pedicularis foliosa U48740 AF489959 AF026836 Schwalbea americana AF055150 AF051998 AY849866 Seymeria pectinata AF055141 AF051999 AF026837 Striga asiatica U48746 AF052000 AF026838 Striga gesnerioides U48747 AF489963 AF026839 Tozzia alpina U48754 AF052001 AF026843 Verbascum blattaria U48763 Verbascum thapsus AF052002 L36452 Veronica arvensis U48768 AF052003 Veronica persica L36453 Taxonomic authorities, localities and voucher information can be found in [25], with the exception of Euphrasia disjuncta Fernald & Wiegand, Hyobanche atropurpurea Bolus., and Veronica persica Poir. Purifying selection is relaxed in the nonphotosynthetic plants for all three genes. The test values are as follows. matK : 2×LR = 10.67, df = 1, p = 0.001; rbcL : 2×LR = 31.3, df = 1, p = 2.2 × 10 -8 ; r ps2 : 2×LR = 8.56, df = 1, p = 0.00343. Another way to describe the difference in the pattern of synonymous and non-synonymous rates is to say that the former are more correlated across genes. This can be seen in Fig. 3 , which shows plots comparing two genes at a time. In comparisons including rbcL nonsynonymous rates, the data point from the Epifagus pseudogene has been excluded. Its unconstrained evolution is not typical of "nonsynonymous" change and its position on the plot made it an extreme outlier with an enormous influence on the regression line. For rps2 and rbcL , the synonymous plots are more highly correlated, whereas for matK , which is relatively unconstrained, they are about the same. Discussion The dramatic rate increase observed in Epifagus [ 14 ], with branches 5–10 times as long as other taxa, can now be seen to have begun earlier in the history of the Orobanchaceae. It is shown to be composed of increases in both synonymous and nonsynomymous rates. The general pattern is that many of the non-photosynthetic plants, such as Epifagus , Cistanche , and the Orobanche species, have increases in both synonymous and nonsynonymous rates, indicating that both (1) selection is relaxed, and (2) there has been a change in the rate at which mutations are entering the population in these species. However, rate increases are not immediate upon loss of photosynthesis, since we do not see increases in Boschniakia , Harveya , Hyobanche , Lathaea , Alectra orobanchoides , and Striga gesnerioides . This pattern is similar to that found using smaller data sets [ 15 , 20 ]. Separate analyses of synonymous and nonsynonymous rates give us some insight into potential mechanisms. The speciation rate hypothesis predicts that more speciose clades should have a faster r s (and therefore larger d s ) than a species-poor sister group. This was suggested as a cause of rate variation in non-coding DNA in the Lentibulareaceae [ 9 ]. In this study, not all genera have been sampled, and those that have are often represented by one or two species. In addition, some genera may not be monophyletic (for example, Orobanche is not). Thus, accurate numbers of species cannot be assigned to individual branches or clades. However, a few things can be noted. Euphrasia , with ~170 spp., is clearly more speciose than its sister group, with 3 spp. It has a somewhat faster d s . However, the Schwalbea lineage, with a single species, has a fairly high d s . Its position is not certain, but its sister group is probably the Bartsia – Melampyrum clade (>300 spp.), the Castilleja – Pedicularis clade, (>700 spp.), or the union of the two. These groups have do not have dramatically higher d s values; in fact the Castilleja – Pedicularis clade's value is slightly lower. Differences in generation time may play some role in the observed rate variation. However, as was found previously [ 15 ], the pattern is not clear. Since most of the genera sampled in this study contain both annuals and perennials, it is likely that most branches on the tree actually represent a combination of annual and perennial evolutionary history. However, there are some intriguing details that might merit further study. The clade containing Bartsia , Euphrasia and Melampyrum contains mostly annuals [ 21 ] and has some high d s values, as one would expect from a generation time effect. However, both Euphrasia and Melampyrum contain almost exclusively annuals and have very different rates. Likewise, the large clade containing Boschniakia and Epifagus contains mostly annuals and has an overall high d s . The perennials Boschniakia and Cistanche have lower d s than their sister taxa, which also supports the generation time hypothesis, but there is as much variation among categories (perennial, annual) as between categories. Conclusions The distinctive pattern of rate increases in Orobanchaceae has at least two causes. It is clear that there is a relaxation of constraint in many (though not all) non-photosynthetic lineages. However, there is also some force affecting synonymous sites as well. At this point, it is not possible to tell whether it is generation time, speciation rate, mutation rate, DNA repair efficiency or some combination of these factors. Clearly, generating additional data from nuclear and mitochondrial genes would help us to more clearly distinguish among these hypotheses. Some of the above-mentioned hypotheses (generation-time, speciation rate) would be expected to affect nuclear and mitochondrial genomes in a similar fashion, whereas factors affecting mutation rate or efficiency of DNA repair would not, as these process involve different, though perhaps overlapping, sets of enzymes in each of the three genomes [ 22 - 24 ]. Methods Sampling We sampled 15 photosynthetic and 16 nonphotosynthetic Orobanchaceae, and eight outgroup taxa. The specimens used and their GenBank accession numbers are given in Table 1 . Table 2 P-values of the likelihood ratio tests. rps2 matK rbcL r N r S ratio (ω) r N r S ratio (ω) r N r S ratio (ω) photo. vs nonphoto. 1.4 × 10 -11 0.0014 0.0034 0 0.077 (NS) 0.001 0 2.3 × 10 -8 2.2 × 10 -8 photo. vs OEC 0 6.3 × 10 -10 7.2 × 10 -8 0 7.6 × 10 -14 0.0038 0 0 2.7 × 10 -11 photo. vs HHB 0.41 (NS) 0.0066 0.24 (NS) 1.4 × 10 -6 4.5 × 10 -9 0.23 (NS) 0.076 (NS) 0.0011 8.1 × 10 -5 The two categories at left were compared for nonsynonymous rate differences ( r N ), synonymous rate differences ( r S ) or rate ratio (ω) differences. "Photo." refers to photosynthetic branches, "nonphoto." refers to nonphotosynthetic branches, "OEC" refers to Orobanche , Epifagus and Cistanche branches, and "HHB" refers to Harveya , Hyobanche and Boschniakia branches. Amplification and sequencing We amplified and sequenced rps2 as in [ 15 ], matK as in [ 25 ], and rbcL as in [ 26 ]. A total of 15 new sequences were generated for this study, including 3 rps2 , 7 matK , and 5 rbcL sequences. Alignment The rps2 alignment was simple, containing only two small indels. For matK and rbcL , a search for the best alignment was conducted using Clustal X and a variety of alignment parameters. Alignments were evaluated according to the following optimality criterion: whichever alignment yields the MP tree(s) with the highest consistency is considered the best alignment[ 27 ]. For alignment assessment, MP analyses with and without indel characters were used. When used, indel characters were generated with the program GapCoder [ 28 ](available from ), which uses the simple indel coding method of Simmons and Ochoterena [ 29 ]. The rescaled consistency (RC) index [ 30 ] of the resulting parsimony analyses was used to assess alignment optimality, with one exception: very low gap opening penalties (GOP), such as 3 or less were excluded. These low GOP values lead to inflated RC values, due to the large numbers of gaps, which reduce the treelength and the homoplasy. rps2 had just one small indel and was aligned by eye. For matK , the optimal computer alignment was generated using GOP = 5 and gap extension penalty (GEP) = 1. Transitions were weighed the same as transitions. The RC from the analysis with indel characters included was 0.3693. The RC without indel characters was 0.3759. The alignment was then adjusted by eye. This final alignment yielded RC values of 0.3878 (indel characters included) and 0.3763 (indel characters excluded). For rbcL , the optimal computer alignment was generated using GOP = 5 GEP = 3. Transitions were weighed the same as transitions. The RC from the analysis with indel characters included was 0.4679. The RC without indel characters was 0.4437. The alignment was then adjusted by eye. This final alignment yielded RC values of 0.4635 (indel characters included) and 0.4499 (indel characters excluded). Phylogenetic analyses For rps2 , positions homologous to positions 48–660 of the Nicotiana gene were used. For matK , the entire gene was used. For rbcL , positions homologous to Nicotiana gene positions 5–1325 were used. The three genes were then combined into a single data set. PAUP* 4.0b8 [ 31 ] was used to conduct a MP heuristic search, including indel characters, and using Nicotiana tabacum as the outgroup taxon, TBR branch swapping and 100 random addition replicates. Bootstrap analyses were conducted with the same settings, except with only 40 random addition orders. 500 bootstrap replicates were performed. ML analyses excluded indels. Using the hLRT method of the program ModelTest 3.06 [ 32 ], the ML model of GTR+G was selected as the best evolutionary model for the combined data set. Base frequencies (A = .298, C = .177, G = .203) and substitution rates (A-C = 1.51, A-G = 2.11, A-T = 0.264, C-G = 0.788, C-T = 2.62, G-T = 1) were obtained from the MP trees. Among-site variation was included in the model, based on a gamma distribution with four categories. A heuristic search was conducted, similar to the MP search, but with only 10 random addition replicates, each limited to the examination of 5000 rearrangements. Rates of DNA change Overall rate heterogeneity was assessed using the K-H test as implemented in PAUP, using the same ML analyses, except that the starting tree was a neighbor-joining tree and the analysis was limited to 40 rearrangements. Nonsysnonymous and synonymous changes were reconstructed on branches using the codon-based likelihood model of Muse and Gaut [ 33 ], as implemented in HYPHY for MacOS, ver. 0.95 beta [[ 34 ] 2004], available at . At least nine of the rbcL "genes" are probably pseudogenes. These are indicated in Figure 2C . Seven of these have already been discussed elsewhere [ 20 ]. The Orobanche caryophyllaceae and Cistanche phelypaea "genes" have internal stop codons and thus are probably also pseudogenes. There may also be other pseudogenes with intact ORFs, making their pseudogene status less obvious [ 35 ]. Once a pseudogene is formed, it is no longer constrained for a protein function, so synonymous and nonsynonymous changes can no longer be formally defined. Moreover, changes that would have been synonymous and nonsynonymous are now expected at equal rates. Thus, by including these sequences in the tests, we get additional evidence that constraint is relaxed in nonphotosynthetic plants. Therefore, even for the pseudogenes, we have still calculated the synonymous and nonsynonymous rates separately, assuming a reading frame based on alignment to the other genes in the data set, and have indicated the pseudogenes in Fig. 2C . Rate increases were compared among categories of taxa (such as photosynthetic and non-photosynthetic), using d N , d S , and the d N / d S ratio (ω) in likelihood ratio tests [ 36 ]. These tests were conducted using HYPHY and the category assignments of the branches are those shown in Figure 1 . In addition, a previous study [ 15 ] indicated that some nonphotosynthetic branches might not have rate increases. Thus, two subsets of the nonphotosynthetic branches were tested: (1) Orobanche , Epifagus and Cistanche branches and (2) Harveya , Hyobanche and Boschniakia branches. Each of these tests used the data set from a single gene and compared two nested hypotheses: H 1 : the photosynthetic and non-photosynthetic branches share a single value (for one of the parameters d N , d S or ω). H 2 : the photosynthetic and non-photosynthetic branches have two separate values. If the tree has a significantly higher likelihood under H 2 , that is taken as evidence that the nonphotosynthetic branches have higher rates. Scatter plots and correlation tests were used to examine the degree of correlation between synonymous and nonsynonymous sites within a gene, and also to see if either class of sites was correlated between genes. Authors' contributions NDY and CWD conceived of and designed the study together. NDY did the sequencing, data analyses and drafted the manuscript. CWD provided the genomic DNA samples and provided the conducive laboratory environment, both physical and intellectual, as well as many suggestions for the manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554776.xml |
549041 | Polymorphisms of the XRCC1, XRCC3, & XPD genes, and colorectal cancer risk: a case-control study in Taiwan | Background Recent studies relating to the association between DNA repair-gene polymorphisms and colorectal cancer risk would, to the best of our knowledge, appear to be very limited. This study was designed to examine the polymorphisms associated with three DNA repair genes, namely: XRCC1 Arg399Gln, XRCC3 Thr241Met and XPD Lys751Gln, and investigate their role as susceptibility markers for colorectal cancer. Methods We conducted a case-control study including 727 cases of cancer and 736 hospital-based age- and sex-matched healthy controls to examine the role of genetic polymorphisms of three DNA-repair genes ( XRCC1 , XRCC3 and XPD ) in the context of colorectal cancer risk for the Taiwanese population. Genomic DNA isolated from 10 ml whole blood was used to genotype XRCC1 Arg399Gln, XRCC3 Thr241Met and XPD Lys751Gln by means of polymerase chain reaction (PCR) and restriction fragment length polymorphism (RFLP) analysis. Results The risk for colorectal cancer did not appear to differ significantly amongst individuals featuring the XRCC1 399Arg/Arg genotype (OR = 1.18; 95% CI, 0.96–1.45), the XRCC3 241Thr/Thr genotype (OR = 1.25; 95% CI, 0.88–1.79) or the XPD 751Gln allele (OR = 1.20; 95% CI, 0.90–1.61), although individuals featuring a greater number of risk genotypes (genotype with OR greater than 1) did experience a higher risk for colorectal cancer when compared to those who didn't feature any risk genotypes (Trend test P = 0.03). Compared with those individuals who didn't express any putative risk genotypes, individuals featuring all of the putative risk genotypes did experience a significantly greater cancer risk (OR = 2.43, 95% CI = 1.21–4.90), particularly for individuals suffering tumors located in the rectum (OR = 3.18, 95% CI = 1.29–7.82) and diagnosed prior to the age of 60 years (OR = 4.90, 95% CI = 1.72–14.0). Conclusions Our results suggest that DNA-repair pathways may simultaneously modulate the risk of colorectal cancer for the Taiwanese population, and, particularly for rectal cancer and younger patients. | Background Humans are routinely exposed to mutagenic and carcinogenic aromatic amines via smoking, well-cooked food and other sources [ 1 ]. These chemicals can form DNA adducts in vivo and thus lead to DNA damage [ 2 ]. The integrity of most of the so-damaged DNAs is typically restored as a consequence of the action of certain DNA-repairing enzymes, the normal function of which is important for maintaining genomic integrity and preventing cellular neoplastic transformation [ 3 ]. Since the genetic polymorphisms of DNA-repair enzymes might be able to influence DNA adduct levels [ 4 - 6 ], the particular degree of DNA-repair capacity has often been associated with the risk of human cancers [ 7 - 11 ]. Amongst the known genetic polymorphisms of the DNA-repair genes [ 12 ], the xeroderma pigmentosum group D (XPD, also known as ERCC2) and x-ray repair cross-complementing groups 1 and 3 (XRCC1 and XRCC3) have been studied most commonly [ 13 ]. The XPD gene encodes a helicase that is a component of the transcription factor TFIIH [ 14 ], this factor being an essential member of the nucleotide-excision repair (NER) pathway that is responsible for effecting repairs to bulky adducts and UV-induced DNA damage [ 15 ]. In 2002, Qiao et al. [ 16 ] reported that individuals featuring XPD 751Gln/Gln did demonstrate suboptimal DNA-repair capacity (DRC) in regard to its ability to remove UV photoproducts when compared to the XPD 751Lys/Lys and Lys/Gln genotypes. The XRCC1 protein is a scaffolding protein directly associated with polymerase beta, DNA ligase III and poly (ADP-ribose) polymerase (PARP) and functions in a complex to facilitate the base-excision repair (BER) and single-strand break-repair processes [ 17 - 19 ]. In 1999, Lunn et al. noted that individuals harboring the XRCC1 399Gln allele were associated, more significantly, with higher levels of both aflatoxin B1-DNA adducts and glycophorin A variants when compared to individuals who exhibited the Arg/Arg genotype [ 4 ]. XRCC3 participates in DNA double-strand break repair and is a member of an emerging family of Rad-51-related proteins that likely participate in homologous recombinational repair (HRR) in order to maintain chromosome stability [ 20 ]. To the best of our knowledge, studies pertaining to these DNA-repair genes focusing on colorectal cancer risk would appear to be limited and controversial. In 2000, Abdel-Rahman et al. [ 21 ] observed that the XRCC1 399Gln allele, compared to the XRCC1 399Arg/Arg genotype, was associated with an increased risk for developing colorectal cancer, especially amongst young urban residents, although in 2003, Mort et al. [ 22 ] failed to reveal any significant associations between colorectal cancer risk and any polymorphisms of four of the NER genes ( XPD , XPF , XPG , ERCC1 ) or XRCC1 . In the present paper, we conducted a hospital-based case-controlled study to examine the role of genetic polymorphisms of three DNA-repair genes ( XRCC1 , XRCC3 and XPD ) in the context of colorectal cancer risk for the Taiwanese population. Methods Subjects Detailed descriptions of the specific characteristics of the study participants have been published previously [ 23 ]. In brief, participants were recruited from the Chang Gung Memorial Hospital between January 1995 and January 1999 inclusively. The colorectal adenocarcinoma cancer cases ( n = 776) participating in this study were newly diagnosed and histologically confirmed. Patients suffering from familial adenomatous polyposis, hereditary nonpolyposis colorectal cancer, or inflammatory bowel disease and other related malignancies were excluded from study participation. Seven hundred and twenty-seven of the original 776 cases (94%) were finally included in this study. Seven hundred and forty-seven age (same age) and sex-matched controls were recruited from the Physical Check-Up Department during the same period. All the participating controls had received comprehensive health examinations including colonoscopies. After excluding individuals diagnosed with other colorectal diseases, a history of other cancers or the existence of a family history of colorectal cancer, 736 controls (98%, 736/747) were finally included in this study. With informed consent, the socio-demographic characteristics of study participants were ascertained by means of the application of a structured questionnaire, at which time 10 ml of venous blood was collected. Genotyping Genomic DNA isolated from 10 ml whole blood was used to genotype XRCC1 Arg399Gln, XRCC3 Thr241Met and XPD Lys751Gln by means of polymerase chain reaction (PCR) and restriction fragment length polymorphism (RFLP) analysis. All of the PCR reactions were carried out by a Mastercycler gradient thermocycler (Eppendorf, Hamburg, Germany) in a final volume of 25 ul containing 200 ng of each primer, 50 ng genomic DNA, 1.5 mM MgCl 2 , 200 ul dNTPs and 1.0 unit of Taq DNA Polymerase in the buffer provided by the manufacturer. In addition, all laboratory genotyping personnel were blind to the case-control status of the samples. The 615 bp XRCC1 PCR products were amplified with the primers 5'-TTGTGCTTTCTCTGTGTCCA-3' (sense) and 5'-TCCTCCAGCCTTTTCTGATA-3' (antisense) and digested with Msp I (New England BioLabs, Beverly, MA, USA). The Arg allele revealed 374 and 221 bp fragments following digestion and polyacrylamide gel electrophoresis, whilst the Gln allele was not digested by Msp I [ 4 ]. The 136 bp XRCC3 PCR products were amplified with the primers 5'-GCCTGGTGGTCATCGACTC-3' (sense) and 5'-ACAGGGCTCTGGAAGGCACTGCTCAGCTCACGCACC-3' (antisense) and digested with Nco I (New England BioLabs). The Met allele revealed 97 and 39 bp fragments following digestion and polyacrylamide gel electrophoresis, while the Thr allele was not digested by Nco I [ 24 ]. The 734 bp XPD PCR products were amplified with the primers 5'-CCTCTCCCTTTCCTCTGTTC-3' (sense) and 5'-CAGGTGAGGGGGACATCT-3' (antisense) and digested with Pst I (Takara, Japan). The Gln allele revealed 646 and 88 bp fragments following digestion and polyacrylamide gel electrophoresis, whilst the Lys allele was not able to be digested by Pst I [ 25 ]. Statistical analysis Tests for Hardy-Weinberg equilibrium amongst controls were conducted using observed genotype frequencies and a chi-square test featuring one degree of freedom. Baseline sociodemographic characteristics between cases and controls were analyzed using the chi-square test and two-sample Students' t -test. Multivariate unconditional logistic regressions were used to examine the association between the XRCC1 , XRCC3 and XPD genotypes and the risk for colorectal cancer. Since the differences in the estimated risks between conditional logistical regression and unconditional logistical regression were small, unconditional logistical regression was used to estimate odds ratio (OR) and 95% confidence interval (CI) with the matching factors (age and gender) included in the model for estimation [ 26 ]. Other potential confounders (such as physical activity, cigarette smoking, alcohol use, coffee intake, and consumption of staple, meat, vegetable/fruit and fish/shrimp) would be also included if they have >10% effect on the gene main effects. However, none of these factors have met the inclusion criteria. Based on the multiplicative scale, the likelihood ratio test was further used to evaluate the interaction between XRCC1 , XRCC3 and XPD genes on the risk for the colorectal cancer. Since the associations between polymorphisms of DNA-repair gene, cancer susceptibility and DNA repair capacity are inconsistent [ 13 ], we defined the risk allele as the allele with OR>1 observed in the present study. Stratified analyses were also conducted to evaluate the differences between specific tumor sites (colon and rectum) and age groups (< = 60 years old and > 60 years old). All analyses were performed using the SAS statistical package (version 8.1 for windows; SAS Institute, Inc., Cary, NC, USA) and all tests were two-sided. Results More men (56%) than women participated in this study (Table 1 ). The mean age for both groups was 60 years. From the cancer cases, 352 patients suffered from colon cancer (48%) and 375 patients from rectal cancer (52%), with most such cases being deemed to be at stage II or stage III (both were 33%). Detailed analyses of the sociodemographic characteristics and potential risk factors associated with colorectal cancer amongst the study population have been published previously [ 23 ]. The genotypic distributions of the three DNA-repair genes for both cancer cases and controls are shown in Table 2 . The frequencies for the XRCC1 399Gln, XRCC3 241Met and XPD 751Gln allele amongst the controls were, respectively, 0.27, 0.05 and 0.07, these genotype frequencies being in Hardy-Weinberg equilibrium. Due to the relatively low frequencies of variant alleles for these genes in the present study, any genotype featuring one or more variant alleles was combined for further analyses. The risk for colorectal cancer was not significantly different for individuals featuring the XRCC1 399Arg/Arg genotype (OR = 1.18; 95% CI, 0.96–1.45), the XRCC3 241Thr/Thr genotype (OR = 1.25; 95% CI, 0.88–1.79) or the XPD 751Gln allele (OR = 1.20; 95% CI, 0.90–1.61). As revealed in Table 3 , those individuals exhibiting a greater number of risk genotypes (genotype with OR greater than 1) faced a greater risk for colorectal cancer when compared to those individuals who did not display any risk genotypes (Trend test, P = 0.03). Subjects who demonstrated two or three of the putative risk genotypes did reveal a significantly greater risk for colorectal cancer (OR = 1.95; 95% CI, 1.08–3.52 and OR = 2.43; 95% CI, 1.21–4.90, respectively) as compared to those individuals who did not feature any putative risk genotypes, although it appears that no gene-gene interactions arose amongst these three genes (all P levels for interaction were >0.21). When stratified by tumor site and age at diagnosis, these combined gene effects upon cancer risk were observed for individuals who revealed that their tumor was located in the rectum (Trend test P = 0.03) and those individuals for whom their tumor was diagnosed prior to their being 60 years of age (Trend test P = 0.004; Table 4 ). The ORs for subjects with three putative risk genotypes were 3.18 (95% CI, 1.29–7.82) for rectal cancer and 4.90 (95% CI, 1.72–14.0) for those individuals diagnosed prior to 60 years of age, respectively. Discussion To the best of our knowledge, few studies have investigated the role of polymorphisms in DNA-repair genes for patients suffering colorectal cancer [ 21 , 22 ]. In this hospital-based case-control study of colorectal cancer-suffering patients in Taiwan, we found polymorphisms in three DNA-repair genes associated with an elevated risk of colorectal cancer. In addition, a gene-dosage effect was found for rectal cancer and younger patients. These findings suggest that those genes involved in different DNA-repair pathways may act simultaneously in the process of carcinogenesis for colorectal cancer. Although we didn't find any significant independent associations between these DNA-repair genes and colorectal cancer risk, the risk appeared to be slightly increased for individuals who featured the XRCC1 399Arg/Arg, XRCC3 241Thr/Thr genotypes and the XPD 751Gln allele. Our results do not appear to be entirely consistent with the results of some previous reports [ 21 , 22 ], the former group reporting that the XRCC1 399Gln allele significantly increased the risk of colorectal cancer (OR = 3.98, 95% CI = 1.50–10.6). In 2003, Mort et al. [ 22 ] noted that the risk of suffering colorectal cancer was significantly heightened for individuals who featured the XRCC3 241Thr allele (OR = 1.52, 95% CI = 1.04–2.22) and only slightly increased for those individuals who revealed the XRCC1 399Gln and XPD 751Gln alleles. In 2001, Park et al. [ 27 ] reported that advanced colorectal cancer-suffering patients who revealed the XPD 751Gln/Gln genotype featured a poorer (positive) response rate to chemotherapy and also a shorter survival period compared with colorectal cancer-suffering individuals who belonged to either the 751Lys/Lys or the 751Lys/Gln group. The apparent divergence between these studies and ours might be due to one of two reasons. Firstly, it may simply be that ethnic differences in allele frequency for the polymorphism might explain the controversial findings. The frequencies for the XRCC1 399Gln, XRCC3 241Met and XPD 751Gln alleles amongst the healthy controls in this study (0.27, 0.05 and 0.07, respectively) were similar to the results for other studies conducted in Taiwan [ 28 , 29 ] and China [ 30 , 31 ], but appeared to be much lower than those reported in 2003 by Mort et al. [ 22 ] (0.42 for XRCC1 399Gln, 0.45 for XRCC3 241Met and 0.36 for XPD 751Gln) for a British population. On the other hand, the XRCC1 399Gln allele frequency observed in the present study is greater than that observed by Abdel-Rahman et al. in 2000 (0.14 for XRCC1 399Gln) [ 21 ]. Abdel-Rahman et al. also found that urban residents have 9.97-fold increased risk of early-onset colorectal carcinoma than rural residents with the XRCC1 399Gln allele [ 21 ]. Therefore, it is possible that the divergence in results from different studies might be related to different levels of carcinogen exposure for different populations. Moreover, inadequate study design such as a too-small sample size and/or the inadequate controlling for certain confounders (such as age and gender) should also be considered as constituting the underpinning for such differing results. Combined effects of polymorphisms of the XRCC1 Arg399Gln, XRCC3 Thr241Met and XPD Lys751Gln genes in regard to colorectal cancer risk were observed in the present study. With the complexity of detail of environmental exposures to various carcinogens, it is plausible that the effective repair of DNA damaged by chemical mixtures necessitates multiple DNA-repair pathways (including BER, HRR and NER pathways). The failure of, or the presence of deficient DNA repair capacity for each specific DNA-repair pathway may contribute to an increase in cancer risk. Additive or multiplicative effects of combined genetic variants for different DNA-repair pathways have been previously reported for lung cancer [ 32 ], melanoma [ 33 ] and breast cancer [ 34 ]. In 2003, Zhou et al. [ 35 ] found that the risk of lung cancer amongst nonsmokers increased progressively with the increase in the number of high-risk alleles of XRCC1 and XPD genes. Hu et al. [ 36 ] also observed that prolonged cell-cycle delay was significantly associated with the number of variant alleles of the APE1 and XRCC1 genes that were present. It would therefore appear reasonable to hypothesize that genetic polymorphism(s) for DNA-repair genes may simultaneously contribute to colorectal cancer susceptibility. In this study, we found that the combined effect of multiple DNA-repair genes upon colorectal cancer risk was significant for our younger age group, but not so for the older age group. This finding appears to be similar to the results of most of the previous studies pertaining to colorectal cancer [ 21 ], basal-cell carcinoma [ 37 ], head-and-neck cancer [ 38 ], hepatocellular carcinoma [ 39 ], and lung cancer [ 6 , 32 , 35 , 40 ] that we reviewed, although we did note that one study reported that such an elevated risk was also observed for old-aged head-and-neck cancer patients [ 41 ]. In addition, few studies that we reviewed failed to find a difference between age groups in regard to for cancer risk [ 31 , 42 - 46 ]. In 2000, Duell et al. [ 47 ] noted that old healthy subjects who featured the XRCC1 399Gln allele appeared to be significantly associated with detectable DNA adducts in their blood mononuclear cells when compared to younger subjects who featured the 399Arg/Arg genotype. In 2001, Hemminki et al. [ 48 ] also found that old subjects who revealed the XPD 751Gln/Gln genotype exhibited a decreased DNA-repair rate for, specifically, UV-specific cyclobutane pyrimidine dimers in the skin. Therefore, it is possible that individuals who demonstrate a less-efficient DNA-repair capacity might develop tumors at a younger age than individuals who reveal an efficient DNA-repair capacity. In our study, we also found that the combined genetic effect of these three DNA-repair genes in regard to cancer risk was more pronounced for the rectum than for the colon. This difference may reflect certain etiological differences between colon and rectal carcinogenesis. In fact, different epidemiological characteristics, etiology, pathogenesis and clinical behavior have been reported for different anatomical sites of colorectal cancer [ 49 , 50 ], this latter group suggesting that carcinogenesis within the distal colon was associated with bulky-adduct-forming (BAF) agents and that these DNA lesions were repaired through NER pathways. In 2001, Hong et al. [ 51 ] also found that much more DNA-repair and apoptosis activity occurred in the distal rather than the proximal colon for the rat azoxymethane carcinogenesis model. Taken together, these observations support the notion that insufficient DNA-repair capacity could contribute to the risk of cancer associated with exogenous carcinogen exposure. To the best of our knowledge, this study is the first to report on XRCC1 , XRCC3 and XPD polymorphisms in relation to the risk of colorectal cancer for the Taiwanese population. Our results suggest that genetic polymorphisms of the XRCC1 , XRCC3 and XPD genes, particularly in combination, may be associated with an individual's susceptibility to colorectal cancer. Acknowledging the relatively limited sample size in the subgroups for the low allelic frequencies, further studies incorporating a larger sample size and/or another ethnic population are needed to confirm the genetic role of DNA-repair mechanisms as regards colorectal cancer susceptibility. Conclusions Our results suggest that DNA-repair pathways may simultaneously modulate the risk of colorectal cancer for the Taiwanese population, and, particularly for rectal cancer and younger patients. Competing interests The author(s) declare that they have no competing interests. Authors' contributions CCY carried out the genotyping analysis, performed the statistical analysis and drafted the manuscript. FCS participated in the design of the study. RT and CRCC participated in the design of the study and provided clinical biospecimens. LLH conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549041.xml |
554986 | The efficacy of playing a virtual reality game in modulating pain for children with acute burn injuries: A randomized controlled trial [ISRCTN87413556] | Background The management of burn injuries is reported as painful, distressing and a cause of anxiety in children and their parents. Child's and parents' pain and anxiety, often contributes to extended time required for burns management procedures, in particular the process of changing dressings. The traditional method of pharmacologic analgesia is often insufficient to cover the burnt child's pain, and it can have deleterious side effects [ 1 , 2 ]. Intervention with Virtual Reality (VR) games is based on distraction or interruption in the way current thoughts, including pain, are processed by the brain. Research on adults supports the hypothesis that virtual reality has a positive influence on burns pain modulation. Methods This study investigates whether playing a virtual reality game, decreases procedural pain in children aged 5–18 years with acute burn injuries. The paper reports on the findings of a pilot study, a randomised trial, in which seven children acted as their own controls though a series of 11 trials. Outcomes were pain measured using the self-report Faces Scale and findings of interviews with parent/carer and nurses. Results The average pain scores (from the Faces Scale) for pharmacological analgesia only was, 4.1 (SD 2.9), while VR coupled with pharmacological analgesia, the average pain score was 1.3 (SD 1.8) Conclusion The study provides strong evidence supporting VR based games in providing analgesia with minimal side effects and little impact on the physical hospital environment, as well as its reusability and versatility, suggesting another option in the management of children's acute pain. | Background Pain is a noxious stimulus which can be interpreted in many ways by different individuals but as yet the mechanisms by which the body manages it are not completely understood. Pain has been defined by the International Association for the Study of Pain as 'an unpleasant sensory and emotional experience associated with actual or potential tissue damage or described in terms of such damage' [ 3 ]. Though it is frequently related to physical causes (such as inflammatory processes and nociceptive transmission of pain messages), the experience is entirely subjective, making objective measurement of pain difficult. [ 4 - 6 ], The way in which pain is perceived depends on many factors, including past experiences, memory, understanding of pain, cultural conditioning, and pain threshold [ 5 , 7 , 8 ]. Children with burn injuries undergo significant physical and emotional trauma, initially from their injury, and subsequently from the dressing changes and related treatment they undergo throughout the healing phase. The latter is referred to, in this article, as 'procedural pain'. Clinicians involved in the care of children with acute burns use the best available methods to reduce procedural pain. Adequate and appropriate pain management is essential to ensure that symptoms secondary to pain experiences do not become habitual [ 9 ]. Moreover, unrelieved pain can produce serious physiological and psychological consequences leading to an increased risk of morbidity and even mortality [ 10 , 1 ]. Therefore pain experiences can significantly impact on immediate and longer-term quality of life and well- being of young people. The burns service at the Women's and Children's Hospital (Adelaide, South Australia) collaborated with University of South Australia, on this project. Pain management comprises administration of a number of medications, including analgesics , muscle relaxants and hypnotics . These drugs help in reducing procedural pain experienced by children, however they frequently have unwanted side effects such as drowsiness, nausea, reduced postural control and lethargy [ 11 , 8 ]. Pain management of children in hospitals during dressing changes has been reported as inadequate [ 1 , 2 ] and is often described by children to be the most distressing part of the hospitalisation [ 12 , 13 ]. Procedural pain, experienced by children with burns is often distressing for health professionals and parents. Therefore an investigation into nonpharmacological strategies of pain relief for children is warranted from the perspectives of improving pain management, decreasing incidences of side effects and distress in children, parents and health professionals involved with burns dressing changes. Virtual Reality (VR) was initially conceived as a tool for pain modulation by Hoffman et al [ 14 - 18 ] who found it to be effective in reducing burns pain in adults [ 15 , 16 ] as well as in other situations to manage pain and phobias [ 14 - 18 ]. In 2003, one of our project teams [ 19 ] reported on a single subject, cross over design pilot study at the Women's and Children's Hospital (WCH), Adelaide for a child during rehabilitation following orthopaedic surgery. This pilot study suggested the possible usefulness of VR to modulate pain in children undergoing burns dressing changes. Why virtual reality? VR can be considered intermediary to reality and computer technology. Owing to its ability to allow the user to immerse and interact with the artificial environment that he/she can visualize, the game-playing experience is very engrossing [ 14 - 18 , 20 - 22 ]. VR games are different to other games as they give the user a perception of actually being in a different environment. Visual, auditory and touch sensations can be modified based on the stimuli. The game used in this study was developed by the Department of Computer and Information Sciences, University of South Australia. A number of criterions had to be taken into account when designing the game, keeping in mind the different characteristics of prospective players (gender, age groups, intellectual capabilities), the amount of violence portrayed, the complexity of the game and being aware of the amount of control and functions given to the child. The game designers had to keep the structure of the game as simple as possible with minimal controls, to minimise the physical movements required to play the game. Methods This project was an interdisciplinary and inter-sectorial collaboration between the Centre of Allied Health Evidence, and the Department of Computer and Information Sciences, both at the University of South Australia (UniSA), Australia, and Women's and Children's Hospital(WCH), Adelaide, South Australia. Ethics Ethical approval for this project was obtained from both the institutions (WCH, Adelaide and UniSA). Study sample All children admitted to one specific ward (Newlands Ward), WCH, aged between 5 and 18 years, having burns to more than three percent of their body surface area, and requiring dressing changes, were eligible for inclusion in the study. Children with burns to their hands, face or head, past history of epilepsy and reduced intellectual capacity were not included, as they would have been unable to appropriately use the VR equipment. Informed consent All eligible subjects were identified by ward staff, and were invited to participate in the project by the project team. Written child and parent consent was obtained at every contact. Interventions The test administrations of routine pharmacological analgesia or routine pharmacological analgesia coupled with virtual reality were randomly assigned to each half of the burns dressing change (removal of existing burns dressings or application of fresh dressings) following a coin toss determining the sequence. The child and parents were given a standard explanation about the VR administration and the VR game. If required, subjects were allowed a short preview to assist them to understand how to play the game. VR equipment • The VR equipment constituted a laptop (Dell Inspiron 5100, Pentium 4 2.4 Ghz CPU with a Radeon Mobility 7500 Video Card) with the game software, developed by the Department of Computer and Information Sciences, UniSA (Based on the game 'Quake' by ID Software), a head-mount display (HMD) (IOGlasses Head Mount Display with a SVGA video resolution of 800 × 600 16 million colours), with a tracking system (Intersense IS300 6 degree of freedom Inertia Cube with a USB-Serial converter, required for Inertia Cube), to allow interaction with the virtual environment by moving the head and neck and decoder and a mouse used as a trigger. The game involved a visual simulation giving the children a feel of being on a track, using a pointer to aim and shoot monsters. Figure 1 illustrates the use of the equipment, and scene from the game. Figure 1 a) Child using the VR equipment, b) Mechanics of the equipment and c) a scene from the game she is playing • The developers considered the applicability of the game through varying age groups, gender, intelligence and intellectual capacities, while designing the game. The game tried to achieve effective distraction via immersion without violence and a simplified game structure requiring minimal control by the player, to allow the smallest possible movement during the dressing change procedure. Administration procedure A within-subject design was used in which the children acted as their own controls. There was no interference with the dosage or type of analgesia which was administered to the children 30 to 45 minutes prior to dressing removal. The burns dressing changes would normally occur every Wednesday morning before the hospital ward, grand rounds. The dressing change involved administration of prescribed medication and application of olive oil on the dressing (if adhesive tape was covering the wound), approximately 30 to 45 minutes prior to the actual procedure. The first treatment half constituted the removal of the adhesive tape/bandages and the under-dressing (acticoat/silver oxide dressing) and the second half comprised of the wound being debrided and a fresh dressing applied, after being assessed by the consultant medical officer/s. Data collection Following the completion of each half of the dressing change (with or without administration of VR), the researcher obtained scores for average pain using the Face Scale (Figure 2 ), and interviewed the child, mother and the nursing staff regarding their perceptions of the procedure Using standardised questionnaires (Appendix II and Appendix III). One researcher only was involved in data collection, and intra-rater reliability was maintained by using standard protocols for introductions, explanations, VR administrations and data collection procedures. Figure 2 Pain rating scale used by the children [23]. Outcome measures The subjects were asked to score their average pain experience at the end of each phase of the dressing change procedure (VR and pharmacological analgesics, and pharmacological analgesics only). Pain was scored using a modified self-report Faces pain scale [ 23 ]. The scale depicts increasing levels of pain and is offered in combination with a visual analogue scale of 0 – 10, associated with each picture representing a level of pain. Parents/carers and nurses were also interviewed by the data collector at these times, using open ended questions to obtain views regarding the child's anxiety and perception of pain, and utility of VR in a clinical setting. Data analysis The data was analysed by a blinded assessor, to reduce any biases and increase the rigour with which a de-identified and coded dataset was probed. Results Subjects There were nine eligible, consenting child subjects (6 boys and 3 girls) in the sample. The average age for both boys and girls was 10.0 years (SD 3.7 and 4.1 respectively), age ranging between 5 to 16 years. The average percent of body surface area burnt was 5.3% (SD 3.4%) and there was no significant gender or age difference in body area burnt. For boys, two had burns from contact with a silencer (muffler) on a four wheeled motor bike, two were burnt from a hot water bag bursting, one was burnt while playing with petrol and fire, and the remaining boy was burnt with hot oil from a BBQ. For the girls, all three were burnt by overturned fluids. All participants were experiencing burns for the first time, and when enrolled into the study, their burns were at either second or third dressing change. Every participants' pain, prior to enrolling in this study, had been managed either with no pain relief, or by pharmacological means. For every child participant, one parent or guardian with one exception (was not available to observe the dressing change) provided data on the effectiveness of the VR for every post-session interview. One key nurse involved in the burns management was also interviewed following each session. Trials Overall, 13 trials were undertaken from nine children (one subject participating in three trials, two subjects in two trials, and the remainder in one trial each). The results of two subjects were withdrawn for further analysis as the respective participants were too drowsy from the effects of analgesia to participate appropriately in the VR section of the session. Thus the remaining seven child subjects were included for analysis, with a total of 11 useable trials (an average of 1.6 trials per subject). The seven participants in the included trials had an average age of 11.1 years (SD 3.5). Time factor There was no significant difference ( p < 0.05 ) in time taken in the two treatment halves (removal and application of fresh dressing). The average difference in administering the two treatment halves was approximately 2 minutes (Figure 3 ). Figure 3 Per trial comparison of time taken to complete each treatment half Pain change With pharmacological analgesia only, the mean pain score (using the Faces Scale), over all included trials was 4.1 (SD 2.9), whilst for VR coupled with pharmacological analgesia, the average pain score was 1.3 (SD 1.8). Because of the small number of child subjects in the study, the data was considered per child, and per trial. Over all included trials, the mean pain score difference between administrations was 3.2 (SD 2.1), which was significant using paired t-tests (p < 0.01). This indicated the importance of the effect of using VR (coupled with analgesia) in reducing pain experiences during burns dressing changes. The per trial pain responses to VR and analgesia, and analgesia alone, compared with the average trial response per administration is shown in Figure 4 . Figure 4 Per trial differences in pain scores compared with average administration scores For each child subject who completed an eligible trial, the average per-child difference in pain scores between administrations of VR & Pharmacological Analgesia, or Pharmacological Analgesia alone, suggested that every child but one obtained an improvement in pain scoring of at least 2 points on the Faces Scale, attributable to VR, as demonstrated in Figure 5 . Figure 5 Per child differences in pain response attributable to VR Comments made by nurses, parent/s and child subject All nurses or parents agreed that VR helped distract the children and was helpful in reducing pain and there were no negative comments regarding the application of VR. (note - delete 'interestingly') Nurses' responses The overwhelming response from the nursing staff was that VR administration was helpful to the child. Comments from the interviews are provided below as evidence of this. "... probably VR helped to take concentration off ... probably helped take away a lot of the anticipation away from the treatment". "... from my past experience, I can tell that it (changing burns dressing) can be a real problem. It was not a problem today. He did not even flinch while the dressing was being taken off." "... communication was good – he understood what I asked him to do. I did not find it invasive or intrusive..." "... it was great to do the changing (of dressing) without dumping him with medication." "... no (communication was not effected) ... he responded well to the requests and commands." "yes (pain was significantly less)... he had lot more pain with cleaning (of the wound) than taking the dressing off... He seemed to cope better with VR than without." "... cleaning the dressing in the bathroom made him more anxious, when he saw the wound – felt more pain." "Yes, he was more anxious when VR was not on." "... he felt worse when he was looking at it (the wound) compared to when he was not." "... was more relaxed and concentrated on the game. You could tell that he could feel the pain, but focussed on the game." The perception of the assisting nurses were the children were more cooperative and distracted from the administration of VR, which helped reduce the difficulty in changing the burns dressing compared to when routine analgesia was used by itself. There appeared to be no problem in physically using VR within the environmental constraints of the burns dressing area, and in no instance did VR impede communication with the child. Parents' responses All parents agreed with the positive effects of VR in pain management for their child. They all commented that the child's anxiety level was perceptibly less when using VR, and the child looked forward to playing the VR game. Comments below from the parent interviews support the positive VR effects: "... was a lot calmer and enjoyed the VR." "...absolutely, she did not remember about the last dressing that was taken off. She had to be prompted, to remind her of the game and change of dressing." "... much happier than usual. He reckoned he felt it but I think he did not. He did not show any of the same signs." "...she was not as anxious. Was afraid before hand, but she was all right after the game was switched on. It took her mind, off the pain." "...was smiling while playing the game." "... compared to the medication which left him groggy, disoriented, lost track of time and anxious, I think this (VR) allows the continuity of time and reduces anxiety." "Yes (pain was significantly less) ... probably judging it from yesterday – medication made him worse – uncooperative and pig-headed, compared to when he was playing the game." "Yesterday he was whinging thinking about the dressing change, this morning, when I told him that you were coming; he had a grin on his face..." Comments from child participants on the VR game However, the current game appeared to have a reasonable level of complexity and engaged the participating children of different age groups. The above comments were randomly selected from the questionnaire deployed at the completion of each trial (to interview the nurses, parents and children participating in the study). Discussion This is the first published randomised clinical trial to our knowledge reporting the use of VR for children with burns. It concurs with the findings of Hoffman et al [ 15 , 16 ] who tested VR on adults with burns, and suggests that VR could provide a significant improvement in the pain management for all children undergoing treatment for this condition. We found that VR coupled with analgesics was significantly more effective in reducing pain responses in children than analgesics only. Although there were 3 occasions where the child, equally scored both treatment halves, the respective carer and nursing staff member, consistently indicated that the child's behaviour was less distressed and calmer during the treatment when VR was applied, suggesting that VR made it less distressing for the child. Thus the feedback given by nursing staff and parents provided additional and important information in interpreting the Faces Scale responses given by the children. Distraction by an interactive game was the putative influence in reducing sensitivity to pain. Clinical implications Given that the application of VR as a method for pain control in the clinical setting is very simple, the results of this study are encouraging with respect to future use. The present prototype VR game being used appears to be cumbersome due to the number of wires attaching to the laptop and the HMD, but this could be simplified considerably so that the equipment required is simply a console with a trigger and a head mount. Applying this equipment would be as simple as providing medication prior to the dressing change procedure. The equipment is reusable and requires minimal technical knowledge for use. Provided a number of different games were available to cater for different age groups, it could be widely applied, and will allow children to relocate themselves to 'another world' during dressing changes, decreasing their attention to painful stimuli. It was noted by nurses during several trials that communication was never a problem; they were able to instruct the child to change, or assist change in position without any difficulty in a compliant and relatively pain-free manner. On the other hand, children without VR were often distressed and crying in pain, decreasing their ability to listen and cooperate. Limitations A number of factors resulted in a small sample size. There were relatively fewer children with burns than anticipated, who fitted the inclusion – exclusion criteria, and who were admitted to the Women's and Children's Hospital, Adelaide, during the data collection period. Several potentially eligible children (2 trials) had such severe side effects from the medication (particularly drowsiness) that they were not able to participate, and two refused to participate. The reduced sample size limits ability to generalise the results, and a study with a larger sample size may provide better understanding of the usefulness of VR as a treatment adjunct for pain relief. Another potential limitation is that some children were tested more than once – in the first instance immediately after their burn and then subsequently after surgery or during another dressing change later in the healing phase. In these cases, there may have been a learning effect which modified the pain scores, or simply an overall decrease in pain due to healing. It was also noted from the feedback that older children found the game too simple and therefore not as absorbing or distracting as the younger children found it. Finally, waterproofing of the VR equipment would allow data collectors to examine pain responses through an entire dressing change including when subjects had their wound debridement carried out in the bathroom. Conclusion There appears to be considerable scope for further research into the potential for using VR in the clinical setting. Larger trials could be conducted, using games appropriate for the varying age groups. The next stage would be to test VR alone against pharmacological pain relief, to investigate whether VR is as effective in isolation, and could decrease use of analgesia, thus avoiding the side effects associated with medication. Another avenue of future research would be to investigate the exact mechanisms by which VR assists pain modulation. It is hypothesized that it works by distracting a child's attention from painful stimuli, which in turn reduces the perceived intensity of pain. Competing interests The author(s) declare that they have no competing interests Authors' contributions DAD was involved with coordination and recruitment of trial subjects, application of the trial, collection of data and writing up the paper. KAG was involved with statistical analysis of the data and supervising the scientific conduction of trials. ALS was a key medical consultant for the trial subjects and coordinated efforts within the hospital to recruit trial subjects. SEM was the first contact with the prospective trial subjects. She identified and provided them with initial information and got verbal consent from them. She also acted as a coordinator for all the nursing staff members on the ward participating in the trial. BAT provided IT support for building and helped in maintaining the inventory (VR equipment) and the game software. Pre-publication history The pre-publication history for this paper can be accessed here: Supplementary Material Additional File 1 Pain scale scoring Click here for file Additional File 2 Caregiver interview Click here for file Additional File 3 Nurse interview Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554986.xml |
549069 | Using ontologies to describe mouse phenotypes | By combining ontologies from different sources the authors developed a novel approach to describing phenotypes of mutant mice in a standard, structured manner. | Background Mutant mice are the premier genetic models for human diseases. An increasing number of laboratories and companies worldwide are now carrying out detailed analyses of mouse phenotypes that have been generated from large-scale mutagenesis of the mouse genome. Description of mouse phenotypes has not traditionally adhered to predefined rules or been recorded in databases. However, the sheer volume of data from high-throughput screens (such as N -ethyl- N -nitrosourea (ENU) mutagenesis [ 1 ]) is now driving the need to manage information about mutants in a paperless environment and to build databases that will allow this data to be shared between laboratories and used to formulate hypotheses about gene function. The key to satisfying this need is the ability to describe different phenotypes in a consistent and structured way. There is a need for consistency in the way different communities of biologists attempt to present this kind of data since consistent representation of phenotypes across different domains (such as pathology and anatomy) and species is crucial for the semantic interpretation and the efficient use of this complex information in different kinds of study, such as comparison of gene functions between species. Ontologies have been an important tool for structuring biological information since the time of Linnaeus. With the advent of the Gene Ontology (GO) in 2000 [ 2 ] these techniques for strictly specifying the semantic relationships between terms have become a standard to support knowledge representation in the field of genomics. Hierarchical ontologies hold information about the structure of a particular domain of knowledge at varying degrees of detail (granularity), thus permitting us to integrate concepts and descriptions at different levels of resolution. This approach is forming the basis of new methods for mining biological data [ 3 , 4 ]. In this article, we describe developments in describing mouse phenotypes using ontologies. Ontologies and knowledge bases The term ontology is derived from the Greek and is used in philosophy to mean 'a description of what exists'. There are many definitions of the word, however, and for the purpose of this article, an ontology is 'a specification of entities and their relationships' [ 5 ]. The key word 'specification' implies a formal organization. Thus, an ontology is a formalism to describe entities and the relationships between them. Ontologies for computing applications are schemas for metadata [ 6 ]. They provide a controlled organization of terms and their relationships that has explicitly defined and machine-processable semantics [ 7 ]. The controlled semantic portrayal of entities and their relationships allows the description of a domain of knowledge. For our purposes ontologies mainly attempt to replace free-text descriptions of phenotypes with equivalent computable descriptions that can be used to draw inferences about these data. An ontology together with a set of individual instances of the kinds of entities it specifies constitutes a knowledge base [ 8 ]. It may be difficult to distinguish between the knowledge contained in an ontology and the knowledge contained in a knowledge base [ 9 ]. In phenotype ontologies the distinction between the ontology and the knowledge base must be clear. The ontology should capture the general conceptual structures necessary to describe the domain, whereas the knowledge base should provide the individual instances that are described using the ontology. So, in the ontology one can first define the entity (class) of 'pain perception' and further, assign to this entity the attribute 'relative sensitivity' and specify for this attribute a range of allowed values using concepts such as 'sensitive' or 'insensitive', and so on, thereby allowing us to describe pain-perception phenotypes. The knowledge base, however, holds data about particular instances [ 10 ], for example a particular mouse with a particular genotype, under defined handling conditions and of a certain age, that has a particular level of sensitivity to pain according to a particular assay. In other words, the ontology constitutes a general theory (how to describe phenotypes), whereas the knowledge base describes particular circumstances, in our case particular instances of phenotype. Why use ontologies? An important question here is why do we need to use ontologies; why not simply use a series of unconnected, standard terms such as provided by a controlled vocabulary? The advantages of using ontologies have been argued extensively, but the main reason is that ontologies are attempting to capture the precise meaning of terms. Furthermore, ontologies can be used for reasoning and inference (for example, consistency checking or drawing conclusions from the knowledge). The most important factor from our perspective is the need to combine information from different phenotypes or from different protocols (assays). For example, if a mutant mouse has six digits in each forelimb we will wish to use this information in a variety of ways (for example, to group mice with limb pattern defects, or with affected forelimbs, or with abnormal numbers of digits in any limb). For this, we need not just a controlled vocabulary of terms, but also information about how these terms relate to one another (for example, that forelimb is an instance of 'limb', that the normal number of digits in the forelimb is five, that the number of digits is an instance of 'pattern', and so on). Current approaches to the description of mouse phenotypes Traditionally, the main source of information for most scientists is the peer-reviewed journal literature. Electronic versions of published information have opened the road to accessing and retrieving information in a much easier and more cost-effective manner. The growth and wider availability of the world-wide web has led to a significant growth in the amount of readily available electronically stored information [ 11 ]. With this surge of readily available information the location and retrieval of relevant information has become a major (commercial) activity [ 12 ]. One of the most important issues in information retrieval is constructing effective indexing methods that are required for the sophisticated querying of the stored data. Free-text searching forms the basis of information retrieval but is extremely limited because of the inherent lack of accuracy and specificity. Complex free-text descriptions, such as are used for phenotypes, are almost impossible to index and retrieve in a useful way directly from the biomedical literature. The potential power of complex searches against information from multiple experiments requires the annotation of free text into structured representations that can be understood and where the power of computational algorithms can maximize the potential of the information to be compared and contrasted. The most comprehensive attempt to annotate mammalian phenotypic data so far, the Mammalian Phenotype Ontology (MP) [ 13 ], is currently under development by the Jackson Laboratory [ 14 , 15 ]. The current structure of the ontology is generated using DAG-Edit [ 16 ], the current GO standard, and allows a hierarchical display of terms and their definitions. These terms include a combination of entities and values, for example, id MP:0001509 corresponds to 'abnormal body position', which at a high level provides a sufficient description of phenotypic data. This approach allows high-level access to the knowledge held in the ontology, but also has certain limitations similar to the GO paradigm. If one attempts to create too much specificity within an ontology of this type it can expand to unmanageable proportions and parentage relationships can be overlooked as their number grows. For example, merely creating new terms by prepending the two qualifiers, 'increased' and 'decreased', everywhere that is applicable, will massively increase the size of the ontology. To allow a systematic approach to the model, combinations would also have to be instantiated that might never be used. Because there is a practical limit to the number of values that can be managed, such an approach is limited. Inevitably, decisions have to be made as to which individual combination describes a particular phenotypic entity best. We note here though that the development of MP is being developed pragmatically, with instances being added as needed to annotate mouse phenotypes, following the paradigm used by GO developers. MP is a cross-product ontology that includes mouse anatomy ontology, GO and other controlled terms as part of the construction of MP terms. Although the cross-reference IDs are not visible, they are part of the design of MP. Some of the work described here reflects insights gained during extensive discussions about the representation of phenotypes at the Phenotype Consortium meeting held in Bar Harbor, ME in September 2003. The developers of the MP ontology are part of this consortium and have intentionally created their ontology in such a way that it can be easily extended to form instances of the compositional approach discussed in the next section. With the objective of capturing information about phenotypes in any organism, Ashburner proposed the Phenotype And Trait Ontology (PATO) [ 17 ] in 2002. PATO is a schema according to which, "phenotypic data can be represented as qualifications of descriptive nouns or nounal phrases" (M. Ashburner, unpublished work). Each noun represents an observable characteristic and for each noun there will be a set of attributes, for each of which is defined a set of appropriate values. In addition to these three semantic classes (namely observable entities together with the associated attributes and values), the concepts that are needed to describe phenotypes include the assays by means of which the phenotypes were determined and the environmental and genetic conditions (Microarray Gene Expression Data Society [ 18 ]) under which these assays were performed. Taken together, the semantic concepts and relationships defined for PATO, assays, genetic and environmental conditions, will form the basis for the systematic description of phenotypes. Results A proposal for describing mouse phenotypes The description of mutant phenotypes must provide a practical way to capture the biologically relevant information about the phenotype in machine-readable form [ 19 ]. It should allow us to compare, combine and analyze different phenotypes. For this, the ontology must first be consistent, and second be able to generate statements that have a logically well-formed structure in order to support reasoning from descriptions of different phenotypes. To provide these functionalities we propose a compositional method of describing phenotypes [ 19 ]. By this we mean that the description of the phenotype combines terms from different standard ontologies, each of which supports a particular domain of knowledge. A list of ontologies that should be included in such a phenotype ontology is given in Table 1 . These ontologies are combined in a specified formula or schema that provides the logical structure of the whole. The schema itself can be considered as a meta-ontology that describes how other ontologies relate to one another. Figure 1 illustrates such a schema. According to the schema in Figure 1 , the whole organism has certain attributes, such as genotype, identity number, and exists under certain handling conditions (Table 2 ). The organism also has a set of core components including its anatomy, development, physiology and behavior. Each of these core components is represented by a separate ontology and each has a set of attributes, again represented by an ontology. For example, the organism may have an anatomical component 'left eye' which is a term from the anatomy ontology. The left eye, in turn, may have attributes of 'color', 'size', and so on, taken from the attributes ontology. This combination of core entity and attribute constitutes a phenotypic character - something that can be measured. Phenotypic characters, in turn, link to 'assays', which return a variety of 'values', again represented by an ontology, which may be applied to the phenotypic character in question. When this schema is used to describe actual phenotypes, instances of single phenotypic characters are linked together to provide a full phenotypic description of an individual organism. Each character can be represented by a line in a table where the table represents the full phenotype. Figure 2 presents this schematically. According to the schema in Figure 1 , five classes of ontology (in circles), namely organism, entity, attribute, assay and value, are required to express a phenotypic instance. Organism This class holds the information (organism attributes) of an organism in which the phenotypic characters are observed (see Table 2 ). Entity Entities will be formed by importing ontologies discussed in Table 1 : behavior, anatomy, and so on. Each entity may be associated with a set of attributes, for example, color and size, that may also be shared with other entities. Attribute Attributes will be provided by PATO [ 17 ]. PATO should hold general attributes that can be applied through different phenotypic ontologies. This has the advantage of economy and also enables cross-referencing between domains. New attributes should be assigned to classes only when they cannot be modeled with existing options. Assay Assays will have a hierarchical structure and will define a range of values that correspond to a particular combination of entity and attribute (that is, phenotypic character). They hold multiple relations to values, qualifiers and free text as well as their own metadata. The slot for free text is included to capture knowledge that cannot be expressed through the ontology as yet. Values Splitting PATO into two different ontologies, PATO attributes (above) and PATO values, allows the PATO ontology to be incorporated into the schema [ 19 ]. Values can thus be either specific values provided by the assay or common values, provided by PATO. A possible relationship between these sets of values would be 'interpretation_of'. Although values provided directly by the assay are usually the objective recordings of a test for a specific phenotypic character, there can be an interpretation of these recordings in terms of a higher level phenotypic character. For example, in an assay of memory in the mouse that uses a water test, the values returned by the test may be that a mouse completed the task in a certain time and manner, but these results may be interpreted to indicate a value corresponding to the phenotypic character comprised by the entity 'memory' that was assayed for the attribute of 'short-term recall' and returned the interpretative value 'loss of memory'. By introducing the 'interpretation_of' relationship, we could make this distinction in a machine-understandable manner and allow the possibility, if required, of expressing the original objective values of the test, thus avoiding information loss. This aspect of the schema remains under study. A central idea in this schema is that of the 'phenotypic character', which we can define as any feature of the organism that is observed or 'assayed'. An example for the mouse is tail length. A phenotypic character is a compound composed of an entity, in this case an anatomical entity 'tail', and an attribute of tail, here 'length'. Similarly the physiological entity 'hearing' (GO:0007605) has the attributes 'sensitivity', 'range', and so on. Thus, 'hearing range' and 'hearing sensitivity' are distinct phenotypic characters. The ideal phenotypic character is one that can be measured independently of others. In practice, however, phenotypic characters are rarely independent. Furthermore, the observations from any particular assay will most probably depend on several different phenotypic characters. For example, the results returned by the click-box test for hearing sensitivity in the mouse actually depend, not only on hearing, but also on the mouse's ability to make a detectable locomotor response (the Preyer reflex [ 20 ]). These multiple dependencies are captured in the schema, enabling the ontology to support the appropriate possible groupings of phenotypes. This will allow us, for example, to group all mutants that have (by direct assay), or may have (for example, those failing the click-box test), an effect on the locomotor system. Conversely, different assays may provide information about a single phenotypic character. For example, an acoustic brain-stem response (ABR, a sound-evoked potential within the acoustic nerve) [ 21 ] can be measured to assay basic hearing ability as well as to give a threshold-response curve for differing frequencies. Linking assays with characters in this way will support machine reasoning, enabling us, for example, to make the hypothesis that a particular mouse has a locomotor rather than hearing defect. Indeed, the need to capture this network of relationships between assays and phenotype is a strong indication of the need for an ontology rather than merely a controlled vocabulary of unrelated terms. The expressivity of representation languages such as DAML+OIL [ 22 ], OWL [ 23 ] and OBO [ 17 ] could also dynamically account for the possibility of a cross product or dependence required for representing a phenotype. For example, if a cross product between ontologies does not exist (that is, one of the required terms is not to be found in an ontology), one can assign an 'anonymous class' that is dynamically defined as being both a class in one case and an instance in another. As an example, one might want to refer to the term cocaine dependence, but that cross product may not exist. An 'anonymous class' can be dynamically defined as being both 'cocaine' (coming from a chemical ontology) and 'dependence' (coming from the behavior ontology) to generate this cross product. Finally, we note here that it should be possible to link current high-level structures (such as the current MP ontology), which are necessary in many cases for annotation purposes, to the more expressive form we propose here, so that it can also be explored computationally. Example In this section we describe an example of the application of the compositional schema. We chose a phenotype example at random from the MP database: 'nest building' [MP:0001447]. Several descriptions of nest-building patterns can be found in the corresponding reference [ 24 ]. For example, the authors comment: "Note the fluffy well formed nests built in the +/+ cages and the huddling of mice in these nests, in contrast to the poorly formed nests in -/- cages with random sleeping patterns." and later: "In addition, +/+ mice built nests from nestlet material that averaged 50 mm in depth, while -/- mice built significantly shallower nests (Figure 4D), with depths that averaged less than 20 mm [ t (10) = 3.754, p < 0.004]." The authors also describe the assays used to record these observations: "Nesting Patterns: six cages of wild-type and six cages of mutant mice ( N = 4 mice per cage) were used to evaluate nesting patterns. A 5 × 5 cm piece of cotton nesting material (Ancare, Bellmore, NY) was placed in each cage. After 45 min, photographs were taken of each nest and the nest depth was measured. Nest height data were analyzed using the Student's t test." For some users/applications the compound term 'abnormal nest building' might be a sufficient description of this particular phenotypic instance, but this would result in information loss. A human would have to retrieve and read the reference to extract further information. Our schema allows the expression of this information in a machine and human readable manner. In Table 3 we provide the relevant part of our ontology modeled according to the schema. One can easily express these phenotypic instances. In order to describe fluffy, well formed nests or poorly formed nests one would use the following combination: Nest building {has_attribute} attribute:quality {characterized_by} defined_quality_assay (described in Nesting Patterns [ 24 ]) {returns_value} well-formed Nest building {has_attribute} attribute:quality {characterized_by} defined_quality_assay (described in Nesting Patterns [ 24 ]) {returns_value} poorly-formed Nest building {has_attribute} attribute:quality {characterized_by} defined_quality_assay (described in Nesting Patterns [ 24 ]) {returns_value} fluffy We note here that had the value 'fluffy' not been included in the standard values for a quality assay, it could be captured in the free-text field provided by the schema. To express a nest of 50 mm depth or significantly shallower: Nest building {has_attribute} attribute:absolute_depth {characterized_by} undefined_ absolute_depth_assay {returns_value} 50 mm Nest building {has_attribute} attribute: relative_depth {characterized_by} undefined_ relative_depth_assay {returns_value} shallow {has_qualifier} significant With this information one could go back to a higher level and still be able to express a more general characterization of this phenotype as 'abnormal nest building' but obviously the opposite is not possible. An important unresolved issue concerning the use of ontologies to describe phenotypes arises from the fact that all the ontological structures developed so far are designed to describe individual mice. Mutagenesis experiments usually characterize a number of mutant mice to take into account variable penetrance of the mutation and other stochastic effects. A strategy will therefore need to be developed to describe the generalized phenotypic properties of a cohort of mice. This may involve the use of more sophisticated relations such as {usually characterized by} or even quantitative relations such as {80% characterized by}. Discussion Importance of the assay The assay plays a central role in our schema (Figure 1 ). Assays are the means of making observations and as they determine what can be observed they are a necessary complement to the attribute ontology. Generally, they are recorded as protocols or even as standard operating procedures (SOPs). However, even a visual observation is a form of assay and this needs to be reported when one expresses a phenotypic instance, for example: eye {has_attribute} attribute:color {characterized_by} visual inspection {returned_value} pink On a practical level, assays can add specificity and functionality to the relationship between entities, their attributes and the corresponding values. Most important, an assay vocabulary allows the entire schema to be dynamic by including new assays and capturing explicit differences between assays in different laboratories. The assay will also allow standardization and definition of values for a given phenotypic character, for example, how abnormal is defined in relation to body position. Implementation Our schema can be expressed using a variety of modeling tools and knowledge representation (KR) languages [ 25 ]. We chose DAG-Edit [ 16 ] (version 1.408) and Protégé-2000 [ 26 ] (version 1.9) which is Java-based, well supported and incorporates multiple inheritance, relation hierarchies, meta-classes, constraint axioms and F-Logic [ 27 ]. Although the complexity of our current models can be described with existing tools, in the future more complex phenotype domains may require migration to a finer-grained conceptualization. Populating the Mouse Phenotype Ontology The schema was designed to be easily populated using extant core ontologies, such as anatomy, and defining attributes related to each entity. The assay vocabulary can be constructed as required. Permitted values are defined in the range of different assay attributes in part devised in the form of a general scheme and in part built from the output of particular assays. Although we include for demonstration purposes three core ontologies, namely behavior, anatomy, and developmental anatomy (Figures 3 and 4 ), we have tested the schema only on behavior. We also include a possible structure for PATO attributes and a separate ontology for common values. We note, however, that the structure of PATO has not been finalized. Figure 3 shows the implementation of the schema in DAG-edit. Figure 4 shows a typical implementation of the Schema in Protégé 2000. Options for providing a definition, definition reference, documentation, associated annotations, synonyms, and so on, are offered in our schema. Similar options can be used for attributes using the metaslot options. Since most of the ontologies we are planning to use were generated using the DAG-edit [ 16 ] format, we had to convert them to the Protégé-2000 format using the tools and methodology described by Yeh et al . [ 27 ], with minor modifications. This task, however, should no longer be necessary as the latest version of DAG-edit allows the export of ontologies in OWL format. Modeling issues Decisions will inevitably have to be made to combine a core ontology with its attributes and then define facets of that relationship, for example, cardinality, attribute value type and attribute range. In our schema, the class hierarchy of all ontologies employed represents an 'is-a' relation. So, mouse social behavior 'is-a' mouse behavior, or mouse social behavior is a 'kind-of' mouse behavior and so forth. All other relationships, including PATO and 'part-of' relationships, are modeled as attributes. However, we note here that efforts are currently being made by the GO consortium to define and formalize the 'part-of' relationship, which is considered vital and special in bio-ontologies, especially anatomy [ 28 ]. Because our phenotype ontology and PATO need to be the result of a collaborative effort within the communities, we feel that it is important to set out the basic modeling concepts that need to be applied upon allocating attributes to the core ontologies. Deciding whether to introduce a new attribute or represent this functionality through an entity is often quite difficult. Several things need to be considered in order to make the best decision, although it should be noted that there are no clear distinction as to what is a right or wrong decision. The first thing to take into account is that subclasses of a class inherit all properties of the parent and could have additional properties and different restrictions from the latter. PATO should remain as general as possible, and, when possible, care should be taken to avoid making PATO domain specific. For example, in the behavior ontology there is a class named 'reflexes' that contains children such as 'blinking reflex', 'Preyer reflex' and 'righting reflex'. It might be worth considering having one 'attribute of reflex' available in PATO rather than creating a separate attribute 'of' for each individual reflex, such as 'attribute of blinking reflex', 'attribute of Preyer reflex', and so on. Then again, if one wishes to assign different functionalities to these properties, creating separate attributes might be useful. As a rule though, one should consider that PATO needs to be consistent, usable and interoperable if it is to be applied to the general domain of phenotypes. Repetition between core ontologies and PATO should be avoided where possible. What is also often not clear is whether one should add a new class to represent functionality or assign attributes to already existent classes. For example, think of the entity 'body position'. There are several ways to model this entity in the mouse behavior phenotype ontology. One could declare 'body position' as a child of a class called 'posture'. An 'attribute of body position' could then be assigned to this class with a range of values that might be specific to an assay, for example SHIRPA [ 29 ] allows the value 'lying on its left side' among other values to an assay for body position. Alternatively, a more general 'attribute of position' could be assigned to this class. The choice depends on the functionality of the ontology and the range of phenotypes we wish to express. If the entity requires more specific attribute values to represent specific functionalities important to the domain of knowledge, we assign more specific attributes. If this functionality is not important for the domain, we assign specific attribute values [ 8 ]. 'Body position' could also be split into an entity of 'body' and an attribute of 'position'. Again, a new class 'body position' should be assigned, if one considers the objects with different attributes as different kind of object and this distinction important in the domain. As a general rule, before assigning new classes and attributes one should consider the functionality and their role in the domain, creating more distinctions as the depth of knowledge that is required to be expressed in the ontology increases. Classes in the hierarchy should not necessarily have to introduce new properties [ 8 ]. Although, in many cases these entities could be represented as attributes, it is not necessary for the functionality of the domain. If the expert thinks that this distinction is significant for the class hierarchy and the logical representation of his knowledge of the domain, then these entities should be represented as classes [ 8 ]. An important additional consideration is whether creating new terms in an ontology results in terms that cannot be consistently distinguished experimentally ('resolution'). Conclusions We have presented here an approach to the use of ontologies in describing mouse phenotypes that could provide a platform for the consistent representation of mouse phenotypic data. We have also described in detail a possible methodology to construct applications of this schema across different domains. We have dealt with modeling issues and provide guidelines to deal with semantic and practical problems. We maintain that such modeling efforts in any domain should be done in a collaborative fashion in the community. Repetition between different parts of the mouse phenotype ontologies is unavoidable. However, the use of consistent IDs, synonyms and records for associated annotations could allow seamless integration of ontology products. The nature of the schema proposed, as well as its components, is extremely dynamic; therefore coordination of efforts is vital. The structure allows extensibility and interoperability. Although an ontology should not cover all possible information about a domain, the main idea behind this concept is to allow the phenotype ontology to cope with novel and unpredictable phenotypes and account for new assays, serving scientific autonomy and information validity and integrity. We have built a software system [ 30 ] which includes a browser that allows searching and viewing the knowledge captured though the complex relations described here and databases that allow the dynamic update of different parts of the core ontologies, including PATO, without the loss of applied facets. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549069.xml |
522804 | Immunohistochemical investigations and introduction of new therapeutic strategies in scleromyxoedema: Case report | Background Scleromyxoedema is a rare chronic skin disease of obscure origin, which may often be associated with severe internal co-morbidity. Even though different casuistic treatment modalities have been described, to date, curing still seems to be impossible. Case presentation We report a 44-year-old Caucasian female presenting with remarkable circumscribed, erythematous to skin-coloured, indurated skin eruptions at the forehead, arms, shoulders, legs and the gluteal region. Routine histology and Alcian blue labelling confirmed a massive deposition of acid mucopolysaccharides. Immunohistochemical investigations revealed proliferating fibroblasts and a discrete lymphocytic infiltration as well as increased dermal expression of MIB-1 + and anti-mastcell-tryptase + cells. Bone marrow biopsies confirmed a monoclonal gammopathy of undetermined significance without morphological characteristics of plasmocytoma; immunofixation unveiled the presence of IgG-kappa paraproteins. Conclusions Taking all data into account, our patient exhibited a complex form of lichen mxyoedematosus, which could most likely be linked a variant of scleromyxoedema. Experimental treatment with methotrexate resulted in a stabilisation of clinical symptoms but no improvement after five months of therapy. A subsequent therapeutic attempt by the use of medium-dose ultraviolet A1 cold-light photomonotherapy led to a further stabilisation of clinical symptoms, but could not induce a sustained amelioration of skin condition. | Background Lichen myxoedematosus (LM) represents a rare chronic skin disorder of unknown aetiology, which may often be accompanied by severe internal co-morbidity such as haematological involvement including paraproteinaemia, neurologic syndromes, gastrointestinal complications or cardiac abnormalities [ 1 - 4 ]. Clinically and histologically, LM is characterised by papular eruptions caused by an extensive dermal deposition of glycosaminoglycans [ 5 ]. Scleromyxoedema (SCL) is a variant of LM exhibiting erythematous, sclerotic and stiffed lesions beside lichenoid papules with only little tendency of spontaneous remission [ 6 , 7 ]. Even though various experimental treatment modalities have been described, to date, curing of SCL is still not possible. The oncoming case presentation focuses on a progressive variant of SCL as referred to clinical, immunohistochemical and laboratory investigations, followed by low-dose methotrexate (MTX) and subsequent medium-dose ultraviolet A1 (UVA1) cold-light treatment. Case presentation We report a 44-year-old Caucasian woman who initially presented in 2003 with a multitude of progressive lichenoid 2–4 mm papules starting two years ago, particularly marked on the forehead right above both eyebrows, on the dorsal aspects of the forearms, shoulders, legs as well as on the gluteal region, accompanied by severe pruritus (Fig. 1 ). Clinically, the papules were judged as discrete, circumscribed, erythematous to skin-coloured, firm skin eruptions associated with an induration and stiffening of the affected lesions. Furthermore, the patient complained about a progressive thickening of the glabella. Even though we cannot ensure continuous clinical deterioration, at the time of the initiation of therapy, there was no hint for a beginning stabilisation or even improvement of symptoms. Otherwise, she felt healthy and well. The general examination was without pathological findings. Neither a topical nor a systemic therapy was yet applied. Figure 1 Lichenoid papules beside thickened skin on the dorsal aspect of the left shoulder. Skin biopsies were taken from the left forearm/wrist, both legs and the left shoulder. Routine histological examination including haematoxylin-eosin, PAS labelling and Alcian blue staining revealed marked mucinous deposition within the upper and mid dermis beside an increased appearance of fibroblasts, collagen bundles and a discrete inflammatory infiltration. Additionally, immunohistochemical investigations were performed in order to enumerate CD4 + (T helper cells), CD68 + (macrophages), anti-mastcell-tryptase + (mastcells), decorin + (collagen fibril stability protein), MIB-1 + (Ki-67 + proliferating cells), CD20 + (B lymphocytes), and FGF-R + (fibroblast growth factor receptor bearing cells) cells taking consecutive sections (Table 1 ). A punch skin biopsy measuring 3 mm in diameter was taken from affected skin of the right forearm. 5 μm paraffin-embedded sections were cut, mounted on slides and routinely preserved. Prior to the single immunolabelling, different pretreatments were performed for antigen retrieval (Table 1 ). The alkaline phosphatase anti-alkaline phosphatase (APAAP) technique using the labelled streptavidin-biotin (LSAB) method was used to enumerate immunopositive cells at an individual dilution (Table 1 ) taking consecutive sections. The alkaline phosphatase fast red detection kit utilised a biotinylated secondary antibody that binds to the primary antibody. This step was followed by the addition of an streptavidin enzyme conjugate binding to the biotin present on the secondary antibody. Afterwards the specific antibody-secondary-antibody-streptavidin-enzyme-complex was detected using a precipitating enzyme reaction product. Each step was incubated for a precise time and temperature. The alkaline phosphatase was used as enzyme; the chromogene fast red could be visualised. Cells were evaluated semiquantitatively (absent (-), rare (o), moderate (+), frequent (++)) directly below the dermoepidermal junction. Immunopositive cells were evaluated 'blind' separately in two view fields in a row (0.25 mm × 0.25 mm each) directly below the dermoepidermal junction resulting in a length of 0.25 mm and 0.50 mm in depth. In order to avoid a sampling error, a number of sections were randomly reevaluated by a second observer. In case of a significant difference, the sections had to be recounted by both observers. In brief, immunolabelling revealed occasional perivascular CD4 + and CD20 + lymphocytes located in the papillary dermis and a high number of anti-mastcell-tryptase + cells within the subepithelial perivascular infiltrate revealing a continuing decrease with increasing depth (Fig. 2 ). Simultaneously, an increased dermal expression of MIB-1 + cells (Fig. 3 ), morphologically predominantly fibroblasts, within the upper and mid dermis and sporadic FGF-R + cells in an unspecific distribution could be detected. CD68 immunohistochemistry and intradermal decorin levels did not alter remarkably as compared to healthy controls (data not shown). Exact results of all immunohistochemical stainings are detailed in Table 2 . Table 1 Overview about the performed immunohistochemistry (alkaline phosphatase anti-alkaline phosphatase (APAAP) technique using the labelled streptavidin-biotin (LSAB) method) Antibody Source Pretreatment* Dilution Incubation time CD4 Novocastra Loxo, Dossenheim, Germany H 1:60 30 min CD68 Dako, Hamburg, Germany P 1:25 30 min Tryptase Dako, Hamburg, Germany P 1:400 28 min MIB-1 Dako, Hamburg, Germany H 1:10 32 min FGF-R Oncogene Research, San Diego, USA N 1:10 30 min Decorin Oncogene Research, San Diego, USA N 1:10 30 min CD20 Novocastra Loxo, Dossenheim, Germany H 1:50 30 min *N = none, P = protease digestion, H = heat (microwave-3-step-technique) Figure 2 Immunhistochemistry unveiling sporadic lymphocytes beside a high number of anti-mastcell-tryptase + cells within the subepithelial perivascular infiltrate. Figure 3 Mucinous deposition of the upper and mid-dermis accompanied by an elevated occurrence of MIB-1 + dermal fibroblasts. Table 2 Semiquantitative data* of the immunohistochemical studies on a patient with scleromyxoedema Antibody CD4 CD68 Tryptase MIB-1 FGF-R Decorin CD20 Upper dermis + o ++ + o - o Mid-dermis - - + + - - - Lower dermis o - o o - - - *- = absent, o = rare, + = moderate, ++ = frequent Complete laboratory measurements unveiled the following pathological results: leucocytes 10870 μL -1 , lymphocytes 13.2%, IgG 2000 mg/dl. There was no increase in B cell count. Immunoelectrophoresis disclosed albumin 51.4%, alpha-2 globuline 10.5%, gamma globuline 22.9%. Cranial x-ray, x-ray of the thorax, ultrasound of the abdominal organs, electrocardiography and urinary investigations were unremarkable. Blood smear cytological evaluation revealed beginning qualitative but still no quantitative changes as defined by leukocytic aberrations pointing towards a leftward shift. Serum immunofixation demonstrated an IgG-kappa paraproteinaemia. No elevation of the IgG-lambda paraprotein was assessed. Bone marrow biopsies displaying reactive lymphoid infiltration including minimal extension of plasma cells with monoclonal immunoglobuline production provided evidence for monoclonal gammopathy of undetermined significance (MGUS) without distinct morphological characteristics of a plasmocytic plasmocytoma or plasmoblasts. Initially, 20 MHz ultrasound scanning producing cross section images of the skin was established in order to measure both structure and thickness of the skin at the dorsal aspects of the left wrist and the right forearm. The total thickness of the skin was measured from the entrance echo to the border between the dermis and the subcutaneous tissue. A cutaneous diameter of 2291 μm at the left wrist (lesional skin) and of 1106 μm at the right forearm (non-lesional skin) could be assessed. In our unit, an experimental treatment modality using oral MTX 12.5 mg once per week followed by a subsequent folic acid application on the following day for a 6-months-period was subsequently initiated. MTX was well tolerated by our patient. After the first three months, the continuous progress of skin lesions during the last two years could be stopped and our patient experienced subjectively an improvement and objectively a stable clinical outcome without new lesions. Subjective impression of amelioration could not be confirmed by means of ultrasound measurement. Within the following two months, no further improvement could be evaluated, whereas no further progression such as formation of new lesions or increase of stiffness could be observed. Due to the unsatisfying clinical results and declining acceptance by our patient, MTX treatment was stopped and a subsequent therapeutic attempt with medium-dose UVA1 cold-light phototherapy was initiated. Irradiation consisting of 50 J/cm 2 single-dose UVA1 (CL 300000 liquid, Photomed, Hamburg, Germany) was performed four times a week for three weeks followed by two times a week for further two weeks resulting in a cumulative dose of 800 J/cm 2 after five weeks. Meanwhile, the skin status again remained stable, whereas no improvement could be observed. Therefore, our patient broke up phototherapy. To date, skin condition has slightly worsened without any current treatment modality. Conclusions The population prevalence of SCL is known to be extremely low. Skin lesions are characterised by an increased deposition of acid mucopolysaccharides within the papillary and upper reticular dermis [ 8 ]. Even today, aetiology and pathogenesis remain hypothetic. Aberrant dermal deposition of monoclonal paraproteins predominantly of the IgG subtype combined with elevated IgG serum levels indirectly stimulating fibroblast activity are frequently found in LM patients [ 9 - 11 ]. Nevertheless, fibromucinous lesions of LM without the presence of paraprotein accumulations have also been described [ 12 ]. Beside typical skin eruptions, LM might also be associated with severe internal and neurological abnormalities such as cardiac irregularities, paralysis, hemiparesis or even progress to coma [ 3 , 4 , 13 ]. Despite sporadic case reports introducing new therapeutic strategies in LM and SCL, common treatment modalities are still disappointing and unsatisfactory. Topical treatment including hyaluronidase and triamcinolone as well as systemic efforts by the use of corticosteroids, cyclophosphamide, electron-beam therapy, hydroxychloroquine, PUVA, extracorporeal photopheresis, plasmapheresis or high-dose intravenous immunoglobulin partly displayed only limited success in individual patients [ 12 , 14 - 20 ]. In our patient, Alcian blue staining disclosed a remarkable deposition of mucinous material within the upper dermal layers combined with an increased appearance of proliferating MIB-1 + and occasional FGF-R + fibroblasts in immunohistochemistry. Ki-67 (MIB-1) and fibroblast growth factors are involved in a variety of mitogen and proliferative activities [ 21 ]. Thus, the enhanced appearance of positive cells might represent an increased overall activation probably resulting in an aberrant release of mucopolysaccharides. Decorin contributes to the collagen fibril stability and high levels of decorin seem to be closely linked to dermal fibrotic stages as known from systemic sclerosis [ 22 ]. Here, an increased intradermal decorin expression could not be demonstrated. Simultaneously, in our patient the mucinous deposition was accompanied by a decreased presence of collagenous bundles. Interestingly, immunohistochemistry also revealed a number of CD4 + and CD20 + dermal inflammatory lymphocytes as well as anti-human mast cell tryptase + cells, which may profoundly contribute to mucinosis formation [ 23 ]. Unfortunately, we were not able to provide a longitudinal analysis of the different stainings due to missing consent of the patient to perform additional experimental biopsies within the course of therapy. By considering clinical appearance, laboratory findings, immunofixation, bone marrow biopsy and histological evaluation, our patient presumably exhibited a complex variant of SCL. Fibroblast activity was supposed to be increased reflected by a corresponding high number of MIB-1 + and FGF-R + cells within the upper dermis beside a massive mucinous deposition. MTX therapy is approved in the treatment of malignant lymphoma. Additionally, low-dose MTX therapy has been established as a potent regimen in the treatment of T cell related skin diseases associated with a subsequent elevated fibroblast activation status (e.g. progressive systemic sclerosis) [ 24 ]. As in our patient paraproteinaemia of the IgG-kappa class, morphological signs of MGUS, proliferating dermal fibroblasts and a discrete T cell weighted lymphocytic dermal infiltration could be verified, we first decided to start a therapeutic attempt by the use of 12.5 mg MTX weekly in order to suppress local cellular activity following the promising reports about application of the anti-metabolite melphalan and the alkylating agent cyclophosphamide in previous studies [ 12 , 14 ]. Follow-up examinations were performed monthly. Three months after initiation of MTX therapy, an encouraging stable clinical outcome as well as an decline of pruritus without further progression of the disease was observed. However, the 5-months-follow-up revealed no apparent improvement of skin status leading to the joint decision of breaking up MTX treatment. UVA1 phototherapy has been shown to be effective in a number of inflammatory and fibrotic skin disorders by the induction of T cell apoptosis, collagenase activity and antiproliferative pathways [ 25 ]. Therefore, we decided to initiate a second attempt by the usage of a common medium-dose UVA1 irradiation protocol. Nevertheless, our patient broke up this regimen even after a 5-weeks-period due to the absence of immediate clinical improvement and an unfavourable time/benefit ratio, while follow-up examinations during this time revealed a stop of the progress anyway. However, in order to interrupt the clinical progression or even therapy-resistance as reflected by our case presentation and by considering the stable skin conditions following UVA1 phototherapy, we are currently discussing a new therapeutic attempt applying extracorporeal photochemotherapy, as proposed by Krasagakis et al. [ 16 ], in order to stabilise or even improve the present slight aggravation without any potent therapy. List of abbreviations LM: Lichen myxoedematosus; SCL: Scleromyxoedema; MTX: methotrexate; UVA1: ultraviolet A1; MGUS: monoclonal gammopathy of undetermined significance Competing interests None declared. Authors' contributions F.B. participated in the design of the study, carried out the immunohistochemistry, performed the statistical analysis and drafted the manuscript. S.R. carried out 20 MHz ultrasound scanning. A.K. conceived of the study. M.F., M.S. and P.A. participated in histology, design and coordination. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC522804.xml |
552313 | Sequential super-stereotypy of an instinctive fixed action pattern in hyper-dopaminergic mutant mice: a model of obsessive compulsive disorder and Tourette's | Background Excessive sequential stereotypy of behavioral patterns (sequential super-stereotypy) in Tourette's syndrome and obsessive compulsive disorder (OCD) is thought to involve dysfunction in nigrostriatal dopamine systems. In sequential super-stereotypy, patients become trapped in overly rigid sequential patterns of action, language, or thought. Some instinctive behavioral patterns of animals, such as the syntactic grooming chain pattern of rodents, have sufficiently complex and stereotyped serial structure to detect potential production of overly-rigid sequential patterns. A syntactic grooming chain is a fixed action pattern that serially links up to 25 grooming movements into 4 predictable phases that follow 1 syntactic rule. New mutant mouse models allow gene-based manipulation of brain function relevant to sequential patterns, but no current animal model of spontaneous OCD-like behaviors has so far been reported to exhibit sequential super-stereotypy in the sense of a whole complex serial pattern that becomes stronger and excessively rigid. Here we used a hyper-dopaminergic mutant mouse to examine whether an OCD-like behavioral sequence in animals shows sequential super-stereotypy. Knockdown mutation of the dopamine transporter gene (DAT) causes extracellular dopamine levels in the neostriatum of these adult mutant mice to rise to 170% of wild-type control levels. Results We found that the serial pattern of this instinctive behavioral sequence becomes strengthened as an entire entity in hyper-dopaminergic mutants, and more resistant to interruption. Hyper-dopaminergic mutant mice have stronger and more rigid syntactic grooming chain patterns than wild-type control mice. Mutants showed sequential super-stereotypy in the sense of having more stereotyped and predictable syntactic grooming sequences, and were also more likely to resist disruption of the pattern en route, by returning after a disruption to complete the pattern from the appropriate point in the sequence. By contrast, wild-type mice exhibited weaker forms of the fixed action pattern, and often failed to complete the full sequence. Conclusions Sequential super-stereotypy occurs in the complex fixed action patterns of hyper-dopaminergic mutant mice. Elucidation of the basis for sequential super-stereotypy of instinctive behavior in DAT knockdown mutant mice may offer insights into neural mechanisms of overly-rigid sequences of action or thought in human patients with disorders such as Tourette's or OCD. | Background Overly rigid sequential patterns of movement and thought characterize several human brain disorders involving dysfunction in basal ganglia systems (i.e. dopamine nigrostriatal projections to the neostriatum and related brain structures). For example, pathological repetitions of spoken words in Tourette's syndrome, and the tormenting habits and thoughts of obsessive-compulsive disorder (OCD), involve overly rigid sequential patterns of action, language or thought [ 1 - 9 ], which in part may be influenced by genetic factors [ 10 - 13 ]. Normal sequential patterns of action, language and thought also have been suggested to depend on proper basal ganglia function [ 14 , 15 ]. For example, Marsden proposed that "The sequencing of motor action and the sequencing of thought could be a uniform function carried out by the basal ganglia" [ 15 ], and a variety of computational models have been proposed to carry out the general sequencing functions of basal ganglia [ 16 - 19 ]. According to this view, basal ganglia systems evolved originally to coordinate syntactic patterns of instinctive movements, and were extended subsequently by natural selection to participate in sequencing cognitive and linguistic functions as well. Almost all behavior is sequential, so what do we mean by 'syntactic sequence'? In the simplest terms, a syntactic sequence is one that follows normative rules that determine the temporal progression of its elements and impart a lawful predictability to the sequence as a whole [ 14 , 20 , 21 ]. Human language has real syntax, as the prototypical example, complete with recursive generative rules [ 14 , 21 , 22 ]. But other behavior can be described as having properties of syntax too, if the behavioral flow is governed by lawful sequential patterns that follow normative rules to produce a complex serial order [ 14 , 20 , 23 - 26 ]. Neuroethological studies of natural behavior in animals have shown that neostriatum, substantia nigra, and their connecting dopamine projections are critical to sequential stereotypy for complex serial patterns of instinctive behavior [ 26 - 35 ]. In particular, a complex fixed action pattern displayed spontaneously by rodents during grooming behavior, called a syntactic grooming chain, has been exploited by neuroethological studies that point to basal ganglia systems as the controlling neural mechanisms for the stereotypy of complex sequential patterns [ 27 , 28 , 36 ]. A syntactic chain is a 4-phase series of up to 25 elements, each phase containing recursive iterations of its characteristic element (Figure 1 ; see Additional movie file 1 ). This syntactic sequence occurs spontaneously during grooming behavior of most rodents. Mice, rats, gerbils, hamsters, guinea pigs, ground squirrels and other species all have their own signature patterns of syntactic chains, with different details, but all follow the syntactic 4-phase rule [ 37 ]. In one squirrel species ( Spermophilus beecheyi ), syntactic chains have been even further ritualized into a stereotyped display, and adapted for territorial communicative use [ 38 ]. As is typical of fixed action patterns, no two syntactic chains may be absolutely identical, but they are highly similar, stereotyped, and easily recognized, and always follow the same serial patterning rule [ 39 , 40 ]. Thus syntactic grooming chains are complex multi-component patterns that are sequentially stereotyped, and capable of interacting with evolutionary selection pressures that alter the genotype to modulate behavioral patterns. They represent precisely the sort of sequencing function that ancestral basal ganglia systems might originally have evolved to perform [ 2 , 9 , 14 , 15 , 24 , 29 - 31 ]. Figure 1 Prototypical syntactic grooming chain pattern. Choreograph shows mouse movements of the left/right paws over the face (time proceeds from left to right). Lines deviating above/below the horizontal axis show the trajectory height of left/right paws. Large black box denotes bout of body licking, and placement of asterisk in box shows which left/right side flank was chosen by the mouse to initiate body licking. Phase I : series of ellipse-shaped strokes tightly around the nose. Left and right paws often take alternating turns as the major/minor trajectory. Phase II : series of unilateral strokes, each made by one paw, that reach up the mystacial vibrissae to below the eye. Mice often make hybrid Phase I/II strokes, in that one paw makes a Phase II unilateral stroke while the remaining paw makes a smaller Phase I type ellipse. Phase III : series of bilateral strokes made by both paws simultaneously. Paws reach back and upwards, ascending usually high enough to pass over the ears, before descending together over the front of the face. Phase IV (strong or classic form) : sustained bout of body licking, preceded by postural cephalocaudal transition to move mouth and tongue from facial and paw grooming to body grooming. Mouse-typical pattern modified from Berridge (1990). See Additional movie file 1 for examples of syntactic grooming chains by DAT-KD mutant mice. The firing of some basal ganglia neurons in neostriatum and in substantia nigra codes the serial pattern of syntactic grooming chains as an entire sequence in rats [ 27 , 28 ]. In addition, the integrity of basal ganglia neurons is necessary for normal sequential stereotypy of the instinctive pattern. For example, after lesions of neostriatum, rats lose the ability to complete the 4-phase pattern properly (especially after lesions of anterior dorsolateral neostriatum, which contains the neurons that particularly code the syntactic pattern), even though the lesions do not impair constituent grooming movements [ 27 , 36 ]. Similar deficits in grooming syntax are caused by disruption of dopamine neurotransmission in mice lacking dopamine D1 receptors [ 41 ], and in normal rats with neostriatal dopamine depletion caused by 6-hydroxydopamine lesions of nigrostriatal projections [ 42 ]. Brain lesions that disrupt behavioral sequences indicate a potential sequencing function for the targeted structures. However, factors besides sequencing loss may contribute to disrupted serial patterns after lesions. An alternative and stronger proof for dopamine mediation of action syntax would be to demonstrate enhanced stereotypy of behavioral sequences, by boosting nigrostriatal dopamine neurotransmission. Enhanced sequential stereotypy would be reflected if the complex serial pattern as a whole entity became more sequentially rigid or persistent. Indeed, in rats, pharmacological boosting by dopamine D1 agonists administered systemically or into brain ventricles produces sequential super-stereotypy of syntactic grooming chains [ 43 - 45 ]. In a state of sequential super-stereotypy, the stereotyped pattern becomes even more predictable than normal, which is evident as an increase in the probability that all four phases will be completed in syntactic order [ 43 , 44 ]. Such rigidity of complex multiple-phase sequences contrasts with simpler repetition stereotypies (e.g., associated with D2 receptor activation), in which the same movement is repeated over and over again [ 46 - 50 ]. In human pathologies such as Tourette's or OCD, complex sequential super-stereotypy often occurs spontaneously in human patients. If sequential super-stereotypy of complex instinctive behavior sequences is to serve as a model of human disorders involving sequential super-stereotypy, it ought to be able to occur spontaneously in some individual animals too. In addition, it should possess features of compulsive behavioral sequences. Compulsive behavior may have several features, including both perseverative tendencies and more rigid sequences of entire serial patterns . To date, prior genetically-modified mouse models of spontaneous compulsive behavior have successfully captured the perseverative feature, but it is not yet clear whether these animal models also share the exaggerated serial pattern feature of compulsive behavior. For example, the Hoxb8 lox mutant model has been reported to exhibit OCD-like increased persistence of self-directed grooming and body-licking, and even mutual grooming of other mice [ 13 , 51 , 52 ]. Similarly, the D1CT mutant mouse, caused by transgenic potentiation of D1-associated brain circuits, shows OCD-like persistence of grooming, as well as persistence of other behaviors such as digging, climbing, and tics [ 3 , 53 - 57 ]. However, it is unknown whether these or any other animal models also show excessively rigid sequences, in the sense of a stronger multi-element and rule-governed sequence that becomes more rigid as a single complex pattern. For modeling the serial rigidity feature of OCD or Tourette's, an animal model is needed that spontaneously produces an overly-rigid and serially-complex sequence of behavior, such as a syntactic grooming chain. Here we show that this serial pattern feature of sequential super-stereotypy indeed appears spontaneously without drugs in DAT-KD mutant mice with genetic knockdown of the dopamine transporter (DAT) [ 58 ]. DAT-KD mutant mice have 10% normal DAT expression in dopamine neurons [ 58 ], which impairs synaptic re-uptake of dopamine, resulting in elevated (170%) levels of extracellular dopamine in neostriatum (wild-type mice = 100%) [ 58 ]. DAT-KD mutant mice show other behavioral evidence for high levels of dopamine activation. They tend to be hyperactive, to walk in perseverative straight paths, and to over-pursue certain incentive stimuli [ 58 - 60 ]. The question asked in the present study was whether these mutant mice would also show sequential super-stereotypy in their syntactic chains – that is, do they have excessively rigid serial patterns of instinctive grooming behavior? Results Syntactic chains Hyper-dopaminergic mutant mice and wild-type control mice each generated syntactic chains of grooming as described above (Figures 1 , 2 & 3 ). Syntactic grooming chains by DAT-KD mice had virtually all the typical features of wild-type chains and of syntactic chains previously reported for outbred mice and D1 receptor knockout mice [ 37 , 41 ] (Figures 1 & 3 ; see Additional movie file 1 ). Figure 2 Sequential super-stereotypy of syntactic pattern. Cumulative rates of full pattern completion by DAT-KD mutant (dark symbols) and wild-type mice (open symbols) for each type of syntactic chain (Perfect, Insertion of unpredicted component, Phase Reversal, Phase Skip, Substitution of paw lick for Terminal Phase IV component). Choreographs at bottom show example for each type of syntactic chain. Mutant mice have higher rates of syntactic completion for all forms of the chain that terminate in the strong form of Phase IV, body licking, which characterizes the prototypical Phase IV for all rodents. Wild-type mice use a weak form of Phase IV (paw lick substitution) to terminate a substantial proportion of their syntactic chains. All mice show less pattern completion when grooming in the laboratory (top) than when grooming in their home cage (bottom), but mutant mice show more rigid sequential patterns than wild-type mice while grooming in both environments. * p < 0.05; ** p < 0.01. Figure 3 Sample choreographs of actual syntactic chains. Both mutant mice and wild-type mice emit every type of syntactic chain described in the text (Perfect, Insertion of unpredicted component, Phase Reversal, Phase Skip, and Substitution of paw lick for Terminal Phase IV). Syntactic grooming chains The sequential pattern of a syntactic grooming chain contains up to 25 movements serially combined into 4 syntactic or rule-governed phases that form one chain pattern [ 61 ] (Figure 1 ; see Additional movie file 1 ). Each of the 4 phases contains recursive repetitions of its particular component movement. Phase I consists of 5–10 rapid elliptical forepaw strokes made with both paws simultaneously over the nose and mystacial vibrissae. In mice, Phase I ellipses are often slightly asymmetrical and alternating, in the sense that the 'major paw' makes a slightly larger stroke than the 'minor paw' [ 37 ]. Typically, the major/minor role alternates over successive Phase I strokes between left and right paws. The entire Phase I lasts for about one second. Phase II is short (0.25 s) and consists of 1–4 unilateral or highly asymmetrical strokes made by one forepaw. The unilateral stroke is typically of small or medium amplitude ascending to about the level of the eye. In mice, the other paw not participating in the Phase II stroke often makes a smaller Phase-I ellipse-type stroke simultaneously [ 37 , 41 ]. Thus, Phase II in mice typically contains several hybrid Phase I-II strokes, in contrast to rats, which move only a single forepaw [ 37 ]. Mice generally alternate between left and right paws in making Phase II strokes (though sometimes the same paw repeats a short series of Phase II strokes). Phase III is highly visually distinctive, and consists of 1–5 large bilateral strokes with both paws. Both paws move very symmetrically almost as mirror images of the other, typically ascending together high up the side of the face, and passing forward synchronously over the ears. Phase III strokes are extremely stereotyped, usually all of the same height, and with both paws traveling back down to the nose between successive Phase III strokes [ 37 ]. The entire Phase III lasts 1–3 s. Phase IV concludes the prototypical chain, and consists of a postural turn to the side and caudally, and lowering of the head to bring the tongue towards the flank or side of the body, followed immediately by a 2–5 s bout of body licking directed to the flank. Syntactic Initiation: rate of starting chains In terms of the number of syntactic chains started during a grooming bout, DAT-KD mutant mice initiated marginally more syntactic chains overall than wild-type mice (F (1,86) = 3.592, p = 0.061; Figure 4 ). The difference in chain initiation was context dependent. All mice were twice as likely to initiate syntactic chains in the laboratory than at home (F(1, 82) = 85.73, p < 0.001), and mutant mice in particular initiated approximately 25% more chains than wild-type mice in the laboratory environment (F (1,86) = 17.315, p < 0.001; Figure 4 ), compared to only 5% more in the home environment. If the laboratory context was considered more stressful than the home cage environment, then stress dramatically promoted the tendency to begin a highly stereotyped sequence, especially for mutants. Figure 4 Initiation of stereotyped syntactic chain pattern. Rates of initiation of syntactic chains are slightly higher for mutant mice, especially while grooming in the laboratory, measured cumulatively across the entire observation period (top). In more detail, initiation rates are broken down as occurring either early versus late in grooming bouts (bottom). All mice tend to start the stereotyped sequential pattern more often early in a grooming bout. Mutant mice are even more likely than wild-type mice to start the pattern in an early grooming bout, both in home and laboratory environments. ** p < 0.01. The nature of the context-dependence of the difference was further clarified by a closer look at the time course of exactly when syntactic chains were begun by mice during a grooming bout. The overwhelming majority of syntactic chains tended to be initiated early in a grooming bout by all mice (Figure 4 ). Mutants initiated up to twice as many chains as wild-type mice per minute of grooming in the first quarter of a bout, whereas by the last quarter of their grooming bouts mutant and wildtype initiation rates no longer differed significantly (Figure 4 ). Mutants tended to begin more grooming bouts than wild-type mice especially in the laboratory (described below), which may have facilitated the mutants' greater tendency to initiate the syntactic chain pattern in the laboratory (Figure 4 ). In short, syntactic chains were initiated early in a grooming bout by all mice, but mutant mice were even more likely than wild-type mice to initiate chains during those early portions of a grooming bout, and the mutant advantage was greatest in the laboratory environment (which might be the most stressful environment). Syntactic completion In mice and rats, once a syntactic chain pattern begins with Phase I, each remaining action can be predicted with roughly 80–90% accuracy. The entire syntactic chain occurs with a frequency over ten thousand times greater than could be expected by chance (based upon the relative probabilities of the component actions). However, several types or degrees of chain completion are possible. Types of syntactic chains A prototypical or perfect syntactic chain requires Phases I, II, III and IV in order, with no deviations, additions or omissions. Perfect chains were occasionally performed by both mutant and wildtype mice. After performing Phase I, II and III strokes over the face, a mouse performs Phase IV by transitioning to body grooming. For this transition to body licking, the mouse must bend down and backward to bring its mouth toward a side flank, and then begin a vigorous bout of body licking that continues for 1–4 s. In addition, several types of imperfect syntactic chains were observed in both mutant and wildtype mice. Imperfect sequences proceed from Phase I to IV with some minor deviation from the prototypical pattern along the way. In this study, we recognized three forms of imperfect completion. All involved a minor imperfection, which was either an insertion, reversal or replacement of a component action within the syntactic chain. Imperfect completion forms were: (i) Reversal of Phases II-III, where Phase II unilateral strokes were emitted after Phase III bilateral strokes (instead of before them), but the chain was otherwise syntactically correct; (ii) Insertion of an unexpected movement component in between phases, usually a quick paw lick or several paw licks inserted between Phases III and IV; (iii) Skip or omission of one phase en route to completion, where a chain lacked either any Phase II unilateral face stroke, or any Phase III face strokes (never both), but was otherwise syntactically correct (e.g. an observed order of I-III-IV). Finally, a fourth type of syntactic chain was observed that failed to be completed in the prototypical sense, but where the mice substituted paw licking in the terminal place of Phase IV (which might provisionally be regarded as an "attempt" to complete syntactically). We called this Terminal substitution: the final Phase IV component (body licking) was completely replaced with a different type of licking movement (paw licking), and the chain was otherwise syntactically correct (e.g. I-II-III-paw lick bout). Terminal substitution never attains a prototypical Phase IV, and so is not really a form of syntactic completion by criteria used in earlier studies. However, the terminal substitution of paw licking might be viewed as an attempt to complete syntactically with a transition from paw strokes to licking, compared to other forms of incompletion such as either simply stopping or immediately launching into a sequentially flexible series of grooming strokes. Thus for the purpose of analysis, we examined the consequences of allowing terminal substitution to count provisionally as a form of "weak" completion. At the completion of Phase IV (strong or weak), over 93% of syntactic chains led to continued grooming of body or face in sequentially flexible and much less predictable patterns compared to syntactic chains. After 7% of chains, the end of Phase IV terminated the entire grooming bout, and the mouse rested quietly afterwards or began to explore the chamber. Syntactic rigidity: strength of pattern completion DAT-KD mutant mice not only started more syntactic chain patterns, they were also more likely than wild-type mice to complete the syntactic chain patterns they started – in both laboratory and home environments (Figures 2 & 3 ). Sequential super-stereotypy (i.e. more predictable and stereotyped completion of entire sequence) of DAT-KD mutant mice was the most consistent and robust finding of our study (F (1,78) = 12.33, p < 0.001; Figure 2 ). The higher syntactic rigidity of mutants was visible qualitatively and verified quantitatively (Figures 2 & 3 ), and it interacted with the various types of syntactic completion described above (interaction between mutant/wildtype × perfect/imperfect types: F (4,184) = 5.96, p < 0.001). Hyper-dopaminergic mutant mice nearly always completed their syntactic chains with the strong form of Phase IV (body licking), whereas wild-type mice completed roughly half their chains with only the weaker form of Phase IV (paw licking). Mutant mice completed a higher percentage of insertion, reversal and omission types of syntactic chains than wild-type mice (F(4, 184) = 129.01, p < 0.001; each subtype; Figures 2 & 3 ). These stronger or more rigid chains of mutant mice more closely corresponded to the prototypical 4-phase syntax pattern (including the prototypical terminal Phase IV component: body licking). In other words, mutant mice were better than wild-type mice at resisting disruption of the pattern by minor flaws that occurred along the way, and mutants more often returned to the full-blown pattern after any distraction. For example, insertion chains included 1 or 2 extraneous movements, such as a nonsyntactic paw lick action inserted between Phases III and IV. After an insertion, mutants were nearly 50% more likely than wild-type mice to reach a strong form of Phase IV completion (mutant vs. wildtype, p < 0.01, Bonferroni). Similarly, a reversal error reversed the serial order of Phases II and III, or followed a Phase II stroke with a late Phase I ellipse stroke, and after a reversal mutant mice were nearly 50% more likely than wild-type mice to go on to complete a strong form of Phase IV. Finally, in an omission chain, a mouse would omit either Phase II or Phase III (never both), and after an omission mutants were again roughly 50% more likely than wild-type mice to successfully return to the full pattern and reach a strong form of terminal Phase IV completion (each p < 0.01, Bonferroni). In contrast, wild-type mice had a greater proportion of terminal substitution chains that never achieved a full syntactic transition to body grooming. Wild-type mice instead substituted a weaker paw-lick form of Phase IV as terminal component. In terminal substitution, a mouse completely omitted the normal Phase IV shift to body licking, and instead simply continued to lick its paws, never changing posture or moving its head caudally out of the normal facial grooming position (the complete failure of transition to body licking after paw licking marked the difference between Insertion and Terminal Substitution chains). Wild-type mice had nearly twice the proportion of terminal substitutions as mutant mice (F(1,78)= 11.47, p < 0.001). If terminal substitution is regarded as failure to complete the pattern, then wild-type mice simply failed to complete over half the syntactic chains they began. More leniently, wild-type mice could approach an 80% – 90% rate of syntactic completion – if we took the unprecedented step of allowing Phase IV terminal substitution to count as weak completion (Figure 2 ). Allowing this weaker criterion was the only way to consider wild-type mice able to achieve the 80%–90% syntactic completion level that mutant mice successfully achieved through the stronger prototypical form of Phase IV. In summary, DAT-KD mutant mice had more rigid sequential patterns than wild-type controls in several ways. Mutant mice were more likely than wild-type mice to proceed syntactically through Phases I, II and/or III to reach the syntactic final Phase IV (body licking). Even after encountering minor imperfections along the way, mutants persevered in the sequential pattern. Wild-type mice introduced the same imperfections in their syntactic pattern, but did not return to the full pattern or complete Phase IV as strongly, ending their chains without ever reaching the full-blown transition to body grooming that normally terminates a syntactic chain pattern. Finally, syntactic completion was highest in home environment grooming for all mice (even though more syntactic chains were begun in laboratory) (F(1,78)14.31, p < 0.001). This difference suggests that stress may promote the initiation of stereotyped sequences, but impede their lawful completion, and is consistent with reports that stress disrupts completion of syntactic chain sequences [ 62 ]. However, mutant mice were equally more likely than wild-type mice to complete strong patterns in both laboratory and home environments. Motor control for movement capacity In order to reject motor confounds that might have provided an alternative explanation of some results, we assessed whether wild-type mice were simply less able to perform body licking movements than mutant mice. If wild-type mice had motor deficits that impaired their ability to perform body-lick posture/movements, then wild-types might have had weakened syntactic chains simply because of their motor incapacity to perform Phase IV movements, rather than because mutants had stronger sequencing tendencies. Therefore we analyzed whether wild-type mice spent a lower proportion of their total grooming behavior time making body licking movements compared to mutant mice. However, wild-type mice did not have significantly lower total cumulative duration scores for body licking overall than mutant mice (F(1,78) = 0.56, n.s.), indicating there was no motor impairment of Phase IV movements. That suggests the difference in tendency to complete syntactic chains represents a true difference in sequence rigidity or pattern strength, and not in simple motor capacity. Overall grooming behavior: amount, bout number, and bout duration All mice groomed twice as much in their home cages than in the laboratory environment, suggesting that the relatively novel laboratory environment might have acted to suppress spontaneous grooming behavior (F (1,82) = 1.773, p < 0.001; Figure 5 ). Grooming behavior in the laboratory was less than half that of the home cage for both mutants and wildtypes (in terms of cumulative grooming duration per hour of observation). DAT-KD mutant mice spent 10%–50% more time than wild-type control mice in grooming behavior overall (F (1,86) = 3.949, p < 0.05), and the mutant propensity to groom more was most visible in the home environment (p < 0.1). Figure 5 General amount and bout features of grooming behavior. Cumulative time spent in grooming behavior during observation (total duration), Duration of individual bouts of grooming, and the Number of bouts of grooming emitted during observation session. Mutant mice tend to spend more time in grooming than wild-type mice, and to have longer grooming bouts, in the home environment. Mutant mice tend to emit a greater number of fragmented bouts when grooming in the laboratory environment. These general features of grooming enhancement in mutant mice are flexible and context-dependent, in contrast to the greater mutant rigidity of sequential pattern that is constant across both environments (shown in Figure 2). * p < 0.05; ** p < 0.01. Closer analysis of grooming compared the relative contributions to increased grooming time of greater bout numbers versus longer bout durations (Figure 2 ). The increased time spent grooming by mutant mice in the home cage was due to longer grooming bouts (but not to a greater number of bouts) compared to wild-type mice. In their home cages, grooming bouts in mutant mice were 80% longer than in wild-type mice (F (1,86) = 4.083, p = 0.008), while bout numbers did not differ. Although mutants emitted only marginally more body licking than wild-type mice in an analysis that combined data from both home and laboratory environments (F (1,60) = 3.403, p < 0.07), a separate analysis of grooming specifically in the home cage showed that mutants at home had longer cumulative durations of body licking (p < 0.05, Bonferroni), consistent with prolongation of the later components of cephalocaudal grooming bouts in that home environment [ 63 ]. However, as a percentage of total grooming, the proportion of mutant body licking to facial stroke components was not higher than for wild-type mice, either overall (F(1,60) = 0.58, n.s.), or even in the home cage (p = 0.32), which suggests that the mutants' longer grooming bouts in the home cage may also have included more facial strokes than wild-type mice. Thus, longer mutant grooming bouts in the home cage likely involved expansion of several components of grooming, including longer body licking bouts and facial strokes. These perseverative features of DAT-KD mutant grooming in the home cage therefore may overlap with perseverative body grooming tendencies reported for other genetic animal models of compulsive behavior, such as Hoxb8 lox and D1CT mutant mice [ 13 , 51 - 53 , 55 , 57 ]. Conversely, in the laboratory environment, DAT-KD mutants' higher grooming was chiefly due to a greater number of grooming bouts (but not longer bouts). In the laboratory environment, mutants began more grooming bouts than wild-type mice (F(1,86) = 3.478, p = 0.026), but their duration of bouts did not differ. Thus, different features of grooming bouts (length versus number) were enhanced in mutant mice depending on their environmental context of the moment. However, as described above, in both home and laboratory the hyper-dopaminergic mutants were always more likely than wild-type mice to perform more rigid and strongly stereotyped syntactic chain sequences. Discussion Sequential super-stereotypy: pattern completion Our results reveal that hyper-dopaminergic mutant mice show excessively strong and rigid manifestations of a complex fixed action pattern compared to wild-type mice. Their sequential super-stereotypy was produced by DAT knockdown mutation, which reduces DAT to 10% of wild-type levels and causes extracellular dopamine elevation to 170% in neostriatum [ 58 ]. Mutant mice showed more stereotyped and predictable syntactic grooming chains, the instinctive fixed action pattern that serially links up to 25 movements into 4 predictable phases that follow 1 syntactic rule. That entire pattern became even more stereotyped and resistant to disruption in hyper-dopaminergic mutant mice. The stronger pattern was evident in several ways. First, DAT-KD mutant mice were more likely to begin a syntactic chain pattern than wild-type mice, especially during the early minutes of a grooming bout (when the highly stereotyped serial pattern is most likely to be produced), and especially in the novel laboratory environment (a potential stressor). Further, once the complex sequence began, DAT-KD mutant mice went on to execute chains that were more stereotyped and rigid, both qualitatively and quantitatively. Qualitatively, mutant mice almost always achieved the strongest form of the terminal phase (Phase IV), successfully making a transition from head grooming to body grooming. By comparison, wild-type mice ended far more of their chains with a weaker terminal substitution for Phase IV, which left them stuck in head grooming without ever making a transition to body grooming. Quantitatively, DAT-KD mutant mice returned more often to the prototypical pattern after minor mistakes, whereas wild-type mice failed to reach full Phase IV after such mistakes. Mutant mice returned more often to the full pattern after extraneous component insertion, phase omission, or serial reversal of phases. The mutants' elevated pattern strength for this complex sequence was evident in both home and laboratory environmental contexts. If the less-stereotyped sequential patterns of wild-type mice are viewed as the norm (and not as a sequential deficit), then the mutant tendency to complete stronger syntactic patterns must be viewed as sequential super-stereotypy , representing the exaggerated serial rigidity feature of compulsive behavior. Here sequential super-stereotypy is manifest in a complex behavioral sequence that is instinctive and naturally stereotyped to begin with, but becomes even more stereotyped or excessively rigid as a consequence of the DAT mutation. It may be important that the mutant pattern strength is revealed not as an elimination of errors, but rather primarily as a resistance to disruption by errors. In other words, mutants did not have more frequent perfect chains than wild-types: both generated similar moderate rates of minor errors (e.g., inserting extra actions, omitting one syntactic phase from where it ought to be, or reversing the order of 2 phases in the 4-phase pattern). Instead, the mutants' stronger syntactic pattern was like a tightened elastic band, pulling them back after such errors to finish the prototypical pattern. Stronger return to the pattern could only be possible if DAT knockdown strengthened the entire pattern as a global whole, facilitating the mutants' ability to maintain a neural representation of the pattern during an error and to resume the remaining pattern after the error. That suggests that neural mechanisms of pattern coordination were better able to persist in rule maintenance in the face of disruption, and to successfully compete to re-establish control of the behavioral stream after the disruption. Thus, stronger patterns were not simply the results of strengthened Markov sequential transitions among individual pairs of actions, producing a stimulus-response (S-R) reflex chain. If sequence composition was simply a probabilistic construction based only on the frequency of transitions between individual pairs of actions, then stronger perfect completion might have been expected in mutants, but not stronger return after an error. Errors would still terminate or weaken the pattern. Instead, we observed the opposite result: mutants kept errors but recovered better after them, and took the full pattern up again where it had left off. Relation to other nigrostriatal manipulations and behaviors These results are the first demonstration to our knowledge of sequential super-stereotypy of a complex behavioral pattern, occurring spontaneously without drugs. In previous studies, dopamine D1 agonists were needed to cause sequential super-stereotypy of syntactic grooming chains, whereas D2 agonists in contrast reduced initiation and completion of syntactic grooming chains (even though D2 agonists can cause simple repetitive movement stereotypies) [ 43 - 45 , 64 ]. Future studies will be needed to confirm whether the sequential super-stereotypy of DAT knockdown mutant mice depends specifically on increased D1 receptor activation. However, it is notable that there is a consistent trend of D1 circuit-activation inducing OCD-like behavioral persistence in D1 agonist-treated rodents, D1-circuit potentiated D1CT mice, and hyperdopaminergic DAT-KD mice [ 43 - 45 , 53 , 55 , 57 , 64 ]. This suggests that the D1 circuit may play an important role in features of compulsive behavior related to perseveration and sequential rigidity. It also would be of interest for future studies to examine if other animal models of perseverant grooming behavior, such as Hoxb8 lox and D1CT mutant mice, also show any exaggerated serial rigidity features in their fixed action patterns similar to those found here [ 13 , 51 , 53 - 55 , 57 , 65 ]. Finally, it would clearly be of interest to examine whether any other instinctive fixed action patterns belonging to those of DAT-KD mutant mice show sequential super-stereotypy similar to syntactic grooming chains. We should note that although our study is the first to produce spontaneous sequential super-stereotypy, several previous studies reported weakening of the syntactic chain pattern by other genetic manipulations. For example, the ability to complete syntactic grooming chains is impaired in several types of mutant mouse, caused by either a knockout of D1 dopamine receptors [ 41 ], or by a Weaver gene mutation that alters the nigrostriatal dopamine system [ 65 , 66 ]. In the D1 knockout study, mutant D1 mice were less able than wild-type mice to complete the full grooming pattern of syntactic chains they started [ 41 ]. Our DAT-KD findings provide an opposite demonstration to complement that D1 knockout study: DAT knockdown strengthens the same pattern presumably by elevating extra-cellular dopamine. Both results might therefore reflect essentially linear effects on the sequential stereotypy of this complex behavior pattern, mirroring up or down changes in basal ganglia dopamine neurotransmission. Evolution co-opts sequential super-stereotypy We acknowledge that there is one other known form of genetically-related sequential super-stereotypy for syntactic grooming chains. However, that sequential super-stereotypy is not caused by a single targeted gene mutation but rather is a naturally evolved adaptation of the fixed action pattern in a species of ground squirrel, Spermophilus beecheyi [ 38 ], which is probably polygenic in origin. California ground squirrels defend their individual mating territories in the Sierra mountains against other same-sex ground squirrels (especially males against other males). One of their behavioral territory displays is a specialized exapted form of the syntactic grooming chain [ 38 ]. Display forms of Spermophilus beecheyi syntactic chains are ritualized, more sequentially rigid and predictable than normal self-grooming chains, and occur as a single grooming chain with no other grooming before or after [ 38 ]. Phase III elements are amplified and made more visually distinctive, and an extra Phase V component is appended to the end of the pattern (the squirrel seizes and licks its tail, which is also visually distinctive). Syntactic grooming chains are usually performed at the boundary where two adjacent territories meet. Syntactic grooming chain displays appear to be communicative, in that they are emitted in conjunction with other territorial displays, such as scent-marking of objects, and have the social consequence of subsequently reducing the likelihood of a physical fight between the two adversaries [ 38 ]. Thus the evolution of Spermophilus beecheyi ground squirrels appears to have exapted the pre-existing pattern of a syntactic grooming chain, which likely evolved in ancestral rodents over 60 million years ago, and co-opted it into a sequentially super-stereotyped form for specific communicative use [ 38 ]. It may have been selected because of the same feature that led us to study syntactic chains, namely its recognizable sequential stereotypy. Also, the sequential pattern appears highly sensitive to the underlying genotype; for example, the detailed 'signature patterns' of the syntactic grooming chains that distinguish mice from squirrels, rats, guinea pigs and other rodents can be used to construct taxonomic trees of relatedness for them (similar to taxonomies based on differences in skull structure or in DNA sequences) [ 37 ]. The genetic sensitivity of the pattern may explain why evolutionary selection exploited it for use by California ground squirrels, and also explain why knockdown of a single gene can change the strength of the entire complex sequential pattern in studies such as ours [ 41 , 65 , 66 ]. Neural systems and clinical implications of sequential super-stereotypy Altered neurochemical signaling within basal ganglia neural circuits may be the mechanism by which DAT knockdown produces sequential super-stereotypy of grooming syntax. Electrophysiological studies have shown that neurons in neostriatum and in substantia nigra pars reticulata code the sequential pattern of syntactic grooming chains and other natural sequences of behavior [ 24 , 27 , 28 ]. For example, 40% of neostriatal neurons in rats code sequential aspects of the syntactic chain pattern, especially in anterior dorsolateral neostriatum [ 24 , 27 ]. Neurochemical boosting of dopamine signalling from substantia nigra pars compacta on to neostriatal neurons might be one candidate mechanism to modulate sequential super-stereotypy of the pattern in DAT-KD mutants. Similarly, neurons in the substantia nigra pars reticulata appear especially to code initiations of the complex behavioral sequence, and so modulated input to them might be more relevant to the elevated mutant tendency to begin the syntactic pattern [ 24 , 28 ]. Nigrostriatal mechanisms for sequencing instinctive action may have been co-opted in subsequent mammalian and human evolution into use in sequencing learned and cognitive psychological elements [ 67 - 69 ]. In that way, the same basal ganglia mechanisms used for movement syntax may participate in sequential habits that result from learning [ 20 , 29 , 70 - 72 ]. A view of basal ganglia as a general purpose sequencing mechanism is compatible also with computational sequencing models of basal ganglia [ 16 - 19 ]. Beyond the basal ganglia, DAT-KD mutant mice might also have elevated extra-cellular dopamine concentrations in other target structures, including prefrontal cortex and amygdala. Such systems might also contribute to OCD and Tourette's syndromes in humans and to some aspects of compulsive-like behavior in mutant mice. Elaborated applications of dopamine-related circuits for sequencing may thus extend from instinctive animal behavior to abstract human cognition and behavior, including syntactic sequencing of action plans, linguistic syntax, and the serial order of streams of thought [ 14 , 73 ]. A clinical implication of the embeddedness of basal ganglia in sequencing function may be a vulnerability to sequential dysfunction in some human disorders involving nigrostriatal systems [ 74 , 75 ]. Both Tourette's syndrome and obsessive-compulsive disorder show symptoms of sequential super-stereotypy, in the form of overly rigid patterns of action, language or thought [ 76 , 77 ]. Basal ganglia are believed to be involved in generating such pathologically-strong and complex sequential stereotypies [ 1 , 2 , 8 , 9 , 74 , 78 - 85 ]. Hyper-dopaminergic function in nigrostriatal and related neural systems might thus play a role in causing the excessive rigidity of behavioral tics, repetitive language utterances, and obsessive chains of thought [ 2 , 74 , 79 , 81 , 86 , 87 ]. Finally, while highly speculative, it is at least conceivable that an evolutionary specialization of dopamine-related neural mechanisms for self-grooming sequences, suggested by our current results, might also influence the theme or content, as well as the syntactic stereotypy, of some human super-stereotypies involving washing or purifying compulsions [ 74 ]. Pathologically-intense rituals of cleanliness, security behavior, or concerns with contamination, all share a focus that might relate to grooming of oneself [ 74 ]. Conceivably, excessive activation in brain circuits linked by evolution to self-grooming behavior might tip the thematic focus of some human stereotyped sequences toward rituals of cleanliness or reaction to perceived contamination, in addition to strengthening their syntactic rigidity. Whether or not such a direct overlap exists between human pathology and animal instinctive behavior, our results indicate that DAT-KD mutant mice show sequential super-stereotypy in a complex instinctive fixed action pattern. Methods Subjects DAT-KD mutant mice (n = 12 male) and wild-type control mice (n = 12 male) were generated at the University of Chicago by breeding heterozygous mutants on a 129 Sv/J genetic background as described earlier [ 58 ]. Such a design minimizes any contribution to behavioral phenotype by genetic background difference or by differences in genetic modifiers that are linked to the Slc6a3 locus. DAT knockdown was achieved by insertion of the tetracycline regulatable system into the 5' untranslated region in the second exon of the DAT gene (Slc6a3). Such an insertion reduced the DAT promoter strength without affecting its expression pattern. It also allows regulation of DAT expression by dietary tetracycline, although that feature was not used in this study. DAT knockdown reduces adult DAT expression to 10% of wild-type levels and raises extracellular dopamine levels in neostriatum to 170% (wild-type control = 100%) [ 58 ]. Once housed at the University of Michigan, mutant and wild-type mice (age 2–4 months) were allowed to habituate to their new surroundings for two weeks before any behavioral testing. Mice were housed at ~21°C on a 12 h light/dark cycle with lights on at 7 a.m., in groups of two to three same-type individuals during the laboratory environment testing phase. During the home cage testing phase of the experiments, mice were housed individually to facilitate videotaping. Food (Purina Rat Chow; St. Louis, MO) and water (tap water) were always available. Behavioral testing It was important to determine whether any sequential stereotypy difference between mutant and wild-type mice in grooming behavior was a stable difference in action syntax strength, and not merely an artifact of testing conditions. Grooming behavior of rodents is sensitive to environmental contexts, both in quantity and in fine structure, and stressors in particular can either suppress or increase grooming behavior depending on type [ 88 ]. All mice were therefore tested for grooming behavior in 2 environmental contexts: 1) a standard behavioral neuroscience laboratory chamber, and 2) their own home cages (a relatively stress-free environment). Laboratory environment Immediately prior to testing, mice were transported in their home cage on a cart down a 30 m hallway to a laboratory testing room with standard white fluorescent lighting, and placed individually in a test chamber (light intensity 550–650 lux; sound intensity 65–70 decibels measured within chamber). The laboratory test chamber consisted of a transparent cylinder (19 cm high, 12.5 cm diameter) suspended over a tilted mirror. A camera lens focused on this mirror gave a close-up view of the mouse's face, forepaws, and upper body. For behavioral testing, each mouse was placed individually in a test chamber and videotaped for 30 minutes. Each mouse received 3 habituation days in the laboratory test procedure before grooming behavior data were collected over the next 2 consecutive days in 30 min sessions. Home environment Testing in the home environment took place during the dark phase under dim red light conditions. Mice were housed singly in transparent rectangular cages (12 cm high × 19 cm long × 10 cm wide). Videotaping of grooming sequences took place from the side of the transparent home cage, for 30 min each day on 2 consecutive days, with the camera focused closely on the mouse. Behavioral video analysis Videotaped grooming behavior was scored in slow motion (frame-by-frame to 1/10 th actual speed; scorer blind to genotype) for grooming amount (cumulative durations), grooming bout number and bout length, and occurrence of syntactic chains. Syntactic grooming chains were identified and classified in frame-by-frame analysis as either Perfect, Imperfect but completed by full Phase IV (omission, insertion, or reversal types), Terminal substitution of paw lick for Phase IV body licking, or Incomplete (grooming stops before Phase IV, or reverts to sequentially flexible facial grooming and paw strokes) [ 37 , 41 ], [ 43 , 44 ]. We also made choreograph diagrams of syntactic chains from each mouse to compare details of their form and sequential pattern [ 61 ]. Behavioral data were statistically analyzed by 3-factor, 2-factor, or 1-factor ANOVA as indicated above. When significant results were obtained, post hoc paired comparisons were subsequently performed using Bonferroni or Tukey tests (alpha set equal to original 0.05 level). Authors' contributions KCB conceived and supervised the study and drafted the manuscript; JWA co-conceived the study and participated in interpretation and writing; KRH carried out behavioral testing, videoanalysis, and statistics; XZ developed and generated the mutant mice, and participated in writing the manuscript. Supplementary Material Additional File 1 Movie: Sequential super-stereotypy of an instinctive fixed action pattern in hyper-dopaminergic mutant mice. Windows Media Player movie file (.avi): DAT Knockdown grooming fixed action pattern.avi Examples of syntactic grooming chains performed by three hyperdopaminergic mutant mice are shown in the accompanying movie file. Choreograph diagrams of component movements' form and sequence are displayed for each syntactic chain, and strokes are illuminated sequentially in synchrony with their corresponding movements. Note that the first two syntactic chains contain insertion or reversal errors (Mutant mouse 1: paw lick insertions in Phase II, between Phases II and III, and between Phases III and IV; also reversal insertion of a Phase I ellipse stroke within Phase II. Mutant mouse 2: paw lick insertions within Phase I, within Phase III, and between Phases III and IV). However, the syntactic chains are not disrupted by these errors, and the mutants continue on with the sequential pattern to successfully complete Phase IV (body licking). Mutant mouse 3 also shows the ventral view that permits the viewer to see both forepaws simultaneously, which was used to score all syntactic chains in the laboratory. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC552313.xml |
522810 | Inter-Platform comparability of microarrays in acute lymphoblastic leukemia | Background Acute lymphoblastic leukemia (ALL) is the most common pediatric malignancy and has been the poster-child for improved therapeutics in cancer, with life time disease-free survival (LTDFS) rates improving from <10% in 1970 to >80% today. There are numerous known genetic prognostic variables in ALL, which include T cell ALL, the hyperdiploid karyotype and the translocations: t(12;21)[ TEL-AML1 ], t(4;11)[ MLL-AF4 ], t(9;22)[ BCR-ABL ], and t(1;19)[ E2A-PBX ]. ALL has been studied at the molecular level through expression profiling resulting in un-validated expression correlates of these prognostic indices. To date, the great wealth of expression data, which has been generated in disparate institutions, representing an extremely large cohort of samples has not been combined to validate any of these analyses. The majority of this data has been generated on the Affymetrix platform, potentially making data integration and validation on independent sample sets a possibility. Unfortunately, because the array platform has been evolving over the past several years the arrays themselves have different probe sets, making direct comparisons difficult. To test the comparability between different array platforms, we have accumulated all Affymetrix ALL array data that is available in the public domain, as well as two sets of cDNA array data. In addition, we have supplemented this data pool by profiling additional diagnostic pediatric ALL samples in our lab. Lists of genes that are differentially expressed in the six major subclasses of ALL have previously been reported in the literature as possible predictors of the subclass. Results We validated the predictability of these gene lists on all of the independent datasets accumulated from various labs and generated on various array platforms, by blindly distinguishing the prognostic genetic variables of ALL. Cross-generation array validation was used successfully with high sensitivity and high specificity of gene predictors for prognostic variables. We have also been able to validate the gene predictors with high accuracy using an independent dataset generated on cDNA arrays. Conclusion Interarray comparisons such as this one will further enhance the ability to integrate data from several generations of microarray experiments and will help to break down barriers to the assimilation of existing datasets into a comprehensive data pool. | Background The advent of DNA microarrays has provided the science community with a tool to concurrently examine the expression of thousands of genes within a given cell or tissue type, thus providing a platform for future diagnoses and prognostic analyses of disease with gene-level specificity [ 1 , 2 ]. Microarray technology is progressing rapidly as better sequencing and prediction algorithms allows for more refined gene prediction. This has prompted the evolution of the probe sets contained within the array chips over the past few years, in both oligonucleotide and cDNA arrays [ 3 ]. This microarray platform expansion has hindered the direct comparison between numerous datasets of a given phenotype that have been produced using several generations of arrays. In microarray analyses of disease, having a large number of samples better accounts for the biological variability between individuals and therefore increases the power to enhance and define a pathogenetic model for that disease. Due to the considerable expense of microarray chips and the equipment required, along with the common problem of sufficient sample acquisition, being able to combine and compare datasets from various laboratories and across all microarray generations would be a benefit to the entire biomedical community. The constant evolution of microarrays has thus resulted in a significant hindrance to their power as a research or diagnostic tool by dividing datasets according to platform and seemingly limiting their interarray comparability. With the large number of microarray datasets available in the public domain for distinct disease phenotypes from various microarray platforms, cross-platform comparisons can currently be attempted. For example, a number of laboratories have been studying diagnostic pediatric acute lymphoblastic leukemia (ALL) samples from human bone marrow on both oligonucleotide and cDNA microarrays and depositing the raw intensity values into the public domain. Additionally, in 2002 at St. Jude Children's Research Hospital, Yeoh et al . generated a list of genes that have distinct expression levels for various karyotypic and phenotypic aberrations common to pediatric ALL [ 4 ]. This set of genes has been useful as a prognostic profile for ALL by identifying subclasses of the cancer using microarray technology [ 4 ]. Importantly, the St. Jude's gene list has been validated on independent datasets both within their own lab and in an independent laboratory by Kohlmann et al . in 2004. They used the Yeoh et al gene list to successfully segregate the various subclasses of adult ALL [ 5 ]. Consequently, this disease provides an excellent model for testing interarray comparability using one common gene list. ALL is the most common pediatric malignancy comprising over 75% of the annual diagnoses of leukemias in children [ 6 ]. In the United States, the outcome for children with ALL has improved dramatically over the past thirty years with the long-term disease-free survival (LTDFS) rates increasing from less than 10% in 1970 to over 80% today [ 7 ]. However, ALL still carries the risk of relapse in over 20% of patients [ 8 ]. ALL survival is largely due to a greater understanding of the risk factors that affect outcome, which has allowed for more intensity-tailored treatment following an assessment of the patient's risk [ 7 ]. Accurate segregation of patients into their proper risk group is critical to allow for a risk-stratified treatment that is effective enough to clear the disease and decrease the risk of relapse while minimizing the negative long-term side effects [ 7 ]. Factors that affect prognosis are age, sex, race, white blood cell count at diagnosis, phenotypic differences, such as T-cell versus B-cell lineage ALL and karyotypic alterations, such as the hyperdiploid karyotype and the translocations t(12;21)[ TEL-AML1 ], t(4;11)[ MLL-AF4 ], t(9;22)[ BCR-ABL ], and t(1;19)[ E2A-PBX ] [ 7 , 9 ]. These genetic lesions can affect the individual's response to treatment. For example, patients with a hyperdiploid karyotype and those with the TEL-AML1 fusion gene have a better prognosis than patients in the other subclasses [ 10 ]. The initial diagnosis and classification of ALL is currently revealed through multiple time-consuming and expensive tests often involving multiple laboratories [ 11 ]. Thus, a tool that could consolidate these tests into one diagnostic platform would be beneficial to both researchers and clinicians working with ALL. In this study, we sought to determine if datasets from different microarray platforms could be compared in a useful manner. We chose to study pediatric ALL because there is already a substantial pool of datasets freely available in the public domain. First, we collected pediatric ALL array data and cDNA array data generated in experiments from various laboratories. In addition to the collection of public data, we supplemented the hyperdiploid karyotype and the T-cell lineage ALL subclasses by expression profiling additional diagnostic pediatric ALL samples from a tumor bank in Children's National Medical Center in Washington, D.C. We used these independent datasets, including the one generated in our lab, to validate the gene predictors, as defined by Yeoh et al . (2002), for each of the aforementioned prognostic genetic variables. Cross-platform array validation was used successfully to ascertain the accuracy, sensitivity and specificity of the gene predictors for the prognostic variables. In addition, we have demonstrated the ability to compare datasets from different microarray platforms. To our knowledge, this is among the first known successful applications of this technique, along with the validation of the Yeoh et al pediatric ALL gene lists on adult ALL by Kohlmann, et al (2004) [ 5 ]. Interarray comparisons such as these will further enhance the ability to integrate data from several generations of microarray experiments and will help to break down barriers to the assimilation of existing datasets into a comprehensive data pool. Results and discussion Expression profiling of ALL diagnostic bone marrow To supplement the ALL subclasses that are under-represented in expression profiling thus far, we collected and extracted the total RNA from sixteen diagnostic bone marrow samples housed at Children's National Medical Center; seven of the hyperdiploid karyotype and nine with T-cell lineage. The extracts were then hybridized to Affymetrix U133A arrays and expression profiled as an independent training data set. Validation of the gene predictors using independent datasets spanning various array platforms In order to validate the portability of gene predictors across microarray platforms we compared the accuracy with which the six prevalent ALL subclasses can be distinguished on disparate array platforms. To do this we used the discriminating gene lists (~40 genes), which were provided by the comprehensive training ALL sample set analyzed and published by Yeoh et al . in 2002 [ 4 ]. In their study they hybridized RNA from ALL bone marrow samples to Affymetrix U95Av2 arrays. The resulting expression data were analyzed by multiple statistical methods to facilitate the generation of lists of genes that represent the greatest difference in expression between the ALL subclasses [ 4 ]. Yeoh et al . used both a training and test dataset in their analysis to first uncover the subclass-specific gene expression profiles and then to test their predictability on independent samples [ 4 ]. The genes are listed hierarchically, along with supplemental information about the statistical methods used, at . We then accumulated the ALL diagnostic bone marrow array data available in the public domain (Table 1 ). Table 1 Training and test datasets used to validate ALL subclass predictors Training Sample Set Microarray Platform Yeoh et al . Cancer Cell , 2002, 1: 133–143 Affymetrix HG_U95Av2 Validation Sample Set Microarray Platform Predictors Armstrong et al . Nat. Genet ., 2002, 30(1): 41–7 Affymetrix HG_U95Av2 Hyperdiploid, MLL-AF4 , TEL-AML1 Mitchell et al . Unpublished data 2003 Affymetrix HG_U133A Hyperdiploid, T-ALL Stephan DA, Golub TR. Unpublished data 2000 Affymetrix HuGene FL TEL-AML1, E2A-PBX1 Golub et al . Science , 1999, 286: 531–7 Affymetrix HuGene FL T-ALL Ramaswamy et al . Proc. Natl. Acad. Sci. USA , 1999, 98(26): 15149–54 Affymetrix Hu6800 and Hu35KsubA T-ALL Moos et al . Clin. Cancer Res ., 2002, 8: 3118–3130 cDNA TEL-AML1, MLL-AF4, BCR-ABL1 , T-ALL Catchpoole et al . Unpublished data 2002. cDNA T-ALL The independent datasets that we accumulated, including the one generated in our lab, spanned four different microarray platforms: Affymetrix HuGene FL, U95Av2, U133A and custom cDNA microarray platforms (Table 1 ). To modify these test datasets into data that could be directly applied to the predictor gene lists from the U95Av2 arrays, we correlated the probe numbers between these different arrays and the U95Av2 set using the probe match spreadsheet, NetAffx, available at . We then used the discriminating gene list for each subclass to extract the appropriate probes and their intensity values from the expansive expression data for each sample of the validation datasets independently. The level of similarity between the probe sets of the two different array platforms was evident through the number of genes within the 40 discriminators that could be found within the validation data (Table 2 ). For example, the data published by Armstrong et al . (2002) was generated on the U95Av2 array platform [ 6 ]. Therefore, expression data for all 40 predictor genes could be correlated and represented in their corresponding MLL-AF4 , TEL-AML1 , and hyperdiploid datasets. Similarly, the U133A arrays that were used to generate expression data in our lab for the hyperdiploid karyotype and the T-cell lineage ALL contained probes representing the majority of the 40 discriminators, with 38 and 35 genes, respectively. The HuGene FL arrays contain significantly fewer probe sets in common with the selected predictors (from the later-generation U95Av2 microarrays). Accordingly, of the 40 original predictor probes, only 25 were present in the TEL-AML1 dataset, 26 in the E2A-PBX1 dataset and 13 in the T-cell dataset. The difficult task of matching probes from the Affymetrix gene chips with cDNA arrays was illustrative of the disparities between the probe sets within these two platforms. For example, the five cDNA predictors built using the datasets produced by Moos et al . (2002) and Catchpoole et al . (unpublished data) contained data for only ten genes or less from the predictor set gene list. Table 2 Prediction accuracies for ALL subclasses as determined by the different microarray platforms. ALL Subclass Microarray Platform # of Samples in the Dataset # of Samples Representing the Predictor Subclass # of Genes in Predictor (out of 40) 1 Accuracy (%) 2 Sensitivity (%) 3 Specificity (%) 4 Hyperdiploid Affymetrix U95Av2 43 a 5 40 97 80 100 Hyperdiploid Affymetrix U133A 16 b 7 38 94 86 100 T-ALL Affymetrix U133A 16 b 9 35 100 100 100 T-ALL Affymetrix HuGene FL 41 c 8 13 100 100 100 T-ALL Affymetrix Hu6800 20 d 10 30 95 100 90 T-ALL cDNA 52 e 7 5 98 86 100 T-ALL cDNA 9 f 3 29 100 100 100 TEL-AML1 Affymetrix U95Av2 43 a 9 40 91 67 97 TEL-AML1 Affymetrix HuGene FL 23 g 14 30 86 79 100 TEL-AML1 cDNA 52 e 12 10 87 83 88 MLL-AF4 Affymetrix U95Av2 43 a 20 40 100 100 100 MLL-AF4 cDNA 52 e 2 7 98 50 100 E2A-PBX1 Affymetrix HuGene FL 23 c 2 26 96 50 100 1 With a few exceptions, the majority of the gene lists published by Yeoh et al (2002) contain 40 genes. 2 The ability of the predictor to correctly classify the blinded test set into the correct subgroup 3 (# of positive samples predicted correctly)/(total #of true positives) 4 (# of negative samples predicted correctly)/(total #of true negatives) a Armstrong et al. (2002) Nat. Genet . 30(1) , 41–7. b Mitchell et al. (2003) Unpublished data. c Golub et al. (1999) Science 286 , 531–7. d Ramaswamy et al. (2001) PNAS 98(26) , 15149–54. e Moos et al. (2002) Clin Cancer Res . 8 , 3118–3130. f Catchpoole et al. Unpublished data. g Stephan et al. (2000) Unpublished data. To validate the gene predictors from Yeoh et al . (2002), using the aforementioned independent test datasets from various array platforms, we employed supervised learning methods using GeneCluster2 software. Prior to analysis, we formatted the discriminating gene expression values from the test datasets onto spreadsheets according to software instruction, and subsequently applied the data to the software. Genecluster2 then generated blinded predictions on the ALL samples of the test datasets through weighted voting with a leave-one-out methodology. This is accomplished by randomly removing one sample at a time from the test dataset of ALL samples and "training" a predictor gene profile to recognize similarities or disparities between the two classes based on the expression profiles of the samples for the genes of interest [ 11 ]. In this manner each sample is assigned to one of the two classes based on their expression pattern of the predictor genes. The prediction accuracy, sensitivity and specificity were calculated for each of the predictors from each array platform and are displayed in both figure 1 and table 2 . The accuracy of our predictors ranged from 86%–100%, with a mean accuracy of 95%. The mean specificity of the predictors was 98%, ten of which provided a specificity of 100%. The sensitivity ranged from 50%–100%. The mean sensitivity was 83% (fig. 1 , table 2 ). Figure 1 Summary of results of the various ALL subclass predictors tested. The predictors are organized according to microarray platform and the results are listed under each class in terms of the accuracy, sensitivity and specificity of the classification. We saw a high accuracy from the predictors employing data from both U95Av2 and U133A arrays, attesting to the fact that nearly all of the 40 discriminating genes were present in the datasets, thus maximizing the possible prediction strength. In the case of the E2A-PBX1 predictor (96%) and the MLL-AF4 predictor (98%), the sensitivities were only 50%. In both cases there were only 2 samples out of the sample pool expressing the respective translocation and in both analyses one of the two was continuously classified incorrectly. This could be due to many factors, misdiagnosis or mislabelling of the sample, poor sample quality or differences in sample handling. It is difficult to draw a conclusion due to the fact that the samples were collected and processed in a laboratory outside of our own. Another problem with these two predictors may simply be the low sample number. Two samples may not provide enough strength to the classification by Genecluster2 simply due to the inability of such a low sample number to account for the biological variability that exists between patients that is independent of their subclass of ALL. The most surprising result was the high accuracy with which the gene lists could classify T-ALL, TEL-AML1 and MLL-AF4 from cDNA data considering the disparity between the probe sets of cDNA arrays and oligo arrays. The accuracies of the classifiers were: T-ALL (Catchpoole data), 100%; T-ALL (Moos data), 98%; MLL-AF4 , 98%; and TEL-AML1 , 87%. Therefore, it appears that the number of genes in the predictor gene list is much less of a factor in the predictor's classification accuracy than the number of samples representing the phenotype of interest, which supports the argument that being able to do cross-platform analyses to increase sample size is crucial for sensitive and specific class prediction using expression data. This is strongly illustrated by the cDNA predictors, which have few probes in common with the arrays used to generate the predictor gene list, but still classify the ALL samples with high sensitivity, specificity and accuracy. On the other hand the E2A-PBX1 predictor and the MLL-AF4 predictor had low sensitivities correlating to a low sample number in these groups. The high number of probes in common between the arrays used to generate these independent datasets and the arrays used to generate the predictor gene list, 26/40 and 40/40, respectively, were not able to rescue the low sensitivity of the predictor. Conclusions Currently the vast majority of expression data from numerous labs is not being used to its highest potential as independent labs continue to move to more expansive array platforms rendering older datasets less informative in the context of new data. Increasingly, progressive technologies in genome databasing and chip construction are prompting this inevitable evolution of microarrays. Until data can be analyzed and directly compared across array platforms, the size of the data pools will remain small and isolated according to platform [ 12 ]. Here we have shown that the previously validated predictor gene list from Yeoh et al . (2002) withstands validation by testing the predictors using a leave-one-out strategy on all publicly available datasets as well as a dataset generated in our own lab regardless of the array platform used. This meta-data analysis of over 200 arrays from diagnostic ALL samples with hyperdiploidy, T-cell lineage and translocation status (previously confirmed through gold standard techniques), shows that expression profiling as an integrated platform is robust and that ALL data, and presumably other disease models, can be interplatform comparable. By validating the comparability between data from distinct microarray platforms we have demonstrated a tool that can enhance the statistical power provided by large sample sets. Thus, we can potentially develop and validate sensitive diagnostic tools based on large training sample sets, to allow for the rapid assignment of individualized therapy to improve disease outcome in pediatric ALL and other diseases. Methods RNA extraction from bone marrow samples ALL diagnostic bone marrow samples were housed in a tumor bank in Children's National Medical Center in Washington, D.C. Mononuclear cells from diagnostic bone marrow aspirates were separated using density centrifugation on Cappel Lymphocyte Separation Medium (ICN Biomedicals, Aurora, Ohio) and immediately flash frozen according to manufacturer's instruction. A total of sixteen samples were obtained with IRB approval; seven with a hyperdiploid karyotype and nine samples of a T-cell lineage as confirmed by immunophenotyping. The frozen samples were placed directly in TRIzol reagent for RNAse-free thawing for total RNA extraction. We extracted a 10 μg–20 μg pellet of total RNA from each sample by centrifugation following phenol-chloroform extraction. The integrity of the resultant total RNA from each sample was quantified by gel electrophoresis before it was considered to be of good quality for cDNA synthesis. Samples were re-extracted if ribosomal bands were not visible. Expression profiling and support vector machine meta-analysis 10 μg of the extracted RNA from each sample was labelled and hybridized to an Affymetrix U133A array (Affymetrix, Santa Clara, CA) according to protocol as previously described [ 13 ]. Intensity values were calculated using Microarray Suite 5.0 (MAS 5.0) and expression values were adjusted to fall within the lower and upper limits of 1 and 45000 as described by Yeoh et al . (2002) [ 4 ]. To create a predictor that allows for the direct comparison between different generation Affymetrix arrays and cDNA arrays, we used the predictor gene list for each subclass provided by Yeoh et al . (2002) from Affymetrix U95Av2 microarrays. The 40 genes that showed the greatest mean difference in expression between the subclass of interest and the remaining subclasses was used as our predictor gene set. The gene lists and additional information, including the statistical metrics used to generate the gene list from the training set, can be viewed at: . To identify comparable data points between the gene lists from the training set (produced on the U95Av2 Affymetrix chip), and the expression values of samples provided by other public datasets on different generation Affymetrix arrays, we used the probe match function within NetAffx . Data for these probe pairs in the validation sets were extracted and expression values were linearly adjusted to fall within 1–45000 [ 4 ]. Affymetrix probes were identified within cDNA data by a combination of BLAST sequence comparison and GenBank accession number queries. Ratios were log-transformed prior to analysis. GeneCluster2 ( ; Center for Genome Research, MIT, Cambridge, MA) was used to perform blinded predictions on the validation dataset using weighted voting with a leave-one-out methodology. Accuracy, specificity and sensitivity values were then generated for each predictor, as a measure of the predictor's ability to correctly group the samples into their respective class in the validation sets. Authors' contributions SAM carried out the accumulation of datasets, preparation of the predictors, data analysis and drafted the manuscript. SAM with the help of KMB and MMH participated in the sample preparation and hybridization to arrays of the in-house gene expression data. KMB also trained SAM and participated in the preparation of the predictors with SAM and also provided essential mentor support. MM participated in finding and extracting the public data and aided in drafting the manuscript. MMH was a tremendous help in editing the final manuscript. BL provided statistical guidance throughout the project. DC provided us with cDNA array data and aided in the analysis of the cDNA expression data. DC also provided much appreciated critical input throughout the entire project. DAS initiated and crafted the idea and provided the necessary mentorship along the way. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC522810.xml |
545647 | Systems for grading the quality of evidence and the strength of recommendations I: Critical appraisal of existing approaches The GRADE Working Group | Background A number of approaches have been used to grade levels of evidence and the strength of recommendations. The use of many different approaches detracts from one of the main reasons for having explicit approaches: to concisely characterise and communicate this information so that it can easily be understood and thereby help people make well-informed decisions. Our objective was to critically appraise six prominent systems for grading levels of evidence and the strength of recommendations as a basis for agreeing on characteristics of a common, sensible approach to grading levels of evidence and the strength of recommendations. Methods Six prominent systems for grading levels of evidence and strength of recommendations were selected and someone familiar with each system prepared a description of each of these. Twelve assessors independently evaluated each system based on twelve criteria to assess the sensibility of the different approaches. Systems used by 51 organisations were compared with these six approaches. Results There was poor agreement about the sensibility of the six systems. Only one of the systems was suitable for all four types of questions we considered (effectiveness, harm, diagnosis and prognosis). None of the systems was considered usable for all of the target groups we considered (professionals, patients and policy makers). The raters found low reproducibility of judgements made using all six systems. Systems used by 51 organisations that sponsor clinical practice guidelines included a number of minor variations of the six systems that we critically appraised. Conclusions All of the currently used approaches to grading levels of evidence and the strength of recommendations have important shortcomings. | Background In 1979 the Canadian task Force on the Periodic Health Examination published one of the first efforts to explicitly characterise the level of evidence underlying healthcare recommendations and the strength of recommendations [ 1 ]. Since then a number of alternative approaches has been proposed and used to classify clinical practice guidelines [ 2 - 28 ]. The original approach used by the Canadian Task Force was based on study design alone, with randomised controlled trials (RCTs) being classified as good (level I) evidence, cohort and case control studies being classified as fair (level II) evidence and expert opinion being classified as poor (level III) evidence. The strength of recommendation was based on the level of evidence with direct correspondence between the two; e.g. a strong recommendation (A) corresponded to there being good evidence. A strength of the original Canadian Task Force approach was that it was simple; the main weakness was that it was too simple. Because of its simplicity, it was easy to understand, apply and present. However, because it was so simple there were many implicit judgements, including judgements about the quality of RCTs, conflicting results of RCTs, and convincing results from non-experimental studies. For example: • Should a small, poorly designed RCT be considered level I evidence? • Should RCTs with conflicting results still be considered level I evidence? • Should observational studies always be considered level II evidence, regardless of how convincing they are? The original approach by the Canadian Task Force also did not include explicit judgements about the strength of recommendations, such as how trade-offs between the expected benefits, harms and costs were weighed and taken account of in going from an assessment of how good the evidence is to determining the implications of the results for practice. The GRADE Working Group is an informal collaboration of people with an interest in addressing shortcomings such as these in systems for grading evidence and recommendations. We describe here a critical appraisal of six prominent systems and the results of the critical appraisal. Methods We selected systems for grading the level of evidence and the strength of recommendations that we considered prominent and that included features not captured by other prominent systems. These were selected based on the experience and knowledge of the authors through informal discussion. A description of the most recent version (as of summer 2000) of each of these systems (Appendix 1 to 6), was prepared by one of the authors familiar with the system, and used in this exercise. The following six systems were appraised: the American College of Chest Physicians (ACCP, [see Additional file 1 ]) [ 21 ], Australian National Health and Medical Research Council (ANHMRC, [see Additional file 2 ]) [ 17 ], Oxford Centre for Evidence-Based Medicine (OCEBM, [see Additional file 3 ]) [ 16 ], Scottish Intercollegiate Guidelines Network (SIGN, [see Additional file 4 ]) [ 18 ], US Preventive Services Task Force (USPSTF, [see Additional file 5 ]) [ 22 ], US Task Force on Community Preventive Services (USTFCPS, [see Additional file 6 ]) [ 25 ]. These descriptions of the systems were given to the twelve people who independently appraised the six systems, all of the authors minus GEV appraised the six systems, three of the authors (DH, SH and DO'C) appraised as a group and reported as one (see contributions). The 12 assessors all had experience with at least one system and most had helped to develop one of the six included systems. Twelve criteria described by Feinstein [ 29 ] provided the basis for assessing the sensibility of the six systems. Criteria used to assess the sensibility of systems for grading evidence and recommendations 1. To what extent is the approach applicable to different types of questions? -effectiveness, harm, diagnosis and prognosis (No, Not sure, Yes) 2. To what extent can the system be used with different audiences? -patients, professionals and policy makers (Little extent, Some extent, Large extent) 3. How clear and simple is the system? (Not very clear, Somewhat clear, Very clear) 4. How often will information not usually available be necessary? (Often, Sometimes, Seldom) 5. To what extent are subjective decisions needed? (Often, Sometimes, Seldom) 6. Are dimensions included that are not within the construct (level of evidence or strength of recommendation)? (Yes, Partially, No) 7. Are there important dimensions that should have been included and are not? (No, Partially, Yes) 8. Is the way in which the included dimensions are aggregated clear and simple? (No, Partially, Yes) 9. Is the way in which the included dimensions are aggregated appropriate? (No. Partially, Yes) 10. Are the categories sufficient to discriminate between different levels of evidence and strengths of recommendations? (No, Partially, Yes) 11. How likely is the system to be successful in discriminating between high and low levels of evidence or strong and weak recommendations? (Not very likely, Somewhat likely, Highly likely) 12. Are assessments reproducible? (Probably not, Not sure, Probably) No training was provided and we did not discuss the 12 criteria prior to applying them to the six systems. Our independent appraisal of the six systems were summarised and discussed. The discussion focused on differences in the interpretation of the criteria, disagreement about the judgements that we made and sources of these disagreements, the strengths and weaknesses of the six systems, and inferences based on the appraisals and subsequent discussion. In order to identify important systems that we might have overlooked following our appraisal of these six systems we also searched the US Agency for Health Care Research and Quality (AHRQ) National Guidelines Clearing House for organisations that have graded two or more guidelines in the Clearing House using an explicit system [ 30 ]. These systems were compared with the six systems that we critically appraised. Results There was poor agreement among the 12 assessors who independently assessed the six systems. A summary of the assessments of the sensibility of the six approaches to rating levels of evidence and strength of recommendation is shown in Table 1 . Discussion The poor agreement among the assessors likely reflects several factors. Some of us had practical experience using one of the systems or used additional background information related to one or more grading systems, and we may have been biased in favour of the system with which we were most familiar. Each criterion was applied to grading both evidence and recommendations. Some systems were better for one of these constructs than the other and we may have handled these discrepancies differently. In addition each criterion may have been assessed relative to different judgements about the evidence, such as an assessment of the overall quality of evidence for an important outcome (across studies) versus the quality of an individual study. Some of the criteria were not clear and were interpreted or applied inconsistently. For example, a system might be clear and not simple or visa versa. We likely differed in how stringently we applied the criteria. Finally, there was true disagreement. There was agreement that the OCEBM system works well for all four types of questions. There was disagreement about the extent to which the other systems work well for questions other than effectiveness. It was noted that some systems are not intended to address other types of questions and it is not clear that it is important that a system should address all four types of questions that we considered (effectiveness, harm, diagnosis, prognosis), although criteria for assessing individual studies must take this into account [ 31 , 32 ]. Most of us did not find that any of the systems are likely to be suitable for use by patients. Almost all agreed that the ACCP system was suitable for professionals and most considered that the USPSTF system was suitable for professionals. There was not much agreement about the suitability of any of the other systems for professionals or about the suitability of any of the systems for policy makers, although most assessed the USTFCPS system to be suitable for policy makers. There was no agreement that any of the systems are clear and simple, although USPSTF, ACCP and SIGN systems were generally assessed more favourably in this regard. It was generally agreed that the clearer a system was the less simple it was; e.g. the OCEBM system is clear but not simple for categorising the level of evidence. There was some confusion regarding whether we were assessing how clear and simple the system was to guideline developers (as some interpreted this criterion) or how clear and simple the outcome of applying the system was to guideline users (as others interpreted this criterion). Either way, the simpler a system is the less clear it is likely to be. Most of us judged that for most of the systems necessary information would not be available at least sometimes. The OCEBM system came out somewhat better than the other systems and lack of availability of necessary information was considered to be less of a problem for the USTFCPS system. However, the OCEBM and USTFCPS systems were considered by most to be missing dimensions which may, in part, explain why missing information was considered to be less of a problem. This would be the case to the extent the missing dimensions were the ones for which information would often or sometimes not be available. The dimension for which we considered that information would most often be missing was trade-offs; i.e. knowledge of the preferences or utility values of those affected. Additional problems were identified in relationship to complex interventions and counselling, particularly with the USTFCPS and USPSTF systems. It was pointed out that the USTFCPS system addressed this problem by including availability of information about the intervention as part of its assessment of the quality of evidence. Most of the systems were assessed to require subjective decisions at least to some extent. The OCEBM system again stood out as being assessed more favourably, although it may be related to omission of dimensions that require more subjective decisions. Judgement is clearly needed with any system. The aim should be to make judgements transparent and to try to protect against bias in the judgements that are made by being systematic and explicit. Inclusion of dimensions that are not within the constructs being graded was not considered a problem for most of the systems by most of us. Several people considered that it might be a problem for the USTFCPS and USPSTF systems. On the other hand, all of the systems were evaluated to be missing at least one important dimension by at least one person. The challenge of missing dimensions were considered less of a problem for the ACCP and ANHMRC systems. There was not agreement about any of the systems having a clear and simple approach to aggregating the dimensions, although this was considered to be less of a problem for the ACCP, SIGN and USTFCPS systems. There was also not agreement on the appropriateness of how the dimensions were aggregated. This was considered to be more of a problem for the ANHMRC and USTFCPS systems than the other four systems, all of which were considered to have taken an approach to aggregating the dimensions that was at least partially inappropriate by more than half of us. Most of us considered that most of the systems had sufficient categories, with the exception of the ANHMRC system. There was almost agreement that the USPSTF system has sufficient categories. We agreed that it is possible to have too many categories as well as too few, the OCEBM system being an example of having too many categories. There was not agreement that any of the systems are likely to discriminate successfully, although everyone thought that the ACCP, SIGN and USPSTF systems are somewhat to highly likely to discriminate. Lastly, we largely agreed that we were not sure how reproducible assessments are using any of the systems, although half of us considered that assessments using the ANHMRC system are unlikely to be reproducible and about 1/3 considered that assessments using the OCEBM and ACCP systems are likely to be reproducible. We identified 22 additional organisations that have produced 10 or more practice guidelines using an explicit approach to grade the level of evidence or strength of recommendations. Another 29 have produced between two and nine guidelines using an explicit approach. These systems include a number of minor variations of the six systems that we appraised in detail. There was generally poor agreement between the individual assessors about the scoring of the six approaches using the 12 criteria. However, there was general agreement that none of these six prominent approaches to grading the levels of evidence and strength of recommendations adequately addressed all of the important concepts and dimensions that we thought should be considered. Although we limited our appraisal to six systems all of the additional approaches to grading levels of evidence and strength of recommendations that we identified were, in essence, variations of the six approaches that we had critically appraised. Therefore we are confident that we did not miss any important grading systems available at the time when these assessments were undertaken. Based on discussions following the critical appraisal of these six approaches, we agreed on some conclusions: • Separate assessments should be presented for judgements about the quality of the evidence and judgements about the balance of benefits and harms. • Evidence for harms should be assessed in the same way as evidence for benefits, although different evidence may be considered relevant for harms than for benefits; e.g. local evidence of complication rates may be considered more relevant than evidence of complication rates from trials for endarterectomy. • Judgements about the quality of evidence should be based on a systematic review of the relevant research. • Systematic reviews should not be included in a hierarchy of evidence (i.e. as a level or category of evidence). The availability of a well-done systematic review does not correspond to high quality evidence, since a well-done review might include anything from no studies to poor quality studies with inconsistent results to high quality studies with consistent results. • Baseline risk should be taken into consideration in defining the population to whom a recommendation applies. Baseline risk should also be used transparently in making judgements about the balance of benefits and harms. When a recommendation varies in relationship to baseline risk, the evidence for determining baseline risk should be assessed appropriately and explicitly. • Recommendations should not vary in relationship to baseline risk if there is not adequate evidence to guide reliable determinations of baseline risk. Conclusions Based on discussions of the strengths and limitations of current approaches to grading levels of evidence and the strength of recommendations, we agreed to develop an approach that addresses the major limitations that we identified. The approach that the GRADE Working Group has developed is based on the discussions following the critical appraisal reported here and a pilot study of the GRADE approach [ 33 ]. Based on the pilot testing and the discussions following the pilot, the GRADE Working Group has further developed the GRADE system to its present format [ 34 ]. The GRADE Working Group has continued to grow as an informal collaboration that meets one or two times per year. The group maintains web pages and a discussion list. Competing interests DA has competing interests with the US Preventive Services Task Force (USPSTF), PAB has a competing interest with the US Task Force on Community Preventive Services (USTFCPS), GHG and HS have competing interests with the American College of Chest Physicians (ACCP), DH, SH and DO'C have competing interests with the Australian National Health and Medical Research Council (ANHMRC), BP has competing interests with the Oxford Centre for Evidence-Based Medicine (OCEBM). Most of the other members of the GRADE Working Group have experience with the use of one or more systems of grading evidence and recommendations. Contributions DA, PAB, ME, SF, GHG, DH, SH, AL, DO'C, ADO, BP, HS, TTTE, GEV & JWW Jr as members of the GRADE Working Group have contributed to the preparation of this manuscript and the development of the ideas contained herein, participated in the critical assessment, and read and commented on drafts of this article. GHG and ADO have led the process. GEV has had primary responsibility for coordinating the process. Pre-publication history The pre-publication history for this paper can be accessed here: Supplementary Material Additional File 1 American College of Chest Physicians (ACCP), a brief description of the ACCP approach. Click here for file Additional File 2 Australian National Health and Medical Research Council (ANHMRC), a brief description of the ANHMRC approach. Click here for file Additional File 3 Oxford Centre for Evidence-based Medicine (OCEBM), a brief description of the OCEBM approach. Click here for file Additional File 4 Scottish Intercollegiate Guidelines (SIGN), a brief description of the SIGN approach. Click here for file Additional File 5 U.S. Preventive Services Task Force (USPSTF), a brief description of the USPSTF approach. Click here for file Additional File 6 U.S. Task Force on Community Preventive Services (USTFCPS), a brief description of the USTFCPS approach. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC545647.xml |
544559 | Role of the p53/p21 system in the response of human colon carcinoma cells to Doxorubicin | Background Colon adenocarcinomas are refractory to a number of widely used anticancer agents. Multifactorial mechanisms have been implicated in this intrinsically resistant phenotype, including deregulation of cell death pathways. In this regard, the p53 protein has a well established role in the control of tumor cell response to DNA damaging agents; however, the relationship between p53-driven genes and drug sensitivity remains controversial. The present study investigates the role of the p53/p21 system in the response of human colon carcinoma cells to treatment with the cytotoxic agent doxorubicin (DOX) and the possibility to modify the therapeutic index of DOX by modulation of p53 and/or p21 protein levels. Methods The relationship between p53 and p21 protein levels and the cytotoxic effect of DOX was investigated, by MTT assay and western blot analysis, in HCT116 (p53-positive) and HT29 (p53-negative) colon cancer cells. We then assessed the effects of DOX in two isogenic cell lines derived from HCT116 by abrogating the expression and/or function of p53 and p21 (HCT116-E6 and HCT116 p21-/-, respectively). Finally, we evaluated the effect of pre-treatment with the piperidine nitroxide Tempol (TPL), an agent that was reported to induce p21 expression irrespective of p53 status, on the cytotoxicity of DOX in the four cell lines. Comparisons of IC50 values and apoptotic cell percentages were performed by ANOVA and Bonferroni's test for independent samples. C.I. calculations were performed by the combination Index method. Results Our results indicate that, in the colon carcinoma cell lines tested, sensitivity to DOX is associated with p21 upregulation upon drug exposure, and DOX cytotoxicity is potentiated by pre-treatment with TPL, but only in those cell lines in which p21 can be upregulated. Conclusions p21 induction may significantly contribute to the response of colon adenocarcinomas cells to DOX treatment; and small molecules that can exploit p53-independent pathways for p21 induction, such as TPL, may find a place in chemotherapeutic protocols for the clinical management of colorectal cancer, where p53 function is often lost, due to genetic or epigenetic defects or to post-transcriptional inactivating mechanisms. | Background Colorectal cancer is the second most common cause of cancer-related mortality in Western countries, with about 1 million new cases every year diagnosed world-wide and 500,000 patients dying from the disease [ 1 ]. Of the patients, 30% have advanced disease at presentation, either locally or at distant sites; in this setting, chemotherapy remains the only viable therapeutic option. However, even this option is severely hindered by the inherent resistance of metastatic colon cancer to many currently used anticancer agents. A variety of mechanisms by which cancer cells resist chemotherapy have been described, including enhanced export of drugs from cancer cells and alterations in drug metabolism and/or in drug-target interactions [ 2 ]. In addition, the response of cancer cells to genotoxic therapies may be critically impaired by defects in the response mechanisms to DNA damage [ 3 ] or in cell cycle regulatory pathways [ 4 ]. Over the past decade, induction of apoptosis has emerged as a major event in tumor cell response to cytotoxic agents (for a recent review see [ 5 ]). This view, although recently challenged by some Authors [ 6 ], has attracted considerable attention on deregulation of cell death pathways as a key determinant of drug resistance. Two separable, although extensively cross-talking, pathways leading to apoptosis have been characterized [ 7 , 8 ]. The extrinsic pathway is initiated by ligation of transmembrane receptors to activate membrane proximal "activator" caspases, which in turn cleave and activate downstream "effector" caspases. The intrinsic pathway requires disruption of the mitochondrial membrane and the release of mitochondrial proteins, two events that are regulated by the opposing actions of pro- and antiapoptotic Bcl-2 family members. "Intrinsic stresses", such as those produced by DNA-damaging agents, activate the intrinsic apoptotic pathway; the multifunctional transcription factor p53 is thought to be part of a "fast track" connection between nuclear DNA damage and the intrinsic pathway machinery [ 9 ]. p53 regulates multiple responses to genotoxic stress by transcriptional activation or repression of a number of genes encoding proteins involved in cell cycle control (p21 WAF1/Cip1 ), DNA repair (gadd45), and apoptosis (e.g. Bax, Bcl2 and survivin) [ 10 ]. Mutations in p53 and in the p53 pathway can produce multidrug resistance in vitro and in vivo , and reintroduction of wildtype p53 into p53 null tumor cells can re-establish chemosensitivity [ 11 ]. p53 status is not a universal predictor of treatment response, in part because not all drugs absolutely require p53 for their apoptotic function [ 12 ] and in some settings, p53 loss can enhance drug-induced apoptotic cell death [ 13 ]. Still, loss of p53 function correlates with multidrug resistance in many tumor types [ 11 ] and the observation that this is a common defect in human tumors has spurred an active search for strategies aimed at directly activating cell death pathways downstream of p53. In this scenario, the role played by the cyclin-dependent kinase inhibitor p21 is particularly intriguing, as this protein can be activated by both p53-dependent and p53-independent mechanisms and can assume pro- or anti-apoptotic functions, depending on the cellular context (for a review see [ 14 ]). The present research focuses on the role of p21 in tumor cell response to treatment with cytotoxic agents, and on the possibility to improve the therapeutic index of such agents by modulating p21 status by p53-dependent and independent pathways. The following issues have been addressed: (a) analysis of the relationship between p21 status and sensitivity to treatment with the cytotoxic anticancer agent doxorubicin (DOX) in p53-positive and -negative colon cancer cell lines; (b) design of treatment strategies based on the use of small molecules able to modulate p21 status; for the present study we have used a low molecular weight, stable nitroxide radical, 4-hydroxy-2,2,6,6,tetramethylpiperidne- N -oxyl (also known as Tempol, TPL; figure 1 ) that was shown to exert an antiproliferative effect against different cancer cell lines [ 15 ] and to increase p21 levels in a p53-null human leukemic cell line [ 16 ]. Our data indicate that p21 modulation may significantly affect cell response to DOX treatment in the colon cancer cell lines tested. Methods Reagents Standard chemicals, including 4-hydroxy-2,2,6,6-tetramethylipiperidine- N -oxyl (Tempol, TPL) and cell culture reagents were purchased from Sigma-Aldrich srl. (Milan, Italy), unless otherwise indicated; doxorubicin (DOX) was kindly provided by Dr. A Suarato (Pfizer-Pharmacia, Milan, Italy). Cell lines The human colon carcinoma cell lines HCT116, HT29 (obtained from ATCC, Rockville, MD) and HCT116 p21-negative cells (HCT116 p21-/-), kindly provided by Dr B. Vogelstein (Johns Hopkins University, Baltimore, MD, USA), were maintained in DMEM medium supplemented with 10% FBS (Mascia Brunelli); the HCT116-E6 cell line, obtained from HCT116 cells by transfection with pCMVneo-E6 plasmid (provided by Dr B. Vogelstein) containing the HPV16-E6 human gene, was maintained in ISCOVE medium supplemented with 10% FBS and geneticin (500 μg/ml). All the cell lines were cultured at 37°C, in an atmosphere of 5% CO 2 and 95% humidity. Cytotoxicity assays The effects of DOX and/or TPL on cell growth were assessed by the MTT assay [ 17 ]. Briefly, cells were seeded onto 96-well plates and allowed to grow for 24 h prior to treatment. Three different treatment schedules were used: a) 24 h medium, 1 h DOX (0.05 – 10 μM) followed by 72 h incubation in drug-free medium; b) 24 h TPL (0.05 – 10 mM) followed by 72 h incubation in drug-free medium; c) 24 h TPL followed by 1 h DOX and by 72 h incubation in drug-free medium. For combination experiments, the whole range of DOX concentrations (0.05 – 10 μM) was tested following pretreatment with fixed TPL concentrations, corresponding to the IC 25 or IC 50 values obtained for each cell line according to schedule (b). At the end of the treatment period, 50 μl of MTT (2 mg/ml in PBS) were added to each well at 37°C for 3 h and the reduction of MTT by viable cells was measured colorimetrically at 570 nm, using a Universal Microplate Reader EL800 (Bio-Tek Instruments). IC 25 and IC 50 values (i.e. the concentrations yielding 75% and 50% cell survival fractions, respectively) were calculated according to the median effect equation and analysis of the interaction between DOX and TPL was performed as described by Chou & Talalay [ 18 ]. Transfection of HCT116 cells Transfection of HCT116 cells with the pCMVneo-E6 plasmid was performed by electroporation as described by Yanez and Porter [ 19 ], using a Bio-Rad Gene Pulser unit at the following conditions: 280 V, 960 μF. 48 h post-transfection the cells were selected by adding 500 μg/ml of geneticin to the culture medium. The efficiency and stability of transfection were checked by Western blot analysis of whole cell lysates. Control cells were mock-transfected with the pCMVneo plasmid. Preparation of cell extracts and immunoblotting The expression of p53 and p21, before and after 1 h exposure to DOX (1.0 and 10 μM) followed by 23 h incubation in drug-free medium, or after 24 h exposure to TPL (1.0 and 2.5 mM), was evaluated by Western blot analysis of total protein extracts (lysis buffer: NP40 1%, leupeptin 10 μg/ml and aprotinin 10 μg/ml in TBS). Protein concentration in the cellular lysates was determined by the BCA assay (Pierce, Rockford, IL, USA). 15 μg of protein extract/lane were loaded onto 11% polyacrylamide gels and separated under denaturing conditions. Protein samples were then transferred onto nitrocellulose membrane and Western blot analysis was performed by standard techniques. using a mouse anti-p53 monoclonal antibody (DO-1; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) and a rabbit anti-p21 polyclonal antibody (C-19; Santa Cruz Biotechnology). Proteins were visualized using peroxidase-conjugated anti-mouse and anti-rabbit secondary antibodies (Amersham Pharmacia Biotech) and the ECL Plus Western Blotting Detection Reagents (Amersham Pharmacia Biotech). Densitometric analysis was performed using the Scion Image software (Scion Corporation, Frederick, MD). Flow cytometric analysis of apoptotic cells The presence of apoptotic cells in HCT116, HT29, HCT116-E6 and HCT116 p21 -/-, before and after 1 h exposure to DOX (1.0 and 10 μM) followed by 23 h incubation in drug-free medium, was evaluated by flow cytometric analysis, using a Becton Dickinson FACScalibur flow cytometer. Cells were detached by trypsinization, washed in phosphate-buffered saline (PBS) and fixed in ice-cold 70% ethanol for 20 min at -20°C. After an additional wash in PBS, DNA was stained with 50 mg/ml propidium iodide in PBS in the presence of RNAse A (30 U/ml) at 37°C for 30 minutes. 5 × 10 5 cell samples were analyzed and data were processed using the CellQuest software (Becton Dickinson). The percentage of apoptotic cells in each sample was determined based on the sub-G1 peaks detected in monoparametric histograms. Statistical analysis Comparisons of IC 50 values and apoptotic cell percentages in the four different cell lines were performed by ANOVA and Bonferroni's test for independent samples. C.I. calculations an relative statistical analysis were performed as described by Chou and Talalay [ 18 ]. According to this method, a combination index (C.I.) can be calculated from dose-response curves obtained following exposure to DOX and/or TPL as single agents and in combination. C.I. values approximating 1.0 indicate additive interactions between the two agents; C.I. < 1.0 indicate synergy and, conversely, C.I. > 1.0 indicate antagonism. Results Cytotoxicity assays Figure 2 shows the dose-response curves for HCT116, HT29, HCT116-E6 and HCT116 p21-/- cells following 24 h incubation in drug-free medium, 1 h exposure to DOX and 72 h incubation in drug-free medium. IC 50 values calculated from these data are reported in table 1 ; the resistance index (R.I.) is calculated as the ratio between the IC 50 value obtained for each cell line and that obtained for HCT116 cells. The HT29 cell line, which carries a mutant form of the p53 gene, is significantly more resistant to the cytotoxic action of DOX than the HCT116 cell line, carrying a wild-type p53 gene (IC 50 values: 2.197 ± 0.11 μM vs. 0.38 ± 0.03 μM, respectively; mean ± s.e.m. of 4–6 experiments, p < 0.05). HCT116-E6 cells are 2-fold more resistant to DOX than HCT116 cells (IC 50 : 0.770 ± 0.06 μM; mean ± s.e.m. of 4–6 experiments, p < 0.05 vs. HCT116), while HCT116 p21-/- cells are 14-fold more resistant to DOX than HCT116 cell line (IC 50 : 5.457 ± 0.163 μM; mean ± s.e.m. of 4–6 experiments, p < 0.05 vs. HCT116), in spite of the presence of a wild type form of p53. Immune detection of p53 and p21 Figure 3 shows the expression of p53 (A) and p21 (C) proteins, before and after exposure to DOX (1 and 10 M for 1 h followed by 23 h in drug-free medium), in the four cell lines tested; panels B and D report the densitometric analysis of p53 and p21 immune reactive bands, respectively. In the absence of drug treatment, it is possible to observe that HT29 cells show higher p53 but lower p21 protein levels, as compared with the HCT116 cell line. DOX treatment induces a dose-dependent increase in both p53 and p21 levels in HCT116 cells, whereas in HT29 cells p53 protein levels are not significantly modified by the treatment and p21 protein levels are only detectable when 25 μg of protein extract/lane are loaded onto the gel (instead of the 15 μg loaded for the other cell lines), and even then only at the highest DOX concentration used. As expected, HCT116 E6 cells do not show detectable p53 levels, both under baseline conditions and following DOX treatment; in contrast in HCT116-E6 cells p21 levels are increased by DOX treatment in a dose-dependent fashion, although to a lesser extent than in HCT116 cells. HCT116 p21-/- cells show higher baseline expression levels of p53 compared with HCT116 cells and DOX treatment in this cell line enhances p53 expression to a an even greater extent than in HCT116 cells line. As expected, p21 is undetectable in HCT116 p21-/- cells, and DOX treatment does not modify the intracellular levels of this protein. Evaluation of apoptotic cells by flow cytometric analysis Figure 4 shows the percentage of apoptotic cells following treatment of the four colon cell lines with DOX (1.0 and 10 μM) for 1 h followed by 23 h in drug-free medium. No significant differences in the percentage of apoptotic cells were observed in untreated HCT116, HT29, HCT116-E6 and HCT116 p21-/- cells. Exposure to DOX induces concentration-dependent increases in apoptotic cells in all the cell lines tested; HCT116-E6 cells were the least susceptible apoptosis induction by DOX. Effects of TPL on cell survival and p53/p21 levels Table 2 reports the IC 25 and IC 50 values obtained for the four cell lines after 24 h of continuous TPL exposure followed by 72 h in drug-free medium. TPL can be observed to inhibit cell growth in all four cell lines; although no significant differences can be detected among IC 50 values, HCT116 p21-/- cells appear to be less responsive than the other three cell lines. Figure 5 shows that 24 h exposure to TPL (1 and 2.5 mM) induces a dose-dependent increase in both p53 and in p21 levels in HCT116 cells, whereas in HT29 cell line TPL treatment only induces a dose-dependent increase in p21 expression. Exposure of HCT116-E6 cells to TPL (1 and 2.5 mM) for 24 h does not induce any variations in p53 expression, while a dose-dependent increase in p21 expression can be observed following treatment with the nitroxide. In HCT116 p21-/- cells TPL induce a slight increase in p53 protein levels but, as expected, p21 levels were unaffected by TPL treatment. Effects of TPL pretreatment on DOX-induced cytotoxicity Figure 6 shows the effect of 24 h pretreatment with TPL, at fixed concentrations corresponding to the IC 25 and IC 50 values obtained for each cell line, on DOX cytotoxicity. The cells' response to DOX is expressed as the IC 50 values derived from dose/response curves obtained after 1 h exposure to DOX with or without pretreatment with TPL (24 h), followed by 72 h in drug-free medium. Analysis of cytotoxicity data shows a synergistic interaction (C.I.<1) between DOX and TPL for both TPL concentrations in HCT116 cells and in HCT116-E6 and HT29 cells at the lower concentration; only additive effects (C.I. ≈ 1) can be observed in HCT116 p21 -/- cells. IC 50 values for DOX and TPL according to the three different schedules are reported in table 3 . Discussion Resistance of colorectal cancer to established treatment regimens remains a major concern in oncology; thus attempts at improving the survival of patients affected by this disease depend largely on strategies targeting tumor cell resistance, which cannot be rationally planned without a detailed knowledge of the mechanisms underlying this phenomenon. A current paradigm regarding cancer chemotherapy indicates disabling of the intrinsic apoptotic pathway as a key factor in the response of tumor cells to anticancer drugs [ 3 , 5 , 12 ]. Therefore, strategies aiming at re-establishing the cell's capability to activate a cell death program are an area of active research. The present study was performed in order to define the role of the p53/p21 pathway in the response of colorectal carcinoma cells to DOX, a cytotoxic agent that is typically devoid of effects in this tumor type. The results obtained in our cytotoxicity studies indicate that in the cell lines examined p53 status is not unequivocally related to the response to DOX: in fact, while p53 -deficient cells (HT-29, HCT116-E6) are indeed less responsive than the p53/wt parental HCT116 cell line, the highest resistance index was obtained for HCT116 p21-/- cells, harboring two wildtype p53 alleles. As expected, treatment with DOX leads to p53 upregulation in the cell lines expressing wildtype p53; this effect has been thoroughly documented in colon cancer cells as well as in tumor cell lines derived from other tissues, and has been attributed to phosphorylation and subsequent stabilization of p53, possibly through activation of DNA-dependent protein kinase or ATM (ataxia-teleangectasia mutated) kinase (see e.g. [ 20 - 22 ]). In HCT116 cells, p21 expression parallels p53 activation; however, data obtained in HT29 and HCT116-E6 cells clearly indicate the existence of p53-independent pathways for p21 induction, that have been extensively characterized (for a review see [ 23 ]) and can be activated to variable extents (HCT116 E6 > HT29) upon exposure to DOX. Interestingly, the extent of the cytotoxic effects observed in the small panel of colon cell lines tested rather seems to parallel the cells' ability to upregulate p21 (HCT116 > HCT116-E6 > HT29 > HCT116 p21-/-). This result is somewhat unexpected: in fact, whereas the function of p21 in cell growth arrest following DNA damage has been established for a long time [ 24 ], the role played by this protein in the ultimate fate of tumor cells exposed to cytotoxic agents is far from clear-cut [ 14 , 25 ]. In a number of studies, p21 has actually been reported to protect tumor cells against cell death induced by enforced p53 expression [ 26 ] or by low doses of cytotoxic agents [ 13 , 27 - 30 ]. However, in other experimental settings, p53-dependent or -independent induction of p21 expression seems to be a prerequisite for apoptosis [ 31 - 34 ] and to sensitize tumor cells to the action of different agents [ 35 - 37 ]. The putative mechanisms by which p21 might actually induce apoptosis have recently been reviewed [ 38 ], but still await full elucidation. Interestingly, the situation outlined by our results does not seem to conform to either view: in fact, while induction of p21 in p53-proficient and -deficient cell lines is associated with increased response to drug treatment, this was not accompanied by a parallel increase in apoptotic cells, as no significant differences in apoptosis were observed between HCT116 cells and the 14-fold resistant HCT-116 p21-/- cell line (figure 4 ). This suggests that modes of cell death other than apoptosis may operate in tumor cells following exposure to DOX, or, more generally, to DNA-damaging agents, a concept that is beginning to be proposed by a number of Authors [ 39 , 40 ]; of note, recent experimental evidence indicates p21 as one of the major determinants of terminal growth arrest induced by cytotoxic agents [ 41 - 43 ]. Therefore, although issues related to terminal growth arrest and senescence have not been specifically addressed in the present study, the possibility that these phenomena might play a role linking cell death to the observed increases in p21 levels should not be disregarded. The hypothesis that the cytotoxic response of the tumor cell lines tested in the present study may depend on p21 induction is further corroborated by data obtained following pre-treatment with the piperidine nitroxide TPL. The choice of this compound was dictated by previous findings indicating that TPL induces cell death in a number of tumor cell lines irrespective of their p53 status [ 15 ], and that it increases p21 levels in p53 -null cells [ 16 ]. The results of the present study show that TPL affects the four colon cell lines to similar extents, thus confirming that its growth inhibitory effect is independent of p53 function. HCT-116 p21-/- cells are actually slightly less responsive than the other cell lines [even though the difference does not attain statistical significance), which suggests the possibility that the effects of TPL are due in part to its ability to increase p21 levels. Interestingly, the nitroxide also induces p21 expression even in p53-deficient cell lines; this observation suggests that TPL can activate p53-independent pathways for p21 induction, as already noted following exposure of HT29 and HCT116 E6 cells to DOX. Moreover, activation of such pathways by TPL appears to sensitize tumor cells to the action of DOX: in fact, synergistic potentiation of DOX cytotoxicity is achieved by TPL in those cell lines where p21 expression can be induced, but only additive effects between TPL and DOX are observed in HCT116 p21-/-, where p21 expression is constitutively absent. Conclusions In summary, the results of the present study strongly suggest that 1) p21 induction may significantly contribute to the response of colon adenocarcinoma cells to DOX treatment; and 2) small molecules that can exploit p53-independent pathways for p21 induction, such as TPL, may find a place in chemotherapeutic protocols for the clinical management of colorectal cancer, where p53 function is often lost, due to genetic or epigenetic defects or to post-transcriptional inactivating mechanisms. Competing interests The author(s) declare that they have no competing interests Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544559.xml |
554992 | The role of mast cells in the pathogenesis of pain in chronic pancreatitis | Background The biological basis of pain in chronic pancreatitis is poorly understood. Mast cells have been implicated in the pathogenesis of pain in other conditions. We hypothesized that mast cells play a role in the pain of chronic pancreatitis. We examined the association of pain with mast cells in autopsy specimens of patients with painful chronic pancreatitis. We explored our hypothesis further using an experimental model of trinitrobenzene sulfonic acid (TNBS) -induced chronic pancreatitis in both wild type (WT) and mast cell deficient mice (MCDM). Methods Archival tissues with histological diagnoses of chronic pancreatitis were identified and clinical records reviewed for presence or absence of reported pain in humans. Mast cells were counted. The presence of pain was assessed using von Frey Filaments (VFF) to measure abdominal withdrawal responses in both WT and MCDM mice with and without chronic pancreatitis. Results Humans with painful chronic pancreatitis demonstrated a 3.5-fold increase in pancreatic mast cells as compared with those with painless chronic pancreatitis. WT mice with chronic pancreatitis were significantly more sensitive as assessed by VFF pain testing of the abdomen when compared with MCDM. Conclusion Humans with painful chronic pancreatitis have an increased number of pancreatic mast cells as compared with those with painless chronic pancreatitis. MCDM are less sensitive to mechanical stimulation of the abdomen after induction of chronic pancreatitis as compared with WT. Mast cells may play an important role in the pathogenesis of pain in chronic pancreatitis. | Background Although pain is the presenting symptom of most patients with chronic pancreatitis, its neurobiological basis remains poorly understood [ 1 ]. In the past, investigators have focused on the role of anatomical abnormalities such as a strictured pancreatic duct or narrowed intraparenchymal ducts. However, mechanical decompression techniques such as endoscopic stent placement or even surgical pancreatojejunostomy frequently do not provide a permanent solution to the pain [ 1 ]. More recently, investigators have begun focusing on the role of neurotransmitters and neurotrophins such as substance P and nerve growth factor with known or suspected roles in nociceptive signaling and/or sensitization and have reported an increased expression of several of them in the pancreas of patients with painful chronic pancreatitis [ 2 ]. Mast cells are also increased in both acute and chronic pancreatitis [ 3 , 4 ] but their role in the generation of pain in pancreatitis has not been investigated. We hypothesized that mast cells are involved in the pathogenesis of pain in chronic pancreatitis. This hypothesis is based on the following observations. First, mast cells have been associated with human conditions in which pain is a predominant symptom. Interstitial cystitis and irritable bowel syndrome are both conditions in which pain is out of proportion to the objective pathological findings [ 5 , 6 ]. In both conditions, an increase in the number of mast cells has been described in the bladder and the colon, respectively [ 5 , 6 ]. Further, mast cells are frequently found in close proximity to nerves in the intestinal mucosa and the bladder [ 7 - 9 ]. This has also been observed in the pancreas – the total number of mast cells was significantly higher in pancreatic tissue from patients with chronic pancreatitis than in the normal pancreatic controls [ 3 ]. One of the preferential locations of mast cells was around and within the perineurium of nerve fibers in tissue samples of patients with chronic pancreatitis, suggesting the potential for interactions between mast cells and the nervous system. Lastly, there is evidence for bi-directional functional communication between mast cells and nerves [ 10 - 12 ]. Mast cells can not only release mediators that increase excitability of neurons but in turn, neurotransmitters such as substance P can trigger mast cell degranulation [ 10 ]. Mast cells may therefore contribute to the pathogenesis of pain in pancreatitis through degranulation products that can sensitize pancreatic afferent neurons in an ongoing vicious circle of neuronally mediated mast cell degranulation. Our first aim was to analyze the presence and distribution of mast cells in autopsy specimens of chronic pancreatitis and study the correlation, if any, with historical documentation of pain. We then explored our hypothesis further using an experimental model of trinitrobenzene sulfonic acid (TNBS)-induced chronic pancreatitis in both wild type and Kit W /Kit W-v mice, a strain deficient in mast cells (MCDM). Methods Data collection autopsy study Autopsy records from the University of Texas Medical Branch from the years 1993 to 2000 were searched electronically for the term "pancreatitis." One-hundred sixty-six patients were identified of which 26 patients carried an autopsy diagnosis of chronic pancreatitis and 140 patients carried a diagnosis of acute pancreatitis. The medical charts from patients with an autopsy diagnosis of chronic pancreatitis were reviewed for documentation of a medical history of chronic pancreatitis. If no such documentation was present in the chart, patients were excluded from the study (12/26). Thus, 14/26 patients with both a documented history and an autopsy based diagnosis of chronic pancreatitis, were included in the study. Patients were categorized as painful chronic pancreatitis (8/26) when they fulfilled one of the following criteria: a documented history of chronic abdominal pain clinically attributed to chronic pancreatitis that required the use of narcotics, and/or frequent admissions for recurrent abdominal pain clinically attributed to chronic pancreatitis, and/or a surgical or endoscopic procedure for refractory abdominal pain clinically attributed to chronic pancreatitis. Patients were categorized as non-painful chronic pancreatitis (6/26) if patients did not fit any of the criteria listed under painful chronic pancreatitis. In addition, the following data were collected: demographic factors (age and race), cause of death, comorbidities, clinical history of pancreatitis, etiology of pancreatitis, diagnostic studies supporting a diagnosis of pancreatitis (amylase, lipase, calcifications on abdominal plain film, CT-scan, ultrasound or ERCP). Human pancreatic control tissue was obtained from 8 arbitrarily chosen patients of whom the autopsy records recorded acute myocardial infarction as the cause of death. Their medical records were reviewed to ensure that they did not have a clinical history of pancreatitis. Therefore there were three categories of patients: one with painful chronic pancreatitis, one with non-painful chronic pancreatitis and non-pancreatitis controls. A pathologist, blinded to the group assignment, verified all histological diagnoses and counted mast cells on a Giemsa stained tissue section (average of 10 high-power randomly chosen (40X) fields per specimen). The protocol was approved by the Institutional Review Board of the University of Texas Medical Branch. Mice strains All mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Male mice were used from the following strain: WBB6F1/j-Kit W /Kit W-v (MCDM) and the respective littermate control mouse strain, Kit W-v – +/+ (WT). The mice were 3 months of age at the onset of the experiment with body weights of 25–30 gram. Experimental protocols involving mice were approved by our Institutional Animal Care and Use Committee (IACUC) in accordance with the guidelines provided by the National Institutes of Health. Induction of chronic pancreatitis Mice were anesthetized with sodium Nembutal (50 mg/kg body weigh, i.p.) Following a midabdominal laparotomy, a canula was introduced into the common pancreato-biliary duct; the duct was ligated proximally and distally to ensure perfusion into the pancreas and prevent entry of the injected substance into the liver or duodenum. 0.1 ml of 1% TNBS in phosphate buffered saline (PBS)-10% ethanol, pH 8, was infused into the pancreas (modified after Puig-Divi [ 13 ]). The canula was removed and the abdomen closed. Control mice were treated in the exact same fashion but were perfused with saline instead (Figure 1 ). Mice were sacrificed 8 weeks after surgery. Von Frey Filament (VFF) measurements VFF hairs consist of a series of filaments of increasing diameter that produce increasing sensations of touch when applied to the skin. When the tip of a fiber of given length and diameter is pressed against the skin, the force of application increases until the fiber bends. After the fiber bends, continued advance creates more bend, but not more force of application. This principle makes it possible to apply a reproducible force to the skin surface. VFF testing is an established behavioral pain assay used to determine mechanical pain thresholds in somatic pain. More recently, VFF testing has been used as a surrogate marker for visceral pain [ 14 , 15 ]. Mice were placed in a cage with a mesh floor and habituated to the environment for 30–60 minutes. Measurements were taken from the abdomen and the plantar surface of both hindpaws over a period of three weeks prior to the surgery and for a total of three weeks after the surgery (Figure 1 ). VFF filaments of various caliber were applied to the mid-abdomen in ascending order 10 times, each for 1–2 seconds with a 10 second interval. A response was defined as: a) sharp retraction of the abdomen; b) immediate licking or scratching of site of application of hair; or c) jumping. The response frequency was defined as the total number of responses out of 10 applications (expressed as a percentage) to the skin per filament. An investigator blinded to the different treatment groups performed the behavioral testing. Pancreatic histology Fresh specimens of the mouse pancreas were fixed in 10% formaldehyde in PBS pH 7.4 containing 1 mM MgCl 2 at 4°C overnight. Sections from paraffin-embedded specimens were stained with hematoxyline and eosin and observed under a light microscope. Pathological changes were scored based on a scale described by Tito et al. by a pathologist blinded to the different treatment groups [ 16 ]. Data expression and statistical analysis Comparisons of the number of mast cells in autopsy specimen were analyzed using the Mann-Whitney U test. For each behavioral experiment (see figure 1 ), the average response frequency was calculated as the mean of the mean response frequencies for each mouse (across four measures). The "post-pre response frequency" was calculated by subtracting the pre-surgical average response frequency from the post-surgical average response frequency. To assess the independent effect of TNBS on VFF response (ie. to control for the effect of the surgery itself), the post-pre response frequency for TNBS infusion was compared with the post-pre response frequency for saline infusion. This comparison was performed using analysis of variance for a two-factor experiment with repeated measures on time at each level of force for each type of mice (WT and MCDM). The two factors were induction of pancreatitis or not (TNBS or saline, respectively) and time (pre-surgical or post-surgical). TNBS infusion was considered to have had an independent effect on the VFF response if the post-pre response frequency was greater for TNBS than for saline infusion. Fisher's least significant difference procedure was used for multiple comparisons of least squares means with Bonferroni adjustment for number of comparisons. All effects and interactions were assessed at the 0.05 level of significance. Data analysis was conducted using PROC MIXED with LSMEANS options in SAS ® , Release 8.2 [SAS Institute Inc., SAS/STAT ® User's Guide, Version 8, Cary, NC: SAS Institute Inc., 1999]. Results Autopsy data Patient demographics are summarized in Table 1 . Alcohol abuse was the most common cause for pancreatitis in both groups. Analysis of our results, using the Mann-Whitney U test, revealed significantly more mast cells in patients with a history of painful chronic pancreatitis (n = 8) when compared to patients with either non-painful chronic pancreatitis (n = 6) (33.8 vs 9.4 average mast cell count/10 high power fields; p < 0.01) or controls (n = 8), (33.8 vs 6.1 average mast cell count/10 high power fields; p < 0.01) (Figure 2 ). The increased number of mast cells in patients with painful pancreatitis was noted predominantly in interstitial areas and, to a lesser degree, in the periacinar space. Withdrawal responses to mechanical stimulation of the abdomen and paw after induction of chronic pancreatitis Figure 3 shows the post-pre surgical response frequency for both WT and MCDM. TNBS had a significant independent effect on abdominal VFF response in WT mice at the force levels 4 and 8 mN (p = 0.007 and 0.037, respectively) (Figure 3A ). There was a trend towards a significant effect at the force level of 16 mN (p = 0.066). In contrast, for MCDM, TNBS had no significant effect on abdominal VFF response at any force level (Figure 3B ). There was no significant TNBS effect on VFF response in the left hindpaw for either WT mice or MCDM (Figure 4 ). Data not shown for the right hindpaw. Histological analysis of mice pancreas following TNBS injection Pancreatic histology confirmed the presence of chronic pancreatitis in both WT and MCDM with marked fibrosis, inflammatory infiltrates and ductular proliferation mimicking changes seen in human chronic pancreatitis (Figure 5A ). The pancreas of saline treated controls was normal. There was no significant difference in the overall inflammatory scores between the WT and MCDM (Figure 5B ). An increased number of mast cells were counted in WT mice with chronic pancreatitis compared to saline treated controls (5.6 vs 1.5; p = 0.05) (Figure 6 ). As to be expected, no mast cells were present in pancreas of MCDM. Discussion Chronic pancreatitis has been defined as a continuing inflammatory disease of the pancreas characterized by irreversible morphologic changes that typically cause pain and/or permanent loss of function [ 17 ]. The pathogenesis of pain in this condition remains to be satisfactorily established. We examined the association, if any, of pain with mast cells as quantified in autopsy specimens of patients with a history of painful and non-painful chronic pancreatitis and normal controls. Significantly more mast cells were present in pancreatic tissue from patients with a history of painful chronic pancreatitis, indicating an association with this condition and a potential role for these cells in the pathogenesis of pain in painful chronic pancreatitis. There are clearly limitations to a retrospective, autopsy-based study such as the one we report here. For instance, we do not know whether pain was present at the time of death and there was incomplete information on the different patterns of pain. Also, our findings pertain mainly to patients with a history of alcoholic pancreatitis. Nevertheless, our findings do suggest an association of painful chronic pancreatitis with an increased number of mast cells. This observation provided the rationale for further experimental testing, which we performed in mice. We first developed a model of chronic pancreatitis in mice following a modified protocol first described by Puig- Divi et al . [ 13 ]. Histological changes consisted of periductal and lobular fibrosis, duct stenosis, chronic inflammatory cell infiltrates, and gland atrophy, mimicking features of chronic pancreatitis in humans. Significantly more mast cells were present in WT mice with chronic pancreatitis, adding to the validity of this model for use in studies on the role of mast cells in pancreatitis. Both WT and MCDM developed histological changes consistent with chronic pancreatitis, indicating that the elimination of mast cells did not modulate the animals' ability to mount an inflammatory response. Therefore, any changes observed in pain behavior are unlikely to stem from differences in underlying inflammation. Next we determined whether this mouse model could be used to study behavioral differences associated with chronic pancreatitis. The assessment of spontaneous pain in a visceral organ presents significant difficulties. We have used a behavioral method to assess this, which relies on the association of visceral pain with sensitization of somatic regions of the body that share segmental innervation at the level of the spinal cord (referred pain). This somatic sensitization can be quantified using VFF to stimulate the somatotopically appropriate abdominal region and measuring the abdominal withdrawal response. Thus, VFF testing of the anterior abdominal wall can be used as a surrogate marker for visceral pain. Although this is the first time that this technique has been used for the measurement of referred visceral hyperalgesia in a mouse model of chronic pancreatitis, this method has previously been described and validated to assess the severity of referred visceral pain for models of colonic hypersensitivity [ 14 ] as well as rat models of acute necrotizing pancreatitis [ 15 ] and chronic pancreatitis [ 18 ]. The abdominal VFF response was compared to the hind paw response to assess the specificity of the interventions to the pancreas. TNBS treated mice, but not the saline control, developed increased abdominal wall withdrawal responses to VFF testing when compared to baseline, suggesting the development of force-dependent referred hyperalgesia of the abdominal wall in WT mice. There was no evidence of referred hyperalgesia in the hindpaws, suggesting that the measured effect on abdominal withdrawal is specific for an intra-abdominal origin of the pain. Vera-Portocarrero et al. previously described similar findings, increased withdrawal frequency after VFF stimulation to the abdominal area, in a rat model of chronic pancreatitis [ 18 ]. These behavioral changes were abrogated by morphine. Rats that demonstrated behavioral changes also expressed increased substance P expression in the nociceptive layers of the spinal cord, suggestive of central nociceptive changes. Mast cells produce a variety of degranulation products in the setting of inflammation that may activate and/or sensitize primary nociceptive neurons. The neurotrophin growth factor (NGF) is one such product [ 19 - 22 ]. NGF is released in the setting of inflammation and can not only function as a chemoattractant for other mast cells, but it can also trigger mast cell degranulation [ 23 ]. We are speculating that NGF production in the inflamed pancreas is responsible for plastic changes in the sensory neurons by activating proalgesic receptors and channels such as the NGF receptor tyrosine kinase A (TrkA) and Transient Receptor Potential Family V receptor 1 (TRPV1; previously known as VR1) thereby contributing to the generation of pain [ 24 - 26 ]. Similarly, other mast cell degranulation products such as tryptase and histamine are capable of modulating neuronal function [ 27 - 32 ]. Tryptase may directly activate the proteinase-activated receptor-2 (PAR-2), a G-protein coupled receptor expressed by pancreatic nerves, important in the pathogenesis of pain in pancreatitis [ 33 , 34 ]. Although the role for mast cells in the mediation of visceral nociceptive signaling needs to be explored further, we speculate that mast cell products released in pancreatitis, contribute to the development of pain by direct effects on nociceptors located on pancreatic afferent neurons (Figure 7 ). However, before concluding a definite role for mast cells from our experimental data, it should be noted that MCDM carry a spontaneous mutation for tyrosine kinase receptor c-kit which not only produces a deficiency of mast cells but may have an independent effect on the function of sensory neurons, which are known to express it [ 35 ]. Therefore, it remains to be determined whether the detected differences in nociceptive responses is due to the absence of mast cells per se or a yet unknown change in the responsiveness of sensory neurons due to a congenital lack of the c-kit receptor. Reconstitution of mast cells into the MCDM mice should restore their nociceptive responses close to the wild type phenotype. Conclusion Our data should increase awareness of the importance of mast cells in the pathogenesis of painful inflammatory conditions such as chronic pancreatitis and encourage experimental studies for further testing of this hypothesis. Abbreviations TNBS – trinitrobenzene sulfonic acid, WT – wild type, MCDM – mast cell deficient mice, VFF – von Frey Filaments, IACUC – Institutional Animal Care and Use Committee, PBS – phosphate buffered saline, NGF – neurotrophin growth factor, TrkA – tyrosine kinase A, TRPV1 – Transient Receptor Potential Family V receptor 1 (previously known as VR1), PAR-2 – proteinase-activated receptor-2 Competing interests The author(s) declare that they have no competing interests. Authors' contributions WAH conceived of the study, participated in its design and coordination and wrote the manuscript. KG completed the behavioral studies. SYX completed the histological analyses and mast cell counts. JHW provided technical assistance with the surgical procedures and participated in the data analyses and review of the manuscript. WDW provided support for the behavioral studies as well as their analyses. PJP provided financial support and participated in the review of the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554992.xml |
522811 | Polyadenylation of ribosomal RNA by Candida albicans also involves the small subunit | Background Candida albicans is a polymorphic fungus causing serious infections in immunocompromised patients. It is capable of shifting from yeast to germinating forms such as hypha and pseudohypha in response to a variety of signals, including mammalian serum. We have previously shown that some of the large 25S components of ribosomal RNA in Candida albicans get polyadenylated, and this process is transiently intensified shortly after serum exposure just prior to the appearance of germination changes. Results We now present data that this process also involves the small 18S subunit of ribosomal RNA in this organism. Unlike the large 25S subunit, polyadenylation sites near the 3' end are more variable and no polyadenylation was found at the reported maturation site of 18S. Similar to 25S, one or more polyadenylated mature sized 18S molecules get intensified transiently by serum just prior to the appearance of hypha. Conclusions The transient increase in polyadenylation of both the large and the small subunits of ribosomal RNA just prior to the appearance of hypha, raises the possibility of a role in this process. | Background Candida species are now among the most important pathogens especially for the immunocompromised host. They are the fourth most common organisms recovered from blood cultures in hospitalized patients [ 1 ]. Candida albicans the most frequently isolated of the species, is a polymorphic organism. It can switch from a yeast form (blastospore) to a filamentous phase (hypha and pseudohypha) in response to a variety of external stimuli, including mammalian serum. Mutants defective in this serum response, also show a reduced capacity to cause disease in a murine model, [ 2 ] suggesting a virulence role for it. It is widely accepted that the production of the RNA components of ribosomes in eukaryotes proceeds through the transcription of large pre-RNA molecules by RNA polymerase I (Pol I), that get processed into the final large and small subsegments [ 3 ]. We have recently reported the unexpected finding that Candida albicans polyadenylates some of its 25S ribosomal RNA (rRNA) and the polyadenylation site corresponded to the large subsegment 3'-end maturation [ 4 ]. We also found that the concentration of the polyadenylated form of 25S was increased transiently by serum just prior to the appearance of filamentous forms, raising the possibility for a role in hyphal transformation. A question raised by these data was whether this event is unique for the large subunit rRNA or it represents a wider function. For example, is a DNA sequence upstream from the 25S subunit functioning serendipitously as a promoter for RNA polymerase II (Pol II), allowing transcription by this enzyme complex and subsequent polyadenylation. Such a process would not likely be important for ribosomal function. Similar involvement of other subunits on the other hand, would increase the likelihood that polyadenylation of rRNA plays a wider role in the biology of this yeast. We now report our observations related to the 18S subunit of rRNA, that indeed there are other polyadenylation sites located near the 3'end of the 18S subunit. Similar to 25S, we found that a polyadenylated 18S transcript, similar in size to a processed mature molecule, is also enhanced early and transiently by serum, further strengthening the possibility of a regulatory role for polyadenylation in the germination process. Results Cloning of poly A -extended 18S subsegments We have found polyadenylation to occur both in yeast grown in YPD and in those exposed to serum for 5 minutes. In all cloned PCR products the number of adenines in the chain exceeded those in the poly-T primer (DT12) used to generate them. Furthermore, the anchor sequence assures us that we did not extend an inappropriately annealed primer. Unlike our data with the 25S subunit, where the attachment was to one of two thymidines one base apart at the reported maturation site [ 4 ], the polyadenylation sites near the 3' end of 18S subunit were congregating either upstream or downstream of the reported maturation site [ 5 ] (Figure 1A ) but none were at the reported site. Six of seven YPD exposed yeast polyadenylation sites were between positions 1625 and 1643, located 148 bases upstream from the reported 3' end, whereas 3 of 4 polyadenylation sites downstream from the maturation site were from yeast exposed to serum. We were able to amplify a full-length clone from YPD exposed yeast and its site of polyadenylation was at position 1643 (Figure 1B ) near the other polyadenylation sites of yeast exposed to YPD. Figure 1 Polyadenylation sites of 18S rRNA subunit and a full length polyadenylated 18S clone. Polyadenylation positions on 18S subunit. (A) represents the clones derived from amplification with primers designed for the 3'end of 18S. Underlined letters represent polyadenylation sites. Y or S over them indicate whether RNA came from yeast exposed to YPD (Y) or serum (S) and superscripted digits over them indicate the number of clones found at that position. The bold enlarged letters represent the sequence with the most polyadenylation sites. The #1624 represents the position from 5' end of 18S. (B) represents partial sequences at the 5' and 3' ends of a full length clone of a polyadenylated 18S molecule. Serum enhancement of polyadenylation Similar to 25S, polyadenylation of 18S subunit was enhanced by serum exposure and this is shown in Figure 2A , representing a Northern blot utilizing poly-A selected RNA, hybridized with an 18S specific probe. By 5 minutes the intensity of the 18S band was more than tripled but back to baseline levels at 15 minutes. UBI4 was also up-regulated by serum but its intensity remained the same at 15 minutes, while the 18S band returned to baseline assuring us that the 18S enhancement was not as a result from an error in RNA loading. We previously found ACT1 [ 4 ] to show the same pattern, suggesting that serum exposure may also up-regulate constitutive genes. As a control for temperature, we also exposed yeast to YPD at 37°C and to serum at 30°C. There was no increase in polyadenylation in YPD at 37°C and there was increase with serum exposure at 30°C (data not shown) indicating that serum caused this increase. Estimates from phoshporimager data indicate that at baseline in YPD less than 1% of the rRNA is polyadenylated (data not shown). Polyadenylation of rRNA in Saccharomyces cerevisiae has recently been described in similar amounts [ 6 ]. Figure 2 Upregulation of 18S rRNA polyadenylation by serum exposure. Northern blot of poly-A selected RNA hybridized by 18S and UB 14 specific probes (A). (a) is RNA derived from cells grown in YEPD at 30°C, (b) is RNA from yeast in serum for 5 minutes at 37°C, and (c) is RNA from yeast in serum for 15 minutes at the same temperature. (B) represents the quantitative phosphorimager data of the Northern shown in (A). Small letters (a, b, c) are the same as in (A). To further confirm that the 18S increase from time zero to 5 minutes is real, we performed real-time PCR reactions and the results are shown in Figure 3 . As can be seen, amplification can be detected 10 cycles earlier when the template is derived from organisms exposed to serum, confirming the increased amount of 18S in the starting material. 5S is detected at the same cycle whether exposed to serum or not. Figure 3 Real-time PCR confirmation of 18S rRNA polyadenylation upregulation by serum. Real-time PCR reactions represented as PCR baseline subtracted relative fluorescence units (RFU) versus cycle number plots. Triangles represent reactions with 18S specific primers and circles represent reactions with 5S specific primers. The 18S band in Figure 2A , lane b is slightly up-shifted as compared to lanes a and c, suggesting that likely one or more molecules, whose polyadenylation sites are downstream from the reported maturation sites are up-regulated by serum. This is consistent with our findings above, that 3 of the 4 polyadenylation sites located in that region were from serum exposed yeast and it suggests that serum up-regulation of polyadenylation may be selective to these downstream sites. When the same filter was hybridized with the 5S specific probe no bands were detected, indicating that our 18S bands were not a result of rRNA contamination (data not shown). Figure 2B represents the phosphorimager generated counts of these bands confirming the visual results objectively. Northerns that included the polyA-minus fractions (data not shown) continue to show the 18S bands indicating that rRNA transcripts with and without polyadenylated extensions are being produced. Discussion These data indicate that 18S subunit mirrors the large 25S molecule as regards to polyadenylation and its response to serum, suggesting that this is not an incidental phenomenon. They do differ from the 25S subunit in that polyadenylation occurs both upstream and downstream to but not at the reported 3'-end maturation site, while for the 25S subunit, the polyadenylation was found to be exclusively at the maturation site. Perhaps the 3'-end of the 18S subunit plays an important role in the recognition of start sites on mRNA, and is vigorously protected from modifications. This would also suggest that polyadenylation of 18S has a role outside of the ribosome. One of the basic questions raised by our original data was whether these polyadenylated transcripts are products of Pol I and get polyadenylated following maturation cleavage or are newly transcribed by Pol II. These data do not resolve this question. While finding of multiple polyadenylation sites in 18S with most of them clearly not corresponding to a reported maturation site, might result from transcription by an enzyme other than Pol I, it is just as likely that they may represent inappropriate cleavage by the ribosomal RNA processing apparatus and these products are being readied for degradation by polyadenylation. The recent report of polyadenylation in Saccharomyces cerevisiae that was found to be increased in mutants lacking the degradative function Rrp6p [ 6 ] favors the latter scenario. RNA polymerase switching from Pol I to Pol II for rRNA transcription has been described for Saccharomyces cerevisiae [ 7 ] in cells where the gene for one of the components of the Pol I transcription factor UAF (upstream activation factor) was mutated. These mutants gave rise to isolates that were utilizing Pol II for their rRNA transcription and this newly switched-on state was heritable even through meiosis. These data indicate that S. cerevisiae has the inherent capacity to utilize Pol II for rRNA transcription but that this capacity is suppressed by a mechanism that includes UAF. These mutants though, switched to Pol II transcription exclusively. Our data with C. albicans differs in that both polyadenylated and non-polyadenylated forms are produced simultaneously. Conrad-Webb and Butow [ 8 ] have described rRNA transcripts of various lengths that were polyadenylated, produced by a respiratory-deficient isolate of S. cerevisiae . The template utilized by this strain was an episomal copy of ribosomal DNA that contained a Pol II promoter sequence overlapping with the Pol I promoter. Recently, circular and linear rDNA plasmids have been reported in C. albicans [ 9 ] for the first time. Thus it is possible that one of these episomal elements also contains Pol II promoters allowing it to function as the template for polyadenylated forms of rRNAs. With our findings that polyadenylation also involves 18S, such Pol II promoters would have to be present for both subunit genes making Pol II role less likely. Polyadenylation of a small percentage of total RNA in Escherichia coli has been reported [ 10 , 11 ] including rRNA and this polyadenylation occurred even in wild type organisms. Hence it appears, that polyadenylation of these stable molecules occurs more widely as a biological phenomenon. The role of polyadenylation of rRNA in C. albicans is unknown. Open reading frame analysis of the 18S subunit indicates that translation into protein is unlikely, as it would result in peptides shorter than 40 amino acids. Multiple polyadenylation sites upstream and downstream from the reported maturation site suggest, that these may be inappropriately processed molecules that are being readied for degradation, though one of the downstream sites may be an A2 processing site. The mere up-regulation of polyadenylation by serum prior to germination also does not indicate a role in hypha formation as other genes such as ubiquitin and actin also respond similarly. There are aspects to our new 18S and our previous 25S data that leave the possibility for a role in germination open. These include the transient nature of this up-regulation for both subunits just prior to germination and the possible selective nature of this process involving 18S. Conclusions The ribosome is central to cellular function and the RNA component of this organelle assumes critical structural and catalytic roles. Our initial unexpected finding of polyadenylation of a portion of the large rRNA subunit is now extended to the small subunit. That this modification also involves the other major component of the ribosome points to a biological role for this process. The fact that the transient up-regulation of RNA polyadenylation from both subunits just precedes the phenotypic expression of germination, suggests a possible role in regulating Candida albicans' polymorphic behavior. Methods Organism and germination conditions Candida albicans SC5314 (obtained from W. Fonzi) [ 12 ] was grown in YPD medium (1%, w/v, yeast extract; 2%, w/v, peptone; 2%, w/v dextrose) at 30°C. Heat inactivated (56°C for 30 minutes) fetal bovine serum (FBS) (10%, v/v in H 2 O) was utilized to induce germination. Yeast cells were grown overnight in YPD at 30°C, harvested by centrifugation, washed once in H 2 O and transferred to FBS pre-heated to 37°C at 1–5 × 10 6 cells ml -1 RNA isolation RNA from cells at various growth conditions was obtained as follows. Incubating mixtures were rapidly cooled in an ice-water bath and were thereafter centrifuged at 4°C and washed with ice cold water once. Cell walls were digested by suspending the pellet in a buffer containing 1 M sorbitol, 0.1 M EDTA, 0.1%, v/v, β-mercaptoethanol and 100 U ml -1 lyticase (ICN), in a volume 1/5 that of the volume of the of the original cell culture. The digestion proceeded for 10–20 minutes at 30°C. Adequacy of the digestion was monitored by testing a small drop of cell suspension in SDS for viscosity [ 13 ]. Mixtures were centrifuged at 800 × g and RNA was isolated by using the QIAGEN total RNA kit, following the manufacturer's protocol. RNA was precipitated with isopropanol, and either used immediately or stored in ethanol at -20°C. Polyadenylated RNA selection was carried out by following the Oligotex mRNA kit protocol (Qiagen). Northern blot analysis Samples of 50 ng of poly(A) RNA were electrophoresed on 1.2% formaldehyde agarose gel blotted to nylon membranes. A 262 bp long fragment of 18S was generated from reverse transcribed total RNA with the primers 5'-TCGATGGAAGTTTGAGGCAA-3' (P1) and 5'-ATTCAATCGGTAGTAGCGACGGGC-3' (P2) based on the previously published sequence (Barnes et al., 1991). After cloning into pCR2.1 (Invitrogen), and sequencing (T7 Sequenase 2.0 kit, Amersham) to confirm that it was 18S specific, the insert was released by Eco RI digestion and gel purification, and was labeled with P 32 by using random priming. As a control a 499 bp UB14 probe was also generated with the primers 5'-GAAGTCGAATCTTCTGACACCATCG-3' (P3) and 5'-TGGTGGAATACCTTCTTTGTCTTGG-3' (P4). The primer design was based on the UB14 sequence (Accession No. Z54197) reported by the Stanford Candida Genome project's World Wide Web site [ 14 ]. This amplified product was also cloned and sequenced to confirm its specificity. To assess the quality of our RNA, a 5S specific probe was generated with the primers 5'-GGTTGCGGCCATATCTAGCAGAA-3'(P5) and 5'-AGATTGCAGCACAATAGTTTCGC-3' (P6). These primers were based on the reported sequence of the 5S component of Candida albicans rRNA [ 15 ]. Phosphorimaging volume report analysis (Molecular Dynamics, Sunnyvale, CA) has been employed to quantify objectively the intensity of individual bands. We estimated the percentage of polyadenylated 18S, from phosphorimaging data derived from Northern blots comparing total RNA and poly-A selected RNA (derived from the same amount of total RNA). Real-time PCR analysis To confirm that the polyadenylated form of 18S is increased by five minutes, total RNA was predigested with RNase-free DNase I (New England Biolabs), then reverse transcribed with Superscript II RT (Invitrogen) utilizing an anchored polyT primer 5'-AATTCGGCGAGCTCCGCGGCCGCGTTTTTTTTTTTT-3' (DT12) to generate cDNAs from polyadenylated RNA molecules. The same reaction also contained the primer P6, specific for 5S rRNA subunit that does not get polyadenylated [ 15 ]. Using these cDNA templates, a 63 nt long sequence specific for 18S at positions 1585–1648 [ 5 ] was amplified by primers P2 and 5'-TCAGCTTGCGTTGATTACGTCC-3' (P7). In a separate reaction, primers P5 and P6 were used to amplify 5S. To insure that there was no genomic DNA contamination, we carried out PCR reactions utilizing primer pairs P2-P7 and P5-P6 on predigested total RNA that was not reverse transcribed and no products were generated (data not shown). Real time PCR reactions were carried out with iQsybr Green Supermix(Bio-Rad) as source of fluorescence, utilizing an iCycler (Bio-Rad) thermocycler. The cycling settings used were; initial denaturation for 3 min at 95°C followed by 40 cycles each consisting of 30s denaturation at 95°C, 30s primer annealing at 55°C and 30s extension at 72°C. Data were analyzed using iCycler iQ version 3.0a. Amplification of poly-A extended 18S rRNA To identify polyadenylation positions involving 18S, total RNA was heated at 70°C for 5 minutes to minimize secondary structures. Reverse transcription was done with the anchored polyT primer DT12, utilizing Superscript II RT (Invitrogen). PCR products were amplified with primers DT12 and P1 which is situated 398 nucleotides from the reported 3' end of the molecule using HotStarTaq (Qiagen). To show the presence of a mature sized polyadenylated molecule, amplification was also carried out with DT12 and 5'-TATCTGGTTGATCCTGCCAGTAGTC-3' (P8) situated at the 5' end of 18S using FailSafe PCR System (Epicentre). Amplified products were subcloned into pCR2.1 (Invitrogen), a number of clones were picked and sequenced (T7 Sequenase 2.0 kit, Amersham). For the full-length clone, only parts of the 5' and 3' ends were sequenced. Authors' contributions HL and CPW carried out and analyzed experiments. JF conceived of the study, designed experiments, analyzed data and wrote manuscript. All authors have read and agree with final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC522811.xml |
522805 | Topically applied vitamin C increases the density of dermal papillae in aged human skin | Background The influence of ageing on the density of the functional entities of the papillae containing nutritive capillaries, here in terms as the papillary index, and the effect of topically applied vitamin C were investigated by confocal laser scanning microscopy (CLSM) in vivo. Methods The age dependency of the papillary index was determined by CLSM on 3 different age groups. Additionally, we determined the effect of a topical cream containing 3% vitamin C against the vehicle alone using daily applications for four months on the volar forearm of 33 women. Results There were significant decreases in the papillary index showing a clear dependency on age. Topical vitamin C resulted in a significant increase of the density of dermal papillae from 4 weeks onward compared to its vehicle. Reproducibility was determined in repeated studies. Conclusions Vitamin C has the potential to enhance the density of dermal papillae, perhaps through the mechanism of angiogenesis. Topical vitamin C may have therapeutical effects for partial corrections of the regressive structural changes associated with the aging process. | Background A consistent feature of aged and photoaged skin is the flattening of the epidermal-dermal junction, evidenced in histological sections as a loss of rete ridges and the disappearance of papillary projections. The depth of interdigitation of the retepegs and the dermal papillae decreases with age [ 1 ]. Using confocal scanning microscopy sections horizontal to the surface can be obtained optically in contrast to conventional transverse sections which are perpendicular to the surface. Comparison of confocal images with corresponding histological sections have been made by the group of Gonzales to validate the method [ 2 - 5 ]. In confocal images the dermal papillae appear as dark circles surrounded by bright reflecting rings of basal cells containing highly reflective pigment [ 6 ]. Each papilla contains a single nutritive capillary loop [ 7 ]. The density of these functional entities of dermal papillae containing a single nutritive capillary loop can be evaluated more accurately and precisely by confocal microscopy than by conventional histology. This is partly due to the avoidance of shrinkage artefacts after fixation for histological sectioning. Confocal microscopy is more suitable to construct a three dimensional image than conventional serial sections are [ 8 ]. In extremely aged skin, papillae virtually disappear and the junction with the atrophic epidermis is a straight line versus undulations in younger skin. The papillary dermis also thins along with a loss of capillaries. Grove described that the corneocytes in aged skin become larger as a result of decreased epidermal turn-over [ 9 ]. Similarly, cells of the granular layer become larger, indicating a slow epidermal turn-over in aged skin [ 6 ]. Unlike for example in mice, in humans like in primates in general and in guinea pigs vitamin C must be supplied exogenously in the diet. Diets deficient in vitamin C cause the multiple clinical signs of scurvy. Stones evolutionary treatise showed that a genetic mutation resulted in the loss of the ability of humans and some other animal species to synthesize vitamin C [ 10 ]. The main mechanism for the symptoms of scurvy seems to be the instability of non-hydroxylated forms of collagen, as vitamin C is needed for the hydroxylation of prolin [ 11 ]. There is evidence that topical vitamin C might be beneficial in several unrelated conditions. Topical vitamin C has been reported to improve wound-healing [ 12 ]. Roshchupkin has shown that topical vitamin C is protective against immediate effects of ultra-violet radiation on human skin leading to an increase in the dose required to induce erythema [ 13 ]. Topical vitamin C protects against ultra-violet-induced carcinogenesis [ 14 ]. The level of vitamin C in the skin decreases with age, especially in the epidermis [ 15 , 16 ]. Topical vitamin C increases the mRNA levels of collagens I and III, and their processing enzymes in humans [ 17 , 18 ]. Humbert showed the potential of ascorbic acid to improve the clinical appearance of photoaged skin and to reduce facial wrinkles [ 19 , 20 ]. Retinoids, too, are known to improve the clinical appearance of photoaged skin and to promote the downgrowth of rete ridges, restoring the undulating dermo-epidermal interface [ 21 ]. A functional dermal-epidermal junction provides a better protection against mechanical stress which might detach the epidermis and lead to erosions [ 22 ]. The increased availability of oxygen and nutrients should also enhance repair after wounding and reverse the structural changes associated with photoaging. The height of the epidermal-dermal junction like it is measured as a parameter in conventional histology, can be determined by confocal microscopy only in a very inaccurate manner because of the horizontal orientation of the images. The aim of this study is first to establish a CLSM measurable parameter comparable to the height of the epidermal-dermal interface. It should be sensitive to find age-associated changes of the dermal-epidermal interdigitation. Second to examine the effects of a topical vitamin C containing moisturizing cream on the developed CLSM parameters. We used a dose of 3% vitamin C as it shows a good stability and a good release of the active ingredient in a special emulsion system [ 23 ]. Methods In the first experiment (I), three groups of volunteers of different ages were compared measuring the density of the functional entities consisting of a dermal papilla and the nutritive capillary (papillary index). In a second experiment (II) a topically applied cream containing 3% vitamin C and its excipient were tested on the volar forearms of 33 volunteers. The effect of vitamin C cream was compared to excipient and to untreated control sites, respectively. (III) This experiment was repeated with a slightly different vehicle and for a two month treatment period only. Instrument The Vivascope 1000 (Lucid Inc., Rochester, N.Y.) is a commercially available confocal laser scanning microscope, which allows to examine human skin in vivo non-invasively. The system uses a Laser source with a wavelength of 830 nm, an illumination power up to 20 mW on the object and water immersion. In all three studies two different fields of view were used to obtain the papillary index (640 μm × 480 μm), and the projection areas of cells in the granular layer (230 μm × 88 μm). The lateral resolution of the instrument is 0.4 μm, the vertical resolution is about 1.9 μm. Parameters In confocal images of the forearm the epidermal-dermal interface shows dark round areas, the dermal papillae, surrounded by bright circles of basal cells, containing melanin granules and therefore reflecting strongly. Capillary loops are located in the centre of dermal papillae as black holes, often showing bright erythrocytes flowing through the capillary. The density of dermal papillae was evaluated by counting the papillae containing a capillary loop. 20 fields of view were investigated on each test site. The size of cells in the upper granular layer (A gran ) was evaluated by saving images of the most apical plane of the epidermis that still showed living dark nucleated cells. Although in the same image you can have virtual sections of different cell layers, the cells belonging to this layer right underneath the stratum corneum can be identified by the characteristic morphologic features the cells display. The size of cells was analysed using the image analysis program image tool. At least 20 cells in at least three different images were analysed by trained examiners. Subjects and treatment (I) In the first study three groups of female volunteers were studied: twelve, ages 18 to 25, eleven, ages 40 to 50 and twelve, ages 65 to 80 years. Informed consent was obtained. Images were obtained on the volar side of the right forearm during the winter months. (II) In a second experiment a cream containing 3% vitamin C was compared to its vehicle on a group of initially 36 volunteers. 33 volunteers finally participated using it twice daily. Applications for four months to mid-volar forearms of healthy female postmenopausal subjects, ages 45 to 67 years (mean 55.3 years), phototype II and III. One area was left untreated. The double-blind study was conducted in winter. Informed consent was obtained. Evaluations were obtained at baseline and each month for four months, with one and two months follow-ups after stopping the treatment. (III) To determine reproducibility a second study was conducted over an eight weeks treatment period with a slightly different vehicle. The parameter of interest was the papillary index. Informed consent was obtained. Measurements were made after 20 minutes of acclimatisation at 21°C and 50% relative humidity. Statistics Morphometric results are expressed in box-whisker-plots as the median +/- the quartile. Statistical analysis used the Student t-test for normally distributed data or the U-Test (Mann-Whitney) for the age comparison study. The Wilcoxon test was used for analysing the effect of the vitamin C studies. Statistical significance was established for the 95% confidence level (p < 0.05). Results (I) The papillary index in the young group, with a median of 54.3 papillae per mm 2 , was significantly the highest of the three age groups. The middle aged group showed a significantly higher index, median 29.6 papillae per mm 2 , than the oldest group, median 24.7 papillae per mm 2 . The decrease of the papillary index was more pronounced between the younger group (18 to 25 years) and the middle-aged group (40 to 50 years) than between this group and the old group (65 to 80 years) (fig. 1 ). Figure 1 Box-plot of papillae per mm 2 in three age groups In the elderly group there were often large segments of flat epidermal-dermal junctions with a complete loss of papillae. In these regions the microvasculature consisted of horizontally orientated vessels having noticeably larger diameters than in young skin, probably venules rather than capillaries. A few curled remnants were occasionally found just beneath the atrophic epidermis. The vessel walls could not be resolved into endothelium, pericytes or smooth muscle cells by confocal examination, owing to low reflectiveness. (II) From four weeks onwards ascorbic acid treated areas showed a significantly higher density of capillary containing papillae than untreated and vehicle treated sites (fig. 2 , p-values in table 1 ). Figure 2 Box-whisker-plots of papillae per mm 2 during topical vitamin C; t0 = baseline, t1 to t4 = after 1 to 4 months, t5 and t6 = 1 and 2 months after stopping Table 1 statistical analysis of the verum treatment study: p-levels, Wilcoxon Matched Pairs Test vehicle vs verum untreated vs verum untreated vs vehicle t0 58,70% 21,10% 36,40% t1 1,40% 0,35% 95,00% t2 0,03% 0,02% 62,60% t3 0,02% 0,03% 97,00% t4 0,01% 0,01% 87,30% t5 0,60% 15,30% 28,10% t6 4,28% 12,50% 77,30% Statistically significant differences were also observed in the vitamin C treated sites from four weeks onward in comparison to the beginning of the study. The untreated as well as the vehicle treated site showed a slight, but significant increase (from median 25.3 to 26.9 papillae per mm 2 for the untreated and from median 22.7 to 30.3 papillae per mm 2 for the vehicle treated sites) over the treatment period as well. The significance of the difference between the verum and the two control sites were no longer evident two months after stopping treatment. Despite of the different median values for the areas at t0, no significant differences between these baseline values could be detected. It is noteworthy that the newly formed papillae could not be distinguished from the preexisting ones. Neither abnormal formations of blood vessels like multiple capillaries in a papilla, parallelisation of the vessels or highly enlarged diameters of the vessels nor signs of inflammatory reactions e.g. rolling or adhesion of leukocytes to the vessel walls were observed. The projection areas of the cells in the upper granular layer showed a significant decrease of about 16% from 472 μm 2 to 383 μm 2 in the vitamin C treated site, in comparison to baseline and the two control sites. Significant difference between vehicle and untreated controls compared with each other and with the initial status at the beginning of the study were not detected (fig. 3 ). Figure 3 Box-whisker-plots of areas (μm 2 ) of cells in the apical granular layer; t0 = baseline, t2 = after 2 months of treatment (III) The increase of the density of the papillae under influence of vitamin C was reproduced by repetition of the treatment schedules, ruling out the likelihood of changes due to chance. Discussion The decrease of the height of the epidermal-dermal junction is a well described histological finding in aged and photoaged skin [ 22 , 24 ]. We conclude from our data that ageing results in a damage of the papillary dermis and of the vascular network in this layer and a decrease of the papillary density over time. Vitamin C has been reported to improve the clinical appearance of photoaged skin and to enhance the synthesis of composite elastin fibres and of collagen [ 19 ]. Our data indicate that the topical application of vitamin C partially restores the anatomical structure of the epidermal-dermal junction in young skin and increases the number of nutritive capillary loops in the papillary dermis close to the epidermal tissue in the aged skin of postmenopausal women. The increase in the density of papillae after vitamin C treatment can not be interpreted only as a sign of epidermal hyperplasia with an enlarged area covered with basal layer and the epidermis growing down into the dermis. More likely it is linked to a restructuring of the papillary dermis, as the top of the newly formed papillae and the capillaries are localized above the average height of the plane basal layer seen predominantly in aged skin. This suggests restoration to a more normal functional state of the epidermal-dermal interdigitation and of the overlying epidermis. In aged skin, the cells in the apical granular layer show larger projection areas than in the skin of younger individuals [ 8 ] similar to the enlargement of the size of corneocytes with age described earlier by Grove and Kligman [ 9 ]. Kligman shows an inverse relationship between turnover and the size of corneocytes [ 9 ]. A similar relationship could be shown for the size of cells in the granular layer [ 25 ]. Parallel to this, the smaller size of granular layer cells in the vitamin C treated areas can be interpreted as a sign for higher proliferative activity of the epidermis. Another implication of the increase in papillae is that new blood vessels are formed during the treatment with vitamin C. The newly formed blood vessels show a normal anatomical structure in confocal microscopical examination and are apparently integrated in a healthy vascular architecture. In the confocal images there are no signs of pathologic changes of the vasculature like, for example, increased diameter, parallel orientation or clew of vessels in a papillae. No perivascular infiltrations of lymphocytes could be observed during the treatment period. The mechanism, by which Vitamin C restores dermal papillae is unknown. Conclusions These are early results which strongly suggest that topical vitamin C may have important anti-aging effects in correcting the structural and functional losses associated with skin aging. Competing intrests The studies were funded by the Beiersdorf AG, Germany. Parts of the results presented in the manuscript are subject of patents pending by Beiersdorf AG. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC522805.xml |
546414 | Modulation of epithelial sodium channel (ENaC) expression in mouse lung infected with Pseudomonas aeruginosa | Background The intratracheal instillation of Pseudomonas aeruginosa entrapped in agar beads in the mouse lung leads to chronic lung infection in susceptible mouse strains. As the infection generates a strong inflammatory response with some lung edema, we tested if it could modulate the expression of genes involved in lung liquid clearance, such as the α, β and γ subunits of the epithelial sodium channel (ENaC) and the catalytic subunit of Na + -K + -ATPase. Methods Pseudomonas aeruginosa entrapped in agar beads were instilled in the lung of resistant (BalB/c) and susceptible (DBA/2, C57BL/6 and A/J) mouse strains. The mRNA expression of ENaC and Na + -K + -ATPase subunits was tested in the lung by Northern blot following a 3 hours to 14 days infection. Results The infection of the different mouse strains evoked regulation of α and β ENaC mRNA. Following Pseudomonas instillation, the expression of αENaC mRNA decreased to a median of 43% on days 3 and 7 after infection and was still decreased to a median of 45% 14 days after infection (p < 0.05). The relative expression of βENaC mRNA was transiently increased to a median of 241%, 24 h post-infection before decreasing to a median of 43% and 54% of control on days 3 and 7 post-infection (p < 0.05). No significant modulation of γENaC mRNA was detected although the general pattern of expression of the subunit was similar to α and β subunits. No modulation of α 1 Na + -K + -ATPase mRNA, the catalytic subunit of the sodium pump, was recorded. The distinctive expression profiles of the three subunits were not different, between the susceptible and resistant mouse strains. Conclusions These results show that Pseudomonas infection, by modulating ENaC subunit expression, could influence edema formation and clearance in infected lungs. | Background The epithelial sodium channel (ENaC) is expressed in epithelial cells of several tissues involved in salt and water reabsorption. The channel is composed of three related subunits (α, β, γ) that are able to reconstitute a functional channel when expressed in Xenopus laevis oocytes [ 1 , 2 ]. ENaC is expressed in a wide range of tissues, including the kidney [ 1 , 3 - 5 ], distal colon [ 1 , 3 , 5 ], lung [ 6 - 8 ], ear epithelium [ 9 , 10 ], papilla of the tongue [ 11 - 13 ], eyes [ 14 ], chondrocytes [ 15 ] and differentiating epithelia [ 16 ]. ENaC synthesis and activity are highly regulated by hormones, such as aldosterone, vasopressin and catecholamines, by intracellular pH, feedback inhibition and extracellular proteases [ 17 , 18 ]. In the lung, vectorial Na + transport from the alveoli to the interstitium is the main force that drives water out of the alveoli [ 19 , 20 ]. This transport mechanism plays a crucial role late in gestation and at birth when sodium transport is involved in lung liquid clearance [ 21 ]. Its importance at birth has been shown unambiguously in αENaC gene knockout mice, where the inability to clear lung water rapidly leads to hypoxemia and death [ 22 ]. Na + transport is also important in adults for lung liquid clearance [ 19 , 23 ]. Increased ENaC expression has been detected in the lung and in alveolar epithelial cells in vitro , following stimulation with steroids, β-agonists, catecholamines, and agents that increase cAMP concentration [ 24 - 27 ]. αENaC expression in the lung is modulated at birth when considerable liquid clearance is required [ 3 , 6 , 27 ]. It is also upregulated during hyperoxia [ 28 , 29 ] and downregulated during hypoxia, which could explain high altitude lung edema (HALE) [ 30 , 31 ]. Several lines of evidence suggest that up-regulation or downregulation of ENaC activity in the lung could be associated with lung infection. In type I pseudohypoaldosteronism, a recessive genetic disease leading to a non-functional ENaC, susceptibility to lung infection has been reported [ 32 - 34 ]. Although ENaC is not the primary defect associated with cystic fibrosis (CF), airway cells from CF patients show a 2–3-fold increase in Na + transport compared to normal cells [ 35 , 36 ]. This sodium hyperabsorption results from the inability of cystic fibrosis transmembrane regulator (CFTR) in CF cells to downregulate ENaC activity [ 37 , 38 ]. Pseudomonas aeruginosa is a bacterium occuring naturally in a wide range of environments such as in soil, fresh and seawater, plants and decomposing organic matter [ 39 ]. Although not usually pathogenic, this common bacterium can evoke opportunistic infections in immunodeficient persons, such as patients with severe burns [ 39 ]. Pseudomonas can promote nosocomial lung infection after artificial ventilation [ 40 ] and is also present in patients with bronchiectasis [ 39 ]. Chronic lung infections are the major cause of morbidity and mortality in CF patients [ 41 ] where Pseudomonas aeruginosa is the main source of chronic lung infection in CF patients [ 42 ]. Instillation of Pseudomonas aeruginosa in the lung of anaesthetised rabbits has been reported to promote acute pneumonia, resulting in alveolar epithelial injury, loss of epithelial barrier integrity, lung edema, pleural empyema and pleural effusions within 8 h of infection [ 43 ]. A more chronic pneumonia model has been developed in the mouse by the intratracheal instillation of P. aeruginosa entrapped in agar beads. In this model, the lungs of susceptible mouse strains develop severe lung infection with a strong inflammatory response and some lung edema [ 44 , 45 ]. Pseudomonas by itself has been shown to inhibit active sodium absorption in cultured airway epithelial cells [ 46 ]. Here, we studied its impact on the expression of genes involved in the modulation of liquid absorption in alveolar and airway epithelium, namely the three ENaC subunits and the catalytic subunit of Na + -K + -ATPase. Pseudomonas entrapped in agar beads was instilled in the lung of resistant (BalB/c) and susceptible (DBA/2, C57BL/6 and A/J) mouse strains, and the expression of α, β, γENaC and α 1 Na + -K + -ATPase mRNA was studied by Northern blotting in lungs infected between 3 hours and 14 days. Methods Infection of mice with P. aeruginosa Clinical strain 508 of P. aeruginosa (provided by Dr. Jacqueline Lagacé, Université de Montréal, Montreal, Canada) was entrapped in agar beads, and 50-μl suspensions containing 2 × 10 5 to 1 × 10 6 CFU/ml were instilled intratracheally in male mice of resistant (BALB/c) or susceptible (DBA/2, C57BL/6 and A/J) strains as described previously [ 44 , 45 ]. Macrophage and polymorphonuclear (PMN) counts in bronchoalveolar lavage (BAL) BAL were performed as described elsewhere with a few modifications [ 44 ]. The infected mice were sacrificed by CO 2 inhalation at different time points after P. aeruginosa instillation in the lungs. The trachea was cannulated, and the lungs were washed seven times with 1 ml PBS. Total cell counts were conducted in a hemacytometer. Differential cell counts were made by Diff-Quick staining (American Scientific Products) of Cytospin preparations. Number of animals: day 1, n = 6; day 4, n = 16; day 6, n = 6; day 14, n = 3. Northern blotting The lungs from infected mice were harvested between 3 h to 14 days after infection, homogenized in 5 ml of 4 M guanidine isothiocyanate, and centrifuged on a cesium chloride gradient [ 44 ]. Fifteen to 20 μg of total RNA purified from the lungs were electrophoresed on 1% agarose-formaldehyde gel and transferred to Nytran membranes (Schleicher & Schuell, Keene, NH, USA) by overnight blotting with 10 X SSC. Hybridization was performed, as reported previously [ 3 ], in Church buffer (0.5 M Na phosphate, pH 7.2, 7% SDS (w/v), 1 mM EDTA, pH 8) [ 47 ]. The nylon membranes were hybridized successively with different cDNA probes. (αENaC, βENaC and γENaC, α 1 Na + -K + -ATPase, glyceraldehyde-3-phosphate dehydrogenase (GADPH) or 18S rRNA). To detect αENaC mRNA, the blots were hybridized with 764-bp mouse αENaC cDNA (His-445 to stop codon) [ 3 ]. The probes for rat β and γENaC cDNA were gifts from Dr. B.C. Rossier (Institut de pharmacologie et de toxicologie de l'Université de Lausanne, Lausanne, Switzerland) and coded for the entire cDNA [ 2 ]. The α 1 Na + -K + -ATPase probe was a gift from Dr. J. Orlowski (Physiology Department, McGill University, Montreal, Quebec, Canada) and consisted of a NarI-StuI 332-bp fragment coding from nucleotide 89 to 421 (from the 5'UTR to Arg-61) of the rat kidney and brain α isoform [ 48 ]. For quantitative study, αENaC mRNA expression was normalized to murine GADPH with a 455 bp cDNA probe cloned between nucleotide 146 and 601 [ 44 ] or with 18S rRNA, using a 640-bp cDNA probe between nucleotidet 852 and 1492 of the rat 18S rRNA sequence [ 26 ]. The blots were exposed to Kodak Xar-film with an intensifying screen, or to a PhosphorImager (Molecular Dynamics, Sunnyvale, CA, USA) for densitometric analysis. Because different strains of mice were investigated in this study (BalB/c, DBA/2, C57BL/6 and A/J), the expression of the different mRNA was calculated at each time point as the % of expression relative to an untreated control from the same strain. The data from the different strains were pooled and subjected to statistical analysis. Between each round of hybridization, the membranes were stripped by treatment with 0.1 X SSC, 1% SDS and 2.5 mM EDTA at 95°C. The blots were allowed to cool gradually with agitation for 30 min at room temperature. The membranes were then rinsed with 5 X SSC and rehybridized. Number of animals: n = between 6 and 8 animals for each time point and each mRNA studied. Statistics For the BAL cell count, the data are presented as means ± SE (standard error). For ENaC and Na + , K + -ATPase mRNA expresion, the comparisons between groups were analyzed by Wilcoxon signed rank non-parametric test using Statsview software (SAS Institute, Inc., Cary, NC, USA). Probability p values < 0.05 were considered to be significant. Results Inflammation in mice infected with P. aeruginosa The inflammation process evoked by Pseudomonas instillation in the lung of C57BL/6 mice was monitored by studying the number of total cells in BAL at different times after infection. As shown in Figure 1 , the inflammation process was more pronounced on days 1 and 4 post-infection. Significant PMN recruitment was noted on day 1 after infection since these cells constituted 90% of the cell population in BAL at that time (Fig. 1 ). The proportion of PMN decreased gradually over time. On days 6 and 14, there was a significant reduction of PMN in BAL (p < 0.05) compared to day 1. PMN still constituted 18% of the cells in BAL on day 14. The infection also led to modulation in the number of macrophages with a significant increase (p < 0.05) on day 4 post-infection (Fig. 1 ). Figure 1 Differential cell counting in C57BL/6 mice infected with 1–2 × 10 5 Pseudomonas aeruginosa embedded in agar beads. Pseudomonas infection leads to strong inflammation with recruitment of PMN and macrophages in bronchoalveolar lavage on days 1 and 4 post-infection. Day 1, n = 6; day 4, n = 16; day 6, n = 6; day 14, n = 3. Differential cell counting: PMN in light grey, macrophages in dark grey. Modulation of α, β and γENaC expression following lung infection with P. aeruginosa Pseudomonas embedded in agar beads was administered intratracheally in resistant (BALB/c) and susceptible strains of mice (DBA/2, C57BL/6, A/J) as described previously [ 44 , 49 ]. α, β and γENaC expression in infected lungs was measured by Northern blot hybridization (Fig. 2 ). Expression of the three subunits was highly modulated in time after lung infection, but showed a similar pattern between the four mouse strains tested. The BALB/c strain that is resistant to Pseudomonas infection [ 45 ], as well as the DBA/2, C57BL/6 and A/J susceptible strains, showed increased α, β and γENaC expression at 24 h, followed by a decrease on day 3 post-infection. The GADPH standard gene did not manifest any modulation of its expression. Densitometric quantitative analyses of the Northern blots were performed for the four mouse strains. The relative expression at each time point was determined relative to uninfected animals of the same strain and the data from the 4 strains were pooled for analysis (Fig. 3 ). αENaC mRNA expression presented a significant decline to a median of 43% on days 3 and 7 post-infection, and was still decreased to a median of 45% on day 14 post-infection compared to uninfected controls (p < 0.05, Fig. 3 ). βENaC mRNA expression was increased to a median of 241% of uninfected control values, 24 h post-infection (p < 0.05), and was followed by a decrease to medians of 42% and 54% on day 3 and 7 post-infection (p < 0.05) (Fig. 3 ). Although the expression of γ ENaC mRNA showed an expression pattern very similar to the α and βENaC subunits, with an increased expression at 24 h (median of 171%) followed by a decreased expression on day 3 (median of 53%) and 7 (median of 66%) of infection, these changes failed however to reach significance (Fig. 3 ). No modulation of α, β or γENaC mRNA was detected when the lungs were instilled with agarose beads only (data not shown). We also investigated the expression of α 1 Na + -K + -ATPase mRNA coding for the catalytic domain of the sodium pump, but could not find any significant change during infection (Fig. 3 ). Figure 2 Expression of α, β and γENaC mRNA in the lung following infection with Pseudomonas aeruginosa . Representative Northern blot of α, β and γENaC mRNA expression following infection with Pseudomonas in resistant (BalB/c) and susceptible (DBA/2, C57BL/6 and A/J) strains of mice. There is a characteristic modulation of the three ENaC subunits that is not different between strains. Figure 3 Densitometric analysis of the modulation of αENaC, βENaC, γENaC and α 1 Na + -K + -ATPase mRNA following Pseudomonas infection. The modulation of αENaC, βENaC, γENaC and α 1 Na + -K + -ATPase mRNA by Northern blots hybridization was subjected to a densitometric analysis. Because different strains of mice were investigated in this study (BalB/c, DBA/2, C57BL/6 and A/J), the expression of the different mRNA was calculated at each time point as the % of expression relative to an untreated control coming from the same strain. The α and βENaC mRNA were modulated at some time point by Pseudomonas infection compared to uninfected animals. There was no modulation for γENaC or α 1 Na + -K + -ATPase mRNA. αENaC mRNA was downregulated compared to uninfected controls on days 3, 7 and 14 post-infection (*, p < 0.05). βENaC mRNA was elevated at 24 h post-infection (*, p < 0.05) compared to uninfected controls and was downregulated thereafter on day 3 and 7 post-infection (*, p < 0.05). Number of animals: n = between 6 and 8 animals for each time point and each mRNA studied. Discussion The instillation of Pseudomonas enmeshed in agarose beads in the lung is a good model to study lung inflammation [ 44 , 45 ] and lung injury [ 43 ] secondary to an infection. For this study, P. aeruginosa enmeshed in agarose beads was instilled into the mouse lung because the model allows the development of chronic lung infection in susceptible mouse strains [ 44 , 45 ]. The infection leads to cellular infiltration and alveolar edema that stand on day 3 post-infection and that can be still demonstrated on day 14 post-infection in Pseudomonas -susceptible mouse strains [ 45 ]. Because the lung inflammation associated with Pseudomonas infection is accompanied by lung injury [ 43 ], and because we have shown recently that ENaC expression can be modulated under conditions that promote lung injury [ 50 ], we tested here if Pseudomonas was affecting the mRNA expression level of the three ENaC subunits as well as the catalytic subunit of the Na + pump since these elements are involved in lung liquid balance across the alveolar epithelium [ 19 , 23 ]. The results reported here indicate that Pseudomonas infection modulated the expression of the three ENaC mRNA with a characteristic pattern. There was no significant difference, however, in the expression profile of ENaC mRNA between the Pseudomonas -resistant (Balb/C) and -susceptible (DBA/2, C57BL/6, A/J) mouse strains. Modulation of ENaC expression is therefore most likely not a genetic marker linked to the susceptibility of mouse strains to establishment of a chronic infection with Pseudomonas . Pseudomonas infection affected ENaC mRNA with a pattern consisting of increased expression at 24 h, followed by a marked decrease on day 3 post-infection. The change in ENaC mRNA was related to bacterial infection, since agarose beads alone failed to evoke any modulation of these RNA. The three ENaC subunits were modulated with a similar profile, with some noticeable differences, however. Although αENaC mRNA expression tends to increase by day 1, the most noticeable feature brought by Pseudomonas infection to αENaC mRNA was the significant decreases after 3 days, 7 days and 14 days post-infection. To the best of our knowledge, this is the first report demonstrating that bacterial infection in vivo can lead to modulation of ENaC mRNA expression. Recently, αENaC mRNA expression was found to be downregulated in the mouse lung after 7 and 14 days of adenoviral infection [ 51 ]. Furthermore, there is some evidence that αENaC expression is also decreased in other models of lung injury. Folkesson et al. [ 52 ] reported a decline in ENaC expression following subacute lung injury, 10 days after intratracheal administration of bleomycin. More recently, we recorded a decrease in ENaC expression after ischemia-reperfusion lung injury [ 50 ]. All these results, and the results reported in the present report, suggest that the modulation of ENaC expression associated with lung infection could be a widespread mechanism, not specific to a given pathogen or injury process, but a general response of the lung to inflammation and injury. The β ENaC subunit was also modulated by Pseudomonas infection. There was a significant increase in the mRNA expression on day 1 post-infection, followed, as for αENaC, by a decreased expression on day 3 and day 7 post-infection. Different stoichiometries have been proposed for ENaC. One model suggests a 2α, β, γ ratio [ 53 , 54 ] whereas others postulate an octomeric [ 55 ] or nonameric structure [ 55 , 56 ]. Although the expression of the α subunit alone is sufficient to allow ENaC activity [ 1 ], the three subunits are needed to get a fully functional channel [ 2 ]. The expression of the three subunits in Xenopus laevis oocytes increases amiloride-sensitive Na + current by 100% compared to αENaC alone [ 2 ]. The α, β and α, γ channels are 20 times less effective in driving amiloride-sensitive current than the native channels and show differences in their biophysical properties [ 57 , 58 ]. Gene inactivation or over-expression of the different ENaC subunits has revealed important differences in the role each subunit plays in lung liquid management. αENaC knockout mice develop respiratory distress and die within 40 h from birth because of their inability to clear lung liquid [ 22 ]. Lung liquid clearance at birth is also slower in γENaC knockout mice [ 59 ], but is not affected in βENaC knockouts [ 60 ]. Increased transgenic expression of βENaC targeted in the airway epithelia, but not α or γ subunits, showed an increase Na + transport across the airway epithelium and a reduced height of the airway surface liquid [ 61 ]. For all these reasons, it is difficult to predict how the modulation of ENaC mRNA expression and its effect on the ratio of the three subunits, would have an impact on ENaC activity. In addition, ENaC mRNA content also does not necessarily reflect the amount of active channel at the membrane. One thing seems clear however, because of its prominence in the lung, the diminution of αENaC expression that we detected in the lung following Pseudomonas infection, could certainly influence amiloride-sensitive current and lung liquid clearance as in αENaC KO mice rescued by transgenic expression of αENaC that has a lower expression of ENaC in the lung [ 62 , 63 ]. In such model, there is a reduced ENaC current in tracheal cells [ 63 ], and a much slower lung liquid clearance following thiourea or hyperoxia-induced lung edema [ 64 ]. The general biphasic modulation of ENaC mRNA expression with an increase at 24 h followed by a decrease thereafter is an interesting finding that could explain some contradictory reports concerning ENaC expression in lung following Pseudomonas infection. Acute bacterial pneumonia in rats has been shown to increase alveolar epithelial fluid clearance [ 65 , 66 ] when in late pneumonia, there is a decrease in the lung liquid clearance ability of the lung [ 66 ]. These contradictory results could be well explained by the modulation of ENaC expression reported here. The long term ENaC downregulation by Pseudomonas infection could be of potential clinical significance to understand the slow improvement in some ARDS patients. In contrast to α and β ENaC, α 1 Na + -K + -ATPase mRNA was unaffected in the course of lung infection. This is similar to what has been reported during adenovirus lung infection [ 51 ] where αENaC, aquaporin 1 (AQP1) and AQP5 mRNA show decreased expression, but not α 1 Na + -K + -ATPase. In ischemia reperfusion injury, there was also no modulation of α 1 Na + -K + -ATPase expression despite significant ENaC downregulation [ 50 ]. These results, as well as the data reported here, suggest that the inflammatory process seems to selectively affect, and not in a non-specific way, some elements of lung liquid clearance. It would be difficult at this time to speculate on the reasons for this modulation. Na + -K + -ATPase is an important element in lung liquid clearance, however, by being one of the key generator of membrane potential, the enzyme also affects other channels and ion transport process. It is possible that by modulating ENaC expression and not α 1 Na + -K + -ATPase, Pseudomonas infection alters the Na + transport system but does not change other important cell functions meditated by Na + -K + -ATPase. Furthermore, despite a similar mRNA expression level, there could be a fall in protein content or activity of the sodium pump. Additional experiments are necessary to answer this question. Several studies report that in lung epithelial cells, viral infection [ 67 , 68 ], mycoplasma [ 69 ], bacterial infection [ 70 , 71 ], and inflammatory cytokines such as tumor necrosis factor-α (TNF-α) [ 70 , 72 ], interleudin-1β (IL-1β) [ 73 ], or TGF-β [ 74 ] decrease the expression of water channels, such as AQP1 and AQP5 and reduce the short circuit current generated by cells. Adenoviral lung infection in mice results in pulmonary inflammation and lung edema with lowered expression of AQP1, AQP5 and αENaC [ 51 ]. All these data, including the results presented here, suggest that lung inflammation, by decreasing the expression of αENaC and water channels, could hamper the liquid clearance ability of the lungs and favour edema formation. Conclusions We have shown in this report that Pseudomonas infection modulates ENaC mRNA expression. This modulation is independent of mouse strain susceptibility to establishment of chronic infection with Pseudomonas . Although there is an elevation of ENaC expression after 24 h, the most important feature is probably the long-lasting decrease of αENaC transcripts on days 3 and 7 post-infection. The lung inflammation induced by Pseudomonas infection therefore seems to favour a reduction in the expression of an essential element involved in lung liquid clearance as well as the regulation of airway surface liquid volume. Authors' contributions AD performed the hybridization, the statistical analysis of the blots and wrote the manuscript. DG performed the Pseudomonas instillation, RNA extraction and Northern blotting of RNA sample. The BAL recovery as well as PMN and macrophage counting was performed by CG. YB and DR designed and co-ordinated the study. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC546414.xml |
554987 | The prevalence of dental erosion in Nigerian patients with gastro-oesophageal reflux disease | Background In various people of the Western world, gastro-oesophageal reflux (GOR) has been reported to be a common problem. Various studies have also assessed the relationship between GOR and dental erosion. The authors are not aware of such studies in Nigerians. It is therefore the aims of the present study to estimate the prevalence of GOR; to estimate the prevalence of dental erosion in patients with GORD; to document the oral findings in patients diagnosed with GORD and to compare these findings with previous studies elsewhere. Methods A total of 225 subjects comprising of 100 volunteers and 125 patients diagnosed with GORD were involved in this study. History of gastric juice regurgitation and heartburn were recorded. Oral examination to quantify loss of tooth structure was done using the tooth wear index (TWI) designed by Smith and Knight (1984). Results Twenty patients with GORD presented with dental erosion in the maxillary anterior teeth with TWI scores ranging from 1–3. The prevalence of erosion was found to be statistically significant between GORD patients (16%) and control (5%) (p < 0.05), but not significant between endoscopic diagnostic groups (p > 0.05). Conclusion The present study supports the consideration of dental erosion as the extra-oesophageal manifestation of GORD. However the association between GORD and burning mouth sensation needs more investigation. | Background Gastro-oesophageal reflux (GOR) is the passage of gastric contents into the oesophagus. Once past the upper oesophageal sphincter, the gastric juice may pass into the oral cavity. The continual exposure of the teeth and other oral structures to gastric refluxate may result in dental erosion and other soft tissue symptoms [ 1 ]. Any acid with a pH below the critical pH of dental enamel dissolution (5.5) can dissolve the hydroxyapatite crystals in enamel. However the critical pH below which enamel dissolves is not constant but is rather inversely proportional to the concentrations of calcium and phosphate in the saliva and plaque fluid [ 2 ]. Gastric refluxate has a pH of less than 2.0 and thus has the potential to cause dental erosion [ 3 ]. Acid regurgitation is a common symptom of upper gastro-intestinal tract disorders and dysfunctions such as peptic ulcer (duodenal and gastric ulcers) and reflux oesophagitis [ 4 ]. Other symptoms include heartburn, non-cardiac epigastric and retrosternal pain [ 5 ]. Prominent among factors precipitating GOR and its complications, gastro-oesophageal reflux disease (GORD) are fatty diets and alcohol. Current understanding of GOR and GORD is that acid reflux into the oesophagus may be caused by three possible mechanisms [ 6 ]: (1) Transient spontaneous or inappropriate relaxations of the sphincter; (2) Transient increase in intra-abdominal or intragastric pressure; (3) Functional abnormality of the lower oesophageal sphincter (LOS). Bargen and Austin in 1937 [ 7 ] first reported the link between dental erosion and gastro-intestinal disturbances. Since then there has been several other studies confirming the relationship between this loss of tooth structure and GORD in the United Kingdom (UK) [ 8 ], United States of America (U.S.A) [ 9 ] and Canada [ 10 ]. A recent survey among young people in the U.K also revealed an association between dental erosion and symptoms of GOR [ 11 ]. Although soft tissue symptoms (non-specific burning sensation in the mouth) have been mentioned in the literature, pathognomonic soft tissue lesions have not been documented [ 1 ]. But some researchers have reported a lack of relationship between periodontal lesions and GOR, since the prevalence of periodontal lesions is similar in patients with GORD and in healthy volunteers [ 12 ]. In various people of the Western world, GOR has been reported to be a common problem, often related to meals and occurring in about 60% of the population at some point in their lives [ 13 ]. Whereas the prevalence of GORD is estimated to range from 6% – 10% [ 14 , 15 ], Meurman et al [ 16 ] examined 117 patients with GORD, of whom 28 (24%) had dental erosion. Also, Schroeder et al [ 17 ] identified dental erosion in 11 (55%) of 20 patients with GORD. The authors are not aware of such figures in Nigerians. A search of the literature also revealed a dearth of information on this condition and its complications among Black Africans. It is therefore the aim of the present study to estimate the prevalence of GOR among patients attending the medical out-patient department of the Obafemi Awolowo University Teaching Hospital Complex Ile-Ife, Nigeria; to estimate the prevalence of dental erosion in patients with GORD; to document the oral findings in patients diagnosed with GORD and to compare these findings with previous studies elsewhere. Methods A total of 225 subjects were involved in the study over a period of 6 months, January – June 2002. The subjects were made up of 100 volunteers attending the medical outpatient department (MOPD) and 125 patients diagnosed with gastro-oesophageal reflux disease among whom the prevalence of GOR and GORD was determined respectively. Their age ranges from 18 – 72 years, with a mean age ± S.D. of 38 ± 10.87 years. Patients presenting primarily with symptoms of asthma, bronchitis and other respiratory disorders were excluded, since they are also at risk of dental erosion [ 18 ]. To assess the prevalence of GOR and GORD, consenting patients attending the medical outpatient department of the Obafemi Awolowo University Teaching Hospital's complex Ile-Ife, Nigeria, were questioned with reference to their experience regarding symptoms of GOR such as regurgitation of gastric juice, epigastric and non-cardiac pain (heartburn). The onset, frequency of occurrence and duration of each of the symptoms were ascertained and recorded. Patients presenting with a history of heartburn 2 or more times per week were diagnosed as having GORD 10 . Oral examination was carried out to quantify any loss of tooth structure using the tooth wear index (TWI) designed by Smith and Knight (1984) [ 19 ]. To document the oral findings associated with GORD, patients referred to the Gastro-intestinal (GIT) endoscopic unit for investigation of gastro-oesophageal tract disease were also evaluated for dental erosion and other soft tissue symptoms such as non-specific burning sensation in the mouth and sensitivity in the tongue. The dental evaluation included history to determine potential eatiological factors responsible for dental erosion. Patients were also examined clinically to quantify loss of tooth structure using the tooth wear index. The oral examination was performed by one of the authors (AO), blind as to the endoscopic diagnosis of subjects. Patients with positive endoscopic findings plus the occurrence of heartburn two or more times per-week were assessed to meet the criteria for GORD. When a clear-cut endoscopic evidence of oesophageal inflammation was seen (marked redness, fibrinous membrane, and or ulcerations), a diagnosis of oesophagitis was made. Gastroduodenal ulcer was diagnosed through endoscopic evidence of an ulcer with a necrotic base or a clear-cut scar. This was grouped into those with gastric ulcer (ulcer in the pylorus, antrum, corpus or fundus) and those with duodenal ulcer (ulcer in the bulbus). Only one diagnosis was given for each patient. Data collected were entered into a computer and analysed using chi-square. P value <0.05 were considered significant. Results Thirty-five of the 100 subjects attending the medical outpatient department had acid regurgitation and acidic taste sensation in the mouth. This occurs most of the time following a full stomach meal and is associated with belching. Table 1 shows the prevalence of GOR and GORD. Eleven reported a history of heartburn 2 or more times per week, while 16 and 10 reported weekly and monthly history of heartburn respectively. Comparison of subjects presenting with heartburn 2 or more times per week (M = 5, F = 6) and those with less than 2 times per week (M = 12, F = 14) shows that more female presented with heartburn. However, there was no statistical difference among the genders with regard to the symptoms recorded. None of these subjects presented with any burning sensation in the mouth or sensitivity in the tongue. In five of those that reported 2 or more weekly history of heartburn, there were minimal losses of tooth structure (TWI scores of 1 and 2 in three and in two subjects respectively) limited to the maxillary central incisors (Table 2 ). Table 1 Prevalence of GOR and symptoms of GORD Symptoms Male (N = 49) Female (N = 51) Total Regurgitation and acidic taste in the mouth 16 19 35 Heartburn ≥ 2 per week 5 6 11 Heartburn per week 7 9 16 Heartburn per month 5 5 10 Chi Square = 0.10; df = 3; p = 0.99. Table 2 Prevalence of dental erosion Dental erosion Subjects/patients No (%) TWI scores M.O.P. (N = 100) 5 (5) 1 – 2 GORD (N = 125) 20 (16) 1 – 3 Mantel Haenszel chi-square = 5.50, p = 0.02. One hundred and twenty five patients were diagnosed with GORD. Twenty presented with dental erosion in the maxillary anterior teeth with TWI scores ranging from 1–3. The prevalence of erosion was found to be statistically significant between GORD patients (16%) and controls (5%) (Mantel Haenszel Chi-square 5.50, p < 0.05), Table 2 . The result of the gastrointestinal endoscopy reported 41(32.8%) as having reflux oesophagitis, 36 (28.8%) with duodenal ulcer and 48 (38.4%) with gastric ulcer. Eight patients with dental erosion came from the group who had reflux oesophagitis, 7 from the group with duodenal ulcer and 5 from the group with gastric ulcer. Comparison of patients with oesophagitis and those without shows that the mean age of patients with oesophagitis was lower than in those without. The prevalence of erosion was not statistically significant between the endoscopic diagnostic groups (chi-square = 1.33, df = 2, p = 0.51), Table 3 . Table 3 Endoscopic diagnostic groups: basic data and number of patients withdental erosion Diagnostic groups Number Mean age (yrs) Gender Dental erosion detected No (%) ± S.D Male Female Reflux oesophagitis 41 (32.8) 36.9 ± 9.7 24 17 8 Duodenal ulcer 36 (28.8) 39.2 ± 11.4 16 20 7 Gastric ulcer 48 (38.4) 38.1 ± 11.0 17 31 5 Prevalence of erosion between diagnostic groups: Chi-square = 1.33, df = 2, p = 0.51. The details of the orodental findings in the twenty patients who had dental erosion are shown in Table 4 . There seems to be a slight association between the duration of gastrointestinal symptoms and the severity of erosion. TWI scores of 3 were seen only in those patients whose abdominal symptoms had lasted 10 years or more. The palatal surfaces of the maxillary anterior teeth were usually involved but the central incisors were the most severely affected. Six patients (4 from reflux oesophagitis and 2 from duodenal ulcer) had dental erosion related symptoms in their teeth: sensitivity to cold and heat. The teeth that presented with dentine sensitivity had TWI scores of 3. Table 4 Orodental findings in the twenty patients with GORD who had erosions S/N Diagnostic groups*** Age (yrs) Gender Duration (yrs) TWI scores Dentine Sensitivity** BMS* 1 RO 26 Female 5 1–2 NP NP 2 RO 43 Female 10 1–3 P P 3 RO 44 Male 10 1–3 NP NP 4 RO 38 Male 5–10 1 NP NP 5 RO 48 Female 15 2–3 NP P 6 RO 33 Female 10 1–3 P P 7 RO 52 Male 15 2–3 P P 8 RO 50 Female 15 2–3 P P 9 DU 30 Male 5–10 1–2 NP NP 10 DU 50 Male 20 2–3 P P 11 DU 45 Female 15 1–3 NP NP 12 DU 47 Female 10 1–2 NP NP 13 DU 33 Female 5 1 NP NP 14 DU 36 Male 10 1–3 NP NP 15 DU 43 Male 15 2–3 P P 16 GU 46 Female 10 1 NP NP 17 GU 35 Female 10 1–2 NP NP 18 GU 32 Female 5 1 NP NP 19 GU 54 Male 20 2–3 NP P 20 GU 45 Male 15 1–3 NP P *BMS = Burning mouth sensation and sensitivity in the tongue **NP = Not present, P = Present ***RO = Reflux oesophagitis, DU = Duodenal ulcer, GU = Gastric ulcer. Burning mouth sensation and peppery sensation in the tongue where reported by 9 patients who had had gastrointestinal symptoms for 10–20 years. Discussion Various methods have been employed in the investigation of GOR including endoscopy with biopsy and contrast radiography. Although twenty-four hour oesophageal pH monitoring is considered the gold standard investigation of GOR [ 20 ], due to non-availability of the ambulatory pH recorder in our center, GOR was diagnosed by endoscopy where visual identification of mucosal inflammation and oesophagitis was used to identify the existence of GOR. Several other authors have used this method [ 4 , 21 ]. The prevalence of GOR and GORD in the present study was 35% and 11% respectively. The prevalence of GOR was much lower than the 60% reported in the U.K. and other Western societies [ 13 ], whereas that of GORD was a little above the reported range of 6%–10% [ 14 , 15 ]. A lower percentage (10%) of the subject studied reported a monthly history of heartburn as compared to 59% of the population reported by Lock et al [ 22 ]. The present study examined 125 patients with GORD of which 20(16%) had dental erosion. This is also lower than the 24% and 55% reported by Meurman et al [ 16 ], and Schroeder et al [ 17 ] respectively. Although the authors cannot propose any reason for the reported lower prevalence of GOR and GORD, the low prevalence of dental erosion may be related to difference in diet. More so that the thrust of the European studies support the view that erosion (resulting from acidic and carbonated foods and beverages) is more important than attrition in the aetiology of tooth wear [ 23 ], whereas, attrition have been reported to be more important than erosion in the aetiology of tooth wear among Nigerians [ 24 , 25 ]. Also in the study of gastro-oesophageal reflux in children and its relationship to erosion of primary and permanent teeth, diet was considered to be a major contributory factor [ 26 ]. The consumption of acidic beverages among the study population was low, hence the silence on its contribution to the overall disease level. Although this is controversial, the authors are of the opinion that intrinsic (gastric) acid results in palatally eroded sites (as shown in Figure 1 ) while extrinsic (dietary) acids lead to labial or vestibular erosions. The authors observed some degree of tooth wear (tooth surface loss) on the molar teeth. It was not documented because we believe they are more likely due to "attrition" resulting from rigorous mastication of the more fibrous Nigerian diet. However, acidic refluxate may have been a contributory factor, since tooth surface loss is a multi-factorial disease. Figure 1 Severe dental erosion affecting the palatal surfaces of the upper anterior teeth in a patient with reflux oesophagitis. In accordance with the result of Gregory-Head et al [ 27 ], patients diagnosed with GORD in this study had higher TWI scores compared with control subjects. All the twenty patients that presented with dental erosions in the present study had an underlying gastrointestinal pathosis (Gastric oesophagitis, Duodenal ulcer, and Gastric ulcer) with increased output of acid secretion into the stomach [ 28 ]. There was however no statistically significant difference in the prevalence of dental erosions in these diagnostic groupings, Table 3 . This is in agreement with the study of Jarvinen et al [ 4 ]. The severity of dental erosions may depend on the frequency of regurgitation and duration of the gastro-oesophageal reflux. In the present study, patients with TWI scores of 3 had their symptoms for more than 10 years as shown in Table 4 . This is supported by the study of Loffeld et al [ 21 ], which revealed a significant association between duration of complaints and presence of damage in the upper incisors, but in contrast to the findings of Jarvinen et al [ 4 ] who reported no direct association between the frequency of regurgitation symptoms and the severity of the erosive lesions. Six patients presented with erosion related symptoms in their teeth, they had dentine sensitivity to cold and heat. All of these patients had dental erosions with TWI scores of 3 (Loss of enamel exposing dentine for more than one-third of the surfaces). This probably explains the sensitivity. Nine patients reported burning mouth sensation and sensitivity in the tongue. They describe the sensation as peppery/burning feeling in the vestibule of the mouth and mostly on the dorsal surface of the tongue. These may have resulted from the prolonged effect of acidic gastric refluxate on the oral mucosa and on the papillae of the tongue. Conclusion The present study supports the consideration of dental erosion as the extra-oesophageal manifestation of GORD. However the association between GORD and burning mouth sensation needs more investigation. Competing interests The author(s) declare that they have no competing interests. Authors' contributions AOO conceived of the study, participated in its design, performed the dental examination including the application of the tooth wear index (TWI), and participated in the initial draft and final write-up of the manuscript. EAA and DAN performed the endoscopic examination of patients with gastro-oesophageal reflux disease, participated in the initial draft and final write-up of the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554987.xml |
544558 | An SVD-based comparison of nine whole eukaryotic genomes supports a coelomate rather than ecdysozoan lineage | Background Eukaryotic whole genome sequences are accumulating at an impressive rate. Effective methods for comparing multiple whole eukaryotic genomes on a large scale are needed. Most attempted solutions involve the production of large scale alignments, and many of these require a high stringency pre-screen for putative orthologs in order to reduce the effective size of the dataset and provide a reasonably high but unknown fraction of correctly aligned homologous sites for comparison. As an alternative, highly efficient methods that do not require the pre-alignment of operationally defined orthologs are also being explored. Results A non-alignment method based on the Singular Value Decomposition (SVD) was used to compare the predicted protein complement of nine whole eukaryotic genomes ranging from yeast to man. This analysis resulted in the simultaneous identification and definition of a large number of well conserved motifs and gene families, and produced a species tree supporting one of two conflicting hypotheses of metazoan relationships. Conclusions Our SVD-based analysis of the entire protein complement of nine whole eukaryotic genomes suggests that highly conserved motifs and gene families can be identified and effectively compared in a single coherent definition space for the easy extraction of gene and species trees. While this occurs without the explicit definition of orthologs or homologous sites, the analysis can provide a basis for these definitions. | Background Several methods have been developed for the detailed global comparison of multiple whole genomes and the production of global phylogenies. Most of these methods require the prior identification and selection of a reasonably small subset of putative orthologs within which individual homologous sites are identified with some degree of confidence using alignment [ 1 - 7 ]. Frequently, detailed alignment information is subdivided and compressed into a smaller number of complex characters (such as gene content or gene order), which are then used for quantitative comparison [[ 4 , 5 ]; see [ 6 ] for review], but the more or less direct use of large scale sequence alignments have also been attempted [ 7 ]. Though generally less developed, many non-alignment methods, considered initially by Blaisdell [ 8 ], are currently being explored for a similar purpose [[ 9 - 15 ]; see [ 17 ] for review]. Rarely do such methods simultaneously provide 1) detailed and unbiased comparisons of a high fraction of biomolecular sequences within full genome datasets, and 2) globally consistent gene and species trees based on this exhaustive comparison. We have recently developed an SVD-based phylogenetic method that provides accurate comparisons of a high fraction of sequences within whole genomes without the prior identification of orthologs or homologous sites [ 13 ]. This method has been successfully applied to a number of diverse genome datasets, including mitochondrial genomes, bacterial genomes, and viral genomes [ 13 - 15 ]. Here we apply this method to a diverse set of nine complete eukaryotic nuclear genomes, resulting in the production of a species tree based on the automatic identification and simultaneous comparison of over 400 conserved amino acid motifs and gene families. Results and discussion Proteome data sets and sequence conversion The nine eukaryotic genomes compared in this analysis are listed in Table 1 . The protein sets obtained from NCBI for the malaria parasite (Pfal) and the budding yeast (Scer) each contributed only 3% of the 175,559 total proteins in the dataset, while the proteins for Frub provided nearly 21% of the total. Only the Frub proteins were obtained from the Ensemble Genome Browser [ 16 ], since protein predictions for this organism were not available from NCBI. Differences in methods used to predict proteins by these two organizations might be responsible for the large difference in the number of proteins predicted for comparable vertebrate genomes (>37,000 for Frub, but only 21–25,000 for all other vertebrates). These differences could, in principle, drastically effect the gene and species trees derived from a global comparison of all proteins. However, the position of Frub in the final species tree suggests that these effects were relatively minor (see below). We have noted previously that even drastic genome size differences can be accommodated by our method [ 14 ]. SVD-derived vector definitions for motifs and gene families All the proteins in the dataset were recoded as overlapping tetrapeptide frequency vectors and the resulting data matrix was decomposed by the SVD. A total of 437 singular triplets were obtained as output. The "protein" vectors provided in the "right" factor matrix are known to provide reduced dimensional definitions for all proteins in the dataset as linear combinations of the orthogonal "right" singular vectors (rsv's). Conversely, the right singular vectors themselves frequently represent "ideal" versions of proteins defining a given gene family [ 13 , 14 ] Protein vectors having the strongest projections on a given rsv are therefore likely to represent members of a given gene family. In this analysis, the proteins with the five strongest projections (referred to as the "top 5") for each rsv were used to identify and summarize a number of gene families. The total number of proteins from each species that appear in the "top 5" for all 437 right singular vectors are listed in Table 1 . Although the fraction of "top 5" proteins identified by the SVD roughly parallels the fraction of total proteins from each species, the mammalian proteins tend to dominate the analysis. Each right singular vector can potentially define two distinct gene families. In this case, the highest positive elemental values within a vector identify proteins associated with one protein family, while the highest negative values identify proteins associated with an anti-correlated family (i.e. proteins that rarely share the same tetrapeptides). Frequently, however, strong family definitions are provided for just one protein family. In this case, the anti-correlated proteins are seen to be derived from a mixture of two or more families. Since the choice of sign is arbitrary, strong family definitions are equally likely to be provided by either the positive or the negative values within a vector. Family definitions provided by positive vector values are denoted below using the simple vector index (e.g. 277 = the 277 th singular vector). Those provided by negative vector values are followed by an "a" (e.g. 277a). Its worth noting at this point that protein family definitions provided by the SVD necessarily account for not only "what is there" (tetrapeptides that form the motifs that define the family), but also "what is not there" (tetrapeptides excluded by that family of proteins, but likely to form anti-correlated motifs within other families of proteins). Protein family definitions provided by right singular vectors An abbreviated list of 58 protein families identified within the 437 SVD-derived singular triplets are provided in Table 2 . For each listed singular triplet, the gi# of an example protein chosen from among the "top 5" values within the right singular vector is provided, along with its corresponding Name and a Protein Description provided within the NCBI annotation for that protein. In general, proteins described by the more dominant singular triplets were selected for presentation from the complete list of 437 triplets. However, some were chosen due to their historical utility for evolutionary comparisons (ribosomal proteins) and/or their tendency to be accompanied by strongly correlated peptide motifs (last column of Table 2 ). Relatively few families appear in the table due to the fact that some vectors strongly describe only one family rather than two, some vectors describe only families from species that lack annotation or are poorly annotated at NCBI (i.e. Frub proteins, Agam proteins, etc.), some vectors describe protein families listed by NCBI merely as "unknown" or "conserved unknown", some vectors describe proteins with weakly conserved motifs, and some vectors describe distinct subfamilies of proteins. In the latter case, multiple right singular vectors are apparently required in combination to describe some of the more diverse families of proteins. Included in Table 2 is the number of singular vectors that include the chosen example protein within its "top 5". When multiple vectors are involved in defining multiple related subfamilies, the most "dominant" vector (the one on which the example protein casts its strongest projection) is listed in the first column. Thus, some proteins are seen to have multiple subfamily affiliations. The multiple vectors observed per family effectively subdivide the 58 families into 179 distinct subfamilies. For instance, Table 2 includes a set of 18 ribosomal protein families described by a total of 65 singular vectors (highlighted in bold). Ribosomal proteins are frequently well conserved, effectively aligned, and commonly used for estimating evolutionary relationships. Their presence within our list of dominant singular vectors suggests their utility for establishing evolutionary relationships even in the absence of explicit alignments and explicit a priori assignments of orthology. The diverse families of ras proteins present within the eukaryotic data set provide good examples of the ability of SVD-derived singular triplets to identify and describe both superfamilies and subfamilies of proteins. The ras proteins are well described by at least 13 vectors, including the 6 dominant vectors highlighted in italics in Table 2 . All the "top 5" members of the protein families identified by these 6 vectors are listed in Table 3 . Vector 197a summarizes the brain-associated ras11 subfamily (Rab11), vector 236a summarizes the Aplysia-related ras subfamily (ApRas), vectors 277 and 277a summarize the brain-associated ras 5 subfamily (Rab5) and the complex Ha/K/Nras subfamily (HaRas) respectively, vector 350a summarizes the ras-related C3 botulinum toxin substrate 1 subfamily (Rac1), and vector 387a summarizes the brain-associated ras1B subfamily (Rab1B). The most dominant ras vector, 389a, appears to describe a more generalized version of the Rab1 subfamily, since this vector includes both Rab1A and Rab1B proteins within the "top five". In addition, as explained below, this vector also summarizes a high fraction of the entire set of 34 ras sequences within all subfamilies. For comparison, KOG and Homologen memberships are also listed, when available, for each of the "top 5" proteins listed in Table 3 . Table 4 provides a similar comparison for a set of four arbitrarily selected protein families unrelated to ras or to each other (potassium channel, enolase, solute carrier protein, and ADP-ribosylation factor). Since most of the genomes used in our study have not yet been included within the KOG classification scheme, only fly and human proteins have official KOG affiliations. However, we expect with high likelihood that most if not all of the top 5 proteins listed in Tables 3 and 4 would also be members of the particular KOG family listed for each vector. Given this, there would be a good correspondence in Tables 3 and 4 between KOG family members and the proteins identified by singular vectors. In contrast, the Homologen resource appears to provide a more selective classification method, dividing the KOG protein families into two or more subfamilies within which members are more likely to represent specific orthologs. Conserved motif definitions provided by left singular vectors Members of any particular ras subfamily represented by a given right singular vector share a uniquely conserved set of correlated tetrapeptides we have previously referred to as a "copep motif". These motifs are explicitly described by the corresponding "left" singular vectors (lsv's) comprising a given singular triplet. The lsv's describe these copep motifs as linear combinations of the 160,000 possible tetrapeptides. Those with high positive values identify peptides found with high probability in the conserved motif of a given subfamily, while those with a high negative value identify peptides excluded with high probability. Therefore, like the rsv's, the lsv's frequently describe two distinct anti-correlated entities (in this case motifs rather than protein families) using either positive or negative values within the vector. Using essentially the same procedure described above for any given rsv, the tetrapeptides having the largest positive or largest negative projections on any given lsv were identified in order to provide a focused summary of the motifs described by that vector. For motif extraction, however, an arbitrary cut-off value (absolute value > 0.025) was used to identify dominant tetrapeptides. In most cases, it is possible to cluster the resulting short list of dominant tetrapeptides into several uninterrupted copep strings formed by tetrapeptides that overlap in 3 of 4 consecutive amino acid positions. Using this procedure, one long copep string was identified for each of the singular triplets listed in Table 2 . The length of the identified long copep string and its corresponding E-value (resulting from pairwise BLAST) are provided as a summary in the last column. The precise amino acid sequences of the long copep strings identified for all listed vectors are provided in a supplementary table [see Additional file 1 ]. The E-values listed provide a measure of the specificity with which each corresponding protein is identified by the copep string extracted from a given lsv. Its important to note that the long copep string provides only an approximate summary of the lsv from which it is extracted, yet the small E-values clearly indicate that the vast majority of the proteins identified in Table 2 are very specifically recognized by their corresponding copep string. Figure 1 provides a more detailed demonstration of how correlated peptide motifs and their associated gene families are simultaneously identified and described by SVD-derived singular vectors. In order to allow a clear comparison of SVD-derived motifs with alignment-derived motifs, the dominant tetrapeptides were superimposed over matching regions of a standard ClustalX alignment of the 34 ras proteins identified in the "top five" of the corresponding right singular vectors listed in Table 3 . In this example, the dominant tetrapeptides extracted from the six selected left singular vectors are demarcated within (shaded/colored) boxes. Many of the dominant tetrapeptides are seen to form extended strings of overlapping peptides that correspond well to conserved contiguous regions within particular subsets of the ras proteins. For example, vectors 350a and 236a identify and provide distinct descriptions for motifs within the Ras-related botulinum toxin C3 substrate proteins (RasC3) and the Aplysia-related ras proteins (ApRas), respectively. The two most dominant left singular vectors of Figure 1 (389a and 387a) describe motifs within overlapping subsets of the nine Rab1 proteins. In addition, the most dominant left singular vector (389a) appears to describe a highly conserved motif within the entire set of 34 ras proteins reasonably well (solid clear boxes). This vector conspicuously identifies dominant tetrapeptides that span the two regions of the alignment in which unbroken strings of two or more invariant amino acids (asterisks) are present. These two regions are known to be required for ras GTPase activity [ 18 ]. It is notable that although these 34 ras proteins have only one stretch with more than two globally conserved consecutive amino acids (DTAGQE), vector 389a is capable of describing large regions of all 34 proteins by recognizing the latent similarity of multiple equivalent tetrapeptides. For example, this single vector recognizes KSAL, KSCL, and KTCL (residues 18–21 of the alignment) as dominant tetrapeptides that occupy equivalent positions within four of the six subtypes of ras proteins (Figure 1 ). Vector 389a also provides a reasonably strong summary of the large number of other ras proteins present within the genomes of these organisms, but not included in Figure 1 (not shown). In general, the most dominant singular vectors appear to identify highly conserved peptides present in a high fraction of individual members of a protein family or superfamily, while the less dominant vectors appear to describe conserved tetrapeptides present within a restricted set of proteins comprising a subfamily. Instead of simply providing restricted motif summaries using the most dominant elements of the left singular vectors, we have also attempted to examine entire vectors in order to gain a better understanding of the motifs (and associated protein families) they describe. A reasonably efficient method for depicting left singular vectors is presented in Figure 2 , using vectors 389 and 277 as examples. Both vectors are shown as frequency distributions (purple bars) that summarize the approximate magnitudes of the projections provided by all 160,000 tetrapeptides on the vector in question. These distributions are compared to a normal distribution having the same standard deviation (blue bars). In both examples, a significant fraction of tetrapeptides have high or low values in considerable excess of that expected from a normal distribution. Many of these also exceed the arbitrary cut-off value of 0.025 (dashed lines) used to extract the dominant tetrapeptides that serve to summarize the corresponding motifs. Parts of the Rab5 and HaRas motifs extracted from vector 277 are shown in Figure 2A as overlapping dominant tetrapeptides with associated projection values. Similar motifs extracted from vector 389 are shown in Figure 2B . In the latter case, a motif from the large subunit ribosomal protein rpL29 represents the "anti-motif"of the Ras/Rab proteins described by the extreme vector elements of opposite sign. Species vectors for the production of species phylogenies The detailed comparative information contained within the hundreds of singular vectors and their corresponding motifs and gene families was subsequently used to build a species phylogeny by summing all the SVD-derived right protein vectors separately for each organism and then comparing the relative orientation of the resulting species vectors [ 13 ]. Figure 3A shows the SVD-based topology obtained for the nine eukaryotes compared in this study. This tree supports a coelomate rather than ecdysozoan lineage. Two distinct re-sampling methods were used to estimate branch statistics for this tree. The top value of each pair of support values for each branch shown in Figure 3A was generated using a traditional bootstrap procedure [ 19 ]. In this case, 100 random sets of 437 re-sampled singular vectors were made and used to construct 100 species trees. Alternatively, a novel "successive, delete one" jackknife procedure [ 14 ] was used to generate the bottom value shown for each branch. In this case, the least dominant singular vector was removed successively (down to 10 vectors) to generate 427 ordered sets of singular vectors, and a new tree was estimated following each removal. Although bootstrap support values for the branches grouping arthropods with vertebrates (37%) and worms with other metazoa (49%) are relatively weak, support values for these branches are strong (100%) using the modified jackknife procedure. All other branches are strongly supported by both procedures. The branch separating Cele from the coelomates is of special interest, since the weak bootstrap support observed (37%) might suggest a significant affinity between Cele and the arthropods consistent with the "ecdysozoan" model (Figure 3A – alternative branching pattern shown in red). Bootstrap support for the alternative ecdysozoan cluster, however, was only 24%. Use of the "successive, delete-one" jackknife procedure as a species tree branch statistic is justified by the fact that SVD provides singular triplets in order of their "dominance" in explaining the data set [ 20 ]. Mathematical dominance provides an objective measure of importance that can be utilized to weight characters. Since the modified jackknife procedure used here deletes the least dominant singular vectors one at a time in order, the more dominant singular vectors (i.e. conserved motifs/families) are automatically weighted more heavily within the consensus tree. Hence, one can argue that our novel jackknife procedure provides stronger support for the derived phylogeny because the most dominant singular vectors generally contain stronger information about gene and species relationships. Poorly described proteins and species tree quality While our SVD-based analysis technically considers all proteins present within all nine genomes of the data set, it is likely that accurate vector definitions are provided for only a small fraction of these proteins. Theoretically, the 437 singular triplets could effectively describe as many as 2 × 437 = 874 protein families. However, many of these vectors appear to best describe particular subfamilies of larger groups of closely related proteins. Thus, the 58 protein families listed in Table 2 are each represented by anywhere from 1 to 8 triplets. Although, as mentioned earlier, some protein families lacking clear functional annotation were omitted from this table, it still serves to provide a conservative lower estimate of the number of well-described protein families provided by the SVD. Assuming the number of identifiable protein families in our nine genome data set significantly exceeds the 58 to 179 protein families unambiguously demarcated and subdivided in our analysis, then hundreds or perhaps thousand of the poorly described proteins included in our species vector sums might be contributing a high fraction of "noise" to the definition of species. In an attempt to increase the fraction of well described proteins used to define species, proteins having poor projections on all 437 right singular vectors were ignored during the summation process. Arbitrary vector magnitude cut-off values of 0.005 or 0.05 were applied to reduce the number of poorly described proteins used to build species trees. Even though the highest and most stringent cut-off value removed the majority of proteins during summation, both new species trees had identical topologies to that of the tree shown in Figure 3A in which all proteins were included. Bootstrap and modified jackknife branch support values for these tree are shown in Figure 3B along with those derived from the inclusive analysis. The removal of only a small fraction of poorly described proteins (cut-off = 0.005, about 10 3 proteins removed) resulted in 22% bootstrap and 100% modified jackknife support for the coelomate lineage, but 0% support for the ecdysozoan lineage. Removal of a much higher fraction of poorly described proteins (cut-off = .05, about 10 5 proteins removed) produced an equivalent result. Hence, poorly described proteins contribute little to the support that our analysis provides for the coelomate model. Conclusions As demonstrated above, an SVD-based analysis of multiple genomes automatically interprets proteins from input genomes as potential members of a limited list of hierarchically defined protein families and subfamilies. Each subfamily is defined in detail by one or more singular vectors as linear combinations of a large number of peptides (160,000 tetrapeptides, in this case). Potentially, a large number of proteomes lacking annotation can be directly interpreted using this method, assuming a sufficient number of annotated proteomes are included in the analysis. Although most of the genomes used in the present analysis were already accompanied by detailed protein annotations, formal annotations of the Frub and Agam proteins were not readily available. Nevertheless, our SVD-based analysis was able to provide precise protein motif descriptions and subfamily affiliations, not only for the six Frub or Agam proteins shown in Figure 1 , but also for any of the hundreds of other Frub or Agam proteins exhibiting strong vector projections on any of the 437 derived singular vectors (see "SVD top five" of Table 1 ). Our method bears partial resemblance to a recently described graph-theoretic method for rapidly clustering massive datasets of whole genome protein sequence [ 22 ]. In this case, the protein definitions generated were not used to derive gene or species trees, but to provide for a comprehensive clustering of all proteins into families having one or more members. The nodes of their graphs, like the vectors from the right matrix in our analysis, represent proteins, while the edges between nodes in their graphs, like the angles between vectors in our analysis, contain the distance information used to compare proteins. However, the distance information in their analysis was obtained ultimately from exhaustive pairwise BLAST alignments. In contrast, our distance information was derived without alignment, by reference to the 437 most dominant SVD-derived orthonormal left singular vectors. These vectors provide "motif models" expressed as particular linear combinations of the 160,000 possible tetrapeptides. The projections of these motif models on a given protein vector serve to quantitatively define the protein. Since no more than 874 motif models would be provided by our truncated SVD, our method would be less effective than other methods for providing comprehensive family designations for all proteins in a dataset [ 22 , 23 ]. However, a high fraction of these protein families are found to contain only one or a few members [ 22 ]. Singletons and small families would generally provide unimportant contributions to relative species definitions, since the majority of species would lack a homolog for comparison. Hence small or poorly conserved protein families, presumably represented by the weaker singular triplets in a complete SVD, are profitably ignored in our analysis. Although our descriptive analysis of singular triplets (e.g. Table 2 , Figure 1 ) suggests that the protein vectors in our high dimensional definition space can be effectively clustered, we have not applied any specific clustering algorithm. Hence no explicit clustering of proteins, equivalent to the identification of orthologs or homologs, is required. Nevertheless, the application of a clustering algorithm to our vector based symmetric protein distances is clearly feasible and results in accurate clustering for a high fraction of proteins. In fact, the accuracy with which proteins are clustered into known families via Neighbor Joining was used previously to establish optimal dimensionality for a well characterized data set [ 13 ]. In addition, unlike other methods, our method provides a straightforward vector addition mechanism for converting relative protein definitions into relative species definitions for the production of species phylogenies. Alternative non-alignment methods exist for comparing sequences [reviewed in [ 17 ]]. Some of these methods may prove to be scalable and adaptable to the problem of whole genome phylogeny. For example, a comprehensive bacterial phylogeny was recently derived using species vectors that include a set of background corrected pentapeptide or hexapeptide (K-tuple) frequency values [ 12 ]. Although apparently effective for producing global species phylogenies, this method fails to provide quantitatively comparable protein definitions or interpretable predictions for conserved motifs. While many phylogenetically informative pentapeptides and hexapeptides are likely derived from homologs or orthologs, no mechanism exists for extracting, summarizing, and interpreting this information in terms of motif and gene family definitions. This high stringency method provides a low false positive rate (strong connections between probable orthologous peptides), but comes at the expense of a high false negative rate (little or no recognition of other homologous regions within proteins). For organisms exhibiting a significant level of horizontal gene transfer [ 24 - 26 ], models for motifs and protein families may be crucial tools for identifying "borrowed" genes and assessing their impact on phylogenetic hypotheses. Our SVD-based species tree supports the traditional "coelomate" model of animal phylogeny. Other large-scale, genome level analyses also tend to support this model [ 27 , 28 ]. The alternative "ecdysozoan" model is supported by comparative analyses of rRNA and analyses that include morphological characters [ 28 , 29 ]. Although genome-scale analyses should perhaps carry considerable weight due to the higher fraction of "total information" used as input, the separation of "signal" from "noise" represents a serious hurdle for these methods. Our method represents a uniquely independent solution that provides a noise-reduced simultaneous global comparison of all proteins within multiple genomes without the need for alignments and without the prior application of operational definitions of orthology. As such, it provides a global perspective on gene and species relationships that is based on a much larger subset of information than that normally used. Since it is a non-alignment method, it provides a fundamentally different kind of analysis, and to the extent that the resulting species phylogenies agree with those provided by other analyses that depend upon highly filtered subsets of aligned orthologs or close homologs, we may derive an additional degree of confidence in these relationships. However, the balanced comparison of a large number of additional whole genome sequences from a variety of animals will likely be required in order to produce an unambiguous and universally accepted animal phylogeny. Methods Datasets Complete reference protein sequences for nine whole eukaryotic genomes ranging from yeast to man were compiled into a single dataset (Figure 1a ). Curated protein sequence files were obtained from NCBI dated as follows: human (Hsap) 10/10/03, mouse (Mmus) 10/31/03, rat (Rnov) 9/23/03, mosquito (Agam) 10/24/03, fly (Dmel) 10/24/03, worm (Cele) 11/12/03, malaria (Pfal) 10/17/02, and yeast (Scer) 11/12/03. Pufferfish (Frub) protein sequences dated 8/26/02 were obtained from the DOE Joint Genome Institute. Peptide frequencies and SVD Each protein sequence in the dataset was recoded as a high dimensional vector containing raw frequencies for each of the 160,000 possible tetrapeptides. Previous work has established that although tripeptides work well for estimating similarities between highly divergent proteins contained within small sets of viral genomes [ 15 ], tetrapeptides work better for larger data sets derived from vertebrate mitochondrial genomes or whole bacterial genomes [ 13 , 14 ]. Although pentapeptides also worked well with the mitochondrial datasets (unpublished), our computational capacity precluded the use of pentapeptides (3.2 million patterns) and larger data sets, like the one used here. Following a log-entropy transformation [ 21 ], the singular value decomposition of the resulting data matrix was computed. The log-entropy transformation tends to down-weight evenly distributed high frequency peptides that are likely sources of homoplasy. After 1500 Lanczos iterations (residual errors less than 10 -6 ), three output matrices were obtained, consisting of 437 singular triplets (left and right singular vectors and their corresponding singular value). Each left singular vector produced by the SVD defines one or two conserved motifs within the dataset as particular linear combinations of tetrapeptides [ 13 , 14 ]. Similarly, each of the right singular vectors defines one or two conserved gene families (or subfamilies) as particular linear combinations of proteins. Each gene family identified by a given right singular vector contains motifs described by the corresponding left singular vector. Two distinct motif/families are frequently identified per triplet, since each triplet describes both a correlated motif/family (positive values) and an anti-correlated motif/family (negative values). Vector based motif and protein family models "Dominant" vector elements (absolute values in excess of 0.025) were extracted from the left singular vectors and summarized using the C++ program "Copepx" [ 14 ]. These values were associated with the most "dominant" (i.e. highly conserved) tetrapeptides found within the motifs described by a given left vector. In addition, the "top five" positive and "top five" negative elements were extracted from the right singular vectors and summarized using the C++ program "Coprotx". These values represent the most dominant members of the gene families described by a given right vector. Species trees and branch support Distance matrices were derived by summing all the SVD-derived right protein vectors for a given organism and then comparing the relative orientation of the resulting species vectors using the program Cosdist [ 13 , 14 ]. Species trees were subsequently derived from distance matrices using Phylip-Neighbor [ 30 ]. Two distinct resampling methods were used to provide branch support: a traditional bootstrap procedure [ 19 ], and a modified jackknife procedure. For the bootstrap, 100 random sets of 437 resampled singular vectors were made and used to construct 100 species trees. For the "successive, delete one" jackknife procedure [ 14 ], the least dominant singular vector was removed successively (down to 10 vectors) to generate 427 ordered sets of singular vectors, and a new tree was estimated following each removal. List of abbreviations used Homo Sapiens (Hsap), Mus musculus (Mmus), Rattus Norvegicus (Rnov), Anopheles gambiae (Agam), Drosophila melanogaster (Dmel), Caenorhabditis elegans (Cele), Plasmodium falciparum (Pfal), and Saccharomyces cerevisiae (Scer), Fugu rubripes (Frub), correlated peptide (copep), correlated protein (coprot), right singular vector (rsv), left singular vector (lsv). Authors' contributions GS conceived the study, gathered the input data, provided primary interpretation of the output, and drafted the manuscript. MB wrote and adapted software, performed computational analysis on the input data, and provided manuscript modifications. All authors read and approved the final manuscript. Supplementary Material Additional File 1 Copep Motifs. Long copep strings identified within the left singular vectors of a given s-triplet. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544558.xml |
549068 | The Mammalian Phenotype Ontology as a tool for annotating, analyzing and comparing phenotypic information | The Mammalian Phenotype (MP) Ontology enables robust annotation of mammalian phenotypes in the context of mutations, quantitative trait loci and strains that are used as models of human biology and disease. | Background Mammalian phenotypes are complex and the term itself is imprecise. Generally, we use the word phenotype in referring to the appearance or manifestation of a set of traits in an individual that result from the combined action and interaction of genotype and environment. Because mouse is the premier model organism for the study of human biology and disease, the goal of comparative phenotyping and building new animal models through genetic engineering holds great promise. The mouse has distinct advantages for studies that translate to humans. It is a small, short-lived mammal with a fully sequenced genome in which all life stages can be accessed, and for which myriad tools are available for precisely experimentally manipulating its genome. Further, the large collection of inbred strains of mice and the controlled environment in which the animals live provides the ability to confirm phenotype observations and to systematically perturb environmental factors and genetic input to measure effects under defined conditions. Current international efforts to 'make a mutation' for every gene through mutagenesis [ 1 ] and genetic engineering [ 2 , 3 ] make it imperative for phenotype data to be represented in standard descriptive formats to enable computational analysis and comparison. Mammalian phenotypes are frequently genetically complex. Mutation of even a single gene almost always produces pleiotropic effects. Conversely, non-allelic mutations can produce indistinguishable phenotypes. Modifier genes and epistatic interactions can markedly alter the phenotype. Combining different allelic combinations of different genes can produce unique phenotypes not found in the single-gene mutation genotype. Imprinting of genes can dramatically affect phenotype. Mutations expressed in different inbred strains of mice can manifest as an increase or decrease of severity or penetrance of the corresponding phenotype. Quantitative trait loci (QTL) can contribute in complex nonlinear ways to the phenotype. In addition, mutations that are 'genomic' in nature, either disrupting or deleting multiple genes or occurring in intergenic regions, can produce distinct phenotypes and challenge us to think beyond gene effects to genomic effects. The outcome of these complex interactions can be dissected and reproducibly examined by characterizing inbred strains that represent the combined phenotype of the 'whole-genome' genotype in its environmental context. The Mouse Genome Database (MGD) at the Mouse Genome Informatics website [ 4 , 5 ] serves as the model organism database for mouse, representing the genetics, genomics and biology of the mouse and as a community resource for mammalian studies. Significant reorganization and modeling of phenotypes is now underway to support these data robustly, to represent phenotypes in ways that are computationally accessible, and to provide human interfaces to these data that will enable knowledge building and hypothesis generation. One component of this work is the development of the Mammalian Phenotype (MP) Ontology, a structured vocabulary that will aid in standardizing annotations and, with its concepts definitions, unambiguously describe phenotypic observations. Results and discussion The problems of text Written descriptions of phenotypes in higher organisms reflect the complexity of the subject, the richness of language, and the phenomenal diversity that these data represent. While text descriptions are commonly used in publications describing phenotype, and have been the basis of electronically accessible phenotypic descriptions (for example, Online Mendelian Inheritance in Man (OMIM) [ 6 ] and the Mouse Locus Catalog (MLC) [ 7 ], text is unreliable for searching, either manually or computationally. From the user's perspective, even the best full-text search including Boolean operators will miss appropriate records (false negatives) and return unwanted records (false positives). Consider the example in Table 1 where searches were done to find spontaneous mutations in which mice were entirely or partially devoid of hair/fur. To obtain a complete result, the user would need to use a number of search terms and synonyms. The wording within the text depends upon the author of the record and his/her particular word usage and editorial style. A minimum of four search terms is needed to recover the 27 relevant mutations displayed in this table and it cannot be ascertained if this is a complete set of mutations for this phenotype. Conversely, the user is returned with 23 irrelevant results. Irrelevant results can be returned for many reasons including, but not limited to, the following: the author of the record is contrasting the phenotype of a mutation in one gene with a mutation in another gene; the author is making a statement that includes the negation of the trait; the match is based on gene name rather than phenotype; the mutant was used as a linkage marker to genetically map another gene. A further detriment to database text records is their difficulty to update and maintain. As new information is learned about a phenotypic mutant, the record must be continually rewritten. Although this practice might be sustained for a small number of records, it does not scale when thousands of mutant records are considered. The alternative of simply adding on another paragraph to existing text records becomes confusing, with potentially conflicting information and different writing styles appearing in one textual description, and unwieldy, with more and more text that may no longer represent a logical synthesis. Nomenclatures, vocabularies and ontologies Formal nomenclatures for genes, mutant alleles and inbred strains of mice have existed since the 1940s [ 8 , 9 ]. The MGD [ 4 ] serves as the authoritative source for the names and symbols associated with mouse genes, alleles and strains. The advantage of applying such nomenclatures has been increasingly recognized as genomes become better defined and the realized power of comparative genomics allows homologous and orthologous gene relationships to be explicitly defined. At present, human, mouse and rat gene nomenclatures operate in parallel, using coordinated symbols for all three species' genes. In addition, mouse and rat strain nomenclatures were merged to one standard strain nomenclature recently, making strain identity and nomenclature conventions consistent. Nomenclature guides for mouse and rat genes, mutant alleles, and strains are available online and regularly revised based on international nomenclature committees' reviews [ 10 ]. Beyond nomenclatures, which are key to object identities and relationships, are vocabularies that can be used to describe broader concepts and categorizations. Vocabularies can take many forms, including simple lists of controlled terms, such as the cytogenetic band designations used to name the bands defined by chromosome staining or the classes of genetic markers, such as gene, pseudogene, expressed sequence tag (EST), and so forth. The annotation of complex biological data and concepts requires more than lists and simple vocabularies. Ontologies, or 'descriptions of what there is', contain both concepts, with precise meanings, and relationships among those concepts. As such ontologies are able to support descriptions of complex biology and are useful in making these data more amenable to computational analyses. The first widely used ontology developed and adopted in the biological domain is the Gene Ontology (GO) [ 11 - 13 ] which contains concepts of molecular function, cellular localization and biological process for annotating the functional aspects of genes. The GO is structured as a directed acyclic graph (DAG), where each vocabulary term (node) may have both multiple parent term and multiple child term relationships. MGD uses GO extensively for gene annotation [ 14 ]. In addition, MGD has adopted the Mouse Embryo Anatomy Nomenclature Database [ 15 ] and the Anatomical Dictionary for the Adult Mouse [ 16 ] for annotating data that include anatomical attributes, such as tissue sources for clones and phenotypes. The Gene Expression Database (GXD) [ 17 ], integrated with MGD through the Mouse Genome Informatics (MGI) system [ 4 ], applies these anatomical ontologies as a central concept in the description of expression data. Mammalian Phenotype Ontology Although the need for vocabularies as key components to consistent phenotype annotations for mammals has been recognized for some time [ 18 ], and many smaller controlled vocabularies have been implemented to describe various aspects of phenotype in MGD (for example, class of mutation, embryonic stem (ES) cell lines used for generating targeted mutations, type of inheritance), much of the data has remained in text form. Over the past two years, the Mammalian Phenotype (MP) Ontology has emerged to more precisely describe phenotypes, and to allow easier access to phenotype-sequence interactions. Our goal is to describe the richness of phenotypes as precisely as they are known, recognizing that phenotype data are by nature complex and usually incomplete. Taking advantage of structural properties of a DAG, we have the ability to annotate phenotypes to the level of data resolution available, whether general or very specific and the ability to query with a high-level term, returning all phenotypes containing annotations to that term or to terms more specific than the query term. Thus, one can query for 'respiratory signs/symptoms' and retrieve all phenotypes annotated to this term and its hierarchical 'children' (abnormal breathing, abnormal respiratory sounds, anoxia, apnea, dyspnea, hypercapnia, and so on), or specifically request annotations to any of these sub-terms. The top level terms of the MP Ontology include physiological systems, behavior, developmental phenotypes and survival/aging. Physiological systems branch into morphological and physiological phenotypes at the level immediately below. A browser to view the ontology is available at [ 19 ] (Figure 1 ). In this browser the DAG structure is flattened into a hierarchy, with multiple hierarchies representing unique paths to a term displayed sequentially. MP terms and synonyms can be searched or users can browse the ontology starting from the high-level terms and open levels continuously to increasingly granular terms. Each MP ontology term has a unique identifier, a definition and synonyms. In the term detail pages, these data and the number of hierarchical paths of the vocabulary where the terms appear are displayed. A plus sign following the term indicates that children of this term exist. In this figure, displayed next to the term, is a link indicating the number of annotation instances in MGD using this term or children of this term. This feature, due to be publicly available in early 2005, will greatly improve phenotype-centric searching in MGD. Developing the MP vocabulary To initiate the vocabulary, we first developed a high-level categorization of phenotypes consisting of approximately 100 terms, such as heart/cardiovascular dysmorphology and skeletal axial defects. As we used this list for annotations, terms were refined and general organizing principles for the MP vocabulary were developed. An important component of our approach has been to address two practical implementation questions. From the biologist's perspective, the question is what term would be used to describe a specific phenotypic trait. From the curation perspective, we ask what terms reflect biological reality and maximize curator productivity. From a purely ontological perspective, every trait could be broken down into a core object, such as 'cornea' or 'gastrulation', defined by anatomical, behavioral or physiological terms, and a series of attribute vocabularies that describe the quality, quantity and character of a trait. For the practical reason of needing robust terms to describe phenotypes up-front to speed curation and the problem of losing biological meaning, particularly for clinical or dysmorphology terms, when terms are completely deconstructed (that is, the sum of the parts is less than the term itself), we have chosen to use compound terms in the MP Ontology. A few examples of terms where it is difficult to preserve the full biological meaning once they are deconstructed are shown in Table 2 . In addition, it should be noted that each of these terms requires multiple annotations to recover all aspects that the single term provides. Use of complex terms in the MP Ontology, however, does not preclude also storing the decomposed version should this later prove desirable (see PATO model discussed in [ 20 ]). More important, the MP Ontology can currently hold, for each term, database cross-references to other ontologies. This is a common practice in GO when compound terms are developed. For the MP Ontology, these cross-references include anatomical terms from the Mouse Anatomy ontologies [ 15 , 16 ] and the GO process terms [ 21 ]. Three major strategies are being pursued to further develop the vocabulary itself. First and most important is through the ongoing process of curating phenotype data. As new phenotypic traits are described and published, the need for new terms is recognized. New terms added in this way may be a simple addition to an existing hierarchical path or may result in the addition of entire new branches in the hierarchy. Second, collaborative efforts between the MGD phenotype curators, the mouse mutagenesis centers and the rat genetics community identify new specific terms and suggest improved organization of terms within particular hierarchical branches. Third, we are recruiting individuals with expertise in specific biological domains to review and evaluate sections of the vocabulary for accuracy, completeness and systematic arrangement. The MP Ontology is a work in progress and remains incomplete in some areas. We welcome the participation of the mammalian research community so that the most useful, definitive and universally applicable terms will be included. Information can be obtained by sending e-mail to pheno@informatics.jax.org . While common pathological and clinical terms are used in the MP Ontology, considerations for term placement within the structure and for precise terminology is often derived from comparison with other open biological ontologies (OBO) [ 22 ]. Recently, a cell-type ontology has become available [ 23 ] and a comparison of terminology to this ontology has not yet been completed. We are working with the mutant mouse pathology database Pathbase [ 24 , 25 ] to map and cross-reference terms from their Pathology Ontology. Vocabulary tools The MP Ontology was built as a DAG using the DAG-Edit software written by John Richter and Suzanna Lewis [ 26 ]. The MP Ontology is updated daily and can be browsed or searched online at [ 19 ]. MP files also are available in flat file format and OBO format from our ftp site [ 27 ] and are posted at the OBO site [ 28 ]. Phenotype data annotation Phenotypes are described in the MGD relative to the genotype of the individual. Genotype objects specifically consist of one or more allele pairs describing mutations or QTLs and the genetic background strain(s) where the phenotype was observed. Each phenotype annotation associates a MP Ontology term with a genotype/strain and the reference or data source supporting this assertion. Additional modifying text may be annotated to describe detail that is not easily standardized. Examples include experimental conditions, age of onset and incidence, and trait penetrance, among others. The annotation note may also include specifics of the phenotype where such details are deemed to be too case specific to be a MP term. In addition, genotypes are associated with OMIM where a particular mouse genotype is a model for human diseases and syndromes. Figure 2 shows the portion of one phenotype record that uses the MP Ontology. Conclusions The MP Ontology and annotation schema was designed to minimize curatorial time, yet remain precise enough to describe phenotypic data. It supports robust phenotypic annotations and querying capabilities for mouse phenotype data. While this vocabulary is far from complete, we have designed strategies for its continued development as a collaborative effort for supporting the representation of existing mutations and those that continue to be created. As of 1 November 2004, over 11,150 phenotypic alleles representing mutations in 5,214 unique genes had been catalogued in MGD. For these alleles, 9,696 genotype records exist, with 21,556 phenotypic annotation instances. The MP Ontology is also used in phenotypic data annotations at the RGD [ 29 ]. As our database groups continue to accumulate annotations, it will be possible to mine these data to ask interesting questions about similarities and differences in comparable allele effects between the species, as well as within species. Comparative phenotype data will potentially uncover new modifier effects and point to new pathway relationships and genetic networks tied to disease processes. The MP Ontology will be critical for enabling computational analyses and providing a framework for improved web views and other human-comprehensible displays for the research community. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549068.xml |
549040 | T lymphocytes among HIV-infected and -uninfected infants: CD4/CD8 ratio as a potential tool in diagnosis of infection in infants under the age of 2 years | Background Serologic tests for HIV infection in infants less than 18 months do not differentiate exposure and infection since maternally acquired IgG antibodies may be detected in infants. Thus, the gold standard for diagnosis of HIV-1 infection in infants under the age of 2 years is DNA or reverse transcriptase polymerase chain reaction. There is an urgent need to evaluate alternative and cost effective laboratory methods for early diagnosis of infant HIV-1 infection as well as identifying infected infants who may benefit from cotrimoxazole prophylaxis and/or initiation of highly active antiretroviral therapy. Methods Whole blood was collected in EDTA from 137 infants aged 0 to 18 months. DNA polymerase chain reaction was used as the reference standard for diagnosis of HIV-1 infection. T-cell subset profiles were determined by flow cytometry. Results Seventy-six infants were DNA PCR positive while 61 were negative. The median CD4 counts of PCR negative infants were significantly higher than those of the PCR positive infants, p < 0.001 . The median CD4/CD8 ratio and the %CD4 of the PCR positive infants were both significantly lower than those of the negative infants, p < 0.001 . The CD4/CD8 ratio had a >98% sensitivity for diagnosis of HIV-1 infection and a specificity of >98%. Conclusion The CD4/CD8 ratio appears useful in identifying HIV-infected infants. The development of lower cost and more robust flow cytometric methods that provide both CD4/CD8 ratio and %CD4 may be cost-effective for HIV-1 diagnosis and identification of infants for cotrimoxazole prophylaxis and/or highly active antiretroviral therapy. | Background Endemic HIV infection, in sub-Saharan Africa, where in many countries more than 20% of pregnant women are HIV seropositive leads to a diagnostic problem in the evaluation of their infants. Without intervention, more than 25% of infants born to seropositive women will acquire HIV infection in the first year of life. HIV testing with enzyme immunoassay-based rapid tests have expanded capacity to identify seropositive women and provide interventions, but even with single dose Nevirapine and other antiretrovirals, infection of infants still exceeds 10% in the first year of life. Serologic tests for HIV do not accurately identify those infants who have acquired infection within the first 18 months of life because of transplacentally acquired maternal IgG antibodies. As antimicrobial and antiviral interventions are developed to reduce morbidity and mortality, among infants born to seropositive women, the early diagnosis of infection is increasingly important. The gold standard for diagnosis of HIV-1 infection in infants under the age of 2 years is DNA polymerase chain reaction (PCR) or reverse transcriptase (RT)-PCR [ 1 - 3 ]. With the increasing availability of single dose nevirapine for prevention of mother-to-child transmission (MTCT) of HIV [ 4 ] and generic antiretroviral drugs for treatment of AIDS in resource-poor countries, there is an urgent need to develop alternative laboratory methods for early diagnosis of infant HIV-1 infection as well as identifying infants who meet the criteria for commencing cotrimoxazole prophylaxis and/or initiation of ARV therapy. The World Health Organization (WHO) recommends initiation of highly active antiretroviral therapy (HAART) in HIV-seropositive infants under the age of 18 months, who have WHO Pediatric Stage III disease and CD4 percentage (%CD4) <20% in resource-poor countries where %CD4 are available but virologic tests (DNA PCR, RT-PCR or immune-complex dissociated p24 antigen) for confirmation of HIV infection are not available [ 5 ]. Thus, in infants under the age of 18 months, a single laboratory test that can identify both HIV-infection and provide %CD4, two important parameters for decision-making in initiating HAART, may be useful. Infants who are positively identified as HIV-infected and meet clinical criteria are likely to benefit from HAART. Immunological changes in HIV-1 infection include a decrease in CD4 + cells, a transient increase in CD8 + cells, total lymphocytes and inversion of the CD4/CD8 ratio [ 6 , 7 ]. As HIV infection progresses, the CD4 + cells decline, while the CD8 + cells which may remain at high levels for long periods, eventually decrease but not to baseline levels. Since in healthy children the CD4 + and CD8 + cells account for 60% and 30% of the T lymphocytes respectively, a normal CD4/CD8 ratio should always be >1.0. Thus, in HIV-1 infection where there is a decrease in CD4 + cells and an increase in CD8 + cells, the reversal of the CD4/CD8 to <1.0 should in theory be useful for diagnosis of HIV-1 infection. Certain flow cytometers when used in tandem with a haematological analyzer can provide absolute CD4 + cell counts, their percentages as well as the CD4/CD8 ratio. The use of the ratio in combination with %CD4, may lead to a timely identification of infected infants, who meet the WHO criteria for initiation of CTX prophylaxis and/or ARV therapy. While the infrastructure needed to conduct flow-cytometric analyses of HIV infection are still largely confined to a few centers, an increasing number of point of care diagnostic testing systems, inexpensive methods to measure CD4 cells are currently in development. These include the development and evaluation of simplified volumetric flow cytometric methods using a low cost flow cytometer that can be powered from a car battery or by solar panels (Cyflow SL, Partec, Munster, Germany) by Cassens and colleagues [ 8 , 9 ] and modification of a commercially available 4-parameter flow cytometer, Luminex 100 (Luminex, Austin Texas, USA) to a compact portable prototype instrument that can operate with a 12-volt rechargeable battery [ 10 ]. Furthermore, use of generic CD4, CD8 and CD45 fluorescence-conjugated monoclonal antibodies can reduce the cost of determining T cell subset profile even when employing standard flow cytometers [ 10 , 11 ]. The main objective of the current study was to evaluate the CD4/CD8 ratio for diagnosis of subtype C HIV-1 infection in infants under the age of 2 years among infants where DNA PCR was performed to diagnose HIV infection. Materials and Methods Study cohort The infant specimens used in this study were obtained from two independent prospective studies; the short course zidovudine (AZT) and Pediatric AIDS Clinical definition (PACD). The Medical Research Council and the Institutional Ethics Committees approved both studies and present study. The infants included in the study were under the age of 2 years. All the infants were breastfeeding at the time of specimen collection. Infants who were followed in the short course AZT study aimed at preventing MTCT were enrolled between May 2001 and June 2002. In the AZT study, mothers received short course AZT starting at 36 weeks gestation and throughout labor. Their infants received AZT for 7 days. In the PACD study, hospitalized children aged between 2 months and 18 months were prospectively enrolled into the study, between July 2002 and July 2003, following informed consent from their mothers. The mothers of these children did not receive antiretroviral therapy for prevention of MTCT nor for HIV disease. The PACD study is a hospital based analytical cross-sectional survey. Laboratory specimens were obtained once from each subject. Children presenting in the moribund state, requiring immediate resuscitation or those with known HIV status or those whose mothers/guardians refused to sign the informed consent were excluded from the study. The objective of the PACD study is to identify clinical symptoms, associated with HIV-1 infection in infants under the age of 18 months, which may be used in the absence of laboratory tests, for HIV-1 diagnosis. Blood collection and processing A total volume of 2 ml whole blood was collected in ethylenediamine tetraacetic acid from each of the 137 infants in the short course AZT and the PACD studies. The whole blood was then aliquoted into two tubes (500 microlitres in each) for determination of T cell subset profiles, and the second for DNA PCR and the rest centrifuged at 200 g to obtain plasma, which was stored at -80°C. The laboratory tests described below were conducted in a blind fashion. Flow cytometry analysis T cell subset profiles were determined by flow cytometry using a Coulter Epics XL equipped with System II software (Beckman Coulter, Miami, Florida, USA) within 4 hours of blood collection. This flow cytometer was run, in a double platform setting where the absolute counts for both white blood cells and lymphocytes were obtained on a Celldyn 3500R haematological analyzer (Abbott, GmbH, Germany). Then from the combined results, the absolute CD4 + and CD8 + cell counts, CD4/CD8 ratios, as well as the %CD4 and %CD8 values among lymphocytes were automatically calculated. DNA PCR analysis DNA PCR Roche amplification assay version 1.5 (Roche Diagnostics, Branchburg, NJ, USA) was employed as the reference standard for diagnosis of infant HIV-1 infection status. DNA extraction, amplification and detection were performed and results interpreted following the manufacturer's instructions (Roche Diagnostics, Branchburg, NJ, USA) as we previously described [ 12 , 13 ]. Infant grouping Reference ranges of T cell subset profiles for infants and children are usually stratified by age as <12 months, 1 to 5 years and 6–12 years. In the current study, in addition to overall evaluation of samples obtained from all the infants aged 0 to 18 months, the evaluated parameters were also compared based on infant age groups 0–11, and 12–18 months. Statistical Analysis Diagnostic tests (sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and test efficiency (TE) with 95% confidence interval) were used to assess the assay under evaluation with DNA PCR as the reference standard. A MTCT prevalence rate of 30.7% (based on the current rate in Zimbabwe) was used in the calculations of PPV and NPV. Sensitivity was defined as the percentage of reference standard HIV positive samples found reactive by assay under evaluation. Specificity was defined as the percentage of reference standard HIV-negative samples that were negative by the assay under evaluation. TE refers to the overall ability of a test to correctly identify all positives and negatives (the absence of false positives and false negatives). It is a combination of the sensitivity and the specificity of an assay and gives an idea of the total effectiveness of the assay. PPV was defined as the probability that a specimen had CD4:CD8 ratio less than 1 when the test was DNA PCR positive. NPV was defined as the probability that a specimen does not have a CD4:CD8 ratio ≥ 1 when the test was DNA PCR negative. Comparisons of T cells between infected and uninfected infants were done using non-parametic equivalent of the T-test (Kruskal-Wallis test). P values less than 0.05 were considered statistically significant. Results Of the 137 infant specimens tested using DNA PCR, 76 were HIV-1 positive and 61 were HIV-1 negative. The 76 PCR positive infants included 9 infants who had evidence of in utero transmission as determined by serial DNA PCR of birth and subsequent samples tested in longitudinal short course AZT MTCT studies (Zijenah et al, unpublished data). T cell subset of infected and uninfected infants T lymphocyte subset profiles were performed for the 137 infants who had whole blood specimens for flow cytometry. The median age of these infants at specimen collection was 5.5 months (Interquartile Range [IQR]: 3–13) and 8.0 months (IQR: 4–14) for HIV infected and uninfected respectively ( p = 0.08 ). As expected, the median CD4 + cell counts of PCR negative infants were significantly higher than those of the PCR positive infants, p < 0.001 (Table 1 ). Inversely, the median CD8 + cell counts were significantly higher among PCR positive infants than PCR negative infants, p < 0.001 (Table 1 ). The median CD4/CD8 ratio of the PCR positive infants (0.4, IQR: 0.3–0.6) was significantly lower than that of the PCR negative infants (1.8, IQR: 1.4–2.3), p < 0.001 . The median %CD4 of PCR positive infants was also significantly lower than that of PCR negative infants, p = < 0.01 (Table 1 ). Table 1 T cell subset profile of HIV-1 infected and uninfected infants PCR Positive (n = 76) PCR Negative (n = 61) P value a Median age (months) 5.5 (IQR: 3–13) 8 (IQR: 4–14) 0.08 Median CD4 + (cells/μL) 521.5 (IQR: 323–805) 1356 (IQR: 916–1769) <0.001 Median CD8 + (cells/μL) 1302.5 (IQR: 829–2054) 799 (IQR: 471–1020) <0.001 Median CD4/CD8 ratio 0.4 (IQR: 0.3–0.6) 1.8 (IQR: 1.4–2.3) <0.001 Median %CD4 13.9 (IQR: 9.2–19.1) 29.9 (IQR: 25.3–34.5) <0.001 Median %CD8 31.3 (IQR: 22.3–42.9) 18.4 (IQR: 14.2–21.5) <0.001 Abbreviations: PCR, polymerase chain reaction; n, number tested; P value a for statistical significance between group medians was estimated using the Kruskal-Wallis test. DNA PCR versus CD4/CD8 ratio Seventy-five infants of the 76 PCR positive group (98.7%) had CD4/CD8 ratio <1 while 60 infants of the 61 who were PCR negative had CD4/CD8 ratio ≥ 1. All the 9 infants infected in utero had a CD4:CD8 ratio less than 1. Their median CD4/CD8 ratio was similar to that of the rest of the PCR positive babies (0.4; IQR: 0.3–0.6), albeit the numbers are too small for statistical significance considerations. The infant (aged 11 months at specimen collection) who was DNA PCR positive with a CD4/CD8 ratio of 1.1 was from the PACD study. The PCR negative infant (aged 4 months at specimen collection) with a CD4/CD8 ratio of 0.3 was also from the PACD cohort. The overall sensitivity and specificity of the CD4/CD8 ratio were 98.7% (95% CI: 96.1–100 and 98.3% (95% Confidence Interval (CI): 95.1–100) respectively with PPV and NPV of 96.3% and 99.4% respectively and a test efficiency of 98.5% (95% CI: 96.5–100) (Table 2 ). Comparison of all the evaluated parameters between the 0–11 and 12–18 months infant age groups showed that the 95% CI overlap between the groups which implies no statistically significant difference in these two age groups. Table 2 Evaluated parameters for CD4/CD8 ratio for the three infant age groups using DNA PCR as reference standard 0–18 months infant age group (n = 137) 0–11 months infant age group (n = 95) 12–18 months infant age group (n = 42) %Sensitivity 98.7 (CI: 96.1–100) 98.2 (CI: 94.7–100) 100 (CI: 100–100) %Specificity 98.4 (CI: 95.2–100) 97.5 (CI: 92.7–100) 100 (CI: 100–100) %PPV 96.4 94.6 100 %NPV 99.4 99.2 100 %TE 98.5 (CI: 96.5–100) 97.9 (CI: 95.0–100) 100 (CI: 100–100) Abbreviations: n, number tested; NPV, negative predictive value; PPV, positive predictive value; TE, test efficiency. Discussion In resource-poor countries, the major constraint in the use of DNA PCR for diagnosis of HIV-1 infection in infants under the age of 18 months is the cost of the equipment and the reagents. In addition, highly trained laboratory personnel and stringent quality assurance measures are needed to run this assay for routine diagnosis of HIV infection in infants. In Zimbabwe, enzyme linked immunosorbent assay (ELISA) is routinely performed in both public and private laboratories for diagnosis of various infections including HIV-1. However, because of the transplacental transfer of maternal IgG antibodies, which may persist in infants for up to 18 months, ELISA is not suitable for diagnosis of HIV-1 infection in these infants. Therefore alternative methods are needed for this purpose. In addition, both public and private laboratories have flow cytometers or FACSCount machines for enumeration of T cell subset profile and automated haematological analysers for routine full blood counts with differential. With this equipment available in the country we evaluated the CD4/CD8 ratio as an alternative diagnostic test for infant HIV-1 infection. In our investigation we used Epics XL Coulter flow cytometer equipped with System II software (Beckman Coulter, Miami Florida, USA) for the following reasons. With this instrument, CD4/CD8 ratios can be conveniently obtained. In addition, when used in double platform setting tandem with a haematological analyzer, the results also show absolute CD4 + cell counts and both %CD4 and %CD8 values among lymphocytes. The %CD4 value among lymphocytes is the recommended parameter for analyzing pediatric samples, as absolute counts for infants are age sensitive and variable. A simpler single platform system such as the FACScount (Becton-Dickinson Immunocytometry Systems, San Jose, CA, USA), is not fully suited for pediatric work as it provides CD4 + , CD8 + , CD3 + T lymphocyte counts and CD4/CD8 ratio but %CD4 are not available. Of note, %CD4 expressed among CD3 + T lymphocytes, is a different parameter from the customary %CD4 expressed among lymphocytes. The %CD4 expressed among lymphocytes and not the %CD4 expressed among CD3 + T lymphocytes is recommended for decision making to initiate ARV therapy in children under the age of 18 months [ 5 ]. In Zimbabwe, CD4 + cell count, is less expensive than PCR and may provide additional information of value to the clinician with respect to prognosis, and the need for prophylaxis and treatment. Optimal flow cytometry for determination of T cell subset profile, offers the added advantage that CD4/CD8 ratio will determine the infection status of the infant while the % CD4 will provide information on decision-making for commencement of HAART. Overall, the CD4/CD8 ratio had a ≥ 98% sensitivity for diagnosis of infant HIV-1 infection and a specificity ≥ 98%. Both sensitivity and specificity were 100% for infants in the 12–18 months age group. Interestingly, in parallel studies performed in 250 HIV-1 seropositive adults, 249 had a CD4/CD8 ratio of <1. The CD4/CD8 ratio of the one patient was 1 at enrollment and has remained so for over one year. When interpreting our data, it is important to note that normal T cell subset values among African children differ from those of other populations [ 14 - 16 ]. A study in Guinea Bissau [ 16 ], reported that Guinean children under the age of 2 years had lower %CD4 and CD4/CD8 ratios and higher %CD8 when compared to their counterparts from developed countries. Interestingly, girls had higher CD4/CD8 ratios and lower %CD8 than boys. In our study, there were no statistically significant differences in absolute T cells, or percentages or CD4/CD8 ratio between boys and girls (Table 3 "refer to Additional file 1 "). A very few studies in Africa have compared T cell subset profiles between HIV-1 infected and uninfected infants under the age of 2 years [ 14 , 17 ]. Moodley and colleagues in South Africa reported that the most marked changes in lymphocyte subset of HIV-1 infected children aged between 3 and 15 months were a lower %CD4 and higher %CD8 relative to uninfected infants [ 17 ]. In addition, CD4/CD8 ratio was a good predictor of poor clinical outcome at 3 months. The authors concluded that CD4/CD8 ratio and %CD4 among lymphocytes are reliable markers of HIV-1 infection in an African pediatric population. Furthermore a raised CD8 + cell count rather than a CD4 + cell count was a more specific prognostic marker of disease progression in HIV infected children. Embree and colleagues, in Kenya, also reported that HIV-1 infected children had lower %CD4 and higher %CD8 by 3 months when compared to uninfected children [ 13 ]. The authors concluded that %CD4 and %CD8 among lymphocytes could be useful as an adjunct in HIV-1 diagnosis. The two African studies mentioned above, have documented the clinical value of %CD4, CD4/CD8 ratio and CD8 counts in HIV-1 infection in infants. In summary, the CD4/CD8 ratio may be useful in identifying infected infants while %CD4 will identify infants who may benefit from cotrimoxazole prophylaxis and/or initiation of HAART, and for management of HIV-infected infants in developing countries in general. We thus propose use of flow cytometry, where available, as a point of care diagnostic tool for ill infants admitted to hospitals with clinical symptoms suggestive of HIV infection and/or AIDS. Competing interest The author(s) declare that they have no competing interests. Authors' contributions LSZ designed the study, analyzed the data and drafted the manuscript. DAK, KJN, AB and GJ participated in the design of the study and preparation of the manuscript. SR conducted the statistical analysis and participated in the preparation of the manuscript. OT performed DNA PCR and analysed the data. AB and CG conducted flow cytometry and analysed the data, with the guidance of GJ. MN and PM participated in the collection of infant specimens and demographic data as well as preparation of the manuscript. All authors read and approved the final manuscript. Supplementary Material Additional File 1 Table 3. Gender-based comparison of T cell subset profiles of HIV infected and uninfected babies. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549040.xml |
544564 | A randomized trial of washed red blood cell and platelet transfusions in adult acute leukemia [ISRCTN76536440] | Background Platelet transfusion is universally employed in acute leukemia. Platelet concentrate supernatants contain high concentrations of biologic mediators that might impair immunity. We investigated whether washed platelet and red cell transfusions could improve clinical outcomes in adult patients with acute leukemia. Methods A pilot randomized trial of washed, leukoreduced ABO identical transfusions versus leukoreduced ABO identical transfusions was conducted in 43 adult patients with acute myeloid or lymphoid leukemia during 1991–94. Primary endpoints to be evaluated were platelet transfusion refractoriness, infectious and bleeding complications and overall survival. Results There were no significant differences in infectious or major bleeding complications and only one patient required HLA matched platelet transfusions. Minor bleeding was more frequent in the washed, leukoreduced arm of the study. Confirmed transfusion reactions were more frequent in the leukoreduced arm of the study. Overall survival was superior in the washed arm of the study (40% versus 22% at 5 years), but this difference was not statistically significant (p = 0.36). A planned subset analysis of those ≤50 years of age found that those in the washed, leukoreduced arm (n = 12) had a 75% survival at five years compared with 30% in the leukoreduced arm (n = 10) (p = 0.037) Conclusion This study provides the first evidence concerning the safety and efficacy of washed platelets, and also raises the possibility of improved survival. We speculate that transfusion of stored red cell and platelet supernatant may compromise treatment, particularly in younger patients with curable disease. Larger trials will be needed to assess this hypothesis. | Background In recent years, data have accumulated that platelet transfusion refractoriness and transfusion reactions in patients with hematologic malignancies can be reduced by use of leukoreduced [ 1 - 3 ] and/or ABO identical [ 4 , 5 ] platelet transfusions. Preliminary data also suggest that use of ABO identical [ 6 ] and leukoreduced transfusions [ 7 ] might potentially affect clinical outcomes such as survival and bacterial infection. Data also exist suggesting that alloimmunization to plasma antigens may play a role in platelet transfusion refractoriness, [ 8 ] and that removal of plasma supernatant can reduce the incidence of reactions to platelet transfusions [ 9 ]. One method for removing plasma supernatant from platelet concentrates is washing. However, washing involves loss of perhaps 20% of platelets. No clinical trial data exist comparing washed and unwashed platelet transfusions in terms of efficacy in preventing bleeding, and safety, in terms of unforeseen complications of transfusing platelets subjected to an additional manipulation. Over the last two decades it has become apparent that allogeneic blood transfusions can modify host immunity and clinical outcomes [ 10 ]. Epidemiologic data, animal models, and, in some instances, randomized clinical trials demonstrate that transfusions reduce solid organ allograft rejection and repetitive spontaneous abortions, and increase the likelihood of post-operative bacterial infections [ 11 ]. Perhaps most controversial is the association between blood transfusion and cancer recurrence, which has been convincingly demonstrated in some animal models, [ 12 ] but for which randomized clinical trial evidence is lacking. We observed that patients with cancer had significantly greater recurrence rates if transfused with whole blood [ 13 , 14 ] rather than red cell concentrates, and this epidemiologic association has been confirmed by others [ 15 ]. Platelet transfusions are almost universally used in the supportive care of acute leukemia in adults. The original study design hypothesized a potential benefit from removing soluble immunomodulatory mediators in red cell and platelet concentrates derived from plasma and white cells. White cells and their secreted products are now largely removed prior to storage through filtration. However, there is now reason to be concerned about soluble platelet derived substances that would not be removed by pre-storage leukoreduction, as well as immunomodulatory mediators from plasma itself, such as IgG and soluble HLA antigens. More recent data document that stored platelet concentrate supernatants accumulate striking levels of biologic response modifiers during storage, including vascular endothelial growth factor (VEGF), soluble CD40L, histamine and transforming growth factor (TGF-β1) [ 16 - 18 ]. There is reason to believe that these molecules are infused at what may be clinically significant doses, and might alter recipient immune function. sCD40L has recently been demonstrated to be a growth promoting and apoptosis inhibiting factor for leukemic cells in vitro [ 19 ]. Platelet transfusions are given to patients undergoing myelotoxic chemotherapy for acute leukemia over a two to three week period when the peripheral blood immune system is regenerating. We speculated that platelet transfusions, in addition to causing platelet refractoriness and transfusion reactions, might impair anti-leukemic immunity because allogeneic transfusions have been shown to favor type 2 immunity (e.g., characterized by cytokines such as IL4 and IL10) and suppress type 1 cellular immunity (e.g., IL2, γ-interferon and TNF-α) [ 20 , 21 ]. There are data suggesting that host type 1 immunity may be important in the eradication of residual malignant cells after therapy [ 22 ]. To investigate the possible efficacy of washed platelet transfusions in preventing platelet refractoriness, transfusion reactions, bleeding and improving long term survival in adult acute leukemia, we performed a randomized trial in patients receiving either washed, leukoreduced, ABO identical platelet and red cell transfusions compared with our standard protocol of leukoreduced, ABO identical transfusions. Methods Patients The diagnosis of acute leukemia was based upon laboratory results from our hematopathology laboratory. Only patients receiving chemotherapy with curative intent were included. Three patients died before receiving a full course of induction therapy but are included in the data. Data retrieval of clinical information was done in a blinded fashion. Patients and clinicians were not blinded as to study allocation due to the obvious difference in packaging of washed versus unwashed transfusions. For acute myeloid or undifferentiated leukemia treatment invariably involved initial remission induction attempts with seven days of cytosine arabinoside and three days of an anthracycline. Patients with acute lymphoid leukemia received vincristine and prednisone, as well as other drugs. Patients with high risk features or those who failed to achieve remission received additional courses of induction therapy and/or additional agents depending on attending physician preference. Cytogenetic results, when available, were classified retrospectively according to a currently used scheme [ 23 ]. All stem cell transplants performed involved autologous bone marrow or that from an HLA identical sibling. Patients entering remission also typically underwent consolidation therapy and bone marrow transplant depending on age, performance status, availability of a sibling HLA matched allogeneic donor and other factors. Our institutional review board for studies involving humans approved the study protocols and informed consent documents. Once placed on a particular transfusion protocol, as described below, a patient received only transfusions of that type throughout their treatment. This included consolidation, transplant and relapses, continuing until they were cured or died of their leukemia. In 1991–94, 43 patients participated in a randomized trial of ABO identical, leukoreduced versus ABO identical, leukoreduced, plasma reduced (washed) transfusions. Study endpoints were platelet transfusion refractoriness, total platelet transfusion requirements during induction therapy, infections during induction therapy and overall survival. Randomization was arranged by our Department of Biostatistics employing computer generated random assignments and a blocked design with separate, sequential, sealed, opaque envelopes for patients with diagnoses of acute myeloid (AML) or acute lymphoid (ALL) leukemias. Leukoreduction for all patients was by post-storage, bedside filtration with Pall filters (RC50 and PL100), and plasma reduction of both red cell and platelet transfusions was by saline washing using the Cobe 2991 by a previously published method [ 24 ]. Neither transfusion service nor clinical staff was blinded to study assignment after opening of the envelope, but this was not considered essential as major clinical outcomes (platelet transfusion responsiveness and survival) were unlikely to be affected by staff knowledge of study assignment. Power calculations were not performed as the trial was considered primarily a feasibility trial. All adult patients with acute leukemia who were to receive full dose induction chemotherapy would be recruited during a three year period beginning in early 1991. A maximum of 60 patients were expected to be accrued. Transfusion practice was consistent during the period of the study, employing almost exclusively whole blood derived random platelet concentrates, either ABO identical, leukoreduced, or washed, ABO identical, leukoreduced, according to protocol. Prophylactic transfusions were consistently given at morning platelet counts of <20 × 10 9 /l and usually prior to invasive procedures at <50 × 10 9 /l. Protocol violations in which patients received a transfusion of the incorrect type were <0.5% of all transfusions. All transfusions were treated with 2500 centigray gamma irradiation. We did not collect data on storage time of the red cells or platelets transfused in the study. Patients were to be excluded from the analysis if they did not have acute leukemia or died prior to randomization. Since randomization occurred upon admission to the hospital with the diagnosis of suspected acute leukemia, some patients were found to have other diseases after randomization. These exclusions involved two patients who were randomized to the unwashed group but found not to have acute leukemia upon further investigation. Subgroup analysis was planned for patients ≤50 years of age because they are known to have substantially better prognosis, and for patients with acute myeloid leukemia, who may have a greater risk of HLA alloimmunization. Fifty years of age was also the maximal age for allogeneic bone marrow transplantation in patients with acute leukemia at the time of the study. Data were collected by a blinded coauthor (JMH) from medical record review of the admission for initial induction, and follow-up data on survival obtained from the local tumor registry. The clinical data for blood component use and morbidity were collected only for the initial admission to achieve treatment time comparability. Bleeding was evaluated for the initial admission for remission induction, and defined as minor (1–2 days not requiring any therapeutic intervention such as wound site hematoma, guaiac positive stools, mild epistaxis not requiring transfusion nor packing, etc.) or major (requiring transfusion of red cells or otherwise mandating surgical or other therapeutic intervention). Refractoriness was evaluated for the entire duration of a patient's transfusion therapy until cessation of transfusions or death, and defined as the need for HLA matched platelet transfusions, as determined by the attending hematologist. Transfusion reactions were reported at the discretion of the nursing staff and evaluated by Transfusion Service residents who were not blinded as to the component received. A new fever or new rigors occurring during or shortly after a transfusion was considered a reaction whereas a fever or rigor that occurred shortly before the transfusion started was considered unrelated. Statistical methods Statview 5.0 (SAS Institute, Cary, NC) was employed to calculate survival curves by the Kaplan-Meier method and Cox proportional hazards regressions. For continuous variables, the Mann-Whitney test was used for bivariate comparisons, and for categorical variables, Fisher's exact test or Chi square with continuity correction were employed as appropriate. No corrections were made for multiple comparisons, because except for the major outcome variables of survival, platelet refractoriness, infectious complications and platelet utilization, the data comparisons are considered exploratory. Results The demographic pre-treatment and initial hospitalization clinical outcomes data for patients in the study are shown in Tables 1 and 2 , and the long term survival results are shown in the Kaplan-Meier plot in figure 1 . Except for cytogenetics there were few differences between the two cohorts in terms of pre-treatment variables. Of those who had cytogenetics successfully determined there were more patients with poor risk cytogenetics in the washed arm (11 of 18; 61%) than in the unwashed arm (2 of 9; 22%) of the study (p = 0.11 by Fisher's exact test). There was no statistically significant increased need for red cell or platelet transfusions, nor increased bleeding in the patients receiving washed transfusions. Table 1 Demographic and pre-treatment clinical variables according to type of transfusions given. ABO Matched, Leukoreduced ABO Matched, Leukoreduced, Washed P value by Mann Whitney or Fisher's Exact Test N 18 25 Age 48 ± 23 47 ± 17 0.89 Male 9 (50%) 11 (44%) 0.76 ALL/AML 4/14 5/20 0.99 Antecedent Hematologic Disorder (MDS) 0 of 18 (0%) 3 of 25 (12%) 0.25 Favorable Risk Cytogenetics 1 (6%) 0 (0%) 0.09* Standard Risk Cytogenetics 6 (33%) 7 (28%) Poor Risk Cytogenetics 2 (11%) 11 (44%) Unknown Cytogenetics 9 (50%) 7 (28%) Admission Blast Count (×1000/μl) 25 ± 36 23 ± 41 0.59 All data are mean ± 1 SD. MDS = myelodysplastic syndrome *For cytogenetics as a category overall Table 2 Outcome variables according to type of transfusions received. ABO Matched, Leukoreduced ABO Matched, Leukoreduced, Washed P value by Mann Whitney or Fisher's Exact Test N 18 25 Red Cells (units) 16 ± 12 16 ± 6.2 0.21 Platelets (units) 84 ± 100 73 ± 49 0.64 Platelets/Day (units) 1.8 ± 1.2 1.9 ± 1.3 0.90 Red Cells/Day (units) 0.4 ± 0.1 0.4 ± 0.3 0.39 Platelets/Red Cell 4.6 ± 2.4 4.4 ± 2.5 0.74 Courses of Induction Chemotherapy 1.3 ± 0.6 1.3 ± 0.4 0.90 Length of stay (days) 43 ± 24 42 ± 17 0.47 Received HLA matched platelets 0 of 18 (0%) 1 of 25 (4%) Days with fever >37 degrees Celsius 16 ± 11 15 ± 9.2 0.90 Days of antibiotics 36 ± 26 36 ± 18 0.25 Positive Microbial Cultures 1.0 ± 1.2 1.3 ± 1.4 0.53 Days with bleeding 0.62 ± 2.2 0.58 ± 1.2 0.19 Reported Transfusion Reactions per Patient 0.4 ± 0.8 0.4 ± 0.7 0.91 Complete remission at discharge 12 of 18 (67%) 17 of 25 (68%) 1.00 Received BMT 8 of 18 (44%) 14 of 25 (56%) 0.54 All continuous variables results are means ± 1 SD. Figure 1 Results of the randomized trial of washed (n = 25 patients) versus unwashed (n = 18) platelet transfusions . There is no significant difference in survival by the logrank test (p = 0.36). Censored data points are patients remaining alive. Clinically evident bleeding was uncommon (8 of 43 patients, 19%) and major bleeding occurred in only 2 of 43 patients (5%). Minor bleeding was more common in those receiving washed platelets (6 of 25, 24%) than unwashed (0 of 18, 0%) (p = 0.03 by Fisher's exact test) but major bleeding occurred in only one patient in each arm of the study (1 of 25 versus 1 of 18; not significantly different). Bleeding was in all instances related to specific anatomic lesions, such as minor epistaxis, hematomas at the site of trauma or invasive procedures, or guaiac positive stools. For six patients these mild bleeding episodes were only seen on one or two days during their admission, without generalized petechiae or purpura. One patient in the washed arm had melena for five days, which resolved. The only life threatening bleeding was non-fatal and occurred in one patient in the unwashed arm of the study who experienced eight days of hemoptysis and received 416 units of platelets and 52 units of red cells during induction therapy. Bleeding did not influence overall survival at last follow-up, which was 38% in those who had bleeding and 35% in those patients with no bleeding (p = 0.94 by logrank test). Patients with bleeding required more platelet (mean of 141 ± 118) (1 SD) and red cell transfusions (21 ± 13) than patients with no bleeding (61 ± 44 platelets and 15 ± 7 red cells), but this was almost exclusively due to the one patient in the unwashed arm of the study with eight days of hemoptysis. When this patient was removed from the analysis, the patients with bleeding received no more transfusions than those without bleeding. Reported numbers of transfusion reactions were similar in both arms of the study (a mean of 0.4 per patient). In the washed arm of the study there were eight transfusion reactions reported due to intercurrent fever and two allergic reactions (rash and/or urticaria) to unwashed platelet transfusions given in violation of protocol due to clinical urgency. In the unwashed arm of the study there were three transfusions with intercurrent fever and four allergic reactions. Reactions were reported in 7 of 25 (28%) of patients in the washed study arm and 7 of 18 (39%) of the patients in the unwashed arm (p = 0.52). When evaluated by the transfusion medicine resident, seven of the reactions in the washed arm were considered to be pre-existing fever and the reported reaction likely unrelated to transfusion. When only "on protocol" transfusions judged causally related by a blood bank physician are considered, the patients in the washed arm were less likely to experience reactions (1 of 25; 4%) than patients in the unwashed arm (7 of 18; 39%) (p = 0.0058 by Fisher's exact test). During the course of the treatment, three of 18 patients randomized and analyzed in the unwashed arm had severe or repeated transfusion reactions and were placed on a washed protocol for future transfusions. As determined by a Cox proportional hazards regression, survival was not associated significantly with type of leukemia (p = 0.37), cytogenetic results (p = 0.63), or receipt of washed blood transfusions (p = 0.62) but was significantly associated with age (p = 0.0002), with younger patients surviving longer. Because long-term survival in acute leukemia is uncommon in those over the age of 50–60 years, we also performed a planned subset comparison of the 22 patients in the trial ≤50 years of age. We recognized that this subset would be small, rendering statistical analysis more difficult. These results are shown in figure 2 , along with demographics on these patients in Tables 3 and 4 . Figure 2 Survival in those patients in the randomized trial of washed (n = 12) versus unwashed (n = 10) platelet transfusions ≤50 years of age is plotted by the Kaplan-Meier method. Those in the washed group had significantly better survival (p = 0.037 by logrank test). Two of the patients in the washed group were both alive and in remission at last recorded follow-up of 47 months. Two of the patients in the unwashed arm were both alive and in remission at last recorded follow-up of 116 months. At the minimum follow-up time of 41 months 9 of 12 patients in the washed arm were alive and in remission, as compared with 3 of 10 in the unwashed arm. Table 3 Demographic and pre-treatment clinical variables according to type of transfusions given in those ≤50 years of age. ABO Matched, Leukoreduced ABO Matched, Leukoreduced, Washed P value by Mann Whitney or Fisher's Exact Test N 10 12 Age 30 ± 9.4 32 ± 9.5 0.77 Male 5 (50%) 6 (50%) 1.00 ALL/AML 1/9 4/8 0.32 Antecedent Hematologic Disorder (MDS) 0 of 10 (0%) 0 of 12 (0%) 1.00 Favorable Risk Cytogenetics 0 (0%) 0 (0%) 0.09 Standard Risk Cytogenetics 5 (33%) 2 (28%) Poor Risk Cytogenetics 1 (11%) 6 (44%) Unknown Cytogenetics 4 (50%) 4 (28%) Admission Blast Count (×1000/μl) 29 ± 49 29 ± 49 0.56 All data are mean ± 1 SD. MDS = myelodysplastic syndrome Table 4 Outcome variables according to type of transfusions received in those ≤50 years of age. ABO Matched, Leukoreduced ABO Matched, Leukoreduced, Washed P value by Mann Whitney or Fisher's Exact Test N 10 12 Red Cells (units) 20 ± 14 16 ± 6.2 0.93 Platelets (units) 127 ± 121 69 ± 52 0.23 Platelets/Day (units) 2.4 ± 1.1 1.4 ± 1.0 0.039 Red Cells/Day (units) 0.4 ± 0.06 0.4 ± 0.1 0.39 Platelets/Red Cell 5.9 ± 2.0 4.1 ± 3.0 0.069 Courses of Induction Chemotherapy 1.5 ± 0.8 1.2 ± 0.4 0.31 Length of stay (days) 50 ± 30 47 ± 18 0.56 Received HLA matched platelets 0 of 10 (0%) 0 of 12 (0%) 1.00 Days with fever >37 degrees Celsius 19 ± 12 14 ± 9.1 0.36 Days of antibiotics 44 ± 31 40 ± 21 0.74 Positive Microbial Cultures 1.1 ± 1.3 1.4 ± 1.5 0.76 Days with bleeding 1.0 ± 2.8 0.46 ± 0.67 0.36 Transfusion Reactions per Patient 0.4 ± 0.7 0.4 ± 0.5 0.73 Complete remission at discharge 9 of 10 (90%) 11 of 12 (92%) 1.00 Received BMT 8 of 10 (80%) 11 of 12 (92%) 0.57 All continuous variables results are means ± 1 SD. There were more patients with ALL (p = 0.32) and more patients with poor risk cytogenetics in the washed arm of the study (p = 0.09). Otherwise there were few differences between the patients ≤50 years of age in the two cohorts. Patients in the washed arm had significantly better overall survival and required fewer platelets per day or per red cell transfused than patients in the unwashed arm. There were no significant differences in bleeding, use of growth factors or white cell transfusions between the two arms of the study, in either the entire or younger cohorts (data not shown but available from the authors on request). Keeping in mind the extremely small number of patients involved, a Cox proportional hazards regression was performed on these younger patients in the randomized trial. Receipt of washed transfusions was a significant predictor of longer survival (p = 0.011), as was cytogenetics (patient's with unknown cytogenetics having poorer survival, p = 0.027) but age and type of leukemia were not statistically significant predictors of survival time. The survival of all patients ≤50 restricted to those with AML is shown in Figure 3 confirming that the patients in the washed arm did not survive longer solely because of the larger number of patients with ALL in that arm of the study. Figure 3 Survival in those patients in the randomized trial of washed (n = 8) versus unwashed (n = 9) platelet transfusions ≤50 years of age with AML is plotted by the Kaplan-Meier method. Those in the washed group experienced better survival but this was not statistically significant (p = 0.10 by logrank test). At a minimum follow-up of 41 months, 6 of 8 patients in the washed arm were in remission and alive compared with 3 of 9 in the unwashed arm (two of the patients in the unwashed arm were alive and in remission at 116 months). In the subset of all patients ≤50 years of age with either type of leukemia who both achieved a complete remission and also received definitive post-remission therapy in the form of bone marrow transplantation, the survival in patients receiving washed transfusions (n = 12) was significantly better than those receiving unwashed transfusions (n = 8) (p = 0.031 by log rank test) (graph not shown). Discussion and conclusions To our knowledge, these data represent the first randomized trial of washed platelet transfusions in any setting. The issues of the efficacy and safety of washed platelet transfusions are of importance because (1) some patients require plasma-reduced transfusions to treat allergic or febrile reactions, (2) new methods of viral and bacterial pathogen-inactivation for transfusions may require washing prior to transfusion, and (3) a growing body of evidence suggests that immunologically important molecules are present in the stored supernatant of blood transfusions that might, speculatively, affect clinical outcomes. Our study design was less likely to detect a benefit of platelet washing because, unlike many centers, we routinely use ABO identical platelet transfusions for patients with leukemia. There is evidence from two small randomized trials in leukemia that this reduces refractoriness [ 4 , 5 ] and increases survival [ 6 ]. Preliminary results exist that ABO matching may also reduce morbidity and mortality in surgical patients [ 25 ]. There was no apparent benefit to washed transfusions in terms of reduced platelet transfusion refractoriness, reduced bacterial infections, reduction in reported febrile or allergic transfusion reactions or reduced length of stay. Confirmed transfusion reactions were less frequent in the washed arm of the study, and 15% of the patients in the leukoreduced arm of the study had severe or repeated reactions that led to their receiving washed transfusions, which abrogated those reactions. Our data do provide evidence, for the first time, that washed platelet transfusions are probably as safe and efficacious as standard leukoreduced transfusions. The mean number of days with bleeding was marginally but not significantly reduced in the patients in the washed group, as were the number of platelet transfusions needed per day in younger patients, and the ratio of platelet transfusions to red cell transfusions. However, there was more minor bleeding in the washed group raising the possibility that washed platelets are slightly less effective. However, the prevalence of major bleeding requiring treatment was the same in each group and very low (<5%). The only life threatening hemorrhage in these 43 patients occurred in a patient in the leukoreduced arm. This study is also the first attempt to investigate whether transfusion practices during initial induction therapy are associated with changes in overall survival in acute leukemia in adults. The underlying rationale for studying this issue are observations demonstrating that allogeneic transfusions alter host T cell and natural killer cell immune function in surgical patients and experimental animals [ 11 ]. The major causes of death in adults with leukemia are failure to achieve complete remission, and relapse after achieving complete remission. There is some evidence from the allogeneic bone marrow transplant literature that host immune function may play a role in preventing relapse during the post-treatment period, but there are no data that demonstrate a role for the immune system in achieving complete initial remissions [ 26 ]. Our study found no difference in complete remission induction success rates with differing transfusion protocols, but does support the possibility of an association between type of red cell and platelet preparation transfused and long term survival. Large numbers of red cell and platelet transfusions are given to patients with acute leukemia during the period of recovery from aplasia that is caused by cytotoxic chemotherapy given during remission induction. Additional transfusions are given during consolidation and bone marrow or peripheral blood stem cell transplantation. Many of these transfusions occur during the period when at least the bone marrow and peripheral blood compartments of the immune system are reconstituting and might be susceptible to the immunomodulatory effects of transfusions. Both white cell and platelet-derived mediators are present in the transfused red cells as well as platelets. Prestorage leukoreduction (not employed in this study) removes the vast majority of the white cells and platelets from stored red cells, and red cell concentrates would be expected to have much lesser concentrations of biologic mediators, even without washing. Platelet derived mediators would, of course, still be present in prestorage leukoreduced platelet concentrates. Transfusions are known to cause suppression of type 1 cellular immunity and upregulation of type 2 humoral immunity [ 20 , 21 ]. Such immune deviation could, speculatively, impair host defenses provided by T cells, dendritic cells and natural killer cells that might be involved in the eradication or control of residual tumor. Potential mediators of down regulation of cellular immunity by transfusions include allogeneic white cells, red cells or platelets, ABO antigen-antibody complexes formed after repeated ABO non-identical platelet transfusions, and the stored supernatant of, in particular, platelet concentrates. Transfusion of stored platelet supernatant plasma might be hypothesized to mediate such effects by immunoregulatory and tumor growth promoting factors, such as sCD40L or angiogenic factors such as VEGF [ 16 - 18 ]. Leukoreduction filters remove white cells from the transfused blood components. In the case of pre-storage leukoreduction, filtration also removes some of the biologic response modifiers secreted by white cells during storage. Platelet concentrate supernatant contains large amounts of mediators such as soluble CD40L (CD154), VEGF, TGF-β1, histamine and other biologic response modifiers that might impair cellular immunity [ 16 - 18 ]. Washing immediately prior to transfusion removes most soluble materials from the transfused platelet and red cell transfusions, including those released during storage from the platelets, red cells and white cells. Our data provide initial support for the novel hypothesis that changing transfusion practices could play a role in long-term survival in acute leukemia, particularly in younger patients who can be potentially cured. There are distinct limitations to what can be concluded from our data. The number of patients studied is very small and almost all of them had AML. The statistically significant improvement in survival in patients ≤50 years of age with AML who had unusually good survival represents a subset analysis, not a primary outcome group and thus may represent the play of chance. The survival of 75% of any cohort of adult patients with acute leukemia for 4–5 years is a distinctly unusual circumstance [ 27 ]. This could be due to the benefits of receiving washed transfusions, but could also be a chance occurrence in a small cohort of patients. Our overall survival in all patients of 40% in the larger cohort of patients of all ages and risk factors in the washed arm is equivalent to the best survival that has been reported in low risk, younger patients in other trials from the period of the early 1990s [ 27 ]. As shown in the figures, there was only one relapse and death that occurred at beyond five years in our cohorts, with a number of patients alive and in remission at 5–12 years after diagnosis. Thus it appears that if there was benefit from washing of transfusions, it probably involves an increased likelihood of durable remission, rather than purely delaying relapse. These data, however promising, raise hypotheses for additional testing rather than proving a principle. If transfusion practice impacts anti-leukemic immunity, and/or survival as our data suggest, a moderate sized randomized clinical trial should be able to confirm this in relatively few years. This is a propitious time for such trials as the need for prophylactic platelet transfusion therapy is being revisited. The question is being raised as to whether platelet transfusions in non-bleeding patients are truly necessary or need be as frequent as currently employed in this disease [ 28 ]. One strategy for randomized trials would be to randomize patients to only therapeutic platelet transfusions (transfusion only for bleeding manifestations) versus current standard practice of prophylactic transfusions at a set threshold such as 10 × 10 9 /μl. Platelet washing is time consuming, delaying transfusion by about 2.5–3 hours, may present an additional opportunity for bacterial contamination, and involves some loss of platelets (about 20%). Platelet washing may not be feasible in clinically urgent situations. Direct costs are modest, at less than $40 per transfusion. Washing adds about $500–2000 to the total cost of caring for patients with acute leukemia in our center from diagnosis to cure or death. This is less than 0.5–1% of the total costs of treating AML or ALL with curative intent. There are few or no proven side effects of platelet washing other than reduced dose of platelets transfused. There are some additional potential clinical benefits of washing including a reduced likelihood of transfusion complications such as febrile non-hemolytic transfusion reactions, transfusion-related acute lung injury, and allergic reactions. These preliminary data, albeit from a very small number of patients, raise the possibility that some patients with acute leukemia might benefit from washed transfusions. Larger studies are indicated to explore this possibility. Competing interests Gambro BCT provided partial support for these studies. Dr. Blumberg has received lecture honoraria and a previous research grant from Pall Biomedical Corp. and lecture honoraria from Baxter. Authors' contributions NB and JMH had the original idea for the study. All authors contributed to design of the study, drafting of the manuscript and all revisions. NB supervised the laboratory preparation of blood components for the study. JMR provided overall clinical supervision of the study and care for many of the patients. JMH collected the data and JMH and NB analyzed the data. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544564.xml |
545646 | Screening large-scale association study data: exploiting interactions using random forests | Background Genome-wide association studies for complex diseases will produce genotypes on hundreds of thousands of single nucleotide polymorphisms (SNPs). A logical first approach to dealing with massive numbers of SNPs is to use some test to screen the SNPs, retaining only those that meet some criterion for futher study. For example, SNPs can be ranked by p-value, and those with the lowest p-values retained. When SNPs have large interaction effects but small marginal effects in a population, they are unlikely to be retained when univariate tests are used for screening. However, model-based screens that pre-specify interactions are impractical for data sets with thousands of SNPs. Random forest analysis is an alternative method that produces a single measure of importance for each predictor variable that takes into account interactions among variables without requiring model specification. Interactions increase the importance for the individual interacting variables, making them more likely to be given high importance relative to other variables. We test the performance of random forests as a screening procedure to identify small numbers of risk-associated SNPs from among large numbers of unassociated SNPs using complex disease models with up to 32 loci, incorporating both genetic heterogeneity and multi-locus interaction. Results Keeping other factors constant, if risk SNPs interact, the random forest importance measure significantly outperforms the Fisher Exact test as a screening tool. As the number of interacting SNPs increases, the improvement in performance of random forest analysis relative to Fisher Exact test for screening also increases. Random forests perform similarly to the univariate Fisher Exact test as a screening tool when SNPs in the analysis do not interact. Conclusions In the context of large-scale genetic association studies where unknown interactions exist among true risk-associated SNPs or SNPs and environmental covariates, screening SNPs using random forest analyses can significantly reduce the number of SNPs that need to be retained for further study compared to standard univariate screening methods. | Background Genome-wide association studies for complex diseases such as asthma, schizophrenia, diabetes, and hypertension will soon produce genotypes on hundreds of thousands of single nucleotide polymorphisms (SNPs). Due to the large number of SNPs tested and the potential for both genetic and environmental interactions, determining which SNPs modify the risk of disease is a methodological challenge. While the number of genotypes produced by candidate gene approaches will be somewhat less daunting, on the order of hundreds to thousands of SNPs, it will still be a considerable challenge to weed out the noise and identify the SNPs contributing to complex traits. A logical first approach to dealing with massive numbers of SNPs is to first conduct univariate association tests on each individual SNP, in order to screen-out those with no evidence for disease association. The primary goal of such a procedure is not to prove that a particular variant or set of variants influences disease risk, but to prioritize SNPs for further study. Using a univariate test at this stage will result in low power for SNPs with very small marginal effects in the population, even if the SNPs have large interaction effects. Of course, in addition to taking all individual SNPs, all SNP pairs could also be tested for association. However, when dealing with multiple thousands of SNPs at the outset, such an approach is cumbersome, and raises the question of where to stop: why not all sets of three, four, or even five SNPs as well? Many model-building methods exist for dealing with large numbers of predictors. For example, stochastic search variable selection (SSVS) [ 1 ], a form of Bayesian model selection, has been explored as a tool to discover joint effects of multiple loci in the context of genetic linkage studies [ 2 - 4 ]. However, these methods are limited in the number of predictors that can be included at one time, causing some researchers to resort to a two-stage approach, in which only main effects are considered in a first stage, and interactions between loci with strong main effects are considered in a second stage. This approach can lead to the loss of important interactions with only weak main effects. Multivariate adaptive regression splines (MARS) models have also been explored in the context of genetic linkage and association studies [ 5 , 6 ] with some degree of success. However, these and other model selection methods appear to be limited in the number of predictors that can reasonably be accommodated in one analysis, and the types of possible interactions that are allowed must be specified in advance. They are not suited to the initial task of identifying from a massive set of SNPs a subset for further analyses. Combinatorial partitioning and multifactor dimensionality reduction [ 7 - 10 ] are closely related methods developed specifically to detect higher-order interactions among polymorphisms that predict trait variation. However, these methods are meant to identify interactions among small sets of SNPs, and have minimal power in the presence of genetic heterogeneity [ 10 ]. They are therefore inappropriate for use as a screening tool for searching through thousands of SNPs to identify those contributing to phenotypes in the context of whole-genome association studies. The problem remains: how do we reasonably weed down from thousands or hundreds of thousands of SNPs to a number that can be used by available modeling methods, without losing the interactions that we hope to model in the first place? An additional concern to be considered is genetic heterogeneity. We define genetic heterogeneity to mean that there are multiple possible ways to acquire a disease or trait that can involve different subsets of genes. Traditional regression models are limited in their ability to deal with underlying genetic heterogeneity (see, e.g ., [ 11 ]). If genetic heterogeneity also leads to phenotypic heterogeneity, then methods that classify individuals into phenotypic subgroups for further analysis can be successful. Likewise, if heterogeneity in genetic etiology is primarily due to ethnic background, sub-dividing samples by self-reported ethnicity or genetically defined subgroups can be a powerful antecedent to data analyses for the identification of complex disease genes. However, even in the realm of Mendelian genetic diseases, heterogeneity is rarely so simple. For example, multiple polymorphisms in each of two different genes are responsible for familial breast cancer in the relatively homogeneous sub-population of Ashkenazi Jewish women [ 12 ]. When the root of the heterogeneity is not known a priori , traditional regression models, which lump all individuals into a single group and estimate average effects over the entire sample, are unlikely to successfully identify the genetic causes of diseases. Classification trees and random forests Tree-based methods consist of non-parametric statistical approaches for conducting regression and classification analyses by recursive partitioning (see, e.g., Hastie et al. [ 13 ]). These methods can be very efficient at selecting from large numbers of predictor variables such as genetic polymorphisms those that best explain a phenotype. Tree methods are useful when predictors may be associated in some non-linear fashion, as no implicit assumptions about the form of underlying relationships between the predictor variables and the response are made. They are well-adapted to dealing with some types of genetic heterogeneity, as separate models are automatically fit to subsets of data defined by early splits in the tree. The ease of interpretation of classification trees, along with their flexibility in accommodating large numbers of predictors and ability to handle heterogeneity, has resulted in increasing interest in their application to genetic association and linkage studies. Classification trees have been adapted for use with sibling pairs to subdivide pairs into more homogenous subgroups defined by non-genetic covariates [ 14 ], thus increasing the power to detect linkage under heterogeneity [ 15 ]. They have also shown promise for the dissection of complex traits for both linkage and association [ 16 , 17 ], and for exploring interactions [ 6 ]. A related adaptive regression method has also shown promise in selecting a small number of predictive SNPs from a set of hundreds of potential predictors [ 18 ]. Tree methods have also been used to identify homogeneous groups of cases for further analyses [ 19 ], and as an adjunct to more traditional association methods [ 20 ]. Classification trees are grown by recursively partitioning the observations into subgroups with a more homogeneous categorical response [ 21 ]. At each node, the explanatory variable (e.g., SNP) giving the most homogeneous sub-groups is selected. Choosing alternative predictors that produce slightly sub-optimal splits can result in very different trees that have similar prediction accuracy. The Random Forests methodology [ 22 ] builds on several other methods using multiple trees to increase prediction accuracy [ 23 - 25 ]. A random forest is a collection of classification or regression trees with two features that distinguish it from trees built in a deterministic manner. First, the trees are grown on bootstrap samples of the observations. Second, a random selection of the potential predictors is used to determine the best split at each node. For each tree, a bootstrap sample is obtained by drawing a sample with replacement from the original sample of observations. The bootstrap sample has the same number of individuals as the original sample, but some individuals are represented multiple times, while others are left out. The left-out individuals, sometimes called "out-of-bag", are used to estimate prediction error. Because a different bootstrap sample is used to grow each tree, there is a different set of out-of-bag individuals for each tree. With a forest of classification trees, each tree predicts the class of an individual. For each individual, the predictions, or "votes", are counted across all trees for which the individual was out-of-bag, and the class with the most votes is the individual's predicted class. Random forests produce an importance score for each variable that measures its importance. This score can be used to prioritize the variables, much as p-values from test statistics are used. Using ensembles of trees built in this manner increases the probability that some trees will capture interactions among variables with no strong main effect. Unlike variable selection methods, interactions among predictors do not need to be explicitly specified in order to be utilized by a forest of trees. Instead, any interactions between variables serve to increase the importance of the individual interacting variables, making them more likely to be given high importance relative to other variables. Thus, random forests appear to be particularly well-suited to address a primary problem posed by large scale association studies. In preliminary studies, we have shown the potential of random forests in the context of linkage analysis [ 26 ]. Other investigators are beginning to recognize the potential of the Random Forest methodology for studying SNP association [ 27 ] and classification [ 28 ]. To fully understand the basis of complex disease, it is important to identify the critical genetic factors involved, and to understand the complex relationships between genotypes, environment, and phenotypes. The few successes to date in identifying genes for complex disease suggest that despite carefully collected large samples, novel approaches are needed in the pursuit to dissect the multiple and varying factors that lead to complex human traits. Ultimately, the challenge in identifying polymorphisms that modulate the risk of complex disease is to find methods that can seamlessly handle large numbers of predictors, capitalize on and identify interactions, and tease apart the multiple heterogeneous etiologies. Here, we explore the use of the Random Forest methodology [ 22 , 29 ] as a screening tool for identifying SNPs associated with disease in the presence of interaction, heterogeneity, and large amounts of noise due to unassociated polymorphisms. Results Genetic models We simulated complex diseases with sibling recurrence risk ratio for the disease ( λ s ) fixed at 2.0 and population disease prevalence K p equal to 0.10. These values are consistent with or lower than estimates from known complex genetic traits, such as Alzheimer disease, where estimates of cumulative prevalence in siblings of affected range from 30–40%, compared to a population prevalence of 10% at age 80 [ 30 ]. Such traits are understood to be caused by multiple interacting genetic and environmental factors. Our genetic models incorporate both genetic heterogeneity and multiplicative interaction as defined by Risch [ 31 ]: we simulate sets of 4, 8, 16, and 32 risk SNPs ("rSNPs") in linkage equilibrium, interacting in independent pairs or quartets to increase disease risk, and contributing equally to the overall sibling recurrence risk ratio of 2 and population disease prevalence of 0.10. For simplicity, we simulated the models such that each rSNP pair or quartet accounts for the same proportion of the genetic risk, and each SNP within a pair/quartet is responsible for an equal proportion of the genetic risk. Thus, all of the rSNPs simulated for a model have the same allele frequency and the same observed marginal effect in the population. We denote the models using the shorthand HhMm, where H = h (=2, 4, 8, 16) is the number of heterogeneous systems, and M = m (=2 or 4) is the number of multiplicatively interacting SNPs within each system. For example, 16 loci are responsible for the total λ s = 2 and K p = .10 for models H4M4 and H8M2, and 32 loci are responsible for models H8M4 and H16M2. Table 1 presents relevant features of our models. The Methods section describes the genetic models in more detail. Table 1 Genetic models used for simulating case-control data. Risk SNPs Case Genotype Correlation Allele Marginal GRR Penetrance Factors Within System Between System Model Number Frequency Het Hom 0 1 2 H2M2 4 0.207 2.85 4.71 2.4E-02 0.51 1 0.30 -0.32 H4M2 8 0.160 1.99 2.96 3.9E-04 0.50 1 0.35 -0.12 H8M2 16 0.104 1.66 2.18 8.0E-06 0.56 1 0.32 -0.05 H16M2 32 0.069 1.46 1.78 1.4E-05 0.59 1 0.28 -0.02 H4M4 16 0.282 1.63 1.79 1.2E-08 0.79 1 0.17 -0.06 H8M4 32 0.214 1.34 1.40 2.8E-03 0.86 1 0.14 -0.02 Simulation and analysis All analyses were performed on 100 replicate data sets of 500 cases and 500 controls. In addition to the rSNPs contributing to the trait, we simulated noise SNPs ("nSNPs"), independent of disease status, with allele frequencies distributed equally across the range .01–.99. To simulate the results of an association study, in which we do not expect to be lucky enough to genotype all polymorphisms related to a trait, we included only a subset of the total number of rSNPs in each analysis. We denote the analysis design using the shorthand KkSsNn, where K = k is the total number of rSNPs genotyped in the analysis, S = s is the number of SNPs within each interaction system genotyped, and N = n is the total number of SNPs genotyped in the design. Thus, N-K is the total number of nSNPs in the analysis. For example, suppose the genetic model is H8M4, and the design is K4S2N100. Then out of the total of 8 × 4 = 32 rSNPs that contribute to the trait, four are genotyped: two interacting SNPs from within each of two heterogeneity systems. Six heterogeneity systems are not represented at all in the analysis. In addition to the four genotyped rSNPs, 100-4 = 96 total nSNPs are also genotyped in the design. Comparison of raw and standardized importance scores Random forests version 5 software [ 29 ] produces both raw ( I T ) and standardized ( Z T ) variable importance scores (see Methods section for definitions of the scores). Little is known about the properties of importance indices under different distributions of the predictor variables. We use simulation to investigate their properties in the context of discrete predictors such as genetic polymorphisms conferring susceptibility to a complex trait. We first compared the raw and standardized scores, in order to determine whether one might outperform the other in screening. We considered a K4S2N100 analysis design for each genetic model described in Table 1 . I T and Z T are highly correlated; the average correlation coefficient over 100 replicate data sets ranged from .93 (H8M4) to >.99 (H2M2 and H4M2) (Table 2 ). The average correlation between the ranks based on I T and Z T for the 100 SNPs over the 100 replicate data sets was 0.98 for each of the six models (Table 2 ). Comparing the ranks of the four rSNPs among all SNPs, neither importance measure outperforms the other for all models (Table 3 ). The mean ranks of the rSNPs for the two measures are significantly different only for the H16M2 and H8M4 models. For H16M2, the average rank of the rSNPs is higher for Z T than for I T . The opposite is true for H8M4 (see Table 3 ). Table 2 Summary of the correlation between I T and Z T ("raw") and rank( I T ) and rank( Z T ) ("rank") for four rSNPs and 96 nSNPs over 100 replicate data sets: K4S2N100 analysis design. H2M2 H4M2 H8M2 H4M4 H16M2 H8M4 r 2 raw rank raw rank raw rank raw rank raw rank raw rank Mean 0.996 0.983 0.990 0.982 0.975 0.982 0.957 0.982 0.964 0.982 0.933 0.982 SD 0.001 0.006 0.003 0.006 0.009 0.006 0.013 0.007 0.013 0.006 0.018 0.007 Min 0.990 0.963 0.980 0.957 0.941 0.960 0.921 0.955 0.926 0.962 0.891 0.953 Max 0.998 0.993 0.996 0.992 0.989 0.993 0.984 0.992 0.983 0.994 0.970 0.992 Table 3 Comparison of ranks based on Z T and I T for the four rSNPs over 100 replicate data sets: K4S2N100 analysis design. H2M2 H4M2 H8M2 H4M4 H16M2 H8M4 Rank: Z T I T Z T I T Z T I T Z T I T Z T I T Z T I T Mean 2.5 2.5 2.5 2.5 2.51 2.52 2.64 2.61 5.16 5.94 9.35 8.69 SD 1.12 1.12 1.12 1.12 1.13 1.16 1.74 1.62 8.06 8.33 13.67 13.9 Max 4 4 4 4 6 8 23 21.5 77 62.5 83 88.5 p-value* 0.94 1.00 0.77 0.12 1.32E-20 2.91E-12 *p-value for the Wilcoxon signed-rank test comparing the rSNP ranks based on I T and Z T Ranking SNPs based on Z T and Fisher p-value We next compared the ranking of rSNPs by importance score (Z T ) to ranking by Fisher Exact test p-value using K4S2N100 and K4S2N1000 analysis designs, where two SNPs from each of the first two interaction systems are in the analysis. Figure 1 shows the proportion of replicates for which the top ranked 1, top 2, top 3, and top 4 SNPs are the four genotyped rSNPs in the data set for each of the four most complex genetic models. For N100, the random forest Z T criterion ranks the four rSNPs as the most significant SNPs more often than the univariate Fisher Exact test association p-value under all genetic models. The difference between the random forest and association p-value ranking is less extreme for N1000. For the H8M4 genetic model, the results do not suggest that one ranking system is better than the other overall. Figure 2 shows the proportion of replicates for which all rSNPs are among the top N SNPs. In other words, it is the proportion of data sets for which none of the genotyped rSNPs are screened out, if the top ranking N SNPs are chosen for further study. For N100, a consistently higher proportion of replicates ranked using Z T contain all of the rSNPs. Thus, for a given probability of retaining all of the rSNPs, more SNPs can be eliminated using the Z T criterion than the Fisher exact test p-value. For example, for model H16M2, only 15 SNPs must be retained to have 80% probability that the 4 rSNPs are in the retained set, while 44 SNPs must be retained if the p-value criterion is used. The difference is less dramatic for H8M4: 37 SNPs give 80% probability that the four genotyped rSNPs are in the retained set for the Z T criterion, compared to 43 for the p-value criterion. For N1000, the advantage of the Z T criterion is clear for the H8M2 and H16M2 models. For H4M4, the advantage of Z T is minor, while for H8M4, ranking by Z T appears to give poorer results than the p-value criterion for the higher cutoff values of N. A second interpretation of Figure 2 is that, for any number of retained (not screened-out) SNPs, the probability that all of the genotyped rSNPs are retained is higher for the Z T criterion than for the univariate p-value criterion for all but the H8M4 model with 1000 total SNPs. Figure 1 Proportion of replicates for which the most significant 1, 2, 3, and 4 SNPs are all rSNPs for K4S2N100 and K4S2N1000 analysis designs. Genetic models are listed on the plots. "RF" and "Fisher" refer to the random forest importance index Z T and the Fisher Exact test p-value. See text for notation description. Figure 2 Proportion of replicates for which all rSNPs are among the top-ranking N SNPs for K4S2N100 and K4S2N1000 analysis designs. Other notation as in Figure 1. Noticing that the analyses with all SNPs from an interacting system (e.g., the H16M2K4S2 simulations) had a more substantial improvement in ranking using Z T over p-value than the analyses with subsets of SNPs from an interacting system, we hypothesized that the interactions among the pairs of analyzed rSNPs influenced the improved ranking performance of the random forests over the univariate tests. To confirm this, we used the H8M4 genetic model and analyzed the data in the following manner. For a constant number of analyzed rSNPs included in the model (K = 4, 8, or 16) and a constant 96 nSNPs, we looked at the effect of increasing S, the number of rSNPs from each interaction system that were genotyped. Thus, for K8S1, along with 96 nSNPs, one SNP from each of the first 8 systems was included in the analysis, while for K8S4, all four SNPs in the first two systems were included in the analysis. For K8S3, three SNPs from the first two systems, and one from the third were included. Assuming that the random forest analysis was taking advantage of the interactions among the rSNPs, and that this was responsible for the improved performance of the random forests over the univariate tests, we expected the Fisher p-values and random forest importance Z T to perform similarly when only a single rSNP was genotyped from each system, and the random forests to perform increasingly better than the univariate Fisher tests as S increased from 1 to 4. Figures 3 and 4 show the results, which are consistent with this hypothesis. For the Fisher p-values, the proportion of replicates for which the N most significant SNPs were rSNPs is similar for each S. For the random forest importance Z T , the S = 1 analyses for each K were similar to the Fisher results, while for each increase in S, the proportion of replicates for which the N most significant SNPs were rSNPs increases (Figure 3 ) and the proportion of replicates for which all rSNPs are present at any cutoff point increases (Figure 4 ). The differences can be substantial: for the H8M4 model, with K = 4 rSNPs in the analysis, the number of most significant SNPs required to have 80% probability that all four rSNPs are included is 50, 34, 22, and 5, respectively for S1, S2, S2, and S4. We conclude that for a given number of rSNPs within a set of potential predictors, the more interacting SNPs there are, and the larger the groups of SNPs that interact, the greater the performance increase of the random forest analysis as compared to a univariate analysis. Figure 3 Proportion of replicates for which the most significant N SNPs are all rSNPs. H8M4 genetic model. Analysis designs include 96 noise SNPs; K and S are listed on the plots. Other notation as in Figure 1. Figure 4 Proportion of replicates for which all rSNPs are among the top-ranking N SNPs for H8M4 genetic model. Analysis designs include 96 noise SNPs; K and S are listed on the plots. Magnitude of difference Beyond simply ranking SNPs, we may wish to use the magnitude of the difference in importance or p-value to determine which subset of top-ranked SNPs should be prioritized for further study. Thus, particularly for the cases where the rSNPs are among the top-ranked SNPs, we want to determine not just that Z T ranks interacting rSNPs higher than the univariate test, but also that the differences in rank correspond to differences in magnitude of Z T that are meaningful. In other words, we want to know how much "better" in terms of Z T (or p-value) the rSNPs are than the nSNPs. Toward this goal, we computed the difference between the importance Z T of the top ranked rSNP and the top ranked nSNP: D max (Z T ) = max rSNP ( Z T ) - max nSNP ( Z T ), as well as the lowest ranked rSNP and the top ranked nSNP: D min (Z T ) = min rSNP ( Z T ) - max nSNP ( Z T ). Thus, D min (Z T ) is positive when the lowest ranked rSNP is larger than the highest ranked nSNP, and negative when the lowest ranked rSNP is smaller than the highest ranked nSNP. We computed the analogous quantities, D max (-log p) and D min (-log p), for the -log 10 transformed Fisher Exact test p-values. In Figure 5 , we have plotted box plots of these differences for several models using analysis designs K4S2N100 and K4S2N1000. P-values for a paired T-test of whether the mean difference is equal to 0 are also placed on the plot. For H8M2, the lowest ranking rSNP has Z T that is significantly higher than the highest ranking nSNP for both N100 and N1000, while the difference in -log10(p) is not significantly different from 0 for N100, and is significantly less than 0 for N1000. These plots illustrate that the positive differences are typically more extreme for Z T than for -log p, and that the negative differences are less extreme. Figure 5 Distribution of difference in importance ZT between the top ranked rSNP and the top ranked nSNP (Dmax(ZT), and lowest ranked rSNP (Dmin(ZT)) and top ranked nSNP. Dmax(-log p) and Dmin(-log p): differences using -log10 p-value from the Fisher exact test. Beside each boxplot is the p-value for the test of whether the mean difference over the 100 replicates is significantly different from 0. Genetic models listed in plot. Analysis design: K4S2, with N100 and N1000 shown on plot. Discussion A key advantage of the random forest approach is that the investigator does not have to propose a model of any kind. This is important in an initial genome-wide or candidate region association study, where little is known about the genetic architecture of the trait. If interactions among SNPs exist, they will be exploited within the trees, and the variable importance scores will reflect the interactions. Therefore, we expect that when unknown interactions between true risk SNPs exist, the random forest approach to screening large numbers of SNPs will outperform a univariate ranking method in finding the risk SNPs among a large number of irrelevant SNPs. Our genetic models for simulation feature both multiplicative interaction and genetic heterogeneity. The multiplicative interaction results in a marginal effect in the population, the size of which is dependent on many factors, including the amount of heterogeneity. Thus, we have highlighted models in which univariate tests still have power, and shown that the random forest analysis can outperform these tests for selecting subsets of SNPs for further study. For models with genetic heterogeneity and interactions resulting in no main effect, similar to the models described by Ritchie et al. [ 10 ], the performance of random forests compares considerably more favorably to univariate tests (data not shown), since the univariate tests have no power when main effects are absent. Further investigation of how to determine a cutoff for SNPs to keep for further analysis is needed. Unfortunately, this task is likely to be strongly dependent on information that is impossible for an investigator to know a priori , such as the underlying genetic model and the ratio of associated risk SNPs to noise SNPs in an analysis. Our results from analyses with four risk SNPs among 1000 SNPs suggest that even when a high proportion of the analyzed SNPs are unassociated, a random forest can rank interacting SNPs considerably higher than a univariate test, and that the proportional difference in importance between the risk SNPs and the best of the noise SNPs can be larger on average for a random forest. In our scale-up from 100 to 1000 total SNPs, we kept the number of risk SNPs constant. In practice, as we increase the number of SNPs genotyped, we expect that we will also increase the number of risk SNPs (or SNPs in linkage disequilibrium with risk SNPs) that are captured in an analysis. Thus, as a larger and large proportion of the genome or candidate region is captured by a scan, the more likely we will be to have all or most of sets of SNPs that interact, and thus the more likely we are to be in situations where random forest screening will outperform univariate screening of SNP data. It is important to consider the tuning parameters for such analyses. Consistent with the recommendations made by Breiman and Cutler [ 22 , 29 ], the number of variables randomly selected at each split seems to have minimal effect over a wide range of values surrounding the square root of the number of covariates (SNPs). Breiman and Cutler do not recommend a method to determine the number of trees necessary for an analysis. The documentation examples typically use on the order of 100–1000 trees, but these examples are primarily in the context of prediction, without computing estimates of variable importance. In our experience with the simulated data sets presented here, in which the truly associated covariates are outnumbered considerably by those that are noise, multiple thousands of trees must be used in order to get stable estimates of the variable importance. In practice, we recommend building several forests for a data set with a given number of trees. If the ranking of variables by importance does not change significantly from forest to forest, then the number of trees is adequate. We have examined the use of random forests in the context of association studies for complex disease with uncorrelated SNP predictors. Random forests can also be used when predictors are correlated, as is the case with multiple SNPs in linkage disequilibrium within a small genetic region. For any analysis procedure, the more highly correlated variables are, the more they can serve as surrogates for each other, weakening the evidence for association for any one of the correlated variables to the outcome. In a random forest analysis, limited simulations suggest that correlated variables lead to diminished variable importance for each correlated risk SNP (data not shown). One way to limit the problems presented by SNPs in linkage disequilibrium is to use haplotypes instead of SNPs as predictor variables in a random forest. Future challenges include quantifying more completely the effect of linkage disequilibrium among SNPs submitted to a random forest analysis, and developing random forests in the context of haplotypes. Conclusions With the increasing size of association studies, two-stage analyses, in which in the first stage a subset of the loci are retained for further analyses, are becoming more common. The most frequently voiced concern for these analyses is that variables that interact to increase disease risk but have minimal main effects in the population will be missed. Random forest analyses address this concern by presenting a summary importance of each SNP that takes into account its interactions with other SNPs. Current implementations of random forests can accommodate up to one thousand of SNPs in one analysis with the computation of importance. Further, there is no reason to restrict the input variables to SNPs. Potential environmental covariates can also be easily accommodated, allowing for SNPs with no strong main effect, but environmental interactions, to be distinguished from unassociated SNPs. We have shown that when unknown interactions among SNPs exist in a data set consisting of hundreds to thousands of SNPs, random forest analysis can be significantly more efficient than standard univariate screening methods in ranking the true disease-associated SNPs highly. After identifying the top-ranked SNPs and other variables, and weeding out those unlikely to be associated with the phenotype, more thorough statistical analyses, including model building procedures, can be performed. Methods Variable importance Rather than selecting variables for modeling, a random forest uses all available covariates to predict response. Here, we use measures of variable importance to determine which covariates (SNPs, in our case) or sets of covariates are important in the prediction. Breiman [ 22 ] proposed to quantify the importance of a predictor variable by disrupting the dependence between the variable and the response and measuring the change in the tree votes compared to the original observations. In practice, this is achieved by permuting the variable values among all observations in the out-of-bag sample of each tree. If the variable is predictive of the response, it will be present in a large proportion of trees and be near the root of those trees. Observations with a changed variable value may be directed to the wrong side of the tree, leading to vote changes from the right to the wrong class. Conversely, if the variable is not related to the response, it will be present in few trees and, when present, it will be near a terminal node, so that few tree votes will be changed. In Random Forests (version 5) Breiman and Cutler [ 29 ] define the importance index as follows. For individual i, let X i represent the vector of predictor variable values, y i its true class, V j ( X i ) the vote of tree j and t ij an indicator taking value 1 when individual i is out-of-bag for tree j and 0 otherwise. Let X (A,j) = ( X 1 (A,j) ,..., X N (A,j) ) represent the vector of predictor variables with the value of variable A randomly permuted among the out-of-bag observations for tree j, and X (A) the collection of X (A,j) for all trees where N is the total number of individuals in the sample. Letting 1(C) denote the indicator function taking value 1 when the condition C is true and 0 otherwise, the importance index averages over the trees of the forest, and is defined as: where N j represents the number of out-of-bag individuals for tree j and T is the total number of trees. The importance index can be standardized by dividing by a standard error derived from the between-tree variance of the raw index I T , . The standardized index is defined as: The variance represents the tree to tree variance of I T , rather than the variance of I T due to the sampling of the individuals from a population: the magnitude of Z T increases with the number of trees in the forest, and the number of trees is limited only by computing time. Thus, this standardized index cannot be treated as a Z-score in the traditional sense. Simulation models and methods For simplicity, assume each locus has the same effect, and let (q 0 , q 1 , q 2 ) represent the penetrance factors for 0, 1, or 2 risk alleles for an individual locus in a given model. Let G = {g 11 , g 12 , . ., g HM } be the multilocus genotype for an individual, where g hm (=0, 1, 2 risk alleles) indicates the individual's genotype at locus m (=1, . ., M) of heterogeneity system h (=1, . ., H). Then the penetrance for genotype G is defined as: For example, for model H2M2, an individual with genotype G = {0101} would have penetrance P G = 1 - (1 - q 0 q 1 ) 2 = 2 q 0 q 1 - ( q 0 q 1 ) 2 . The penetrance factors (q 0 , q 1 , q 2 ) and risk allele frequencies, as well as other features of our genetic models, are listed in Table 1 . For a given model type, such as H4M4, and a given λ s and K p , there is a unique allele frequency when we make the assumption that each SNP subunit has equal effect (the given penetrance factor vector) in the population. We chose penetrance factors such that the risk alleles at each locus for the H2M2, H4M2, H8M2, and H16M2 models are approximately additive in effect on the penetrance factor scale. For H4M4 and H8M4, we chose penetrance factor vectors such that the risk alleles show a moderate degree of dominance. The marginal genotype relative risks (GRRs) listed in Table 1 are the relative penetrances for heterozygote and homozygote carriers of each risk allele, as compared to non-carriers in the population, which would be observed if only a single rSNP were considered at a time. Thus, this is a measure of the observed effect size of each of the rSNPs in the population. The marginal GRRs are modest, in line with what might be expected when there are a large number of small effects contributing to a complex phenotype. For cases, the genotypes for pairs/quartets of SNPs within an interacting system are positively correlated, while SNPs from distinct systems are negatively correlated. The magnitude of the correlations decreases with increasing number of heterogeneity systems and increasing number of equal-effect SNPs interacting within each system. Analysis All analyses were performed on 100 replicate data sets of 500 cases and 500 controls. We treated the SNPs as ordinal predictors. Random forests have one primary tuning parameter: "mtry" the number of randomly picked covariates to choose among for each split. The Random Forest v5 manual [ 29 ] recommends trying the square root of the number of predictors, along with values smaller and larger than the square root, and choosing the value that minimizes the out of bag prediction error rate. We considered both the prediction error and the stability of the variable importance estimates when determining the values of mtry to use and the number of trees to grow. We found that the prediction error rate was very stable over a wide range of mtry for the number of trees we required for consistent measures of importance. We analyzed each replicate data set with 4–16 rSNPs and 96 nSNPs using a random forest of 5000 trees, choosing the best split from among a different randomly-selected set of 35 SNPs at each node (mtry = 35). On average, each replicate data set with 100 total SNPs took 40 minutes to complete on a 2.6 Ghz Intel Xeon processor. For data sets with 4 rSNPs and 996 nSNPs (1000 SNPs total), we used 15000 trees, and chose from among 125 SNPs at each node (mtry = 125). Analysis of each replicate of these data sets took 123 minutes on average. User time could potentially be substantially decreased by parallel-processing: trees could be grown on separate nodes, and combined for analysis of importance. However, parallel tree-building is not yet available in the Random Forest progream. To compare the performance of random forests with that of a univariate, one-SNP-at-a-time approach, we tested for association between genotypes for each individual SNP and disease status using a Fisher Exact test [ 32 ]. Ranking of rSNPs The random forest analysis produces the raw and standardized importance indices ( I T , Z T ,), which can be used to rank order the importance of SNPs much as p-values from association tests are used. Using either method, the ranking of the SNPs in an analysis can be used to prioritize which sets of SNPs (or genome regions) should be followed up with further genotyping and/or additional analyses. We use the convention that a rank of 1 is the highest ranking SNP. We compare the ranking of the raw and standardized importance measures, and further compare these with the rank based on p-values from a test of association, the Fisher Exact test, to determine whether the random forest can better discriminate susceptibility SNPs from SNPs unrelated to disease status when there is interaction and heterogeneity among and between SNPs. Authors' contributions KLL conceived of and led the design of the study, coordinated all phases of simulation and analysis, and drafted the manuscript. LBH and JS automated the simulation and random forest analysis procedures, and participated in the design and analysis of study results. PVE added critical insight to the development of the genetic models and participated in the interpretation of study results. All authors provided comments on a draft manuscript and read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC545646.xml |
532394 | Computer Simulation of the Movement of Listeria Bacteria | null | Each of the cells in our body is a complex machine. Within each cell, thousands of proteins and other molecules interact to produce the highly organized cellular events that are needed for life. However, there is no cellular line manager telling the individual proteins where to go and what to do. It's rather like an ant colony. No one tells the individual ants how to build a nest. They just do it. Similarly, no one tells the individual actin molecules, for example, to get together and polymerize to form the cytoskeleton that is essential for cell movement and other cellular processes. It just happens. Ant colonies and actin polymerization are both examples of emergent behavior. Broadly defined, emergent behavior is when a collection of individuals interact without central control to produce results that are not explicitly programmed. Scientists are only just beginning to understand how complex behaviors of this type can emerge from a myriad of individual interactions, many of which are well-defined experimentally. Jonathan Alberts and Garrett Odell have turned to an unlikely ally—the bacterium Listeria monocytogenes —in their study of the complexities of cellular actin polymerization. This rod-shaped microbe, which lurks in well-ripened brie and other unpasteurized cheeses, can cause the sometimes fatal disease listeriosis, particularly in young babies or people with weakened immune systems. When Listeria microbes in food reach a person's gut, they penetrate the cells lining the gut, reproduce, and then spread from cell to cell without ever exposing themselves to the extracellular environment. In this way, they avoid the host's immune system. One particular protein produced by Listeria is central to this sneaky intracellular lifestyle: ActA. Expression of ActA allows the microbes to hijack the machinery in the host cell that controls the growth of actin networks. Some of the cellular actin, instead of forming the cytoskeleton of the cell, polymerizes around the bacterium, forming a dense “comet tail” of actin that generates a force to move the bacterium around the cell and push it into neighboring cells. This clandestine use of our cellular machinery for actin polymerization is far simpler to understand than the elaborate use our cells normally make of it, so the study of Listeria propulsion provides a scientific stepping-stone to understanding the involvement of the actin cytoskeleton in normal cell movements. Alberts and Odell have used established data on the biochemical and mechanical details of actin polymerization and the physiological concentrations of the proteins involved in the process to build a computer simulation of how an actin network assembles around a moving, rod-shaped bacterium. Their “in silico” reconstitution, which considers the behavior of individual actin filaments and requires a cluster of 80 computer processors to be run for several days at a time, produces realistic bacterial motion in terms of speed and persistence, and models the actin tail shape. The model also reproduces smaller scale emergent behavior. Real Listeria cells do not move smoothly. Instead, the microbes move jerkily, with runs interspersed with pauses. The simulation faithfully reproduces this type of movement and offers a mechanistic explanation for it. The approach described by Alberts and Odell can now be used to investigate many other, more complex cell mechanics problems, such as how the cellular movements involved in cell division are achieved. With the availability of detailed biological information and powerful computers, we may at last start to solve the mystery of how interactions among maybe 100,000 gene products can produce the organized cellular processes that cell biologists have been watching under the microscope for years. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC532394.xml |
552306 | Tai Chi and vestibular rehabilitation improve vestibulopathic gait via different neuromuscular mechanisms: Preliminary report | Background Vestibular rehabilitation (VR) is a well-accepted exercise program intended to remedy balance impairment caused by damage to the peripheral vestibular system. Alternative therapies, such as Tai Chi (TC), have recently gained popularity as a treatment for balance impairment. Although VR and TC can benefit people with vestibulopathy, the degree to which gait improvements may be related to neuromuscular adaptations of the lower extremities for the two different therapies are unknown. Methods We examined the relationship between lower extremity neuromuscular function and trunk control in 36 older adults with vestibulopathy, randomized to 10 weeks of either VR or TC exercise. Time-distance measures (gait speed, step length, stance duration and step width), lower extremity sagittal plane mechanical energy expenditures (MEE), and trunk sagittal and frontal plane kinematics (peak and range of linear and angular velocity), were measured. Results Although gait time-distance measures were improved in both groups following treatment, no significant between-groups differences were observed for the MEE and trunk kinematic measures. Significant within groups changes, however, were observed. The TC group significantly increased ankle MEE contribution and decreased hip MEE contribution to total leg MEE, while no significant changes were found within the VR group. The TC group exhibited a positive relationship between change in leg MEE and change in trunk velocity peak and range, while the VR group exhibited a negative relationship. Conclusion Gait function improved in both groups consistent with expectations of the interventions. Differences in each group's response to therapy appear to suggest that improved gait function may be due to different neuromuscular adaptations resulting from the different interventions. The TC group's improvements were associated with reorganized lower extremity neuromuscular patterns, which appear to promote a faster gait and reduced excessive hip compensation. The VR group's improvements, however, were not the result of lower extremity neuromuscular pattern changes. Lower-extremity MEE increases corresponded to attenuated forward trunk linear and angular movement in the VR group, suggesting better control of upper body motion to minimize loss of balance. These data support a growing body of evidence that Tai Chi may be a valuable complementary treatment for vestibular disorders. | Background Vestibulopathy decreases whole body dynamic postural control and causes functional limitations [ 1 - 4 ]. Limitations in ambulation, dynamic balance and trunk control, for example, can lead to disability and contribute to decreased quality of life [ 5 ]. Vestibular rehabilitation (VR) is a well-accepted exercise program intended to remedy balance impairment caused by damage to the peripheral vestibular system [ 6 ]. Vestibulopathy impairs both the vestibulo-ocular reflex (VOR) and the vestibulo-spinal reflexes (VSR) [ 7 ]; hence, VR is designed to adapt the CNS to diminished vestibular input and to compensate for VOR and VSR loss, via gaze and balance retraining, which in turn should improve whole body dynamic stability [ 8 - 10 ]. Alternative therapies, such as Tai Chi (TC), have recently gained popularity as a treatment paradigm for a variety of human ailments, including balance impairment [ 11 - 13 ]. TC employs detailed regimens of physical movement, breathing techniques, and cognitive tools to strengthen the body, relax the mind, and balance the flow of life force [ 14 ]. The purported improvements in overall body control and mind-body focus with TC may offer an improved approach to treating balance dysfunction [ 11 , 13 , 15 - 17 ]. Where the explicit objective of many VR exercises is to improve gaze stability, TC emphasizes a 'soft' unfocussed gaze during the prescribed balance exercises. Although there is strong evidence that VR [ 4 , 8 - 10 , 18 ], and more recently TC [ 11 , 13 , 15 - 17 ], can benefit people with vestibulopathy, the degree to which gait improvements may be related to neuromuscular adaptations of the lower extremities are unknown. Thus, although the end result of both TC and VR should be improved dynamic stability during locomotor activities of daily living, including gait, we hypothesized that the mechanisms underlying such improvements should differ substantially. A better understanding of how balance dysfunction interventions affect lower extremity neuromuscular function during ADL may be useful for developing gait training exercises, and for providing a fuller understanding of the link between motor function and balance. In this report, we present preliminary data from a blinded randomized clinical trial comparing the effects of VR and TC on gait function, joint kinetics and trunk kinematics in older adults. While the overall aim of this study was to determine the effects of balance rehabilitation on gait characteristics, we directed our efforts in this paper to better understand the relationship between mechanical energy transfers along the lower extremity kinematic chain (ankle-knee-hip), and forward and side-to-side velocity of the trunk. Our general hypothesis was that adults with balance impairment from vestibulopathy who receive the VR or TC intervention will improve gait function as indicated by time-distance measures. Our specific hypotheses are based on the following rationale: Recent studies in healthy older adults [ 19 - 21 ], with general impairments such as strength loss [ 22 ], or pathologies such as knee arthritis [ 23 , 24 ], show that the hip musculature often aids, or compensates for, ankle plantar-flexor muscles in providing both forward propulsion and trunk stability [ 25 ]. These prior studies have shown a consistent decline in plantar-flexor muscle power during gait, with an increase in hip muscle power, in older adults with, and without, known mobility impairments. As shown recently by Neptune and colleagues [ 26 ], the ankle plantar flexors contribute significantly to both forward propulsion and vertical trunk stability. Thus, one would expect that improvements in lower extremity motor control, aimed at increasing forward propulsion and trunk stabilization, would be represented by decreases in hip mechanical energy expenditures and increases in ankle (and perhaps knee) mechanical energy expenditures, and be directly related to improved kinematics of the trunk. Based on the above rationale, and the specific treatment programs described in the following sections, we hypothesized: 1) TC treatment will improve lower extremity motor control by increasing ankle mechanical energy expenditure (MEE) contribution, and decreasing hip MEE contributions, to total energy of the leg, more than VR; and 2) that improved trunk control following TC will be positively correlated with improvements in lower extremity motor control, while improvements in trunk control following VR will not be correlated with improvements in lower extremity motor control. The latter may indirectly implicate other mechanisms, most likely improvements in VOR/gaze stability [ 27 ]. Methods Subjects Fifty-three patients with balance impairment due to vestibular hypofunction were recruited and randomized into two treatment groups: VR, a group vestibular rehabilitation intervention, and TC, a group Tai Chi exercise. Of the 53 patients admitted, 15 dropped out or were excluded prior to completing the intervention. The majority of drop outs were due to a new medical condition unrelated to the study preventing participation (e.g. fractured foot, acute back pain) or due to the sudden need to care for an ill family member. Two subjects were eliminated because of lack of force plate data to use in the data analysis (see Gait Analysis section for more detail). Of the 36 subjects remaining for analysis, 17 subjects were randomized into the VR treatment group (12 unilateral and 5 bilateral) and 19 subjects in the TC group (11 unilateral, 8 bilateral). Unilateral or bilateral vestibular hypofunction (UVH or BVH) diagnoses were obtained as previously described [ 10 ]. Briefly, all patients had gait unsteadiness without evidence of central nervous dysfunction. All patients were referred to the study because they had locomotor instability for which they sought treatment from project physicians. Patients with bilateral vestibular hypofunction had bilaterally decreased caloric responses (total slow phase velocity of ≤10 degrees·sec -1 for the sum of right and left ear caloric stimulation at 27 and 44°C warm water stimulation of both ears and ≤8°·sec -1 slow phase velocity for the sum of 35 cc of ice water stimulation in each ear) and decreased VOR gains during passive rotational testing at up to 50°·sec -1 (at least 2.5 SD below normal mean values at frequencies of rotational testing from .01 to 0.5 Hz). Patients with unilateral vestibular hypofunction had damage only on one side, including at least 30% unilaterally reduced caloric response, positional nystagmus while lying with the damaged ear down, and/or confirmatory abnormalities on rotational testing (mildly decreased low frequency gains, increased phase leads and asymmetrical rotation induced nystagmus, i.e., decreased vestibular time constant). Patients with bilateral deficits are typically more disabled than are those with unilateral deficits. The average time post-onset of vestibulopathy for the 36 subjects included in the analysis was 3.05 years (range 0.58 – 12 years). All subjects were community dwelling and reported varying degrees of limitations in locomotor activity. Twenty of the 36 subjects were female, and the 36 subjects were 59.5 ± 11.5 years old (range, 41–81), 1.70 ± .11 m tall and 83.6 ± 16.5 kg in weight (breakdown by treatment group is shown in the Results section). All subjects had at least one course of VR since the time of onset of their vestibular symptoms. Inclusion criteria required that each subject did not have VR for >6 months from study enrollment. The testing protocol was approved by MGH institutional review board, and all subjects provided written informed consent according to institutional guidelines on human research. Interventions The VR and TC treatment interventions were provided in a total of six (3 VR and 3 TC groups) small groups with an average of 8 subjects per group. Each intervention program met once weekly, on separate weekdays, for 10 weeks in the same exercise room. The weekly sessions for both intervention groups lasted approximately 70 minutes. Each treatment program was lead by the same instructor for the three treatment cohorts. The instructors were blinded to the exercises provided to the other treatment program. One or two assistants were available for each session for all treatment groups to insure participants' safety. All treatment sessions included time to: 1) review material introduced in prior sessions; 2) introduce new material; 3) ask questions and share personal experiences or concerns regarding the practices; and 4) cool down and rest. Tai Chi intervention The TC intervention incorporated three objectives outlined in a balance-related TC program developed by Wolf and colleagues [ 12 , 17 ]. First, it emphasized movements that are easily comprehensible. Second, the sequence of exercises introduced reflected a progression that increasingly challenges postural stability, with a shift in weight bearing from bilateral to unilateral support. Third, the program emphasized increasing the magnitude of trunk and arm rotation while diminishing the base of support. The five specific TC movements employed in this study – 'raising the power', 'withdraw and push', 'wave hand like clouds', 'brush knee twist step', and 'separate right and left legs' – are described and illustrated in a training manual for the Cheng Man-Ch'ing's Yang-style short form [ 28 ]. In addition to these five formal TC movements, the intervention also included a short set of traditional TC warm-up exercises focused on loosening up the physical body and incorporating mindfulness and imagery into movement. Warm-up exercises included: gentle stretches sequentially targeting the shoulders, necks, arms and legs; a torso stretching exercise that coordinated weight shifts with rotations of the trunk and passive arm swinging; and a 5 minute seated meditation emphasizing relaxed diaphragmatic breathing. Approximately 20 minutes of each class was devoted to warm-up exercises, of which 10–12 minutes was spent in standing. Following an additional 40 minutes of formal TC practice, 10 minutes was allowed for group discussion. Vestibular rehabilitation intervention The VR intervention used in this study was a comprehensive exercise program designed to improve the problems specifically associated with damage to the peripheral vestibular system [ 4 , 6 , 10 , 29 ]. Each treatment session focused on the three main objectives of the VR intervention. Firstly, a series of eye-head coordination exercises were performed to promote gaze stability during both quiet standing and dynamic functional activities (such as combining movement of an image across the retina with head movement). Subjects progressed to performing these eye-head exercises with the target on a more complex background (to simulate real world activities), at increasingly faster speeds of head movements (eg, 2–3 Hz), and during more dynamic standing and locomotor activities. A second treatment objective included VOR training in a group format with subjects standing. Target foveation objects (words of various sizes) were fixed to a large checkerboard background covering one wall of the exercise room, enabling us to provide the appropriate visual stimuli. Patients were progressed by increasing the speed (frequency and amplitude) of head movement to train the VOR more appropriately at speeds consistent with everyday locomotor activities. The third main component of the VR program was upright balance retraining exercises that enhance the use of various sensory cues for gaining posture control [ 8 , 29 , 30 ]. Examples of these exercises include subjects maintaining their balance while decreasing their base of support (such as standing on one foot, marching or walking heel to toe) and while walking on various floor surfaces (such as the pliable surfaces of a foam mat). Subjects were further challenged by incorporating head and trunk movements or with eyes closed during standing and walking exercises. All exercises were performed in an upright position (either standing or during locomotion) based on individual tolerance. If required, a seat was provided and the exercise performed in a seated position until the subject was able to tolerate the activity in standing. Each group treatment session lasted 60 minutes allowing 20 minutes for each of the 3 main exercise components. There was an additional 10 minutes for questions and answer time and for assistance with individual progression of home exercise programs. Gait analysis Subjects performed two-to-four gait trials along a 10 m level walkway at baseline testing, and at testing following the intervention program, at their freely selected pace upon the command "Please walk as you normally would, as if taking a brisk walk in the park". Body segment kinematics were acquired at 150 Hz with a four-camera Selspot optoelectric tracking system (Selective Electronics, Partille, Sweden), used to acquire position and orientation data of 11 segments (both feet, shanks, thighs and arms, and the pelvis, trunk and head). Collection of kinematic data and processing is described in more detail elsewhere [ 31 ]. Kinetic data consisted of ground reaction forces acquired from two adjacent piezoelectric force platforms (Kistler Instruments, Winterthur, Switzerland), synchronously sampled with body segment kinematic data at 150 Hz. Subjects were required to have at least one good gait trial, both at baseline and post-intervention testing sessions, to be included in the data analysis. A "good gait trial" was one that satisfied the following criteria: 1) one foot was required to be in whole contact with only one or both force platforms without interference from the other foot, 2) all body segments were visible, and tracked without artifact, during the stance portion of gait. Two subjects were excluded on the basis of failing one or both of the above criteria. Data analysis Parameters selected for data analysis consisted of dynamic gait function (time-distance measures), lower extremity neuromuscular control (sagittal plane mechanical energy expenditures, MEE), and trunk stability (sagittal and frontal plane kinematics). Dynamic gait function Gait function was assessed with standard time-distance measures [ 4 , 32 ], including: gait speed, step length, step width and stance duration. Gait speed was measured as the average anterior-posterior velocity component of the whole-body center of gravity over stance phase of gait. Step length was measured as the anterior-posterior distance between right and left ankle centers when each foot was flat on the floor during its respective mid stance portion. Stance duration was measured as the time elapsed between heel strike and toe off (duration of stance phase), and step width was measured as the lateral distance between ankle centers at the foot positions used for step length calculation. Lower extremity neuromuscular function Neuromuscular function of the lower extremities was assessed using mechanical energy expenditure (MEE) of the ankle, knee and hip, relative to the total MEE of the leg, and were computed as described previously [ 23 , 33 ]. Briefly, the mechanical power profile of the joint, the scalar product of net joint moment and angular velocity, is integrated over specific time intervals to arrive at mechanical energy expended, MEE, or work done. The intervals are defined by periods of concentric transfer (MEE (+) , the amount of concentric mechanical energy expended with segment-to-segment energy transfer), eccentric transfer (MEE (-) , the amount of eccentric mechanical energy expended with segment-to-segment energy transfer), and no-transfer (MEE (o) , the amount of concentric and eccentric energy expended without segment-to-segment energy transfer) conditions. The total joint MEE is simply the sum of these components (MEE (t) = MEE (+) + MEE (-) + MEE (o) ). Leg MEE is the sum of joint MEE (ankle, knee and hip) for different conditions, or totals. Percentage contribution of joint MEE (for each condition and total) to leg MEE (for each condition and total) was then calculated. Trunk stability Trunk stability was assessed using kinematics of the trunk center of mass [ 34 ], and consisted of anterior-posterior trunk velocity (peak and range) as well as lateral trunk velocity (peak and range); sagittal plane angular (pitch) velocity of the trunk (peak and range) and frontal plane angular (roll) velocity of the trunk (peak and range). The rationale for using kinematic measures of trunk stability, instead of overall stability (such as whole-body CG sway or kinematics), however, was to enable us to examine the relationship between lower extremity neuromuscular function (using the mechanical energy analysis as such a measure) and the kinematics of the upper body, as a mass to be controlled apart from the legs. Peaks and ranges were taken from stance phase of the same leg used for the mechanical energy analysis described above. Statistical analysis One-way ANCOVA was used to compare change scores between the two groups, using the baseline values as covariates. Paired samples t-tests compared the change in each variable for each group from baseline to post-intervention testing. Pearson correlations were used to examine associations between change scores in lower extremity MEE and change scores in trunk velocities, for each treatment group. Due to the large number of comparisons in this exploratory study, a Ryan-Holm step down Bonferonni approach was used to control for type I errors [ 35 ], using a family-wise α = .05. Using this scheme, families of three members (MEE (+) , MEE (-) and MEE (o) contributions) required significance at α = .017 for at least one comparison, α = .025 for the second comparison, and α = .050 for the third comparison. Families of four members (anterior-posterior peak and range, and lateral peak and range of trunk velocity) required significance at α = .013 for at least one comparison, α = .017 for the second comparison, and so on. All p-values given will be unadjusted, but the adjusted α is given for each comparison where appropriate. SPSS for Windows (v10, SPSS Inc. Chicago, IL) was used for all statistical analyses. Results The two groups were not different in age (VR: 56.9 ± 11.6 yrs; TC: 61.7 ± 11.3 yrs; p = .223), height (VR: 1.69 ± .11 m; TC: 1.71 ± .11 m; p = .712) or weight (VR: 81.1 ± 19.3 kg; TC: 85.8 ± 13.6 kg; p = .399). There was no significant difference in proportion of UVH and BVH in the treatment groups (Chi-square, p = .429), or proportion of men and women in the treatment groups (Chi-square, p = .709). There were no significant between-groups differences (using ANCOVA for controlling for baseline differences) for any of the variables examined. There were, however, significant changes pre- and post-treatment within each group. These latter results appear to suggest that clinically important differences in each group's response to the therapies exist. Thus, the remainder of the results presented will focus on the within-groups comparisons. Time distance measures Both groups improved (unadjusted α = .05) following intervention in time-distance measures (see Table 1 ), with the TC group showing greater overall improvements; the VR group improved significantly in stance duration ( p = .044) and step length ( p = .045), but not in gait speed ( p = .060) or step width ( p = .390); the TC group improved in gait speed ( p = .009) and step length ( p = .010), but not in stance duration ( p = .055) or step width ( p = .313). Table 1 Time distance measures before and after intervention. Variable Baseline Value Post-Treatment Value Mean Standard Dev Mean Standard Dev p -value* VR Gait speed (m/s) 1.180 .312 1.235 .229 .060 Step length (m) .616 .119 .639 .116 .045 Stance duration (s) .667 .065 .653 .047 .044 Step width (m) .093 .045 .096 .040 .390 TC Gait speed (m/s) 1.090 .275 1.170 .261 .009 Step length (m) .582 .110 .612 .118 .010 Stance duration (s) .715 .089 .684 .055 .055 Step width (m) .109 .046 .114 .042 .313 * Within-groups paired t-test significance, based on unadjusted α = .05. Mechanical energy expenditures Figure 1 shows the changes in joint and leg MEE (t) for each joint, and the sum of all the joints (leg). Although the total leg MEE (t) change was similar, the distribution of joint MEE (t) were quite different for the two treatment groups. Comparison of the change in percent contribution of MEE for each transfer condition for each joint to leg MEE showed that only the TC group had significantly reduced ( p < .001, adjusted α = .017) relative hip concentric MEE (+) and increased ( p = .019, adjusted α = .025) relative ankle concentric MEE (+) , following training. These data are shown in Figure 2 . Figure 1 Change scores in ankle, knee and hip total MEE (t) and leg total MEE (t) for VR and TC groups (in J %BW). Error bars represent 95% confidence intervals on the mean. Figure 2 Change scores in percent contribution of ankle, knee and hip concentric MEE (+) to leg concentric MEE (+) for VR and TC groups (in J %BW). Error bars represent 95% confidence intervals on the mean. Trunk kinematics TC group had significantly increased ( p = .009, adjusted α = .013) peak trunk forward velocity during stance phase of gait following treatment, while the VR group's increase was similar though not statistically significant ( p = .018, adjusted α = .013). There were no significant changes in forward velocity range, nor were there significant changes in peak or range of lateral trunk velocity for either group. The VR group, however, did show a significant increase in peak trunk angular velocity ( p = .007, adjusted α = .017) and range of trunk angular velocity ( p < .001, adjusted α = .013) in the frontal plane. There were no significant changes in trunk angular velocity in the frontal plane for the TC group, and neither group showed significant changes in peak and range of trunk angular velocity in the sagittal plane. These data are summarized in Figure 3 . Figure 3 Change scores in trunk velocity. (a) Linear velocity: anterior-posterior (A/P) velocity peak and range, and medial-lateral (M/L) velocity peak and range; (b) Angular velocity: pitch (sagittal plane) velocity peak and range, roll (frontal plane) velocity peak and range. Error bars represent 95% confidence intervals on the mean. Relationships between MEE and trunk kinematics Correlation analysis between changes scores in leg MEE and trunk kinematics revealed significant relationships for both treatment groups. Most striking was the consistent directional relationship between trunk velocity and leg MEE within each of the treatment groups. For the VR group, changes in range and peak of forward velocity of the trunk was negatively correlated with changes leg MEE (range: r = -.536, p = .013, adjusted α = .013; peak: r = -.431, p = .042, adjusted α = .017). For the TC group, however, changes in range and peak of forward velocity of the trunk was positively correlated with changes in leg MEE (range: r = .620, p = .003, adjusted α = .013; peak: r = .451, p = .026, adjusted α = .017). Figure 4 shows scatter plots depicting the positive and negative relationships between change scores in leg MEE and trunk velocity range for TC and VR groups, respectively. There were no significant relationships detected between change in joint or leg MEE and change in lateral linear velocity of the trunk, nor in sagittal or frontal plane angular velocity of the trunk. Figure 4 Change scores in trunk forward velocity range (in m/s) versus change scores in total leg MEE (t) (in J %BW) for VR (top plot) and TC (bottom plot) groups. Dashed lines represent the 95% confidence intervals on the mean. Discussion and conclusions Little is known about the mechanisms of improved balance and postural control following rehabilitation in people with vestibulopathy. Although VR has shown promise for improving patients balance and gaze stability [ 4 , 9 , 10 , 29 , 36 - 40 ], just over 65% of people treated respond to the therapy [ 10 ]. Improvements in function are not ubiquitous with VR treatment. Alternative therapies, such as TC, offer a complementary approach to improving balance and postural control by teaching body control and awareness [ 12 , 15 ]. The purpose of the present study was to examine lower extremity neuromuscular function during gait in patients receiving either VR or TC treatment, and to examine how changes in neuromuscular patterns relate to changes in trunk control. Our first hypothesis, that patients receiving TC treatment will improve lower extremity motor control by increasing ankle MEE contribution and decreasing hip MEE contribution more so than the VR patients was not supported by analysis of between-group differences. However, in examining the within-groups changes (pre versus post-intervention), some potentially important biomechanical observations were made. Our second hypothesis, that trunk control in patients receiving TC will be positively correlated with improvements in lower extremity motor control, but trunk control in those receiving VR will not be positively correlated with improvements in lower extremity motor control, was supported. Although an overall improvement in gait function (as indicated by time-distance measures) for both treatment groups was observed, confirming our general hypothesis, our data suggest that the mechanisms underlying those improvements differ, and appear to be linked to differences in neuromuscular responses of the lower extremities to the treatment programs. Specifically, our data suggest that changes in the relative contribution of individual joints to total leg mechanical energy expenditure (MEE), and the relationship between changes in lower extremity mechanical energy expenditure and changes in upper body kinematics, are different between TC and VR interventions (Figure 4 ). Further, these data highlight the importance of assessing gait not only with time-distance functional gait measures (Table 1 ), but also with measures that assess neuromuscular function of the lower extremities and control of the body's most massive segment, the trunk. We found that TC patients significantly increased the contribution of ankle MEE to total leg MEE and decreased contribution of hip MEE to total leg MEE following the treatment program, while the VR group showed no significant change in ankle or hip MEE contributions following intervention (Figure 2 ). Although the changes were not statistically different between groups, the within groups comparisons suggest that clinically important trends may nonetheless be present in terms of biomechanical responses to the different therapies. Figure 1 shows the total joint MEE (t) (sum of all transfer components) for ankle, knee and hip, and total leg MEE (t) , for each group. The increases in leg MEE (t) , which were similar for both groups, were apparently achieved by different neuromuscular adaptations of the individual joints. Where the VR group appear to increase total MEE (t) of each joint to gain a total MEE increase of the leg, the TC group show a distinctive pattern of substituting ankle plantar-flexor contribution for hip extensor/flexor contribution. Prior studies on the relative roles of ankle and hip kinetics in gait [ 19 , 20 , 24 ] suggest that the result observed for TC patients indicates a trend toward a reduction in hip compensation and increased use of ankle muscles to provide both propulsion and stability. Figure 2 shows the percent contribution of the changes to the concentric leg MEE (+) for the changes in concentric MEE (+) of individual joints. While both groups decreased the relative contribution of concentric hip MEE (+) , only the TC group increased the contribution of ankle concentric MEE (+) , while the VR group increased the contribution of knee concentric MEE (+) . Concentric energy transfer represents the energy expended by muscles in concentric contraction when energy is being transferred between segments. Because concentric contraction represents work being done by the muscles (as opposed to eccentric contraction, which is work being done on the muscles), we can interpret the above finding as meaning that, for the TC group, a greater proportion in the change in concentric work done by the leg muscles is attributed to the change in concentric work of ankle plantar-dorsiflexors, while for the VR group, this contribution decreases. One possible reason concentric ankle MEE contribution may have increased significantly in the TC group, but not the VR group, is because the TC and the warm-up exercises improved ankle flexibility. Tight ankles (limited range of motion, ROM) may preclude the optimal structural alignment to coordinate mechanical energy sufficiently to increase propulsion [ 41 ], perhaps at the expense of trunk stabilization. Ankle function is important for balance corrections in both healthy elderly and vestibulopathic subjects [ 42 - 44 ]. A study by Van Deusen et al. [ 45 ] found that Tai Chi-like exercises for elders with arthritis resulted in a significant increase in ankle plantar flexion; this finding supports the above contention that the TC group in our study may have increased ankle MEE contribution as a result of increased ankle ROM, ankle moment, or both. The tight coupling between ankle and hip power in gait [ 19 ] would also explain the neuromuscular adaptive decrease in hip MEE contribution. Given the importance of ankle-plantar flexors in both propulsion and trunk stability, we conclude that TC teaches optimization of MEE in an effort to control the trunk while improving lower extremity function. The relationship between lower extremity MEE and trunk kinematics for the two treatment groups lends further credibility to this conclusion. As shown in Figure 4 , the relationship between change in leg MEE and change in the range of forward trunk velocity was positive for the TC group, and negative for the VR group. Similar relationships were also observed between change in leg MEE and change in peak forward trunk velocity. The observed direct relationship for the TC group suggests that the redistribution of power among ankle, knee and hip joints, which resulted in a net increase in the total MEE of the leg, enabled these patients to attain a faster gait. This observation is expected based on the principles of TC, which emphasize a vertical alignment integrating the head, torso, hips and legs. This concept of integrated alignment is reflected in phrases from the TC classics such as ". suspend the spine like a necklace of pearls" and "movements are initiated in the feet, steered by the waist and administered through the hands." [ 46 ]. In contrast, for the VR group, however, the increase in leg MEE was associated with a decrease in both peak and range of trunk velocity. This finding suggests that VR subjects, when increasing power generation/absorption with their lower extremities, reduce trunk oscillations during gait, possibly as a way to stabilize the trunk and head. This corrective procedure may not be necessary for TC subjects as they learned to move the trunk more proportionately to total lower extremity MEE, without need to explicitly attend to additional factors or mechanisms to stabilize the head. Although speculative, these scenarios correspond with the observed high positive correlation between change in leg MEE and change in trunk velocity peak and range seen after TC training but not after VR rehabilitation. Because the VR exercise program may increase subjects' awareness of eye and head movement strategies that cause dizziness and instability [ 8 , 47 ], a more rigid head and trunk strategy during dynamic activities such as gait would be expected. The VR program's balance retraining exercises do not emphasize dynamic whole body movement patterns that improve overall postural control [ 10 ]. Subjects practice maintaining balance in challenging postures (narrow base of support such as feet together and one-legged standing still) using a variety of self selected movement patterns. It is probable that subjects would make the trunk more rigid to lessen head movement during these tasks. The VR group's decrease in trunk velocity range with increase in mechanical energy of the lower extremities during gait appears to support this explanation. Within the TC group, the subjects practice series of movement patterns that include elements of controlled trunk rotation without instruction on eye fixation. The training of smoothly transitioning body segment motions may provide these subjects a different mode of compensation for their instability. Practice of the TC movements may promote more natural trunk movements similar to healthy persons as shown in our biomechanical findings during gait. Although the preliminary results presented here suggest the lower extremities may play an important role in the ability of vestibulopathic patients to improve gait function, several limitations of the present study may prevent broad generalization of the results. Our small sample size was perhaps the most important limitation. It is probable that the lack of between-group significant changes, particularly in light of many significant within group differences (pre-post intervention), was due to high variances obscuring group mean differences. Indeed, a larger sample size for controlling type II errors (increasing power), and better control of type I errors for multiple statistical tests, is warranted for future full-scale studies. The large age range within groups may have also contributed to high variability, but note that more heterogeneous samples in fact enhance external validity, including generalizability, of the results. We also did not include a no-treatment control group in the experimental design of this preliminary study. Because vestibulopathic patients may learn to compensate spontaneously, a no-treatment or sham-treatment group would be necessary to determine if changes in gait function are truly a result of the interventions. This is unlikely, however, given the inclusion criteria that all subjects must have had stable symptoms for 6 months and were on average 3 year post-onset of vestibulopathy. Given that both groups improved gait function in our randomized design comparing substantially different interventions suggests that the effects observed were not spurious. It must be recognized, however, that assumed improvements in function, via increased gait speed for example, may be limited [ 48 , 49 ]. As well, we only analyzed the mechanics of the lower extremities in the sagittal plane. It is highly likely that compensations for lower limb power impairments occurred in frontal and transverse planes as well. Also, the different number of patients in each group having a diagnosis of UVH and BVH is a potential limitation. Although there were no significant differences in proportion of UVH and BVH between the two treatment groups, that the BVH patients were much more disabled than the UVH patients in this study may be important, even when the difference in proportions of diagnostic categories (BVH or UVH) within treatment groups is small. Lastly, although there were no significant differences in age and gender distribution between treatment groups, a larger study sample would allow such subgroup effects to be studied. We conclude that VR and TC can successfully improve gait function, as determined by common time-distance measures, in patients with vestibulopathy. We further conclude, however, that TC improves lower extremity motor control more than VR, by selective redistribution of joint energetics, which appears to engender a more vigorous gait and better trunk control. TC, as a complementary treatment to VR, may allow for better control of the trunk through reorganization of lower-extremity motor patterns, elicited from the flowing, controlled TC exercises. Competing interests The author(s) declare that they have no competing interests. Authors' contributions All authors participated in the overall study design, contributed to the interpretation of data and writing/editing of the manuscript, and have read and approved the final manuscript. CAM conceived the hypotheses for this manuscript and carried out the data analysis; DEK was the principal investigator of the project; SWP was the neurologist associated with the project; DMS conducted the patient testing and assisted in development of the vestibular rehabilitation program; PMW conducted the Tai Chi intervention; and SLW was the project consultant. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC552306.xml |
554993 | Association between a variation in the phosphodiesterase 4D gene and bone mineral density | Background Fragility fractures caused by osteoporosis are a major cause of morbidity and mortality in aging populations. Bone mineral density (BMD) is a useful surrogate marker for risk of fracture and is a highly heritable trait. The genetic variants underlying this genetic contribution are largely unknown. Methods We performed a large-scale association study investigating more than 25,000 single nucleotide polymorphisms (SNPs) located within 16,000 genes. Allele frequencies were estimated in contrasting DNA pools from white females selected for low (<0.87 g/cm 2 , n = 319) and high (> 1.11 g/cm 2 , n = 321) BMD at the lumbar spine. Significant findings were verified in two additional sample collections. Results Based on allele frequency differences between DNA pools and subsequent individual genotyping, one of the candidate loci indicated was the phosphodiesterase 4D ( PDE4D ) gene region on chromosome 5q12. We subsequently tested the marker SNP, rs1498608, in a second sample of 138 white females with low (<0.91 g/cm 2 ) and 138 females with high (>1.04 g/cm 2 ) lumbar spine BMD. Odds ratios were 1.5 (P = 0.035) in the original sample and 2.1 (P = 0.018) in the replication sample. Association fine mapping with 80 SNPs located within 50 kilobases of the marker SNP identified a 20 kilobase region of association containing exon 6 of PDE4D . In a second, family-based replication sample with a preponderance of females with low BMD, rs1498608 showed an opposite relationship with BMD at different sites (p = 0.00044-0.09). We also replicated the previously reported association of the Ser37Ala polymorphism in BMP2 , known to interact biologically with PDE4D, with BMD. Conclusion This study indicates that variants in the gene encoding PDE4D account for some of the genetic contribution to bone mineral density variation in humans. The contrasting results from different samples indicate that the effect may be context-dependent. PDE4 inhibitors have been shown to increase bone mass in normal and osteopenic mice, but up until now there have been no reports implicating any member of the PDE4 gene family in human osteoporosis. | Background The postmenopausal loss of bone mass and subsequent increased risk of low-energy (fragility) fractures is an important public health problem, especially in countries with a high proportion of elderly individuals. More than 1 million fragility fractures, primarily in postmenopausal women, occur each year in the US. The annual direct medical costs exceed US$10 billion [ 1 ]. Bone mineral density (BMD) measured with dual energy X-ray absorptiometry (DEXA) has been widely used to estimate the risk of fracture in epidemiological studies and to study treatment effects of antiresorptive agents in clinical trials. There are several well documented environmental and biological factors known to influence bone mineral density and the risk of fragility fractures including female gender, age, previous fragility fracture, low body weight, reduced lifetime exposure to estrogen, low calcium intake, physical inactivity, vitamin D deficiency, smoking, and excessive alcohol consumption [ 2 - 5 ]. There is also a strong genetic component to interindividual BMD variability, with heritability estimates ranging from 0.46 to 0.84 at different body sites [ 6 - 8 ]. Numerous candidate genes have been tested for association to BMD and fragility fractures. A polymorphism in a transcription factor-binding site of the collagen 1A1 ( COL1A1 ) gene has shown one of the most consistent associations to osteoporosis, even if the association is generally weak for BMD and varies between populations [ 9 - 11 ]. Linkage studies have also been performed with the aim of locating genetic loci influencing BMD variation [ 12 - 19 ]. So far, the genes responsible for the resulting linkage peaks have not been identified. Recently, linkage of a compound osteoporosis phenotype was reported to chromosome 20p12. Subsequent positional cloning efforts implicated BMP2 , the gene encoding for bone morphogenetic protein 2, as responsible for the linkage [ 20 ]. Nevertheless, the associations reported thus far that have been independently validated account for only a small portion of the genetic contribution to BMD and osteoporosis. Studies that rely on direct association approaches based on linkage disequilibrium within populations are expected to have greater statistical power and be more feasible to implement than traditional linkage studies to identify common variations that influence common, complex traits such as osteoporosis [ 21 ]. Recently, there has been increasing interest in the use of whole-genome association methods to identify genes that are involved in complex trait variation. To date, however, few such large-scale studies have been reported. In an effort to identify genes and variants that influence risk of osteoporosis, we conducted a large-scale study using more than 25,000 single nucleotide polymorphisms (SNPs) located within approximately 16,000 genes in DNA pools of unrelated females at the extremes of the lumbar spine bone mineral density distribution. A number of intriguing associations identified in this study are currently being scrutinized in further detail. In this paper we report the most advanced of these, which is the association of a variation in PDE4D , the gene encoding cyclic AMP-dependent phosphodiesterase 4D, with lumbar spine BMD. PDE4D selective inhibitors have been shown to promote osteoblast differentiation in progenitor cells [ 22 ] and to increase bone mass by promoting bone formation in normal mice [ 23 ] but the gene has not until now been implicated in human bone metabolism. Methods Subjects Discovery sample: unrelated females from UK twin collection The population sample from which the discovery samples were chosen consisted of 5,436 female twins collected at the Twin Unit at St Thomas Hospital, London, England. They were selected without regard to health or trait. The volunteers had been recruited through advertisements and had undergone extensive investigation at the Twin Unit at St Thomas Hospital. Investigations included several questionnaires inquiring about present and past diseases, symptoms, family history, socio-economical factors, and medication. Subjects underwent an extensive clinical assessment including DEXA measurements of bone mineral density and anthropometric measurements [ 17 , 24 ]. All individuals with data on lumbar spine BMD were considered for inclusion. To exclude relatives, the individual with the most extreme BMD was kept in each twin pair. Individuals with diseases or medication known to influence BMD were excluded, as were individuals younger than 40 years because of the observed complex relationship between age and BMD. In addition, individuals with fractures were excluded from the high BMD group. BMD values were adjusted for age, weight, BMI, self reported leisure time physical activity, smoking, and menopausal status using an ordinary least squares model including second and third order polynomial terms for age and second order terms for weight and BMI. We included BMI and weight as covariates because both were independently associated with BMD in this sample. Based on the trade-off between group sizes and separation, target sizes of 350 were chosen, resulting in a separation of approximately 1.9 SD between groups. After assessment of DNA availability sufficient for such a large scale study, group sizes were reduced to 319 and 321 individuals in the high and low groups, respectively. Lumbar spine BMD T-scores were calculated with the females between 20 and 35 years of age as the reference population. Based on this, 32% of the women in the low BMD group had osteoporosis and an additional 58% had osteopenia according to WHO criteria. The characteristics of the selected individuals are reported in Table 1 . Replication sample: Australian twin collection A twin sample from Royal North Shore Hospital, Sidney, Australia was collected similarly as the UK twin collection. 731 individuals including twin pairs and singletons with lumbar spine BMD assessments were available for genotyping (Table 3 ). Groups of unrelated subjects corresponding to the lower and upper quartiles of the age- and BMI-adjusted lumbar spine BMD distribution were defined similarly to the discovery sample. The characteristics of the selected individuals are reported in Table 1 . Replication sample: international multi-center sib-pair study The second replication sample was a multi-center (Australia, UK, New Zealand, Belgium) study that collected sib pairs concordant and discordant for bone mineral density [ 17 ]. Probands (BMD Z-score <-1.5 at lumbar spine, femoral neck, or hip) were identified and their siblings were contacted and underwent DEXA measurements at the lumbar spine and hip. Participants had to be between 25–85 years of age. Exclusion criteria included steroid medication, hyperparathyroidism, immobility, amenorrhea, anorexia nervosa, and unstable thyroid disease. Nine hundred and eight individual samples were genotyped from 392 families. In the present analysis we included only females older than 40 years of age. In this sample the distribution of family sizes were 164 singletons, 248 families of 2, 34 families of 3, 7 families of 4, and 3 families of 5 members (Table 3 ). Lumbar spine BMD levels were adjusted by age and BMI as described for the Australian sample. Human subjects protection All studies were approved by the appropriate research ethics committees. All participants gave their informed consent to participate in genetic studies before enrollment. Bone mineral density Bone mineral density was estimated from the L1-L4 vertebrae, hip, and forearm using DEXA according to the user's manual for the Hologic QDR 4500W, (Hologic, Waltham, Massachusetts, United States) at all collection sites. SNP markers and genotyping A set of 25,494 SNP markers was selected from a collection of 125,799 experimentally validated polymorphic variations [ 25 ]. This set was limited to SNPs located within gene coding regions, minor allele frequencies greater than 0.02 (95% have frequencies greater than 0.1), and a target inter-marker spacing of 40 kb. SNP annotation is based on NCBI dbSNP database, refSNP, build 118. Genomic annotation is based on NCBI Genome Build 34. Gene annotation is based on LocusLink genes for which NCBI was providing positions on the Mapview FTP site. For pooled DNA assays, 25 ng of case and control DNA pools was used for amplification at each site. All PCR and MassEXTEND™ reactions were conducted using standard conditions [ 26 ]. Relative allele frequency estimates were derived from area under the peak calculations of mass spectrometry measurements from four analyte aliquots as described elsewhere [ 26 ]. For individual genotyping, the same procedure was applied except only 2.5 ng DNA was used and only one mass spectrometry measurement was taken. Primers used to genotype rs1498608 were ATAACCTCGGGGTCCAGAAA (forward PCR primer), GAATCCCTGTTCATTCCTTG (reverse PCR primer) and CCCTAAAAACTGTTCCAGGTA (extension primer). The primers used to genotype the Ser37Ala polymorphism in BMP2 were AGCTGGGCCGCAGGAAGTTCG (forward PCR primer), TCGTCAGAGGGCTGGGATGAG (reverse PCR primer) and TGAGGGGCGGCCCGACG (extension primer). Statistical analysis Tests of association between adjusted lumbar spine BMD group and each SNP using pooled DNA were carried out in a similar fashion as explained elsewhere [ 27 ]. Briefly, the test statistic is based on the difference in allele frequencies between the two groups divided by the known sources of variation in each allele frequency estimate, including sampling and pool-specific measurement variation. Sources of measurement variation incorporated in the test statistic are pool formation, PCR/mass extension, and chip measurement. When three or more replicate measurements of a SNP were available within a model level, the corresponding variance component was estimated from the data. Otherwise, the following historical laboratory averages were used: pool formation = 5.0 × 10 -5 , PCR/mass extension = 1.7 × 10 -4 , and chip measurement = 1.0 × 10 -4 . Tests of association using individual genotypes were carried out using a chi-square test of heterogeneity to compare allele frequencies, and Fisher's exact test to compare genotype frequencies (due to low frequency contingency table cells). Confidence intervals and P-values for odds ratios were derived using Fisher's exact test. When one or more cell counts were zero, non-infinite odds ratios were estimated by adding 0.5 to each cell [ 28 ]. In the samples that included a combination of singletons, sib pairs, and occasionally additional relatives, we estimated the relationship between genotypes and phenotypes using the generalized estimating equations (GEE) approach with a Gaussian link by clustering on family using an exchangeable correlation matrix [ 29 ]. Hypothesis testing was carried out with a Wald test statistic. The geepack implementation of GEE in the R statistical software platform was used [ 30 ]. No attempt was made to correct P-values for multiple testing. Rather, P-values are provided to compare the relative strength of association. P-values less than 0.05 are referred to as statistically significant. Results Initial genome scan in UK sample We carried out a genome-wide association study using 25,494 SNPs located within 10 kb of 15,995 LocusLink annotated genes. An overview of the investigative process is shown in Figure 1 . The basic design was a two-group study in subjects from the tails of the adjusted lumbar spine BMD distribution. We selected lumbar spine BMD as the phenotype to create the contrasting groups because it had a high estimate of heritability in our twin sample (h 2 = 0.82) [ 7 ]. The selected low and high BMD groups consisted of 319 and 321 individuals, respectively. The adjusted BMD range was 0.56–0.87 g/cm 2 in the low BMD group and 1.11–1.60 g/cm 2 in the high BMD group, corresponding to the upper and lower 22 nd percentiles. Other characteristics of the samples are described in Table 1 . To facilitate the screening of such a large number of SNPs, we employed a high-throughput approach using DNA pools, chip-based mass spectrometry [ 26 , 31 - 33 ], and a three-phase SNP selection strategy (Figure 1 ). In the first phase, we performed a single PCR and primer extension reaction for each SNP on two DNA pools consisting of equimolar amounts of DNA from each individual in the low BMD group and high BMD group, respectively. Relative allele frequencies obtained from four mass spectrometry measurements of the extension products were compared between pools. In the second phase, the 1,520 SNPs (~6%) with the most statistically significant associations (nominal P-values < 0.05) were re-measured in triplicate on each DNA pool. In the third phase, we genotyped the 140 most significant SNPs (9%) from step two (nominal P-values < 0.02) on all individuals comprising the pools. Based on the genotype results, 78 SNPs were confirmed to have statistically significant allele frequency differences between cases and controls (P < 0.05). The liberal criteria for selecting SNPs from each phase represent a practical trade-off between following up false positive versus overlooking false negative associations. We chose to follow up as many SNPs from each phase as seemed reasonable. One of the associations was found with rs1498608, an A/T polymorphism within intron 5 of PDE4D on chromosome 5q12. Allele frequencies for the T allele based on genotyping were 0.91 in the low lumbar spine BMD pool and 0.88 in the high BMD pool (OR = 1.5, P = 0.035). Complete genotype counts and summary statistics are reported in Table 2 . Observed genotype frequencies were consistent with expected frequencies under Hardy-Weinberg equilibrium within each study collection. Menopausal status did not have a significant influence on the effect (P = 0.87). Genome-wide studies using tens of thousands of SNPs and liberal statistical selection criteria are expected to yield a high proportion of false positive associations. To distinguish the true genetic effects from among the false positives, the 78 selected SNPs were genotyped in a second twin sample from Australia. Replication in Australian sample The Australian replication sample, a combination of female and male twin pairs and singletons, was analyzed in two ways. First, to create a design and carry out an analysis comparable to the discovery sample, unrelated individuals were selected from the lower and upper quartiles of the sex-specific adjusted lumbar spine BMD distribution (Table 1 ). A similar effect was observed for the marker SNP rs1498608 in females (OR = 2.14, P = 0.018) and males (OR = 1.55, P = 0.35) as in the original UK collection (Table 2 ). The second method of analysis utilized generalized estimating equation (GEE) models to take into account all of the available genotype information by carrying out a regression-type analysis while accounting for familial covariance. The characteristics of this sample are reported in Table 3 . The regression of marker genotypes on adjusted lumbar spine BMD, with sex included as a covariate, found the AA genotype to be associated with significantly higher levels than the AT (β = 7.8 g/cm 2 , P = 0.049) or TT (β = 8.0 g/cm 2 , P = 0.037) genotypes, thus confirming the results observed in the unrelated tails of this sample. Similar GEE analyses carried out for femoral neck and hip BMD were not statistically significant. Replication in international multi-center family study Being a sample of mostly affected sib pairs, this sample was unsuitable for formation of groups with contrasting BMD because of the preponderance of individuals with low BMD (Table 3 ). Therefore, we restricted the analysis to using a generalized estimating equation, regressing marker genotypes on BMD values. Surprisingly, the estimates in this sample were opposite to that in the Australian sample, as the AA genotype was associated with lower adjusted lumbar spine BMD values than both the AT (β = -5.3 g/cm 2 , P = 0.11) and the TT (β = -5.4 g/cm 2 , P = 0.09) genotypes. Using Z-scores at the femoral neck (P = 0.0007 and 0.0004), total hip (P = 0.003 and 0.007), and lumbar spine (P = 0.03 and 0.02) as dependent variables confirmed this pattern of association. In all cases the AT and TT genotypes had very similar point estimates. Association fine mapping In order to better define the extent of the region of association and possibly identify other SNPs more strongly associated with BMD, we performed DNA pool based association fine mapping in the UK sample using 80 publicly available intronic SNPs in the 100 kb region surrounding the incident SNP (Figure 2 ). This analysis identified a 20 kb region of association encompassing exon 6 of PDE4D . Replication of BMP2 association As described in the discussion below, PDE4D inhibition is known to influence BMP2-induced alkaline phosphatase activity in osteoblast precursor cells. Recently, variation in the gene encoding BMP2 was found to be associated with osteoporosis in a study employing whole genome linkage and subsequent positional cloning [ 20 ]. Since we were unaware of any published independent attempts to replicate this finding, we genotyped the Ser37Ala polymorphism in our UK and international samples. In the UK sample, the allele frequency of the rare allele was 2.2% in the low BMD group and 1.4% in the high BMD group, with an odds ratio of 1.6 (P = 0.28). We tested for, but were unable to detect, an interaction between the Ser37Ala polymorphism and rs1498608 on the association with lumbar spine BMD. In the international sample we performed an allele based general estimating equation to estimate the effect of the rare allele on BMD in that sample. The allele frequency of Ala37 in this sample, mainly selected for low BMD, was 1.9%. The effect of the Ala allele was estimated to decrease the adjusted lumbar spine BMD by 0.06 g/cm 2 (P = 0.0029). There were no homozygous Ala individuals in this sample. Discussion In an association study using SNPs in nearly 16,000 genes we obtained evidence that variation in the SNP rs1498608 located within PDE4D is associated with low bone mineral density at the lumbar spine in females. PDE4D encodes cyclic AMP-dependent phosphodiesterase 4D. Phosphodiesterases are a superfamily of enzymes involved in degradation of cyclic adenosine monophosphate (cAMP) and cyclic guanosine monophosphate (cGMP) [ 34 , 35 ]. cAMP and cGMP are important second messengers participating in the response of various cells to hormones. In osteoblasts, cAMP produced in response to parathyroid hormone or prostaglandins regulates osteoblastic differentiation [ 36 - 39 ], which leads to increases in cancellous bone volume as indicated by experiments in animal models [ 40 - 45 ]. The intracellular level of cAMP is regulated by G protein-coupled adenylyl cyclase [ 46 ], and degradation is mediated by the phosphodiesterases. The phosphodiesterase superfamily consists of seven families, PDE1-7, distinguished by substrate specificity, chromatographic behaviour during purification, and affinity for biochemical activators and inhibitors. Of these, the PDE4 family is specific for cAMP and is selectively inhibited by rolipram. Four PDE4 genes, 4A, 4B, 4C, and 4D, have been cloned from rat and humans, all of which are predicted to have multiple protein products due to alternate spicing of RNAs. PDE4 inhibitors have been shown to increase bone formation in normal mice [ 23 ] and to ameliorate loss of bone mass in animal models of osteopenia [ 47 , 48 ]. PDE4A and PDE4D are expressed in two common mouse osteoblastic cell lines, ST2 and MC3T3-E1, that represent different stages in the osteoblast differentiation pathway [ 22 ]. PDE4 inhibition with rolipram increased BMP2-induced alkaline phosphatase activity, a marker of early osteoblast differentiation in ST2 cells. Furthermore, rolipram increased the expression of alkalic phosphatase, osteopontin, collagen type I and osteocalcin in the same osteoblast precursor cells [ 22 ]. In spite of these experimental data, we are not aware of any published attempts to investigate the role of PDE4 genes in human osteoporosis. However, variation in PDE4D was recently reported to be associated with the risk of ischemic stroke [ 49 ]. Given the central role of PDE4 in second messenger signalling, it is quite conceivable that PDE4D variants may have effects on the risk for different common diseases. There are other examples of genes having such pleiotropic effects, the most notable being APOE in hyperlipidemia and Alzheimer's disease [ 50 , 51 ]. It should also be noted that Gretarsdottir et al found that the PDE4D association with stroke was strongest for a region in the recently extended 5' end of the gene, which is close to 1,000 kb upstream of rs1498608 [ 49 ]. Assuming a contribution of PDE4D to the risk of osteoporosis as well as stroke, it is possible that different domains are involved in the different diseases. Given the interaction between BMP2 and PDE4 for the inhibition on osteoblastic differentiation in vitro, it is interesting to note that variants in the gene encoding for BMP2 have also been found to increase risk of osteoporosis in humans [ 20 ]. In the current study, we replicated the association between the Ser37Ala variant in BMP2 and measures of osteoporosis in an international family-based sample ascertained via low BMD probands. Although not statistically significant, this finding was supported by the results in the discovery sample of unrelated high and low spine BMD subjects. The allele frequency in the low BMD group was 2.2% and in the high group 1.4%, with an odds ratio of 1.6 (P = 0.28). The rare allele was less common in our low BMD group than the low spine BMD group (3%) in the Icelandic sample. However, our allele frequencies in the low and high BMD groups and the resulting OR corresponded well with the figures in the Danish sample (1.8% vs 1.0%, RR = 1.8) reported in the same paper [ 20 ]. We found no evidence for statistical interaction between the variations in BMP2 and PDE4D in either sample. However, given the low minor allele frequencies of each SNP, there was very little power to test for interaction effects. The starting point of the present study was a large-scale association study of more than 25,000 SNPs located in 16,000 genes. After a stepwise selection process an association between an intronic SNP in PDE4D and lumbar spine bone mineral density was detected, providing the first evidence that a variant of this gene could contribute to the risk of osteoporosis in humans. The effect was similar in size in premenopausal and postmenopausal women, indicating that the effect would be on the attainment of peak bone mass rather than the rate of decrease in BMD after menopause. The lack of a detectable interaction with female sex hormones is supported by having observed a similar genetic effect in the small sample of males in our study. The genetic contribution to peak bone mass is possibly bigger and definitely better documented than the as yet unproven genetic influence on postmenopausal bone loss [ 52 ], and it is possible that PDE4D could contribute to this effect, especially in light of the documented anabolic effect on bone by PDE4 inhibitors. An association with an intronic SNP provides little evidence for a change in amount or function of the protein that could explain the association. None of the 80 SNPs investigated as part of the association fine mapping were non-synonymous coding changes, which is consistent with the extraordinary lack of variation that others have reported for all PDE classes [ 53 ] and PDE4D in particular [ 49 ]. This makes it unlikely (but still possible) that the observed association would be due to a non-synonymous and disruptive single-base coding change in linkage disequilibrium with our marker SNP. Therefore it is more likely that the effect is mediated by a change in RNA splicing or expression. Given the functional similarity between different PDE4 enzymes, we went back and scrutinized our data for associations with SNPs in the other PDE4 genes that may have been overlooked during the first stage of the scan. The only SNP in PDE4B in our assay set, rs1318475, was taken through to the second stage (Figure 1 ) where it was estimated to have an OR of 0.78 (P = 0.041), but failed the criteria to be taken forward to the genotyping stage. Similarly, a SNP roughly 18 kb downstream of PDE4C , rs874628, was also taken forward to the second stage where it displayed an OR of 1.3 (P = 0.08). These results suggest that further investigation into possible associations between variants in all PDE4 genes and bone mineral density may be justified. The route by which these genetic associations were arrived at and the potential for spurious association must be considered. Recent published work has brought light to the need for proper validation to verify genetic findings for complex traits [ 54 - 56 ]. In the current study, the initial association found between the PDE4D marker and bone mineral density was one result from over 25,000 hypothesis tests. A conservative Bonferroni adjustment to yield an experiment-wide type I error rate of 0.05 would demand a test-wise p-value on the order of 10 -6 . Given the modest sample size, only common variations with relatively large effects (OR > 2) would reach such significance levels. Instead, we chose to be more mindful of the role of type II error rates and apply a more liberal set of criteria in the initial phases of the study and verify true genetic effects by independent replication. The analysis of 78 selected markers in the Australian replication sample resulted in multiple associations of continuing interest, with rs1498608 displaying one of the strongest associations. A one-sided test of association comparing the results in the discovery and replication samples yields a p-value of 0.0074. This would not be considered significant on an experiment-wide level after Bonferroni adjustment, which would require a p-value on the order of 0.0006 or lower. The analysis in the international replication sample produced contradictory data in that the A allele, which in the first two samples was associated with increased lumbar spine bone mineral density, was associated with decreased BMD at all tested sites. The pattern of association evident from the first two samples, with AT and TT genotypes having very similar point estimates, was preserved in this sample, even in the face of the reverse direction of association. The highly statistically significant association between rs1498608 and femoral neck and hip BMD in this third sample and the consistency in the pattern of association would be unexpected from a spurious result. A possible explanation for the contradictory results could be the fact that the international sample consists mostly of individuals with low BMD since the probands all have a BMD Z-score < -1.5, and most of the siblings also have low BMD. It is possible that within such a selected sample the relationship between rs1498608 and BMD could be altered due to interactions with other genetic or environmental factors. The well-documented anabolic effect on bone by PDE4 inhibitors lends indirect support for the association reported here, and it would seem that the possible role of PDE4D variants in the genetic contribution to BMD in humans merits further investigation. Conclusion The result of the present large scale association study together with data from previously published animal models suggest that genetic variation in the gene encoding PDE4D contributes to the variation in lumbar spine BMD in humans. Competing interests The authors of this manuscript affiliated with Sequenom, Inc. (RR, SM, CH, GM, SK, MN, AB) declare competing financial interests, which may include current or prior receipt of salary and/or stock ownership, as Sequenom, Inc. may be affected financially by the publication of this manuscript. Authors' contributions RR drafted the manuscript and participated in study design and data analysis. SM was study project leader and participated in data analysis. SK supervised the operational aspects of the study, and was responsible for the development of the SNP marker assay set. CH and GM participated in the development of the SNP marker assay set. SW was project manager for the international multicenter study. PS was principal investigator for the Australian Twin collection. TS was principal investigator for the UK twin and international collections. MN participated in study design, performed the statistical analyses and helped draft the manuscript. AB had the overall scientific responsibility for the study. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554993.xml |
509302 | The Telomeric Protein TRF2 Binds the ATM Kinase and Can Inhibit the ATM-Dependent DNA Damage Response | The telomeric protein TRF2 is required to prevent mammalian telomeres from activating DNA damage checkpoints. Here we show that overexpression of TRF2 affects the response of the ATM kinase to DNA damage. Overexpression of TRF2 abrogated the cell cycle arrest after ionizing radiation and diminished several other readouts of the DNA damage response, including phosphorylation of Nbs1, induction of p53, and upregulation of p53 targets. TRF2 inhibited autophosphorylation of ATM on S1981, an early step in the activation of this kinase. A region of ATM containing S1981 was found to directly interact with TRF2 in vitro, and ATM immunoprecipitates contained TRF2. We propose that TRF2 has the ability to inhibit ATM activation at telomeres. Because TRF2 is abundant at chromosome ends but not elsewhere in the nucleus, this mechanism of checkpoint control could specifically block a DNA damage response at telomeres without affecting the surveillance of chromosome internal damage. | Introduction Telomeres prevent the recognition of natural chromosome ends as double-stranded breaks (DSBs). When telomeres become dysfunctional due to shortening or loss of protective factors, chromosome ends activate a DNA damage response mediated (in part) by the ATM kinase ( Karlseder et al. 1999 ; Takai et al. 2003 ). A major challenge in telomere biology is to define the mechanism by which functional telomeres prevent these events. Here we show that the human telomere-associated protein TRF2 is an inhibitor of the ATM kinase, suggesting a mechanism by which the telomeric protein complex prevents the activation of this DNA damage response transducer. TRF2 is a small multimeric protein that binds to duplex telomeric (TTAGGG) repeats and recruits hRap1, ERCC1/XPF, WRN, and the Mre11/Rad50/Nbs1 complex to chromosome ends ( Li et al. 2000 ; Zhu et al. 2000 , Zhu et al. 2000 2003 ; Opresko et al. 2002 ; Machwe et al. 2004 ). TRF2 can be inhibited with a dominant-negative allele, TRF2 ΔBΔM , which removes the endogenous TRF2 complex from telomeres ( van Steensel et al. 1998 ). Upon expression of TRF2 ΔBΔM , telomeres become uncapped and experience a series of deleterious events, including association with DNA damage response factors such as 53BP1, cleavage of the telomeric 3′ overhang by ERCC1/XPF, and telomere–telomere ligation by DNA ligase IV ( van Steensel et al. 1998 ; de Lange 2002 ; Smogorzewska et al. 2002 ; Takai et al. 2003 ; Zhu et al. 2003 ). The DNA damage response to uncapped telomeres induces phosphorylation of DNA damage response proteins, including H2AX, SMC1, Rad17, CHK1, and CHK2, and upregulation of p53, p21, and p16, resulting in a G1 arrest ( Karlseder et al. 1999 ; Smogorzewska and de Lange 2002 ; d'Adda di Fagagna et al. 2003 ). Primary human cells with telomere damage undergo apoptosis or senescence ( Karlseder et al. 1999 ; Smogorzewska and de Lange 2002 ). An important transducer of the DNA damage signal is the ATM kinase (reviewed in Shiloh 2003 ). ATM activation requires autophosphorylation on S1981 and concomitant dissociation into monomers, the presumed active form of the kinase ( Bakkenist and Kastan 2003 ). DSBs and other genome stress lead to a rapid conversion of the ATM pool into active S1981–P monomers, which can phosphorylate regulators of the G1/S, intra-S, and G2/M cell cycle transitions ( Bakkenist and Kastan 2003 ). Activation of ATM also takes place in response to telomere damage. When telomeres become uncapped due to inhibition of TRF2, S1981-phosphorylated ATM associates with telomeres ( Takai et al. 2003 ). Furthermore, ATM targets become phosphorylated in aging cells with shortened telomeres ( d'Adda di Fagagna et al. 2003 ). Genetic evidence for a role of ATM in the telomere damage pathway is provided by the diminished ability of ataxia telangiectasia (A-T) cells to mount a DNA damage response after telomere uncapping ( Karlseder et al. 1999 ; Takai et al. 2003 ). However, several lines of evidence suggest that a second PIKK (phosphatidylinositol 3-kinase-like kinase), such as ATR or DNA-PKcs, can transduce the telomere damage signal in the absence of ATM ( Takai et al. 2003 ; Wong et al. 2003 ). One proposed mechanism of telomere protection is based on the ciliate telomere proteins, which envelop the single-stranded telomere terminus ( Horvath et al. 1998 ). Such a protein cap, if sufficiently stable, could simply hide chromosome ends from the DNA damage surveillance machinery. Both budding and fission yeast also contain protective single-stranded telomere-binding proteins, but it is not known whether these proteins function similarly by forming a physical cap over the telomere terminus ( Garvik et al. 1995 ; Baumann and Cech 2001 ). TRF2 must represent a different mechanism for telomere protection since it only binds to the duplex part of the telomere. TRF2 has been proposed to promote the formation of t-loops ( Griffith et al. 1999 ; Stansel et al. 2001 ). In the t-loop configuration, the 3′ overhang of TTAGGG repeats is strand-invaded into the duplex part of the telomere. Although this could be an effective way to protect chromosome ends from nucleases and ligases, t-loops have several structural features resembling DNA lesions, including single strand to double strand transitions, 3′ and 5′ ends, and single-stranded DNA. Therefore, human telomeres may need additional mechanisms to circumvent checkpoint activation. The results presented here argue for a model in which TRF2 directly blocks activation of the ATM kinase. Results TRF2 was overexpressed in IMR90 primary fibroblasts using a retroviral vector. Under these conditions, TRF2 saturates its telomeric binding sites and is present in the nucleoplasm. While control IMR90 cells showed the expected reduction in mitotic index after ionizing radiation (IR), TRF2 overexpression partially abrogated this checkpoint response, increasing the percentage of cells entering mitosis from 0.3% to 8% ( Figure 1 ). The inappropriate entry into mitosis is indicative of a failure of the cell cycle checkpoints. Since cell cycle arrest after IR largely depends on ATM (reviewed in Shiloh 2003 ), we asked whether TRF2 was acting on this kinase. Caffeine, an inhibitor of ATM and the related kinase ATR, suppressed IR-induced arrest to a similar extent as TRF2 ( Figure 1 ). Furthermore, caffeine had no additional effect on the ability of TRF2-overexpressing cells to bypass the DSB checkpoint, suggesting that TRF2 and caffeine target the same step in the response pathway. Figure 1 TRF2 Inhibits the IR-Induced Cell Cycle Arrest (A) Retrovirally infected IMR90 cells were treated with 4 Gy IR (left and right) or treated with 4 Gy IR and exposed to caffeine (10 mM) directly after irradiation (middle). After 16 h, during which the cells were incubated in 1 μg/ml colcemide, the DNA was stained with DAPI and mitotic cells were identified by immunofluorescence with an antibody to phosphorylated histone H3. (B) Quantification of bypass of IR-induced cell cycle arrest. The mean percentage of phosphorylated histone H3-positive cells and SDs from three experiments are given. The low maximal incidence of phosphorylated H3-positive nuclei (approximately 18%) is due to loss of mitotic cells during processing; loss of mitotic cells occurred at the same level in control and experimental samples. We then asked whether TRF2 overexpression inhibited other ATM-dependent readouts of the DNA damage response. ATM phosphorylates and stabilizes p53 in response to DNA damage (reviewed in Kastan and Lim 2000 ). Quantitative immunoblotting showed that cells overexpressing TRF2 had a diminished ability to induce p53 after irradiation ( Figures 2 A and 2 B). Both the relative level of p53 protein and the induction of its downstream targets p21, Bax, and Hdm2 were dampened. By contrast, the p16/Rb pathway was not affected by TRF2 ( Figure 2 A). We also examined the phosphorylation of Nbs1 on S343, a target of ATM ( Gatei et al. 2000 ; Lim et al. 2000 ; Wu et al. 2000 ; Zhao et al. 2000 ). Phosphorylation of this residue causes a change in electrophoretic mobility shift and can also be detected using an antibody specific for the phosphorylated form of Nbs1. Extracts from irradiated control cells showed the previously reported retardation of Nbs1 and its reactivity with the S343–P-specific Nbs1 antibody ( Figure 2 C). Both alterations could be reversed by phosphatase treatment of the Nbs1 immunoprecipitations (IPs). In contrast, irradiation did not appear to induce phosphorylation of Nbs1 in cells overexpressing TRF2, indicating that TRF2 diminished the ATM-dependent phosphorylation of Nbs1 ( Figure 2 C). Figure 2 Effect of TRF2 on Downstream Readouts of the IR-Induced ATM Response (A) Retrovirally infected IMR90 cells were exposed to 5 Gy IR and harvested after 0, 12, 24, and 36 h. Levels of p53, Bax, p21, Hdm2, p16, pRB, and γ-tubulin (loading control) were detected by immunoblotting of equal cell number equivalents. (B) Amount of p53 protein at the indicated timepoints (hours) was determined by densitometry of p53 immunoblots such as shown in (A). Amounts were normalized to the vector control at 0 h. Mean values from three experiments and standard deviations are shown. (C) Retrovirally infected IMR90 cells were exposed to 20 Gy IR and harvested after 45 min. Nbs1 was immunoprecipitated and subsequently detected by immunoblotting using a general Nbs1 antibody and a phosphospecific Nbs1 S343 antibody. IPs were treated with λ-phosphatase where indicated. Because TRF2 overexpression blunted several cellular responses that depend (in part) on the ATM kinase, we determined TRF2's effect on the activation of ATM itself. Phosphorylation of ATM on S1981, an early and essential step in the activation of this kinase, can be detected rapidly after IR, even when a low level of DNA damage is induced ( Bakkenist and Kastan 2003 ). To test whether TRF2 affected the autophosphorylation of ATM, the two proteins were expressed in 293T cells and ATM was activated with low doses of IR. TRF2 inhibited the activation of ATM as monitored by immunoblotting with an antibody specific for ATM S1981–P ( Figure 3 A). The relative level of ATM S1981–P (normalized to total ATM protein) at 0.3 Gy was 49% of the vector control value (p = 0.002, Student's t test; n = 7). The TRF2 paralog TRF1, which also binds the duplex telomeric repeats, did not have a significant effect on ATM activation, demonstrating that the effect on ATM is specific to TRF2 ( Figure 3 A). Overexpression of TRF2 also diminished the IR-induced ATM autophosphorylation of endogenous ATM in IMR90 fibroblasts to 55% of vector control value at 0.3 Gy and 60% of vector control value at 0.6 Gy ( Figure 3 B). Figure 3 Effect of TRF2 on IR-Induced ATM Phosphorylation (A) Overexpression of TRF2 inhibits IR-induced phosphorylation of transfected ATM in 293T cells. 293T cells co-transfected with ATM and either TRF2, TRF1, or vector were treated with the indicated doses of IR. After a 30-min recovery, cells were harvested and immunoblot analysis was performed on whole-cell lysates. (B) Overexpression of TRF2 inhibits IR-induced phosphorylation of endogenous ATM in primary fibroblasts. IMR90 primary fibroblasts infected with a retroviral construct expressing TRF2 or an empty virus were treated with the indicated doses of IR. After a 1 h recovery, cells were harvested and ATM was immunoprecipitated from whole-cell lysates. Immunoblot analysis was performed on immunoprecipitated ATM. In order to understand the mechanism by which TRF2 inhibited ATM, we determined whether they interacted in vivo. IPs of the ATM kinase from primary human IMR90 fibroblasts resulted in recovery of a small fraction (approximately 1%) of endogenous TRF2 ( Figure 4 A). This association was accentuated when TRF2 was overexpressed from a retroviral vector. TRF2 was not recovered in anti-ATM immunoprecipitates from A-T cells even when TRF2 was overexpressed ( Figure 4 A), demonstrating that the recovery of TRF2 is dependent on the presence of functional ATM. The co-IP of TRF2 with ATM from IMR90 cells was resistant to the addition of ethidium bromide (data not shown), arguing that DNA tethering is not responsible for the association. A control IP with antibodies to the CycD1/Cdk4/Cdk6 kinase complex did not precipitate TRF2, and an irrelevant nuclear protein (Nova1), overexpressed in parallel in IMR90 cells, was not recovered in the ATM IP ( Figure 4 A). The association of TRF2 with ATM was not dependent on the presence of DNA damage, since neither IR nor UV treatment enhanced the recovery of TRF2 in ATM IPs (data not shown). Figure 4 TRF2 Interacts with the ATM Kinase In Vivo and In Vitro and TRF2 Does Not Localize to IRIF (A) Co-IP of TRF2 with ATM. Protein extracts from IMR90 cells and A-T cells (AG02496) infected with an empty virus or a TRF2-overexpressing virus were incubated with anti-ATM or anti-Cyclin D1 antibodies as indicated, and TRF2 was detected in the IP pellets by immunoblotting. The right panel represents IPs with anti-ATM antibodies from IMR90 cells infected with a retrovirus overexpressing Nova1 or the empty vector and detection of Nova1 by immunoblotting. For each extract 1% of the IP input (input) was processed for immunoblotting in parallel. (B) Bacterially expressed ATM–GST fusion proteins were purified on glutathione agarose beads and visualized by Western blotting with anti-GST antibody (Upstate Biotechnology [Lake Placid, New York, United States] #06–332) (top). Unfused GST was run on a separate gel because of its low molecular weight. Equal amounts of fusion proteins and GST alone were incubated with purified baculoviral TRF2 (middle) or TRF1 (bottom), bound to glutathione beads, spun down, washed, and bound proteins were processed for immunoblotting with an anti-TRF1 or anti-TRF2 serum. (C) TRF2 does not localize to IRIFS. IMR90 primary fibroblasts infected with a retroviral construct expressing TRF2 or an empty virus were treated with 5 Gy IR. After a 90 min recovery, cells were fixed and processed for immunofluorescence with or without Triton X-100 extraction before fixation. Arrowheads denote foci of TRF2 signal previously demonstrated to represent telomeres. When overexpressed, some TRF2 is localized to nucleolus. The nature of the association of TRF2 with ATM was explored further using in vitro pulldown experiments. GST-tagged fragments of ATM were tested for their ability to bind purified TRF2 protein expressed in a baculovirus system. In parallel, we used baculovirus-derived TRF1, which was previously shown to interact with ATM by co-IP ( Kishi et al. 2001 ). TRF2 bound to two overlapping fragments of ATM spanning amino acids 1439 to 2138 ( Figure 4 B). This region contains the FAT domain and S1981, the critical target of autophosphorylation ( Bakkenist and Kastan 2003 ). TRF1 bound a different region of ATM ( Figure 4 B), demonstrating the specificity of the observed interactions. Blunting of the DNA damage response was observed when TRF2 was overexpressed throughout the nucleus. Because TRF2 is chiefly present at telomeres, the simplest interpretation is that the observed activity reflected a telomeric function. However, we also considered the possibility that TRF2 may have a heretofore clandestine role in the general DNA damage response. If this were true, TRF2 might be expected to localize to IR-induced foci (IRIF), where it would be in a position to modulate ATM. Previous data had shown that the endogenous TRF2 does not relocate from telomeres to IRIF ( Zhu et al. 2000 ). Similarly, immunofluorescence analysis showed that overexpressed TRF2 did not form IRIF ( Figure 4 C): the pattern of TRF2 localization was unchanged by IR, and there was no detectable colocalization with the known IRIF component 53BP1 ( Schultz et al. 2000 ). This was also the case when TRF2 localization was examined in cells from which the nucleoplasmic proteins were extracted with a mild detergent ( Figure 4 C). Thus, the inhibitory effect of TRF2 on ATM signaling does not reflect association of TRF2 with sites of DNA damage. Instead, we propose that the inhibition of ATM by TRF2 is an innate property of the protein, important at its natural location: telomeres. However, we cannot exclude the transient presence of TRF2 at DNA lesions and/or a role for TRF2 in the general DNA damage response. Discussion Natural chromosome ends require mechanisms to prevent the activation of the DNA damage response. Inhibition of the ATM kinase at human telomeres is particularly important since the telomeric complex contains the Mre11 complex, one of the DNA damage sensors of the ATM pathway ( Carson et al. 2003 ; Petrini and Stracker 2003 ; Uziel et al. 2003 ). The telomeric protein TRF2 appears to play a central role in preventing telomeres from activating ATM. Removal of TRF2 from telomeres results in the localization of the active, phosphorylated form of ATM to unprotected chromosome ends ( Takai et al. 2003 ) and induces ATM-dependent apoptosis ( Karlseder et al. 1999 ). The data reported here are consistent with the hypothesis that TRF2 protects telomeres through a direct interaction with ATM that blocks its activation. As a result, TRF2 abrogates the downstream outcomes of the ATM-dependent DNA damage response, including phosphorylation of various ATM targets and cell cycle arrest. We feel that the interaction of TRF2 with ATM is likely to be relevant to the mechanism by which TRF2 blocks ATM signaling. TRF2 binds ATM in a region surrounding S1981, which is functionally linked to the oligomerization state of ATM. Phosphorylation on S1981 occurs concomitant with the dissociation of ATM dimers (or oligomers), forming the monomeric, active form of the kinase ( Bakkenist and Kastan 2003 ). TRF2 is also an oligomer of four to eight subunits, which are held together by the TRFH dimerization domain as well as other, yet to be defined, protein interactions ( Broccoli et al. 1997 ; Fairall et al. 2001 ; Stansel et al. 2001 ). Owing to its oligomeric nature, TRF2 could potentially cross-link ATM monomers and thus hold the kinase in its inactive dimeric (or oligomeric) state. In this manner, TRF2 could abrogate the ATM pathway since it would block amplification of the ATM signal at an early step. In agreement with this idea, the in vitro GST pulldown experiments showed that TRF2 can interact with ATM when it is not phosphorylated on S1981. Because mutations in the TRF2 dimerization domain destabilize the protein, it has not been possible to test the contribution of TRF2 oligomerization on ATM repression directly. Alternatively, the interaction of TRF2 with the region surrounding S1981 may prevent ATM autophosphorylation or TRF2 binding could block a presumed interaction between ATM and a DNA damage sensor, such as the Mre11 complex. It is unlikely that TRF2 acts as an ATM target mimetic that titrates out genuine ATM targets since TRF2 is not a target of the ATM kinase (R. Drissi, M. B. Kastan, and J. Dome, unpublished data). Furthermore, TRF2 does not block ATM kinase activity in an in vitro assay (S. Kozlov, J. Karlseder, and M. F. Lavin, unpublished data). These findings are consistent with TRF2 acting at one of the earlier steps in the activation of the ATM kinase, including the interplay between ATM and DNA damage sensors, ATM autophosphorylation, or dissociation of ATM dimers. In this study we have expressed TRF2 at high levels throughout the nucleus, whereas endogenous TRF2 is localized primarily to telomeres ( van Steensel et al. 1998 ). Our estimates suggest that TRF2 is extremely abundant at telomeres. Human cells contain on the order of 1 million copies of TRF2 (X.-D. Zhu and T. de Lange, unpublished data), sufficient to position thousands of TRF2 molecules at each chromosome end. This number is consistent with the presence of thousands of TRF2-binding sites per telomere and the oligomerization potential of the protein. Thus, for every ATM kinase that could be activated at a chromosome end, there is a vast molar excess of its potential inhibitor, TRF2. Since TRF2 is specifically lodged at telomeres and remains there when DNA damage is induced, it is unlikely to interfere with activation of the ATM kinase at sites of DNA damage elsewhere in the genome. Hence, TRF2 could act as a telomere-specific inhibitor of ATM. Previous studies have shown that overexpression of TRF2 can protect critically short telomeres generated by replicative aging ( Karlseder et al. 2002 ). TRF2 reduced the incidence of end-to-end chromosome fusions in this setting and also delayed the onset of senescence. These findings suggested that senescence is induced by an altered telomere state, in which the telomere has become so short that the amount of TRF2 it can recruit is insufficient for the protection of the chromosome end. When TRF2 is overexpressed, this deficiency in TRF2 recruitment may be overcome. One possibility is that the altered state of critically short telomeres represents a situation in which telomeres have a diminished ability to form t-loops. The current findings raise the possibility that the altered state may also include a situation in which the telomere contains insufficient TRF2 to repress ATM. However, the ability of TRF2 to delay senescence was also observed in A-T cells ( Karlseder et al. 2002 ), indicating that ATM repression is not the only pathway by which increased TRF2 loading can protect critically short telomeres. As TRF2 can bind ATM, it has the inherent ability to recruit this protein to telomeres. ATM has not been observed at undamaged telomeres, but its abundance may be too low for detection. The idea that ATM could be recruited to telomeres by TRF2 is interesting considering that ATM-like kinases are necessary for telomere maintenance in Saccharomyces cerevisiae ( Craven et al. 2002 ) and Schizosaccharomyces pombe ( Matsuura et al. 1999 ). It is not excluded that human telomere maintenance similarly requires ATM. TRF2 could function to recruit ATM in an inactive form, perhaps allowing for highly regulated activation of ATM at appropriate times. Such regulation of DNA damage signaling and repair pathways at telomeres has been proposed previously in the context of the nonhomologous end-joining pathway and the role of the nucleotide excision repair endonuclease ERCC1/XPF ( Smogorzewska et al. 2002 ; Zhu et al. 2003 ). In both cases, proteins with the potential to have detrimental effects on telomeres appear to be regulated such that their activities can be employed for telomere function. Materials and Methods Cell culture and IR treatment IMR90 primary lung fibroblasts (ATCC, Manassas, Virginia, United States) and AG02496 and AG04405 primary A-T fibroblasts (Coriell Cell Repository, Camden, New Jersey, United States; PD 12 and PD 15) were grown and infected with retroviruses as described elsewhere ( Karlseder et al. 2002 ). For γ-irradiation, 3 10 5 cells were seeded in 5-cm culture dishes and exposed to a Ce 137 source. Where indicated, the medium was replaced with medium containing 10 mM caffeine (Sigma, St. Louis, Missouri, United States) and 1 μg/ml colcemide (Sigma). Indirect immunofluorescence was performed as described ( Smogorzewska et al. 2000 ; Takai et al. 2003 ). Co-transfection assay for S1981–P inhibition One day prior to transfection, approximately 5 10 6 293T cells were plated in 10-cm dishes. Cells were transfected with 1 μg of FLAG–ATM DNA ( Canman et al. 1998 ) and 9 μg of N-terminally Myc-tagged TRF2 ( Karlseder et al. 2002 ) or TRF1 in a pLPC vector backbone (gift of S. Lowe, Cold Spring Harbor Laboratory) or vector alone using CaPO 4 coprecipitation. Two days after transfection, cells were harvested in media and divided into three equal fractions, which were exposed to 0, 0.3, or 0.6 Gy IR. Cells were allowed to recover for 30 min, washed with PBS, and resuspended in 250 μl of lysis buffer (50 mM Tris [pH 7.4], 1% Triton X-100, 0.1% SDS, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 1 mM PMSF, with a complete mini-protease inhibitor tablet [Roche, Basel, Switzerland] per 10 ml). The NaCl concentration was raised to 400 mM, and the lysate was incubated on ice for 5 min. The NaCl concentration was reduced to 200 mM, cell debris was removed by centrifugation, and an equal volume of Laemmli buffer was added to the lysate. Immunoblotting For ATM immunoblots in the ATM S1981–P suppression assays: 40 μl of 293T cell lysate or ATM immunoprecipitated from approximately 5 10 6 IMR90 fibroblasts were run on 7.5% precast polyacrylamide Bio-Rad (Hercules, California, UnitedStates) Ready Gels. PVDF Immobilon TM Transfer Membrane (Millipore, Billerica, Massachusetts, United States) was prepared for protein transfer according to the manufacturer's instructions and the gel was transferred for 2 h at 90 V. Membranes were preincubated in 10% milk, 0.1% Tween-20 in PBS for 30 min at room temperature and subsequently incubated with primary antibodies: polyclonal rabbit ATM S1981–P ( Bakkenist and Kastan 2003 ) and mouse monoclonal ATM antibody MAT3 (gift from Y. Shiloh) diluted in 0.1% milk, 0.1% Tween-20 in PBS overnight at 4 °C followed by three 10 min washes. Membranes were incubated for 45 min with HRP-conjugated secondary antibodies, washed, and developed using the ECL system (Amersham, Little Chalfort, United Kingdom). Immunoblots of Myc-tagged proteins (using Ab-1; Oncogene Research, Cambridge, Massachusetts, United States) in the ATM S1981–P suppression assays were performed as above, except that nitrocellulose (Schleicher and Schuell, Keene, New Hampshire, United States) filters were used. For all other immunoblots, cells were trypsinized, washed once with PBS, and subsequently lysed in Laemmli buffer at 10 4 cells/μl. Lysates (10 μl) were separated on SDS-polyacrylamide gels (29:1 acrylamide: bisacrylamide, 8% for p53, TRF2, and γ-tubulin, 6% for pRB and Hdm2, 12% for Bax, p21, and p16) and transferred onto nitrocellulose membranes (Schleicher and Schuell) for 60 min at 90 V (Bio-Rad Mini-Protean II Cell). Membranes were preincubated in 10% nonfat dry milk, 0.1% Tween-20 in PBS for 30 min and subsequently incubated with primary antibodies: p53 D01 (Santa Cruz Biotechnology, Santa Cruz, California, United States); TRF2 serum 647 ( Zhu et al. 2000 ); pRB #554136 (PharMingen, Uppsalla, Sweden); Hdm2 #3F3 (gift from, A. Levine); Bax #sc-7480 (Santa Cruz Biotechnology); p21 sc-7480 (Santa Cruz Biotechnology); p16 #15126E (PharMingen); γ-tubulin GTU88 (Sigma); Nova1 ( Luque et al. 1991 ) in 5% dry milk, 0.1% Tween-20 in PBS overnight. Secondary antibody incubation and ECL were performed as described above. To quantify signals, band intensities were determined using an AlphaImager TM 2200 using the SpotDenso function of AlphaEaseFC TM Software Version 3.1.2 (Witec, Littau, Switzerland). IP For co-IP of ATM and TRF2, proteins were extracted from subconfluent cells by incubating trypsinized cells (approximately 10 7 cells/0.1 ml buffer) in 20 mM HEPES (pH 7.9), 0.42 M KCl, 25% glycerol, 0.1 mM EDTA, 5 mM MgCl 2 , 0.2% NP40, 1 mM DTT, 0.5 mM PMSF, 1 μg/ml leupeptine, 1 μg/ml aprotinin, 10 μg/ml pepstatin on ice for 30 min. Debris was removed by centrifugation (14,000 rpm, 4 °C, 10 min). Protein concentration in the supernatant was determined using the Bradford assay and 400 μg of protein was diluted to 150 mM KCl and incubated for 20 min with 100 μl of protein G–Sepharose beads (Amersham), blocked with 1% fetal bovine serum in PBS. The beads were collected at 14,000 rpm for 1 min and the supernatant was incubated with 5 μg of anti-ATM antibody (AB3, Oncogene Research) or anti-Cyclin D1 antibody (#sc-6281, Santa Cruz Biotechnology) for 1 h at 4 °C on a nutator. Protein G–Sepharose beads (30 μl) were added, and the mixture was incubated 1 h at 4 °C on a nutator. The beads were collected at 4,000 rpm at 4 °C, washed three times with wash buffer (150 mM NaCl, 1% NP40, 50 mM Tris [pH 8.0] with protease inhibitors as described above) by vortexing the suspension for 10 s. The beads were resuspended in Laemmli buffer and boiled 5 min, and proteins were separated by polyacrylamide gel electrophoresis. For IP of endogenous ATM from IMR90 fibroblasts for the phosphorylation assay, cells were resuspended in lysis buffer (50 mM Hepes [pH 7.5], 150 mM NaCl, 50 mM NaF, 1% Tween-20, 0.2% NP40, 1mM PMSF, with 1 complete mini-protease inhibitor tablet [Roche, Basel, Switzerland] ) and centrifuged at 13,000 rpm for 10 min. 400 μl of lysate was incubated with 30 μl of blocked protein G beads and 100 μl of D16.11 monoclonal supernatant ( Alligood et al. 2000 ) for 1.5 h. Beads were washed once in lysis buffer and twice in RIPA buffer and resuspended in 60 μl of Laemmli buffer. Pulldown assays GST–ATM fusion plasmids ( Khanna et al. 1998 ) were transformed into BL-21 cells. A 10-ml overnight culture was used to inoculate 500 ml of LB-Amp (50 μg/ml) and at OD 600 , 0.5–0.7, 0.3 mM IPTG (final) was added. After 3 h at 30 °C, cells were harvested, resuspended in 8 ml of lysis buffer (50 mM Tris [pH 7.9], 100 mM KCl, 1% Triton X-100, 2 mM DTT, 0.1 mM PMSF, 1 complete protease inhibitor tablet [Roche]) and sonicated three times for 30 s on ice. The lysate was cleared by centrifugation at 50,000 g at 4 °C and incubated with 600 μl of equilibrated glutathione beads for 2 h at 4 °C. The beads were washed three times for 10 min each (washes 1 and 3: PBS, 1% Triton X-100, 2 mM DTT, 0.1 mM PMSF, 1 mM benzamidine, 1 complete protease inhibitor tablet [Roche]; wash 2: 300 mM NaCl, 50 mM Tris [pH 7.9], 2 mM DTT, 0.1 mM PMSF, 1 mM benzamidine, 1 complete protease inhibitor tablet [Roche]) and a fourth time in wash 4 (50 mM Tris [pH 7.9], 100 mM KCl, 10% glycerol, 2 mM DTT, 0.1 mM PMSF). Fusion proteins were eluted in 500 μl of wash 4 containing 15 mM glutathione (reduced form). Two subsequent elutions were collected. Five micrograms of GST fusion proteins or GST alone were incubated with 2 μg of baculoviral TRF1 or TRF2 in binding buffer (150 mM NaCl, 100 mM KCl, 50 mM Tris [pH 8.0], 1% NP40, 0.1% SDS, 100 g/ml BSA) at 4 °C for 1 h. Glutathione beads (20 μl) were added and incubated for 1 h at 4 °C. Beads were collected by centrifugation at 5,000 rpm at 4 °C and washed three times for 10 min each with binding buffer, and bound protein was eluted by boiling the samples in Laemmli buffer. GST fusion proteins, TRF1, and TRF2 were detected by immunoblotting. Cell cycle arrest assay Cells were seeded on microscope coverslips and irradiated with 4 Gy of γ-irradiation as described above. Cells were incubated in growth medium containing 1 μg/ml colcemid for 16 h. Cells were fixed, and phosphorylated histone H3 was detected by indirect immunofluorescence using a phosphospecific antibody (6G3 monoclonal; Cell Signaling Technology, Beverly, Massachusetts, United States). Cells in mitosis were counted and expressed as a percentage of total cell number. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC509302.xml |
509316 | Innate Immunity in Fruit Flies: A Textbook Example of Genomic Recycling | Drosophila serve as a wonderful model for studying aspects of innate immunity, i.e. the physical, cellular, and molecular features that provide the first lines of defense against infections in flies and man | Organisms of vastly differing morphologies, ecologies, and behaviors—such as fruit flies and humans—are now known to share a multitude of molecular, cellular, and developmental processes. Not only is there extensive similarity in the sequences of fly and human genes, but in addition, almost all of the proteins and major signal transduction pathways that control cell division and differentiation in mammals are also found in the fruitfly Drosophila melanogaster ( Rubin et al. 2000 ; http://flybase.bio.indiana.edu/ ). Components in these pathways perform the same biochemical functions and act in the same order in both fruitfly and mammalian cells. Evolutionary conservation is of considerable practical and theoretical importance to biologists. First, it provides a valuable source of data for the reconstruction of phylogeny ( Salemi and Vandamme 2003 ). Evolutionary connections between organisms that were once hidden by morphology have now been exposed in genomic analyses. Second, the conservation of evolutionary processes or traits is a prime area of investigation in theoretical evolutionary biology ( Gould 2002 ). What can, and cannot, be changed evolutionarily? In a constantly evolving world, how can any biological system or trait survive unchanged ( Van Valen 1973 )? Finally, conservation provides fundamental insights into how complex biological systems, such as immunity, are assembled, maintained, and altered in evolution. Elements of Immunity “Innate” immunity refers to the variety of physical, cellular, and molecular features that provide the first lines of defense against infections. The relatively quick innate immune responses operate along with slower but more targeted adaptive immune responses that generate antigen-specific mechanisms that eventually lead to the destruction and elimination of the pathogen. In mammals, the skin and the epithelial lining of the mucosal tissues act as the primary nonspecific barriers, impeding infectious agents from entering the body. The mucous membrane barrier traps microorganisms, and the cilia present on the epithelial cells assist in sweeping the microbes towards the external openings of the respiratory and gastrointestinal tracts. If infectious agents gain entry into the body, internal innate immune responses become activated and rapidly eliminate the infection. Internal innate immune agents and responses include (amongst others) low pH of the stomach and vagina, proteolytic enzymes and bile in the small intestine, and phagocytosis. Phagocytosis is a fundamental innate immune mechanism carried out by a number of different cell types, including macrophages. Specific macrophage subpopulations are associated with different tissues (alveolar macrophages in the lung, microglial cells in the central nervous system, etc.). Their main function is to consume microorganisms, other foreign substances, and old, dying cells. Innate immunity is present from birth, and the information for innate immune responses is inherited. Cells in the mammalian innate immune system (e.g., macrophages) detect “microbial nonself” by recognizing pathogen-associated molecular patterns (PAMPs; Janeway 1989 ). PAMPs are products of microbial metabolism that are conserved over evolution, distributed in a wide variety of pathogens, and not found in host cells. Lipopolysaccharide is an example of a PAMP and is found in bacteria, viruses, and fungi. Receptors, called pattern recognition receptors, are present on surfaces of host cells and recognize PAMPs. When activated, pattern recognition receptors induce intracellular signaling via the transcription factor NF-κB, resulting in the activation of genes involved in host defense. Adaptive immunity is characterized by greater specificity than innate immunity, as the adaptive immune response can not only distinguish foreign cells from self, but can also distinguish one foreign antigen from another. Another hallmark of adaptive immunity is memory, which enables the body to remember specific adaptive responses in response to specific antigens. Immunological memory allows the body to make a greater and more rapid second response when the body is reinfected by the same pathogen. Immunological memory underlies both immunization and resistance to reinfection, conferring a tremendous evolutionary advantage to vertebrates. The adaptive immune response has nearly infinite flexibility: the T and B lymphocytes of the acquired immune system can rearrange the elements of their immunoglobulin and T-cell receptor genes to create billions of clones with distinct antigen receptors. In organisms where both innate and acquired immune systems are present, there is a clear interdependence between the two systems. For a fully functional immune system, these components must act in synergy. Innate Immunity in Drosophila Because it lacks an adaptive immune response, Drosophila melanogaster serves as a wonderful model for studying aspects of the innate immune system that might otherwise be obscured by the actions of the adaptive immune response. Insects defend themselves against parasites and pathogens by invoking a multitude of innate immune responses ( Figure 1 ; for more details, see recent reviews by Hoffmann and Reichhart [2002] , Hultmark [2003] , Brennan and Anderson [2004] , Meister [2004] , and Theopold et al. [2004] ). Like humans, Drosophila protects itself against microbes and parasites via epithelial barriers: for example, epithelial cells of the trachea, gut, genital tract, and Malpighian tubules produce antimicrobial peptides (local response). Figure 1 Innate Immune Responses of Drosophila (A) Posterior region of a third instar larva showing the cuticle and the trachea. These structures provide a physical barrier against infections. Cellular immune reactions consist of phagocytosis, encapsulation, and melanization. (B) A dead and melanized crystal cell phagocytosed by a plasmatocyte. (C) Encapsulation of an egg of a Drosophila parasite. The parasite is a wasp that normally infects larvae. Cells surrounding the egg are lamellocytes. The cells and the egg are stained with a fluorescent nuclear stain. (D) Clot formation occurs during wound healing. (E) Crystal cells in contact with the larval cuticle. The contents of the crystal cells are melanized. Melanization occurs in response to intruding pathogens or parasites and is also observed during wound healing. (F) Humoral immune reaction. The expression of antimicrobial peptides in the larval fat body is induced by microbes. Cells of the fat body appear green due to the presence of a transgene encoding the green fluorescent protein, under the control of the drosomycin promoter. The drosomycin promoter is activated in response to fungal infections and is under the control of the Toll pathway (see Figure 2 ). Antimicrobial peptides are released from the fat body into the hemolymph. This response is therefore systemic. A similar antimicrobial gene activation response can occur locally in specific body parts such as the trachea or the gut (not shown). Once within the body cavity, microbes may be consumed by the phagocytic blood cells called plasmatocytes ( Figure 1 ). Larger pathogens (such as eggs of parasitic wasps) are inactivated by encapsulation, an immune response carried out by specialized cells called lamellocytes ( Figure 1 ). Lamellocytes differentiate in response to macroscopic pathogens, and their precursors are thought to reside in the larval lymph gland. The transcription factors (GATA, Friend-of-GATA, and Runx family proteins) and signal transduction pathways (Toll/NF-κB, Serrate/Notch, and JAK/STAT) that are required for specification and proliferation of blood cells during normal hematopoiesis, as well as during the hematopoietic proliferation that accompanies immune challenge, are conserved ( Evans et al. 2003 ; Meister 2004 ). In this issue of PLoS Biology , Crozatier et al. (2004) identify the transcription factor Collier as being critical for the differentiation of lamellocytes in Drosophila . The mammalian ortholog of Collier (Early B-cell Factor) is involved in B-cell differentiation in mice. In addition to triggering cellular immune responses, invading pathogens also activate humoral reactions. Microbes induce the rapid secretion of antimicrobial peptides from the cells of the fat body into the larval or adult body cavity (systemic response; Figure 1 ). A microbial infection initiates a zymogen cascade that plays a crucial role in the activation of the antimicrobial genes in the fat body. Infection or wounding also triggers a protein-cleaving cascade that results in the production of toxic intermediates and melanin around microbes or wound sites. This proteolytic cascade is similar to the vertebrate clotting cascade. Drosophila hemolymph also coagulates and participates in host defense and wound healing ( Figure 1 ; Theopold et al. 2004 ). Given the evolutionary success of insects, this combination of defense mechanisms has proven to be extremely effective, allowing insects to thrive in septic environments. NF-κB Activation: The Toll and Imd Pathways of Drosophila The Drosophila genome encodes several members of the multifunctional Toll family of receptors ( Beutler and Rehli 2002 ). Mutations in the Drosophila Toll gene (as well as in other components in the pathway) make the fly susceptible to fungal or gram-positive bacterial infections. However, Toll does not act as a pattern recognition receptor in the fly; instead its activation depends on the presence of the processed (active) form of the growth-factor-like polypeptide Spätzle. Processing of Spätzle depends on a serpin-controlled proteolytic cascade ( Figure 2 ). Figure 2 Molecular Components of the Toll and Imd Pathways Involved in Drosophila Immunity Toll is activated by the processed Spätzle (left). Toll activation leads to intracellular signaling via cytoplasmic proteins Tube and Pelle, leading to the degradation of Cactus and nuclear localization of NF-κB proteins Dorsal and Dif. These transcription factors bind to promoters of target genes, such as drosomycin , activating their transcription. The NF-κB protein for the Imd pathway, Relish, activates diptericin transcription. The signaling events resulting in Dorsal/Dif/Relish activation in the fly are “recycled” in mammals in the activation of mammalian NF-κB. See reviews and De Gregorio et al. (2002) for more details. While components of the Drosophila Toll pathway were identified in earlier genetic screens for developmental mutants, those in the Imd pathway have been the focus of more recent studies, mainly in the context of Drosophila immunity ( Hoffmann and Reichhart 2002 ; Hultmark 2003 ). The effector NF-κB transcription factor of the Imd pathway is Relish, which upon immune activation is cleaved by the Dredd caspase ( Figure 2 ). Using a combination of the RNA interference approach of silencing gene function and a high-throughput cell culture assay, Foley and O'Farrell (2004) report the identification of two new conserved members of this Imd pathway: Sickie is a novel protein required for Relish activation, and Defense repressor 1 is a novel inhibitor of the Dredd caspase. The impressive progress in our understanding of innate immunity in Drosophila is now guiding scientists to explore the immune system of other insects such as the mosquito, Anopheles gambiae , that spreads human malaria. Immune responses in this mosquito are linked to the elimination of the malarial parasites ( Osta et al. 2004 ). A comparison of the immunity-related genes in Anopheles and Drosophila reveals the presence of the Toll signaling pathway in the mosquito genome, even though there are some differences in genes encoding pathogen recognition and signal transduction molecules ( Christophides et al. 2002 ). A detailed and comparative view of the genetic mechanisms underlying their host defense will contribute to the identification of new targets for insecticide development, and provide opportunities for controlling the transmission of pathogens. Concluding Remarks The homologs of many genes involved in innate immune responses in flies and humans have also been found in mice, sharks, nematodes, and plants (e.g., Pujol et al. 2001 ; Nurnberger and Brunner 2002 ). In species studied to date, host defense appears to be mediated by homologous proteins. Taken together, these findings suggest that the regulatory mechanisms of host defense may be hard-wired in the genome much as DNA replication and cell division are. Protein motifs, domains, and signaling elements have, for millions of years, not only retained their ancestral biochemical features but have also continued to participate in similar physiological responses. It is crucial that our evolving knowledge of “genomic recycling” be used to enhance our understanding of the evolution of humans, not only in the context of “descendants of ancient apes,” but in the larger context of our fundamental unity and shared genetic history with all other species. This simple but fundamental idea has yet to be adopted by the majority of our students and teachers. Unless we do more to overcome resistance to the idea that humans share deep evolutionary connections with all animal life, students will become increasingly isolated from an understanding of, and participation in, the genomics and bioinformatics revolution that is transforming the biological and biomedical sciences. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC509316.xml |
546415 | In vitro activity of vancomycin, quinupristin/dalfopristin, and linezolid against intact and disrupted biofilms of staphylococci | Shed cells or disrupted parts of the biofilm may enter the circulation causing serious and very hard to treat biofilm-associated infections. The activity of antimicrobial agents against the shed cells/disrupted biofilms is largely unknown. Methods We studied the in vitro susceptibility of intact and disrupted biofilms of thirty clinical isolates of methicillin-resistant and methicillin–susceptible Staphylococcus aureus (MRSA and MSSA) and Staphylococcus epidermidis to vancomycin, quinupristin/dalfopristin, and linezolid and compared it to that of the suspended (planktonic) cells. Results Bacteria in the disrupted biofilms were as resistant as those in the intact biofilms at the minimum inhibitory concentrations of the antibiotics. At higher concentrations, bacteria in the disrupted biofilms were significantly ( P < 0.001) less resistant than those in the intact biofilms but more resistant than the planktonic cells. Quinupristin/dalfopristin showed the best activity against cells of the disrupted biofilms at concentrations above MICs and vancomycin, at 500 and 1,000 μg/ml, was significantly more active against the biofilms of MRSA and S. epidermidis Conclusion The difficulty of treating biofilm-associated infections may be attributed not only to the difficulty of eradicating the biofilm focus but also to the lack of susceptibility of cells disrupted from the biofilm to antimicrobial agents. | Introduction Gram-positive infections have become a serious problem, especially in the nosocomialsetting, and treatment of these infections is complicated by the emergence of multidrug-resistant pathogens [ 1 ]. Infections caused by Staphylococcus aureus and Staphylococcus epidermidis are among the most frequent causes of both healthcare-associated and community-onset infections [ 2 ]. Staphylococci cause a large percentage of infections by forming biofilms on medical implants, damaged tissues, and most commonly on indwelling vascular catheters [ 3 - 7 ]. Biofilm-associated infections are becoming more common, and occur largely because of the increase in the use of indwelling medical devices. Central venous catheters (CVCs), for example, are inserted in more than 20 million hospitalized patients in the United States alone each year [ 8 ]. The mortality rate due to CVC-related bloodstream infections is estimated to be 12–25%, with additional healthcare costs in the order of $33,000 to $35,000 per event [ 9 ]. The predominant microorganisms associated with CVC-related infections are Staphylococcus epidermidis and Staphylococcus aureus where they are often found in biofilms upon removal of the devices [ 10 - 12 ]. Biofilm associated infections are difficult to treat due to the inherent antibiotic resistance of the sessile bacteria [ 4 , 6 , 13 ]. A number of factors contribute to this resistance such as a slow growth rate [ 14 , 15 ], failure of the agent to penetrate the biofilm [ 16 , 17 ], physiological changes and gene expression or repression due to the biofilm mode of growth [ 18 , 19 ]. Other factors such as age of the biofilms [ 20 ], production of extracellular polymeric substance (EPS) [ 21 - 23 ], and presence of biomaterials [ 24 ] also play a role in decreasing susceptibility of the bacteria within the biofilms to antimicrobial agents. Routinely, the diagnostic laboratories report the susceptibilities done on planktonic bacteria only. Although many studies have focused on the antimicrobial susceptibility of bacteria grown in biofilms [ 25 - 28 ], none of these studies included bacteria that disrupted from the biofilms. Disruption of the biofilm can occur during the removal of colonized catheters or during fluid infusion through them. The result is the entrance of bacteria or groups of bacteria shed from the biofilm into circulation causing bloodstream infections. These bacteria may not belong to either the planktonic or the biofilm phase and consequently may have a different pattern of antimicrobial susceptibility. For this reason, we studied the in vitro susceptibility of intact and disrupted biofilms of MRSA, MSSA and S. epidermidis to vancomycin, quinupristin/dalfopristin, and linezolid and compared it to the susceptibility patterns of the same bacteria in suspension. 1. Materials and Methods Unless otherwise indicated, all chemicals (analytical grade) were purchased from Sigma Chemical Co., St. Louis, Missouri, USA. Antibiotics Vancomycin (VAN) was purchased from Sigma Chemical Co., St. Louis, Missouri, USA. Quinupristin/dalfopristin (Q/D) was provided by Rhone-Poulenc Rorer, Collegeville, PA, USA and Linezolid (LNZ) was provided by Pharmacia & Upjohn, Kalamazoo, MI, USA. Microorganisms Ten isolates each of MRSA, MSSA, and S. epidermidis were used in this study. The microorganisms are clinical isolates from patients with blood stream infections which were provided by the microbiology laboratories at St. John's Hospital, Springfield, Illinois. These isolates were screened for biofilm formation on polystyrene microliter plates as previously described [ 27 ]. Antimicrobial susceptibility in suspension The minimum inhibitory concentrations (MICs) of the antibiotics were determined by using the broth microdilution technique as described by the National Committee for Clinical Laboratory Standards (NCCLS) [ 29 ]. The minimum bactericidal concentrations (MBCs) were determined by mixing the contents of each well at MIC and higher concentrations. Ten-microliter portions were then taken from each well and streaked onto the surface of blood agar. After 24 h incubation, the number of colony forming units per milliliter (CFU/ml) were counted and the MBCs, defined as the concentration at which 99.9% of bacteria was killed, were determined. The MIC 90 and MBC 90 obtained in susceptibility testing on planktonic bacteria were used in interpreting the results of the experiments with intact and disrupted biofilms. Biofilm formation and quantification To form biofilms, 100 μl portions of Tryptic Soy Broth (TSB) (Difco laboratories, Detroit, MI, USA) containing 1 × 10 6 CFU/ml of the microorganisms were delivered to flat bottom 96 polystyrene plates (Falcon No. 353072, Becton Dickinson and Company, Franklin Lakes, NJ, USA). After 24 h incubation at 37°C, the supernatants were aspirated and the remaining biofilms were washed twice with distilled water. TSB with or without the antibiotics at MIC values or at 50,500, or 1,000 μg/ml was added to the wells. Biofilms in the plates used for disrupted wells were then dislodged by using sterile wooden sticks and all plates were incubated again for 24 h. Plates with disrupted biofilms were centrifuged at 3,000 rpm for 15 min. to sediment the biofilm particles and all plates were then cautiously aspirated. The intact biofilms and the sediments of the disrupted biofilms were then determined by using a modified colorimetric assay previously described by Roslev & King [ 30 ]. On this assay, bacteria with an active electron transport system reduce the tetrazolium salt (redoxdye) to water soluble orange formazan product. Briefly, 100 μl lactate Ringers solution containing tetrazolium sodium 3'-{1- [(phenylamino)-carbonyl]-3,4-tetrazolium}-bis (4-methoxy-6-nitro) benzene sulfonic acid hydrate (XTT) (0.5 gm/L) and menadione (1 μM) was added to the intact biofilms and the disrupted biofilm sediments. The contents of the plates were mixed via plate shaker (Lab-Line Instruments Inc., Melrose Park, IL, USA) for 5 min followed by incubation for 1 hat 37°C in the dark. Plates with disrupted biofilm were first centrifuged for 15 min at 4°C, and the supernatants containing the soluble colored formazon were transferred to new plates. The intensity of the color was then measured via micro plate reader (Multiscan Plus, Thermosan Systems, Finland) at 490 nm and compared to that of drug-free wells. For plates with intact biofilms, the intensity of the color of the soluble formazan was measured directly and compared to drug-free wells. Confocal Scanning Laser Microscopy (CSLM) One-milliliter portions of TSB containing 1 × 10 6 CFU/ml of S. epidermidis isolate (SE6) were used to inoculate sterile plastic cover slips placed in a 4 well multidish (Nunc No. 176740, Roskilde, Denmark). After 24 h incubation at 37°C, the cover slips were moved to new plates and washed twice with distilled water. Fresh TSB (1 ml) with 500 μg/ml of vancomycin was added to the wells. Biofilms on cover slips designed to study the disrupted biofilms were then carefully dislodged. After incubation for another 24 h, the biofilms (intact and disrupted) were stained with LIVE/DEAD BacLight bacterial viability stain (Molecular Probes, Eugene, OR, USA) following manufacturer's instructions. The biofilms (intact and disrupted) were then examined by Olympus Fluoview CSLM (model IX 70, Olympus America Inc. NY, USA). In another set of experiments, the susceptibility of planktonic cells was examined by growing the bacteria in 1 ml portions of TSB. After 24 h, the bacterial suspensions were centrifuged at 10,000 rpm for 10 minutes, washed twice with sterile distilled water and finally dispersed in 1 ml portions of fresh TSB with the antibiotic. After another 24 h of incubation, the bacteria were stained and examined as previously mentioned. Statistical Analysis The mean and S.D. were calculated from the results of 10 isolates of each of the Staphylococcal species. One-way analysis of variance (ANOVA) was used to determine the differences between various antibiotic treatments. Tukey's pair comparison test was used at the chosen level of probability ( P < 0.05) to determine significance difference between means. Results In suspensions, all isolates were susceptible to all antibiotics tested (Table 1 ). At MICs, the antibiotics showed very little effect on the viability of bacteria within the biofilms (intact or disrupted). At higher concentrations (50,500 and 1000 μg/ml), the biofilms of all isolates were significantly (P < 0.001) less susceptible to the antibiotics compared to disrupted biofilms (Figures 1 , 2 , 3 ). Linezolid was less active than quinupristin/dalfopristin and vancomycin in killing the bacteria, especially in the biofilms. Table 1 Susceptibility of the tested isolates to vancomycin, quinupristin/dalfopristin and linezolid in suspension. Microorganism a Antimicrobial agents (μg/ml) Vancomycin Quinupristin/dalfopristin Linezolid Methicillin-susceptible S. aureus (MSSA) MIC Range 0.50–1 0.125–0.25 1–2 MIC 90 1 0.25 2 MBC Range 8–16 8 >64 MBC 90 8 8 >64 Methicillin-resistant S. aureus (MRSA) MIC Range 0.5–1 0.25–50 1–2 MIC 90 1 0.50 2 MBC Range 2–16 4–16 >64 MBC 90 8 16 >64 S. epidermidis MIC Range 2–4 0.06–2 0.50–1 MIC 90 2 0.50 1 MBC Range 2–16 0.25–16 >64 MBC 90 8 8 >64 a The MIC defined as the minimum concentration of antibiotic at which growth was completely inhibited while the MBC defined as the minimum concentration at which 99.9% of bacteria was killed. Figure 1 Susceptibility of intact and disrupted biofilms of MSSA, MRSA, and S. epidermidis at different concentrations of vancomycin. Figure 2 Susceptibility of intact and disrupted biofilms of MSSA, MRSA, and S. epidermidis at different concentrations of quinupristin/dalfopristin. Figure 3 Susceptibility of intact and disrupted biofilms of MSSA, MRSA, and S. epidermidis at different concentrations of linezolid. Quinupristin/dalfopristin showed the best activity against cells of the disrupted biofilms at concentrations above MICs. It was also more active than vancomycin against biofilms of both S. aureus and S. epidermidis at 50 μg/ml. Vancomycin at 500 and 1,000 μg/ml, was significantly more active against the biofilms of MRSA and S. epidermidis but not MSSA. Killing of the bacterial cells in intact or disrupted biofilms by quinupristin/dalfopristin and linezolid was independent of antibiotic concentrations over the range of 50–1,000 μg/ml, but for vancomycin, this was observed at higher concentration range (500–1,000 μg/ml). The ratios of viability of disrupted biofilms to that of intact biofilms were calculated for the isolates with each antibiotic concentration (Figure 4 ). The ratio values were similar for the three antibiotics at MICs. At other concentrations, the highest viability ratio was observed with linezolid and the lowest with quinupristin/dalfopristin. Figure 4 Viability ratios of MSSA, MRSA, and S. epidermidis of disrupted to intact biofilms at different concentrations of vancomycin, quinupristin/dalfopristin and linezolid. CSLM (Figure 5 ) demonstrated resistance of the intact biofilms to vancomycin, indicated by large number of viable cells, and resistance of the disrupted biofilm compared to the planktonic cells. It is also clear that the disrupted biofilm consists of clumps of larger size compared to that of the planktonic cells. Figure 5 CSLM images of S. epidermidis (SE6) intact biofilms (A), disrupted biofilm (B) and planktonic cells (C) on plastic coverslips after incubation for 24 h with 500 μg/ml of vancomycin. The bacterial cells were stained with LIVE/DEAD BacLight bacterial viability stain to directly visualize the effects of the antibiotic. The green fluorescence reflects processing of the dye by metabolically active cells while the red fluorescence is characteristic of dead cells. Note that while the green fluorescence was considerably more prominent in the intact biofilm image, the disrupted biofilm does display more green fluorescence than the planktonic cells. Also, note that the disrupted biofilm consists of large clumps and aggregates compared to the typical clusters of planktonic cells. Discussion Coagulase-negative staphylococci and S. aureus (mostly methicillin-resistant) are among the leading causes of nosocomial blood stream infections in the USA [ 31 ] with a crude mortality of 21–25% respectively [ 32 ]. S. aureus accounted for up to 13% of isolates recovered from patients with nosocomial infections from 1979 through 1995, and the percentage has increased in recent years [ 33 , 34 ]. It has been estimated that 65% of nosocomial infections are biofilm associated [ 5 , 35 ]. S. epidermidis is a common cause of blood stream infections associated with indwelling medical devices. S. aureus , in addition to causing blood stream infections, is a significant cause of tissue infections such as pneumonia and osteomyelitis [ 5 ]. Three phases of bacteria were used in this study; planktonic, biofilms and disrupted biofilms. Planktonic cells were used for determination of MICs and MBCs. This phase of bacteria has routinely been used as gold standard for determination of susceptibility of bacteria and prediction of clinical efficacy of antimicrobial agents. Planktonic cells grown in in vitro batch cultures are usually in nutrient-rich medium. Bacteria grown in biofilms differ greatly from the same organisms grown in suspensions by having different growth characteristics and taking up nutrients and drugs differently [ 36 , 37 ]. When a biofilm is at steady state, cells are shed from it at a constant rate [ 38 ]. These cells may enter into circulation and cause blood stream infection. Biofilm associated infections are 10 to 1,000 times more resistant to the effects of antimicrobial agents [ 5 , 35 , 39 ]. Bacteria are shed through biofilm disruption, which may result in entrance of biofilm pieces into circulation causing systemic infections. The question that needed to be answered is whether these cells behave as planktonic cells or as biofilm in regards to susceptibility to antimicrobial agents. We used 24 h incubation for interaction with antimicrobial agents in keeping with the procedures followed in diagnostic laboratories. Although it is possible that during this long incubation period the disrupted biofilm could have re-adhered to the polystyrene, we believe this was not the case. This is based on finding in our and other laboratories that antimicrobial agents are able to prevent the adherence of bacteria when exposed to them prior to formation of the biofilm [ 25 - 27 ] All isolates in suspension were susceptible to the antibiotics as determined by NCCLS guidelines. Vancomycin and quinupristin/dalfopristin were capable of killing 99.9 % of the bacteria in suspension at concentrations up to 16 μg/ml, while linezolid, a bacteriostatic antibiotic, did not show such effect even at the maximum concentration used. At the MICs, the antibiotics exerted little effect on the viability of the intact and disrupted biofilms. This was expected because even in suspension the MIC 90 s were 4–8 times lower then the MBC 90 s for vancomycin and 16–32 times lower for quinupristin/dalfopristin. At higher concentrations, although the intact biofilms were significantly more resistant than the disrupted biofilms, the recalcitrance of the latter was clear. The antibiotic concentrations were 100–4,000 times the MIC 90 s for quinupristin/dalfopristin and 25–1,000 times the MIC 90 s for vancomycin and linezolid. None of these concentrations were able to kill 99.9 % of the microorganisms in the disrupted phase. This was further demonstrated by examining the viability of S. epidermidis cells in the three phases in the presence of vancomycin by using CSLM. The susceptibility of the disrupted biofilm lies between the highly resistant biofilm and the susceptible planktonic cells. This may be attributed to the fact that disrupted biofilms consist of fragments that may retain some features of the intact biofilms. It is obvious from the images that the disrupted biofilm consists of large clumps and aggregates compared to the typical clusters of planktonic cells. The semi quantitative assessment of viability by CSLM was similar to those obtained by the colorimetric assay. Killing of the bacterial cells in intact or disrupted biofilms by quinupristin/dalfopristin and linezolid was independent of antibiotic concentrations over the range of 50–1000 μg/ml and over a higher range of vancomycin concentrations (500–1,000 μg/ml). Hamilton-Miller & Shah [ 40 ] found that killing of S. epidermidis in the biofilm by quinupristin/dalfopristin was independent of antibiotic concentrations over the range of 20–200 times the MIC. This lack of dose response at high concentrations favors the hypothesis that the bacteria disrupted from the biofilm are more or less similar to those in the biofilm rather than the bacteria in suspension. For better comparison, the viability ratios of the disrupted to the intact biofilms were calculated at different concentrations of the antibiotics. Linezolid was less efficient in killing bacterial cells in intact or disrupted biofilms which explains its highest viability ratio. Quinupristin/dalfopristin, with the lowest viability ratio, was more active against the cells of the disrupted biofilms at concentrations above MICs for both S. aureus and S. epidermidis . It was also more active than vancomycin against biofilms of both S. aureus and S. epidermidis at 50 μg/ml. On the other hand, vancomycin at 500 and 1,000 μg/ml, was significantly more active against the biofilms of MRSA and S. epidermidis but not MSSA. It has been reported that vancomycin accumulates at high concentration in the biofilms of gram positive bacteria especially S. epidermidis compared to linezolid [ 41 ]. This may be attributed to the ability of glycopeptides to bind to exopolysaccharides produced by the bacteria. However, such high concentrations of vancomycin or quinupristin/dalfopristin are not achievable in clinical practice. In general, our data show that Quinupristin/dalfopristin is more active than vancomycin and linezolid against the disrupted biofilms and there was no difference between methicillin-susceptible and methicillin resistant staphylococci. We conclude that the difficulty in treating the infections related to indwelling medical devices may not be only due to lack of eradication of the cells in the biofilm phase, but also due to resistance of bacteria disrupted from the biofilm. Authors' contributions ME did the biofilm work and participated with NK in the design of the study, review and interpretation of the data and discussion. SR and TK participated in the determination of MICs and MBCs of the antibiotics. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC546415.xml |
554777 | Out of Tanganyika: Genesis, explosive speciation, key-innovations and phylogeography of the haplochromine cichlid fishes | Background The adaptive radiations of cichlid fishes in East Africa are well known for their spectacular diversity and their astonishingly fast rates of speciation. About 80% of all 2,500 cichlid species in East Africa, and virtually all cichlid species from Lakes Victoria (~500 species) and Malawi (~1,000 species) are haplochromines. Here, we present the most extensive phylogenetic and phylogeographic analysis so far that includes about 100 species and is based on about 2,000 bp of the mitochondrial DNA. Results Our analyses revealed that all haplochromine lineages are ultimately derived from Lake Tanganyika endemics. We find that the three most ancestral lineages of the haplochromines sensu lato are relatively species poor, albeit widely distributed in Africa, whereas a fourth newly defined lineage – the 'modern haplochromines' – contains an unparalleled diversity that makes up more than 7% of the worlds' ~25,000 teleost species. The modern haplochromines' ancestor, most likely a riverine generalist, repeatedly gave rise to similar ecomorphs now found in several of the species flocks. Also, the Tanganyikan Tropheini are derived from that riverine ancestor suggesting that they successfully re-colonized Lake Tanganyika and speciated in parallel to an already established cichlid adaptive radiation. In contrast to most other known examples of adaptive radiations, these generalist ancestors were derived from highly diverse and specialized endemics from Lake Tanganyika. A reconstruction of life-history traits revealed that in an ancestral lineage leading to the modern haplochromines the characteristic egg-spots on anal fins of male individuals evolved. Conclusion We conclude that Lake Tanganyika is the geographic and genetic cradle of all haplochromine lineages. In the ancestors of the replicate adaptive radiations of the 'modern haplochromines', behavioral (maternal mouthbrooding), morphological (egg-spots) and sexually selected (color polymorphism) key-innovations arose. These might be – together with the ecological opportunity that the habitat diversity of the large lakes provides – responsible for their evolutionary success and their propensity for explosive speciation. | Background "At some stage in the past the waterways of Africa were, from the fishes' point of view, accessibly interconnected." P. H. Greenwood (1983) With estimated numbers of about 1,000 and 500 species respectively the assemblages of cichlid fishes of East African lakes Malawi (LM) and Victoria (LV) are, by far, the most species-rich species flocks [ 1 - 6 ]. In Lake Tanganyika (LT), Africa's oldest lake, "only" about 200 to 250 cichlid species occur, but they are phenotypically, behaviorally and genetically more diverse [ 7 - 9 ]. Remarkably, almost 100% of the species of all of these flocks are endemics and the three East African Great Lakes do not have a single cichlid species in common [ 10 ]. While in LT at least twelve eco-morphologically distinct tribes can be clearly distinguished phylogenetically [ 7 , 9 , 11 ], the species flocks of LM and LV are entirely comprised of cichlids assigned to only one of these tribes, the Haplochromini [ 1 , 4 , 10 , 12 ]. Additionally, several less species-rich flocks of haplochromines are found in smaller lakes in East Africa [ 10 ], some of which have been combined with the LV radiation into a "superflock" of closely related species [ 4 , 12 , 13 ]. Not all haplochromines are lacustrine, however, and some 200 species inhabit rivers and occur in northern-, eastern-, southern- and central-Africa but are virtually absent from West Africa [ 10 , 12 ] (Fig. 1 ). Traditionally, it was believed that riverine haplochromines seeded the cichlid radiations in all East African Great Lakes [ 8 , 10 ]. Molecular-based phylogenies indeed uncovered haplochromine lineages that are ancestral to the species flocks of LM and LV (see e.g. , [ 1 , 4 , 9 , 14 ]. In contrast, the primary radiation of LT's cichlid assemblage was found to predate the origin of haplochromine lineages [ 9 , 15 , 16 ] suggesting a close phylogenetic affinity of the ancestor(s) of the haplochromines to other tribes in LT [ 9 ]. However, only a relatively small fraction of the genetic and geographic diversity of the haplochromines has been included in previous phylogenetic and phylogeographic studies. Figure 1 Distribution of the major haplochromine lineages in Africa with special emphasis on the East African lakes (according to our phylogenetic and phylogeographic analyses and references [4, 12, 13, 34]). Note that the color scheme is carried throughout this publication. That the haplochromines have a particularly strong propensity for explosive speciation among cichlids is best illustrated by their unparalleled species-richness and diversity. With the exception of two LT lineages (Lamprologini: ~70–100 species; Ectodini: ~25–30 species), discrete adaptive radiations of non-haplochromines are comprised by about a dozen species at most. Thus, which evolutionary novelties might be causally related to the explosive speciation that distinguish the haplochromines from all other cichlid lineages remains a crucial question for the understanding of the explosive patterns of cichlid evolution in East Africa. Haplochromines furthermore represent prime examples for parallel evolution, and it is particularly the lacustrine haplochromine species flocks that independently evolved morphologies and color-patterns that are convergent between species of different lakes [ 2 , 17 , 18 ]. In order to gain a deeper understanding of the adaptive radiations of cichlids in general and the formation of East African cichlid species flocks in particular, several crucial questions still remain to be answered: (i) What evolutionary lineages from which geographic ranges make up the diversity of haplochromines? (ii) Which were the founding lineages of the lacustrine adaptive radiations and were they riverine generalists? (iii) Can particular behavioral and/or morphological key-innovations be identified that might be causally related with the diversification of haplochromines? With the aim of addressing these questions, we conducted the most extensive phylogenetic and phylogeographic study of haplochromine cichlids so far, analyzing a portion of up to 2,000 bp of the mitochondrial genome of about 100 species. We included representatives of relevant cichlid tribes from LT [ 7 , 9 , 15 ] as well as members of all major riverine and lacustrine haplochromine lineages and all but one haplochromine genus (following [ 12 ]). Results The neighbor-joining analysis based on 304 complete mitochondrial control region sequences (Fig. 2 ) could not resolve the phylogenetic relationships between haplochromine lineages and the LT lineages with convincing bootstrap support. However, this analysis, which primarily aimed to provide a basis for the selection of taxa for the second set of analyses, already indicated the existence of a monophyletic clade that is comprised by the Tanganyikan Tropheini, sister-group to a clade consisting of the LM cichlids plus several East African riverine and lacustrine lineages and the representatives of the LV region superflock [ 4 ]. Figure 2 50% majority-rule bootstrap consensus tree of 304 cichlid fish specimens based on 862 bp of the mitochondrial control region (neighbor-joining, HKY85 model with gamma substitution correction, 5,000 replicates). Relevant bootstrap values are depicted on the respective branches. The branches colored in red indicate the taxa that were chosen for the phylogenetic analyses combining the control region with sequences of the NADH dehydrogenase subunit II gene (see Fig. 3; for Ctenochromis oligacanthus from GenBank no control region sequence was available). The colors of the boxes that indicate the major clades refer to Figs. 1 and 3, the labels of the clades correspond to Table 1 [see Additional file 1 ]. Note that Haplochromis bloyeti (marked by an asterisks) had a control region sequence identical to H. sp. 1533 of [25], which was collected in the Malagarasi River, and grouped – together with other fishes from the Malagarasi area and from the Lake Edward/George region – into their group VII. Likewise, our H. sp. Tanzania I (marked by a circle) was identical to H. sp. 1738 of [25], which was collected in Lake Chala and clustered with other taxa from Tanzania into their group VI. In addition to Haplochromis gracilior (endemic to Lake Kivu), which was recently identified as close relative of the Lake Victoria superflock [4], we found another sister group to the superflock. This lineage includes Haplochromis paludinosus that occurs in the Malagarasi, as well as undescribed species from Tanzania and Lake Edward (see also Fig. 3). Like Haplochromis gracilior from Lake Kivu, all these taxa have the diagnostic character state 'Adenine' in position 630 of the control region alignment and root to the Lake Victoria superflock through the central rift valley haplotype [4], corroborating the view that Lake Kivu is the main reservoir from which the Lake Victoria superflock evolved [4]. Figure 3 Maximum likelihood phylogeny of the haplochromine cichlids [general time-reversible model with gamma correction] based on 100 taxa. Numbers above the branches represent Bayesian posterior probabilities obtained with Mr. Bayes, numbers below the branches represent maximum-likelihood bootstraps (100 replicates, obtained with PAUP*). Boulengerochromis microlepis and Oreochromis tanganicae , two tilapiine cichlids in LT, were used as outgroup taxa [9, 15, 16, 63]. In accordance to previous studies [9, 15] we find that within the LT species flock the Eretmodini (Ere.) are placed as sister group to the Lamprologini (Lampr.) plus several LT tribes ("Tanganyika") including the Orthochromis assemblage from the Malagarasi plains, and the haplochromines sensu lato . The latter clade combines four distinct lineages, a Congolese/South-African- (CSA; ~150 species), the Pseudocrenilabrus - (3 species), the Astatoreochromis -lineage (3 species), and the modern haplochromines (~1,800 species). The modern haplochromines combine the LT Tropheini, the species flock of LM, several riverine lineages as well as the LV region superflock according to [4]. The haplochromines sensu lato are characterized by their breeding behavior; true egg-spots ( ocelli ) are likely to have evolved in the ancestor of the Astatoreochromis -lineage and the modern haplochromines. By contrast, the Malagarasi River Orthochromis are biparental caregivers [34] providing behavioral support for our molecular-based classification that excluded these fish from the haplochromines sensu lato . We note that several genera are polyphyletic and major taxonomic revisions will be required in the future to take our phylogenetic results into consideration. For example, Orthochromis of the Malagarasi River plains form a clade outside the remaining haplochromines in close affinity to the LT Ectodini (see also [9]) whereas Orthochromis polyacanthus and O. stormsi , which share derived features [12], fall – in accordance to their distribution – into the Congolese/South African clade. Other polyphyletic genera are Astatotilapia , Ctenochromis , and Haplochromis . The grey arrows next to some species names refer to the pictures on the right, the asterisk symbol marks the ancestor of the haplochromines sensu lato , the circle symbol marks a "piebald" ("orange blotched") form as found in the modern haplochromines only. In the maximum likelihood tree of the dataset including 100 taxa (Fig. 3 ) the LT Eretmodini were placed as sister group to the LT Lamprologini, followed by a clade comprised by the two representatives of the Limnochromini ( Limnochromis auritus , Triglachromis otostigma ) plus Cyphotilapia frontosa , and all remaining taxa. Among these, a clade that includes four LT tribes (Cyprichromini, Ectodini, Limnochromini, Perissodini) plus the Orthochromis species from the Malagarasi drainage, which had so far been considered to belong to the Haplochromini, was recovered as sister group to the remaining haplochromine representatives. The latter, the haplochromines sensu lato , clustered into four distinct groups, with a Congolese/South-African lineage (CSA) sister to the Pseudocrenilabrus -, the Astatoreochromis -lineage, and the 'modern haplochromines' ( i.e. a clade comprised by the LT Tropheini sister to the LM representatives, several East-African riverine lineages and the members of the LV region superflock). Bayesian inference revealed the same branching order for the different lineages. In the strict consensus topology of the 78,617 most parsimonious trees (unweighted tree length: 4400; tree not shown), the Eretmodini were again resolved as sister group to the Lamprologini, the Limnochromini plus C. frontosa , four LT tribes plus the Malagarasi Orthochromis and the haplochromines sensu lato . Here, the Astatoreochromis -lineage was resolved as most ancestral lineage, and as sister group to a clade comprised by the Pseudocrenilabrus - and the CSA lineage, and the modern haplochromines. Also, the neighbor-joining analysis recovered a tree (not shown) with the Eretmodini as sister-group to the Lamprologini, the Limnochromini plus C. frontosa , the four LT tribes including the Malagarasi River Orthochromis species, and the haplochromines sensu lato . In the neighbor-joining tree, the Pseudocrenilabrus -lineage occupied the most ancestral branch in the haplochromines sensu lato , followed by the Congolese/South-African clade, the Astatoreochromis -lineage and the modern haplochromines. A Shimodaira-Hasegawa test [ 19 ] revealed that there are no significant differences between the topologies obtained with the different algorithms (P < 0.05). Similarly, in the four-cluster likelihood mapping analysis [ 20 ] none of the three possible alternative branching orders among the four main lineages of the haplochromines sensu lato received support greater than 50%. The dating of the major cladogenetic events (Fig. 4 ) found an age of 2.4 MYA (1.22 – 4.02 MYA) for the most recent common ancestor of the four lineages of haplochromines, and about 1.8 MYA (0.66 – 3.78 MYA) for the most recent common ancestor of the modern haplochromines. This latter value was smaller in all resampling replications pointing to a younger age of the modern haplochromines compared to the remaining three lineages of haplochromines sensu lato . The most recent common ancestor in the CSA-clade was estimated to have lived about 2.0 MYA (1.15 – 3.89 MYA), the split of the CSA lineage from the common ancestor with the Pseudocrenilabrus -, Astatoreochromis -, and modern haplochromine lineage was dated to about 2.4 MYA (1.22 MYA – 4.02 MYA). The test for shifts in the probabilities of speciation in the haplochromines sensu lato according to [ 21 ] suggested increased rates of lineage diversification ( p c < 0.01) along three branches in the maximum likelihood, the Bayesian inference and the neighbor-joining trees: (i) the branch leading to Astatoreochromis -lineage plus the modern haplochromines; (ii) the branch leading to the modern haplochromines; and (iii) the branch leading to the LV region superflock (see Fig. 4c for the maximum likelihood tree). In the maximum parsimony strict consensus tree an increased rate of lineage diversification was found: (i) for the branch leading to the modern haplochromines; and (ii) for the branch leading to the LV region superflock. Figure 4 The "out of Tanganyika" scenario of haplochromine evolution in Africa . ( a ) Several haplochromine lineages independently left Lake Tanganyika and colonized large parts of Africa via past and present river connections. Some of these lineages seeded cichlid radiations in distant lakes. The phylogeographic scenario is in agreement with palaeo-geological reconstructions of the evolution of the East African Rift region. LT is the oldest of the rift lakes. Its central basin began to form between 9 and 12 MYA, the northern (8-7 MYA) and the southern basin (2–4 MYA) began to fill at later periods [76]; deepwater conditions exist since about 5–6 MYA [78]. LM (2–4 MYA) and LV (750,000 years) are considerably younger. ( b ) Proposed lake level of Lake Tanganyika during the last four million years [76, 77, 79] indicating major low- and high-stands. ( c ) Chronogram of the haplochromine evolution in Africa as reconstructed with r8s [72, 73] based on the maximum likelihood topology. The size of each clade represents its species number. The modern haplochromines are a recent and rapidly speciating lineage. Our molecular clock calibration suggested about 2 MYA (1.15 – 3.89 MYA) for the most recent common ancestor in the Congolese/Southern African lineage and ca. 2.4 MYA (1.22 – 4.02 MYA) for their split from the common ancestor with the Pseudocrenilabrus -, Astatoreochromis -, and modern haplochromine lineage. This lies in the range of the proposed high lake-level stand of LT between the minima at 3.5 MYA and 1.1 MYA (650–700 m below present level) [76, 77] making an overflow through the Lukuga valley possible, thus opening the connection between LT and the Congo drainage. The asterisks mark nodes with a significant burst of lineage diversification ( p c < 0.01) [21]. Aa... Astatoreochromis alluaudi , CSA... Congolese/South African lineage, Hb... Haplochromis bloyeti , LMF... Lake Malawi species flock, LVS... Lake Victoria Region Superflock, MO... Malagarasi Orthochromis assemblage, Pm... Pseudocrenilabrus multicolor , Pp... Pseudocrenilabrus philander , Ps... Pseudocrenilabrus- lineage, Tr... Tropheini. The maximum parsimony and maximum likelihood character state reconstructions revealed that the characteristic maternal mouthbrooding behavior, where only the females incubate their fry in their buccal cavities, evolved in the common ancestor of the CSA lineage, the Pseudocrenilabrus -, the Astatoreochromis -lineage, and the modern haplochromines (see asterisk in Fig. 3 ). The true haplochromine-like egg-spots [ 10 , 22 , 23 ] are likely to have evolved in the common ancestor of the Astatoreochromis -lineage and the modern haplochromines (Fig. 3 ). Some members of the CSA lineage also show yellow or reddish markings on their anal fin. However, these markings do not represent real haplochromine-like egg-spots, which are characterized by an inner yellow or orange ring and an outer transparent and colorless ring [ 10 , 22 , 23 ]. Instead, the more homogenous markings seen in the CSA lineage might be viewed as an intermediate character state in the evolution of the species-specific egg-spots as found in the Astatoreochromis -lineage and in the modern haplochromines. The mapping of riverine versus lacustrine lifestyle onto the maximum likelihood topology suggested that the ancestor of the modern haplochromines was riverine (Fig. 5 ). Figure 5 Maximum parsimony reconstruction of habitat type (lake versus river) using MacClade (the maximum likelihood reconstruction with Mesquite revealed analogous results) based on the maximum likelihood topology (see Fig. 3). The ancestor of the modern haplochromines (marked by a circle) is likely to have been a riverine species. This implies that also the Tanganyikan Tropheini originated from a riverine ancestor and re-colonized the lake where they presently form an abundant group in the rocky littoral zones. Discussion Phylogeny and evolutionary origin of haplochromines Mitochondrial DNA (mtDNA) sequence data have a long and successful history in the study of East African cichlid evolution (see e.g., [ 1 , 4 , 9 , 17 , 24 , 25 ]). Limitations with mtDNA have only been encountered when focusing on the phylogeny among extremely closely related species due to the possibility of the persistence of ancestral polymorphism (see e.g., [ 26 ]), or because of hybridization events (see e.g., [ 27 - 29 ]). Nevertheless, mtDNA sequences proved to be particularly suitable for the reconstruction of the East African cichlid phylogeny at the tribal level and for tribal assignments [ 9 , 15 ]; for phylogenetic reconstruction within older tribes [ 30 , 31 ]; and for phylogeographic analyses [ 4 , 32 ]. Also, because of the extremely fast rate of lineage formation in cichlids, nuclear and even some mitochondrial genes [ 1 ] are too slowly evolving to contain phylogenetic information (reviewed in [ 6 ]). The different phylogenetic algorithms, with which we analyzed our data, revealed largely congruent results. In all analyses, and in agreement with previous results [ 9 ], we found that the Eretmodini are placed as sister group to the substrate spawning Lamprologini – with an estimated number of up to 100 species the most species-rich tribe in LT – plus several LT tribes and all haplochromine representatives (see Fig. 3 for the maximum likelihood tree). The molecular phylogenies thus corroborate that all haplochromines are ultimately derived from LT cichlids and that their ancestor(s) are likely to have left LT secondarily. We consider all species that belong to the monophyletic group descending from this ancestor (asterisk in Fig. 3 ), to being haplochromines sensu lato . These are further divided into four distinct groups, a Congolese/Southern African lineage (CSA), the genera Pseudocrenilabrus (Ps.) and Astatoreochromis (As.), and the modern haplochromines (MH). While the respective monophyly of these four lineages was supported by high bootstrap values and Bayesian posterior probabilities, our analyses could not unambiguously resolve the exact relationships between these four lineages [maximum likelihood and Bayesian inference (see Fig. 3 ): (CSA, (Ps., (As., (MH)))); maximum parsimony: (As., ((Ps., CSA), (MH))); neighbor joining: (Ps., (CSA, (As., (MH))))]. An evaluation of these alternative hypotheses by means of a Shimodaira-Hasegawa test [ 19 ] and a four-cluster likelihood mapping analysis [ 20 ] indicated that none of these alternative branching orders receives significantly more support than the others. This suggests that the four lineages of the haplochromines sensu lato evolved almost contemporaneously from a common ancestor. This is further supported by the observation of relatively short branches interrelating these four lineages and the generally low bootstrap values and Bayesian posterior probabilities supporting the corresponding relationships proposed by the different algorithms. The CSA lineage is composed of several widespread and moderately species-rich groups of the Congo drainage and Southern Africa and consists of two main clades: A clade with a species from the Congo drainage ( Ctenochromis oligacanthus ) ancestral to Southern African genera ( Serranochromis , Sargochromis , Pharyngochromis ) was resolved as sister group to a clade comprised by solely Congo drainage taxa ( Orthochromis , Haplochromis sp. nov., Cyclopharynx and Thoracochromis ). Within the Pseudocrenilabrus - and Astatoreochromis -clades, branch lengths were relatively short and a more detailed phylogeographic sampling would be necessary to resolve the relationships between the different geographic lineages. The modern haplochromines consist of species flocks of an unparalleled diversity. They include the endemic LT tribe Tropheini (~25 species) sister-group to a clade comprised by the entire species flock of LM (~1,000 species) and several East African riverine and lacustrine lineages (~200 species) plus the LV region superflock (~600 species). With approximately 1,800 species this – here phylogenetically defined – monophyletic lineage makes up about 7% of all known teleost fish. Astatotilapia , Thorachochromis , and Orthochromis are polyphyletic genera Our phylogenies show that several genera are in fact polyphyletic, and major taxonomic revisions will be required in the future to take our phylogenetic results into consideration. For example, and in agreement with previous studies (see e.g. , [ 4 , 14 , 33 ]), Astatotilapia emerges as polyphyletic genus, with representatives assigned to both the East African riverine clade in the modern haplochromines and the LV region superflock. The genus Thoracochromis , represented in our analysis by T. brauschi , has also been shown to be polyphyletic before [ 14 ], with T. brauschi from the Congo drainage as a more ancestral lineage, and T. petronius and T. pharyngealis from the Nile drainage with affinities to the LV region superflock (note that the Nile River Thoracochromis of [ 14 ] are consequently listed as Haplochromis in [ 34 ]). The placement of T. brauschi as sistergroup to T. petronius and T. pharyngealis plus the remaining representatives of the LV region superflock in the AFLP based phylogeny of [ 14 ] seems to contradict our mtDNA based results in which T. brauschi was identified as member of the CSA lineage. However, the reported branching order did not receive considerable bootstrap support in [ 14 ]. Also, the choice of Astatoreochromis alluaudi (mislabeled as Astatotilapia alluaudi in [ 14 ]) as single outgroup species seems problematic, as our present analyses (see above) and former results [ 4 ] indicate that A. alluaudi and not T. brauschi is more closely related to the modern haplochromines (and, thus, also to the LV region superflock). Further analyses including nuclear DNA sequence data and more taxa assigned to the genus will be necessary to address this problem. Based on the phylogeny it is apparent that the Orthochromis lineage, which is confined to the Malagarasi River system and two isolated rivers, East of LT [ 35 ], is not part of the radiation of haplochromine cichlids. This is further supported by the breeding behavior of these fish: While the Malagarasi River Orthochromis are biparental caregivers [ 35 ], the haplochromines sensu lato are all maternal mouthbrooders. Thus, the Orthochromis -species from the Malagarasi area, a group of exclusively riverine fish, should be placed into its own tribe. The name Orthochromis is, however, also used for riverine species from the Congo drainage. In our analyses, Orthochromis polyacanthus and O. stormsi from the Congo River system fall – according to their distribution – into the CSA-clade leaving also the genus Orthochromis polyphyletic (see also [ 9 ]). We suggest to using Orthochromis for the CSA lineage representatives, since O. polyacanthus was the first species of the genus to be described [ 36 ], and Schwetzochromini (as tribe name) and Schwetzochromis (as genus name) for the Malagarasi area species, since this name was repeatedly used for some species of that complex (see e.g. , [ 9 , 34 , 35 ]). Phylogeography and phylochronology We note that the application of a molecular clock for estimating divergence times has the potential of not being without problems for several reasons (see e.g. , [ 37 , 38 ]). However, a molecular-clock-based time estimate does surely provide an approximate framework for phylogeographic inferences. Our phylogeographic scenario (Fig. 4a ), which is derived from the maximum likelihood phylogenetic and molecular clock analyses, suggests that several lineages independently left LT to colonize surrounding river systems and consequently other lakes in East Africa. The molecular clock calibration based on the chronogram generated with r8s [ 39 ] (Fig. 4b,c ) yielded about 2.4 MYA (1.22 – 4.02 MYA) for the split of the CSA lineage from the common ancestor of the haplochromines sensu lato , and about 2 MYA (1.15 – 3.89 MYA) for the first branching events within the CSA clade. The spread of Congo drainage taxa into southern river systems occurred at a later stage, most likely at the relatively shallow watershed between upper branches of the Congo River and the Zambezi River – a scenario that is also supported by the placement of Serranochromis sp. (from Lake Mweru in the upper Congo) as sister group to the Zambezi/Southern African genera Sargochromis and Pharyngochromis in our phylogenies. However, further sampling in that area would be necessary to reconstruct the southward spread of the CSA lineage. At essentially the same time as the CSA lineage, the ancestors of the Pseudocrenilabrus- and the Astatoreochromis- lineage diverged from their common ancestor. Despite their large distributional ranges – they also colonized the LV (both lineages) and LM ( Pseudocrenilabrus ) region – the genera Astatoreochromis and Pseudocrenilabrus never underwent considerable speciation. The three described Astatoreochromis species occur in the LV region including Lakes Edward and George ( A. alluaudi ), in the rivers Rusizi and Lukuga ( A. straeleni ), and in the Malagarasi River ( A. vanderhorsti ). The three species of Pseudocrenilabrus occur from the Nile system to the LV region ( P. multicolor ), in Eastern- and Southern Africa including LM ( P. philander ), and in the central Congo basin ( P. nicholsi ). All analyzed representatives are relatively closely related suggesting a recent spread of these lineages in East Africa. However, we did not include P. nicholsi , which is morphologically different from P. multicolor and P. philander and would – if it really belonged to the Pseudocrenilabrus -lineage – represent the only haplochromine in the Congo drainage that is not a member of the CSA lineage. The most recent common ancestor of the modern haplochromines was dated to have existed about 1.8 MYA (0.66 – 3.78 MYA) (Fig. 4 ). This ancestral lineage forms the crucial phylogenetic and biogeographic link between the species flocks of all three East African Great Lakes, and its discovery documents the existence of much earlier hypothesized fish-accessible waterways between these waterbodies [ 40 , 41 ]. Apparently, the Malagarasi River (and possibly the Rusizi) played a major role for the dispersal of these fishes, since many modern haplochromine lineages occur in these drainages and in lakes South-Eastern and North of LT exclusively, which argues against the view that LM haplochromines originated from Zambezi River stocks [ 41 ]. Whether or not Lake Rukwa has ever acted as link between the faunas of LT and LM [ 41 ] cannot be answered by our data. Lake Rukwa seems to have overflowed at its maximum levels into LT several times. However, Lake Rukwa has also become very shallow in recent geological times and it might have dried up completely [ 41 ] eradicating its original fauna. At present, Lake Rukwa harbors haplochromines that belong to the East-African riverine clade in Figs. 2 , 3 [ 4 , 25 ]. Our analyses also recovered another closely related lineage to the LV region superflock in the East-African riverine clade, in addition to Haplochromis gracilior form Lake Kivu [ 4 ]. This lineage includes H. paludinosus that occurs in the Malagarasi (which was already suggested by [ 33 ]), as well as undescribed species from Tanzania and Lake Edward (Figs. 2 , 3 ). It is, however, unclear by which waterway haplochromine cichlids once colonized Lake Kivu. The flow of the Rusizi, presently from Lake Kivu into LT with the Panzi falls as strong barrier for fish migration, might actually have been reversed before the uplift of the Virunga volcanoes north of Lake Kivu as suggested by deposits of fossil LT mollusks and fluviatile sands in the upper Rusizi valley [ 41 , 42 ]. This connection could possibly explain how haplochromines of LT origin were able to colonize Lake Kivu about 1.5 million years ago (Fig. 4 ). Evolutionary key-innovations of haplochromines One of only few synapomorphies of the haplochromines sensu lato is the particular type of cranial apophysis for the upper pharyngeal bones [ 12 ]. The distinctive organization of the pharyngeal apophysis, a second set of jaws that is functionally decoupled from the oral ones [ 43 ], is characteristic to all cichlids and has been interpreted as prominent feature that – because of its adaptability – contributes to the cichlids' evolutionary success [ 2 , 3 , 10 , 43 ]. It is, however, not evident how the relatively minor morphological modification of part of that structure in the haplochromines [ 12 ] might function as an evolutionary key-innovation. Interestingly, however, all haplochromines sensu lato are maternal mouthbrooders with the females alone incubating the eggs in their buccal cavities [ 10 , 12 ]. Mouthbrooding, which is regarded as an adaptation to predation pressure [ 44 - 46 ], has evolved several times independently and in diverse behavioral modes in cichlids [ 10 , 22 , 47 , 48 ]. The characteristic maternal mouthbrooding behavior displayed by haplochromines is believed to being a derived character state [ 35 , 46 , 47 ]. Mouthbrooding strongly limits the number of eggs and fry that can be raised and might have led to generally much smaller population sizes, which has, for example, population genetic implications on fixation of alleles, and might result in smaller effective population sizes. Furthermore, mouthbrooding species may be considered to being promising colonizers of new habitats, since only a single mouthbrooding female is necessary for the founding of a new population. An eminent feature of several female mouthbrooding cichlid genera is the occurrence of egg-spots on the anal fins of males. In some species also females show such ovoid markings, but these are smaller and much less conspicuous than in males. Also, some species of the modern haplochromines, e.g. , some deep-water lineages of LM, have lost their egg-spots secondarily. In mimicking real eggs to attract females, these egg-spots function as natural releasers [ 22 , 23 ], or intra-specific sexual advertisement [ 44 ], apparently serving to ensure a greater fertilization success of the eggs by bringing about greater proximity of the female's mouth to the male's genital opening. Based on the molecular phylogeny, we could trace the origin of the characteristic egg-spots ( ocelli ) [ 10 , 23 ] to the common ancestor of the Astatoreochromis -lineage and the modern haplochromines. There are other cichlid species in which males display yellow or red marks on their pelvic, dorsal or anal fins, but only in these lineages true egg-spots on the males' anal fins with a yellow, orange or red center and a colorless/transparent outer ring [ 10 , 22 ] are found. Interestingly, the branch leading to the Astatoreochromis -lineage and the modern haplochromines is the one with a pronounced potential for an increased rate of speciation (see Fig. 4c ). Based on the character state reconstructions (Fig. 5 ) it seems likely that this ancestor was riverine. Thus, it may be concluded that the egg-spots first evolved in a haplochromine cichlid that inhabited a turbid riverine environment, where these conspicuous markings would seem to be particularly effective and necessary for intra-specific communication. Another innovation that further distinguishes the exceptionally species-rich modern haplochromines from all other cichlids is the occurrence of numerous color morphs, often accompanied by sexual color dimorphism. Inter- and intra-specific polychromatism combined with maternal mouthbrooding involving egg-spots as releasers can be hypothesized to being permissive for sexual selection through female choice and, hence, the haplochromines' propensity for species formation, as sexual selection is probably a major causal factor in the origin of isolating mechanisms and the maintenance of reproductive isolation [ 18 , 49 - 53 ]. These distinctive features of the modern haplochromines, that have arisen just in their ancestor, in combination with the numerous ecological niches that are provided by the large East African lakes might thus have induced a considerable increase of the haplochromines' evolutionary potential. The importance of large waterbodies for the evolution of the modern haplochromines is reflected by the fact that these cichlids only radiated in lakes (and species number rather correlates to the size, but not to the age, of a lake), whereas the riverine lineages are all species-poor albeit often widespread (Figs. 1 , 3 ). Replicate adaptive radiations of the 'modern haplochromines' A common feature of many adaptive radiations is that their founders are believed to have had a more generalist's lifestyle, while adaptive radiations themselves are defined by being composed of highly specialized species with narrower niche widths [ 54 , 55 ]. Theory predicts that generalists more likely have better dispersal abilities and are expected to be able to adapt readily to novel environmental settings [ 55 ]. A generalist ancestor scenario fits well with the diversification of haplochromine cichlids. Generalist riverine species of the genera Astatoreochromis , Astatotilapia , Pseudocrenilabrus , and Haplochromis ( e.g. , bloyeti ), are ancestral to the adaptive radiations of the Tropheini of LT, and/or the radiations of LM and the LV region superflock. These genera are widely distributed and not confined to Eastern Africa, and they are the only ones that could inhabit the waterways that – over geological time spans – connected the lakes of Eastern Africa. The phylogeny presented here (Fig. 3 ) reveals that modern haplochromines gave rise to several major adaptive radiations; the most prominent ones are those of LM and LV. Interestingly, it uncovers that also the radiation of the Tropheini from LT [ 30 ] must now be considered as an additional radiation of the modern haplochromines, corroborating the much older perception that LT accommodates several independent species flocks [ 56 ]. It is further suggested by mapping the fishes' lifestyle onto our molecular phylogeny that the highly specialized Tropheini are descendents of a river-living species. This implies that the ancestor of the Tropheini successfully re-entered the lake habitat and evolved into the presently dominant group in the rocky littoral zone of LT. Thus, this lineage of modern haplochromines managed to occupy "empty niches" in an apparently "full" ecosystem, as all remaining tribes, which now account for about 200 species, had already been established when the ancestor of the Tropheini secondarily entered LT (Figs. 2 , 3 ). The observation that these fish underwent an independent adaptive radiation in LT underlines the haplochromines' propensity for speciation. In an apparent contrast to most other known examples of adaptive radiations [ 55 ] is the finding that the generalist ancestors of the haplochromine species flocks were derived from already highly diverse and specialized LT endemics (Fig. 3 ). Therefore, specialization may not be an "evolutionary one-way street", but rather some lineages have reversed their level of specialization, i.e. , generalists arose from highly specialized lineages, yet, apparently retained their high propensity for speciation and level of evolvability (see [ 57 ]). The faunal revolution of LT's radiation of cichlids was thus not confined to the lake habitat itself (see also [ 8 , 9 , 58 ]), but it effectively involved large parts of Africa via the intermediate step of repeatedly evolving generalist riverine lineages – in much the same way as the adaptive radiation in LV produced haplochromine species that secondarily colonized surrounding rivers [ 4 ]. Our phylogeny of haplochromines provides strong support for replicate adaptive radiations in East African cichlids. The concept of replicate radiations, in which the same sequence of adaptations to ecological niches evolved repeatedly in lineages that inhabit similar environments, has been developed based on sympatric species pairs of fishes in postglacial lakes and on the Anolis lizard ecomorphs on different islands of the Greater Antilles [ 59 - 61 ]. Our inclusive phylogenetic and phylogeographic study shows that similar ecological types of cichlids in the different East African lakes evolved independently (see also [ 2 , 17 ]), yet it also shows that the convergent ecotypes in the species flocks of LM, LV, Lake Kivu as well as in the Tropheini [ 2 , 10 , 12 , 13 , 17 ] arose from the same ancestral phenotype in the ancestor of the modern haplochromines. We suggest that a combination of behavioral (maternal mouthbrooding) and morphological innovations (egg-spots, color polymorphisms, pronounced sexual dichromatism) as well as ecological opportunities (after the colonization of large lakes) might have predestined this particular lineage to give rise to these replicate adaptive radiations. It has been noted before that lineages of LT origin have left the lake secondarily (there are, for example, about five lamprologine species that are found in the Congo and Malagarasi Rivers) [ 9 , 58 ]. Here, we show that the entire haplochromine diversity has its origin in LT corroborating the view that ancient lakes not only preserve biodiversity but also act as biodiversity hotspots, genetic reservoirs and cradles from which new lineages evolve [ 4 , 8 , 9 ]. What remains to be answered is where the LT cichlids originated and to what extent a proposed and meanwhile desiccated Pliocene lake in the Congo plains [ 41 , 42 ] was the source of the ancient LT lineages, pushing back even further the onset of replicate adaptive radiations in East African cichlids. Conclusion Our phylogenetic analyses that include representatives of all major haplochromine lineages show that all haplochromines are derived from Lake Tanganyika cichlids. While the Orthochromis species of the Malagarasi area apparently do not belong to the radiation of the haplochromine cichlids and should be placed into a new tribe, we defined four new lineages within the haplochromines sensu lato : A clade combining Congolese and South-African genera (CSA-lineage), the Pseudocrenilabrus -, the Astatoreochromis -lineage, and the exceptionally species-rich modern haplochromines. The ca. 1,800 species of modern haplochromines are comprised of the entire haplochromine species flocks of Lake Malawi and the Lake Victoria region, some 200 riverine and lacustrine species, as well as the Tanganyikan Tropheini, which are likely to have evolved from a riverine ancestor and secondarily colonized Lake Tanganyika. This proposed "out of Tanganyika" scenario of haplochromine evolution is in agreement with the geological and palaeo-geological history of East Africa. Based on a character-state reconstruction from this new phylogeny, we were able to discover the evolution of several key-innovations that arose in the lineage leading to the modern haplochromines. These character reconstructions suggest that a combination of behavioral (maternal mouthbrooding) and morphological characteristics (egg-spots, color polymorphisms, pronounced sexual dichromatism) as well as ecological opportunities (after the colonization of large lakes) might have predestined this particular lineage to give rise to replicate adaptive radiations and, therefore, be causally related to the extraordinary success of these particular cichlid fish radiations. Methods Specimen information and DNA methods For this study, a total of 304 specimens were analyzed. We combined all available GenBank entries from previous studies [ 4 , 9 , 15 , 25 , 62 ] with 180 newly determined DNA sequences. Table 1 [see Additional file 1 ] lists specimen information, geographic origin of the specimens, names of collectors, and GenBank accession numbers. When available, voucher specimens have been deposited at the Royal Museum for Central Africa, Tervuren, Belgium. Sample preparation, polymerase chain reaction (PCR) amplification and DNA sequencing have been performed as described elsewhere [ 9 ] for both mitochondrial DNA segments the complete non-coding control region and the entire NADH Dehydrogenase Subunit II (ND2) gene. Forward and reverse sequences have been assembled using the computer programs Sequence Navigator (Applied Biosystems, USA) and Sequencher (GenCodes, USA). Phylogenetic reconstruction and hypotheses testing The complete sequences of the mitochondrial control region (895 bp) and the ND2 gene (1,047 bp) were aligned using the computer program Clustal W [ 63 ]; alignments have been further adjusted by eye. Up to 34 gaps had to be included in the control region alignment, which were coded as indels. Due to missing data on the 5'-end of the control region in about one quarter of the taxa, a terminal section of 34 bp has been excluded from the phylogenetic analysis, leading to an alignment of 1,908 bp for the combined dataset. For phylogenetic reconstruction we performed maximum likelihood, maximum parsimony and neighbor-joining methods in parallel using the computer program PAUP* 4.0b10 [ 64 ]. Two taxa belonging to tilapiine cichlids were included as outgroup based on previous phylogenetic analyses using mitochondrial DNA [ 9 , 15 , 16 ], nuclear DNA [ 65 ] and SINE insertion patterns [ 66 , 67 ] demonstrating that the Tilapia-lineages are ancestral to all remaining LT tribes (but excluding Tylochromis , which is ancestral to the Tilapiini). We did not include representatives of three LT tribes – the Bathybatini, Trematocarini and Tylochromini -, as these ancient lineages were shown to have existed before the primary radiation of mouthbrooders in LT [ 9 , 15 , 67 ]. After an initial neighbor-joining analysis including the control region sequences of all 304 specimens, we reduced the dataset to 100 taxa based on the obtained topology. This reduction was necessary to allow computational feasibility for maximum likelihood and maximum parsimony analyses. The optimal model of molecular evolution for the maximum likelihood analysis was determined in a likelihood-ratio test running the computer program Modeltest v3.06 [ 68 ]. For the heuristic maximum likelihood search of the combined dataset we used the general-time-reversible model of molecular evolution, with a gamma shape parameter of 0.7937 and a proportion of invariable sites of 0.3426. Due to the many closely related taxa in the dataset, maximum parsimony analyses were completed for 10 10 rearrangements. For neighbor joining, we used the HKY model and conducted a bootstrap analysis with 5,000 replicates. Bootstrap analyses for maximum likelihood were performed with 100 replicates and for maximum parsimony with 1,000 replicates. We also applied Bayesian inference of phylogeny with Mr. Bayes 3.0b4 [ 69 ] running four Metropolis Coupled Monte-Carlo-Markov-Chains in parallel for 250.000 generations, using the general-time-reversible model with gamma correction (six types of substitutions), and excluding 5 % of the trees as burn-in. The obtained topologies from the different phylogenetic algorithms were evaluated by means of a nonparametric Shimodaira-Hasegawa test [ 19 ] under a resampling-estimated log-likelihood as implemented in PAUP* [ 64 ]. To estimate the support for distinct internal branches critical for our interpretations, we performed a four-cluster likelihood mapping analysis with the program PUZZLE 5.0 [ 20 , 70 ] in which we grouped the taxa of the haplochromines sensu lato into four groups according to the four lineages that were recovered from the phylogenetic analyses (CSA-, Pseudocrenilabrus -, Astatoreochromis -lineage, and modern haplochromines). To test whether or not shifts in the probabilities of speciation occurred on certain branches of the obtained phylogeny, we quantitatively tested the fit of this tree to a Markovian null model in which the probability of speciation is equally distributed along branches. Therefore, we calculated the cumulative probability p c [ 21 ] for relevant branches based on the different trees. We then mapped lifestyles (riverine versus lacustrine), breeding behavior, and the occurrence of true egg-spots on the maximum likelihood tree using MacClade 4.0 (Sinauer, Sunderland, MA) for maximum parsimony character state reconstructions and Mesquite [ 71 ] for maximum likelihood reconstructions. To tentatively date the major cladogenetic events in the haplochromines, we constructed a chronogram based on a maximum likelihood tree with constraint molecular clock (see above for search parameters). Therefore, we used the computer program r8s [ 72 ] applying the local molecular clock method and an optimization via the truncated newton method [ 73 ]. Confidence intervals were assessed by means of a bootstrap approach. We simulated 25 bootstrap matrices with Mesquite and, for each matrix, constructed a maximum likelihood tree (general-time-reversible model; model parameters, gamma shape correction and proportion of invariable sites being estimated from each matrix; rearrangements limited to 1,000). The resulting trees were then analyzed with r8s as described above, and in addition with the minimum and maximum values of geological datings separately, in order to define upper and lower bounds. As calibration points, we used an age estimate for the LM species flock of about one million years [ 74 , 75 ], the maximum age of 200.000 years for the LV region superflock [ 1 , 4 , 25 ] as well as the time window for the Lukuga connection between LT and the Congo system (between the minima at 3.5 MYA and 1.1 MYA) [ 76 , 77 ]. An escape of LT lineages into river systems draining into the lake was possible at any time, whereas the only connection to the Congo system, the Lukuga, was available at periods of high lake level stands only. The Lukuga channel, the only outlet of LT, was dry when first seen in 1874 by Cameron, but four years later the lake overflowed. Since then, the Lukuga was repeatedly flooded. Intermittent connections with the Congo River system in the late Pliocene/early Pleistocene were suggested on the basis of large lake level fluctuations in LT during periods with increased precipitation [ 41 , 76 - 79 ]. The resulting molecular clock rate fits well with previously used rates in East African cichlids [ 4 , 75 ]. List of abbreviations used As., Astatoreochromis ; bp, base pairs; CSA, Congo/South Africa; LM, Lake Malawi; LT, Lake Tanganyika; LV, Lake Victoria; MH, modern haplochromines; mtDNA, mitochondrial DNA; MYA, million years ago; ND2, NADH Dehydrogenase Subunit II; Ps., Pseudocrenilabrus . Authors' contributions WS, EV, and AM designed the study and were involved in sampling. WS and TM carried out the molecular work and the analyses. All authors contributed to the preparation of the manuscript. They read and approved the final version. Supplementary Material Additional File 1 Table 1 – Specimen information, geographic origin and GenBank accession numbers of all taxa included in this study . This table lists the species and tribe names, geographic origin, source of specimens, names of collectors and collection numbers (if available), and GenBank accession numbers for both mitochondrial gene segments. The taxa that are included in Fig. 3 are marked by a circle, specific taxon labels in Fig. 3 are depicted in the "Label" column. The clade names according to Figs. 2 and 3 are shown in the last column. Tribe names are according to [ 7 ]. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554777.xml |
544570 | Regulation of cell cycle by the anaphase spindle midzone | Background A number of proteins accumulate in the spindle midzone and midbody of dividing animal cells. Besides proteins essential for cytokinesis, there are also components essential for interphase functions, suggesting that the spindle midzone and/or midbody may play a role in regulating the following cell cycle. Results We microsurgically severed NRK epithelial cells during anaphase or telophase, such that the spindle midzone/midbody was associated with only one of the daughter cells. Time-lapse recording of cells severed during early anaphase indicated that the cell with midzone underwent cytokinesis-like cortical contractions and progressed normally through the interphase, whereas the cell without midzone showed no cortical contraction and an arrest or substantial delay in the progression of interphase. Similar microsurgery during telophase showed a normal progression of interphase for both daughter cells with or without the midbody. Microsurgery of anaphase cells treated with cytochalasin D or nocodazole indicated that interphase progression was independent of cortical ingression but dependent on microtubules. Conclusions We conclude that the mitotic spindle is involved in not only the separation of chromosomes but also the regulation of cell cycle. The process may involve activation of components in the spindle midzone that are required for the cell cycle, and/or degradation of components that are required for cytokinesis but may interfere with the cell cycle. | Background Microtubules undergo striking reorganization during anaphase and telophase. During anaphase, antiparallel, interdigitating microtubules and many associated proteins become organized into discrete bundles in the spindle midzone [ 1 ], the region between separated chromosomes. As the cell enters cytokinesis, these midzone microtubule bundles merge into a single compact, electron-dense structure called midbody. It is generally recognized that, at least for cultured animal cells, midzone microtubules play a major role in cytokinesis. For example, cleavage furrows, both normal and ectopic, are associated with similar microtubule bundles [ 2 - 4 ], while regions physically blocked from midzone microtubules by micromanipulation are unable to undergo cytokinesis [ 5 ]. Moreover, continuous interactions of midzone microtubules with the cell cortex are required for sustaining the cytokinesis of cultured animal cells [ 6 ]. Recent progress suggested that, in addition to cytokinesis, the spindle midzone might be involved in additional functions. For example, midzone and midbody microtubules are associated with many regulatory proteins apparently unrelated to cortical contraction, such as the DNA replication initiator Orc6 [ 7 ], the inhibitor of apoptosis survivin [ 8 ], and the tyrosine kinase binder Nir2 [ 9 ]. In addition, mother centrioles were found to migrate to the midbody during telophase before returning to their interphase position, possibly activating some centrosomal components for cell cycle progression [ 10 ]. Furthermore, treatment of dividing cells with dihydrocytochalasin B, an inhibitor of cytokinesis, caused not only the inhibition of cytokinesis but also G1 arrest after mitosis [ 11 ], raising the possibility that ploidy, cortical contraction, and/or activation/deactivation of proteins during cytokinesis, may play a role in the regulation of the following cell cycle. To address this possibility, we severed cells at anaphase or telophase by microsurgery, to bypass the normal mechanism of cytokinesis. The functional role of the spindle midzone or midbody in the following cell cycle was then tested by manipulating the position of the microsurgery or by applying pharmacological agents. Using extended time-lapse microscopy, we found that anaphase midzone microtubules play an important role in the progression of the subsequent interphase. However, there was no evidence of the involvement of telophase midbody in cell cycle progression. Results and discussion To investigate if anaphase spindle midzone affects interphase progression in daughter cells, we cut NRK52E cells at early anaphase, as soon as sister chromosomes have separated completely, with a fine glass fiber to form daughter cells with or without spindle midzone structures. Staining of severed daughter cells with antibodies against aurora B (Figure 1A ), which is known to relocate from centromeres to the spindle midzone during early anaphase [ 12 , 13 ], confirmed that the spindle midzone was removed from one of the daughter cells. However, both daughter cells showed typical interphase microtubules organization (Figure 1B ), and normal cellular and nuclear morphology (Figure 2, d ), indicating that cells recovered fully from the microsurgery and exited mitosis successfully. Figure 1 Microsurgery of dividing NRK cells. The site of severing is indicated by dotted lines (A, a; B, a). Immunofluorescence indicates that aurora B, a protein associated with midzone microtubules, is partitioned predominantly to the cell with midzone microtubules (A, d). Microtubules, staining of another cells cut in a similar way, showed an indistinguishable network in both daughter cells after recovering for one hour (B, d). Bar, 20 μm. Figure 2 Inhibition of the interphase progression following surgical removal of the spindle midzone structures. An NRK cell was cut at anaphase to form daughter cells with (top) and without (bottom) the spindle midzone structures (cutting site indicated by the dotted line, a). Subsequent time-lapse recording indicates that both daughter cells formed nuclear envelop (arrows, d). However, only the daughter cell with spindle midzone (top) showed cytokinesis-like contractions (c, arrows). The daughter cell with spindle midzone entered the subsequent mitosis 11 hours after microsurgery (e, arrow), whereas the daughter cell without spindle midzone entered mitosis 22 h after microsurgery (f, arrow). Time elapsed since cutting is shown in hours:minutes. Bar, 20 μm. A total of eight manipulated cells were followed by extended time-lapse imaging through the following interphase. All the daughter cell with spindle midzone showed cytokinesis-like cortical contraction activities, while the daughter cell without spindle midzone showed no cortical contraction (Figure 2 , see Additional file 1 ). The daughter cell with spindle midzone subsequently progressed with normal timing through the following interphase, entering mitosis at a time similar to that of adjacent control cells (Figure 2 , Table 1 ). However, in 5 out of 8 cases, the cell without midzone showed a duration of interphase more than twice that of the sister cell with spindle midzone (Table 1 ). In the example shown in Figure 2 , interphase lasted for 22 h, as compared to the normal 11 h for the sister cell with the spindle midzone. In two cases, cell cycle appeared to be arrested in interphase since no mitosis was observed over a period of 34 – 46 h. Table 1 Duration of interphase for daughter cells with and without the spindle midzone or midbody # midbody/spindle midzone + - spindle midzone (8) 721 ± 66 1444 ± 236** spindle midzone (5)* 671 ± 28 1724 ± 301** midbody (9) 610 ± 92 678± 113 control*** (66) 611 ± 16 # Average time in minutes ± SEM *Average of 5 experiments that showed at least a doubling in the length of interphase for the daughter without the midzone. **In 3 cases, the daughter cell without spindle midzone failed to enter mitosis during the period of time-lapse recording. For these cells, the length of interphase was assumed to equal the length of observation. Therefore, this average value represents an under-estimate. ***Two unmanipulated cells near the manipulated cell were measured in each experiment as controls. The inhibition of interphase progression was affected by the position of severing. When cells were severed near the equator to divide midzone structures between the two daughter cells, both cells showed cytokinesis-like cortical contractions and no effect on the progression of interphase. This observation was repeated with 4 cells. To test the possibility that the duration of interphase was affected by the size of the daughter cell, we took pairs of normally divided daughter cells and cut away a lateral region near the pole from one of the daughter cells. In all 5 cases, both daughter cells entered the subsequent mitosis with a timing similar to that of neighboring unmanipulated cells (Figure 3 ). Together, these results indicate that the progression of interphase was affected directly or indirectly by the presence of the anaphase spindle midzone. Figure 3 Normal interphase progression following manipulation of the size of a daughter cell. A lateral region near the pole was cut away from one of the daughter cells at the end of cytokinesis (cutting site indicated by the dotted line, a). Subsequent time lapse imaging revealed a normal interphase duration in both daughter cells (b, c). Time elapsed since cutting is shown in hours:minutes. Bar, 20 μm. To test if cortical contraction is the primary determining factor of the rate of interphase progression, we treated cells with cytochalasin D at early anaphase for 10 min then severed them at various positions in the spindle midzone, at a time when midbody started to form and ingression started to appear in control cells. Cytochalasin D was removed 20 min later to ensure that there was no lingering cytokinesis activity upon removal of the drug [[ 14 ] see Methods]. In all 5 cases, both daughter cells entered the subsequent mitosis with a timing similar to that of neighboring cells, despite the complete inhibition of cortical contraction (Table 2 ). In addition, unmanipulated, binucleated cells, which failed cytokinesis spontaneously, also entered subsequent mitosis with a similar timing. Thus, the present observation is distinct from the "tetraploidy checkpoint", which was identified with dihydrocytochalasin B-treated REF52 cells [ 11 ], but remained controversial with regard to its universal existence [ 15 ]. Table 2 Duration of interpahse for cells severed in the presence of cytoskeletal inhibitors # Severed cells ## control ###, ## cytochalasin D 599 ± 49 (5) 587 ± 24 (10) nocodazole 892 ± 71 (7) 621 ± 30 (14) # Average time in minutes ± SEM ## Statistics was calculated by grouping the daughter cells. Therefore, the number of daughter cells analyzed was twice the number of experiments as indicated in the parentheses. ### Two unmanipulated, dividing cells near the manipulated cell were followed in each experiment as controls. We then asked if midzone microtubules, or proteins associated with the anaphase spindle midzone, are involved in interphase progression. Treatment of cells with nocodazole for 6–13 min at early anaphase before severing, with or without additional incubation with the drug, caused a significant delay (~4.5 h; p = 0.002) in the progression of interphase for both daughter cells, as compared to neighboring cells (Table 2 ). Aurora B in these cells showed a complete dispersal from the midzone (Figure 4 ) [ 12 ], with a similar amount distributed in the two severed cells. These observations suggest the midzone microtubules may provide a scaffold for the activation/deactivation of some components to allow normal progression of the cell cycle. We suspect that the weaker effect compared to that caused by severing may be due to the known resistance of some midzone microtubules to nocodazole [ 6 ]. Alternatively, a brief association of the components with the midzone microtubules before and during nocodazole treatment may be sufficient for partial activation/deactivation. Figure 4 Presence of aurora B in both daughter cells following microsurgery of nocodazole-treated cells. An NRK cell at early anaphase was treated with 10 μM nocodazole for ~10 min before microsurgical cut between segregated chromosomes (cutting site indicated by the dotted line, a). Then the daughter cells were released from nocodazole by washing twice with fresh medium. Immunofluorescence of aurora B showed that both daughter cells contained dot-like structures of aurora B along the cell cortex (c, d, arrows). Bar, 20 μm. We also tested if telophase midbody is required for interphase progression. Cells were severed at mid- or late cytokinesis to form daughter cells with and without the midbody (Figure 5 , see Additional file 2 ). Using cells expressing aurora B-GFP, which is known to associate with the midbody [ 12 , 13 , 16 ], we confirmed that the midbody was completely segregated from one of the daughter cells (Figure 5B ). All nine manipulated pairs showed a normal progression of interphase indistinguishable from that of unmanipulated control cells, irrespective of the presence of the midbody (Table 1 ). We conclude that the presence or absence of telophase midbody no longer affects the progression of subsequent interphase, despite the similarity of molecular components to those in the earlier spindle midzone. Thus, most likely it is transient catalytic reactions in the anaphase midzone that are crucial for the progression of cell cycle. Figure 5 Normal interphase progression following removal of the midbody. A cell was cut at the end of cytokinesis to form daughter cells with or without the midbody (A, b, arrow). Subsequent long-term time-lapse imaging indicated a normal phase morphology for both cells (A). The cell with midbody entered mitosis 10 h 40 min after cutting, whereas the cell without midbody entered mitosis ~1 h afterwards (e, f, large arrows). Fluorescence imaging of aurora-B-GFP, a midbody component, confirmed that the midbody is completely segregated into one of the daughter cells (B; cutting indicated by dotted lines) Bar, 20 μm. Some of major mitotic regulators are degraded by anaphase promoting complex/cyclosome (APC) during anaphase [ 17 ]. Degradation of the polo-like kinase (Plk1) and aurora A by APC occurred while they were localized along midzone microtubules [ 17 ], raising the possibility that interphase progression may require the degradation of some mitotic/cytokinetic proteins, which may then cause activation of downstream components crucial for cell cycle. In addition, some molecules associated with midzone microtubules may be directly involved in cell cycle events such as DNA synthesis, as suggested by the chromosomal passenger protein-like dynamics of a DNA replication initiating factor, Orc6 during cell division [ 7 ]. Conclusions Our results suggest that anaphase midzone not only play a role in the stimulation of cytokinesis in cultured cells, but also provide a scaffold for the activation/deactivation of factors essential for the progression of subsequent cell cycle. Methods Cell culture, microscopy, and image processing Normal Rat Kidney epithelial cells (NRK-52E; American Type Culture Collection, Rockville, MD) were cultured in Kaighn's modified F12 (F12K) medium supplemented with 10% FBS (JRH Bioscience, Lenexa, KS), 50 U/ml penicillin, and 50 μg/ml streptomycin, on glass chamber dishes as previously described [ 18 ]. The cells were maintained at 37°C in a stage incubator built on top of a Zeiss Axiovert S100TV or an Axiovert 35 inverted microscope (Carl Zeiss, Thornwood, NY), and viewed with 10X, NA 0.25 Achrostigmat, 40X, NA 0.75 Plan-Neofluor or 100X, NA 1.30 Plan-Neofluor lens. All images were acquired with a cooled charge-coupled device camera (ST133 controller and CCD57 chip; Roper Scientific, Trenton, NJ) and processed with custom software for background subtraction. Microsurgery and drug treatment Glass needles for microsurgery were prepared with a David-Kopf Model 700 vertical puller. The tip of the needle was melted and elongated into a fine fiber with a Narishige microforge (Model MF900). Microsurgery of the cells was achieved by carefully lowering a fiber onto the target cell followed by slow dragging with a micromanipulator (Leica, Deerfield, IL). Cytochalasin D (Sigma, St. Louis, MO) were stored at -20°C as 2.5 mM stock in DMSO, and diluted into warm medium before application to cells. We found that treatment of early anaphase cells with cytochalasin D for 30 min completely inhibited cytokinesis even upon removal of the drug, similar to what was reported with dihydrocytochalasin B [ 14 ]. Thus, early anaphase cells were treated with cytochalasin D at a final concentration of 2 μM for 10 min before microsurgery. The cells were then incubated for additional 20 min and washed at least twice with fresh medium. Nocodazole (Sigma, St. Louis, MO) was stored at -20°C as 10 mM stocks in DMSO and was diluted with warmed medium before use. Early anaphase cells were incubated with nocodazole for 6–13 min and then cut into two daughter cells with microsurgery. Immediately after microsurgery or following 50 min incubation, the daughter cells were released from nocodazole by incubating with two changes of fresh medium for at least 5 min each. The significance of these results was assessed using analysis of variance (ANOVA) and t-test in Microsoft Excel. Transfection and immunofluorescence Aurora B-GFP was constructed and transfected into NRK cells as described previously [ 12 ]. Immunofluorescence of tubulin and reconstruction of microtubules images were carried out as described previously [ 19 ]. For immunofluorescence, cells were rinsed with warm cytoskeleton buffer and fixed with 4% paraformaldehyde (EM Science, Gibbstown, NJ) in warm cytoskeleton buffer for 10 min [ 6 ]. They were then rinsed thoroughly in the cytoskeleton buffer and permeablized by incubation with 0.5% Triton X-100 in cytoskeleton buffer for 5 min. Fixed cells were rinsed with the cytoskeleton buffer, blocked for 10 min with 1% BSA (Boehringer Mannheim, Indianapolis, IN) in PBS, then incubated with anti-AIM-1 monoclonal antibodies (BD Biosciences, San Jose, CA) at a dilution of 1:200 in PBS with 1% BSA for 45 min at 37°C. After washing with PBS/BSA thoroughly, cells were incubated with Alexa 546-conjugated goat anti-mouse antibodies (Molecular Probes, Eugene, OR) at a dilution of 1:100 for 30 min at 37°C. Authors' contributions MMH carried out all the experimental work and drafted the manuscript. GS and YLW participated in its design and coordination of the research and edited the manuscript. All authors read and approved the final manuscript. Supplementary Material Additional File 1 Interphase progression of the daughter cells with and without spindle midzone. An NRK cell was severed at early anaphase to form daughter cells with and without spindle midzone. Subsequently, time-lapse imaging was performed to investigate if anaphase spindle midzone was involved in interphase progression. The daughter cell with spindle midzone showed a cytokinsis-like cortical contraction and entered into the subsequent mitosis 11 h 23 min after microsurgery, while the daughter cell without spindle midzone showed no cortical contraction and entered into mitosis 23 h 8 min after microsurgery. Click here for file Additional File 2 Interphase progression of the daughter cells with and without midbody. An NRK cell was severed at telophase to form daughter cells with and without midbody. Subsequently, time-lapse imaging was performed to investigate if midbody was involved in interphase progression. The daughter cell with midbody entered into the subsequent mitosis 7 h 35 min after microsurgery, while the daughter cell without midbody entered into mitosis 9 h 25 min after microsurgery. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544570.xml |
543579 | prot4EST: Translating Expressed Sequence Tags from neglected genomes | Background The genomes of an increasing number of species are being investigated through generation of expressed sequence tags (ESTs). However, ESTs are prone to sequencing errors and typically define incomplete transcripts, making downstream annotation difficult. Annotation would be greatly improved with robust polypeptide translations. Many current solutions for EST translation require a large number of full-length gene sequences for training purposes, a resource that is not available for the majority of EST projects. Results As part of our ongoing EST programs investigating these "neglected" genomes, we have developed a polypeptide prediction pipeline, prot4EST. It incorporates freely available software to produce final translations that are more accurate than those derived from any single method. We show that this integrated approach goes a long way to overcoming the deficit in training data. Conclusions prot4EST provides a portable EST translation solution and can be usefully applied to >95% of EST projects to improve downstream annotation. It is freely available from . | Background The need for more sequence Complete genome sequencing is a major investment and is unlikely to be applied to the vast majority of organisms, whatever their importance in terms of evolution, health or ecology. Complete genome sequences are available for only a few eukaryote genomes, most of which are model organisms. The focus of eukaryote genome sequencing has been on a restricted subset of known diversity, with, for example, nearly half of the completed or draft stage genomes being from vertebrates. While Arthropoda and Nematoda have two completed genomes each, with a dozen others in progress, compared to predicted diversity (over a million species each) current genome sequencing illuminates only small parts of even these phyla. The disparity between sequence data and motivation for biological study is significant. Allied to this bias in genome sequence is a bias in functional annotation for the derived proteomes: a vertebrate gene is more likely to have been assigned a function due to the focus of biomedical research on humans and closely related model species such as mouse [ 1 ]. Shotgun sample sequencing of additional genomes through expressed sequence tags (EST) or genome survey sequences (GSS) has proved to be a cost-effective and rapid method of identifying a significant proportion of the genes of a target organism. Thus many genome initiatives on non-traditional model organisms have utilised EST and GSS strategies to gain an insight into "wild" biology. An EST strategy does not yield sequence for all of the expressed genes of an organism, because some genes may not be expressed under the conditions sampled, and others may be expressed at very low levels and missed through the random sampling that underlies the strategy. However the creation of EST libraries from a range of conditions, such as different developmental stages or environmental exposures, promotes a closer examination of the biology of these species. The well documented phylogenetic sequence deficit [ 2 ] has led us to coin the term "neglected genomes". Currently many groups are sequencing ESTs from their chosen species to perform studies in a wide-range of disciplines, from comparative ecotoxicology [ 3 ] to high-throughput detection of sequence polymorphisms [ 4 , 5 ]. The contribution of EST projects for neglected but biologically relevant organisms is highlighted in Figure 1 . As with all sequence data, obtaining high quality annotation requires prior information and is labour intensive. The "partial genome" information that results from EST datasets presents special problems for annotation, and we are developing tools for this task. The need for high quality translation The PartiGene software suite [ 6 ] simplifies the analysis of partial genomes. ESTs are clustered into putative genes and consensuses determined. All the data is stored in a relational database, allowing it to be searched easily. While preliminary annotation based on BLAST analysis of nucleotide sequence can be performed, more robust methods are needed to allow high-quality analysis. The error-prone nature of ESTs makes application of most annotation tools difficult. To improve annotation, and facilitate further exploitation, a crucial step is the robust translation of the EST or consensus to yield predicted polypeptides. The polypeptide sequences present a better template for almost all annotation, including InterPro [ 7 ] and Pfam [ 8 ], as well as the construction of more accurate multiple sequence alignments, and the creation of protein-mass fingerprint libraries for proteomics exploitation. High quality polypeptide predictions can be applied to functional annotation and post-genomic study in a similar way to those available for completed genomes. Translating Expressed Sequence Tags Prediction of the correct polypeptide from ESTs is not trivial: 1. The inherent low quality of EST sequences may result in shifts in the reading frame (missing or inserted bases) or ambiguous bases. These errors impede the correct recognition of coding regions. The initiation site may be lost, or an erroneous stop codon introduced to the putative translation. 2. ESTs are often partial segments of a mRNA, and as most cloning technology biases representation to the internal parts of genes, the initiation methionine codon may be missed. This is a problem for some of the de novo programs which use the initiation methionine to identify the coding region (described below). Sequence quality can be improved by clustering the sequences based on identity. For each cluster a consensus can be determined [ 9 ]. This approach, however, will not address the whole problem as poor quality EST sequences may not yield high quality consensuses and for smaller volume projects, most genes have a single EST representative. Therefore additional methods must be applied to provide accurate polypeptide predictions. Similarity-based methods A robust method to determine the correct encoded polypeptide is to map a nucleotide sequence onto a known protein. This concept is the basis for BLASTX [ 10 ], FASTX [ 11 ] and ProtEST [ 12 ]. BLASTX and FASTX use the six frame translation of a nucleotide sequence to seed a search of a protein database. The alignments generated for each significant hit provide an accurately translated region of the EST. BLASTX is extremely rapid, but the presence of a frameshift terminates each individual local alignment, ending the polypeptide prematurely. FASTX is able to identify possible frameshifts, but its dynamic programming approach is significantly slower than BLASTX. These methods require that the nucleotide sequence shares detectable similarity with a protein in the selected database. Many genes from both well studied and neglected genomes do not share detectable similarity to other known proteins. For example, the latest analysis of the Caenorhabditis elegans proteome shows that only ~50% of the 22000 predictions contain Pfam-annotated protein domains [ 8 , 13 ], and 40% share no significant similarity with non-nematode proteins in the SwissProt/trEMBL database [ 14 ]. This feature is not unique to the phylum Nematoda, and is likely perhaps to be more extreme for neglected genomes, given the phylogenetic bias of most protein databases. ProtEST uses a slightly different similarity-based approach [ 12 ]. A protein sequence is compared to an EST database. phrap [ 9 ] is used to construct a consensus sequence from the ESTs found to have significant similarity. These consensuses are then compared to the original sequence using ESTWISE (E. Birney, unpublished [ 15 ]) giving a maximum likelihood position for possible frameshifts. The system is accurate but is not readily adaptable to the high-throughput approach necessary when dealing with very large numbers of ESTs. More crucially, an EST that does not show significant similarity to a known protein is not translated. 'de novo' predictions To overcome the reliance upon sequence similarity, de novo approaches based on recognition of potential coding regions within poor quality sequences, reconstruction of the coding regions in their correct frame, and discrimination between ESTs with coding potential and those derived from non-coding regions have been developed [ 16 - 18 ]. DIANA-EST [ 16 ], combines three Artificial Neural Networks (ANN), developed to identify the transcription initiation site and the coding region with potential frameshifts. ESTScan2 [ 18 ] combines three hidden Markov models trained to be error tolerant in their representations of mRNA structure (modelling the 5' and 3' untranslated regions, initiation methionine and coding region). DECODER [ 17 ] uses an essentially rule-based method for identifying possible insertions and deletions in the nucleotide sequence, as well as the most likely initiation site, and was developed for complete cDNA sequence translation. Each of these methods has different strengths in their attempt to identify the precise coding region; all require prior data to train their models. Published descriptions of their utility are based on training with human full length coding sequences (mRNAs), and thus tens of thousands of training sequences (many million coding nucleotides) were used to achieve optimum results. As stressed above, this amount of prior data is not available for the vast majority of EST project species (Figure 1 ). New solution – prot4EST Prior to this project, nematode ESTs available through NEMBASE [ 19 ] had been translated using DECODER, as a preliminary study had suggested that it outperformed the other available methods (DIANA-EST and ESTScan1 [ 20 ]) (Parkinson pers. com.). 7388 out of the 40000 resulting predicted polypeptides were likely to be poorly translated (<30 amino acids), and we suspected many more contained errors. This motivated the creation of a solution using several methods to enhance the quality of the polypeptide predictions, exploiting their strengths while recognising their short-comings. prot4EST is an EST translation pipeline, written in Perl, with a user-friendly interface, that links some of these described methods together. It carries out retrieval and formatting of files from online databases for the user. It has been designed to be used as a stand-alone tool, or as an integral part of the PartiGene process [ 6 ]. Implementation DECODER The DECODER program [ 17 ] was developed to define start codons and open reading frames in full-length cDNA sequences. It exploits the quality scores for the sequence produced from base-calling software, such as phred [ 21 , 22 ], and additional text-based information to identify all possible coding regions. In regions of low sequence quality up to 2 nucleotides are removed or inserted, representing possible frameshifts. A likelihood score is calculated for each possible coding sequence (CDS), and the one with the lowest score is chosen as the correct CDS. The score is computed from the probability of generating a random sequence with a better Kozak consensus (the nucleotide sequence surrounding the initiation codon of a eukaryotic mRNA), ATG position and codon usage. DECODER requires a codon bias table, which is used to determine the putative coding regions optimal codon usage. A penalty term limits the number of insertions/deletions in the corrected CDS. ESTScan2.0 Hidden Markov models (HMM) can represent known sequence composition in a probabilistic manner [ 23 ]. This has been exploited recently in applications to find genes in genomic sequence [ 24 , 25 ], predict domain composition in protein sequences [ 26 ], and align multiple sequences [ 27 ]. ESTScan [ 18 ] exploits the predictive power of Hidden Markov models by combining three models: 1. Modeling mRNA structure: ESTScan separates the probable CDS from the untranslated regions (UTRs). The core of the coding sequence is represented by a 3-periodic inhomogeneous hidden Markov model. Flanking this core model are start and stop profiles for the codons observed at these positions. The profiles for untranslated regions flank the start and stop states. 2. Error tolerance: ESTScan allows insertions and deletions (indels) in the EST sequence. For example, if it is more probable that a particular nucleotide is the result of an insertion event then it is omitted from the 'corrected' sequence. Conversely, if the HMM probability scores suggest that a nucleotide has been deleted then the model inserts an X into the 'corrected' sequence to denote this prediction. 3. EST structure: ESTScan recognises that the EST may be composed of a combination of 5' UTR, CDS and 3' UTR. ESTScan's hidden Markov models are trained using complete CDS entries from either the EMBL or RefSeq databases. Scripts included with the distribution parse the data files, extracting the necessary sequence information to produce the model files. The major issue considered at this point is redundancy. If the training data is internally redundant then the resultant model will be fully successful only in finding what is known and will have reduced power in detecting novel transcripts. Default parameters were used in ESTScan for building the HMM and in predicting polypeptides. HSP tiling The BLASTX program [ 10 ] allows a nucleotide sequence to be searched against a protein database. The nucleotide query is translated in all six frames and these are used as the query sequences for a BLASTP search. High scoring segment pairs (HSP) are identified that maximise a bit score derived from an amino acid similarity matrix. If a single indel occurs in the nucleotide sequence, causing a frameshift, the HSP is either terminated at this position or continues out of frame. Downstream of this frameshift the query sequence may be long enough to result in another significant HSP to the same protein sequence, this time in a different frame. Simple extraction of the best BLAST HSP will miss such features. prot4EST implements a rule-based method that considers all the HSPs for a match to a database sequence and considers whether a frameshift can be identified. Where a frameshift is identified the HSPs are joined. Where two HSPs overlap the sequence with the better bit score is used. The prot4EST pipeline prot4EST is an integrated pipeline utilising freely available software in a tiered, rule-based system (Figure 2 ). Tier 1: Identification of ribosomal RNA (rRNA) genes The protein databases contain (probably spurious) translations of ribosomal RNA genes and gene fragments, and thus it is important to identify and remove putative rRNA derived sequences before further processing. A BLASTN search is performed against a database of rRNA sequences obtained from the Ribosomal Database II (Table 1 ; [ 28 ]). A BLAST expect value cutoff of e-65 is used to identify matches. The cutoff is a conservative one to reduce the number of false positives. Those nucleotide sequences with significant matches are annotated as rRNA genes and take no further part in the translation process. Tiers 2 and 3: Similarity search The second and third stages are similar. First a BLASTX search is performed against proteins encoded by mitochondrial genomes. The mitochondrial protein database is obtained from the NCBI ftp site (Table 1 ). Any sequences with significant hits (cutoff e-8) are annotated as mitochondrion-encoded genes for the remainder of the process, and the relevant mitochondrial genetic code is used for translation. Sequences that do not have significant similarity to mitochondrial proteins are compared using BLASTX to the SwissProt database [ 14 ]. Sequences that yield no significant similarity are moved onto tier 4 of the process. For those sequences that show significant similarity to a protein sequence from either database a HSP tile path is constructed. prot4EST then considers whether the nascent translation can be extended at either end in the same reading frame. Tier 4: ESTScan prediction The hidden Markov models used by ESTScan to identify the coding region are constructed from EMBL format files for complete CDS using scripts supplied with the package. Preprocessing is integrated within prot4EST, including the downloading of the EMBL files. A pair of length threshold criteria are applied to each putative polypeptide before it is accepted. The open reading frame must be at least 30 codons in length, and cover at least 10% of the input sequence. Polypeptides that satisfy these criteria undergo the extension process described above, sequences that fail any of the criteria are passed onto the next tier. The extension process is carried out on those sequences that exceed the thresholds. Tier 5: DECODER prediction The DECODER program is used to predict CDS and thus polypeptide translations for the remaining nucleotide sequences. For each sequence a quality file in phrap format is required. When a quality file is unavailable a file with quality values of 15 is generated for each sequence. The codon usage table required by DECODER can be specified by the user or downloaded from CUTG, the codon usage table database [ 29 ]. By default DECODER only processes the forward strand of each sequence, and therefore the reverse complement of each sequence is taken and processed through DECODER. Two putative polypeptides are generated for each nucleotide sequence. The longer polypeptide is selected as the more probable translation. The polypeptide predictions are checked using the same length threshold criteria as for ESTScan (above). Tier 6: Longest ORF This last attempt to provide a putative polypeptide translation determines the longest string of amino acids uninterrupted by stop codons from a six-frame translation of the sequence. If a methionine is present in this string it is flagged as a potential initiation site. Output The primary output from prot4EST consists of the putative polypeptides in FASTA format, complemented with files containing information describing the translated sequences. This information includes: position of the translation with respect to the nucleotide sequence, the genetic code used for translation, position and BLAST statistics of HSPs used in the tile path. All this additional information is stored in two CSV format files, permitting parsing and simple insertion into a database. Speed This is highly dependent upon the composition and size of the dataset. As a guide, each prot4EST run carried out in the benchmarking (below), took less than an hour for a 2316-sequence input with an Athlon 1400 Mhz processor. The BLASTX searches were carried out separately and used as input to prot4EST (for details see the userguide, availabile from the program web page). Benchmarking EST translation methods We benchmarked five translation methods to test their relative performance. DECODER is designed to consider only the forward strand of the nucleotide sequence, as it was originally designed for full-length CDSs. When applied to ESTs it is imperative that both strands are analysed, as both 5' and 3' ESTs are generated. Therefore the reverse complement of each nucleotide consensus was also analysed. DECODER_default (1) considers only the prediction from the forward strand, whilst DECODER_best (2) uses the more accurate prediction. ESTScan (3) considers both strands of the nucleotide sequence, and was run as a stand-alone process with default settings. Two arrangements of components within prot4EST were tested. prot4EST_ed (4) implements ESTScan before using DECODER on any remaining untranslated sequences. Conversely, prot4EST_de (5) uses DECODER first followed by ESTScan. The DECODER module in prot4EST considers translations on both the foward and reverse strands of the query sequence. 1 Data Sets Test EST dataset for translation We randomly selected 4000 Caenorhabditis elegans ESTs from dbEST [ 30 ]. To reduce redundancy, the ESTs were clustered using CLOBB [ 31 ]. phrap [ 9 ] was then used to derive a consensus sequence for each cluster. This resulted in 2899 nucleotide sequences. To ensure that the consensuses corresponded to a coding region, we carried out a BLASTN search for each consensus against the complete C. elegans cDNA dataset available from Wormbase (version 117) [ 32 ]. Significant matches were found for 2372 consensuses. Finally, this set was used to query the C. elegans protein dataset (Wormpep version 117), thus associating each nucleotide sequence with a corresponding reference polypeptide. A final test set of 2316 consensus sequences was produced. Training datasets 1: Caenorhabditis elegans Both ESTScan and DECODER require prior gene sequence. The C. elegans RefSeq collection was obtained, comprising 21033 entries (December 2003; [ 33 ]). A Perl script constructed random training sets giving differing totals of coding nucleotides from 10000 to 350000. Four sets were assembled for each level. The build_tables script (part of the ESTScan package) was used to filter out sequences [ 18 ]. We used the same training sets to build the codon usage tables required by DECODER. CUSP from EMBOSS [ 34 ] was used to build the tables, and a separate Perl script written to convert the output to that required by DECODER. For any given run of prot4EST the ESTScan HMM training set and codon usage table used were derived from the same training set of C. elegans cDNAs. 2: Prokaryote genomes GenBank entries from 167 complete prokaryote genomes were obtained (May 2004). A Perl script was written to extract the CDS entries and construct a RefSeq-style resource for each prokaryote species (available upon request). If a taxon's genome consisted of more than one megaplasmid the sequences were combined. CDS annotation was not available for 11 genomes. We used the CDS collections for the 156 taxa to determine AT content, construct hidden Markov models and codon usage tables. 3: Arabidopsis thaliana 28960 complete CDS entries for A. thaliana were obtained from the RefSeq database [ 35 ]. 4: Spirurida (Nematoda) We queried GenBank for all complete CDS entries from species in the nematode order Spirurida. BLAST databases SwissProt (release 42.7) and TrEMBL (release 25.7) [ 14 ] were combined to give a SwissAll database. To recreate the situation facing neglected genome analysis, the accession numbers for all proteins from species in the nematode orderRhabditida were retrieved from the NEWT taxonomic database [ 36 ] and these entries (~23000) were removed from SwissAll. 2 Data collection and analysis Comparison of predicted polypeptides to the 'true' polypeptide We compared each putative polypeptide predicted from the C. elegans test dataset to its cognate reference protein using bl2seq from the NCBI distribution. Default parameters were used except for the theoretical database size (-d), set to 130000, the size of SwissProt. The blast reports were parsed using BioPerl modules [ 37 ]. Each C. elegans reference protein sequence was also compared to itself using bl2seq with default parameters. The raw and bit scores were recorded. Calculation of comparison statistics The raw and bit scores were normalised for length and against their theoretical maximum using equation 1, where: BITlocal is the bit score of the local alignment between the predicted polypeptide and its cognate reference protein, BITmax is the bit score for the alignment between the reference protein and its self, WPlength is the length of the wormpep protein that is the reference of the nucleotide consensus translated, ESTlength is the length of the nucleotide consensus that has been translated. (equation 1) Results and discussion To measure the accuracy of translation two statistics were derived from the comparison of the predicted and reference polypeptides. The coverage is the percentage of the predicted polypeptide that aligns with the reference. The bit score represents the total of the alignment's pair-wise scores, normalised with respect to the substitution matrix used to calculate these scores. In this study the bit score was itself normalised to compensate for EST length and the maximum possible bit score for each comparison (see Methods, equation 1). The number of consensuses translated that had a significant match to their cognate reference C. elegans protein was also recorded for each run. The influence of number of training codons Both variants of DECODER were unable to produce robust translations for over half the nucleotide sequences no matter how many nucleotides were in the training set (Figure 3 ). As expected, the inclusion of the reverse complement in the DECODER analysis improved its performance. The inability of DECODER to translate more than 50% of the polypeptides can be traced to its core assumptions. One criterion used is the determination of the most likely initiation methionine. While this is almost always present in full length cDNAs (for which it was designed), the occurrence of any ATG codon in EST consensuses is less certain. We noted that DECODER will try any ATG codon to start its prediction, even if this results in a polypeptide of 2 amino acids in length. The effect of the number of training nucleotides on ESTScan performance is pronounced. For the majority of the replicates, at each training set size the fraction of predictions that have significant matches to their reference sequence was around 75%, but the number of translations dropped significantly below 250000 training nucleotides. However, for 10000 coding nucleotides or less no robust translations are produced. Additionally, there was variance in the performance of ESTScan when there were between 20000 and 50000 training nucleotides. Examination of these training sets showed no difference in AT content compared to larger training sets, but did suggest that fluctuations in codon usage bias might be involved. The replicates that performed less well comprised sequences with shorter mean length, and had codon biases that were at the extremes of the distribution (not shown). This variation in sequence composition clearly has an effect on the probabilities that populate the HMM used by ESTScan. We suspect that the ability of ESTScan to predict robust translations when trained by datasets of 150000 to 200000 coding nucleotides is inflated as a consequence of the random selection of the training set from the complete C. elegans transcriptome. In a genuine situation, when only a small number of full-length CDS exist in the public databases, a significant number will be from highly expressed genes with atypical codon bias and structure. This bias will be evident in real-world CDS sets with fewer than 200 members (150000–200000 coding nucleotides). When the training sets contained a large number of non-redundant coding nucleotides (> 150000), prot4EST_ed and ESTScan performed equally well (Figure 3a ). When the number of coding nucleotides available for training and codon bias determination were reduced, prot4EST translations still showed significant similarity to the correct protein in at least 80% of instances. The translations produced by prot4EST_ed were the most robust across all totals of coding nucleotides, for both coverage and bit score (Figures 3b & 3c ). As the number of coding nucleotides used in training decreased, both measures showed slight reductions. Performance of alternative prot4EST architectures prot4EST_ed produced more robust translations for higher numbers of training sequences. However when smaller totals of training nucleotides were used the translations produced by the alternative architecture, prot4EST_de, were slightly better (Figure 3c ), although a smaller proportion of translations were produced with this setup (Figure 3a ). The better performance of prot4EST_ed was examined by following the fate of individual test sequences through the prot4EST pipeline. By employing ESTScan before DECODER, larger training sets allowed the deployment of well trained HMMs (Figure 4 ). All predictions satisfied length and quality filters, and so were accepted as robust. The corresponding DECODER predictions, while satisfying length filters, were not as robust. As the training sets decreased in size, the ESTScan predictions failed the filters and so were ignored, and DECODER used instead. Performance of similarity search Seven sequences out of 2316 were identified as rRNA in tier 1. Tiers 2 and 3 of the prot4EST pipeline exploit any significant sequence similarity between the query sequence and known proteins for coding region determination. This approach identified coding regions from just under half of the consensuses, 1131. Nineteen were identified as mitochondrial genome derived. To benchmark the similarity approach against the other probabilistic methods, the accuracy of predictions from 1131 consensuses were compared. Translations derived from prot4EST tiers 2 and 3 were more robust than those from ESTScan or DECODER (Figure 5 ). Given that an increase in the number of non-redundant coding nucleotides used to train ESTScan produces more robust translations, we attempted to use coding regions determined thus far to create larger training sets, with the expectation of improved translations. The results from the BLASTX search against the SwissAll database were checked for matches where the alignment included the start of the protein sequence. These results contained the information required to construct pseudo-CDS entries which can be added to the training set for populating the HMMs of ESTScan. In this study there were only six BLASTX alignments that provided suitable pseudo-CDS, failing to provide any significant increase in the level of non-redundant coding nucleotides. However other species we study have produced higher numbers of pseudo-CDS which prot4EST uses to give improved translations (data not shown). Effect of training set and target set sequence composition As a significant proportion of any EST set will not share similarity with known sequences, de novo translation methods need to be trained to as high a level as possible. The question is how this should be done, given the paucity of prior sequence data for individual species. Should CDS from species considered phylogenetically related be combined or should a large set from a model organism be used? A recent study of gene finding in novel genomes has shown a significant effect of sequence composition upon gene structure prediction, with more closely related model genomes providing poor training if the codon bias differs significantly from the genome of interest [ 25 ]. The performance of ESTScan was affected by even slight fluctuations in sequence composition. We examined the effect of AT content on the accuracy of translation. The complete CDS complements of 156 prokaryotes were assembled as described in the Methods. This gave a range of AT contents from 28% ( Streptomyces coelicolor ) to 78% ( Wigglesworthia glossinidia ), independent of any bias due the organisms' relatedness to C. elegans . The lowest number of non-redundant coding nucleotides was 461,299, in excess of the minimum number suggested for robust training. To explore datasets from more closely related sources all available CDS entries for the nematode order Spirurida (last common ancestor with C. elegans was 475–500 MYA [ 38 ]), and the plant Arabidopsis thaliana [ 39 ] were obtained. There was a significant correlation between AT content of the training set and the coverage by the putative polypeptides of their reference C. elegans proteins (r = 0.49 P > 0.001) (Figure 6 ). The most robust predictions were produced by HMMs trained on datasets with an AT content similar to that of C. elegans . For the prokaryote training sets, the number of nucleotides used had no significant effect upon performance (data not shown). We note that some prokaryote training sets with AT contents close to C. elegans performed poorly: homogeneity of AT content is thus not a panacea. The best performance was obtained using the A. thaliana training set, with significantly better coverage than achieved with the more closely related Spirurida. As the plant dataset contained 130 times as many coding nucleotides as did the Spirurida training set, four random A. thaliana training sets of comparable size to the Spirurida were built. These smaller training sets still performed better than the Spirurida training set, though not as well as the full CDS collection. Conclusions prot4EST is a protein translation pipeline that utilises the advantages of a number of publicly available tools. We have shown that it produces significantly more robust translations than single methods for species with little or no prior sequence data. Around three quarters of current EST projects are associated with training sets of < 50000 coding nucleotides (Figure 1 ). Thus prot4EST offers significant improvement in this real world situation. Even with substantial numbers of coding nucleotides, the use of similarity searches means prot4EST is able to outperform the best de novo methods. Given the increase in protein sequences submitted to SwissProt/TrEMBL, prot4EST's ability and accuracy can only increase over time. These more accurate translations provide the platform for more rigorous down-stream annotation. Currently we are using the prot4EST pipeline to translate ~95000 nematode consensus sequences from 30 species. These translations will then be passed onto other tools we are developing for EST analysis and annotation (see ). Availability and requirements Project name: prot4EST Project home page: Operating system(s): Fully tested on Linux – Redhat9.0, Fedora2.0. Programming language: Perl Other requirements: ESTScan2.0 DECODER rgscerg@gsc.riken.go.jp BioPerl 1.4 Transeq License: GNU GPL Any restrictions to use by non-academics: None for prot4EST source code. DECODER requires a license. See User Guide. Authors' contributions JW performed all the analyses and wrote all the Perl code. MB oversaw the project and suggested additional features. Both authors shared responsibility for writing this manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC543579.xml |
449901 | Dances as Windows into Insect Perception | Honeybees signal the location of food sources to their hive- mates using a "dancing" flight pattern. Translating these patterns, scientists learn what bees perceive | Experimental psychologists working with humans have a fundamental advantage over scientists studying the behaviour of other animals. This is because human subjects can give a verbal account of their experience. For example, they can report: ‘These two lights of different colour look equally bright’ or ‘This object looks further away than that one’. Such direct reports facilitate studying how information from the sensory periphery, that is, the sense organs that actually interface with the environment, is processed in the brain. The perceptual world of animals is often very different from that of humans. Many animals have sensory facilities that we humans lack; for example, insects can see ultraviolet and polarised light. But how they actually perceive the world, based on information from their sensory periphery, is often beyond our grasp. Because animals cannot describe their sensations, our access to them is often based on indirect psychophysical tests, where animal performance depends fundamentally on motivation and training method ( Chittka et al. 2003 ). However, some animals do in fact describe the world around them, but not necessarily in ways that we might intuitively understand. Perhaps the best example of this are the honeybees (genus Apis ), which have a symbolic ‘language’ that nestmates use to communicate with each other about profitable food sources. By eavesdropping on this communication, scientists have recently obtained a unique perspective into the perceptual world of insects. How does the dance language work? A triumphant scout bee returns from the field, and advertises the location of a newly discovered food source to nestmates. To do this, the forager performs a repetitive sequence of movements, the so-called waggle dance, which is one of the most intriguing examples of complex animal behaviour. The successful forager wiggles her abdomen provocatively from side to side, moving forward in a straight line. Then she runs in a half circle to the left, back to her starting point, performs another straight wiggle run along the path of her first, and then circles to the right ( Figure 1 ). This pattern is repeated multiple times, and is eagerly attended by unemployed bees in the hive. Shortly after such dances commence, dozens of newly recruited foragers arrive at the food source being advertised. Figure 1 Figure-Eight-Shaped Waggle Dance of the Honeybee ( Apis mellifera ) A waggle run oriented 45° to the right of ‘up’ on the vertical comb (A) indicates a food source 45° to the right of the direction of the sun outside the hive (B). The abdomen of the dancer appears blurred because of the rapid motion from side to side. (Figure design: J. Tautz and M. Kleinhenz, Beegroup Würzburg.) In the 1940s, Nobel laureate Karl von Frisch deciphered the code hidden in this seemingly senseless choreography performed on vertical honeycombs in the darkness of the hive (reviewed in von Frisch 1967 ). He found that the angle of the waggle run from the vertical is equal to the angle between the sun's azimuth and the indicated food source outside the hive. For example, if a food source is found in the direction of the sun, the dancer will waggle ‘straight up’ the vertical comb. If food is found 45° to the right of the sun's direction, the waggle run will be oriented 45° to the right of vertical on the comb ( Figure 1 ). The distance to the target, a flower patch with abundant nectar or pollen, is encoded in the duration of the waggle run: the longer the bee waggles, the larger the distance of the food from the hive. No other species (besides humans) uses a similarly symbolic representation to communicate information from the real world. But how do bees measure the flight distance that they communicate so precisely? It was previously thought they do this by measuring the energy used as they fly ( Heran 1956 ). However, doubts emerged when it was found that distance estimation by bees could be manipulated by altering the number of landmarks between the hive and a food source, suggesting bees were counting landmarks encountered en route ( Chittka and Geiger 1995 ). In an elegant experiment, Esch and Burns (1995) tapped into the bees' dance language to access their subjective assessment of flight distance. They let bees forage from a food source 70 m from the hive and recorded the dance distance code of the returning foragers. Subsequently, the feeder was attached to a weather balloon, and slowly lifted to an altitude of 90 m—so that the distance between the hive and the food now increased from 70 m to 114 m. Correspondingly, foragers should have indicated a longer distance, by stretching their waggle run duration. But, in fact, the perceived distance (as indicated in the dance) decreased by more than 50%! This clearly shows that bee perception of distance cannot solely be based on energy expenditure, since a longer flight that cost more energy was danced as a shorter ‘distance’ in the waggle run. So what actually drives the bee odometer? Because the landscape bees pass in flight moves more slowly when bees fly at higher altitudes, Esch and Burns (1995) conjectured that foragers process the speed with which visual contours move across the eye (optic flow), and integrate this with travel time. To confirm this hypothesis, Srinivasan et al. (2000) further exaggerated the experienced image flow, by training bees to fly through narrow chequered tunnels. These bees grossly overestimated actual travel distance, bragging to their nestmates that they had flown 195 m when in fact they had flown 6 m. Attendees of these dances promptly believed the high-class swindle, and searched for food at remote locations that the dancers had never even visited ( Esch et al. 2001 ). The quality of information available about the velocity of the passing landscape will depend, of course, on the sensitivity of the eyes. The eyes of bees contain three types of colour receptors, with maximum sensitivity in the ultraviolet, blue, and green domains of the spectrum ( Autrum and von Zwehl 1964 ). Their excellent colour vision is optimal for flower identification ( Chittka 1996 ), but do they also use it to measure the image velocity of the passing landscape? Surprisingly, the answer is no—bee odometry is in fact totally colour blind. Chittka and Tautz (2003) found that bees use exclusively the signal from their green receptors for measuring image velocity ( Figure 2 ), confirming earlier reports that motion vision in bees is mediated only by this receptor type ( Giurfa and Lehrer 2001 ; Spaethe et al. 2001 ). Thus, the level of intensity contrast present in the scene strongly influences the bees' subjective experience of flight distance ( Chittka and Tautz 2003 ; Si et al. 2003 ). Figure 2 Bees Use Different Visual Cues When Viewing Flowers and Landscape Image Motion Although bees see flowers in colour, they do not analyse the colours of the landscape image that moves across the eye as they fly. Their perception of landscape motion is colour-blind; motion vision is driven solely by a single spectral receptor type, the bees' green receptor. This is reflected in the distance code of the dance: the more green contrast is present in the scene, the further bees ‘think’ they have flown. (Figure design: F. Bock, Beegroup Würzburg.) With so many external variables influencing distance estimation, it seems unlikely that the honeybee odometer would be very robust in natural conditions. Now, as reported in this issue of PLoS Biology , Tautz et al. (2004) have quantified the bees' subjective experience of distance travelled when they fly over natural terrain with varying levels of contrast. Specifically, they compared the dances of bees flying over water (scenery with low visual contrast) with those of bees flying over land (scenery with relatively high contrast). They trained bees to forage at a feeder on a boat, which was paddled increasing distances from the hive, until it reached an island. All the while, observers at the hive deciphered the dances of the bees returning from the feeder. Interestingly, bees flying 200 m over water hardly appeared to register an increase in travel distance, whereas the same increase in distance flown over land resulted in a substantial increase in perceived flight distance. This is consistent with the hypothesis that the bees' odometer is largely based on visual, external cues and demonstrates that this system is sensitive to visual contrast. But there must be something else beside visual cues. Navigation over water, in the near absence of visible ground features, is extremely difficult without a reliable internal instrument measuring travel speed. This is the case even for us humans with sophisticated measuring devices: malfunctioning air speed indicators have been responsible for several airplane crashes into water, for example Birgenair Flight 301 and AeroPeru Flight 603 in 1996. Heran and Lindauer (1963) likewise observed that honeybees flying over lakes sometimes lost altitude and plunged into the water. However, the new study by Tautz et al. (2004) also shows that most bees will reliably fly over prolonged stretches of water without accident. Furthermore, even though bees experience only a small increase in subjective travel distance when flying over water, it is not zero. This indicates that bees do perhaps resort to an internal measure of flight distance when other cues fail. For example, bumblebees walking to a food source in absolute darkness, that is, in the complete absence of visual cues, are able to correctly gauge travel distance ( Chittka et al. 1999 ), indicating that an internal odometer, possibly based on energy consumption, also exists. It appears that animal navigation, just like aviation, relies on multiple backup systems that support each other and can compensate if one system fails in a certain context. Spying on honeybee dances can not only tell us about the cues they use for navigation, but also allows insights into the cognitive architecture that governs other aspects of bee behaviour, such as the assessment of flower quality. We've learned that bees prefer high over low nectar concentrations because this is reflected in their dances. When bees find better nectar, they dance more enthusiastically, that is, the number of dance circuits per minute increases ( Seeley et al. 2000 ; Waddington 2001 ). However, Waddington (2001) found that the relationship between actual and perceived nectar quality is nonlinear. In fact, it is a positive but decelerating relationship, so that an increase in sucrose concentration from 10% to 20% results in twice the difference in dance rate that an increase from 50% to 60% does. Interestingly, the perceived change in quality is stronger when there is a decrease than when there is an increase in nectar quality of the same magnitude. Such asymmetric perception of gains and losses is well known in humans, where it has been linked to risk-aversive behaviour ( Tversky and Kahnemann 1981 ). Unfortunately, animal subjects often do not yield this type of information very readily. Only in their own language do they reveal many of their perceptual peculiarities. Using the bee language as a window into insect visual perception has been a wonderful tool and is a promising avenue for further research into the question of how miniature brains encode the world around them. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC449901.xml |
549083 | Resection of small plexiform neurofibromas in neurofibromatosis type 1 children | Background Plexiform neurofibromas (PNF) are benign tumors of the peripheral nerve which mostly develop in patients with neurofibromatosis type 1 (NF1). Surgical interventions are usually not applied to children with small tumors. These are rather restricted to debulking of larger tumors in adults that cause clinical complications or aesthetic disfigurement. In most cases, a total resection of PNF is not possible due to the network-like growth of the tumors. Patients and methods Early surgical intervention was carried out for 9 small PNFs in 7 NF1 children. Tumor resection was performed following the graphical delineation of the affected skin and according the MRI findings. Results Total resection was achieved for all 9 PNF without causing any neurological or organic deficit. Annual magnetic resonance tomography over a period of four years did not reveal any relapse of the tumors. Conclusions Early surgical intervention for small superficial PNFs in NF1 children have various advantages and may especially be considered a strategy to prevent progression. | Background Plexiform neurofibromas (PNF) are benign tumors originating from nerve sheath cells, subcutaneous, or visceral peripheral nerves and can involve multiple fascicles [ 1 ]. PNF occur almost exclusively in patients with NF1, an autosomal dominant disorder caused by defect of one allele of the tumor suppressor gene, NF1 on 17q [ 2 - 5 ]. At least 30% of NF1 patients suffer from PNF [ 2 , 6 , 7 ], which are often present at birth and progress during the first years of life. The growth rate and pattern of PNF vary to a large extent and their growth spurts are unpredictable. PNF can arise in various parts of the body, for example as anterior mediastinal masses, sciatic nerve lesions with pelvic extension, or perirectal plexiform and uterine tumors, often leading to severe clinical complications [ 7 ]. Especially tumors occurring in the head and neck area often lead to facial disfigurement and functional deficits. Due to large size of the tumors and involvement of multiple fascicles of nerves and tissues, the risk of neurological and functional destruction upon tumor resection is high. Surgical interventions are thus commonly postponed as long as possible. In addition, most surgical interventions are limited to debulking and rests of tumors often re-grow afterwards leading to the requirement of repeated intervention [ 8 - 10 ]. Previously we reported that PNF can be distinguished into three growth categories using magnetic resonance imaging (MRI): superficial, displacing and invasive. Superficial PNF arise from subcutaneous or cutaneous nerves and may remain within the upper layer of the skin – usually not involving major nerves [ 10 ]. Subtotal and total resection without functional destruction is often possible for superficial PNF, as demonstrated in our recent study [ 11 ]. In contrast, invasive PNF infiltrate multiple tissue planes and are thus much more difficult or impossible to resect. Superficial PNF show progressive growth and it is unknown whether or not a superficial PNF may change to displacing or invasive types. Early resection of small superficial PNF may thus be considered as an advantageous treatment option. In this study, we present the results of early intervention for 9 PNFs in 7 NF1 children aged 3 to 15 years. Patients and methods The 7 children were examined in our NF-Clinic in the Department of Maxillofacial Surgery, University Hospital Eppendorf, Hamburg. Diagnosis of NF1 was based on the NIH criteria [ 6 ]. Informed consent was obtained from parents of these children. All patients received dermatological, neurological and ophthalmological examinations as well as an ultrasound of abdominal organs. PNF were diagnosed based on the following indications: subcutaneous location on palpation, associated with thickening of the skin, local hypertrophy, hair excess and hyperpigmentation on inspection, and prepubertal occurrence from the medical report. Magnetic resonance image (MRI) was done for the tumor regions at 1.5 Tesla with T1- and T2- weighted sequences including a short-tau inversion-recovery (STIR) sequence. Ultra rapid half Fourier single-shot turbo spin-echo (HASTE) sequences were used for imaging the trunk. Intravenous contrast medium was given to all patients. Tumor resection was performed following the graphical delineation of the affected skin and according the MRI findings. Tumor resection included the epidermal and subcutaneous layer with safety margins of 1 cm in all directions. Exploration of the underlying muscles was mandatory. Results Seven NF1 children, five boys and two girls, aged 3 to 15 years were included in this study. All of them met the NIH diagnostic criteria for NF1 [ 6 ]. PNF was initially diagnosed clinically and further ascertained by MRI (Figs. 1 , 2 , 3 , 4 , 5 , 6 , 7 ). Five tumors were histologically confirmed as PNF (Fig. 8 ). Other four were reported as diffuse neurofibromas due to lack of nerve fascicle in the examined sections. However, these tumors were parts of PNF. Five PNF in four children exhibited hyperpigmentation while the other two had hair excess in the tumor area. Two children had two and the other five had one PNF each. Size of tumors varied from 2 cm to 8 cm in greatest diameter (table 1 ). All PNF were superficial based on MRI. The tumor location was variable (table 1 ). Neither functional deficit nor pain was caused by the tumors at the time of surgery. Neither was there significant aesthetic disfigurement caused by the tumors. Figure 1 1A: Pre-surgical T2-weighted STIR-sequence, axial section: in right ventral chest wall bright superficial thickening of cutis and subcutis without involvement of muscles. 1B: Pre-surgical T2-weighted STIR-sequence, axial section: very bright, flat superficial cutaneous PNF in proximal part of right upper arm. 1C: Pre-surgical clinical frontal view of both PNF on right upper arm and thorax wall. 1D: Post-surgical T1-weighted sequence, axial section: small defect in cutis and subcutis, no PNF visible anymore. 1E: Post-surgical T2-weighted STIR-sequence, axial section: complete removal of tumor with hypointensive induration. 1F: Post-surgical clinical frontal view of both scars on right upper arm and thorax wall. Figure 2 2A: Pre-surgical T2-weighted STIR-sequence, axial section: superficial tumor covering the laryngeal prominence. 2B: Pre-surgical clinical frontal view of PNF of ventral neck. 2C: Post-surgical T2-weighted STIR-sequence, axial section: scar tissue praelaryngeal visible. 2D: Post-surgical clinical frontal view of laryngeal scar. Figure 3 3A: Pre-surgical T2-weighted STIR sequence of left ventral abdominal wall, paraumbilical. Bright plexiform neurofibroma with thickening of cutis without involvement of muscle. 3B: Pre-surgical T2-weighted STIR sequence, axial section of left ventral abdominal wall, costal margin. Small bright plexiform neurofibroma in cutis and subcutis ventral left. 3C: Pre-surgical clinical frontal view of abdomen with both superficial plexiform neurofibromas visible. 3D: Post-surgical MRI control from A. Complete removal of plexiform neurofibroma. 3E: Post-surgical MRI control from B. Complete removal of plexiform neurofibroma. 3F: Post-surgical clinical frontal view of abdomen. Only two visible scars left after surgery. Figure 4 4A: Pre-surgical T2-weighted STIR-sequence, axial section: discrete signs of flat funicular, cutaneous and subcutaneous PNF near the left iliac crest with involvement of soft tissue, but without visible infiltration of abdominal muscles. 4B: Pre-surgical clinical view of PNF in the left flank. 4C: Post-surgical T2-weighted Haste-sequence: complete resection of tumor, only a smooth fibrous post-surgical induration and small scar can be identified. 4D: Post-surgical clinical view of hypertrophic scar. Figure 5 5A: Pre-surgical T2-weighted STIR-sequence, axial section; small nodular hyperintensive PNF in the cutis and subcutis on left ventral chest wall. 5B: Pre-surgical clinical view of PNF in left chest wall, parasternal. 5C: Post-surgical T2-weighted Turbo Spin Echo-sequence, axial section: tumor no longer visible, only adipose tissue visible in the subcutis. 5D: Post-surgical clinical view of scar. Figure 6 6A: Pre-surgical T2-weighted STIR-sequence, transversal section: bright, flat cutaneous and subcutaneous PNF of right back without involvement of abdominal wall and muscles. 6B: Pre-surgical clinical view of PNF on right back with hypertrichosis. 6C: Post-surgical T2-weighted Turbo Spin Echo-sequence, axial section: complete removal of PNF, thin scar, no subcutaneous fatty tissue visible in scan. 6D: Post-surgical clinical view of scar after tumor removal on back. Figure 7 7A: Pre-surgical T2-weighted STIR-sequence, transversal section: Cutaneous and subcutaneous PNF of forearm, bright signal without involvement of muscles or fascia. 7B: Pre-surigcal clinical view of hyperpigmented PNF of left forearm. 7C: Post-surgical T2-weighted STIR-sequence, axial section: complete removal of tumor. 7D: Post-surgical clinical view of scar on left forearm. Figure 8 Photomicrograph of PNF. Plexiform neurofibroma WHO grade I invading fascicles of a peripheral nerve. Tumor growth is primarily confined to the endoneurial space; the perineurium (arrow) forms a natural border for the tumor. Note the loose myoxid texture of the tumor tissue Table 1 Clinical information Patient number Sex Age at surgery Tumor location Hyper pigmentation hair growth Size (greatest diameter) Histology post-operative MRI Annual MRI follow up 1 f 5 upper arm right no no 4 cm PNF yes yes 5 thorax right yes no 4 cm PNF yes yes 2 m 15 Adams apple no no 5 cm PNF no yes 3 f 14 rib bow left no no 4 cm PNF no yes 14 paraumbilical left no no 3 cm PNF no yes 4 m 3 thigh left yes yes 7 cm PNF no yes 5 m 4 thorax left yes no 2 cm PNF yes yes 6 m 7 back right yes yes 8 cm PNF yes yes 7 m 11 lower arm left yes no 5 cm PNF no only 2003 In 2001, all seven children underwent surgical intervention for their nine PNF. During the mobilization of the underlying skin layers, three patients showed enlarged nerves running from subcutis to superficial muscle layers, which were identified and additionally resected. These enlarged nerves were proven to be PNF in all cases. Primary wound closure was achieved following the mobilization of the marginal skin. Healing followed without complications in all cases. The scars are relatively small and no hypertrophy was observed. The intervention was well tolerated by all children and recovery occurred within a few days. Postoperative MRI was done for three children. Six children underwent follow-up examinations annually by MRI for three years. One child was only examined in 2003, two years after the operation. MRI revealed successful total resection for all 9 tumors (Figs. 1 , 2 , 3 , 4 , 5 , 6 , 7 ). So far, no tumor re-growth has been detected in any of the children. Discussion In this study we demonstrated that small superficial PNF can be completely resected, leaving scars without hypertrophy. Operation is relatively uncomplicated, allows healing by primary intention, requires only a few days of hospitalization and is thus no burden for even the youngsters. Within the clinical and radiological follow-up period of three years there was no re-growth of the tumors. However, it cannot be excluded that all tumor cells have been resected. Therefore those patients warrant further annual clinical examination. The scars in some children may seem more obvious and disfiguring after the operation. However, in comparison to the complications the tumors may cause later, they are only of minor concern. Furthermore, these scars may be corrected later surgically. Without resection, these small PNF are likely to grow continuously to a large or even very large size which often causes aesthetic disfigurement, functional deficits and pain. Resection of large tumors is much more difficult and total resection is usually not possible any more. Currently it is not clarified whether or not growth patterns of PNF might possibly change from superficial to displacing or to invasive types over time. Early surgical intervention of small superficial PNF may thus be considered as a preventive strategy for later disfigurement and functional deficits. Adequate diagnosis of NF1 and thus small PNF in the pediatric age group is challenging. Children usually do not complain about subtle pain or discomfort in the affected skin region. Even in the presence of some symptoms parents are often not aware of the origin of the tumor. Only small PNF of the face and neck causing disfigurement in children tend to be recognized early. On the other hand, small PNF of the trunk and extremities which present with subtle hyperpigmentation, hair excess and palpable tumor tissue are frequently not noticed by the physicians. These tumors are rarely diagnosed correctly as PNF. In our cohort only one child received the diagnosis of a PNF correctly by the referring family physician. For better management and adequate treatment, efforts should be made to diagnose NF1 and PNF in their early stage. Conclusions Early surgical intervention of small superficial PNFs is uncomplicated, without burden for even the youngsters and enables total resection of the tumors. It may be considered as a preventive strategy for later disfigurement and functional deficits. Competing interests The authors declare that they have no competing interests. Authors' contributions REF and RS carried out the surgical intervention for the tumors and post-operational care of the patients. CF carried out the MRI and was involved in the diagnosis of PNF. MH and VFM did the diagnosis and were responsible for management and consulting of the patients. REF and VFM designed the study and prepared the manuscript. All authors read and approved the final version of the manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549083.xml |
546401 | A temperament for learning: The limbic system and myelomeningocele | This article was the winner of the triennial Casey Holter Memorial Prize awarded by the Society for Research into Hydrocephalus and Spina Bifida, 2004. Abstract This essay explores the link between the limbic/hypothalamic systems within the complex conditions of hydrocephalus and myelomeningocele. Acknowledging the neuroanatomical and neuroendocrine risks inherent in the developing brains of these individuals, we focus on the converging components of temperament, cognition, and language. | Introduction Children with myelomeningocele (MM) do not easily fit into stereotypical profiles. Professionals, families, and inter-disciplinary teams, devoted to the support of these individuals and the reduction of secondary disabilities, see daily testaments to individual variations. Clinical commonalities among those with MM are frequently noted and perhaps too often – and too simplistically – considered inherent to the condition. Behavioral characteristics have long been one such category for easy generalities. As modern investigators study the underlying keys to cognitive and behavioral disabilities, theories related to cortical development and differentiation have offered insightful possibilities [ 1 ]. Much remains to be discovered in the realm of behavior, emotions, and personality profiles within this group and the biologic differences that may explain them. This essay is submitted with the suggestion that in our ongoing quest to understand the learning and behavioral characteristics in youth with MM, focus might reasonably be placed on cellular and neurobiological mechanisms specifically related to the limbic and hypothalamic systems. Given the developmental shifts in neuro-architecture in the fetus and infant with MM, we offer the limbic system with its cortical, and brain stem interconnections, and particularly its close association with the hypothalamic region, as an area wherein many of the phenotypic commonalities may arise. Similarities and variations in temperament profiles among children with MM present an opportunity to explore and enhance our understanding of this ontologically old but, as yet, not fully understood portion of the brain. Temperament patterns among this group, differing from other cohorts, may be reflective of altered integrity within this intricate neural system. The converging components of temperament, memory, and language in this group of children with developmental differences may help the researcher and the clinician to better address the so-called behavioral issues that impact academic learning. We will re-visit established and emerging information related to the limbic and hypothalamic systems. We will review reports from recent studies on language and on temperament among children with MM, and comparative work focused with different at-risk populations of children. These will be considered along with other clinically relevant phenotypic findings among children with MM that have clear links to the limbic and hypothalamic systems. Le Grand Lobe Limbique The limbic system consists of structures reminiscent of the old mammalian brain corresponding to that of the so-called higher mammals [ 2 , 3 ]. The various components of the limbic system influence a diversity of functions that are integral to the cognitive aspects of autonomic, affective, and sexual behavior [ 4 ]. Components of the system and their general activities that might be expected to impact memory and/or temperament include the following: Amygdala – little almond shaped structure near the temporal pole. It receives catecholamine and 5-HT containing projections from the brain stem. Its most prominent projection (stria terminalis) runs in the wall of the lateral ventricle, ultimately connecting to the hypothalamic center. Along with other behaviors (particularly sexual), it mediates the major affective activities describe as fear [ 3 , 5 , 6 ]. Anterior thalamic nucleus – associated with variations in emotional reactivity [ 3 ]. Cingulate gyrus – located on the medial side of brain near the corpus callosum. Implicated in encoding and retrieval of semantic and episodic memories, attention, drive and pain perception [ 7 ]. Hippocampus – formed from the inferior portion of the temporal lobe into the lateral ventricle. It is involved in memory, especially long-term memory. When both hippocampi are completely ablated, nothing can be retained in the memory. The intact hippocampus allows comparison of the conditions of a present challenge with similar previous experiences. This process is central to the ability to make choices [ 3 , 8 ]. This can be critical for survival in the wild – or perhaps survival in the classroom. Entorhinal, subicular cortices – are implicated in spatial memory storage, connect hippocampus to remainder of temporal cortex [ 2 ]. Septal nuclei – a subcortical target of hippocampus [ 2 ]. Mamillary complex – is the major limbic-hypothalamic area [ 2 ]. From a cellular and structural aspect, the proximity of limbic structures to the ventricles hints at the possibility of neuropathological changes in these structures either directly through mechanical compression effects of the dilated ventricles, or indirectly via alterations of metabolic pathways in children with MM [ 9 - 11 ]. Structural changes, and potential functional sequellae, may also arise through deficiencies in the developmental process when HC occurs in the fetus. Behaviorally, an emerging body of neuropsychological data suggests a critical relationship between these structures, their functions, and their relationship to emotion, memory, and learning. Tell-tale signs of limbic influence: Learning and Temperament Much has been written over the past decade describing phenotypic profiles related to executive functions among youth with MM. While the term executive function continues to be somewhat enigmatic, it has generally been used to encompass a set of abilities including: sustained focused attention and organizational skills, goal-directed behaviors, flexibility with novel situations, generation of unique plans, and working memory functions [ 12 , 13 ]. Long relied-upon anecdotal concepts (short attention spans, poor organization skills, and memory problems) have given way to more controlled studies [ 14 - 19 ]. Language skills evaluated in recent research described both language production and comprehension within theoretically grounded language subsystems [ 20 - 30 ]. From this, a modal profile for MM has been proposed: strengths in syntax and lexicon; and weaknesses in pragmatic communication, making inferences, and understanding nonliteral language [ 31 ]. Research supporting this modal profile for children with MM (especially as it relates to higher-order language skills) is based on their inclusion within the larger heterogeneous samples of children with hydrocephalus (HC) whose conditions were associated with multiple etiologies, both congenital and acquired (e.g. refs. [ 22 , 30 ]). Recent studies of a large sample of children with shunted HC and MM only , demonstrated that these children scored lower than age-matched controls in all areas of language skills – even in the more basic lexical semantic skills which, have long been considered "strengths" [ 26 , 29 ]. But two areas provided the children with particular difficulty: supralinguistic skills (making inferences and understanding nonliteral language) and pragmatic judgment. Cognitive skills of attention/executive function and memory are deeply embedded within acquisition and appropriate application of language [ 32 , 33 ]. The ability to make inferences and understand ambiguities (both supralinguistic tasks) involves the integration of previously learned world and linguistic knowledge with the new knowledge presented [ 34 , 35 ]. For a pragmatically competent communicator, successful language outcomes depend on adequately functioning systems of memory (encoding and retrieval of experiential and linguistic knowledge), attention (ability to focus on the presented communicative task), and problem solving (including the ability to judge and utilize social situations). The few studies that measured memory function in mixed samples of children with MM and HC have demonstrated memory deficits in encoding and retrieval both for verbal and nonverbal information [ 16 , 18 ]. Similarly, current research seems to suggest a prevalence of attention/executive deficits in children with MM [ 14 , 15 , 17 , 19 , 36 , 37 ]. A more than moderate interaction between the successful development and function of the limbic structures and successful learning experiences among children with MM seems likely. Structural or functional malformations of this system should impact the unique learning and cognitive profiles of these children. Prefrontal regions have been identified as critical in functions of memory, language and logical reasoning [ 38 - 40 ]. Animal models and clinical evidence suggest the integral role of the limbic system. The uni-modal and poly-modal inputs from neocortical association areas (including prefrontal), important for information processing, storage, encoding and retrieval, all rely on competent structures within the limbic system [ 8 , 41 - 46 ]. How might these facts relate to our children with MM? Shortly after the neural tube is formed, its caudal portion develops into the spinal cord and the rostral portion into the brain [ 3 ]. Of the three early brain vesicles, the forebrain gives rise to the telencephalon and diencephalon from which structures of the limbic system and the hypothalamus arise. The limbic system, arising from this phylogenetically ancient region of the brain, incorporates the hippocampus, the amygdala, and numerous receptor sites for adrenal and endocrine neuromodulators. MM, resulting from a developmental disruption of this normal neuroembryogenic process, is commonly characterized by its more visual and tangible manifestations: defects within the bony vertebra, associated impairment of the spinal cord, and the functional motor and urological deficits. The clinical manifestations of underlying neurodevelopmental aberrations centrally (the Arnold-Chiari II malformation, ventricular variations, callosal variations, and associated migrational anomalies) are frequently as disabling as the more obvious motor deficits. Structural and functional differences can result from developmental and/or mechanical disruptions in the formative stages – prenatally or thereafter. These central nervous system anomalies, often accompanying the MM, would be expected to influence behavior and learning in these children [e.g. [ 47 , 48 ]]. Developmental disruptions such as those related to MM can also provide "spin-off" developmental deficits related to anatomy and physiology arising from structures that form the limbic system and hypothalamus. Similarly, the potential exists for the developing HC, with its cerebrospinal fluid blockage within and around the brain, to impact the architecture of the emerging limbic system and its neuronal projections. For example, vacuolization and degeneration of neurons in the hippocampal formation have been observed both within hydrocephalic rabbits and humans [ 9 ]. Dendritic changes were observed in the hippocampus of neonatal rats with kaolin-induced HC [ 10 ], and HC exacerbated hypoglycemic injury in rat hippocampal cells [ 49 ]. Indeed in a rat model of kaolin-induced HC, the resultant impairment of glucose metabolism was first observed in the CA3 region of the hippocampus [ 50 ], suggesting this area is metabolically vulnerable to dysfunction. Further support of limbic involvement in HC comes from a recent study that reports damage to the fimbria/fornix of the hippocampus from autopsy-acquired brains of humans with chronic HC [ 51 ]. Clinical evidence from patients, and recent brain imaging studies suggest lesions of the right hippocampus result in spatial memory deficits while left lesions impact verbal memory [ 3 , 8 , 52 ]. Thus, resultant left and right CA3 dysfunction potentially contributes to the verbal and nonverbal memory deficits noted in children with MM and HC. Further, findings by Dolan and Fletcher [ 53 ] provide evidence that the left hippocampus is active in encoding of episodic memories by registering and processing novel verbal material, the end-product being the so-called memory trace or engram. The inability of children with MM and HC to respond efficiently to novel stimuli is well documented [ 54 , 55 ]. Anatomical and central molecular dynamics involved with these language and memory deficits remain critically important areas for future research. Equally intriguing are those – or other – molecular substances and pathways that mediate the response to the perceived frustrations associated with functional deficits, which come into play daily and hourly. What underlies and mediates the style of behavioral responses? In other words, how does the child's temperament phenotype become an integral part of the clinical management, and neurophysiological understanding of learning differences among children with MM? This task of describing individual differences among individuals within a society is ancient in its origin. Early Chinese tradition offers a view of human nature differences based on a dark, female, earthly force ( yin ) and the accompanying active, light, heavenly force ( yang ). The great 2 nd century C.E. Roman physician, Galen, conceived a system of balanced equilibrium of four basic humors, which allowed the physician clues as to the nature of an illness by his awareness of personality traits. This system, based on yellow or black bile and phlegmatic or sanguine natures, represented a culmination of prior ideas (Hindu, Greek, Chinese, others) and provided the understanding of the relationship of temperament to health and function for more than a millennium [ 56 ]. More modern scientists have provided differing slants regarding temperament characteristics of children. The list includes Bates, Buss, Carey, Chess and Thomas, Goldsmith, Kagan, and Plomin [ 56 - 61 ]. The nature of temperamental categories, taking the various researchers' ideas into account, seems to embody four qualities described by Kagan: 1) variability among individuals, 2) relative stability over time and situation within the individual, 3) under some genetic influence, and 4) emergence early in life [ 56 ]. Temperament has been described in typically developing children as the "how " of behavior distinct from ability or "what" of behaving, and motivation or "why" of behavior [ 60 ]. One recurring theme among the conceptual descriptions of temperament remains the notion of multiple behaviors, along a continuum, which combine to interact either favorably or unfavorably with the surrounding environment. The nature of these behaviors is not at the level of psychopathology. Rather, the potential for pathologic development lies within the process of the interaction – the "goodness of fit" – between the environment and the individual's behavioral responses. Consonance between the child and his/her environment potentiates optimal development and supportable teachable moments [ 62 ]. Conversely dissonance between the child's capacities, and style of behavior and the environment demands result in maladaptive functioning. Such was the case with Annie. Annie's Story Annie is an attractive little girl, eight years old, and with abundant red-haired curls. Her diagnosis of MM and shunted HC requires part-time wheelchair use as she attends a public school and participates in recreation activities in the community. Intelligent and engaging, she has, nonetheless, met frustrations during her days at school. Variably described as shy, capable, anxious, and/or self-doubting, teachers and family members know of Annie's apparent emotional lability. Recent academic testing reconfirmed her cognitive strengths with a full scale IQ in the above-average range. Language evaluation showed "relative strengths" in lexicon and syntax, lesser scores in pragmatics, and significant problems with higher-order language, and cognitive tasks that required quick retrieval of stored information. Of note, during the initial testing, Annie became tearful quite quickly. Her negative responses related to the new tasks, new environments, and her persisting fear of "not being able to finish" (on any given activity). The examiner, familiar with children with MM, recognized the triggers and the responses before her. By careful explanations, negotiating cues to assure Annie of her successes in the testing, and agreements for "breaks" as needed, Annie calmed, warmed to the tasks at hand, and completed the testing successfully in the allotted time. The developmental pediatric team was recently notified by the family that Annie was experiencing episodes at school of skin flushing, sweating, sensations of nausea, and emotional stress. On one occasion, urinary incontinence accompanied the episode. These symptoms were eventually noted to be related to those (infrequent) occasions when the teacher felt it necessary to enter a note onto Annie's personal folder for parent review. Beginning hints of school avoidance prompted the parents' concern. Academic challenges and barriers as in the story of Annie are quite common among youth with MM. Layers of seemingly separate but simultaneous factors – cognitive, memory, language, temperament, physical – can elevate even daily and mundane tasks into sources of stress not experienced by the typical classmate. Recent studies of youth with MM and shunted HC, ages 5–12, demonstrated clusters of findings different from those classically described by Chess and Thomas ("easy", "difficult", or "slow to warm") or by Kagan ("inhibited" or "uninhibited"). Children with MM in these studies were reported to be significantly different from normative profiles in five areas: 1) adaptability – less; 2) distractibility – more; 3) approach – guarded; 4) persistence on task – less; and 5) predictability – less [ 55 ]. Given these temperament characteristics in a child with MM, perhaps related to the unique differences in both central molecular dynamics and neurodevelopmental anatomy, responses such as Annie's begin to be understood. While the focused study of the biological basis for temperament dates to mid-20 th century, only recently have physiologic variables and temperament categorical descriptors been paired for critical comparisons. These investigations stem from relatively firm notions about the anatomy and physiology of the central nervous system. Researchers have tracked differing patterns of social response, and encoding of emotional memories to the limbic system, and particularly amygdala [ 63 , 64 ]. The amygdala has been implicated in social cognition, involving adequate recognition and judgment of facial expressions [ 3 , 65 ]. Preliminary studies within our institution reveal our patients with MM and shunted HC had difficulty understanding nonverbal facial cues and expressions of annoyance (explaining in part their poor pragmatic skills). Opioid-mediated inhibitory activities, or responses to the dozens of neurotransmitters or neuropeptides both have genetic implications and biochemical importance as they potentially affect excitability centrally within the amygdala or connecting structures [ 56 ]. Potential changes in functional outcomes might be anticipated. These limbic structures – and their vulnerability in the developing brain affected by MM – offer fertile ground for clinical exploration in the broader study of cognition, memory, and temperament. The segregation of emotion, temperament, language, and cognitive performance is difficult to accomplish in the typical child; it is decidedly more complex in the child with MM and HC who has co-existent ventriculomegaly from early gestation, mid-brain variations, and cerebellar differences. What then are the capacities and activities of these central nervous system components – the limbic and hypothalamic systems – that are typically relied upon for daily function, and which might have significant impact on those outward actions and responses which we categorize as temperament? Emotional triggers, goal direction for problem solving, assimilation of information from the various senses (vision, touch, auditory) – all of these relate to the anatomy and associated function of the limbic system (notably the amygdala), and the closely related hypothalamic region. Evidence from human and animal studies indicates that the amygdala intervenes between the regions concerned with the somatic expression of emotion (hypothalamus, brainstem) and the neocortical areas concerned with conscious feeling [ 3 ]. Similarly, the hypothalamus is a central player in the integration of information. For the specific areas throughout the neocortex to finalize and actualize those processes described as "executive functioning", regulation of stimuli from the environment and peripheral systems of the body is aided through hypothalamic function. Some actions might well be expected to impact behavior we characterize as components of temperament. The anatomy of the hypothalamus is important when considered along with the variations commonly noted in our individuals with ventriculomegaly and Chiari malformations. It lies on the ventral surface of the cortex. The third ventricle divides the hypothalamus and lies in close proximity to hypothalamic projections [ 66 ]. Disturbances or disruptions in these structures due to HC would not be unexpected. Generally, signs or symptoms associated with hypothalamic dysfunction might include: disorders of satiety, appetite and/or caloric homeostasis; sexual dysfunction including precocious puberty or hypogonadism; central autonomic disorders (sleep and consciousness regulation, gastrointestinal function, thermoregulatory control, sphincter disturbance); and affective variations [ 3 , 66 ]. Corticotrophin releasing factor, related to the paraventricular nucleus, is integrally involved in the management of autonomic centers within the brainstem and, via regulation of corticosteroids, with stress adaptation [ 3 ]. The effects of HC on neuronal pathology and function in these regions have been demonstrated in both human studies and animal models [ 10 , 49 , 50 , 67 ]. Such variations in anatomy and fetal development of the midbrain structures – areas of tremendous variations among individuals with MM – could play important roles in the ultimate behavioral phenotypes described by temperament. For example, in the temperament profile of the MM cohort described above, attention was a significant outlier. Mirsky [ 68 ] has reminded us of the multi-component nature of attention. The nature of the neuropsychological model of attention depends upon the data used to generate it. The network involved in the distribution of attention to extra-personal targets implicates cortico-limbic-reticular and hypothalamic circuits; damage to either impacts the attention process [ 68 ]. The ultimate functions completed within cortical regions are clearly important. But dysfunction in the limbic and hypothalamic regions seems intuitive within this population. They can be measured in some instances (particularly neuroendocrine), and they are consistent with the underlying biologic bases of temperament variation as posited by Kagan and others. Temperament as a Component for Assessment of Learning in Youth with MM Annie's tale is not unusual for our children with MM and shunted HC. Clinicians and researchers alike have experienced stories of lesser or more severe impairments. For the parents and teachers working daily with children such as Annie, an understanding of the normal process of neuro-development leads to clearer understanding and a better differentiation from the abnormal. The research and subsequent writings of T. Berry Brazelton have provided ready information to parents of typically developing children by blending the Gesellian milestones of development with the temperament concepts as outlined by Carey, Chess, and Thomas [ 69 ]. The interplay of cognition, temperament, language, executive function, given the underlying neuro-endocrine/central molecular dynamics that accompany MM, offers a greater challenge to clinicians, teachers, and parents. For the classroom, identification of stressors and methods to minimize autonomic over-response can be as critical to successful learning as the understanding of memory strengths, or language problems. As in our story, Annie demonstrated persistence to task, willingness to approach novel situations, positive mood, and the ability to adapt to the testing situation – all temperament domains with which she has struggled. The "goodness of fit" between Annie and the examiner was critical before any progress could be made in better understanding her relative academic and learning strengths. The autonomic dysfunction, manifested in her excessive sweating, flushing, and urinary incontinence in school, only served to heighten both the emotional and physical symptoms. The differences in temperament profiles among children with MM are unlike those in other diagnostic cohorts of children with special health care needs [ 70 , 71 ]. Hughes et al. [ 72 ] have described temperament patterns in preterm infants over the first year of life. Descriptors of temperament characteristics differed from that seen in typically developing full term infants and from that of our children with MM. Because temperament is believed to be influenced by heredity, biology, and experience, parenting must also be considered as a variable (moderator) in temperament development. This concept of "goodness-of-fit" plays an important role in the parents' development, their perception of the infant, and the ultimate quality of the parent-child interactions [ 72 ]. Learning, for the child with MM, certainly takes place beyond the confines of preschool and classroom. Negotiating physicians, curious strangers, medical emergencies, educational diagnosticians, environmental barriers, and other stressors is a continual process. Raising a happy, self-reliant, and resilient child is the dream of parents. Perrin [ 73 ] has suggested that "adjustment" might be conceptualized as 1] the presence of fewer behavioral problems, 2] inter-personal functioning or competence, and 3] development of a positive self-concept. Parental perceptions of their child's actions and reactions contribute greatly to the moderation of, or exacerbation of, affective differences. It is incumbent on the professional to accurately discriminate between temperament characteristics that are barriers to inter-personal optimal functioning and truly psychiatric disorders. Given the neuro-anatomical and neuro-endocrine risks associated with MM, heightened awareness and assessment of temperament in this discriminating process is important. Conclusion T.S. Eliot wrote, We shall not cease from exploration And the end of all our exploring Will be to arrive where we started And know the place for the first time . As we continue, on behalf of children and youth with MM, to explore and better understand the cognitive and learning differences within the group, the developmental variations related to underlying central nervous system structures may continue to hamper the human tendency to categorize. The modal profile may not be singular. As assessments of strengths and differences continue; as descriptors of quality of life are derived; as personality variances are accounted for; as curricula and "home programs" are constructed, the bio-psycho-social model increasingly becomes the clear route for professional and family partnership. For the researcher trying to join un-connected dots into a cohesive whole, we, like Eliot, may need to return to the early components long-standing in the evolution of the brain and its function. The ancient limbic and hypothalamic systems leave indelible tell-tale hints that should not be left behind in our quest for understanding and raising the resilient child. List of abbreviations MM, myelomeningocele; HC, hydrocephalus Competing interests The author(s) declare that they have no competing interests. Authors' contributions Both authors contributed equally to this work. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC546401.xml |
552312 | Integrating quantitative and qualitative methodologies for the assessment of health care systems: emergency medicine in post-conflict Serbia | Background Due to the complexity of health system reform in the post-conflict, post-disaster, and development settings, attempts to restructure health services are fraught with pitfalls that are often unanticipated because of inadequate preliminary assessments. Our proposed Integrated Multimodal Assessment – combining quantitative and qualitative methodologies – may provide a more robust mechanism for identifying programmatic priorities and critical barriers for appropriate and sustainable health system interventions. The purpose of this study is to describe this novel multimodal assessment using emergency medicine in post-conflict Serbia as a model. Methods Integrated quantitative and qualitative methodologies – system characterization and observation, focus group discussions, free-response questionnaires, and by-person factor analysis – were used to identify needs, problems, and potential barriers to the development of emergency medicine in Serbia. Participants included emergency and pre-hospital personnel from all emergency medical institutions in Belgrade. Results Demographic data indicate a loosely ordered network of part-time emergency departments supported by 24-hour pre-hospital services and an academic emergency center. Focus groups and questionnaires reveal significant impediments to delivery of care and suggest development priorities. By-person factor analysis subsequently divides respondents into distinctive attitudinal types, compares participant opinions, and identifies programmatic priorities. Conclusions By combining quantitative and qualitative methodologies, our Integrated Multimodal Assessment identified critical needs and barriers to emergency medicine development in Serbia and may serve as a model for future health system assessments in post-conflict, post-disaster, and development settings. | Background In June 1991, a series of civil wars began as the Socialist Federal Republic of Yugoslavia dissolved violently into the four independent republics of Slovenia, Croatia, Bosnia-Herzegovina, the Former Yugoslav Republic of Macedonia, and the semi-autonomous nations of Serbia and Montenegro. Kosovo remains a province of Serbia but under the administration of the international community. The events of the last decade in the Balkans have significantly impacted the health care system of Serbia. The nation and its health care system were devastated by international sanctions, soaring unemployment, political instability, near economic collapse, and the North Atlantic Treaty Organization (NATO) air campaign [ 1 ]. Consequently, the health care budget was cut dramatically, further limiting the system's capabilities of providing adequate health care. These difficult years resulted in a substantial fall in major health indices and left a crippled health system struggling to provide for the needs of its citizens [ 2 ]. A country's emergency medical services are a particularly important component of the health care system for they provide an essential front-line resource for trauma, medical emergencies, and those without other access to health care. Unfortunately, the system of emergency services in Serbia did not escape the devastation of the 1990s and continues to suffer from problems common throughout the health care system. Limited health care funding has resulted in a lack of necessary equipment, supplies, medications, personnel wages, and an economic infrastructure unable to support necessary health care reforms in the public and private sectors. Furthermore, the country has received little external support for health system reform, underscoring the need for focused assessments to best determine priorities for health care development. There is a desperate need for reorganizing and restructuring emergency medical services throughout Serbia. To better understand the problems, needs, and obstacles to development, a team of investigators from Johns Hopkins University Schools of Medicine and Public Health collaborated with the Belgrade University Emergency Center to perform a multimodal assessment of emergency medicine in Serbia. Our goal in this study was to identify the most pressing needs and achievable goals, as perceived by emergency health care personnel, in order to ensure that development programs and available funds are appropriately directed [ 3 ]. Evaluating health care systems through the use of demographic data collection, surveys, or focus groups alone – as is often done in health system assessments – cannot adequately elucidate the intricacies of a country's health care system. In order to fully appreciate the complex dynamics of the health care system, it is necessary to perform a complete initial assessment; include strong local participation; identify and address significant barriers to change; and identify unique local needs and other cultural dimensions. To do this effectively, it is important to avoid simply introducing a "carbon copy" of a Western health care system that would further burden the system with inappropriate and unsustainable programs. We propose for the rapid assessment of health care services an Integrated Multimodal Assessment that uniquely combines diverse methodologies, whose individual effectiveness has been well described in the literature. Methods This study used a multimodal approach to assess the strengths, needs, problems, and obstacles related to the development of emergency medical services in Belgrade. The four study modalities were 1) demographic information and observational data; 2) focus group discussions with 68 emergency medicine personnel; 3) individual, free-response questionnaires with 39 emergency medicine personnel; and 4) by-person factor analysis (or Q-methodology) of the attitudes and opinions of 36 emergency medicine personnel regarding the development of emergency medical services. Participants in the study were a non-probabilistic sampling of emergency medicine physicians, nurses, and department directors from all pre-hospital and hospital-based emergency medicine institutions in Belgrade. Many participants took part in more than one study modality. Data collection occurred during July-August 2002. The authors received institutional review board approval and the written informed consent of all participants. Characterization of emergency medical services The capacity of emergency medical services in Serbia to provide for the needs of the country's citizens was characterized through health system data collection. Health system information was obtained through meetings with key officials in established health surveillance institutions such as the Emergency Center of the Clinical Center of Serbia; World Health Organization Belgrade office; Public Health Institute of Belgrade; Republic of Serbia Ministry of Health; European Agency for Reconstruction; Belgrade University, Institute of Social Medicine, Statistics and Health Research; as well as several non-governmental organizations with offices in Belgrade. Key officials were asked a series of quantitative, closed-response questions about the Serbian health care system and emergency medical services. Developed from a pilot health care assessment in Serbia the previous year, the questions focused on health system organization; patient population size and demographics; health care personnel; medical education and training; materials, supplies, and medications; and provider aptitude and morale [ 4 ]. Focus group discussions The two-fold purpose of the focus group discussions was 1) to better understand the needs, problems, and obstacles to system development as perceived by those involved directly in providing emergency medical services and 2) to collect a wide variety of opinions and attitudes for subsequent by-person factor analysis. The focus group discussions were held with 68 emergency medicine physicians, nurses, and directors through a non-probabilistic sample taken from the institute for pre-hospital emergency services, the Emergency Center of the Clinical Center of Serbia, and four of Belgrade's six community emergency departments. The discussions were conducted by the primary author in the Serbian language. The audio recordings of the 30- to 60-minute discussions were transcribed and translated into English by a native-English speaker fluent in Serbian (BDN) and a native-Serbian speaker fluent in English (MŠ). The translation was later verified by additional native-Serbian speaking co-authors fluent in English (MV, LM). The investigative field team (BDN, KD, MŠ, MV, LM) summarized the final transcripts and extracted common themes and divisive statements for by-person factor analysis. Emergency medical provider free-response questionnaire Free-response questionnaires enabled assessment of emergency health care needs and priorities for change as perceived by those most directly involved in patient care: the emergency health care providers. A non-probabilistic sample of 39 participants was obtained by open invitation, without compensation, through the director's office of each institution. As with the focus group discussions, the institute for pre-hospital emergency services, the Emergency Center of the Clinical Center of Serbia, and four of the city's six community emergency departments were represented. All subjects were administered a written Serbian-language questionnaire comprised of 10 free-response items (Table 1 ). The questions were developed through discussion with emergency medicine leaders at the University of Belgrade as well as through adaptation of earlier health care assessments conducted by the authors in the region. Each questionnaire was translated from Serbian to English by two bilingual translators (BDN, MŠ) and verified by additional authors (MV, LM) to ensure accuracy. Translated responses were coded and assessed for content by a researcher (KD) blinded to the demographics and specialty training of the respondents. More than one response was coded for each subject when necessary. Duplicate answers were only coded once. Illegible, blank, and off-subject answers were coded as missing data. Data were analyzed using SPSS 11.0 for Windows. Table 1 Free-response questionnaire questions. Investigators developed and administered to participants a Serbian-language questionnaire containing the following open-ended questions. Free-response questionnaire questions • What functions best in emergency medicine in your country and why? • How is the status of physicians in emergency medicine in comparison to that of physicians in other specialties? • With regard to education and training, what are the obstacles to becoming a physician in the field of emergency medicine? • What are the current problems in the system of emergency medicine? And what is being done to solve these problems? • What are the priorities for improving the system of emergency medicine? • What are the barriers to the future development of the system of emergency medicine? • What types of training should be offered for the improvement of the system of emergency medicine? • What are the strong and weak points of the training of physicians in the system of emergency medicine? • What do you think would be the reaction of the health care community (physicians, nurses, administrators) to development of the system of emergency medicine? • What do you think would be the response of the public and government to development of the system of emergency medicine? By-person factor analysis By-person factor analysis, or Q-methodology, avoids many of the limitations of other modalities by allowing the grouping of participants based on their subjective responses to an issue while preventing investigator preconceptions from influencing the grouping structure. In a letter to Nature in 1935, physicist and psychologist William Stephenson introduced by-person factor analysis as a means for the scientific study of subjectivity [ 5 ]. The methodology uses a unique combination of qualitative and quantitative methods to subdivide a study population, evaluate the degree of consensus among the participants, and identify any discordant opinions. The process begins by creating a concourse, or collection, of opinions and perceptions toward the subject of interest – in this case, emergency medicine in Serbia. From this concourse, the investigators develop statements representing the spectrum of opinion and request that respondents assign each statement to a position within a quasi-normal grid distribution as to whether they completely disagree with, feel indifferently or ambivalently towards, or completely agree with the statement [ 6 , 7 ]. In our present study, a concourse was developed through the aforementioned focus groups and free-response questionnaires. Directly from this concourse, the field investigators (BDN, KD, MŠ, MV, LM) produced a Q-sample of 23 statements (Table 2 ), which was administered to 36 Belgrade emergency medicine physicians, nurses, and department directors. The participants were attained by open invitation through the director's office of each institution. Collected demographic information of participants included age, gender, and specialty training. Respondents were asked to sort the Q-sample using the condition of instruction, "Please sort these statements with respect to your opinion of the current system of emergency medicine in Serbia". Sorting involved the forced ranking of all statements within the grid of quasi-normal distribution. Statements with which the participant most strongly disagreed were placed toward one end of the grid (-3 agreement score). Toward the grid's other extreme (+3 agreement score) were placed statements with which the participant most strongly agreed. The participant placed statements that evoked ambivalence or neutrality near the center (0 agreement score) of the quasi-normal distribution. Factor analysis was then performed on the responses using PQMethod 2.10 [ 8 ] followed by manual factor rotation. This analysis leads to the subdivision of the study population into distinct "respondent types" (i.e. factors), which are clusters of participants grouped by their common opinions and perceptions. To define and characterize the resultant respondent types, all participants who loaded heavily on a respondent type (>60% concordance) were selected as respondent type "loaders". The participants who loaded heavily and specifically for a single respondent type (i.e. >60% concordance on the type of interest and <30% concordance on remaining types) were designated respondent type "definers" and closely reviewed to assist in characterizing each respondent type. Table 2 Responses of emergency medicine physicians and administrators to Q-statements regarding emergency medicine in Serbia. Statements are listed from greatest to least participant consensus. With each statement, an averaged agreement score is calculated for all participants and for each identified respondent type. Scores represent the spectrum of participant agreement/disagreement (i.e. "strongly disagree" (-3), "ambivalent/neutral" (0), or "strongly agree" (+3) with the statement). To aid discussion of respondent types, summary labels ("Utilize", "Develop", and "Invest") characterize unique qualities of each respondent type. "Utilize" respondent type is most concerned with the poor utilization of emergency services. "Develop" respondent type advocates the further development of emergency medicine. "Invest" respondent type emphasizes the need for greater investment in emergency medicine. Q-statements (listed in order of greatest consensus to least consensus) Averaged participant agreement score Agreement scores of identified respondent types "Utilize" (5 loaders, 4 definers) "Develop" (6 loaders, 3 definers) "Invest" (4 loaders, 2 definers) It is NOT necessary for patients to be seen by physicians in the field, but rather patients should be brought immediately to an emergency department for care. -2.0 -2 -1 -3 Health management training for health care leaders is essential for the improvement of emergency medical services. -0.1 -1 -1 -2 The public overuses ambulance services because there is no charge for the use of these services. 0.6 1 0 1 The first priority for the development of emergency medicine should be to improve the organization of emergency services. 0.4 2 1 0 Patients arriving to the emergency facility should be taken, according to their illness, directly to a specific specialty department. -0.4 -1 -1 -1 Protocols should be developed to standardize the treatment of patients throughout Serbia. 1.2 0 3 -1 Emergency medicine should be taught as a required course during the last year of medical school. 0.8 -2 2 0 Emergency medicine in Serbia would function better if it were financed by the federal budget rather than by the social health insurance fund. 0.0 1 0 -1 Emergency medicine should be a separate specialty in which physicians are trained to exclusively practice emergency medicine. 0.0 1 2 -2 Primary health care providers in the health houses are sufficiently trained in the triage of emergent and non-emergent patients. -1.8 -1 -2 -3 All institutions that provide emergency medical services should be open 24 hours a day. 1.4 0 2 1 A priority in the development of emergency medicine is to increase the number of appropriately equipped ambulances. -0.6 -2 -3 1 There should be national guidelines to determine which illnesses/injuries should be treated at each type of health care facility. 0.2 2 0 1 Much of the burden on emergency health care providers is due to the time spent caring for non-emergent patients. 1.3 3 0 0 Continuing medical education should be required by law of all physicians working in emergency medicine. 1.0 0 1 2 There is poor coordination among the various specialties that provide emergency medical services. -0.3 -1 1 -1 The public should be better educated about the level of care that each health care institution provides in order to properly use the available health care services. 0.3 3 0 0 The medical school is currently playing a sufficient role in the development of emergency medicine in Serbia. -1.6 -3 -1 -1 There is poor cooperation between the emergency centers, clinical-hospital centers, pre-hospital emergency services, and health houses. 0.6 1 1 3 Physicians working in emergency medicine in Serbia need greater expertise and technical skills to provide an appropriate level of care. 0.8 -2 3 3 The problems in emergency medicine would be solved if there were money and equipment with which to work. 0.4 2 -2 2 There is an appropriate balance of theoretical and practical training for physicians in emergency medicine. -2.0 -3 -3 -2 Radio communication does not function effectively between the ambulances and the medical institutions. -0.1 0 -2 2 Results Characterization of emergency medical services Emergency medicine is not a recognized health care specialty in Serbia. In smaller communities, emergency services are provided by primary health stations ( zdravstvene stanice ) and primary health centers ( domovi zdravlja ). In the capital city of Belgrade, there exists a network of 6 part-time, hospital-based ( kliničko-bolnički centri ) emergency departments that are supported by around-the-clock pre-hospital emergency services ( hitna pomoć ) and the academic Emergency Center ( Urgentni centar ) at the Clinical Center of Serbia. Like many European countries, Serbia has adopted a largely Franco-German model of emergency medicine in which pre-hospital emergency services are provided in the field by physician-staffed ambulances [ 9 ]. Care is subsequently provided in-hospital by physicians of multiple medical specialties. Patients access emergency services through referral from Belgrade's 16 primary health centers, telephoning "94" for pre-hospital emergency services, or by personal referral to the emergency institutions. The academic Emergency Center at the Clinical Center of Serbia receives patients directly as well as by referral from community emergency departments that are closed or overwhelmed. During 2001, the Emergency Center received 130,877 patients. Focus group discussions Through focus group discussions, the investigators were able to determine the opinions and perceptions of emergency service providers and administrators on the current emergency medical system in Serbia. Sixty-eight physicians participated in one of six focus group discussions (female: 39.7%; average work experience: 19.3 years; specialties: surgery (34%), internal medicine (29%), anesthesia (12%), general medicine (7%), emergency medicine (3%), other (15%)). Discussions were held at the institute for pre-hospital emergency services, the Emergency Center, and four hospital emergency departments. The participants' comments from the focus group discussions are collectively summarized in Table 3 . Most providers believe the lack of sufficient funding for emergency medical services is one of the greatest problems affecting emergency medicine in Serbia. Financial support for medications, medical supplies, modern equipment, employee salaries, and facility maintenance is extremely limited and restricts the capabilities of health care providers. Poor organization of emergency medical services, including the lack of government regulation, absence of uniform treatment protocols, improper system management, and poor triaging and routing of patients between facilities, also dominated the focus group discussions. Table 3 Foremost problems of emergency medicine in Serbia and priorities for system development. A summary of the findings from focus group discussions with providers of emergency medical services. Results of focus group discussions with providers and administrators of emergency medical services CITED PROBLEMS IN THE SYSTEM OF EMERGENCY MEDICINE IN SERBIA FINANCE: ≺ Inadequate financial resources for essential equipment, supplies, and medications ≺ Discouraged emergency medical service personnel due to meager salaries, difficult work conditions, and large workloads ≺ Inadequate number of properly equipped ambulances and functioning radio equipment ≺ Very few computers and no health information systems to track patient health records ORGANIZATION: ≺ Need for federal regulation of emergency medical services ≺ Lack of sufficient protocols for the standardization of triage and treatment ≺ Inadequate coordination between the institutions providing emergency medical services ≺ Need for further development of emergency medicine as its own specialty EDUCATION: ≺ Inadequate training in emergency medicine during medical school ≺ Insufficient practical training of emergency medical service providers ≺ Few opportunities for professional development and continuing education of emergency service providers ≺ Limited access to medical innovations through the internet, foreign professional journals, conferences, courses, and seminars ≺ Lack of health management training for leaders of health care institutions ≺ Need for public education about the emergency medical services system and how to properly utilize them SUGGESTED PRIORITIES FOR THE DEVELOPMENT OF EMERGENCY MEDICINE IN SERBIA FINANCE: ≺ Secure funding for essential medications, supplies, equipment, employee salaries, and maintenance of health care facilities ≺ Consider long-term sources of continuous funding for emergency services such as the government budget instead of the social health insurance fund ORGANIZATION: ≺ Develop national protocols for the standardization of emergency triage and treatment ≺ Further develop emergency medicine as its own specialty ≺ Clearly define the responsibilities and emergency services of physicians in each health care facility ≺ Institute a system to promote better coordination between the primary health centers, the hospital emergency departments, and the Emergency Center ≺ Implement quality control measures for the delivery of emergency medical services ≺ Establish a health information system to facilitate the tracking of patients EDUCATION: ≺ Introduce required continuing medical education supported by legislation that would provide health care professionals leave from work to attend this periodic training ≺ Provide health care professionals with access to continuing medical education through the internet, professional journals, conferences, seminars, and practical training ≺ Develop a fellowship program for emergency medicine physicians ≺ Increase the level of practical emergency medical experience provided in medical school and postgraduate training ≺ Implement training in BLS, ALS, and emergency triage for all health care providers ≺ Institute a health management training courses for leaders of health care institutions ≺ Educate the public regarding the level of emergent care that each health care institution provides and how to properly utilize the available health care services In addition, physicians and medical directors report that their emergency medical system is in significant need of improved education and training programs. Suggested improvements include further training in emergency medicine during medical school, bedside training to develop practical skills, development of standardized treatment protocols, and access to continuing professional education. Emergency medical provider free-response questionnaire Questionnaires were completed by 39 physicians at the institute for pre-hospital emergency services, the Emergency Center, and four hospital emergency departments (female: 41.0%; average work experience: 20.8 years; specialties: surgery (31%), internal medicine (31%), anesthesia (13%), general medicine (10%), emergency medicine (5%), other (10%)). Respondents show several areas of high agreement – particularly remarkable considering the questionnaire's free-response format. The most frequently reported strength of Serbia's current health care system are the health care workers (69%). Worker enthusiasm (33%) and the collaboration of various medical specialties providing emergency services (21%) are most frequently cited as the greatest positive attributes of emergency service providers. In addition, 46% of respondents report that the various emergency medical service providers (especially the Emergency Center and the pre-hospital emergency services) are other strengths of emergency medical services in Serbia. Thirty percent of the participants surveyed believe that lack of incentives to enter the specialty, difficulty of the work, and poor financial compensation are important barriers to becoming a physician in emergency medicine in Serbia. The poor reputation of physicians in emergency medicine relative to other physicians (reported by 46% of participants) may contribute to lower interest in specializing in emergency medicine. Poor organization (26%) and insufficient opportunities for professional development (15%) were also reported as major impediments to being an emergency service provider. Participants were asked what they believed were major problems of emergency medical services in Serbia. Inadequate finances, medications, medical supplies, and modern medical equipment were cited by 54% of the emergency medicine personnel as being a major problem (Table 4 ). Of these respondents, 62% feel that they have sufficient technical skills but that insufficient medical equipment significantly restricts their capabilities as health care providers. Another 24% believe that the absence of adequate and continuous funding for emergency medicine is a critical problem. With improved funding they could obtain the necessary supplies, equipment, and medication they need, as well as improve employee salaries and facility maintenance. As a way to ensure a more reliable source of funding, 78% of the emergency medicine personnel surveyed emphasized the need for government financing of emergency medical services through the national budget rather than through the unpredictable social health insurance fund. Table 4 Quantitative results of emergency medical provider questionnaires. Positive responses were calculated as a percentage of the number of providers that included the statement in their free response. More than one response was coded per subject when applicable. Italicized responses denote breakdown of individuals with the above response. Select questionnaire responses of emergency medical providers (n = 39) Positive responses (%) CURRENT PROBLEMS IN EMERGENCY MEDICINE Organization 59 An organized system doesn't exist 35 No government regulation 22 Lack of treatment/triage protocols 22 Poor coordination between health care facilities / lack of team work 17 Poor division of labor of those providing emergency services 17 Other 26 Lack of supplies, equipment, and medications 54 Insufficient equipment 62 Inadequate funding 24 Lack of optimal ambulances 5 Poor diagnostic & therapeutic procedures 5 Training and education 36 Lack of incentives / difficult field of work / poor compensation 23 No answer / there is no system of emergency medicine 18 Pre-hospital emergency services / ambulance services 5 PRIORITIES FOR REFORM Organization 77 Treatment protocols/guidelines 27 Improve coordination between health care facilities (team work) and within hospitals (between specialties) 27 Increase the number of physicians (personnel) trained in EM 17 Government regulation and organization 13 Legal regulation 10 Increase the number of beds 10 Establishment of EM as a separate specialty 7 Develop a computer database for tracking patients 3 Improve efficiency of system 3 Supplies, equipment, and medication (improved diagnostics) 54 Training and education 36 Financing 33 Incentives / work conditions / compensation 21 BARRIERS TO FUTURE DEVELOPMENT Economics/resources 69 Organization 36 Political/government 26 Inadequate education and training 18 Reorganization of emergency medical services is a priority for development according to 77% of participants. Of these responses, poor coordination and team work between health care facilities (27%), the absence of treatment protocols (27%), insufficient numbers of emergency medical physicians (17%), and the lack of government regulation (13%) were the most frequently cited priorities for improving the organization of emergency medical services. Other concerns include lack of quality control measures, burden of caring for non-emergent cases, inadequate admission and triage areas, and insufficient hours of operation of hospital-based emergency departments. Many emergency medicine personnel are also critical of the education and training of physicians in emergency medicine. Although 58% of those surveyed feel that physicians receive good theoretical training, 77% report that their educational system is in significant need of improved medical school and postgraduate training programs. Of the individuals advocating for improved education and training, 27% feel that further training in emergency medicine should be included in the medical school curriculum, and 53% believe that additional bedside training to develop practical skills is imperative. Furthermore, 56% of participants in this study express a need for better access to continuing professional education, including shared practical experiences and training between health care institutions. A large number of emergency medicine personnel (36%) stress the need for greater access to international journals, conferences, seminars, and the internet. Many (15%) also communicate a need for international collaboration and training programs between Serbia and foreign medical institutions. Survey participants cite numerous obstacles to the development of emergency medical services in Serbia, including insufficient funding and resources (69%), poor organization of emergency services (36%), lack of governmental support (26%), and inadequate education and training (18%). Also vital to development efforts are the perceptions and attitudes of the people impacted by and involved in these efforts – the public, health care community, and government officials. A slight majority (62%) of individuals surveyed believe that the health care community will support further development of emergency medical services. However, only 46% and 41% of participants feel that the public and government, respectively, will respond positively to development. By-person factor analysis Of the 36 emergency medicine personnel invited to participate in factor analysis, 33 individuals (91.7%) completed the exercise correctly (female: 45.4%; average work experience: 16.8 years; specialties: surgery (21%), internal medicine (27%), anesthesia (27%), general medicine (6%), emergency medicine (6%), other (12%)). The institute for pre-hospital emergency services, the Emergency Center, and two hospital emergency departments participated in this component of the study. Factor analysis followed by manual factor rotation determined levels of agreement among the 33 participants and identified three unique types of respondents. Table 2 displays the averaged level of agreement and the level of agreement for each respondent type (the spectrum of agreement includes "I strongly disagree" (-3), "I feel ambivalent/neutral" (0), or "I strongly agree" (+3)). With statements written in the form of agreement, a majority of participants believe patients should be seen by physicians in the field (+2.0), protocols should be developed to standardize the treatment of patients throughout Serbia (+1.2), primary health care providers are not sufficiently trained in the triage of emergent and non-emergent patients (+1.8), emergency medical institutions should be open 24 hours a day (+1.4), much of the burden on emergency health care providers is due to time spent caring for non-emergent patients (+1.3), continuing medical education should be required by law of all physicians working in emergency medicine (+1.0), the medical school is not playing a significant role in developing emergency medicine in Serbia (+1.6), and there is not an appropriate balance of theoretical and practical training for physicians in emergency medicine (+2.0). In addition to assessing levels of agreement among participants, by-person factor analysis allowed us to identify three different respondent types, each involving multiple respondents with diverse demographics but who share common and unique perspectives relative to the remaining participants. To aid discussion of the three respondent types, each was assigned a brief label: "Utilize", "Develop", and "Invest". Respondent type "Utilize" was heavily loaded by a total of 5 respondents (15.2% of participants) and defined by 4 (12.1%). The principal concerns of the "Utilize" respondent type include the poor utilization of emergency medical services. These individuals believe that emergency personnel are well trained, educated, and prepared to address the emergent needs of the community. However, they consider their role ill-defined and misunderstood, leading to the ineffective use of their services by the public. Respondent type "Develop" was characterized by 6 loaders (18.2%) and 3 definers (9.1%). These individuals support the further development of emergency medicine as an independent specialty in Serbia. They do not consider inadequate finances, supplies, or equipment to be the chief concern of the system of emergency medicine. Instead, the "Develop" respondent type encourages the development of standardized treatment guidelines, improved coordination between the specialties providing emergency care, and additional education and training in the specialty of emergency medicine. The third respondent type, "Invest", included 4 loaders (12.1%) and 2 definers (6.1%). These individuals consider the system of emergency medicine appropriately organized but lacking adequate investment. They believe greater resources should be applied towards equipment and professional training and development. They are also concerned with an apparent lack of coordination between institutions providing primary health care and emergency services. Additional respondent types were defined by single respondents, which prevented adequate characterization. However, the demographic data on each of these isolated respondent types were closely examined to assure that the individual was not in a unique position of authority that could disproportionately influence future health system reform. The specific responses of respondent types "Utilize", "Develop", and "Invest" to the 23 Q-statements are listed in Table 2 . Discussion The purpose of a rapid assessment is to provide organizations with timely and reliable information to aid targeted interventions. Many rapid assessment methodologies have been utilized, each with its own strengths and weaknesses. One of the most commonly used methodologies is the KAP (Knowledge, Attitudes, and Practices) survey, which is traditionally a standard questionnaire with pre-developed, closed-response questions [ 10 ]. Because of its ease in study development, administration, and analysis, a KAP survey can provide very rapid results for truly emergent situations. However, KAP surveys are limited in flexibility, community involvement, and internal validity. An alternative assessment method, Rapid Assessment Procedures (RAP), successfully addresses some of KAP's limitations through an ethnographic, participatory problem-solving process [ 11 ]. Nevertheless, while these methods may provide limited quantitative information for use in health sector assessment, they can lack sufficient detail to identify major barriers to system improvements. The use of more detailed assessment tools – employing a combination of qualitative and quantitative methodologies – can more accurately characterize the country's health care needs and the significant political and cultural barriers to sustainable health care reform. Our proposed Integrated Multimodal Assessment utilizes a diverse approach for strong internal validity and by-person factor analysis to uniquely identify critical subpopulations. By understanding the concerns of subpopulations, more targeted interventions can be developed to directly address these concerns. A trial study evaluating primary health care in Serbia has demonstrated the advantages of using a combination of qualitative and quantitative tools for health system assessment [ 12 ]. Each of the integrated methodologies contributes its own strengths to the overall assessment, complementing shortcomings of the other methodologies. For example, while the questionnaires provide concise, quantifiable, and less equivocal responses, the focus group discussions allow investigators to seek clarification and have the unique benefit of participant interaction often leading to entirely novel ideas. Out of the near collapse of the Serbian health care system comes the opportunity to establish a health system more effective than ever in meeting the needs of its citizens. Our focus group discussions, questionnaires, individual surveys, and by-person factor analysis reveal that most emergency medical physicians believe the greatest problems in their system are poor organization of emergency medical services; lack of essential funding, medical supplies, medication, and technical equipment; and inadequate education and training. Several respondents also share the idea that poor incentives for specializing in emergency medicine, difficulties of emergency service work, and inadequate compensation are significant barriers to advancing the field of emergency medicine in Serbia. Although many of these are problems ubiquitous to developing health systems, this study illuminates specific opportunities for emergency medicine providers, the Serbian government, and international institutions to work together to address these issues. According to the majority of emergency medicine providers and administrators surveyed, of foremost priority is reorganization of emergency medical services. In order to begin addressing this need, several participants propose the development of triage and treatment protocols. To facilitate improved routing of patients within the hospitals and to the appropriate emergency health care facility, all health care providers, including those within the primary health centers, should receive formal training in triage. Several participants suggest that leadership development and management training should be provided to medical directors and leading health care providers to improve the organization within and between emergency health care centers. Public education programs should also be developed to increase awareness of the level of care that each health care institution provides so that the public can more properly utilize available health care services. Study participants further reported that the attainment of necessary medications, supplies, and equipment should be a top priority in the development of emergency services. The professionals working in the health care system of Serbia are severely limited by a shortage of these resources and, therefore, are unable to apply their knowledge and skills for proper diagnosis and treatment. Obsolete medical equipment further constrain their diagnostic capabilities, and lack of basic medications and supplies impedes their treatment strategies. This multimodal assessment also shows that the education and training of physicians in emergency medicine need further development. Elements of the educational system needing improvement include medical school and postgraduate bedside training programs; training between health care institutions; access to continuing professional education materials including international journals, conferences, seminars, and internet access; and international collaboration between Serbia and foreign medical institutions, including opportunities for Serbian physicians to gain international clinical experience. Several participants also suggest that all health care workers, including physicians working in the primary health centers, receive periodic training in Basic Life Support (BLS), Advanced Life Support (ALS), and emergency triage. Further development of emergency medicine in Serbia will be complex and will face numerous challenges. Reform will require funding, political commitment, supportive legislation, a revised medical curriculum, multi-phasic implementation, and post-interventional evaluation. Respondents believe that the funding needed for development of emergency services represents a significant barrier to change, but that, however, it is not the universal solution. Although a majority of the emergency medicine personnel surveyed feel that the health care community will support further development of emergency services, just under half of the participants expect similar support from the public and government. Additional studies specifically assessing the needs and concerns of the public and government are recommended. The use of multiple assessment modalities and the resulting complexity of this study involve several limitations that require consideration. All participants in this study were invited to participate by the director's office at each hospital or clinic. The study participants were aware that their participation was entirely voluntary and that they would not receive compensation. Although the study did not provide an economic incentive for participation, it did provide the opportunity for those providers with strong opinions to openly voice their concerns and suggestions. As a result, the study population may not represent all emergency health care providers. While every emergency medicine institution in Belgrade participated, this study was limited to the pre-hospital and hospital-based institutions within the capital city (covering 14.9% of the population of Serbia (excluding Kosovo)). Nevertheless, the participants represent the breadth of medical specialties in Serbia, and they constitute a large number of the stakeholders involved in further development of emergency medicine. Conclusions Despite the challenges, development of emergency services in Serbia can be accomplished through the dedication and commitment of providers. As a result of studies on emergency services in Kosovo, developmental programs similar to those suggested here have been successfully implemented to improve the training in and provision of Kosovo's emergency medical services [ 13 ]. Visiting experts currently travel to the province to provide didactic and bed-side training to physicians and other health care providers on the latest advances in emergency medicine, internal medicine, pediatrics, and several other fields of medicine. Similarly, a one-year fellowship program in emergency medicine and a leadership training program have been established at the University of Prishtina. It is the hope of the authors that similar reforms will also be implemented to address the specific situation in the Republic of Serbia and that this multimodal assessment will assist by identifying the critical needs, barriers, and priorities for sustainable development. By utilizing both quantitative and qualitative methodologies, the authors submit that this Integrated Multimodal Assessment tool offers a more robust alternative to standard surveys, KAP studies, or the use of anecdotal information for quickly identifying priorities in health system reconstruction in the post-conflict, post-disaster, or development setting. List of abbreviations KAP: knowledge, attitudes, and practices RAP: rapid assessment procedures Competing interests The authors declare that they have no competing interests. Authors' contributions Each author has participated sufficiently in the work being reported to take public responsibility for the content. BDN, KD, and MVR conceived and designed the study and obtained research funding. MVR and MM supervised the completion of the study. BDN, KD, MS, MV, and LM undertook recruitment of participating centers and physicians. BDN, KD, and MS completed the data collection, statistical analysis, and first draft of the manuscript. All authors reviewed and contributed to the revision of the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC552312.xml |
516236 | A steady state analysis indicates that negative feedback regulation of PTP1B by Akt elicits bistability in insulin-stimulated GLUT4 translocation | Background The phenomenon of switch-like response to graded input signal is the theme involved in various signaling pathways in living systems. Positive feedback loops or double negative feedback loops embedded with nonlinearity exhibit these switch-like bistable responses. Such feedback regulations exist in insulin signaling pathway as well. Methods In the current manuscript, a steady state analysis of the metabolic insulin-signaling pathway is presented. The threshold concentration of insulin required for glucose transporter GLUT4 translocation was studied with variation in system parameters and component concentrations. The dose response curves of GLUT4 translocation at various concentration of insulin obtained by steady state analysis were quantified in-terms of half saturation constant. Results We show that, insulin-stimulated GLUT4 translocation can operate as a bistable switch, which ensures that GLUT4 settles between two discrete, but mutually exclusive stable steady states. The threshold concentration of insulin required for GLUT4 translocation changes with variation in system parameters and component concentrations, thus providing insights into possible pathological conditions. Conclusion A steady state analysis indicates that negative feedback regulation of phosphatase PTP1B by Akt elicits bistability in insulin-stimulated GLUT4 translocation. The threshold concentration of insulin required for GLUT4 translocation and the corresponding bistable response at different system parameters and component concentrations was compared with reported experimental observations on specific defects in regulation of the system. | Background In living systems, extracellular information is processed through signal transduction machinery to appropriately regulate cellular function. This information processing machinery is made up of a complex web of enzyme cascades, allosteric interactions and feedback loops. Depending on their regulatory design these signaling networks elicit diverse responses, but display many common operating principles. A recurring theme in signaling systems is switch-like responses to graded or transient input signal. Various mechanisms are known to generate such all-or-none responses [ 1 ]. Bistability is one such system level property, in which, the system switches between two discrete stable steady states without being able to rest in an intermediate state. Bistable systems exhibit hysteresis wherein, the value of input stimulus required for system transition from one state to another is quite different from the value required for reverse transition. Both computational and experimental analyses have shown that bistability plays a significant role in cellular differentiation and cell cycle progressions [ 2 - 5 ], production of biochemical memory [ 6 ], microbial metabolic systems [ 7 ], lateral signal propagation [ 8 ] and protein translocations [ 9 ]. Existence of bistability in cellular regulation has been attributed to nonlinearity embedded in positive feedback loop or double negative feedback loop [ 10 ]. Here, we present steady state simulation results of metabolic insulin signaling pathway comprising of positive feedback loops and show that this system can convert graded inputs into switch-like bistable output response. Insulin is the most potent anabolic peptide hormone known that elicits myriad biological responses by specifically binding to insulin receptor and simultaneously stimulating multiple signaling pathways to regulate growth, differentiation and metabolism. Insulin maintains glucose homeostasis by stimulating the uptake, utilization and storage of glucose in muscle and adipose tissue, and inhibits hepatic glucose production [ 11 ]. Defects in any of the pathway components lead to disturbance in growth, differentiation, and in the homeostasis of glucose and lipid levels. This leads to disease conditions such as type 2 diabetes, hypertension, obesity and a cluster of abnormalities characterized by insulin resistance or deficiency. In such a condition, normal circulating concentration of insulin is insufficient to elicit appropriate response [ 12 , 13 ]. Studies over the last century have identified the major insulin signaling components involved in the regulation of glucose uptake into cells and its various defects in diseased states. A wide family of glucose-transporter proteins localized in the plasma membrane, facilitate uptake of glucose from the blood into tissues. Among different isoforms, only glucose transporter isoform-4 (GLUT4) is specifically expressed to promote glucose uptake in insulin sensitive tissues, viz. muscle and adipose, and in response to insulin, GLUT4 gets translocated to the plasma membrane from intracellular vesicles [ 14 ]. The biological action of insulin is initiated by binding to the tyrosine kinase receptor and its subsequent activation. The activated tyrosine kinase receptor undergoes autophosphorylation and catalyzes the phosphorylation of several intracellular substrates including the insulin-receptor substrate (IRS) proteins (Fig. 1 ). The activated IRS isoform-1 protein further activates downstream components to elicit translocation of GLUT4 [ 11 ]. There are several downstream kinases like PI-3 kinase, Akt (or protein kinase B) and protein kinase C-ζ (PKC-ζ) demonstrated to be potentially capable of phosphorylating upstream proteins like IRS-1 and tyrosine phosphatase 1B (PTP1B) thus serving as negative and positive feedback loops respectively [ 15 ]. Other than feedback loops, crosstalk between multitudes of signal transduction pathways have also been reported, thus making the insulin-signaling pathway a highly intricate network [ 11 ]. Figure 1 Simplified representation of molecular mechanism involved in insulin signaling pathway that regulates glucose transporter (GLUT4) translocation to cell membrane. Some of the details like, other isoforms of insulin receptor substrate and multiphosphorylation of insulin receptor substrate are not shown here. Nomenclature: GLUT4: Glucose-transporter isoform 4; IRS-1: Insulin receptor substrate-1; PI3K: Phosphatidylinositol-3-kinase; PI (3, 4, 5) P3: Phosphatidylinositol (PI)-3, 4, 5-tiphosphate; PDK1: phosphosinsositide-dependent kinase 1; Akt: Protein kinase Akt or protein kinase B (PKB); PKC: Protein kinase C-ς; PTP1B: Protein tyrosine phosphatase 1B; PTEN: 3' lipid phosphatase; SHIP2: 5' lipid phosphatase; Detailed description of signaling events are given in the methods section. Letter 'P' indicates phosphorylated species. Although studies on various cell lines, transgenic and knock-out mice, have helped to uncover and characterize the different components involved in insulin signaling pathway, there are many voids in our understanding of the precise molecular mechanisms of signal transduction and cellular effects of insulin [ 16 , 17 ]. The major hurdles are complexity of insulin signaling pathway and technical problems like experimental methodology employed for system level quantification. For example, depending upon different techniques employed, quantification of GLUT4 translocation in response to insulin binding yielded different results in the same cell type [ 18 ]. Recent technical developments however have helped in studying the localization and translocation of signaling proteins and overall quantification of signaling processes in single cells has been possible [ 19 ]. In such a scenario, it is pertinent to ask questions regarding the design principles involved in intracellular regulation. For example, what does a particular regulatory structure accomplish and how does it help in exhibiting different physiological responses. Based on available experimental data, computational and mathematical analysis can answer some of these questions and possibly propose new experiments and hypotheses. Earlier mathematical modeling studies of insulin signaling pathways have focused on subsystems of the pathway, like insulin receptor binding kinetics [ 20 , 21 ], receptor recycling [ 22 ] and GLUT4 translocation [ 23 - 25 ]. Recently a comprehensive dynamic model of metabolic insulin signaling pathway was presented, which involved most of the known signaling components [ 26 ]. Although the model correlated well with the published experiment data, authors did not discuss the system level regulatory design of insulin signaling system. In the present work, we have developed a steady state model of insulin signaling to generate dose response curves for fractional translocation of GLUT4 to varying input insulin stimuli. One of the main objectives was to investigate the effect of inherent signaling structure made up of phosphorylation cycles, allosteric interactions and feedback loops on the system level response of insulin on GLUT4 translocation. Furthermore, we were interested in examining whether the regulatory design consisting of positive feedback loops in insulin signaling pathway exhibits bistable response. We solved the steady state equations for the entire metabolic insulin pathway including the positive feedback loops numerically, and found that GLUT4 gets translocated to the plasma membrane in an all-or-none manner in response to a varying concentration of input insulin stimuli. We show that GLUT4 translocation switches between the on-state and off-state and exhibits hysteresis in its response to increasing and decreasing input insulin concentration. This input-output relationship was then studied at various concentration of signaling components and system parameters in order to monitor the range over which this response persisted. We discuss these results by comparing with the known specific defects in regulation of the system (insulin dependent diseases) that lead to improper glucose uptake into the cell. Methods Figure 1 shows a simplified representation of molecular mechanisms involved in insulin signaling pathway. The metabolic insulin-signaling pathway used for the steady state simulation in the present work is shown in Fig. 2 . This schematic representation is a compilation of various interactions in insulin pathway which have been very well reviewed [ 11 - 27 ]. We have used the framework of Goldbeter and Koshland [ 28 ] to model the insulin system at steady state and accordingly an equivalent rate constant and Michaelis-Menten constant nomenclature scheme is applied. The detailed list of the steady state equations for covalent modification cycles, equilibrium relationships for allosteric interactions, mass balance equations for total species and parameters used in the simulations are provided in Appendix. All component enzyme concentrations are represented with respect to whole cell volume. Most of the kinetic/equilibrium constants are taken from the literature. In this analysis, the reactants like ATP and PPi concentrations are assumed to be constant. In the following paragraphs we present the system considered and assumptions made during the analysis. Figure 2 Schematic representation of metabolic Insulin signaling pathway used for the steady state analysis. Nomenclature: I, Insulin; X, unbound surface insulin receptor; XI, unphosphorylated once-bound surface receptor; XIP, phosphorylated once-bound surface receptor; XI 2 P, phosphorylated twice-bound surface receptor; X i represents intracellular receptor pool; XIP i and XI 2 P i are internalized form of XIP and XI 2 P; phosphatase PTP catalyzes the dephosphorylation of AP, XIP, XIP i and XI 2 P i . A, unphosphorylated IRS-1; AP, phosphorylated IRS-1; B, inactive PI3-kinase; APB, phosphorylated IRS-1 and PI3-kinase complex; CP3, lipid PI[3,4,5]P 3 ; CP2, lipid PI[4,5]P 2 ; CP2', lipid PI[3,4]P 2 ; phosphatase SHIP2 catalyzes dephosphorylation of CP3 to form CP2', phosphatase PTEN catalyzes dephosphorylation of CP3 to form CP2; F, inactive Akt and PKC-ς; FP, phosphorylated Akt and PKC-ς; E 8 dephosphorylates FP; E 6 phosphorylates CP2' to form CP3; FP activates GLUT4 from intracellular location to plasma membrane. G C and G M represent GLUT4 in cytoplasm and on plasma membrane respectively. Kd 1 to Kd 3 are dissociation constants; Kd 4 and Kd 5 are distribution coefficients; Km r , Km, Km 1 to Km 8 are Michaelis-Menten constants; k, k p , k d , k s , k 0 , k 1 to k 13 are reaction rates as shown in the figure. Insulin initiates its biological action by interacting with the insulin receptor, which belongs to a superfamily of tyrosine kinase receptors. On binding to the first insulin molecule, the receptor gets auto-phosphorylated and is dephosphorylated by phosphatase PTP1B [ 12 ]. The phosphorylated insulin receptor can either bind with another insulin molecule or undergoes dissociation. Binding of the second insulin molecule does not affect the phosphorylation state of the receptor. Here we have assumed that the concentration of unbound phosphorylated receptor is negligible. Thus, phosphorylated receptors can exist as species bound to either singly or doubly bound molecules of insulin. Insulin bound phosphorylated receptor rapidly gets internalized into the endosomal apparatus of the cell before it gets dephosphorylated by PTP1B and incorporated into intracellular receptor pool [ 29 ]. However recent studies indicate that, PTP1B might interact with insulin receptor directly and deactivate it without internalization [ 30 ]. We have assumed that, the membrane bound phosphorylated insulin-receptor and its internalized form, both get dephosphorylated by PTP1B. The rate equation for intracellular receptor at steady state is represented as where k p is rate constant and Km r is Michaelis-Menten constant for dephosphorylation of internalized insulin receptors XIP i and XI 2 P i . The term k d is first order degradation rate constant and k s is zero order synthesis rate constant of intracellular receptor X i . The receptor exocytosis and endocytosis are assumed to be at quasi-equilibrium because of their faster time scales than the synthesis and degradation of receptors [ 26 ]. The phosphorylated active receptors further catalyze phosphorylation of several intracellular substrates including the IRS proteins, GAB-1, Shc and c-Cab1 [ 16 ]. Among these, IRS-1 protein is known to participate in the regulation of GLUT4 translocation. In the present study we have assumed that, at steady state the twice-bound phosphorylated receptor catalyses the phosphorylation of IRS-1 protein while neglecting the activation of GAB-1, Shc, c-Cab1. The phosphorylated active IRS-1 further binds and activates PI3 kinase and this association is assumed to occur with a stoichiometry of 1:1. Activated PI3 kinase further phosphorylates phosphatidylinositol-(4,5)-bisphosphate (PI-4,5-P2) to form phosphatidylinositol -3,4,5-triphosphate, (PIP3). The dephosphorylation of PIP3 to form PI-4,5-P2 is catalyzed by phosphatase PTEN, whereas, PIP3 is dephosphorylated to form PI-3,4-P2 by phosphatase SHIP2. Active PIP3 then is known to interact allosterically with phosphosinsositide-dependent kinase 1 (PDK1) and which in turn appears to phosphorylate kinase Akt (or protein kinase B) and protein kinase C-ζ (PKC-ζ) [ 11 ]. However, as the interaction due to PDK1 is unclear, active PIP3 is assumed to play a role in phosphorylation of Akt and PKC-ζ. Since the parameters affecting the modification-demodification of Akt and PKC-ζ are considered to be similar, their modification is represented as a single enzyme cascade (Fig. 2 ). The downstream elements of Akt and PKC-ζ, which effect GLUT4 translocation, are also unknown [ 11 - 13 ]. Therefore, we have assumed that phosphorylated Akt and PKC-ζ directly activate the GLUT4 translocation to the plasma membrane. In the basal state, GLUT4 slowly recycles between the plasma membrane and intracellular vesicular compartment. The phosphorylated Akt and PKC-ζ favor GLUT4 translocation (exocytosis) to the plasma membrane and thus increase glucose uptake as a response to insulin binding to the receptor [ 14 ]. Here, total GLUT4 (G t ) is assumed to be sum of GLUT4 concentration in the cytosol (G C ) and on the membrane (G M ). The rate equation for GLUT4 species in cytoplasm at steady state is represented by, where, k 9 is the basal zero order synthesis rate of GLUT4, k 10 is basal first order degradation rate, k 11 is the insulin-activated GLUT4 exocytosis, k 12 and k 13 are basal first order rate of exocytosis and endocytosis, respectively. As assumed by Sedaghat, et al . [ 26 ], the basal equilibrium distribution of cell surface GLUT4 and GLUT4 in the intracellular pool are taken as 4% and 96%. The insulin signaling pathway has been shown to consist of multiple feedback loops [ 15 ]. Active Akt is known to phosphorylate and thereby negatively regulate the upstream phosphatase PTP1B. This phosphorylation impairs the ability of PTP1B to dephosphorylate insulin receptor and IRS-1 by 25% [ 31 ]. This represents overall positive feedback loop as Akt inhibits signal attenuation enzyme PTP1B. The resulting circuit also represents a double negative feedback loop, in which phosphorylated protein negatively regulate the phosphatase that dephosphorylates it. To incorporate these feedback loops we assumed that active Akt affects the total active PTP1B enzyme and thus inhibits the dephosphorylation of the receptor and IRS-1. The feedback effect of Akt on PTP1B was incorporated by following relationship where, [PTP] max is maximum PTP1B concentration, PTP t is the total active PTP1B concentration after incorporating the effects of feedback, AktP represents the phosphorylated Akt concentration influencing the PTPase activity, and k f represents the half saturation constant quantifying feedback. The value of k f was estimated based on the assumption that 25% of PTP1B is inactivated by total AktP [ 31 ]. Thus, k f is appropriately calculated so that the first term [k f /[k f + AktP]] is equal to 0.75. In absence of feedback effects, PTP t equals PTP max . The set of equations given in 'appendix' and in 'methods' section were solved numerically using fsolve program of Matlab (The MathWorks Inc. USA). The accuracy of the simulation was verified by numerically checking the mass balance of all species. The steady state modeling of entire insulin signaling was evaluated including the feedback loops and estimating the fractions of GLUT4 translocated to the plasma membrane for a particular concentration of insulin. Thus, the overall action of insulin on GLUT4 translocation is quantified as, where, f is fractional GLUT4 on plasma membrane, G M is GLUT4 concentration on plasma membrane and G t is total GLUT4 concentration with respect to whole cell volume. Results Bistability in GLUT4 translocation to plasma membrane Fig. 3A shows the predicted dose response curve of steady state fraction of GLUT4 bound to the plasma membrane at different concentrations of insulin. The predicted dose response curve indicates that, there are three steady states exist between 0.01 nM and 0.05 nM of insulin for GLUT4 translocation (curve b, Fig 3A ). Out of these three steady states, GLUT4 gets distributed between two discrete stable steady states, either at plasma membrane or in the cytosol without settling in an intermediate unstable state, thus showing a typical hysteresis response. Due to hysteresis, the dose response curve appears to split and we obtain two distinct half-maximal concentrations (K 0.5 , insulin concentration required for 50% of GLUT4 to reside on the plasma membrane). This represents two threshold concentrations of insulin required for GLUT4 translocation switching on (GLUT4 translocation to plasma membrane at 0.05 nM) and switching off (GLUT4 translocation from to plasma membrane at 0.01 nM). Figure 3 Hysteresis and bistability in insulin-stimulated GLUT4 translocation. A . Dose response curve of insulin stimulated fractional GLUT4 on plasma membrane. Curve 'a' is sigmoidal dose response curve [~Hill coefficient of 3.1] obtained in absence of feedback loop. Curve 'b' represents hysteresis in insulin-stimulated fractional GLUT4 on plasma membrane in presence of feedback loop which impairs the ability of PTPase by 25%. Arrows indicate the switching-on [up arrow] and switching-off [down arrow] GLUT4 translocation. B . A simulated type 2 diabetic condition represented by dose response curve of insulin-stimulated fractional GLUT4 on plasma membrane at higher phosphatase PTP1B concentration. Curve 'a' is typical bistable response obtained in presence of positive feedback loops [PTP1B conc. 0.039 nM]. Curve 'b' represents dose response curve when PTPase concentration was increased by 3 fold [PTP1B conc. 0.098 nM]. A 3-fold increase in the PTPase concentration increased the half-maximal concentration by 100 fold and the response looses bistability. The observed hysteresis is characteristic of a bistable response obtained due to negative feedback regulation of upstream signal attenuation enzyme PTP1B by downstream kinase Akt. Experimental data available in the literature indicates a subsensitive response of insulin, requiring ~130 fold change in insulin concentration for the maximal GLUT4 translocation to plasma membrane [ 32 ]. Our results show an ultrasensitive response in insulin-stimulated GLUT4 translocation due to bistability (~4-fold change in insulin concentration); however, the half saturation values match with that of experimental data. The response was ultrasensitive (Hill coefficient ~3.1) and not bistable in absence of feedback loops (curve a, Fig 3A ). Effect of system component concentration on GLUT4 translocation To examine the influence of pathological conditions arising due to variations in protein expression levels on final output response of insulin, we varied the concentration of individual signaling components IRS-1, PI3K, lipids, PKC-ζ, Akt and phosphatases, PTP1B, PTEN and SHIP2 over a wide range. For each case, the dose response curve of fractional GLUT4 on the plasma membrane at various insulin concentrations was obtained and the response was quantified in-terms of half saturation constant. To illustrate this, we consider a case of increase in PTP1B concentration. Fig. 3B shows the dose response curves for insulin stimulated GLUT4 translocation at plasma membrane at two different concentrations of PTP1B. At high PTP1B concentration, the bistable dose response curve becomes monostable (but, still ultrasensitive) and shifts to the right. This indicates a nullifying effect of negative feedback regulation on PTP1B by Akt and higher requirement of insulin for maximal translocation of GLUT4. Thus, in Fig 3B curve 'a' and curve 'b' can be characterized by two and one half saturation values respectively. Fig. 4A and 4B show the distinct half saturation constant values obtained for switching-on and switching-off of GLUT4 translocation at various concentrations of IRS-1 and Akt respectively. Such an increase or decrease in the half-maximal concentration of insulin characterizes the decrease and increase in insulin sensitivity found in diseased conditions. The threshold concentration of insulin required for switching-on GLUT4 translocation decreases with increase in IRS-1 concentration. This implies that, increase in IRS-1 concentration amplifies the input signal and beyond a certain concentration of IRS-1 [~3 nM], the system looses bistability. Similar results were obtained for variations in lipid, PI3K and insulin receptor concentration (results not shown). GLUT4 translocation at various concentrations of Akt shows that the system becomes monostable when Akt concentration is decreased. However, the degree of bistability (i.e., difference between half maximal concentrations for switch-on and off) increases with increase in Akt concentration and furthermore, the threshold value to activate GLUT4 translocation decreases. Figure 4 Half-maximal concentration of insulin required for 50% GLUT4 translocation at elevated levels of signaling components. Curve 'a' shows half maximal concentration of insulin required to switch-on GLUT4 translocation. Curve 'b' shows half maximal concentration of insulin required to switch-off GLUT4 translocation. Arrow indicates physiological concentration of particular signaling components. A . Half saturation constant at various concentration of IRS-1. Simulated results indicate increased insulin sensitivity when IRS-1 overexpressed. B . Half saturation constant at various concentration of Akt. Simulated results indicate increased insulin sensitivity when Akt overexpressed and loss of bistability when Akt concentration decreased below 0.01 nM. C . Half saturation constant at various concentration of PTP1B. Simulated results indicate decreased insulin sensitivity when PTP1B overexpressed. D . Half saturation constant at various concentration of PTEN. Simulated results indicate decreased insulin sensitivity when PTEN overexpressed. To study the effect of signal attenuation enzymes such as phosphatases on the output response, the concentrations of PTP1B, PTEN and SHIP2 were altered over a wide range, keeping other parameters constant. Fig. 4C and 4D show the influence of variation in concentrations of PTP1B and PTEN on half saturation constant of insulin. Increase in PTP1B and PTEN concentration results in a drastic increase in the threshold concentration of insulin required to switch-on or switch-off GLUT4 translocation. This illustrates that more insulin than physiological concentration is required at higher phosphatase (PTP1B or PTEN) concentrations to translocate GLUT4 from cytoplasm to plasma membrane. For example, around 16-fold change in the insulin concentration is observed for a 1.5-fold increase in PTP1B concentration from 0.039 nM to 0.06 nM. The system looses bistability beyond a narrow range of PTP1B concentration between 0.02 nM to 0.05 nM. Thus, the response of GLUT4 translocation to insulin is particularly sensitive to PTP1B concentration. Influence of feedback on GLUT4 translocation The feedback effect of active Akt on PTP1B was studied by increasing the Akt concentration (Fig. 5A ) and by changing the percentage feedback at a fixed Akt concentration (Fig. 5B ). As shown in Fig. 5A , increase in Akt concentration amplifies the signal by maintaining bistable response. Similarly, by increasing the percentage feedback at a fixed Akt concentration, (Fig. 5B ) the degree of bistability dramatically increased, while not influencing the threshold concentration required for switching-on the response. The bistable response was not observed when percentage feedback was smaller or in absence of feedback loops. In absence of receptor internalization, 65% inhibition of PTP1B by Akt was required to display a bistable response, whereas, inclusion of receptor internalization demonstrated bistability even at 25% inhibition of PTP1B. Figure 5 Influence of feedback effects on bistable insulin-stimulated GLUT4 translocation. A . Bistable response with increase in the concentration of Akt representing increased non-linearity due to zero order ultrasensitivity. Dose response curves obtained at different Akt concentrations: Curve 'a', 0.01 nM; Curve 'b', 0.03 nM; Curve 'c', 0.05 nM. B . Influence of percentage of feedback effects on dose response curve of insulin-stimulated GLUT4 translocation. The percentage feedback represents the percentage by which the dephosphorylation ability of PTP1B is impaired. Dose response curves obtained: Curve 'a' in absence of feedback; Curve 'b' 25% feedback effect; Curve 'c' 67% feedback effect; Curve 'd' 90% feedback effect. The steady state analysis of metabolic insulin-signaling pathway demonstrated signal amplification as signal propagates down the cascades. The amount of insulin required for 50% activation of insulin receptor, IRS-1, PIP3, Akt, PKC-ζ and GLUT4 was estimated to decrease in presence or absence of feedback loops (results not shown). Effect of system parameter values on GLUT4 translocation In addition to genetic variation at the protein expression levels in diseased conditions, mutational changes can also alter the system parameters and thereby modify the final output response. To examine the influence of system parameter values on insulin-stimulated GLUT4 translocation, we have analyzed the performance of insulin signaling pathway to variations in key parameter values such as, dissociation constant and Michaelis-Menten constant. Increase in dissociation constant quantifying the interaction between insulin-receptor and phosphorylated IRS-1-PI3K shows an increase in the half saturation constant indicating higher requirement of insulin over the physiological concentration (Fig. 6A and 6B ). The system becomes monostable at very low values of dissociation constants. Similarly, decrease in the Michaelis-Menten constant of the dephosphorylation cycles, also increases the half saturation constant, thus decreasing the insulin sensitivity (Fig. 6C ). Simulation results indicate that, the alterations in binding constant of allosteric interactions and Michaelis-Menten constants in modification-demodification cycles in the insulin-signaling pathway can result in insulin resistance or diabetes. Figure 6 Effect of key system parameter values on GLUT4 translocation. Curve 'a' shows half maximal concentration of insulin required to switch-on GLUT4 translocation. Curve 'b' shows half maximal concentration of insulin required to switch-off GLUT4 translocation. Arrow indicates parameter used in the simulation. A . Half maximal concentration of insulin required for GLUT4 translocation at different values of dissociation constant [Kd 2 ] for binding of second molecule of insulin to phosphorylated insulin bound receptor. Simulated results indicate decreased insulin sensitivity when Kd 2 increased. B . Half maximal concentration of insulin required for GLUT4 translocation at different values of dissociation constant [Kd 3 ] for binding of phosphorylated IRS-1 to PI3K species. Simulated results indicate decreased insulin sensitivity when Kd 3 increased. C . Half maximal concentration of insulin required for GLUT4 translocation at different values of Michaelis-Menten constant [Km 2 ] for dephosphorylation of phosphorylated IRS1 by PTP1B. Simulated results indicate decreased insulin sensitivity when Michaelis-Menten constant [Km 2 ] was decreased due to increased affinity with dephosphorylating enzyme. Discussion In this work we have demonstrated that, the dose response curves of fractional GLUT4 concentration on plasma membrane at various concentration of insulin exhibit hysteresis-a property of bistable systems. The analysis of bistable response in presence of feedback loops was done at varying concentration of signaling components and system parameters in physiological range. The overall response of insulin demonstrated signal amplification as the signal propagates down the cascade, thus requiring less insulin for GLUT4 translocation. The insulin sensitivity increased by increasing the concentration of proteins that amplify the insulin action and decreasing the concentration of proteins that attenuate insulin-signaling pathway. This indicates that the bistability and the half saturation constant are dependent on the component concentrations and system parameters. It is known that defects in insulin signaling pathway leads to pathological conditions like diabetes, wherein normal or elevated levels of insulin produces impaired biological response. This characteristic decrease or increase in insulin sensitivity is mainly attributed to post-receptor defects including mutational changes in protein expression levels or other parameters like dissociation constants and Michaelis-Menten constants [ 13 , 33 ]. Numerous experimental studies like targeted deletions/mutations of signaling components have yielded insights about the disease states. In the present work, to study the influence of pathological conditions on final output response of insulin, the concentration of individual signaling components was varied over a wide range, by keeping other parameters constant. The predicted results are consistent with various reported experimental observations and thus validate our steady state model. ( i ) Decreased concentration of phosphorylated insulin receptor and IRS-1 are observed in muscle from morbidly obese subjects [ 34 ] and those with diabetes [ 35 ]. This decreased phosphorylation can be either due to decrease in protein concentration itself or impaired phosphorylation event. ( ii ) Increase in the level and activity of several tyrosine phosphatases (PTP1B) was found to be associated with insulin resistance and reduced insulin sensitivity [ 12 , 13 , 33 , 36 ]. ( iii ) Overexpression of PI3K and its downstream targets Akt and PKC in cell culture models increased GLUT4 translocation [ 12 ]. (i v ) Decrease in the association of PI3K with phosphorylated IRS-1 and subsequent activation appears to be a characteristic abnormality in type 2 diabetes and obesity [ 13 , 33 - 35 ]. ( v ) Single gene knockout experiments have shown that, mice with PTP1B knockout [ 37 ], mice with SHIP2 knockout [ 38 ] and targeted deletion of PTEN in murine lever [ 39 ], all results in hypersensitivity towards insulin. In the present work, though we have not done in-silico perturbation analysis by deleting a particular protein, we have changed the concentration of specific protein over wide range to bring about the similar effect of deficiency. Thus, our simulation results show that the insulin sensitivity dramatically increased when we decreased the concentration of phosphatases like PTP1B, PTEN and SHIP2. Increase in the concentration of Akt, makes the signal amplification increased along with slight increase in the degree of bistability. This effect is brought about by the enhanced nonlinearity in the feedback loop due to zero order ultrasensitivity [ 28 ] imposed by increasing the concentration of Akt or percentage feedback. At high Akt concentration (or when overexpressed), the system can respond in constitutively active mode or might also function as a memory module. That is, once insulin switches on the system, the high Akt concentration or percentage feedback by itself can keep the switch on even after insulin is removed. This may be the reason for the experimental observation of insulin independent GLUT4 translocation to plasma membrane when Akt is overexpressed [ 12 , 40 ]. This insulin independent translocation of GLUT4 is thought to be due to activation of PI3K independent pathway or by amplification of residual signal. Our analysis indicates that the inherent feedback structure present in the insulin-signaling pathway by itself can induce this phenomenon. Does GLUT4 translocation show a bistable response to insulin in-vivo? Bistability has been shown to be the outcome of regulatory structure with feedback loops and non-linearity in the system [ 41 ]. The non-linearity in the system is brought about by an ultrasensitive response typically obtained through enzyme cascades. Such ultrasensitive responses exhibit steep dose response curves with Hill coefficient greater than one [ 1 ]. The cascade systems use energy for their operation and are optimally operated under zero order sensitivity (i.e., cascades operating under saturating interconvertable enzymes) requiring minimum energy [ 42 , 43 ]. Presence of feedback loops may further optimize the requirement of energy. Enzyme cascades and multiple positive feedback loops are observed in insulin-signaling pathway. Experimental results have shown that the dose response curve of insulin-stimulated glucose uptake is subsensitive with a Hill coefficient of about 0.8 [calculated from ref. [ 32 ]]. Thus the question arises as to what may be the significance of the cascade signaling system with positive feedback loops in insulin signaling pathway. The reason for this discrepancy may be because, the experimental data represents a profile of glucose uptake in ensemble of cells [ 32 ], and furthermore, glucose uptake may not be proportional to the amount of GLUT4 on cell surface [ 18 ]. Recently, bistability has been experimentally demonstrated in many cellular regulation systems [ 10 ]. Experiments on cellular differentiation and cell-cycle progression have shown that, to demonstrate bistability it is essential to measure the amount of input required to switch-on the system from a low activity state to a high activity state, and amount of input required to hold-on the system in high activity state [ 3 - 5 ]. Reynolds et al . [ 8 ], have shown experimentally that, the reaction network of PTPase inhibition by activated epidermal growth factor receptor (EGFR, a receptor tyrosine kinase, RTK) through reactive oxygen species, can generate highly amplified and switch like bistable response to a threshold concentration of ligand stimulus. In this system, PTPase is a negative regulator of active RTK and thus, PTPase inhibition by reactive oxygen species serves as a positive feedback loop. Our simulation results indicate that similar bistable response can be obtained in insulin-stimulated GLUT4 translocation because of the positive feedback loops (inhibitory action of Akt on PTP1B). Though experimental verification of this property is awaited, there are indications that insulin signaling pathway possesses many requisite components to exhibit bistable response. The simulation results presented here showed that, the ultrasensitivity in absence of feedback effects and the regulatory structure of feedback loops are main reasons for a bistable response. Other than the positive feedback loops considered in the present work, Insulin signaling system is also known to contain many feedback loops which have not been entirely characterized [ 15 ]. One possible reason for having so many interlocking positive feedback and negative feedback loops may be to cause oscillations in GLUT4 translocation or to make the bistability of GLUT4 translocation – more robust. Recently, it has been shown that insulin stimulation in a variety of cell types elicit a rapid production of hydrogen peroxide and which reversibly inhibit PTP1B enhancing propagation of the early insulin signal [ 44 ]. This regulatory mechanism was also found to be essential for PI3K mediated Akt activation, GLUT4 translocation to plasma membrane and insulin-stimulated glucose uptake [ 45 ]. However, unlike EGFR system [ 8 ] existence of bistable behavior is yet to be shown in insulin signaling system. In insulin signaling pathway other than GLUT4, proteins like Akt and PKC get translocated to plasma membrane and such inter-compartmental translocations can also exhibit switch like bistable response [ 9 ]. It is becoming clear that quantification studies have to be performed in single cell rather than cell populations [ 19 ]. This is true especially while addressing the system level questions like ultrasensitivity, bistability and oscillations [ 4 - 7 , 46 ]. Recently, this was also found to be of immense value in case of insulin signaling pathway to uncover the PIP3 activation mode [ 47 ]. Simultaneous measurement of PIP3 production and GLUT4 insertion in individual 3T3L1 adipocytes suggested that a threshold for PIP3 signals exists. Below this threshold, GLUT4 transporters are minimally inserted into the plasma membrane and their insertion increases once this threshold is overcome. In summary, it is essential to show through experiments that, the stimulus-response curve for insulin-stimulated GLUT4 translocation exhibits hysteresis, – a distinctive splitting in stimulus response curve. There should be a particular concentration of insulin, which is not sufficient to translocate GLUT4 to plasma membrane, but is sufficient to maintain GLUT4 on plasma membrane. What may be the significance of such a bistable behavior in GLUT4 translocation? Though there is no obvious answer to this question yet, it is known that, bistability can maintain a biological response even when the input stimulus is brief and the high activity level is maintained only as long as the system requires. Insulin binding to its cell surface receptor is rapidly followed by internalization of insulin-receptor complex. This internalization of receptor has been implicated in receptor down regulation, attenuation of insulin sensitivity and insulin clearance from the circulation [ 12 , 13 ]. Thus a brief input stimulation should be sufficient to balance the translocation of GLUT4 to plasma membrane and its reversal depending on glucose concentration. Additionally, due to small absolute stimulus concentrations, the probability of noise occurrence is high. A bistable response having distinct threshold concentration to switch-on and switch-off offers advantage in handling noise. In cellular regulation, different structural motifs such as enzyme cascades, feedforward control and multiple feedback loops yield complex regulatory networks. These are shown to be responsible for exhibiting system level properties including bistability and robustness [ 10 , 48 , 49 ]. Other than multiple feedback loops, structural regulatory motifs like multisite covalent modification cycles have been shown to induce bistability [ 50 ]. The interconnections between these regulatory motifs at the system level may elicit a multistable response to particular stimuli, which have to be theoretically uncovered and demonstrated through experiments. Abbreviations used GLUT4: Glucose-transporter isoform 4; IRS: Insulin-receptor substrate; PI3K: Phosphatidylinositol-3-kinase; PIP3: Phosphatidylinositol (PI)-3,4,5-tiphosphate (PI-3,4,5-P3); Akt: Protein kinase Akt or protein kinase B (PKB); PKC: Protein kinase C; PTP1B: Protein tyrosine phosphatase 1B; PTEN: 3' lipid phosphatase; SHIP2: 5' lipid phosphatase; Competing Interests None declared. Author's Contributions LG, VKM and KVV conceived and designed the experiments. LG performed the experiments. LG, VKM and KVV analyzed the data. VKM and KVV conceptualize the manuscript. All authors have read and approved the final manuscript Appendix Equilibrium relationships, rate equations, mass balance equations and model parameters used for simulation of metabolic insulin signaling system at steady state (refer Fig. 2 for nomenclature and interaction details). Equations were solved simultaneously, for evaluating fractional GLUT4 translocation at a particular insulin concentration, using fsolve function of Matlab (The MathWorks Inc. USA). Most of the values of model parameters for steady state analysis are taken from Sedaghat et al . [ 26 ]. Nomenclature, parameter values are: Component concentrations I t , total insulin concentration varied over a wide range; X t , total receptor = 0.003 nM; A t , total IRS-1= 0.01 nM, B t , total PI3-Kinase = 0.5 nM, PTEN t , total PTEN= 0.007 nM; CP2 t , total lipid = 0.01 nM; SHIP2 t , total SHIP2 = 0.001 nM; F t , total Akt+PKC-ξ = 0.02 nM, PTP max , total PTP1B= 0.039 nM; G t , total GLUT4 = 9 nM; E 6t , total E 6 = 0.001 nM; E 8t , total E 8 = 0.001 nM; Rate constants k 0 = 2500 min -1 ; k = 0.2 min -1 ; k 1 = 4.16 min -1 ; k 2 = 1.4 min -1 ; k 3 = 50 min -1 (assumed); k 4 = 42.1 min -1 ; k 5 = 2.8 min -1 ; k 6 = 3 min -1 ; k 7 = 20 min -1 (assumed); k 8 = 6.9 min -1 ; k 9 = 0.11 min -1 ; k 10 = 0.0012 min -1 ; k 11 = 3.47 min -1 (assumed); k 12 = 6.96*10 -3 min -1 ; k 13 = 0.17 min -1 ; k p = 0.461 min -1 ; k d = 1.67 × 10 -18 min -1 ; k s = 1.67*10 -25 nM min -1 ; Dissociation constants Kd 1 = 3.33 nM; Kd 2 = 333.3 nM; Kd 3 = 0.014 nM; Distribution coefficients Kd 4 = 9 nM; Kd 5 = 0.1 nM; Michaelis-Menten constants Km r , Km 1 to Km 8 = 10 -4 nM The total molar balances for different species are given below. I t = I + XI + XIP + 2 XI 2 P + PTP . XIP + A . XI 2 P + XIP i + 2 XI 2 P i + PTP . XIP i + PTP . XI 2 P i [A1] X t = X + X i + XI + XIP + XI 2 P + XIP i + XI 2 P i + PTP . XIP + A . XI 2 P + PTP . XIP i + PTP . XI 2 P i [A2] A t = A + AP + APB + A . XI 2 P + PTP . AP + APB . CP2 [A3] CP 2 t = CP 2 + CP 3 + CP 2' + APB . CP 2 + PTEN . CP 3 + SHIP 2. CP 3 + SHIP 2. CP 3 + E 6 . CP 2' + F . CP 3 [A4] F t = F + FP + CP 3 F + E 8 . FP [A5] SHIP 2 t = SHIP 2 + SHIP 2. CP 3 [A6] PTEN t = PTEN + PTEN . CP 3 [A7] PTP t = PTP + PTP . XIP + PTP . AP + PTP . XIP i + PTP . XI 2 P i [A8] B t = B + APB + CP 2. APB [A9] E 6 t = E 6 + E 6 . CP 2' [A10] E 8 t = E 8 + E 8 . FP [A11] G t = G M + G C [A12] Equilibrium relationships Rate expression with pseudo-steady state representation of complexes for modification-demodification cycles Receptor autophosphorylation and dephosphorylation cycle IRS-1 phosphorylation and dephosphorylation cycle Phosphorylation and Dephosphorylation of PI-4,5-P2, PI-3,4-P2 and PIP3 Phosphorylation and Dephosphorylation of Akt + PKC | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516236.xml |
546398 | Reaching in reality and virtual reality: a comparison of movement kinematics in healthy subjects and in adults with hemiparesis | Background Virtual reality (VR) is an innovative tool for sensorimotor rehabilitation increasingly being employed in clinical and community settings. Despite the growing interest in VR, few studies have determined the validity of movements made in VR environments with respect to real physical environments. The goal of this study was to compare movements done in physical and virtual environments in adults with motor deficits to those in healthy individuals. Methods The participants were 8 healthy adults and 7 adults with mild left hemiparesis due to stroke. Kinematics of functional arm movements involving reaching, grasping and releasing made in physical and virtual environments were analyzed in two phases: 1) reaching and grasping the ball and 2) ball transport and release. The virtual environment included interaction with an object on a 2D computer screen and haptic force feedback from a virtual ball. Temporal and spatial parameters of reaching and grasping were determined for each phase. Results Individuals in both groups were able to reach, grasp, transport, place and release the virtual and real ball using similar movement strategies. In healthy subjects, reaching and grasping movements in both environments were similar but these subjects used less wrist extension and more elbow extension to place the ball on the virtual vertical surface. Participants with hemiparesis made slower movements in both environments compared to healthy subjects and during transport and placing of the ball, trajectories were more curved and interjoint coordination was altered. Despite these differences, patients with hemiparesis also tended to use less wrist extension during the whole movement and more elbow extension at the end of the placing phase. Conclusion Differences in movements made by healthy subjects in the two environments may be explained by the use of a 2D instead of a 3D virtual environment and the absence of haptic feedback from the VR target. Despite these differences, our findings suggest that both healthy subjects and individuals with motor deficits used similar movement strategies when grasping and placing a ball in the two reality conditions. This suggests that training of arm movements in VR environments may be a valid approach to the rehabilitation of patients with motor disorders. | Introduction Virtual reality (VR) is a computer-based, multisensory interactive simulation occurring at the same speed and time as events in the physical world. Different levels of immersion can be achieved ranging from complete 3D (cave, head-mounted display) to partial 2D (computer display, TV screen) with different hardware configurations. Interface devices (computer mouse, joystick, force sensor, cyberglove) allow the user to move in and interact with objects in the virtual environment. Of crucial relevance to rehabilitation is the potential for increasing the user's level of interaction with their real physical environment so as to maximize their return to community life [ 1 ]. The efficacy of using VR to retrain movement and the issue of whether training in a virtual environment will transfer to meaningful function in the real physical world has been explored in a number of studies with encouraging early results [ 2 - 4 ]. Neurophysiologists and rehabilitation specialists like physical and occupational therapists are beginning to be interested in VR as a tool to study motor control and to evaluate and treat motor deficits secondary to central nervous system lesions such as stroke [ 5 ]. The use of virtual computer-based interventions for telerehabilitation is also gaining in popularity because of the possibility of providing extended practice in the patient's own home or community environment [ 6 , 7 ]. The advantage of using VR in community, clinical and laboratory settings is that by virtue of its programmability, environments and the amount and type of feedback can be modified according to the user's motor capacities, motivation and therapeutic goals [ 5 , 8 ]. In addition, sensory parameters of the environment can be creatively adapted to evoke responses to a larger number of situations in a shorter amount of time than is available in physical set-ups. For example, in research studies, when determining the capacity to reach and grasp static targets, methodologies are often limited to one or two tasks because of the inability to easily adapt the experimental hardware. VR permits the use of more dynamic experimental set-ups in which object locations and orientations can be reliably and rapidly modified. This study focused on the possibility of using virtual environments for the retraining of arm motor function in individuals with hemiparesis due to stroke. Major barriers to arm motor recovery after stroke are coordination deficits and the use of maladaptive movement strategies for reaching and grasping. Patient motivation and movement repetition are key factors in motor recovery [ 9 - 12 ]. Current practice of rehabilitation of reaching deficits after stroke is based on movement repetition of targeted tasks. However, improvements in tasks practised in clinical settings have not been shown to have adequate carry over into real world activities of daily living [ 10 ]. One of the factors that may decrease the real world relevancy of practice in the clinical setting is the lack of attention to the retraining of varied goal-directed, effector-relevant whole arm movements. VR is an ideal medium in which to create such practise environments that have the advantage of providing additional motivation to patients to perform repetitive movement and can be available in the home or community following formal rehabilitation [ 13 ]. Indeed, some studies have reported that motor gains achieved by patients with stroke in VR environments may transfer to physical tasks and be measurable using common clinical scales [ 2 , 5 , 14 ]. Despite the growing interest in the use of VR for motor retraining, it is not known if movements involving reaching and grasping objects in VR environments are performed in a manner similar to those done in the physical world. Thus, the goal of this study was to validate VR as a tool for studying reaching and grasping in healthy subjects and in individuals with hemiparesis by comparing movement kinematics of identical tasks made in a physical and a virtual environment. Since reaching and grasping deficits have been well characterized in individuals with hemiparesis [ 15 - 17 ], the purpose of the study was not to compare movements between groups but to establish the validity of using a VR environment for the study of movement in each group. Preliminary results have appeared in abstract form [ 18 ]. Methods Eight healthy subjects (4 males and 4 females; 56.8 ± 17.1 years) and 7 adults with hemiparesis (3 males and 4 females; 48.9 ± 18.6 years) with no prior experience with VR participated in the study. Potential participants were identified from discharge lists of Montreal area rehabilitation centres. Out of 17 medical charts screened, 12 patients met eligibility requirements according to study inclusion and exclusion criteria. Patients were included if they were under 60 years old, had sustained a single, non-traumatic unilateral stroke in the territory of the left middle cerebral artery and had arm paresis (3/7 for the hand and 6/7 for the arm on the Chedoke-McMaster Stroke Assessment Scale [ 19 ] (Table 1 ). Patients were excluded if they had cerebellar or brain stem lesions, shoulder pain or other neurological/orthopaedic conditions affecting reaching ability, visual field deficits, uncorrected problems of visual acuity or severe perceptuo-cognitive deficits (heminegligence, ataxia, receptive aphasia) determined by standard clinical tests. Of these 12 individuals, 10 expressed willingness to participate. After obtaining informed consent approved of by the institutional Ethics Committee, they were assessed by a physical therapist and 3 individuals were excluded because of inability to perform the task. Healthy subjects were recruited from the community. They had no orthopaedic or neurological disease. All patients had been discharged from all in- or out-patient clinical services. Table 1 Demographic characteristics and clinical scores of participants with hemiparesis Subject Age (yrs)/sex Time since injury (months) Type of lesion CM: arm CM: hand 1 63/M 28 Temporo-parietal 6 6 2 42/F 34 Parietal 7 6 3 27/F 63 Parietal 7 3 4 51/F 51 Frontal 7 6 5 31/F 33 Temporo-parietal 6 6 6 47/M 64 Fronto-temporo-parietal 7 6 7 81/M 33 Temporo-parietal 7 6 Mean ± SD 48.9 ± 18.6 43.7 ± 15.3 6.7 ± 0.5 5.6 ± 1.1 M = male, F = female, CM = Chedoke-McMaster stroke assessment Subjects performed 6 trials each of two near identical tasks set in the physical world or in a virtual environment. In both tasks, seated subjects grasped a real or virtual ball of 7 cm diameter with their right hand, beginning from the edge of a real or virtual table, reached forward by leaning the trunk and then placed the ball within a 2 cm × 2 cm yellow square on a real or virtual target (Figure 1 ). Care was taken to set-up the physical task so that the initial position of the arm, ball, table and wall were identical to that of the virtual task. Thus, in both environments, the initial position of the arm was about 0° flexion, 30° abduction and 0° external rotation (shoulder), 80° flexion and 0° supination (elbow) with the wrist and hand in the neutral position. The fingers were slightly flexed. The initial position of the ball was 13 cm in front of the right shoulder, 7 cm above and 3 cm to the left of the subject's hand. The target was placed 31 cm in front of the shoulder, 12.5 cm above and 14 cm to the right of the initial position of the ball (Figure 1 ). Figure 1 Experimental set-up. Physical ( a ) and virtual reality condition (b) . For the VR task, the ball appeared on a computer screen inside a cube that also displayed the position of the subject's hand. The VR target was the upper right back corner of the cube. Subjects had to grasp the virtual ball, transport it to the VR target and release it. The VR environment was displayed in 2 dimensions (2D) on a computer screen placed 75 cm in front of subject's manubrium (Figure 1 ). The virtual representation of the subject's hand was obtained using a 22 sensor fibre optic glove (Cyberglove, Immersion Corp.) and an electromagnetic sensor (Fastrak, Polhemus Corp.) that was used to orient the glove in the 2D environment. Data from these devices were synchronized in real time. To enable the subject to "feel" the virtual ball, a prehension force feedback device (Cybergrasp, Immersion Corp.) was fitted to the dorsal surface of the hand. The Cybergrasp delivered prehension force feedback in the form of extension forces to the distal phalanxes of the thumb and each finger. Forces applied to the fingers were calibrated for each subject while he/she was wearing the Cyberglove. These ranged from 6 to 8 N per finger and all subjects perceived that they were holding a spherical object in their hand. Prior to data collection, all participants practised the tasks in physical and virtual conditions (20 – 40 min). To better compare the participants' performance in the two environments, the glove and grasp devices were worn on the hand in both conditions (Figure 1 ). Kinematic data from the right arm were recorded with 6 infrared-emitting diodes (IREDs) placed on the distal phalanx of the index and thumb, the distal head of the first metacarpal, the radial styloid process, the lateral epicondyle of the humerus and the acromion (120 Hz, Optotrak Motion Analysis System, Northern Digital Corp.). Each trial was divided in two phases: 1) reaching and grasping the ball and 2) ball transport and release. For the first movement phase, 4 temporal and 4 spatial parameters of reaching and grasping were determined. Temporal parameters were movement time, time to peak wrist velocity (RPV), time to maximal hand aperture (RMGA), and the delay between them (RPV-RMGA). Spatial parameters were endpoint path curvature, maximal grip aperture, angular ranges of joint motion and elbow-shoulder interjoint coordination [ 16 ]. For the second movement phase, we determined one temporal (movement time) and 4 spatial (endpoint path curvature, trajectory length, angular ranges of joint motion and interjoint coordination) parameters. Movement onsets and offsets of each phase were defined as the times at which the tangential velocity of the IRED on the index finger surpassed and remained above or fell and remained below 10% of the maximal peak velocity respectively. The temporal parameters (time to peak wrist velocity, time to maximal grip aperture) were normalized to movement time and the delay between them was calculated. For the spatial parameters, the curvature of the trajectory of the IRED on the index finger was estimated as the ratio between the actual trajectory length and a straight line segment between the initial and final positions [ 20 ]. Joint angular excursions were expressed as the difference in degrees between the angle at the beginning and the end of movement, according to movement times defined above. For interjoint coordination, we determined the slopes between elbow extension and 1) shoulder flexion and 2) shoulder abduction. The slope of the angle-angle relationship describes the relative contribution of each joint throughout the movement where a slope of 1 indicates an equal contribution of each joint. Slopes greater than 1 indicated a larger contribution of elbow extension than shoulder movement and vice versa. The relationship between both angles was considered linear since all regression correlation coefficients were = 0.8. However, since a linear approximation was used, the slope provides only a general estimate of the contribution of each angle. Statistical analysis Both parametric and non-parametric statistics were used. For within-group comparisons between the two conditions of reality, Student t-tests were used. However, since variances were not homogeneous (Levene's test) for healthy subjects and participants with hemiparesis, non-parametric tests were used for between-group comparisons (Kruskal-Wallis ANOVA). A significance level of p < 0.05 was used, adjusted for multiple comparisons by type using the Bonferroni correction. Results All healthy subjects and participants with mild upper limb motor deficits were able to reach, grasp, transport, place and release the virtual ball using movement strategies that were similar to those used for the physical ball (Tables 2 and 3 ). Arm movement trajectories (Figure 2 ) were smooth and followed similar paths for movements made in both environments for both subject groups. Trajectory lengths were similar in both conditions for healthy subjects (289 ± 28 mm in real compared to 302 ± 55 mm in VR) and for participants with hemiparesis (251 ± 25 mm in real compared to 260 ± 30 mm in VR). Table 2 Comparisons between reality conditions for the first phase of movement: reaching and grasping the ball. Healthy Stroke Physical condition VR condition Physical condition VR condition Mean SD Mean SD Mean SD Mean SD Temporal parameters Movement time – onset to grasping (s) 0.68 0.17 0.95 0.35 1.23† 0.27 1.43† 0.41 RPV (%) 44.9 10.4 41.8 6.1 34.3 12.6 40.8 2.9 RMGA (%) 72.5 12.5 60.9 11.8 73.5 16.1 65.3 9.1 Delay between RPV and RMGA (%) 31.6 16.6 19.2 10.7 34.5 15.0 24.3 8.1 Spatial parameters Curvature index 1.39 0.16 1.62 0.44 1.76 0.62 1.97 0.86 Wrist extension at grasping (°) 1.4 9.1 -3.9 8.2 12.1 2.2 4.5 14.1 Slope elbow extension/shoulder flexion 0.70 0.34 0.60 0.26 0.53 0.33 0.47 0.27 Slope elbow extension/shoulder abduction 2.30 2.02 2.31 2.54 2.65 1.66 2.30 1.26 Maximal grip aperture (mm) 95.7 16.4 90.2 20.5 89.8 20.2 84.9 19.3 RPV = relative time to peak velocity of the wrist; RMGA = relative time to maximal grip aperture; VR = virtual reality † Significant difference between groups, Kruskal-Wallis, p < 0.05 Table 3 Comparisons between reality conditions for the second phase of movement: ball transport and release Healthy Stroke Physical condition VR condition Physical condition VR condition Mean SD Mean SD Mean SD Mean SD Temporal parameters Movement time – onset to placing (s) 0.84 0.29 1.18 0.31 1.49† 0.45 2.28† 0.82 Spatial parameters Curvature index 1.14 0.03 1.23 0.29 1.37† 0.27 1.42 0.57 Wrist extension at placing (°) 18.2 12.1 4.0* 8.1 20.3 8.2 6.4 16.0 Elbow extension (°) 25.6 6.9 38.4* 10.9 26.9 11.0 37.3 19.6 Shoulder flexion (°) 24.8 5.2 33.0 7.1 28.4 4.9 35.5 18.2 Shoulder abduction (°) 13.6 5.2 16.4 4.6 18.5 6.9 20.7 8.9 Slope elbow extension/shoulder flexion 1.08 0.08 1.21 0.16 1.01 0.26 1.18 0.15 Slope elbow extension/shoulder abduction 2.36 0.76 2.88 1.19 1.51† 0.69 1.48† 0.51 * Significant difference between reality conditions, Student t-tests with Bonferroni correction (p < 0.05/4 angles = 0.013); VR = virtual reality † Significant difference between groups, Kruskal-Wallis, p < 0.05 Figure 2 Mean endpoint (marker on the index finger) trajectories for the two phases of the movement task for one healthy subject in the two reality conditions. In healthy subjects, the temporal and spatial aspects of the two phases of the task were almost identical between the physical and virtual conditions (Tables 2 , 3 ). However, there was a non-significant tendency to make movements more slowly and to use less wrist extension for grasping during the first phase of the movement (reaching and grasping the ball) in the virtual condition (Table 2 ). During the second phase, healthy subjects used significantly less wrist extension (paired t-test, p < 0.05) and more elbow extension (paired t-test, p < 0.05) to place the ball on the virtual vertical surface (Table 3 , Figure 3 ). In these subjects, there were no other differences between any other temporal or spatial parameter for both movement phases (peak wrist velocity, relative time to peak wrist velocity, timing of maximal grip aperture, trajectory curvature or interjoint coordination). Figure 3 Interjoint coordination. Relationship between elbow extension and shoulder horizontal adduction (mean traces per condition) during the second phase of the movement (placing) for both conditions in two healthy subjects ( A,B ) and in two individuals with hemiparesis ( C,D ). In all examples, subjects used more elbow extension in the virtual reality condition. Movements made by individuals with hemiparesis in the physical environment differed from those made by healthy subjects in three ways. In both phases, movements were significantly slower and in the second phase, trajectories were more curved and interjoint coordination was altered (Tables 2 and 3 ). In particular, the slope of the relationship between elbow extension/shoulder abduction was lower than in healthy subjects during the second phase of the movement (p < 0.02, Figure 3 ). This decrease in slope was due to a more abducted position of the shoulder in the patient group. Despite these differences, patients showed tendencies similar to healthy subjects when reaching and grasping in the VR environment compared to the real environment (Tables 2 , 3 ). They tended to decrease the speed of movements made in VR compared to the physical environment, to use less wrist extension in both movement phases and to use more elbow extension in the second phase of the movement. In addition, 5 out of 7 participants with hemiparesis significantly decreased the wrist extension while 4 increased elbow extension at the end of the second phase of the movement (at the time of placing the ball) in the VR condition. Discussion The similarity in movement kinematics between physical and virtual reaching and grasping suggests that virtual reality may be an effective environment for rehabilitation. Interest in training in virtual environments is increasing amongst rehabilitation professionals in light of recent evidence suggesting that neuronal recovery after stroke-related brain damage critically depends on the motivation of the individual and the intensity of training [ 10 , 11 ]. Virtual reality represents a novel training environment in which a wide variety of tasks can be easily practiced. It is also becoming increasingly accessible with the advent of home-based computers and telerehabilitation technology [ 7 ]. Thus, the demonstration that movements practiced in a virtual environment are kinematically similar to movements with physical objects is essential to ensure the transfer of training benefits to the real-life situation. Our results show that subjects tended to decrease wrist extension and increase elbow extension in the virtual compared to the physical condition. Two principal factors may explain those differences: the absence of depth perception in the VR condition and the absence of tactile feedback at the end of the reach. Binocular vision enables humans to perceive depth in a 3 dimensional (3D) environment. This faculty is called stereopsis (for a review see [ 21 ]). Since the VR condition in our experimental set-up was a 3D task presented on a 2D display, the results can be compared with those investigating reaching and grasping movements made under conditions of monocular vision in which depth perception is reduced. Indeed, such studies have shown that reaching and grasping movements are characterized by shorter movement time and shorter relative time to maximal grip aperture [ 22 ]. The difference in depth perception in the 2D virtual environment may also be responsible for the tendency to increase elbow extension in both groups. Previous studies have shown that fine motor corrections are produced by distal joints [ 23 ]. The 2D display resulted in the subject underestimating the real distance to the wall so that during the course of the second phase of the movement, the subject had to compensate by increasing the extension of the limb until the screen display indicated that the ball had reached the target distance. This caused a slight change in strategy for the second phase of the movement requiring an increase in the amount of elbow extension. Participants with hemiparesis showed the same tendencies as the healthy group but differences were not significant due to subject variability. Changes in motor patterns may be avoided by using 3D immersive environments, such as those visualized through a head-mounted display. The absence of depth perception cannot explain the decrease of wrist extension at the end of the second phase of the movement in VR compared to the physical condition. A more likely explanation involves the type of haptic feedback provided to the subject. In the physical condition, subjects had to extend the wrist so that the ball and not their fingers would make contact with the target. In the VR condition, wrist extension was not necessary because subjects only had to place the ball at the coordinates of the virtual wall without encountering a physical barrier. To avoid such differences between physical and VR environments, relevant haptic feedback is necessary to indicate contact of the hand with the object or target. Another way is to integrate physical objects into the VR environment such as the manipulation of a real paper envelope in real-time in a VR environment [ 6 ], or the mimicking of irregularities in a walkway with a multi-dimensional movement platform [ 24 , 25 ]. Overall, the finding that both healthy subjects and individuals with motor deficits used similar movement strategies in a physical and a limited virtual environment suggests that VR technology is a valuable tool for studying and retraining reaching, grasping and placing movements. Whether movement kinematics may be improved with the use of interfaces providing stereopsis and more relevant haptic feedback should be investigated by comparing movements made in immersive to those made in non-immersive virtual environments. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC546398.xml |
554763 | Consequences of the historical demography on the global population structure of two highly migratory cosmopolitan marine fishes: the yellowfin tuna (Thunnus albacares) and the skipjack tuna (Katsuwonus pelamis) | Background Yellowfin and skipjack tuna are globally distributed in the world's tropical and sub-tropical oceans. Since little, if any, migration of these fishes occurs between the Atlantic and Indo-Pacific Oceans, one might expect to see genetic differences between sub-populations in these ocean basins. However, yellowfin and skipjack tuna have extremely large population sizes. Thus, the rate of genetic drift should be slower than that observed for other tunas. Results Low levels of genetic differentiation were observed between Atlantic and Pacific samples of yellowfin tuna. In contrast, no genetic differentiation was observed between Atlantic and Pacific samples of skipjack tuna. Conclusion Much lower levels of genetic differentiation were found among sub-populations of yellowfin tuna compared to those observed for other large tunas, probably due to the large population size of yellowfin tuna. Since skipjack tuna appear to have even larger population sizes, it is not surprising that no genetic differentiation was detected between Atlantic and Pacific samples of these fish. | Background The yellowfin tuna, Thunnus albacares , has a global distribution in tropical and sub-tropical oceans. Annual catches of yellowfin tuna have averaged 1.2 million metric tons since 1998 with sizes ranging from 5 to 20 kg [ 1 ]. If the average size of a harvested fish is 10 kg, then the harvest of 1.3 million metric tons in 2002 represents approximately 130 million individual fish. In an earlier study, Scoles and Graves [ 2 ] were unable to find evidence of genetic differentiation between small samples of yellowfin tuna ( Thunnus albacares ) from the Atlantic (n = 20) and Pacific Oceans (n = 100) using an RFLP analysis of whole mitochondrial DNA (mtDNA). Subsequently, Ward et al. [ 3 ] found significant differentiation at the GPI-A* allozyme locus but only weak evidence for genetic differentiation with an RFLP analysis of whole mtDNA. In the whole mtDNA study, much larger sample sizes were employed to increase the sensitivity of the mtDNA assay, but only two restriction enzymes were used and consequently, only a few restriction sites were analyzed. The GPI-A* data were consistent with earlier studies that demonstrated genetic differentiation between Eastern and Western Pacific samples [ 4 , 5 ]. Skipjack tuna ( Katsuwonus pelamis ) are found in tropical and warm temperate waters of the world's oceans. They are present in the three major oceans in large numbers and comprise approximately 40% of the annual catch of the world's tunas. Annual catches are on the order of 2 million metric tons or approximately 670 million individuals per year [ 1 ]. Despite their huge numbers, the skipjack tuna are not as well studied as most members of the genus Thunnus . Tagging studies have demonstrated limited seasonal movements, but not much transoceanic movement. Thus, they probably do not spawn at discrete locations [ 6 ]. In the Pacific, genetic studies using isozymes have demonstrated an East-West cline in a serum esterase allele [ 7 - 9 ]. Fujino et al. [ 10 ] also demonstrated differences in esterase allele frequencies in samples from the Atlantic, Indian, and Pacific Oceans. However, small samples of Atlantic (n = 7) and Pacific (n = 9) skipjack appeared identical when mtDNA was examined [ 11 ]. After the formation of the Isthmus of Panama, the potential contact between Atlantic and Indo-Pacific populations was limited to the waters around southern Africa. It appears that this separation of the Atlantic and Pacific Oceans resulted in significant genetic drift for many large pelagic fishes since these subpopulations are now genetically differentiated. Examples include bigeye tuna [ 12 , 13 ], albacore [ 14 , 15 ], swordfish [ 16 , 17 ], blue marlin [ 18 ] and sailfish [ 19 ]. In addition, even more pronounced differentiation produced species pairs of Atlantic and Pacific bluefin tunas and the Atlantic white marlin and the Pacific striped marlin [ 19 - 22 ]. To measure the degree of genetic differentiation between the Atlantic and Pacific sub-populations of either yellowfin or skipjack tuna, we examined the hypervariable control region I (CR-I) and a segment of a coding region gene of the respective mitochondrial DNAs. Genetic differences were observed between the Atlantic and Pacific yellowfin tuna samples with PCR-RFLP data of the ATCO gene region, but not with CR-I sequence data. In contrast, no differences between the Atlantic and Pacific skipjack tuna samples were detected with either type of data. Information contained in the CR-I reveals very different demographic histories for yellowfin tuna and skipjack tuna. However, very large long-term female effective population sizes (N e ) were estimated for both species, which may explain the observed levels of inter-oceanic genetic partitioning. Results Control region nucleotide sequence analysis A total of 333 bp of the nucleotide sequence of the mitochondrial DNA (mtDNA) control region was determined for 148 yellowfin tuna (Table 1 and GenBank accession numbers AY899520 – AY899681). For the pooled sample of yellowfin tuna, 110 variable sites defined 130 haplotypes ( h = 0.997) and a nucleotide diversity (π) of 3.5%. Diversity indices for each locality sampled were also high, although values of π and h were slightly lower for the NW Atlantic yellowfin sample where five haplotypes were repeated twice. For skipjack tuna, a total of 394 bp was determined for 115 individuals (Table 2 and GenBank accession numbers AY899405 – AY899519). In the pooled skipjack tuna sample, there were 157 variable sites defining 111 haplotypes, resulting in a very high value of haplotypic diversity ( h = 0.999). Nucleotide diversity (π = 8.4%) was more than twice as high in skipjack tuna as in yellowfin tuna. All of the sampling localities of skipjack tuna had very high diversity values (π >7.7%; h > 0.998). For both species, the high haplotypic diversity values are consistent with the observed large census population sizes (N c ). Phylogenetic analyses resulted in very different trees for each species. The CR-I gene-tree topology of skipjack tuna is much larger and better structured than the yellowfin tuna phylogeny, and contains multiple branches with high bootstrap proportion support (Figure 1 ). However, there is no obvious phylogeographic association in either species, with CR-I lineages from different basins scattered throughout the phylogenetic trees. As a consequence, in both yellowfin tuna (Table 3 ) and skipjack tuna (Table 4 ), the majority of the genetic variation corresponds to differences between individuals within populations, and only a minor fraction of the variation corresponds to differences among-groups. Thus, in both species, the control region sequences provide no evidence of genetic differentiation between Atlantic and Indo-Pacific sub-populations. Table 1 Genetic diversity indices and demographic parameters of yellowfin tuna CR-I. N number of individuals; M, number of haplotypes; π , nucleotide diversity; h , haplotypic diversity. Tajima's D neutrality test and associated probability in parentheses. Mismatch distribution parameters τ, Θ O, Θ 1. Population N M π h Tajima's D τ Θ O Θ 1 All 148 130 0.035 ± 0.018 0.997 ± 0.001 -1.588 (0.022) 8.516 0.047 566 NW Atlantic 31 26 0.027 ± 0.014 0.987 ± 0.012 -0.908 (0.185) 9.746 0.000 87 Ivory Coast 32 29 0.029 ± 0.016 0.994 ± 0.009 -1.254 (0.085) 8.531 0.008 1097 Pacific Ocean 41 40 0.033 ± 0.017 0.999 ± 0.006 -1.155 (0.111) 9.250 0.000 6655 Indian Ocean 44 41 0.032 ± 0.017 0.997 ± 0.006 -1.346 (0.059) 8.391 0.708 352 Table 2 Genetic diversity indices and demographic parameters of skipjack tuna CR-I. Abbreviations and notations as in Table 1. Population N M π h Tajima's D τ Θ O Θ 1 All 115 111 0.084 ± 0.041 0.999 ± 0.001 -0.419 (0.424) 15.82 12.34 3795 NW Atlantic 31 30 0.082 ± 0.041 0.998 ± 0.009 -0.343 (0.408) 15.32 13.50 6655 Brazil 17 17 0.084 ± 0.043 1.000 ± 0.020 -0.354 (0.431) 28.38 4.03 98 E. Pacific Ocean 32 31 0.077 ± 0.038 0.998 ± 0.008 -0.084 (0.537) 14.00 16.10 6655 Solomon Islands 35 34 0.083 ± 0.041 0.998 ± 0.007 -0.266 (0.445) 14.18 16.75 4683 Table 3 AMOVA of the patterns of sequence variation contained in the CR-I of yellowfin tuna. Localities were assigned into three regional groupings: Atlantic, Pacific and Indian. The Atlantic region included two samples, NW Atlantic and Ivory Coast, whereas the Pacific and Indian regions included only one sample each. Source of Variation Variance components Percentage variation Fixation Indices Probabilities Among groups 0.00155 Va 0.31 Φ CT : 0.003 0.17 ± 0.01 Among populations within groups -0.00041 Vb -0.08 Φ SC : -0.001 0.51 ± 0.01 Within Populations 0.49818 Vc 99.77 Φ ST : 0.002 0.038 ± 0.006 Table 4 AMOVA of the patterns of sequence variation contained in the CR-I of skipjack tuna. Localities were assigned into three regional groupings: Atlantic, Pacific and Indian. The Atlantic region included two samples, NW Atlantic and Brazil, whereas the eastern Pacific and Indian Ocean (Solomon Islands) regions included only one sample each. Source of Variation Variance components Percentage variation Fixation Indices Probabilities Among groups 0.296 Va 1.89 Φ CT : 0.019 0.16 ± 0.00 Among populations within groups 0.349 Vb -2.23 Φ SC : -0.023 0.72 ± 0.01 Within Populations 15.712 Vc 100.34 Φ ST : -0.003 0.72 ± 0.01 Figure 1 Unrooted neighbor-joining (NJ) trees showing the relationship of 111 yellowfin tuna and 130 skipjack tuna haplotypes estimated from a matrix of Tamura Nei (α = 0.5) distances. Values of bootstrap proportion support above 60% are included and the geographical origin of haplotypes is identified with symbols (see inset). The bar indicates the size of a line representing differences of 2% with both trees drawn to the same scale. RFLP analysis of the mitochondrial ATCO gene region We hypothesized that the high mutation rate of the control region might have resulted in homoplasy and high levels of haplotypic diversity that masked the actual genetic divergence of Atlantic and Pacific mtDNAs in these species. Analysis of a gene that has a lower mutation rate than that of the control region has been used to reveal genetic divergence in bigeye tuna [ 13 ]. Therefore, we performed a nucleotide sequence analysis on the yellowfin tuna mitochondrial ATCO gene region since variation had been shown previously for this gene [ 23 , 24 ]. We found mutations at two positions that resulted in the loss of restriction sites for the enzymes Dde I and Hpy CH4III, respectively. When amplified Atlantic and Indo-Pacific yellowfin ATCO DNAs were analyzed with these two enzymes, most DNAs were cut with both enzymes (Table 5 ). However, DNAs that had lost either one of the two restriction sites were more common in the Atlantic sample than in the Indo-Pacific sample. When the haplotype distributions were compared, the differences were found to be significant with 7% of the variation occurring between samples (Table 6 ). Thus, low levels of genetic differentiation have occurred between the two yellowfin sub-populations. When the haplotypes were mapped to a neighbour-joining tree of the control region sequences, the mutations affecting the restriction sites were clustered, suggesting that each mutation had occurred once prior to separation of the two sub-populations (data not shown). This result is consistent with a slower rate of mutation in the ATCO gene region compared to that of the control region. Table 5 Allele frequencies at the yellowfin tuna mtDNA ATPase6 locus Sample N ATCO Haplotypes D H DH Atlantic 138 0.13 0.22 0.65 Pacific 96 0.02 0.12 0.86 Haplotype designations: D, cut with Dde I; H, cut with Hpy CH4III; DH, cut with both enzymes. N is the sample size. Table 6 Analysis of molecular variance between Atlantic and Pacific yellowfin tuna samples Source of Variation d. f. Sum of Squares Variance Components Percentage of Variation Between Populations 1 1.19 0.015 Va 7.0 Within Populations 232 46.69 0.201 Vb 93.0 Totals 233 48.60 0.216 Fixation Index F ST : 0.070 Va and F ST : P (random value > = observed value) = 0.00098+/-0.00098 Similar experiments were performed with the skipjack tuna samples. Two variable sites had been observed in the mitochondrial cytochrome b ( cytB ) gene by Terol et al. [ 25 ]. Since one of these sites could not be assayed with restriction enzymes, we determined the nucleotide sequence of this portion of the cytB gene for each DNA in the Atlantic and Pacific skipjack tuna samples. Little additional variation was observed, and allele frequencies were similar in the two samples (Table 7 ). When the haplotype distributions were compared, all of the variation occurred within samples. Thus, there is no evidence for genetic differentiation between the Atlantic and Pacific sub-populations of skipjack tuna. Table 7 Allele frequencies at the skipjack mtDNA cytB locus Allele* Region n AT AC GC Pacific 80 0.31 0.24 0.45 Atlantic 49 0.31 0.18 0.51 *Variable base at the Psh A1 site and a position 19 bases upstream as determined by nucleotide sequence analysis. Demographic history and effective population size The unimodal mismatch distribution for the pooled sample of yellowfin tuna (Figure 2 ) and a significant Tajima D test (Table 1 ) both suggest the historical expansion of this population. In sharp contrast, the skipjack mismatch distribution is bimodal (Figure 2 ) and the neutrality test is not significant (Table 2 ), suggesting that the effective size of the skipjack population has been large and stable for a long period. In fact, the value of θ 0 suggests that the number of females in the skipjack population was originally large (N e0 = 320,000) whereas the effective number of yellowfin tuna females prior to expansion was small (N e0 = 823). However, the values of θ 1 suggest large long-term numbers of effective female breeders (N e1 ) of about 98 million for skipjack tuna and about 10 million for yellowfin tuna. Figure 2 Mismatch distributions for the entire sample of a) yellowfin tuna and b) skipjack tuna. The solid bars in the histograms represent the observed pairwise differences between haplotypes and the curves the expected distribution under the sudden expansion model. Discussion Compared to other scombroid species with cosmopolitan distributions, we found substantially less genetic differentiation between Atlantic and Pacific sub-populations of yellowfin tuna and no inter-oceanic genetic differentiation of skipjack tuna. For instance, in bigeye tuna, there are two highly divergent mtDNA clades, one is cosmopolitan and the other is endemic to the Atlantic [ 12 , 13 ]. This asymmetric distribution of Atlantic and Pacific clades has been observed in other large pelagic fishes as well, including swordfish [ 16 , 17 ], blue marlin [ 18 ], and sailfish [ 19 , 26 ]. Similarly, the striped marlin (Pacific) and white marlin (Atlantic) and the Atlantic bluefin tuna and the Pacific bluefin tuna are considered to be pairs of sister species [ 19 - 21 , 26 ]. Thus, significant differentiation has occurred between the Atlantic and Pacific populations of many large pelagic species. Why are skipjack and yellowfin tuna populations different? One possibility is that in contrast to other large pelagic species, sufficient gene flow occurs to prevent inter-oceanic genetic differentiation. However, this explanation is contrary to the distribution patterns of tunas and billfish [ 27 , 28 ]. For instance, bigeye tuna has a distribution of catches that would suggest a population continuum from the Indian Ocean to South Atlantic waters along the east and west coasts of Africa. However, bigeye samples show a marked inter-oceanic differentiation [ 12 , 13 ]. In the Indian Ocean, the presence of skipjack and yellowfin tuna south of 20° S, is confined to the warm waters associated with the Agulhas currents. Seasonal movements do occur around the Cape of Good Hope via the Agulhas current, but it appears that the migrant tuna and other pelagic species return to the Indian Ocean as the seasons change [ 13 , 29 ]. In fact, the distribution patterns of these two species in the South Atlantic along the African coast of are remarkably similar to both blue marlin and sailfish, two species that show a pronounced differentiation between Indo-Pacific and Atlantic populations. It should be noted, however, that the presence of "Pacific" mitochondrial DNA clades in Atlantic subpopulations of many tuna species indicates that inter-oceanic migration has occurred in the past. However, contemporary levels of inter-oceanic genetic differentiation for these species indicate that current levels of gene flow are absent or severely reduced. Thus, current distribution patterns of skipjack tuna and yellowfin tuna are not consistent with substantial levels of inter-oceanic gene flow. Certainly, there is no evidence to suggest that gene flow is occurring or has occurred at levels higher in these two species than in any of the other pelagic species where interoceanic differentiation has been demonstrated. An alternative explanation to account for the lack of inter-oceanic differentiation in skipjack tuna and yellowfin tuna, is that the time since population expansion began has not been sufficient to allow for the populations to become differentiated. Assuming the very conservative mutation rate for CR-I of 4.9% per million years, a generation time 3.5 years, and the tau value 8.52, expansion of yellowfin tuna occurred about 522 Ky ago. By comparison, the estimated time for genetic differentiation of the Atlantic and Pacific populations of swordfish and of the Atlantic and Pacific bluefin tunas, using the same mutation rate, is very similar (450–470 Ky), and an even shorter time (170 Ky) is sufficient to explain the substantial genetic differentiation between the Atlantic and Mediterranean swordfish populations [ 30 ]. Thus, time since expansion cannot explain the absence of inter-oceanic differentiation in these species. Therefore, we propose that much larger effective population sizes are the primary factor responsible for the similarity of the Atlantic and Pacific sub-populations of skipjack and yellowfin tuna. Among large pelagic species, the female effective population size can be estimated from the demographic estimates obtained from the CR-I sequences of Atlantic bluefin tuna and swordfish. Assuming a mutation rate of 4.9% per million years and a generation time of 6 years, the female N e estimates are 900,000 for bluefin tuna and 800,000 for swordfish [ 30 ]. Effective population sizes have not been estimated for other species of tunas, but annual harvest data are generally used as a proxy for abundance since the commercial harvest of large fish has become quite efficient throughout the world's oceans and harvest restrictions have only incremental impact on the total harvest of any species. Furthermore, haplotypic diversities of tuna mitochondrial DNAs are approximately 99% in all tuna species [ 12 , 31 ], suggesting that female reproductive variance is small. Accordingly, census population size can be expected to be proportional to effective population size in tuna species. The estimated number of skipjack and yellowfin caught in the year 2002 is 670 million and 130 million, respectively (Table 8 ). The catch of all other tunas was 24 million individuals or less. Thus, the abundance of skipjack tuna is approximately 5 times that of yellowfin, and more than 300 times that of Atlantic bluefin tuna and swordfish. These estimates of abundance from fisheries data appear to correspond well with the estimated female N ef values. The observed correspondence does not take into consideration dramatic changes in abundance of some of these species over the last 20 years, nor sex ratio differences, or the age distribution of the catch (e.g., number of mature females). However, the comparison supports the hypothesis that skipjack tuna and yellowfin tuna, which are the two most abundant species, also have the largest effective population sizes, and the lowest amounts of genetic partitioning compared to other scombroid fishes. Table 8 Worldwide Tuna Catch Data for the Year 2002 [1] Species Catch (metric tons) Ave. Wt. (kg) Est. Number Fish Harvested* skipjack 2.0 × 10 6 3 670 × 10 6 yellowfin 1.3 × 10 6 5–20 130 × 10 6 bigeye 0.43 × 10 6 15–20 24 × 10 6 albacore 0.24 × 10 6 9–20 16 × 10 6 longtail 0.125 × 10 6 15–20 6.9 × 10 6 Pacific bluefin 0.024 × 10 6 7 3.4 × 10 6 Atlantic bluefin 0.036 × 10 6 17 2.1 × 10 6 *Number of Atlantic bluefin tuna harvested obtained from ICCAT [51]. Bigeye tuna, and bluefin tuna have much greater levels of genetic differentiation between Atlantic and Pacific subpopulations when mitochondrial DNA control region sequences are compared [ 12 , 31 ]. However, their population sizes are 6 and 15 times lower than that of yellowfin and 30 and 75 times lower than that of skipjack, respectively. Furthermore, since bluefin tuna are a more temperate species, their effective population sizes may have been significantly lower than those observed in recent times whenever the northern hemisphere experienced glacial maxima. Thus, the relative differentiation of the Atlantic and Pacific subpopulations is consistent with the demographics of these species of tuna. One exception to the patterns described above is albacore tuna. The distribution of albacore tuna mitochondrial DNAs does not appear to fit the patterns described above for the other temperate tuna species. Albacore abundance is similar to that of bigeye tuna and both species have two mtDNA clades. However, in albacore the two clades are not as well differentiated nor do they display the phylogeographic association observed for the bigeye tuna clades [ 12 ]. Instead, the bimodal mismatch distribution of pairwise differences in albacore mtDNA is concordant with very large long-term effective population sizes in contrast to the contemporary population size. Conclusion Much lower levels of genetic differentiation were found among sub-populations of yellowfin tuna compared to those observed for other large tunas, probably due to the large population size of yellowfin tuna. Since skipjack tuna appear to have even larger population sizes than yellowfin tuna, it is not surprising that no genetic differentiation was observed between Atlantic and Pacific samples of these fish. Methods Samples of yellowfin tuna were obtained from the eastern Pacific Ocean (near the equator at 110° W; n = 41), the Indian Ocean (n = 63), the Gulf of Mexico and the East Coast of Florida (n = 38) and the Gulf of Guinea (n = 100). Samples of skipjack tuna were obtained from the Northwest Atlantic (n = 31), off the coast of Brazil (n = 19), the eastern Pacific Ocean (n = 43), and the south Pacific near the Solomon Islands (n = 37). DNA isolation, mitochondrial DNA D-loop region amplification, and nucleotide sequence analyses have been described previously [ 12 , 32 ]. The number of segregating sites (S) was estimated with MEGA. Values of haplotypic diversity (h) [ 33 ], nucleotide diversity (π) [ 34 ] and the mean number of pairwise differences (K) were computed in ARLEQUIN ver. 2.0 [ 35 ]. Mitochondrial DNA haplotype phylogenies were estimated using neighbour-joining analyses [ 36 ] with Tamura-Nei distances (α = 0.5)) in MEGA [ 37 ]. The pair-deletion option was used when missing data, insertions, or deletions were present. Maximum-Parsimony (MP) [ 38 , 39 ] was carried using heuristic searches with the default options in PAUP* 4.0b10 [ 40 ]. Statistical support for the nodes was estimated with 1000 non-parametric bootstrap replicates [ 41 ]. All trees were rooted at midpoint. Analyses of molecular variance (AMOVA) [ 42 ] were performed to estimate the partitioning of genetic variation in regional hierarchical arrangements using ARLEQUIN. The demographic history contained in the mtDNA CR-I sequence data was inferred using two approaches. First, the null hypothesis of neutrality may be rejected when a population has experienced population expansion [ 43 ]. Accordingly, Tajima's D test of neutrality [ 43 , 44 ] and its significance levels were estimated using DnaSP 4.00 [ 45 ] based on 1000 simulated re-samplings replicates. Alternatively, a population that has experienced a rapid expansion in the recent past shows smooth wave-like mismatch distribution [ 46 , 47 ]. Thus, mismatch distribution analyses, under the assumption of selective neutrality, were also used to evaluate possible historical events of population growth and decline [ 47 , 48 ]. Past demographic parameters, including τ [ 49 ], θ 0 and θ 1 and their probabilities [ 47 ] were estimated in ARLEQUIN taking into account the heterogeneity of mutation rates [ 35 ]. For the analysis of the mitochondrial cytochrome b gene, a 650 bp fragment of the skipjack tuna cytochrome b gene was amplified using primers CB3 (GGCAAATAGGAARTATCATTC) and GLUDG (TGACTTGAARAACCAYCGTTG) [ 50 ]. Alleles were identified by determining the nucleotide sequence of the amplified fragment. For yellowfin tuna, the ATCO gene region was amplified using primers H9342 (GCCATATCGTAGCCCTTTTTG) and L8562 (CTTCGACCAATTTATGAGCCC) [ 24 ]. The amplified fragments were digested with either Dde I or Hpy CH4III and the digestion products were resolved by electrophoresis in a 1.2% agarose gel. Authors' contributions BE, JV, and JAB conceived the study, supervised the genetic studies, analyzed the data and wrote the manuscript. JAB, AL, and ET performed the DNA sequence analyses. BE, DB, ET, LL, and KC performed the DNA amplifications and RFLP analyses. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554763.xml |
547910 | Toxicity after prolonged (more than four weeks) administration of intravenous colistin | Background The intravenous use of polymyxins has been considered to be associated with considerable nephrotoxicity and neurotoxicity. For this reason, the systemic administration of polymyxins had been abandoned for about 20 years in most areas of the world. However, the problem of infections due to multidrug-resistant (MDR) Gram-negative bacteria such us Pseudomonas aeruginosa and Acinetobacter baumanniii has led to the re-use of polymyxins. Our objective was to study the toxicity of prolonged intravenous administration of colistin (polymyxin E). Methods An observational study of a retrospective cohort at "Henry Dunant" Hospital, a 450-bed tertiary care center in Athens, Greece, was undertaken. Patients who received intravenous colistin for more than 4 weeks for the treatment of multidrug resistant Gram-negative infections were included in the study. Serum creatinine, blood urea, liver function tests, symptoms and signs of neurotoxicity were the main outcomes studied. Results We analyzed data for 19 courses of prolonged intravenous colistin [mean duration of administration (± SD) 43.4 (± 14.6) days, mean daily dosage (± SD) 4.4 (± 2.1) million IU, mean cumulative dosage (± SD) 190.4 (± 91.0) million IU] in 17 patients. The median creatinine value increased by 0.25 mg/dl during the treatment compared to the baseline (p < 0.001) but returned close to the baseline at the end of treatment (higher by 0.1 mg/dl, p = 0.67). No apnea or other evidence of neuromuscular blockade was noted in any of these patients who received prolonged treatment with colistin. Conclusions No serious toxicity was observed in this group of patients who received prolonged intravenous colistin. Colistin should be considered as a therapeutic option in patients with infections due to multidrug resistant Gram-negative bacteria. | Background The worldwide spread of the problem of infections due to Gram-negative bacteria such us Pseudomonas aeruginosa and Acinetobacter baumannii resistant to most classes of antimicrobial agents has made the medical community to rethink of colistin, an old, basically abandoned for the last two decades antibiotic. Intravenous colistin (polymyxin E) has been used in the industrialized countries during the last years mainly in patients with cystic fibrosis infected with Pseudomonas aeruginosa resistant to other available antimicrobial classes [ 1 ]. Excluding this population, the agent has been infrequently used during the last two decades because of major concerns about nephrotoxicity and neurotoxicity [ 2 ]. Several reports during the early years of use of the medication, mainly in the decade 1960 to 1969 left the medical community with the impression that the medication is very toxic [ 3 , 4 ]. However, the recent experience of several clinicians worldwide with the use of the medication has not verified the old reports about the serious or common toxicity of colistin [ 5 - 8 ]. We analysed data of a group of patients who received colistin for more than four weeks to investigate further the concerns about the toxicity of the medication. Methods Design of the study-patient selection Patients who received intravenous colistin from 1/October/2000 to 31/January/2004 at "Henry Dunant" Hospital, a 450-bed tertiary care center in Athens, Greece, were identified by the pharmacy electronic databases. All patients who were found to have received intravenous colistin for more than four weeks continuously, or in two courses of colistin but with no more than seven days interruption between the courses were included in this study. Administration of intravenous colistin in two courses for more than four weeks in each course were studied as two different units in the analysis of toxicity, if there was more than one month colistin-free period between the two courses. One milligram of the colistin formulations used is approximately equal to 12.500 IU (Colomycin, Forest Laboratories ® , Kent, UK or Colistin, Norma ® , Athens, Greece). The study was approved by the Institutional Review Board (IRB) of the hospital. Definitions of infections Diagnosis of pneumonia required two or more serial chest radiographs with at least one of the following: new or progressive and persistent infiltrate, consolidation, cavitation, or pleural effusion. In addition, patients must have had fever >38°C with no other recognized cause, or abnormal white blood cell count [leukopenia (< 4000 WBC/mm 3 ) or leukocytosis (= 12.000 WBC/mm 3 )], and at least two of the following: new onset of purulent sputum or change in character of sputum, increased respiratory secretions or increased suctioning requirements, new onset or worsening of cough or dyspnea or tachypnea, rales or bronchial breath sounds, or worsening gas exchange [ 9 ]. Bacteremia required either growth of a recognized pathogen from one or more blood specimen cultures or at least one of the following signs or symptoms: fever (>38°C), chills, or hypotension and at least one of the following: a) common skin contaminant (e.g., diphtheroids, Bacillus sp., Propionibacterium sp., coagulasenegative staphylococci, or micrococci) grown from two or more blood cultures drawn on separate occasions or b) common skin contaminant (e.g., diphtheroids, Bacillus sp., Propionibactαium sp., coagulase-negative staphylococci, or micrococci) grown from at least one blood culture from a patient with an intravascular line and physician- instituted antimicrobial therapy [ 9 ]. Infections at other body sites or fluids, such as urinary tract infections and surgical site infections were defined based on guidelines from the Centers for Disease Control and Prevention [ 9 ]. Clinical specimens of all body sites were considered when defining the etiologic microbiologic agent of the infection. Data collection-entry Data for several variables including demographic and clinical information as well as results of laboratory and imaging tests were collected and entered in a computer database. All available results of renal function tests (serum creatinine, urea, creatinine clearance, urinalysis), liver function tests (SGPT, SGOT, alkaline phosphatase, γ-GT, bilirubin), creatine phosphokinase (CPK) during the course of colistin treatment and at hospital discharge were collected. Data analysis Creatinine values at the beginning of colistin treatment were compared with the maximum value of creatinine during therapy as well as with the creatinine value at the end of treatment using a non-parametric test (Wilcoxon). The data analysis was performed using SPSS and S-plus software. Results Study population During the period from 1/October/2000 to 31/January/2004, 152 patients received intravenous colistin. Data about the efficacy of the medication were analysed and reported elsewhere [ 5 ]. In this report, we present detailed data regarding the toxicity of colistin in a subgroup of 17 patients who received more than four weeks of intravenous colistin. Table 1 describes various characteristics of this group of patients including demographic and clinical data, such as comorbidity, site of infection, and responsible pathogens. Two patients received two prolonged courses of intravenous colistin each with more than one month colistin-free period between the courses. Subsequently, there were 19 courses of prolonged colistin for the analysis of toxicity. Among these 19 courses, the administration of intravenous colistin was without interruption in 15 courses, while there was interruption of colistin administration in 2 courses for two days, interruption in 1 course for 3 days, and interruption in 1 course for 4 days. The cause for the interruption of colistin in these cases was the attending physicians' attempts to obtain cultures without the confounding influence of antimicrobial treatment in patients with puzzling continuing symptoms of infection. The mean duration of colistin administration (± SD), for each course of colistin therapy was 43,4 days (± 14,6 days). The mean daily dosage ± SD for each course was 4.4 million IU (352 mg) ± 2.1 million IU (168 mg) and the mean cumulative dosage of colistin ± SD for each course was 190.4 million IU (15.232 mg) ± 91.0 million IU (7.280 mg). The mean daily dosage for a patient with body weight of 70 kg was 62.857 IU/kg (5 mg/kg). For patients with impaired renal function, dosage adjustments were done mainly after consulting the ICU director or the Infectious Disease specialists of the hospital, based on the following protocol: if serum creatinine level was 1.3–1.5 mg/dl, 1.6–2.5 mg/dl or = 2.6 mg/dl, the dosage of colistin administered was 2 million IU (160 mg) every 12 hours, 24 hours, or 36 hours, respectively. Patients who were on dialysis treatment received 1 million IU (80 mg) of colistin after dialysis. Colistin was given for long duration in patients mainly in the ICU setting, because of persistence of infections due to Gram-negative bacteria sensitive only to colistin or bacteria sensitive also to antibiotics of other classes that were previously administered (based on in-vitro susceptibility test results) but failed. Subsequently, no other therapeutic strategy was considered better than the continuation of colistin (as combination therapy or monotherapy) because of failure of previous antibiotic regimens and the lack of other therapeutic options. Table 1 Demographic and clinical features of patients managed with prolonged intravenous colistin for infections caused by multidrug-resistant Gram-negative bacteria. Characteristics of patients Median (range) n (%) Demographic Age, years [median (range)] 51 (18 – 79) Sex, male 12/17 (70%) APACHE II score On admission to ICU [median (range)] 14 (7 – 35) On 1 st day of colistin treatment [median (range)] 14 (6 – 22) Comorbidity Malignancy 2/17 (11%) Hemodialysis 2/17 (11%) Urogenital disorders 3/17 (18%) Heart dysfunction 5/17 (29%) Diabetes mellitus 5/17 (29%) Lung dysfunction 5/17 (29%) Liver failure 1/17 (6%) Hematological disorders 3/17 (18%) Neurological disorders 8/17 (47%) Neuropathy/myopathy 4/17 (23%) Trauma 7/17 (41%) Transfusion 15/17 (88%) Prior hospitalization 12/17 (70%) Prior surgery 14/17 (82%) Elective 9/14 (64%) Emergency 5/14 (36%) Prosthetic material 11/17 (64%) Catheters/Invasive devices Tracheostomy 14/17 (82%) CSF drainage 5/17 (29%) Prior medications Prior antibiotic use 17/17 (100%) Prior antifungal use 8/17 (47%) Anti-tumor treatment 0/17 (0%) Cortisone treatment 9/17 (53%) Duration of hospitalization (days) [median (range)] 152 (29 – 591) Duration of stay in ICU (days) [median (range)] 70 (22 – 134) Site of infection* Pneumonia 13/19 (68%) Bacteremia 1/19 (5%) Urinary tract infection 2/19 (11%) Meningitis 2/19 (11%) Surgical site infection 1/19 (5%) Pathogens† Pseudomonas aeruginosa 12/20 (60%) Acinetobacter baumannii 5/20 (25%) Klebsiella pneumoniae 2/20 (10%) Enterobacter cloacae 1/20 (5%) Mortality 7/17 (41%) * The analysis of the site of infection was based on 19 episodes of infection in 17 patients. † The analysis of the responsible pathogens was based on 20 isolates from 19 episodes of infection in 17 patients [two pathogens were isolated in one episode of pneumonia ( Klebsiella pneumoniae + Acinetobacter baumannii mixed infection)]. The analysis of the responsible pathogens was based on 20 isolates from 19 episodes of infection in 17 patients (two pathogens were isolated in one episode). In 2 episodes of infection, both caused by Pseudomonas aeruginosa strains, the minimum inhibitory concentrations (MICs) were not available. In the remaining 10 Pseudomonas aeruginosa isolates, the MICs to colistin ranged from = 0.5 mg/l to 2 mg/l. In vitro susceptibility testing for the 5 isolated Acinetobacter baumannii strains revealed MICs to colistin = 0.5 mg/l, for the 2 Klebsiella pneumoniae strains MICs were 0.5 mg/l and 2 mg/l, and for the Enterobacter cloacae strain MIC was 1 mg/l. The all-cause in-hospital mortality was 41,2% (7 out of 17 patients). Clinical cure and improvement, defined as resolution and partial resolution, respectively, of presenting symptoms and signs of the infection by the end of colistin treatment was observed in 14 of the 19 episodes of infection (73.7%) [cure 10/19 episodes (52.6%), improvement 4/19 episodes (21.1%)]. Unresponsiveness, defined as persistence or worsening of presenting symptoms and/or signs of the infection during intravenous colistin administration, was observed in 5 of the 19 episodes of infection (26.3%). All patients had been admitted to the ICU at some time during their hospitalisation. In 12 courses out of the 19 analyzed the whole colistin therapy was administered in the ICU, in 1 course it was administered in a hospital ward, and in 6 courses it was given both in the ICU and a hospital ward. Renal toxicity Figure 1 depicts the box plots of serum creatinine and blood urea values of patients on the first day of colistin, the maximum values during the course of administration of colistin, and the values at the end of the treatment with colistin (one patient, who was on hemodialysis treatment prior to and during the colistin course, was removed from the analysis of the comparisons of the creatinine and urea values). The median value of baseline creatinine at the beginning of colistin therapy for the 18 prolonged courses of colistin was 0,6 mg/dl. Compared to this value, there was a slight increase of the median of values of creatinine at the end of the colistin course by 0,1 mg/dl, without statistical significance (p = 0,67). The median of the maximum values of creatinine observed was 0,85 mg/dl and was found to be statistically different from both the median of start (p < 0,001) and end of treatment values (p < 0,001). The maximum absolute increase of creatinine, compared to baseline, observed in an episode was 1,4 mg/dl. Only one patient had an increase of more than 50% of the baseline creatinine level to a value higher than 1.3 mg/dl at the end of colistin treatment. For the subgroup of 12 patients with 14 courses of prolonged colistin administration, with normal serum creatinine at the initiation of treatment, the median serum creatinine values at baseline and at the end of colistin treatment were 0,60 mg/dl and 0,55 mg/dl, respectively. In addition, the median maximum value of serum creatinine during the course of colistin treatment was 0,80 mg/dl among this group of patients. Figure 1 The distribution of serum creatinine and blood urea levels on the 1 st day of colistin treatment (start), at the peak value (max), and the end of colistin treatment (end) in all studied courses of prolonged treatment with colistin. (The horizontal line within the boxes represents the median creatinine or blood urea baseline value at the 1 st day of colistin treatment. "Shaded" boxes in the figures represent the distribution of the laboratory values contained between the 25 th and 75 th percentiles (the interquartile range). Dotted lines (whisker lines) extend from the box, down and up, to the minimum and maximum values of the distributions that are not outliers (i.e. that are within 1.5 times the interquartile range); these values of the distribution are shown as "brackets". Any value, which is an outlier, is drawn as a "horizontal line"). At the initiation of colistin administration the median value of baseline blood urea for the 18 courses was 42 mg/dl. During the 18 prolonged courses of colistin administration there was a slight increase in blood urea values; the median maximum value was 60 mg/dl. However, at the completion of colistin therapy median blood urea value returned to 41 mg/dl. During the 19 courses of colistin administration, other potentially nephrotoxic drugs were given. Specifically, aminoglycosides were co-administered in 12 courses [median duration of co-administration (± SD) in these 12 courses 12,5 (± 16,8) days], teicoplanin in 10 [median duration of co-administration 16,0 (± 9,0) days], vancomycin in 8 [median duration of co-administration 23,0 (± 11,2) days], amphotericin in 4 [median duration of co-administration 8,0 (± 9,0) days]. Four patients had history of chronic renal dysfunction, with only one of them receiving hemodialysis treatment prior to admission. This patient continued the hemodialysis during his treatment with colistin and died due to sepsis. For the other 3 patients the baseline, the maximum, and the end-of-therapy creatinine values (mg/dl), were 1,8/2,1/2,1 for the first patient, 3,7/3,7/2,4 for the second patient and 2,4/3,8/2,3 for the third patient. Only one patient had acute renal failure prior to colistin use, which was due to severe haemorrhage during surgery. This patient received 13 courses of hemodialysis. One day after the last course, colistin therapy had to be initiated for the management of Acinetobacter baumannii pneumonia, sensitive only to colistin and gentamicin. Apart from a slight elevation of serum creatinine (from 2,3 mg/dl to 2,7 mg/dl) on the 5 th day of treatment no other sign of nephrotoxicity was observed during a 34-day administration of colistin (the 9 th day creatinine returned to 0,7 mg/dl). Neuromuscular toxicity No apnea or other evidence of neuromuscular blockade was noted in any of these patients who received prolonged treatment with colistin. Four patients were clinically diagnosed to have polyneuropathy and/or myopathy before or during colistin treatment. In the first patient the polyneuropathy symptoms appeared while she was on her 25 th day of colistin treatment and became worse on the 27 th day. From then on, and although colistin was continued for 11 more days, the symptoms gradually subsided. No confirmatory electromyographic testing was performed. The second patient was transferred from another ICU where he was diagnosed as having myopathy due to sepsis. Fifteen days after his admission, and while his myopathy was improving, therapy with colistin was initiated; the treatment did not affect the gradual improvement. An electromyography performed on his 21 st day in our ICU, showed mild axial sensory-motor polyneuropathy and myopathy. He received colistin for a total of 52 days and made full recovery from his neuropathy/myopathy during the course of colistin administration. The third patient developed ICU polyneuropathy one week prior to colistin administration and, again, recovered gradually from it while he was on colistin treatment. The fourth patient was transferred from another ICU, where he had developed ICU polyneuropathy. Only this patient had moderate worsening of the neuropathy while he was receiving colistin. Despite that, he received colistin for a total of 35 days for a respiratory infection with persisting Pseudomonas aeruginosa sensitive only to colistin. His neuropathy improved gradually after the end of colistin treatment. Thus, our assessment regarding these 4 patients who experienced polyneuropathy and/or myopathy is that colistin therapy was probably associated with the development of neurotoxicity only in one of the 4 patients. Liver and biliary tree toxicity Data for the possible hepatobiliary toxicity of colistin were collected. Both laboratory and clinical findings were taken into consideration. In subgroup analyses of patients for whom data were available, no substantial changes on liver function tests was found, as described in Figure 2 (data for SGPT, alkaline phosphatase, direct and indirect bilirubin are not shown). Three of our patients were found to have increased levels of hepatic and cholestatic enzymes during colistin administration. One of them was found to have acute cholecystitis, the second had severe inflammatory systemic reaction (SIRS), and the third had transient elevation of transaminases while he was receiving anti-epileptic medications together with colistin. Figure 2 The distribution of liver function tests [SGOT (AST = aspartate aminotransferase), γ-GT, and total bilirubin] on the 1 st day of colistin treatment (start), at the peak value (max), and at the end of colistin treatment (end) in studied courses of prolonged treatment with colistin. (The horizontal line within the boxes represents the median creatinine or blood urea baseline value at the 1 st day of colistin treatment. "Shaded" boxes in the figures represent the distribution of the laboratory values contained between the 25 th and 75 th percentiles (the interquartile range). Dotted lines (whisker lines) extend from the box, down and up, to the minimum and maximum values of the distributions that are not outliers (i.e. that are within 1.5 times the interquartile range); these values of the distribution are shown as "brackets". Any value, which is an outlier, is drawn as a "horizontal line"). Conclusions The main finding of the study is that no significant deterioration of renal function was found in patients with baseline normal serum creatinine value during prolonged administration of intravenous colistin. In addition, in the group of patients with pre-existing chronic renal dysfunction (not on hemodialysis) the administration of the medication for a long duration did not influence further their renal function. The observed slight increase in the serum creatinine values at the end of treatment (0,1 mg/dl increase in the median creatinine values from the start to end of colistin administration) cannot be attributed only to the prolonged colistin administration as other factors such as concomitantly administered potential nephrotoxic agents, and sepsis per ce may have had an effect. In addition, apnea or other evidence of neuromuscular blockade was not observed in any of the 17 patients who received prolonged treatment with colistin. Early experience with colistin revealed a high incidence of toxicity, mainly nephrotoxicity. In 1969, Ryan et al. reported the fatal case of a previously healthy 4-year-old child with persistent fever after appendectomy who died due to cardiac arrest [ 10 ]. The child had received colistin intramuscularly due to the growth of Pseudomonas aeruginosa and E. coli strains from the excised specimen. This patient developed acute tubular necrosis (confirmed by autopsy) and apnea requiring intubation and mechanical ventilation. However, in this case the dosage of colistin used was 30 mg/kg of body weight every six hours (120 mg/kg/day) whereas the recommended dose is 2,5 – 5 mg/kg/day. One year later, several reports showed high renal adverse reaction rates after colistin administration [ 2 , 4 , 11 ]. In many of them the dosage used was also higher compared to the recommended, i.e. 6,3 mg/kg/day in one report, and 26 mega units in another administered as 10 mega units by intramuscular injection, 10 mega units by intravenous injection and 6 mega units by inhalation (no details are provided in the report about the ratio of IU of colistin per mg). In addition, dosage adjustments in patients with renal dysfunction were not always followed, as well as there were also other factors which perhaps contributed to toxicity, such as the co-administration of other medications with potential toxicity, including anesthetic drugs and muscle relaxants. However, recent data from published reports do not corroborate this finding [ 14 ]. Notably, one study conducted in 35 patients with ventilator-associated pneumonia due to multidrug-resistant Acinetobacter baumannii , who received either intravenous colistin (21 patients) or imipenem (14 patients), showed an almost double increase in nephrotoxicity rates among the patients in the imipenem group compared to colistin group [ 12 ]. In addition, Stein et al. recently reported their experience from 3 patients with orthopedic infections who received colistin for 3 – 6 months without developing any adverse effect requiring discontinuation of the treatment [ 13 ]. The conclusion in the majority of the published studies during 1960 – 1970 was that colistin must be used in serious infections where other antimicrobial agents have failed, always with close monitor of renal function, and with precaution in the co-administration of other possible nephrotoxic or neutotoxic agents [ 2 , 4 ]. Subsequently, the considerable difference regarding reporting toxicity of colistin between old and recent studies deserves an explanation. First, the formulations of currently used colistin may be more purified compared with the old ones. The experience that was accumulated from the use of vancomycin may support this possibility. It is known that the concerns about serious and/or common nephrotoxicity of vancomycin overwhelmed the older literature. Second, fluid supplementation and supportive treatment of severely ill patients has been improved nowadays. Third, higher doses of colistin were administered to patients during the first years of its use. Our study is not without limitations. It is a retrospective study with the inherited problems related to this study design, including difficulties in the determination of colistin associated sensory neurotoxicity. In addition, the number of studied patients is relatively small and there is no control group. However, our study offers insight related to interesting experience from a group of patients who received prolonged treatment with colistin for various infections, mainly as a salvage treatment of persisting and serious infections. In conclusion, intravenous colistin seems to be a safe therapeutic intervention for serious infections due to multidrug-resistant Gram-negative bacteria even when it is administered for a prolonged duration (more than four weeks). More studies will be needed to further clarify the correct use of this old antibiotic in the 21 st century. Competing interests The author(s) declare that they have no competing interests. Authors' contribution MEF and AM conceived the idea; MR, IAB, KR, and SKK collected the data; all authors contributed in the writing and preparation of the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC547910.xml |
524509 | Low dose pramipexole is neuroprotective in the MPTP mouse model of Parkinson's disease, and downregulates the dopamine transporter via the D3 receptor | Background Our aim was to determine if pramipexole, a D 3 preferring agonist, effectively reduced dopamine neuron and fiber loss in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) mouse model when given at intraperitoneal doses corresponding to clinical doses. We also determined whether subchronic treatment with pramipexole regulates dopamine transporter function, thereby reducing intracellular transport of the active metabolite of MPTP, 1-methyl-4-phenylpyridinium (MPP+). Methods Ten 12-month old C57BL/6 mice were treated with MPTP (or saline) twice per day at 20 mg/kg s.c. (4 injections over 48 h ). Mice were pretreated for 3 days and during the 2-day MPTP regimen with pramipexole (0.1 mg/kg/day) or saline. Stereological quantification of dopamine neuron number and optical density measurement of dopamine fiber loss were carried out at 1 week after treatment, using immunostaining for dopamine transporter (DAT) and tyrosine hydroxylase (TH). Additional wild-type (WT) and D 3 receptor knockout (KO) mice were treated for 5 days with pramipexole (0.1 mg/kg/day) or vehicle. The kinetics of [ 3 H]MPP+ and [ 3 H]DA uptake ( V max and K m ) were determined 24 h later; and at 24 h and 14 days dopamine transporter density was measured by quantitative autoradiography. Results Pramipexole treatment completely antagonized the neurotoxic effects of MPTP, as measured by substantia nigra and ventral tegmental area TH-immunoreactive cell counts. MPTP- induced loss of striatal innervation, as measured by DAT-immunoreactivity, was partially prevented by pramipexole, but not with regard to TH-IR. Pramipexole also reduced DAT- immunoreactivity in non-MPTP treated mice. Subchronic treatment with pramipexole lowered the V max for [ 3 H]DA and [ 3 H]MPP + uptake into striatal synaptosomes of WT mice. Pramipexole treatment lowered V max in WT but not D 3 KO mice; however, D 3 KO mice had lower V max for [ 3 H]DA uptake. There was no change in DAT number in WT with pramipexole treatment or D 3 KO mice at 24 h post-treatment, but there was a reduction in WT-pramipexole treated and not in D 3 KO mice at 14 days post-treatment. Conclusion These results suggest that protection occurs at clinically suitable doses of pramipexole. Protection could be due to a reduced amount of MPP + taken up into DA terminals via DAT. D 3 receptor plays an important role in this regulation of transporter uptake and availability. | Background An interesting development in the use of dopamine (DA) agonists for treatment of Parkinson's disease (PD) is that some of them have proven to be neuroprotective in animal models of PD. Antiparkinsonian agents that are direct DA agonists, such as apomorphine [ 1 ], bromocriptine [ 2 ], and pramipexole [ 3 ], are neuroprotective against 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-induced damage to the DA system in mice. Administration of MPTP, which is converted to 1-methyl-4-phenylpyridinium (MPP+) and intracellularly transported into DAergic neurons [ 4 ], provides a good model for studying neuroprotection in PD. MPTP produces Parkinsonism in humans and in subhuman species through selective loss of DAergic neurons of the substantia nigra (SN) [ 5 , 6 ], and a number of related compounds to MPTP also produce nigral cell loss in primates [ 7 ]. MPTP causes apoptosis associated with PD [ 8 - 10 ] ;MPTP produces progressive cell death in humans for decades after the initial insult [ 11 ]. Hence, drugs that reduce the neurotoxicity of compounds like MPTP may be neuroprotective in PD. In fact, it is now hypothesized that direct DA agonists may slow the loss of DAergic terminal function upon long-term administration to PD patients [ 12 - 15 ]. Dopaminergic neurons are tonically inhibited by dendritic and terminal autoreceptors, operating in interaction with DA transporters (DAT) and pharmacologically of the D 2 receptor subtype [ 16 - 19 ]. However, Zapata et al [ 20 ] have reported that the D 3 preferring agonist (+)-PD 128907 regulates extracellular DA levels via interactions with D 3 autoreceptors. If D 3 preferring agonists are potent autoreceptor agonists, then hypothetically long-term changes in expression of DAT or the functional properties of DAT might occur following subchronic treatment. Since intracellular accumulation of MPP+ following systemic injection of MPTP requires DAT [ 4 ], then when DAT is downregulated by D 3 preferring agonists, this could result in lower intracellular accumulation of MPP+ and reduced neurotoxicity to MPTP. The D 3 receptor preferring agonists, pramipexole and ropinirole, are the most potent of the DA agonists affording neuroprotection at 1 mg/kg for pramipexole against MPTP-induced neurodegeneration [ 3 , 21 ] and at 2 mg/kg for ropinirole against 6-OHDA lesions in rats [ 22 ]. Doses 10–30 times higher of DA agonists with low D 3 receptor affinity such as apomorphine [ 1 ] and bromocriptine [ 2 , 23 ] are needed against MPTP-induced neurodegeneration. Because neuroprotection by pramipexole is most evident with concurrent treatment with MPTP and not with post-MPTP treatment [ 24 ], i.e. when autoreceptor contributions should be most pronounced, regulation of DAT may be important. In addition, while the lowest effective dose reported is 1.0 mg/kg for mice, this is significantly greater than a clinically relevant dose in humans (1.5 mg t.i.d., p.o.[ 25 ]). Based on information from Pharmacia Corporation, equivalent plasma levels obtained with 1.5 mg t.i.d., p.o. in humans could be produced with 0.1 to 0.5 mg/kg in the mice. We tested whether 0.1 mg/kg pramipexole would be neuroprotective in aging mice against MPTP-induced neurodegeneration to the DA system, and if this effect could be due to regulation of DAT function. Results Neurohistopathology Male C57BL/6 mice of 8–10 months of age were pretreated with saline or pramipexole (0.1 mg/kg/day) followed by MPTP. At the end of the 7-day recovery period following the last injection of MPTP or vehicle were assessed for the degree of toxicity to the dopamine system by MPTP. MPTP produced a marked loss of tyrosine hydroxylase-immunoreactive (TH-IR) neurons in the substantia nigra (SN), but had less impact in the ventral tegmental area (VTA) (Figs 1 and 2 ), Unbiased stereological quantification of the number of Nissl-stained and TH-IR neurons in the SNpc and VTA was made in the midbrains of the treated groups. MPTP produced a 31% loss of TH-IR neurons in the SN and 17% loss in the VTA. Pramipexole administered once a day for 5 days (i.e. 3 days prior to and during the 2-day vehicle treatment) did not alter the total number of TH-IR neurons in the SN or VTA. Pramipexole administered for 3 days prior to and during the 2-day administration of MPTP completely prevented TH-IR neuron loss in the SN and VTA of MPTP treated mice. To confirm that TH-IR neurons were dead and not simply exhibiting reduced TH-IR, neurons in Nissl stained sections were counted. The results confirmed the data that D 3- preferring agonists can protect against MPTP in vivo as well as against MPP+ in vitro [ 26 , 27 ]. Visualization of DA fibers with dopamine transporter immunoreactivity, DAT-IR (Fig 3 ), and TH-IR (Fig 4 ) demonstrated uniform staining of the caudate-putamen (CPu) and nucleus accumbens (Nac) in the saline/saline cases. In saline pretreated mice, MPTP reduced DAT-IR by 52% in the CPu (Figs 3 and 5A ) and had a smaller but significant impact on DAT-IR labeling of DA fibers in the Nac. DAT-IR in the CPu in the pramipexole plus vehicle (PPX-SAL)-treated mice was reduced by 17%. Furthermore, in pramipexole and MPTP treated mice, a significant attenuation of the impact of MPTP in the CPu (-27% vs -52% loss) and Nac of MPTP treated mice was seen. In contrast to DAT-IR, levels of TH-IR (Figs 4 and 5B ) was not significantly reduced (~11%) by pramipexole in PPX-SAL group. MPTP produced a significant (34%) loss of TH-IR labeling of DA fibers in the CPu of WT mice and to a lesser degree in the Nac SAL-MPTP group. Pramipexole did not significantly attenuate the effects of MPTP on TH-IR labeling of DA fibers in the CPu of WT mice (Fig 5B ). [ 3 H]MPP+ and [ 3 H]dopamine uptake in mouse striatal postnuclear preparations Since pramipexole treatment in non-MPTP treated mice altered levels of DAT-IR, it was important to identify if pramipexole treatment altered the kinetics of [ 3 H]DA and [ 3 H]MPP+ uptake through the DAT. Kinetics of [ 3 H]DA uptake in mouse striata demonstrated a higher V max but a similar K m to that reported for rat fresh striatal postnuclear preparation [ 28 ]. Uptake was sodium dependent and the K m was similar due to inclusion of COMT inhibitor in the preparations [ 29 ]. C57BL/6 mice treated with pramipexole once a day for 5 days exhibited significant differences from saline treated mice (Fig 6 ). The V max (P = 0.0008) and the K m (P = 0.016) for [ 3 H]DA uptake was significantly decreased in pramipexole treated mice. The V max (P < 0.001) and the K m (P = 0.001) for [ 3 H]MPP + uptake were also significantly decreased in pramipexole treated mice. To determine if the reduction in V max and the K m for [ 3 H]DA uptake by subchronic treatment with pramipexole was due to the D 3 receptor, the values for V max and the K m for [ 3 H]DA uptake was measured in WT (C57BL/6) and D 3 KO littermate mice (bred according to our previous methods [ 30 ]), treated with vehicle or pramipexole for 5 days. ANOVA showed that there were group differences for V max values (F = 72.41, P < 0.0001) and for K m values (F = 9.78, P = 0.0007). Synaptosomes prepared from D 3 KO mice had significantly lower V max (P = 0.001, by 59%) and K m (P = 0.01, by 69 %), values for [ 3 H]DA uptake than WT mice (Table 1 ). WT mice treated with pramipexole significantly lowered V max (P = 0.001, by 65%) and K m (P = 0.05, by 50%) values for [ 3 H]DA uptake compared with WT mice treated with saline. D 3 KO mice treated with pramipexole had significantly lower V max (P = 0.001, by 57%) and K m (P = 0.051, by 77%) values for [ 3 H]DA uptake than WT mice treated with saline, but D 3 KO mice treated with pramipexole were not different from D 3 KO mice treated with vehicle. WT mice treated with pramipexole were also not different from D 3 KO mice treated with pramipexole. Dopamine transporter (DAT) autoradiography To determine if the reduction in V max and the K m for [ 3 H]DA uptake by subchronic treatment with pramipexole at 24 h post-treatment was due to the D 3 receptor regulation of the number of DAT sites, the density of DAT sites was measured in WT and D 3 KO mice treated with vehicle or pramipexole for 5 days. ANOVA showed that there were no group differences (Table 1 ) at 24 h post-treatment, but by 14 days there was a significant reduction of DAT sites in WT mice treated with pramipexole (P < 0.01), but not in D 3 KO mice (Fig 7 ). WT mice treated with pramipexole exhibited a 27% reduction in the CPu, a 31% reduction of [ 125 I]RTI binding to DAT in the Nac, and a 11% reduction in the Nas compared to WT mice treated with vehicle. Discussion Antiparkinsonian agents that are direct DA agonists (e.g. apomorphine[ 1 ], bromocriptine [ 2 , 23 ], and pramipexole [ 3 ]) are almost completely neuroprotective against MPTP-induced loss of striatal DA in mice. MPTP administration to mice can produce varying degrees of effect on permanent damage to the nigrostriatal DA system depending on the schedule of MPTP administration and survival period after MPTP [ 31 - 36 ]. In addition, under some conditions dramatic recovery from the initial loss of striatal DA levels can occur without any intervention [ 33 , 34 ]. Therefore, studies employing measures of striatal DA levels as the sole criteria for impact of MPTP and protection by DA agonists do not offer convincing evidence of protection [ 1 - 3 , 23 , 37 ]. Three studies have employed both unbiased stereological counts of TH-IR neurons in the midbrain along with estimate of striatal DA content of animals administered MPTP [ 21 , 24 ] or intracerebroventricular 6-hydroxoydopamine [ 38 ] to produce permanent damage to the DA system with protection by pramipexole. In all 3 studies the impact of MPTP on striatal DA levels was reduced by 50% with pretreatment or concurrent plus post-MPTP treatment with pramipexole (1 to 3 mg/kg). Protection against midbrain TH-IR neurons was greater, even at 12 to 14 days post MPTP/6-OHDA treatment than against DA levels. In this study of a much lower dose of pramipexole (10 to 30 times lower) given 3 days prior to and during the 2-day MPTP treatment, at 1-week post-MPTP pramipexole treatment completely spared the substantial MPTP-induced loss of TH-IR neurons from the SNpc, but was less effective against the loss of DA fibers in the CPu (TH-IR and DAT-IR measures). Measurements of striatal DA [ 21 , 24 ] which show more protection by pramipexole than our estimates of DA fibers in the CPu (TH-IR and DAT-IR measures) may reflect upregulation of DA synthesis in the remaining fibers rather than protection against fiber loss. In addition, pramipexole treatment by itself reduced DAT-IR of DA fibers, suggesting that protection against MPTP might be greater than our immunocytochemical measures indicate. The interesting observation that pramipexole down-regulated DAT-IR in saline treated mice suggests that DAT regulation plays a role in neuroprotective effects of pramipexole. D 3 receptors are most concentrated in the brain on neurons in the nucleus accumbens, however DAergic neurons do express both D 2 and D 3 receptors [ 39 , 40 ], and both D 2 and D 3 receptors are functional autoreceptors [ 16 - 20 , 41 , 42 ]. Dopaminergic neurons are tonically inhibited by dendritic and terminal autoreceptors operating in interaction with DA transporters and pharmacologically of the D2 receptor type [ 16 - 19 ]. Zapata et al [ 20 ] have reported that the D 3 preferring agonist (+)-PD 128907 does interact with D 3 autoreceptors to regulate extracellular DA levels. Those data are consistent with acute treatment with D 2 /D 3 agonists leading to increased V max [ 42 , 43 ], but less is known about subchronic treatment. We observed that subchronic treatment with pramipexole reduced the V max , and K m , of [ 3 H]DA and [ 3 H]MPP+ uptake in mice at 24 h post-treatment. Interestingly, D 3 receptor KO mice exhibited substantially lower V max and K m values of [ 3 H]DA uptake than WT mice, and pramipexole did not further reduce V max and K m values in D 3 receptor KO mice. However, the actual density of sites on the DA terminals was not reduced 24 h after pramipexole treatment in WT mice (or in D 3 receptor KO mice), as determined by DAT autoradiography. The inability to modulate the number of DAT binding sites 12 h after termination of subchronic treatment with D 2 /D 3 receptor agonists has previously been reported [ 44 ], suggesting that the initial alteration of V max and K m with pramipexole might be due to modification of the kinetics of DAT. However, pramipexole treatment reduced DAT-IR of DA fibers 7 days post-treatment, suggesting a reduction in DAT sites. Consistent with this, there was a reduction of [ 125 I]RTI binding to DAT sites in pramipexole treated WT mice 14 days after pramipexole treatment, but not in D 3 receptor KO mice. Thus, there might be both rapid and slower modifications in DAT function produced by D 3 /D 2 agonist treatment, ultimately resulting in lower DAT number, and mediated by the D 3 receptor. It is known that DAT half-life in the striatum is decreased by D 2 /D 3 receptor agonists, and increased by the dopamine D 2 /D 3 receptor antagonist, but not by D1 agonists and antagonists [ 45 ]. Furthermore, the D2 agonist-induced change in DAT kinetics iss inhibited by the co-administration of an antagonist. The absence of DA receptors can also influence DAT function, as shown by dopamine D 2 receptor-deficient mice, which exhibit decreased striatal DA uptake [ 19 ]. The present results can be compared with those reported by Saunders et al. [ 46 ] using hDAT-FLAG expressed in human embryonic kidney 293-EM4 cells, who showed by confocal microscopy and whole-cell current recordings that 2 μM d -amphetamine increased internalization of surface DAT within 1 h. Treatment with DA in HEK-hDAT cells also reduced V max , due to a diminished presence of DAT at the surface of synaptosomes [ 47 ]. Subchronic pramipexole might lead to redistribution DAT from the plasma membrane to endosomal compartments, and regulated, in part, by the D 3 receptor. The initial change in V max , and K m , could be related to a more rapid turnover of DAT, and the longer-term reduction in Bmax to greater internalization and/or reduced synthesis. Thus, D 3 preferring agonists might be potent autoreceptor agonists, and long-term changes in expression of the DAT or the functional properties of DAT could occur following subchronic treatment. This, in turn, could lead to reduced MPP + (and other neurotoxins) uptake into DA neurons, and reduced toxicity in animal models of PD. Ramirez and associates [ 37 ] reported that the neuroprotective effects of pramipexole against MPTP-induced DA loss in mice was attenuated by the selective D 3 antagonist A-437203. Furthermore, the neuroprotective effect of a low dose of pramipexole was attenuated in D 3 transgenic knockout mice and protection by pramipexole was not further attenuated by treatment with a D 3 antagonist. These in vivo data support an important role for the D 3 receptor in the neuroprotective effects of DA agonists, and our data suggest that this, in part, is due to reduced MPP + uptake into DA neurons. Conclusions We have identified that subchronic treatment with a clinically relevant dose of pramipexole beginning before initiation of MPTP treatment affords neuroprotection against DA neuron loss and, to a lesser extent, DA fiber loss. This might involve down-regulation of DAT and reduced MPP+ uptake into DA fibers. Since intracellular accumulation of MPP+ following systemic injection of MPTP requires DAT [ 4 ], then if DAT function and/or number are reduced by D 3 preferring agonists this could result in lower intracellular accumulation of MPP+ and reduced neurotoxicity to MPTP. The importance of knowing the targets of pramipexole, and other D 3 preferring agonists, in neuroprotection in animal models of PD cannot be understated, given the possibility that this protection could be extended to humans [ 12 ]. However, in vivo imaging of DAT as a tool for analyzing the neuroprotective effects of DA agonists could be difficult to interpret, since DAT might be regulated by DA agonists [ 25 , 48 ]. Our data are consistent with this hypothesis and suggest that multiple measures of DA fiber integrity are required to assess neuroprotection by agents [ 49 , 50 ]. Methods All animals were treated in accordance with a protocol approved by the Sun Health Research Institute Animal Care and Use Committee. MPTP Treatment We bred C57BL/6 mice from breeding pairs obtained from Jackson Laboratories. Eighteen male C57BL/6 mice 8–10 months of age were used for the experiments and were handled for 1 week prior to treatment. They had free access to food and water, and were maintained in a 12 h light/dark cycle prior to treatment. Mice were divided into 4 groups: the first received only vehicle (0.9% saline, 0.1 ml/10 mg body wt) (SAL-SAL), the second received pramipexole plus vehicle (PPX-SAL), the third received vehicle plus MPTP (RBI, MA) (SAL-MPTP), and the fourth received pramipexole plus MPTP (PPX-MPT). For those receiving pramipexole (Pharmacia Corporation, Kalamazoo, MI), the drug was dissolved in 0.9% sterile saline and administered by i.p. injection (0.1 ml/10 mg body wt). A single daily dose of 0.1 mg/kg body weight of pramipexole was given for 5 days. Those not receiving pramipexole were given identical injections of saline., Following pramipexole or vehicle injections on days 4 and 5, mice were given injections of either MPTP (20 mg/kg; s.c.) or vehicle (saline) twice daily at 8 h intervals. After 7-day recovery period following the last injection of MPTP or vehicle, animals were euthanized by intracardiac perfusion with 4% paraformaldehyde in 0.15 M phosphate buffer (pH 7.2) following overdose with pentobarbital 120 mg/kg body weight i.p. Brains were removed, postfixed in the perfusion fixative for 24 h at 4°C and transferred to 30% sucrose solution for additional 24 h incubation at 4°C. The tissue was frozen on dry-ice and sectioned in cryostat at 20 μm thickness. Sections were placed in cryoprotectant solution for long-term storage at -80°C. Neurohistopathology Every 10th section at the level of striatum was processed for visualization of DAT and tyrosine hydroxylase (TH), and every 5 th that of the midbrain processed for the visualization of TH-positive cell bodies using the avidin-biotin procedure [ 26 , 51 ]. Immediately adjacent sections from the midbrain were stained for cresyl violet for detection of cells. The sections were washed in phosphate buffer to remove cryoprotectant, incubated with 5% goat serum for 30 min to block background staining and incubated with anti-DAT (Chemicon, CA) or anti-TH (Chemicon, CA) at 1:1000 dilution overnight at room temperature. Control sections were treated with identical solutions but with no primary antibody. Sections were rinsed and incubated with biotinylated secondary anti-rabbit antiserum at a 1:500 dilution (Vector, CA) for 90 min at room temperature. Sections were again rinsed, incubated in streptavidin-peroxidase complex (Vector, CA) at a 1:250 dilution for 2 h at room temperature. After more thorough rinsing, sections were processed for DAB with nickel enhancement. Sections were then rinsed in phosphate buffer, mounted on gelatin-coated slides, air-dried, dehydrated, and cleared in xylene and mounted with Permount. Unbiased stereological quantification of Nissl-stained and TH-IR neurons in the SNpc and VTA was used to estimate cell number in the midbrain. The general routine at low magnification involved use of a sampling grid for the SNpc and VTA. At high magnification the computer-based imaging system randomly selected a region of the grid and clearly definable neurons was counted within the 3-dimensional block. This was repeated for every 10 th region of each section. Estimation of total neuron number was based upon actual cell counts, tissue thickness, total area of designated region, and the total number of sections analyzed per animal. Group means and variances were calculated. Our routine quantitative measurement of the optical density of regions of the striatum stained for TH-IR and DAT-IR [ 26 , 51 ] was employed. Using a Macintosh-based image analysis system with CCD camera and imaging software (BRAIN version 3.0, Drexel University) optical density measurements calibrated to an external standard (Kodak density step tablet) of the region of interest (ROI) and a control region (corpus callosum) of each section were made, the ratio of the ROI to control region was calculated, and the average for each animal determined. Group means and variances were estimated. Statistical analysis of group differences were assessed by ANOVA with pairwise comparisons performed using post-hoc t-tests and the Bonferroni correction. Testing of whether pramipexole administration alters dopamine transporter function Fourteen C57/Bl6 mice (25–30 g, 6 months) were divided into 2 groups: one group received pramipexole (0.1 mg/kg) once a day for 5 days and the other group received the vehicle (saline). An additional 10 WT and 10 D 3 receptor knockout mice (25–30 g, 6 months), bred according to our previous methods [ 30 ], were treated similarly. Twenty-four hours after the last injection of pramipexole or vehicle the mice were euthanized using CO 2 narcosis and the brains rapidly removed and snap-frozen in liquid nitrogen. K m and V max values for [ 3 H]1-methyl-4-phenylpyridinium (MPP+) and [ 3 H]DA uptake in synaptosomes derived from the striatum were deterimined by the method of Eshleman et al [ 28 ], with minor modifications. Comparison of mean values for the K m and V max of [ 3 H]MPP+ and DA uptake were made by t-test (0.05 level of significance). [ 3 H]MPP+ and [ 3 H]dopamine uptake in mouse striatal postnuclear preparations Mouse striata were dissected and homogenized with a glass-Teflon homogenizer in ice-cold modified HEPES (1 ml). The sample was centrifuged at 1000 g, for 10 min at 4°C. The supernatant was collected and centrifuged at 14,000 g for 10 min at 4°C. The pellet was resuspended in 8 ml of HEPES buffer (HEPES 25 mM, NaCl 122 mM, CaCl 2.5 mM, MgSO 4 1.2 mM, pargyline 10 uM, glucose 0.2%, ascorbic acid 0.02%, pH7.4). To the 8 ml sample, butaclamol was added at a final concentration of 100 nM, 4 ml of sample was than removed and placed in a separate 15 ml centrifuge tube (VWR, Pennsylvania) and Mazindol (Sigma, Missouri) was added at a final concentration of 40 uM. 50μl of the samples were added to borosilicate tubes (Fisher, Texas) and placed in a 25°C water bath with the drugs for a 10 min preincubation. The assay was initiated by adding 50μl concentrations of unlabeled MPP+ or DA ranging from 0–300 nM with [ 3 H]MPP+ or [ 3 H]DA at a final concentration of 20 nM. The samples were incubated at 25°C for 10 min. Specific uptake was defined as the difference in uptake observed in the absence and presence of mazindol (40μM). Uptake was terminated after 5 min by filtration through Whatman GF/C filters presoaked in HEPES buffer. Scintillation fluid was added to each filtered spot and radioactivity remaining on the filters was determined using a Wallac β-scintillation spectrometer. Each experiment involved triplicate determinations, and 6 independent experiments for each drug competition curve were performed. Dopamine transporter (DAT) autoradiography 10 WT and 10 D 3 receptor knockout mice (25–30 g, 6 months) were divided into 2 groups: one group received pramipexole (0.1 mg/kg) once a day for 5 days and the other group received the vehicle (0.9% saline). Twenty-four h after the last injection of pramipexole or vehicle, the mice were overdosed as above and the brains rapidly removed and frozen on dry- ice. An additional group of 10 WT and 10 D 3 receptor knockout mice (25–30 g, 6 months) were similarly treated but their brains were processed 14 days after the last treatment. Autoradiography of DAT sites were quantified following labeling with [ 125 I]RTI-55 (3ß-(4-iodophenyl)tropan-2 ß-carboxylic acid methyl ester) (Dupont, New England Nuclear, Boston, MA) in the presence of 100 nM paroxetine (Smith Klein Beecham BRL 29060A) to block the serotonin transporter, according to published methodology [ 30 ]. Specific binding was defined with 40μM benztropine (Sigma, St. Louis MO), and amounted to 95% of total binding. Sections were apposed to 3 H-Hyperfilm for 18 h for DAT. Autoradiographs were analyzed using a computer-based image analysis system (AIS, Imaging Research Inc., Ontario Canada) that converts transmitted optical density to the amount of radioligand bound in pmol per microgram of protein. List of abbreviations D 3 KO = D 3 receptor knockout DA = dopamine DAT = dopamine transporter MPP+ = 1-methyl-4-phenylpyridinium MPTP = 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine PD = Parkinson's disease SN = substantia nigra TH = tyrosine hydroxylase VTA = ventral tegmental area WT = wild type Competing interests The author of this manuscript received a grant from Pharmacia Corporation to partially support the costs of the research. This company is no longer in existence. Authors' contributions CW and HR carried out the MPTP treatment, performed the immunocytochemistry and participated in the statistical analysis. SB and DH carried out the uptake assays, performed the DAT autoradiography and participated in the statistical analysis. JNJ conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC524509.xml |
553971 | Measurement of fractionated plasma metanephrines for exclusion of pheochromocytoma: Can specificity be improved by adjustment for age? | Background Biochemical testing for pheochromocytoma by measurement of fractionated plasma metanephrines is limited by false positive rates of up to 18% in people without known genetic predisposition to the disease. The plasma normetanephrine fraction is responsible for most false positives and plasma normetanephrine increases with age. The objective of this study was to determine if we could improve the specificity of fractionated plasma measurements, by statistically adjusting for age. Methods An age-adjusted metanephrine score was derived using logistic regression from 343 subjects (including 33 people with pheochromocytoma) who underwent fractionated plasma metanephrine measurements as part of investigations for suspected pheochromocytoma at Mayo Clinic Rochester (derivation set). The performance of the age-adjusted score was validated in a dataset of 158 subjects (including patients 23 with pheochromocytoma) that underwent measurements of fractionated plasma metanephrines at Mayo Clinic the following year (validation dataset). None of the participants in the validation dataset had known genetic predisposition to pheochromocytoma. Results The sensitivity of the age-adjusted metanephrine score was the same as that of traditional interpretation of fractionated plasma metanephrine measurements, yielding a sensitivity of 100% (23/23, 95% confidence interval [CI] 85.7%, 100%). However, the false positive rate with traditional interpretation of fractionated plasma metanephrine measurements was 16.3% (22/135, 95% CI, 11.0%, 23.4%) and that of the age-adjusted score was significantly lower at 3.0% (4/135, 95% CI, 1.2%, 7.4%) (p < 0.001 using McNemar's test). Conclusion An adjustment for age in the interpretation of results of fractionated plasma metanephrines may significantly decrease false positives when using this test to exclude sporadic pheochromocytoma. Such improvements in false positive rate may result in savings of expenditures related to confirmatory imaging. | Background Pheochromocytoma is a rare tumor of the adrenal medulla or sympathetic ganglia, which can secrete excessive catecholamines [ 1 ]. Signs and symptoms of pheochromocytoma may include hypertension, pain (including headache, flank pain, abdominal pain, or chest pain), hyperhidrosis, anxiety or panic attacks, cardiac arrythmias, or sudden death [ 1 - 6 ]. A pheochromocytoma may also be detected as an asymptomatic incidental adrenal mass seen on abdominal imaging [ 7 ]. Metabolites of norepinephrine and epinephrine, specifically normetanephrine and metanephrine, may be measured in the plasma by high performance liquid chromatography with electrochemical detection, as described by Lenders et al. [ 8 ]. Measurement of fractionated plasma metanephrines has been called "the best test for excluding or confirming pheochromocytoma" and some investigators have recommended that such measurements "should be the test of first choice" [ 9 ]. In a recent systematic review of the world literature, we have observed that measurement of fractionated plasma metanephrine measurements have a high sensitivity ranging of 96 to100% and a variable specificity ranging from 82% to 100% [ 10 ]. The specificity of fractionated plasma metanephrines in excluding pheochromocytoma appears lowest in populations without known genetic predisposition to disease (those in whom sporadic disease is sought), with a false positive rate of up to 18% in such patients [ 9 ]. We have previously observed that the normetanephrine fraction is elevated in the majority of false positive test results and that false positives are associated with increasing age [ 11 ]. Indeed, investigators from the National Institute of Health have agreed that, "measurements of plasma normetanephrine and metanephrine provide a highly sensitive test for diagnosis of pheochromocytoma, but false positive results remain a problem" [ 12 ]. False positive biochemical test results may result in needless imaging procedures and generate excessive healthcare expenditures in detection of sporadic pheochromocytoma [ 13 ]. Reasons for false positive fractionated metanephrine test results have been explored and alternatives for further evaluation of patients with positive test results have been proposed. It is known that acetaminophen may interfere with measurements of fractionated plasma metanephrines using the Lenders' method [ 8 ], so this drug has traditionally been avoided prior to testing. Eisenhofer and colleagues have also suggested that that tricyclic antidepressants and phenoxybenzamine, respectively, may result in false positive tests [ 12 ]. Of note, in a recent Mayo study, 13% of subjects with false positive fractionated metanephrines used tricyclic antidepressants [ 11 ] but 9% of subjects who had normal fractionated metanephrine measurements also used these drugs. Thus, tricyclic use did not seem to explain the majority of false positives seen at the Mayo Clinic. Furthermore, given that phenoxybenzamine is rarely used in patients without known pheochromocytoma, use of this drug does not explain why there are so many false positive fractionated metanephrine results observed in clinical practice. As a method to distinguish false positives from true positives, Eisenhofer and colleagues have recommended clonidine-suppression testing in patients with positive fractionated plasma metanephrine measurements [ 12 ]. Eisenhofer and colleagues have recommended that plasma norepinephrine and normetanephrine levels be measured three hours after a dose of 0.3 mg clonidine in such patients [ 12 ]. We have proposed an alternative strategy to deal with false positive test results in patients without known genetic predisposition to disease [ 13 ]. As normetanephrine is the fraction responsible for the majority of false positive results, we have proposed that 24-hour urinary measurements of fractionated catecholamines and metanephrines be performed in patients with normetanephrine elevations that are approximately one and two times the upper limit of the normal range, to confirm the biochemical presence of sporadic pheochromocytoma [ 13 ]. Of note, in the case of high risk patients with known genetic predisposition to pheochromocytoma, the pre-test probability of disease may be sufficiently high that measurement of fractionated plasma metanephrines, without confirmatory biochemical studies may be reasonable [ 11 ]. Thus, the issue of lack of specificity of fractionated plasma metanephrine measurements is applicable primarily to non-genetically predisposed individuals in whom sporadic pheochromocytoma is sought. Our objective was to determine whether an adjustment of fractionated plasma metanephrines for age may improve the specificity of interpretation of these measurements when biochemically excluding sporadic pheochromocytoma, and thus result in costs savings in confirmatory imaging studies. Methods Study populations The age-adjusted fractionated plasma metanephrine score was derived from a previously described dataset of 349 subjects (including 33 people with pheochromocytoma) who underwent measurement of fractionated plasma metanephrines as part of an evaluation of suspected pheochromocytoma (also known as the derivation set) [ 11 ]. The diagnostic efficacy of this logistic-regression derived prediction rule was then tested in a second dataset of 158 subjects (including 23 with sporadic pheochromocytoma) who had measurements of fractionated plasma metanephrines performed at the Mayo Clinic Rochester the following year (the validation dataset). None of the patients in the validation set had known genetic predisposition to pheochromocytoma. All patients with pheochromocytoma had histologic confirmation of the diagnosis and those without pheochromocytoma had an alternative diagnosis assigned at the completion of their evaluation based on a combination of other biochemical test results (such as normal 24-hour urinary fractionated metanephrine and catecholamine measurements with or without normal imaging of the adrenals in the form of computerized tomography scanning [CT] or magnetic resonance imaging [MRI]). All data were obtained by retrospective chart review. The Institutional Review Board at Mayo Clinic Rochester approved the study. Measurement of fractionated plasma metanephrines The technique of Lenders (by high performance chromatography and electrochemical detection) was used to measure fractionated plasma metanephrines [ 8 ]. The traditional criterion for positivity is a metanephrine fraction greater than or equal to 0.5 nmol/L or a normetanephrine fraction greater than or equal to 0.9 nmol/L, based on a 95% reference range derived by Mayo Medical Laboratories. Subjects were advised to avoid acetaminophen for 48 hours prior to measurement of fractionated plasma metanephrines. Fractionated plasma metanephrine measurements were taken in the sitting position, with no indwelling venous cannula, and no dietary restrictions prior to testing. The lower limit of detection for the normetanephrine and metanephrine fractions was 0.20 nmol/L. Therefore, normetanephrine and metanephrine fractional measurements reported as being below the detection limit were given the value of 0.19 nmol/L for use in the logistic regression formula. Subjects who had "interfering substances" reported by the laboratory on measurement of fractionated plasma metanephrines were excluded from analyses but the number of such subjects was recorded. Statistical methods Clinical characteristics of subjects in the derivation set who did not have pheochromocytoma but had measurements of the normetanephrine or metanephrine fraction above the upper reference limits (false positive tests using traditional positivity criteria) were compared to those without pheochromocytoma who had true negative tests (χ 2 was used for categorical variables and Student's t-test for independent samples was used for continuous variables). Variables which were different between both groups at a significance level of 0.1 were then entered into a multivariable logistic regression model predicting pheochromocytoma. Age was the only variable of statistical significance distinguishing true positive from false positive fractionated plasma metanephrine measurements in the derivation set. Thus, we forced age with measurement values of normetanephrine and metanephrine fractions in a multivariable logistic regression model predicting pheochromocytoma in the derivation set (SPSS 10.0, Chicago, ILL). The formula for this age-adjusted fractionated plasma metanephrine score is shown below: -4.188 + -0.07(age) + 4.516(metanephrine) + 3.129(normetanephrine) . Age was in years, metanephrine fraction in nmol/L and normetanephrine fraction in nmol/L in this formula. In the derivation set, the Hosmer and Lemeshow test had χ 2 = 4.73, df = 8, p = 0.79 and the Cox and Snell r 2 = 0.38 showing significant model goodness-of-fit. A positivity cut-off age-adjusted metanephrine value of = -1.4752 was chosen as it carried an acceptable sensitivity of = 90.9% (30/33 patients, 95% CI = 76.4%, 96.9%) and specificity of 96.8% (300/310 patients, 95% CI = 94.2%, 98.2%) in the derivation set. The sensitivity level of over 90% was chosen because such a sensitivity level was believed to be clinically reasonable and at this level, specificity was still acceptable. We were aware that the lower the cut-off, the higher the sensitivity, but this would be at the expense of specificity. The formula for adjustment of age was applied to fractionated plasma metanephrine measurements in the second dataset (validation dataset) for testing of sensitivity and specificity in a population in whom sporadic pheochromocytoma was sought. In the validation set, individuals with known genetic predisposition to pheochromocytoma were excluded. Sensitivities were calculated by division of subjects with true positive test results by all the subjects with pheochromocytoma, and specificities were calculated by division of subjects with true negative test results divided by all subjects without pheochromocytoma. For sensitivities, specificities, and likelihood ratios, 95 percent confidence intervals (CI) were calculated using Wilson's method (except the Score Method was used for calculation of 95% CI of likelihood ratios when a zero cell was noted) [ 14 ]. The specificity of the age-adjusted metanephrine score were compared (at the same level of sensitivity) to traditional interpretation of fractionated plasma metanephrine measurements using McNemar's test [ 15 ]. Economic evaluation (decision analysis) We investigated whether use of an age-adjusted fractionated plasma metanephrine measurement could result in cost savings in imaging expenditures, compared to use of fractionated plasma metanephrine measurements interpreted in a conventional fashion, for detection of sporadic pheochromocytoma in a hypothetical tertiary care hypertensive population. We thus performed a decision analysis, with resource implications defined by costs of confirmatory imaging (CT and MRI), interpreted from a third party payer perspective. In the decision analysis, we compared algorithm "A" in which biochemical testing consisted of measurement of fractionated plasma metanephrines with these measurements interpreted relative to the 95% reference range (defined by a normetanephrine fraction above 0.9 nmol/L or a metanephrine fraction above 0.5 nmol/L) to algorithm "B", in which fractionated plasma metanephrine measurements were interpreted by using the age-adjusted metanephrine score. The sensitivity and specificity of biochemical tests was based on data from the validation set. In each algorithm, all patients with positive biochemical testing would undergo confirmatory imaging. The imaging protocol for patients with positive biochemical tests in either strategies began with computerized tomography ([CT] with and without intravenous contrast) of the abdomen, then if negative, I-131 or I-123 metaiodobenzylguanidine (MIBG) scintigraphy (efficacy for I-131 and costs for I-123 shown). The horizon (endpoint) of the analyses was positive diagnosis or exclusion of pheochromocytoma, for hypothetical hypertensive patients subjected to each strategy. The outcome of the analyses was the number of patients with pheochromocytoma expected to be detected by each strategy. The costs of false positive biochemical tests were reflected only in the costs of subsequent imaging and not in potential costs of needless surgery or its possible complications. The diagnostic efficacy of imaging studies was based on respective estimates from the literature: The sensitivity of CT imaging of the abdomen was assumed to be 98% with a specificity of 70% [ 16 ]. The sensitivity of MIBG in detecting benign sporadic pheochromocytoma was assumed to be 87.4% with a specificity of 98.9% [ 17 ]. All costs were reported in 2002 US dollars. Costs of imaging investigations were obtained from the Mayo Clinic Rochester Business Office. For the purpose of the decision analysis model, the prevalence of pheochromocytoma in the hypertensive population that would typically be screened was assumed to be 0.5% [ 18 ]. Results Findings in the derivation set The derivation set (from which the age-adjusted score was developed) consisted of 349 consecutive subjects (including 33 people with pheochromocytoma) who underwent fractionated plasma metanephrine measurements as well as 24-hour urinary total metanephrine measurements with or without 24-hour urinary catecholamine measurements in testing for pheochromocytoma at Mayo Clinic Rochester. In the derivation set, 8 of the 33 individuals had clinically-diagnosed genetic syndromes predisposing to pheochromocytoma (three familial malignant paraganglioma, two von Hippel-Lindau, one had multiple endocrine neoplasia 2a, one had multiple endocrine neoplasia 2b, and one had familial multiple benign paraganglioma). The 316 individuals in the derivation set who did not have pheochromocytoma underwent such testing the following reasons: refractory hypertension (174, 55%), spells (periodic episodes of symptoms such as palpitations, headache, or sweating, 124, 39%), adrenal mass (45, 14%), previous pheochromocytoma or known genetic predisposition to pheochromocytoma (24, 8%). The mean age of subjects with pheochromocytoma was 48 years (SD 18 years, range 16 to 60 years), whereas the mean age of subjects without pheochromocytoma was 52 years (SD 15 years, range 10 to 73 years). Six of the 316 subjects without pheochromocytoma in the derivation set did not have a plasma metanephrine fraction recorded secondary to "interfering substances" and therefore were excluded from the analyses. In the derivation set, the sensitivity of traditionally interpreted fractionated plasma metanephrine measurements (using 95% reference ranges) was 93.9% (95% CI, 80.4, 98.3) (31/33 subjects), with a specificity of 85.2% (95% CI, 80.8, 88.7) (264/310 subjects). Baseline characteristics of individuals in the derivation set without pheochromocytoma were compared for individuals who had true negative fractionated metanephrine measurements (n = 264) to those who had false positive results (n = 46) (Table 1 ). The individuals with false positive fractionated plasma metanephrine measurements in the derivation set were significantly older than those with true negative measurements (p = 0.007), whereas blood pressure, antihypertensive medication use, and rates of obstructive sleep apnea were not significantly different between these groups. Thus, age was chosen as an important variable to adjust for in interpretation of fractionated plasma metanephrines and an age-adjusted metanephrine score was developed from the derivation set data using logistic regression (as described in the Methods). At a cut-off value of -1.4752, the sensitivity of the age-adjusted metanephrine score was 90.9% (30/33 patients, 95% CI, 76.4%, 96.9%), with a specificity of 96.8% (300/310 patients, 95% CI, 94.2%, 98.2%). In this derivation set, which included individuals genetically predisposed to pheochromocytoma, one individual with a dopamine-secreting paraganglioma, another patient with a von Hippel-Lindau disease (diagnosed clinically), and a third patient with sporadic pheochromocytoma had false-negative age-adjusted metanephrine scores. The efficacy of the age-adjusted metanephrine score was then validated in the validation set. Table 1 Clinical characteristics of subjects without pheochromocytoma from the derivation set Clinical characteristic True negative Measurements of fractionated plasma metanephrines (n = 264) False positive Measurements of fractionated plasma metanephrines (n = 46) Significance testing results Gender (females/n) 142/264 26/46 P = 0.73 (χ 2 = 0.12, df = 1) Age (mean, SD years) 50.7 (15.03) 57.3 (14.6) P = 0.007 (t = 2.81, df = 63) Systolic blood pressure (mean, SD mmHg) 147 (26) (n = 260) 153 (32) P = 0.24 (t = 1.19, df = 56) Diastolic blood pressure (mean, SD mmHg) 87 (12) (n = 260) 89 (15) P = 0.39 (t = 0.87, df = 57) Number of antihypertensive agents currently used (mean, SD) 1.4 (1.5) 1.6 (1.3) P = 0.28 (t = 1.09, df = 70) Known diagnosis of obstructive sleep apnea 15/264 4/46 P = 0.43 (χ 2 = 0.62, df = 1) Findings in the validation set In the validation set of 158 subjects, 23 patients had histologically-proven sporadic pheochromocytoma (17 adrenal, 6 extra-adrenal, 8 malignant). Of the patients with pheochromocytoma, none were known to be genetically predisposed to pheochromocytoma and 14/23 were women (61%). The mean age of subjects with pheochromocytoma was 50 years (SD 16 years, range 16 to 83 years), whereas the mean age of subjects without pheochromocytoma was 55 years (SD 16 years, range 7 to 86 years). Of the 135 subjects without pheochromocytoma, 83 (62%) were women. Reasons for measurement of fractionated plasma metanephrines in the subjects without pheochromocytoma were as follows: hypertension (55, 41%), spells with or without hypertension (44, 33%), an incidentally discovered adrenal mass (20, 15%), and previously surgically cured pheochromocytoma (16, 12%). In the validation set, the sensitivity of the age-adjusted metanephrine score was the same as the traditional interpretation of fractionated plasma metanephrine measurements at 100% (23/23, 95% CI, 85.7%, 100%). The specificity of the traditional interpretation of fractionated plasma metanephrine measurements was 83.7% (113/135 patients, 95% CI, 76.6%, 89.0%) and the specificity of the age-adjusted plasma metanephrine score was 97.0% (131/135 patients, 95% CI, 92.6%, 98.8%). Thus, the false positive rate with traditional interpretation of fractionated plasma metanephrine measurements was 16.3% (22/135, 95% CI, 11.0%, 23.4%) and with the age-adjusted score it was significantly lower at 3.0% (4/135, 95% CI, 1.2%, 7.4%) (Figure 1 ) (p < 0.001 using McNemar's test). Figure 1 Percentage of false positive test results (and 95% confidence interval) at 100% sensitivity in using a traditional interpretation of fractionated plasma metanephrine measurements or an age-adjusted metanephrine score. Legend – Pmet(s), fractionated plasma metanephrine measurements; traditional fractionated plasma metanephrine measurements are considered positive if the metanephrine fraction is greater than or equal to 0.5 nmol/L or the normetanephrine fraction is greater than or equal to 0.9 nmol/L; an age-adjusted metanephrine score is positive if it is greater than -1.4752. The difference between false positive rates is statistically significant with p < 0.001 using McNemar's test. Imaging cost implications of screening strategies for pheochromocytoma In the decision analysis, biochemical testing by measurement of fractionated plasma metanephrines (traditional versus age-adjusted interpretation) was followed by CT imaging for all positive biochemical tests and if CT imaging was negative, then MIBG (I-123 or I-131) would be performed (Figure 2 ). For the purpose of the economic evaluation, in all three screening strategies, a 0.5% prevalence of pheochromocytoma was assumed in a target hypertensive population, so 500 patients with pheochromocytoma would be expected in a sample of 100,000 hypertensive subjects (Figure 2 ). Mayo Clinic Rochester charges for diagnostic studies were used: CT scan of the abdomen (with and without contrast) $1460, I-123 MIBG scan (with and without spect) $1875. Figure 2 Decision analysis: Testing algorithm for pheochromocytoma in 100,000 hypothetical hypertensive subjects (including 500 individuals with pheochromocytoma) If 100,000 subjects with hypertension would be screened using algorithm "A" (beginning with biochemical testing by measurement of fractionated plasma metanephrines, traditional interpretation), 499/500 patients with pheochromocytoma (overall sensitivity 99.8%) would be expected to be detected (1 patient expected to have false negative CT and MIBG imaging); furthermore 94,510/99,500 of subjects without pheochromocytoma would be reassured with a negative diagnosis (overall specificity 95.0%). In algorithm "A", 16,718 individuals would undergo CT scanning and 11,363 individuals would undergo I-123 or I-131 MIBG imaging. The total cost of imaging for algorithm "A" would be estimated to be 45.7 million dollars. If algorithm "B" (biochemical testing using the age-adjusted metanephrine score) would be used in 100,000 subjects with hypertension, 499/500 patients with pheochromocytoma (99.8%) would be expected to be detected and 98592/99,500 individuals without pheochromocytoma would be reassured with a negative test result (overall specificity of 99.1%). In algorithm "B", 3,485 individuals would undergo CT scanning and 2,100 individuals would undergo I-123 or I-131 MIBG imaging. The cost of imaging for algorithm "B" would be approximately 9.0 million dollars. Thus, use of the age-adjusted plasma metanephrine score for biochemical testing for sporadic pheochromocytoma in a hypothetical population of 100,000 tertiary care hypertensive patients could result in a cost savings of 36.7 million dollars with equal detection of pheochromocytoma cases, relative to using the same biochemical testing but interpreting fractionated plasma metanephrine measurements in a traditional fashion. Discussion We agree with observation by Eisenhofer and colleagues that when it comes to measurement of fractionated plasma metanephrines for exclusion of pheochromocytoma, "false-positive results remain a problem" [ 12 ], particularly when attempting to exclude sporadic disease. Originally, it was the hope was that measurement of fractionated plasma metanephrines could result in cost savings because of avoidance of multiple biochemical tests [ 19 ]. However, investigators from the National Institute of Health have recommended that clonidine-suppression tests need to be done in order to distinguish true positives from false positives [ 12 ]. An alternative to clonidine-suppression testing may be measurement of 24-hour urinary metanephrines and catecholamines in patients with mild to moderate elevations of the normetanephrine fraction (for normetanephrine values approximately one to two times the upper limit of the normal range) [ 13 ]. In our study, we have provided a unique alternative approach for improving specificity of interpretation of measurements of fractionated plasma metanephrines. By adjusting the metanephrine score for age, we have shown that it may be possible to improve specificity of interpretation of fractionated plasma metanephrines, with no loss of sensitivity in detection of sporadic pheochromocytoma, and potential savings in imaging expenditures. Of note, the sensitivity of the age-adjusted metanephrine score was superior in the validation set (100%) to that observed in the original dataset from which it was derived (91%). An explanation for this finding may be that the validation set included only people who were at risk for sporadic pheochromocytoma (in other words, non-genetically predisposed individuals), whereas genetically predisposed individuals were included in the original derivation dataset. We have previously observed that fractionated plasma metanephrine measurements may be normal in genetically-predisposed individuals with small pheochromocytomas [ 11 ]. Moreover, the physiologic cause for the observed relationship of normetanephrine measurements with age is unclear. Of note, Raber et al have noted exaggerated increases in plasma normetanephrine after exercise in hypertensive individuals with type 2 diabetes, compared to normotensive individuals with or without diabetes [ 20 ]. Furthermore, Raber et al have suggested that the excessive response of plasma normetanephrine to exercise may serve as a marker of exaggerated sympathoadrenal function in hypertensive type 2 diabetics [ 20 ]. Fractionated plasma metanephrine measurements were performed only at rest in our study and we did not examine any potential relationship with diabetes. Systolic and diastolic blood pressures were not significantly different between individuals with false positive fractionated metanephrine measurements and those with true negative measurements in the derivation set in our study. Our study is subject to several limitations. Firstly, limited clinical data on each studied individual were collected so variables that could be of interest such as: body mass index, creatinine-clearance, and rates of diabetes mellitus were not recorded. Furthermore, without autopsy confirmation, one cannot be absolutely certain that individuals labelled as not having a pheochromocytoma did not have an occult paraganglioma or pheochromocytoma. However, we believe that reasonable clinical criteria were used in excluding pheochromocytoma in our study. Another limitation is that we used an assay for measurement of fractionated plasma metanephrines that may be have been subject to interference with acetaminophen [ 8 ], whereas other assays, such as the one described by Roden et al, could have been preferable due to lack of acetaminophen interference [ 21 ]. The cut-off that we chose for positivity of the age-adjusted metanephrine score was also arbitrary and use of a lower cut-off could have resulted in improved sensitivity, albeit with likely some expense of specificity. Finally, our findings have not been validated outside a single institution. Is calculation of an age-adjusted metanephrine score practical for use in daily clinical practice? In this day of palm pilots and desktop computers, we believe that it may be feasible for clinicians to enter the formula for age-adjustment into standard desktop spreadsheet software (such as Excel, Microsoft) and perform such adjustments in the clinic. Alternatively, laboratories can provide age adjusted values to physicians when reporting test results. Thus, we do believe that calculation of an age-adjusted score is feasible to assist physicians in interpretation of fractionated plasma metanephrine measurements. Indeed, such calculations may be less cumbersome and may generate fewer healthcare expenditures than alternative procedures such as supplemental clonidine-suppression testing or collection of 24-hour urinary measures. Our observations should, however, be validated in another population outside of Mayo Clinic. Of particular note, our findings are applicable only to the screening of pheochromocytoma in low risk subjects who do not have known genetic predisposition to disease. In high risk, genetically predisposed individuals, mild elevations of normetanephrine or metanephrine measurements may be clinically significant and should prompt imaging. Conclusion An adjustment for age in interpretation of results of fractionated plasma metanephrine measurements may significantly improve the high false positive rate seen with this test when aiming to exclude sporadic pheochromocytoma. This improvement in specificity may result in savings in expenditures related to confirmatory imaging. Additional research is needed to investigate the generalizability of these findings in other clinical centres. Competing interests Dr. Sawka is a Skeletal Health Scholar funded, in part, by the Canadian Institutes of Health Research. Dr. Sawka was also a Fellow in Health Economics at McMaster University, partly funded by an unrestricted educational grant from Hoffmann-La Roche. The other co-authors have no competing interests to declare. Authors' contributions All co-authors reviewed the manuscript and made suggestions for revisions. The project idea was conceived by A.M. Sawka. Analyses were performed by A.M. Sawka, with input from Dr. Thabane. The manuscript was written and revised by Dr. Sawka. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC553971.xml |
534097 | A probabilistic model for the evolution of RNA structure | Background For the purposes of finding and aligning noncoding RNA gene- and cis -regulatory elements in multiple-genome datasets, it is useful to be able to derive multi-sequence stochastic grammars (and hence multiple alignment algorithms) systematically, starting from hypotheses about the various kinds of random mutation event and their rates. Results Here, we consider a highly simplified evolutionary model for RNA, called "The TKF91 Structure Tree" (following Thorne, Kishino and Felsenstein's 1991 model of sequence evolution with indels), which we have implemented for pairwise alignment as proof of principle for such an approach. The model, its strengths and its weaknesses are discussed with reference to four examples of functional ncRNA sequences: a riboswitch (guanine), a zipcode (nanos), a splicing factor (U4) and a ribozyme (RNase P). As shown by our visualisations of posterior probability matrices, the selected examples illustrate three different signatures of natural selection that are highly characteristic of ncRNA: (i) co-ordinated basepair substitutions, (ii) co-ordinated basepair indels and (iii) whole-stem indels. Conclusions Although all three types of mutation "event" are built into our model, events of type (i) and (ii) are found to be better modeled than events of type (iii). Nevertheless, we hypothesise from the model's performance on pairwise alignments that it would form an adequate basis for a prototype multiple alignment and genefinding tool. | Background One of the promises of comparative genomics is to annotate previously undetectable functional signals in genomic sequence, by identifying and characterising evolutionarily conserved elements. A principled way to extract such signals is by fitting the data to probabilistic models of the molecular evolutionary process. The logic runs as follows: suppose there are various kinds of conserved element x, y, z ... (e.g. exons, bits of RNA, promoters, etc ) that might explain an observed sequence homology. For each of these scenarios, we can construct a probabilistic model M x , M y , M z ... and compare the likelihood of the observed data under each of these models. The model with the best fit indicates the type of functional element present in the sequence. A groundbreaking example of how this probabilistic approach can be used is the QRNA program, designed as a comparative RNA gene predictor [ 1 ]. The three types of element considered by QRNA are noncoding RNA (called RNA ), protein-coding exons (called COD for codon), and unidentified DNA homology (called OTH for other). The former ( RNA ) was modeled using a Pairwise Stochastic Context-Free Grammar (Pair SCFG); the latter two ( COD and OTH ) using Pairwise Hidden Markov Models (Pair HMMs). The noncoding RNA predictions generated a high yield of experimental hits, and offered an information-theoretic glimpse into a modern-day RNA world [ 2 ]. It is natural to consider how such an approach might be applied to a pairwise comparison where the evolutionary "distance" between the two sequences can vary. One approach, analogous to the BLOSUM series of BLAST matrices for proteins [ 3 ], is to partition a set of training alignments into an ad hoc number of bins based on the percentage sequence identity. Alignments in the same bin (i.e. having comparable sequence identity) then represent pairs of sequences at approximately equivalent distances. For example, the BLOSUM62 substitution matrix was estimated from pairwise alignments with at least 62% identity. This sort of approach is used by the RIBOSUM basepair substitution matrices developed for RSEARCH [ 4 ], recent versions of QRNA, and the stemloc program in the author's DART software package. An alternative approach, analogous to the PAM series of BLAST matrices [ 5 ], is to treat the "distance" as a time measurement, by postulating an underlying evolutionary stochastic process or continuous-time Markov chain whose mutation rate parameters are constant over time (called stationarity in stochastic process theory). This evolutionary rate approach uses fewer parameters – and makes fuller use of the data – than the dividing-into-bins approach, since it postulates an infinitesimal generator for all time-scales of the process. For the PAM series, this generator takes the form of an instantaneous substitution rate matrix ; for a primary-sequence model, the generator is a conditionally-normalized Pair HMM or transducer [ 6 ]; for an RNA secondary-structure model, we will see that the generator is a Pair SCFG; and so on. Furthermore, the evolutionary rate model is supremely compatible with likelihood-based phylogenetic methods [ 7 ]. It's therefore worth considering such evolutionary rate-based models , although (since they're trickier to analyse mathematically) they're less suited to quick software prototyping than the "bin-by-percent-ID" approach. With this in mind, we can consider the evolutionary rate-based equivalents of the three pairwise grammars used in QRNA. The OTH model, for noncoding DNA sequence, is a Pair HMM with affine gaps; the closest evolutionary equivalent is the "long indel" model [ 8 , 9 ]. The long indel model incorporates multi-residue indels and single-residue (point) substitutions; it is based on the TKF91 model, which only allows single-residue indels [ 10 ]. In contrast, the current best evolutionary versions of the COD [ 11 ] and RNA [ 12 ] models do not attempt to model indels, changes in exon/intron structure or changes in RNA secondary structure. These are deficiencies which must eventually be addressed; ultimately they will limit the usefulness of the models. For example, the lack of a treatment of indels means that these models can only be used on a pre-generated alignment; they cannot, by themselves, be used to align sequences. In this report we present a simple but improved model of RNA structure evolution, called the TKF91 Structure Tree (Figure 1 ). This model allows not just covariant point substitutions of nucleotides, but also covariant insertions and deletions of bases, base-pairs, whole stems and multi-stem structures (Figure 2 ). Although we have not, in this paper, applied the Structure Tree to multiple alignment, or adapted it to include "long indels", the similarity to existing models [ 8 , 13 ] suggests very natural forms for such adaptations of our model. Furthermore, the TKF91 Structure Tree is algebraically tractable, yielding SCFG-based scoring schemes for simultaneous RNA alignment and structure prediction (from which alignment algorithms naturally follow). To our knowledge, this is the first such model for the evolution of RNA structure to be described within an evolutionary rate framework. A computer program for simultaneous pairwise alignment and secondary structure prediction using the TKF91 Structure Tree has been developed in C++. The potential of the model for RNA sequence alignment has been demonstrated by testing the pairwise aligner on four functional elements from the RFAM database [ 14 ]: the purine riboswitch, the nanos translational control element, the U2 splicing factor and the bacterial nuclear RNase P gene. The TKF91 Structure Tree is a very simple evolutionary model lacking some "obvious" features, such as natural selection to favour the thermodynamically stable overlap of π -orbitals between adjacent stacked bases in RNA double helices. The fact that the model appears to work reasonably well, despite the exclusion of such features, suggests that very simple models of RNA evolution may turn out to be sufficient to uncover a surprisingly large proportion of RNA sequence homology. Methods We begin by reviewing the TKF91 model [ 10 ]. This model describes the evolution of a single sequence under the action of two kinds of mutation event: (i) point substitution events, which act on a single residue only; and (ii) single-residue indel events, which insert or delete a single residue. The rates of both types of event are independent of the neighboring sequence. The TKF91 model, as defined by Thorne et al , is time-reversible. This has the implication, called the pulley principle by Felsenstein, that the position of an ancestral node in a phylogenetic tree can be slid around like a pulley without changing the likelihood of the observed data [ 7 ]. Aligning a pair of observed present-day sequences is therefore identical to aligning an ancestral sequence with its descendant, and we can talk about ancestor-descendant alignment without loss of generality. The TKF91 model can be analysed algebraically [ 10 ], and the probability distribution function (PDF) over ancestor-descendant alignments can be expressed as a Pair HMM [ 13 ] and extended to multiple sequences (using a "Multiple HMM") [ 13 ]. While it is straightforward to define a more general "long indel" model allowing multi-residue deletions and insertions [ 8 ], the only Pair HMMs for this general model that have been described to date are approximations, inspired by the form of the TKF91 model: so far there is no exact Pair HMM solution of the long indel model [ 8 , 9 , 15 , 16 ]. In this paper, we will not be considering such long-indel models. Definition of the TKF91 model The state of the TKF91 process is described by a TKF91 link sequence : a permanent immortal link at the left end of the sequence, followed by zero or more mortal links . Over time, mortal links can be deleted, and new mortal links can be inserted to the right of either immortal or mortal links. This can be treated as a birth-death process ( λ 0 , μ 0 ) with constant immigration ( λ 0 ), where "births" are identified with single-link insertions occuring to the immediate right of the parent mortal link. and "immigration" with insertions immediately right of the immortal link. A further site-independent labeling is introduced on mortal links using the singlet nucleotide alphabet , Ω = { A, C, G, U }. Each site's alphabet label evolves as an independent four-state reversible continuous-time Markov chain (RCTMC) with substitution rate R 0 ( i, j ) from state i to state j . Labels for newly inserted mortal links are drawn from the equilibrium distribution p 0 ( i ) of this substitution process. By reading off the labels of mortal links, the state of the TKF91 process can be equated to a sequence in Ω*. Analysis of the TKF91 model The following functions of ( λ n , μ n ) arise in analyses of equilibrium and transition probabilities in the TKF91 model [ 10 ]. Here t is a time parameter. Here exp( R n t ) ≡ exp( A ) is the exponential of the matrix with elements A ij = R n ( i, j ) t . The meaning of the above functions is as follows. α n is the probability of non-deletion; β n , γ n are the probabilities of insertion, following (respectively) an insertion and a deletion; κ n is the probability of continuing the ancestral sequence; and M n ( i, j ) is the conditional substitution probability from i to j . Note (1 - γ n ) κ n (1 - α n ) = β n (delete → delete and insert → insert transition probabilities are equal). Note also lim t →∞ β n = κ n . The equilibrium probability distribution over sequences in the TKF91 model is a geometric distribution with parameter κ 0 . The residues at individual positions of the sequence are independently, identically distributed at equilibrium and are sampled from the equilibrum distribution of the point substitution process. The TKF91 singlet grammar is shown in Figure 3 . The TKF91 pair grammar is shown in Figure 4 . Note that two alternate sets of rule probabilities, jointly and conditionally normalised, can be read off from Figure 4 : the conditional probabilities can be read off from column P ( d | a ), while the joint probabilities can be found by multiplying the expressions in columns P ( a ) and P ( d | a ) to obtain P ( a, d ). Extending the TKF91 model Various extensions to TKF91 have been proposed [ 8 , 9 , 15 ]. The most tractable kind of extension changes the meaning of a "link" but leaves the indel process on links intact [ 15 ]. Our RNA model is one such extension, allowing two different kinds of TKF91 model that can be mutually nested to form loop and stem regions. Consider the following extension to the TKF91 model, which we call the TKF91 Structure Tree , and which is shown in Figures 1 and 2 . This model uses the fact that an RNA secondary structure (excluding pseudoknots, kissing loops and other "tertiary" interactions) can be identified with a tree. The state of our stochastic process can thus be described by a rooted tree: every node in this tree is either a singlet , paired , loop or stem node. The tree can be broken into overlapping loop sequences and stem sequences , which correspond to strands of unpaired RNA (loops) or double helices of basepaired RNA (stems). Loops are allowed to contain unpaired nucleotides, and can also serve as a branching-off point for nested stems. Stems, on the other hand, are allowed to contain paired nucleotides, and are terminated by a loop (this reflects the smallest unit of RNA structure, which is a stem terminated by a loop). The tree is rooted by a loop sequence. The above description will now be made more precise. Definition of the TKF91 Structure Tree The state of the TKF91 Structure Tree is described by a rooted tree where each node has degree ≤ 3. There are four basic kinds of node in the tree: singlet, paired, loop and stem. Singlet and paired nodes correspond to observable nucleotides. Singlet nodes (labeled from Ω) represent independently evolving nucleotides, as in TKF91. Paired nodes (labeled from Ω 2 ) represent covariant basepairs. Loop and stem nodes determine the tree structure (Figure 1 ). Loop nodes (labeled L ), of which the root node is one, are present at the beginning of loop sequences, which contain singlet and stem nodes and are written horizontally. Stem nodes (labeled S ) are present at the beginning of stem sequences, which contain paired nodes, are terminated by a loop node, and are written vertically. The set of loop and stem node labels is written Φ. The full set of node labels is Ω ∪ Ω 2 ∪ Φ. Φ = { L, S } Ω = { A, C, G, U } Ω 2 = { AA, AC, AG, AU, CA, CC, CG, CU, GA, GC, GG, GU, UA, UC, UG, UU } Loop sequences A loop sequence is very similar to a TKF91 link sequence: as with TKF91, we have a leftmost immortal loop link followed by zero or more mortal loop links . The mortal links are inserted and deleted with rates λ 1 and μ 1 , in the style of TKF91. Each link is also a node in the Structure Tree. Links are labeled from Ω ∪ Φ: the immortal loop link is labeled L , while the mortal loop links are labeled from { A, C, G, U, S }. As with the TKF91 model, the alphabet labeling of each mortal link evolves as an independent five-state RCTMC with substitution rate R 1 ( i, j ) from i to j and equilibrium probability p 1 ( i ) of being in state i , plus the additional restriction that R 1 ( X, S ) = R 1 ( S, X ) = 0 for all X ∈ Ω: in other words, embedded stems can't interconvert with singlet nucleotides. See step 1 → 2 of Figure 2 for examples of single-nucleotide substitution in loop sequences, and steps 3 → 4 and 4 → 5 for single-nucleotide insertion and deletion. The S -labeled links possess an independently evolving embedded stem sequence that can be considered to "nest" inside the loop sequence. If the S -link is deleted, then the embedded stem (and all its children) is deleted with it. Conversely, when a new S -link is inserted, it is inserted with a complete subtree that is sampled from the equilibrium distribution over Structure Trees. See steps 6 → 7 and 7 → 8 of Figure 2 for examples of substructure insertion and deletion. Since a loop sequence is effectively a TKF91 sequence with a special "fifth nucleotide" character representing an embedded stem (the S link), it obeys the same statistics as a TKF91 sequence. In particular, the probability distribution over loop lengths at equilibrium is a geometric distribution with parameter κ 1 . Stem sequences A stem sequence is also derived from a TKF91 link sequence. Unlike the TKF91 link sequence or the loop sequence, however, it is written vertically (rather than horizontally) in Figure 1 . It consists of a topmost immortal stem link , zero or more mortal stem links , and a bottommost, terminating loop link. Again, each link is also a node in the Structure Tree. Links are labeled from Ω 2 ∪ Φ: the immortal stem link is labeled S (this is the node in the parent loop sequence), the mortal links are labeled with the paired nucleotide alphabet Ω 2 (each with an independent sixteen-state RCTMC modeling covariant pair substitution along RNA stems, with substitution rate matrix R 2 ( i, j ) and equilibrium p 2 ( i )), and the terminating loop link is labeled L . The mortal stem links experience TKF91-style insertion and deletion with rates λ 2 and μ 2 (although, in the diagrammatic form of Figure 1 , newly inserted links are placed immediately under their parent link, rather than immediately to the right). The terminating loop link L does not contribute to insertion or deletion (so is effectively immortal but inert) but possesses an independently evolving loop sequence. See step 2 → 3 of Figure 2 for examples of covariant basepair substitution in stem sequences, and step 5 → 6 for covariant basepair insertion and deletion. Note that the immortal stem link, S , is only immortal from the point of view of the stem sequence beneath it. The S is itself a mortal link in a parent loop sequence, and may be deleted as that sequence evolves. In this event, the loop link L will also be deleted, along with all its children (step 7 → 8, Figure 2 ). Thus, the only truly immortal link is the loop node at the root of the Structure Tree, which has no parents to deal death from above. As with the loop sequence, a stem sequence is effectively a TKF91 sequence with minor modifications, and it obeys the same statistics as a TKF91 sequence. The probability distribution over stem lengths at equilibrium is a geometric distribution with parameter κ 2 . Analysis of the TKF91 Structure Tree Figure 5 shows the SCFG for generating the TKF91 Structure Tree at equilibrium. There are two nonterminals, Φ, and four terminals, Ω. Figure 6 shows the pair stochastic context-free grammar for an ancestor and descendant sequence separated by evolutionary time t . Again, conditional and joint probabilities can both be read from the figure. Nonterminals are Φ 1234 ; terminals are Ω a for the ancestor and Ω d for the descendant. Φ 1234 = { L 1 , L 2 , L 3 , L 4 , S 1 , S 2 , S 3 , S 4 } Ω a = { A a , C a , G a , U a } Ω d = { A d , C d , G d , U d } Dynamic programming alignment of sequences to these grammars has the typical complexity for single-sequence [ 17 ] and pairwise [ 18 ] SCFGs. That is, for Figure 5 , the time complexity is O ( L 3 ) and the memory complexity O ( L 2 ), while for Figure 6 , the time complexity is O ( L 3 M 3 ) and the memory complexity O ( L 2 M 2 ), where L and M are sequence lengths. This is also the complexity of the single-sequence and two-sequence Sankoff algorithm [ 19 ], for which SCFGs may be regarded as a probabilistic scoring scheme. The time and memory complexity may be reduced by the use of "banding" techniques [ 20 , 21 ], that restrict the dynamic programming computation to the (typically) highest-scoring central diagonal band of the dynamic programming matrix, or by more flexible constraints on the DP iteration [ 18 ]. Grammar transformations We now describe some transformations of Figures 5 , 6 performed before implementing the grammar parsers. Null cycles The presence in a grammar of "null cycles" – sequences of production rules which cause no net change – complicates the parsing algorithms for that grammar. Generally, null cycles are avoided by programmers designing SCFGs or HMMs for sequence analysis [ 17 ]. However, in the grammars derived automatically for the TKF91 Structure Tree, null cycles arise naturally due to the possibility of zero-length loop or stem sequences in the model. There are several classes of null cycle in the grammars for the Structure Tree model, shown in Table 1 . Degeneracies As well as null cycles, there are other undesirable degeneracies in the Structure Tree grammars. Grammatical degeneracy occurs when more than one parse has the same meaning, so parses are degenerate rather than unique . Most stochastic grammars useful for bioinformatics are degenerate in the sense that there are always many folds or alignments consistent with the observed sequence data; this sort of degeneracy is technically called ambiguity . We are more concerned with other forms of degeneracy, such as structural degeneracy (multiple parses denote a single pattern of basepairing) and alignment degeneracy (multiple parses denote a single alignment). TKF91, in effect, skirts alignment degeneracy by assigning meaning to the ordering of deletions and insertions in an alignment, but alignment degeneracies arise in the Structure Tree model because there are multiple ways to delete and insert things (e.g. deleting a whole stem, versus deleting all its elements individually). There are also structural degeneracies arising from "silent" (i.e. non-emitting) loops or stems. In addition to the null cycles described above, these include (for the singlet grammar) the undesirable "loop bifurcation" L → LL and the "silent bulge" S → S (a null cycle). A full list of degeneracies for the singlet and pair grammars is shown in Table 1 . Prevention of zero-length stems The null cycles all involve zero-length stems and can be broken (NB not marginalised; the likelihood is discarded) by adding extra nonterminals before the corresponding S k , copying all outgoing rules except the nonemitting S k → L k . This also removes the loop bifurcations, but leaves silent bulges of the form S k → . The silent bulges can be removed by adding nonterminals before L k , copying all outgoing rules except L k → ε , changing to so as to prevent escape from without an unpaired emission, and adding new rules of the form to allow escape if there is a genuine bifurcation. A more careful analysis, marginalising null cycles and silent bulges rather than simply ignoring them, is almost certainly possible. Transformation to canonical form Figures 7 and 8 show the singlet and pairwise grammars with null cycles and silent bulges removed, in the canonical form used by the DART software package [ 18 ]. As well as the new sets of nonterminals described above (Φ' for singlet, for pair) the grammar includes nonterminals dedicated to bifurcations ( for singlet, for pair) and emissions ( for singlet, for pair). The separation of the nonterminals into null, bifurcation and singlet/pair emission sets puts the grammar in the form understood by the DART library [ 18 ]. The full nonterminal alphabets are Ψ for singlet states and Ψ 1234 for pair states. The asymptotic complexity of the dynamic programming recursions implied by these grammars is unchanged by the transformation to DART form. For Figure 7 , the time complexity is O ( L 3 ) and the memory complexity O ( L 2 ), while for Figure 8 , the time complexity is O ( L 3 M 3 ) and the memory complexity O ( L 2 M 2 ), where L and M are sequence lengths. Again, the complexity may be reduced by the use of "banding" [ 20 , 21 ] or other [ 18 ] constraints. Parameterisation of the TKF91 Structure Tree The Expectation Maximisation (EM) algorithm is often used for training BLOSUM-like models, e.g. estimating emission and transition probabilities for Pair HMMs [ 17 ] or Pair SCFGs [ 1 ]. It is also useful for training evolutionary rate models, which have roughly the same number of parameters and can make use of larger training sets (since the training data don't have to be "binned" by percent identity). The EM algorithm for the TKF91 Structure Tree can be separated into two parts, one for the substitution process and one for the indel process. Earlier work [ 22 ] showed how to estimate the maximum-likelihood substitution rate matrix R n using the EM algorithm, given the following sufficient statistics: , the expected number of insertions of state d ; , the expected number of aligned sites with ancestral state a and descendant state d . A forthcoming paper describes how to estimate the maximum-likelihood indel rates λ n , μ n for a TKF91 model using the EM algorithm, given the following sufficient statistics: , the expected number of deleted links not followed by an insertion; , the expected number of surviving links not followed by an insertion; , the expected number of deleted links followed by an insertion; , the expected number of surviving links followed by an insertion. We can calculate all the above update statistics simultaneously from data (the E-step) using a constrained version of the Inside-Outside algorithm [ 18 ] for the grammar in Figure 8 , as follows. Assume the joint normalisation, P ( d , a ), and suppose that is the posterior expectation of the number of times rule m of Figure 8 was applied, as returned by the Inside-Outside algorithm. For emit rules, let be the expected number of times rule m was used to emit the specific nonterminals X , Y ... ∈ Ω. Then The terms in parentheses are to be omitted if the conditional normalisation, P ( d | a ), is used. The relationship between the expected insert and match usage , and the expected start, wait and transition usage of the previous work [ 22 ] is as follows where is defined as in the previous work [ 22 ] Results The pairwise aligner for the TKF91 Structure Tree is distributed as part of the DART package at the following URL: The aligner is based on the Stochastic Context-Free Grammars (SCFGs) shown in Figures 7 and 8 , as explained in the Methods section. The specific implementation uses a general Pair SCFG dynamic programming (DP) engine with accelerating heuristics, to be described in a later paper (manuscript in preparation). To test the performance of the model at aligning and predicting structure of RNA sequence, we considered pairs of RNA sequences from four different families, with varying degrees of homology at the level of secondary structure. The four families were the purine riboswitch (Figure 9 ), the nanos translational control element (TCE) from Drosophila (Figure 10 ), the U2 spliceosomal factor (Figure 11 ) and bacterial nuclear RNase P (Figure 12 ). For a given family, denote the two sequences in the family by A, B . The following computations were performed: (1A) , (1B) For each of the two sequences ( A, B ) taken individually, the secondary structure was predicted without the aid of comparative information from the other sequence, using the single-sequence SCFG of Figure 7 . (2) The two sequences ( A, B ) were then aligned using the TKF91 model, without making use of any model of RNA structure, using the Pair HMM of Figure 4 . (3) Finally, the two sequences ( A, B ) were aligned using the TKF91 Structure Tree model introduced in this paper, using the Pair SCFG of Figure 8 . These computations allow a comparison between the TKF91 model, the single-sequence SCFG of Figure 7 and the TKF Structure Tree. The results, including structure and alignment predictions, are illustrated in a compact visual representation that we call a "fold/alignment dotplot". The key to interpreting the fold/alignment dotplot is shown in Figure 13 . The subregions labeled a-f have the following meaning: (a) This triangular dotplot illustrates the single-sequence structure prediction for sequence A of computation (1A). The pixel color at co-ordinates ( x, y ) represents the posterior probability that residues x and y of A are base-paired, in the absence of any information from sequence B . (b) This triangular dotplot illustrates the single-sequence structure prediction for sequence B of computation (1B). The pixel color at co-ordinates ( x, y ) represents the posterior probability that residues x and y of B are base-paired, in the absence of any information from sequence A . (c) This rectangular dotplot illustrates the structure-free pairwise alignment of computation (2). The pixel color at co-ordinates ( x, y ) represents the posterior probability that residue x of A is homologous to residue y of B , in the absence of any structural information from the two sequences. (d) This triangular dotplot illustrates the comparative structure prediction for sequence A of computation (3). The pixel color at co-ordinates ( x, y ) represents the marginal posterior probability that residues x and y of A are base-paired, summed over all alignments to sequence B . (e) This triangular dotplot illustrates the comparative structure prediction for sequence B of computation (3). The pixel color at co-ordinates ( x, y ) represents the marginal posterior probability that residues x and y of B are base-paired, summed over all alignments to sequence A . (f) This rectangular dotplot illustrates the structural pairwise alignment of computation (3). The pixel color at co-ordinates ( x, y ) represents the marginal posterior probability that residue x of B is homologous to residue y of A , summed over all secondary structures of sequences A and B . Note that the orientation of this plot is flipped (reflected about the diagonal axis) relative to (c). In addition, the "true" (published) structures and alignments are overlaid on the computational results as blue squares (or blue dots, on the larger images). The rate parameters used for the TKF91 Structure Tree were obtained by maximum likelihood training from a random selection of structurally-annotated RFAM alignments, as follows: λ 1 = 0.027, μ 1 = 0.03; λ 2 = 0.007, μ 2 = 0.01; p 1 ( S ) = 0.01. The substitution rate parameters were taken from the PFOLD program [ 12 ]. The evolutionary "time" between the two sequences was set to 1 in each case. In the case of the RNase P and U2 genes, the DP algorithms were constrained to a band along the main diagonal of the DP matrix; this constraint was imposed due to limited memory. No such constraint was imposed for the purine riboswitch computations. The posterior probabilities of folding and alignment (dotplots a-f) obtained by DP on these three classes of element are shown in Figure 14 (for the purine riboswitches), Figure 15 (for the nanos TCEs), Figure 16 (for the U2 snRNAs) and Figure 17 (for the bacterial nuclear RNase P genes). In all cases, the Pair SCFG sharply resolves the most probable stems in the sequences; for the nanos, U2 and RNase P sequences, it also resolves the pairwise alignment. Purine riboswitch The purine riboswitches are a class of cis -acting regulatory elements that specifically bind adenine or guanine and are involved in the post-translational regulation of purine transport and biosynthesis [ 23 ]. Figure 9 shows the alignment of the two riboswitch sequences, from Bacillus halodurans and Streptococcus pneumoniae , which was taken from the RFAM database [ 14 ]. The two secondary structures of this pair are exactly identical, although the primary sequences are considerably diverged. Figure 14 shows the posterior dotplots for the purine riboswitches. This is an easy case for the model, with a strong signal and few gaps. The TKF91 Structure Tree grammar (Figure 8 ) is able to identify all stems correctly, with some slight uncertainty over the alignment. The primary-sequence TKF91 grammar (Figure 4 ) is similarly able to find the correct alignment, although the singlet folding grammar (Figure 7 ) has difficulty resolving the stems (note that this grammar does not model basepair stacking effects). Nanos translational control element The translational control element (TCE) is a regulatory sequence from the 3' untranslated region of the Drosophila nanos gene, involved in post-translational degradation and transport of nanos mRNA, which localises to the posterior of oocytes and other cell lines [ 24 ]. Figure 10 shows the alignment of the two TCE sequences, from Drosophila virilis and Drosophila melanogaster , which was curated by hand from the description by Gavis et al [ 24 ]. The two secondary structures of this pair share the same overall bifurcating-stem structure, but with some changes in stem length. Figure 15 shows the posterior dotplots for the nanos TCEs. This time the TKF91 Structure Tree grammar (Figure 8 ) does considerably better than the primary-sequence TKF91 grammar (Figure 4 ) at finding the correct alignment, probably due to the gaps at the end (the TKF91 grammar in Figure 4 is effectively a global aligner with linear gaps, so that the alignments it produces tend to form a continuous line from corner to corner of the DP matrix, without major discontinuities, as can be seen in region (c) of Figure 15 ). Again, the Structure Tree does much better than the singlet folding grammar (Figure 7 ) at distinguishing real stems from background noise, since it is able to use covariation of basepaired residues as a clue. U2 snRNA The U2 small nuclear RNA recognizes and binds the branch point region of introns in pre-mRNA [ 25 ]. Figure 11 shows the alignment of the two splicing factors, from Tetrahymena thermophila and Leptomonas collosoma , was taken from the RFAM database [ 14 ]. The secondary structures of the two sequences are quite similar, but the Leptomonas U2 has a deletion of roughly 35 bp that eliminates an entire stem (stems 4–6 on Figure 11 ). Figure 16 shows the posterior dotplots for the U2 snRNAs. As before, the Structure Tree's stem predictions (regions (d) and (e), above the main diagonal of Figure 16 ) are far more specific than the singlet grammar's predictions (regions (a) and (b), below the main diagonal). The primary-sequence TKF91 grammar (Figure 4 ) is, again, hampered by its global alignment and linear gap penalty, and the alignment in region (c) is stretched and also uncertain. However, the Pair SCFG (Figure 8 ) manages to identify the 35-bp deletion and correctly finds stem 4 of Figure 11 , though stems 5–6 have a lower probability (when predicting the structure of this deleted region, the Pair SCFG is unable to use covariation and must rely on basepairing information alone). Bacterial nuclear RNase P Nuclear RNase P is a class of endoribonuclease ribozyme involved in the production of mature 5' ends of transfer RNAs during tRNA biosynthesis [ 26 ]. Figure 12 shows the alignment of the two ribozyme sequences, from Pichia canadensis and Clavispora opuntiae , which was taken from the RFAM database [ 14 ]. The secondary structures of the two sequences are quite different, with major change in stem length and deletion of whole stem structures, characteristic of this gene family (stems 0–2 and 8–9 of Figure 12 ). Figure 17 shows the posterior dotplots for the RNase P genes. This family is one of the most mutable in RFAM, and the TKF91 Structure Tree performs poorly on this case. Both the Pair HMM (Figure 4 ; region (c) of Figure 17 ) and the Pair SCFG (Figure 8 ; region (f) of Figure 17 ) get the alignment almost entirely wrong, except for a region toward the 3' end that doesn't contain any stems (the region just before stem 8 of Figure 12 ). As a consequence, the Pair SCFG also fails to predict any stems correctly; the singlet SCFG (Figure 7 ) does no better. Region (f) of Figure 17 displays the continuous-line alignment from corner-to-corner, that is characteristic of global aligners with linear gaps: unlike the case of the U2 alignment, the structural signal here is insufficient to compensate for the indel-modeling deficiencies of the TKF91 Structure Tree. The log-odds score of the "true" alignment (Figure 12 ) under the Structure Tree model is highly negative (-547 bits), suggesting that the model is poorly adapted for this example. Compare this with the scores for the previous examples: Figure 9 scored 2 bits, Figure 10 scored -82 bits and Figure 11 scored 35 bits. The low score for the nanos TCEs (Figure 10 ) was due primarily to the deletions in the outermost stem; the score rose to -5 bits with judicious trimming and careful choice of the "time" parameter. Discussion We have described a reversible continuous-time Markov chain, called the "TKF91 Structure Tree", that describes both (i) covariant substitutions and indels in RNA sequence contingent upon a particular secondary structure, and (ii) changes in the underlying RNA secondary structure, corresponding to gain and loss of substructures. A pairwise alignment algorithm based on the model has been implemented in C++ and tested on four homologous pairs of RNA functional element from RFAM [ 14 ]. As with the TKF91 model on which the TKF91 Structure Tree is based [ 10 ], it should be possible, systematically, to design corresponding algorithms for multiple sequences, using either exhaustive dynamic programming [ 6 , 27 ] or Markov Chain Monte Carlo [ 13 ]. It should be noted that the present implementation of the TKF91 Structure Tree is not designed to be a direct competitor to programs like FOLDALIGN [ 20 ], DYNALIGN [ 21 ] or CARNAC [ 28 ]. Such pairwise alignment programs are optimized for criteria like alignment accuracy and sensitivity. The TKF91 Structure Tree, on the other hand, was designed as an expository evolutionary model, ultimately aimed at phylogenetic analysis of multiple RNA sequences in an evolutionary likelihood context. The pairwise alignment program reported in this paper was implemented to demonstrate the potential of this evolutionary model, rather than for use as a practical alignment tool. The author's STEMLOC program, which is similarly based on Pair SCFGs, has been optimized for practical applications (preferring short-term performance advantages over long-term design considerations) and may be freely downloaded from . The results of our tests on pairwise alignments from RFAM reveal the strengths and weaknesses of our model. When RNA structure is very strongly conserved and indels are few, as with the purine riboswitches selected for this comparison (Figure 14 ), the TKF91 Structure Tree performs well at both structure prediction and alignment. On such alignments, the model is expected to be similar to PFOLD [ 12 ], which uses an SCFG and an evolutionary substitution model but lacks an evolutionary treatment of gaps. When the alignment has numerous indels in loops and stems, as in the selected nanos TCEs (15), or even minor rearrangements of structure, as in the selected U2 splice factors (16), the Structure Tree still seems to work well. However, beyond a certain level of structural change, as in the selected RNase P alignment (17), the model performs poorly and leaves considerable room for improvement. In view of the room for improvement, we can identify a number of weaknesses of the TKF91 Structure Tree that could be improved in future models: • Sources of degeneracy such as zero-length stems and loops were removed "by hand" from the Pair SCFG (Table 1 ). These degeneracies could have beeen specifically excluded from the evolutionary model, but with the apparent cost of making an exact solution much harder to find. One might expect the nondegenerate grammars of Figures 7 and 8 to approximate the transition probabilities of such a nondegenerate model. • Indel rates for whole stems/multistems are same as for unpaired residues. In nature, stem gain and loss is much slower than unpaired residue insertion/deletion, since the former is a structural change while the latter is not. • Multiple-residue indel events, and hence affine gap penalties, are not allowed. Again, the poor performance on the RNase P alignment may in part be due to this: the alignment generated has many small gaps scattered throughout, whereas the "true" alignment has fewer, longer gaps. This is a characteristic artefact of using a point indel model (linear gap penalty) where a multi-residue indel model (affine penalty) would be more appropriate. • Stems cannot be deleted without deleting all their "children" as well (i.e. all stems nested inside). Empirical inspection of alignments in RFAM, however, reveals many structures where an outer stem has been deleted or truncated, while inner stems are preserved. Again, perhaps an affine gap penalty for covariant indels (i.e. indels in stems) would address this. Alternatively, one might contrive some kind of "ragged-end" local alignment model, e.g. by embedding the finite TKF91 Structure Tree in an infinite, unobserved Structure Tree (c.f. [ 8 ]), though this may not be the ideal way to model such effects. • The equilibrium distribution over structures is highly simplified. For example, there is currently no modeling of fine-scale energetics such as basepair stacking propensities due to π -orbital conjugation. Mathematically, the complexities of modeling such effects are somewhat similar to those involved in modeling nearest-neighbor substitution biases in DNA (such as methylation-induced CpG deamination). Since recent progress has been made with such models [ 29 , 30 ] we might eventually expect inclusion of stacking effects in models of covariant RNA substitution, as well. • Bulges cannot be inserted into stems, except via the following awkward mechanism: the insertion of a bulge into a stem requires the pre-existence of a null S → L → S transition, where the L is empty. To fix this, L nodes could be allowed in stem sequences, just as S nodes are allowed in loop sequences (in fact, one should probably introduce "left" and "right" L -nodes, corresponding to left & right bulges). However, this would increase the potential for degeneracies. • We have assumed that all stems and loops evolve at the same rate, whereas empirical inspection of RFAM of suggests otherwise. It is known that the analogous assumption in proteins (that all sites evolve at the same rate) can skew phylogenetic distance estimation [ 31 ], and perhaps a similar correction to the discretized gamma priors used in proteins could be applied here [ 32 ]. • There is no special treatment of structural features such as triloops, tetraloops, triple-A platforms, U-turns and the like. Such features are often observed to be evolutionary conserved [ 33 , 34 ] and seem likely to be involved in intermolecular interactions [ 35 , 36 ]. It would be relatively easy to incorporate such features into the TKF91 Structure Tree, as special classes of L - or S -branch. • While the lengths of stem sequences are geometrically distributed in the TKF91 Structure Tree, due to their roots in the TKF91 model, empirical observations of real RNA structures suggest that real stem lengths follow a fairly tight length distribution. Such approximations in modeling stem lengths could conceivably contribute to poorer performance of the model. (In practise, we have not observed unnaturally long stems in the output of the TKF91 Structure Tree aligner, but the existence of a long, perfect inverted repeat in the sequence could conceivably bring out this problem.) Despite these drawbacks, the results of our preliminary benchmark suggest that the TKF91 Structure Tree may be useful for aligning (at least the better-conserved) RNA functional elements. Given the recent growth of RNA sequence and structure databases such as RFAM [ 14 ] and SCOR [ 34 ], it would be interesting to carry out a broad-scale, empirical study of the mutations of RNA structures. This could then be used as a starting point for systematically designing and benchmarking an improved evolutionary model for RNA. In the meantime, the results presented here suggest new ways of designing evolutionary grammars that recognise higher-level structural change as well as point substitutions and indels, offering new ways of using high-throughput comparative sequencing to interpret the contents of genomes. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC534097.xml |
553965 | The domain architecture of large guanine nucleotide exchange factors for the small GTP-binding protein Arf | Background Small G proteins, which are essential regulators of multiple cellular functions, are activated by guanine nucleotide exchange factors (GEFs) that stimulate the exchange of the tightly bound GDP nucleotide by GTP. The catalytic domain responsible for nucleotide exchange is in general associated with non-catalytic domains that define the spatio-temporal conditions of activation. In the case of small G proteins of the Arf subfamily, which are major regulators of membrane trafficking, GEFs form a heterogeneous family whose only common characteristic is the well-characterized Sec7 catalytic domain. In contrast, the function of non-catalytic domains and how they regulate/cooperate with the catalytic domain is essentially unknown. Results Based on Sec7-containing sequences from fully-annotated eukaryotic genomes, including our annotation of these sequences from Paramecium, we have investigated the domain architecture of large ArfGEFs of the BIG and GBF subfamilies, which are involved in Golgi traffic. Multiple sequence alignments combined with the analysis of predicted secondary structures, non-structured regions and splicing patterns, identifies five novel non-catalytic structural domains which are common to both subfamilies, revealing that they share a conserved modular organization. We also report a novel ArfGEF subfamily with a domain organization so far unique to alveolates, which we name TBS ( TB C- S ec7). Conclusion Our analysis unifies the BIG and GBF subfamilies into a higher order subfamily, which, together with their being the only subfamilies common to all eukaryotes, suggests that they descend from a common ancestor from which species-specific ArfGEFs have subsequently evolved. Our identification of a conserved modular architecture provides a background for future functional investigation of non-catalytic domains. | Background Guanine Nucleotide Exchange Factors (GEFs) are obligatory components of signaling cascades regulated by small GTP-binding proteins (called small G proteins hereafter). Their biochemical activity is to stimulate the dissociation of the tightly bound GDP nucleotide from the small G protein in response to cellular signals. Thereby, they favor the binding of the more abundant cellular GTP, organizing the active conformation of the small G protein which can recruit its effectors (reviewed in [ 1 ]). Each small G protein family features its own ensemble of GEFs characterized by a conserved catalytic domain responsible for nucleotide exchange, which is generally combined with non-catalytic domains that define the spatio-temporal conditions of activation. In the case of small G proteins of the Arf family, which are major regulators in membrane trafficking (reviewed in [ 2 ]), the exchange domain is a conserved module of ~200 amino acids called the Sec7 domain [ 3 ]. Its biochemical (reviewed in [ 4 ]) and structural [ 5 , 6 ] mechanisms have been investigated in detail. Remarkably, the Sec7 domain is the only domain that is conserved in all ArfGEFs (reviewed in [ 7 , 8 ]) and it is to some extent interchangeable between species [ 9 ]. In contrast, little is known about the functions of the other domains, which are likely to determine intracellular localization of ArfGEFs and their responsiveness to specific signals. As for most small G proteins, Arf family members are outnumbered by ArfGEFs in many species. In humans for instance, 5 Arf proteins have been identified, and there are at least 13 proteins carrying a Sec7 domain, of which most have been characterized as bona fide ArfGEFs (reviewed in [ 7 , 8 ]). Thus an individual Arf protein may be activated by more than one GEF, emphasizing that essential aspects in building up the Arf responses may be encoded by the modular architecture of their GEFs. Sequence similarity in the non-catalytic regions forms the basis for the classification of ArfGEFs into subfamilies. 8 subfamilies are currently identified in eukaryotes with sizes ranging from small (~40–80 kD including CYH, EFA6 and FBS), to medium (~100–150 kD, including BRAG/LONER, SYT1, SYT2) and large (~170–200 kD) ArfGEFs (reviewed in [ 7 , 8 ]). Large ArfGEFs comprise two subfamilies which we will refer to as the BIG and GBF subfamilies after the name of their human representatives. The GBF subfamily includes human GBF1 [ 10 ], Arabidopsis GNOM [ 11 ] and Saccharomyces Gea1 and Gea2 [ 12 ], the BIG subfamily human BIG1 and BIG2 [ 13 , 14 ] and yeast Sec7p [ 15 ]. An additional subfamily called RalF is found in Rickettsie and Legionella bacteria, likely acting on an host Arf pathway [ 16 ]. Analysis of the CYH and EFA6 subfamilies, present only in multicellular animals, and that of the large ArfGEFs, found in all eukaryotes, have yielded most of the functional data currently available. CYH and EFA6 are active on Arf6 at the plasma membrane where they may function in the crosstalk of membrane traffic, cytoskeleton dynamics and signalling in endosomal pathways (reviewed in [ 17 ]). Most members of the BIG and GBF subfamilies characterized so far function in vesicular trafficking at the Golgi [ 12 , 14 , 18 ], except for BIG2, which also localizes on recycling endosomes [ 19 ], and GNOM which acts in the endosomal recycling pathway [ 11 ]. The domain architecture of non-catalytic regions of ArfGEFs, hence their contribution to specific aspects of the build-up of the Arf response, is essentially not established except for those ArfGEFs with domains found in other classes of cellular regulators. The known domains include membrane-interacting PH domains in the CYH (reviewed in [ 20 ]), EFA6 [ 21 ] and possibly BRAG/LONER[ 22 ] subfamilies, and a putative F-box in the FBS subfamily [ 23 ], a protein-protein interaction domain that has been involved in the recruitment of substrates to the SCF ubiquitination machinery. Coiled-coil structures have also been predicted in the N-terminus of the CYH subfamily and in the C-terminus of the EFA6 subfamily. In CYH, they are involved in dimerization [ 3 ], recruitment of partners [ 24 ] and Golgi targeting [ 25 ], and in actin remodeling functions in the case of EFA6 [ 21 ]. On the other hand, although the functions of BIG and GBF subfamilies have been the subject of many investigations, their architecture is barely described, making it difficult to associate biochemical activities with their molecular structure. Here we investigate the domain architecture in the BIG and GBF subfamilies, including all sequences from fully annotated eukaryotic genomes and our novel annotation of Sec7-containing proteins from the Paramecium tetraurelia alveolate. Sequence comparisons combined with secondary structures and splicing patterns analysis identifies five novel domains that are conserved between BIG and GBF subfamilies, thus unifying them as a higher order subfamily with a probable common ancestor. Our analysis of Sec7-domain containing sequences from Paramecium also introduces a novel subfamily of ArfGEFs unique to alveolates, which we call TBS ( TB C- S ec7). Results and discussion A conserved domain architecture in BIG and GBF subfamilies The BIG and GBF subfamilies are the only ArfGEFs subfamilies common to all eukaryotes [ 8 ] and the sole ArfGEFs present in plants [ 26 ] (Figure 1 ). They are therefore possible representatives of ancestral ArfGEF functions and may provide a model to understand the nature and implementation of activities associated with the exchange function carried by the conserved Sec7 domain. However, domain 'hunting' in BIG and GBF subfamilies was complicated by the facts that the Sec7 domain is their only domain that could be identified from known domain repertoires, and that their poorly characterized non-catalytic regions were not found outside these ArfGEF subfamilies. Alternatively, we based our search of candidate structural domains in BIGs and GBFs on the bioinformatics analysis of their own sequences, taking advantage of the growing number of sequences from fully annotated genomes from mammals, insects, plants, nematode, and fungi, to which we included our annotation of Sec7-containing proteins from the newly sequenced genome of Paramecium. Figure 1 Venn diagram of the nine Sec7-containing subfamilies sorted according to the species where each subfamily has been found. The TBS subfamily was identified in this study. The BIG and GBF subfamilies are merged in a higher order subfamily (GBG), and are the only subfamily common to all eukaryotes. Multiple alignments of 42 sequences (listed in Table 1 ) revealed that the BIG and GBF subfamilies share an unexpected conserved architecture (schematized in Figure 2 ). Two homology domains are located in N-terminus of the Sec7 domain – the DCB (~150 aa) and HUS ( H omology U pstream of S ec7, ~170 aa) domains – and three in its C-terminus -the HDS1 ( H omology D ownstream of S ec7, ~130 aa), HDS2 (~160 aa) and HDS3 (~120 aa) domains (Figure 3 , 4 , 5 , 6 , 7 ). In Arabidopsis GNOM, the DCB domain is included in an N-terminal region of ~250 residues involved in dimerization and possibly binding to cyclophilin5 and called the D imerization/ C yclophilin B inding region [ 27 ], after which the new domain was named (Figure 3 ). All domains are predicted to have a high content of α-helices that co-align in the multiple sequence alignments, reinforcing the prediction of sequence similarities and suggesting that these domains form folded structural units that may share common functional features. Except for the N-terminal DCB domain which is also found in the yeast protein Ysl2p [ 28 ], all of them are unique to these two ArfGEFs subfamilies within the detection limits of the BLAST search. The HUS domain features a remarkably conserved N(Y/F)DC(D/N) motif, which we call the HUS box, which is predicted to locate in a loop where it may be available for functional interactions (Figure 4 ). The N- and C-terminal ends of BIGs and GBFs are more variable, including an unusual enrichment in Asp/Glu or Pro residues in some members. A specific feature of BIG members is that their C-terminus is in general less variable than that of GBFs, and is predicted with a significant amount of secondary structures. In contrast to the predicted structural domains, the intervening regions are highly variable in length and do not yield aligned sequences. Analysis of their amino-acid composition reveals a paucity of hydrophobic residues which is predicted to associate with an essentially unfolded conformation, suggesting that they act as linkers to tether the functional domains together. Table 1 BIG and GBF protein sequences used in this study. Species Protein name a Accession Number Size in amino acids Metazoa Ag Q7PWN5 EAA14874 1522 Q7PXQ7 EAA00837 1285 Ce Q9XWG5 NP_493386 1628 Q9XTF0 NP_499522 1820 Dm Q9VJW1 AAF53331 1653 Q9V696 AAF58532 1983 Hs BIG1 Q9Y6D6 1849 BIG2 Q9Y6D5 1785 GBF1 Q92538 1859 Rn BIG1 XP_232614 1987 BIG2 Q7TSU1 1791 GBF1 XP_347197 1883 Fungi Ca EAL04295 EAL04295 1839 EAL02873 EAL02873 1015 Nc Q7SAX4 EAA33549 1940 Q7SAL8 EAA33457 1626 Sc SEC7 P11075 2009 GEA1 P47102 1408 GEA2 P39993 1459 Sp SC71 Q9UT02 1811 SC72 Q9P7V5 1822 Q9P7R8 NP_596613 1462 Viridiplantae At At1g01960 Q9LPC5 1750 At3g43300 NP_189916 1728 At3g60860 Q9LZX8 1793 At4g35380 O65490 1711 At4g38200 NP_195533 1698 GNOM Q42510, At1g13980 1451 GNL1 Q9FLY5, At5g39500 1443 GNL2 NP_197462, At5g19160 1375 Os 9631.m01366 Q8S565 1789 9630.m00920 Q9XGN9 1687 9634.m04029 - 1704 9635.m03752 - 1680 9631.m04495 - 1456 9630.m02122 - 1396 9632.m00175 Q7XT11 1407 Alveolata Pt GGG1 CR533425 1615 GGG2 CR533424 1628 GGG3 CR533423 1598 GGG4 CR533422 1599 GGG5 CR533421 1435 a Unnamed sequences are designated by their NCBI accession number, AGI (Arabidopsis Genome Initiative) locus numbers for At and TIGR model temporary IDs for Os. BIG and GBF subfamily members are in normal and bold characters respectively, except for Pt members which have not been assigned to either subfamily (see also Figure 8). Species abbreviations are: Ag, Anopheles gambiae ; Ce, Caenorhabditis. elegans ; Dm, Drosophila melanogaster ; Hs, Homo sapiens ; Rn, Rattus norvegicus ; Ca, Candida albicans ; Nc, Neurospora crassa ; Sc, Saccharomyces cerevisiae ; Sp, Schizosaccharomyces pombe ; At, Arabidopsis thaliana ; Os, Oryza sativa ; Pt, Parameciumtetraurelia ). Figure 2 The common domain architecture of the BIG and GBF subfamilies . From N- to C-terminus : DCB , HUS, Sec7, HDS1, HDS2, HDS3. Linker regions of variable length and sequence are shown in grey, with alternate splicing sites in human GBF1, BIG1 and BIG2 in black, white and grey diamond shapes respectively. Interactions reported in the litterature are indicated in boxes of width corresponding to the mapped regions, except for myosin IXb interaction which was studied only with full-length BIG1. Arrows indicate predicted Protein kinase A-anchoring motifs. 1 [45]; 2 [27]; 3 [46]; 4 [47]; 5 [48]; 6 [49]; 7 [50]; 8 [51]. Figure 3 The conserved domains of the BIG/GBF subfamily: DCB domain . Multiple sequence alignement of the conserved domains from BIG and GBF representative sequences showing secondary structure predictions that co-align in all sequences. Colour coding is red for invariant residues, yellow for a sequence similarity score threshold of 0.15 using the BLOSUM62 matrix. The gap in helix 4 is due to an insert in the drosophila Q9V696 sequence, and may be resulting from a sequence annotation error. Figure 4 The conserved domains of the BIG/GBF subfamily: HUS domain. See Figure 3 legend for alignment details. The highly conserved HUS motif is boxed in blue. The gap in helix 5 domain is due to an insert in the Arabidopsis 3g43300 sequence, and may be resulting from a sequence annotation error. Figure 5 The conserved domains of the BIG/GBF subfamily: HDS1 domain. See Figure 3 legend for alignment details. Figure 6 The conserved domains of the BIG/GBF subfamily: HDS2 domain. See Figure 3 legend for alignment details. Figure 7 The conserved domains of the BIG/GBF subfamily: HDS3 domain. See Figure 3 legend for alignment details. To further investigate the predicted organization of BIGs and GBFs in 6 conserved helical domains connected by variable linkers, splicing patterns of human BIGs and GBFs were analyzed in the large number of cDNAs and ESTs in the databases that correspond to GBF/BIG transcripts. This revealed the use of alternate splice donor and acceptor sites predicted to yield proteins with insertions and deletions ranging from 1 to 38 residues, and a number of splice variants arising from exon skipping (Table 2 ). Strikingly, all observed sequence variations occur in regions identified as linkers between conserved domains (Figure 2 ). Together with our domain analysis, this suggests that splicing at non-canonical exon/intron boundaries is only tolerated in regions of the protein where the impact upon folding of domains with essential function would be minimal. Table 2 Alternate splice variants of human GBF1, BIG1 and BIG2 a,b Change in protein Apparent cause of variation in transcript GBF1 Extra Q at 337, 55 residues upstream of HUS domain Insertion of 3 nucleotides (nt) resulting from use of alternate 3' acceptor site within intron during splicing of exons 10 and 11 New Ser and loss of 14 residues at 613, between HUS and Sec7 domains Loss of 36 nt resulting from use of alternate 5' donor site within exon 15 during splicing with exon 16 Loss of VSQD at 1494, 38 residues upstream of HDS3 Loss of 12 nt resulting from use of alternate 5' donor site within exon 33 during splicing with exon 34 Frame-shift at 1625 causing loss of last 19 residues of HDS3 Intron retention between exons 36 and 37 leading to frame shift and premature termination Loss of 38 residues starting at 1784, near C-terminus Loss of 114 nt resulting from use of novel cryptic splice donor and acceptor sites within exon 40. BIG1 Frame-shift at 1340, 32 residues upstream of HDS3 Loss of 59 nt resulting from use of alternate 5' donor site within exon 28 during splicing with exon 29 Loss of VSEKPL at 1557, 68 residues downstream of HDS3 Loss of 18 nt resulting from use of alternate 5' donor site within exon 33 during splicing to exon 34 New T and loss of 33 residues at 1607, 118 residues downstream of HDS3 Loss of 96 nt resulting from use of alternate 3' acceptor site within exon 35 during splicing with exon 34 BIG2 Frame-shift at 1542, 106 residues downstream of HDS3 Loss of exon 35 resulting from splicing of 5'donor site of exon 34 with 3' acceptor site of exon 36 a All changes were expressed relative to the reference sequence stored under accession number NM_004193 (hGBF1), NM_006421 (hBIG1) and NM_006420.1 (hBIG2). b All variants are supported by one or more cDNA/ESTs as detailed in the Aceview for each gene that can be obtained at [38]. Evolution of BIGs and GFBs from a common ancestor Combined, our analysis reveals that the BIG and GBF subfamilies share the same overall domain organization, and are likely to descend from a common ancestor gene that duplicated first to form the BIG and GBF groups, and again within these groups to yield species-specific BIG and GBF members. These two subfamilies can therefore be unified as a higher order ArfGEF subfamily (called below GBG for G BF/ B IG G EFs), from which unrooted phylogenetic trees can be built (Figure 8 ). Unlike previous phylogenetic analysis which compared ArfGEFs based on their Sec7 domains after diverging non-catalytic regions have been trimmed [ 8 ], our trees were established from the simultaneous alignment of all 6 conserved domains (DCB, HUS, Sec7, HDS1, HDS2 and HDS3), excluding variable linkers. The same tree topology was obtained with both neighbor-joining and maximum likelihood methods, and was retained using any one of the new conserved domains alone (data not shown). Bootstrap analysis strongly supported this topology for most branches. Only a few small branches located at the base of the groups were found in less than 60% of the trials in one of the two methods, but this never occurred with both methods simultaneously. Figure 8 Unrooted neighbour-joining phylogenetic tree of the BIG/GBF subfamily. Colour coding for the main groups is green for plants, marine blue for fungi, orange and red for animals, cyan for protists. Branches found in less than 60% bootstrap trials by either the neighbor-joining or the maximum likelihood method are in dotted lines. Species abbreviations are as in Table 1. The tree topology strongly suggests that in most organisms, GBG members sort in separate branches, corresponding to their classification in BIGs and GBFs. Remarkably, our annotation of Sec7-containing proteins in the genome of Paramecium reveals the first departure from this distribution, as all GBGs in this species are located in a single branch, which is closer to the BIGs. This unexpected tree topology may indicate that alveolates diverged from animals/fungi and plants before the duplication of an ancestral GBG into the BIG and GBF, and that GBGs in that organism are representative of this ancestral gene. Alternatively, duplication may have been followed by loss of GBF genes. Current knowledge of the phylogenetic branching of alveolates relative to the plants and animal/fungi branches does not permit resolution between those two possibilities. GBGs in plants: refining the functional evolution of BIGs and GBFs In fungi and mammals, BIGs and GBFs are represented by only one or two members, whose functions in vesicular trafficking at the Golgi within each group appear largely overlapping [ 12 , 14 ]. In contrast, plants encode a large number of GBGs in both the BIG and GBF branches but lack other ArfGEFs (Figures 1 , 8 ). In Arabidopsis, none of the GBGs map to duplicated chromosomes where identical functions may be encoded [ 26 , 29 ]. In addition, comparative analysis with the rice genome nearing completion identifies at least five branches each represented by one rice and one or two Arabidopsis homologs (Figure 8 ). This correspondence between two highly divergent plant species indicates that GBGs diversified early during plant evolution, probably reflecting functional specialization along with the establishment of plant multicellularity. While GNOM has a plant-specific function in recycling plasma-membrane proteins needed for cell-cell communication and cell polarity establishment [ 11 ], possibly closer to the function of EFA6 or CYH subfamilies in metazoans, other plant GBGs are expected to fulfill the presumed ancient function of regulating Golgi trafficking exemplified by mammalian and yeast GBGs. Comparison of orthologous pairs in plants further reveals that they have different sensitivities to Brefeldin A (a widely used fungal inhibitor of Golgi traffic) as predicted from the sequences of the binding site of the drug carried by the Sec7 domain [ 6 ]. This observation clearly illustrates that differences in outcome following BrefeldinA treatment may not reflect differences in underlying molecular mechanisms, but instead simply reflect neutral sequence differences at the Sec7 domains between species. In particular, not all BIGs may be BFA-sensitive or GBFs BFA-resistant, unlike suggested by their original nomenclature. A novel ArfGEF subfamily in alveolates A remarkable evolutionary feature of ArfGEFs is that while GBGs seem to be ubiquitous to all eukaryotes, fungi and animals kingdoms evolved their own ArfGEFs subfamilies unrelated to those of the other kingdoms. We thus addressed the question of whether Paramecium, which has a large number of GBGs (at least five, of which four are present as pairs as the result of recent duplications) but appears to lack the specialization into the BIG and GBF subgroups, has the same ArfGEF distribution as plants or features a second ArfGEF subfamily. We thus searched the newly sequenced genome from Paramecium tetraurelia and the available alveolate genomes from Cryptosporidium parvum and Tetrahymena thermophila for additional Sec7-containing proteins. This identified a novel putative ArfGEF subfamily characterized by the association of the Sec7 domain with a TBC (Tre/Bub2/Cdc16) domain (Figure 9 ), which was found only in the protists kingdom. The TBC domain is predicted to carry a GAP (GTPase activating protein) activity towards small G proteins of the Rab family [ 30 ], suggesting a potential crosstalk between Rab and Arf pathways. Such a relationship between these two small G proteins families, which are major regulators of membrane traffic, would not be unprecedented, as for example the SYT1 ArfGEF gene was identified in yeast by its genetic interactions with Rab proteins in the exocytic pathway [ 31 ]. Interestingly, alveolates have specialized exocytic pathways based on a membrane organelle lying beneath the plasma membrane, the trichocyst, where this unique ArfGEF family may potentially function. Figure 9 TBS: a novel ArfGEF subfamily in alveolates. Top: Domain structure of the TBS subfamily. Below: Sequences of the TBC domain from Paramecium TBS aligned with TBC domains from known RabGAPs. Secondary structures are from the crystal structure of yeast GYP1 [30]. Conclusion A conserved scenario for the activation of Arf proteins by their GEFs? The identification of a conserved modular architecture in all GBG subfamily members suggests that the mechanistic basis for their activation of Arf is likely to follow a similar scenario. Candidate functions for the conserved domains include oligomerization, the collection of input signals, membrane localization, regulation of the exchange activity, scaffolding of Arf proteins to their downstream effectors, not excluding signaling to partners outside the Arf pathways. Dimerization has been reported in the BIG subgroup for BIG1, which forms heterodimers with the highly homologous BIG2 ArfGEF [ 14 ], and in the GBF subgroup for GNOM, which forms homodimers [ 27 ]. The conservation of the DCB domain in GBGs, which is responsible for the dimerization of GNOM, suggests that such a dimerization function may be general to this domain in GBGs. Another unresolved issue is the conservation of the cellular partners effecting the functions associated with the conserved domains. Our identification of an almost invariant motif in the HUS domain argues in favor of this domain interacting with a conserved partner. However, the ancient divergence into the BIG and GBF groups and their subsequent divergence into species-specific members suggest that specialized requirements are likely to have evolved in most organisms, possibly yielding less conserved partners outside the Sec7 and HUS domains. Finally, whereas in plants all ArfGEFs are predicted to function according to the scheme defined by the conserved domains, other species have additional ArfGEF subfamilies with a modular architecture unrelated to that of the GBG subfamily. It is not known to what extent the GBG's scenario for Arf activation will also apply to non-GBGs ArfGEFs, acting alone or in association with protein partners. In the case of the GBGs, our definition of the structural homology domains as reported here should now provide a robust background for future investigations of their interactions and functions. Methods Protein sequence databases were searched with amino acid sequences from human BIG1, human GBF1 and Arabidopsis GNOM using the BLAST algorithm [ 32 ]. Paramecium tetraurelia genes were identified with the BLAST algorithm using genome sequence data from Genoscope [ 33 ] and manually annotated using Artemis [ 34 ]. Tetrahymena sequences were retrieved from the Tetrahymena thermophila genome sequencing project server [ 35 ]. Arabidopsis sequences were retrieved from the Arabidopsis Genome Initiative database [ 36 ], rice sequences from the TIGR Rice annotation project [ 37 ]. Splice variants for hGBF1, hBIG1 and hBIG2 were identified from information provided under Aceview in the December (03) release for their respective listings at the NCBI [ 38 ]. Multiple sequence alignments were performed using ClustalW [ 39 ] with default alignment parameters or T-coffee [ 40 , 41 ]. Reliability of the alignments was evaluated according to the T-coffee score, and ranged from average to good for all predicted domains. Average sequence identities were respectively 24 % (DCB domain), 26 % (HUS domain), 44% (Sec7 domain), 26% (HDS1 domain), 28% (HDS2 domain) and 21% (HDS3 domain). Aligned sequences were displayed with ESPript [ 42 ] using a similarity global score of 0.15 calculated using the BLOSUM62 matrix. Unrooted phylogenetic trees were generated using the neighbor-joining algorithm of ClustalW excluding gapped regions, and with a maximum likelihood method using the PHYML package [ 43 ]. Phylogenetic trees for individual domains was performed on the subset of sequences used in Figure 3 . The reliability of the trees was assessed by a bootstrap analysis (1000 replicates). Trees were drawn with TreeView version 1.6.6. Secondary structure predictions on aligned sequences were carried out with the PHD program along with the ClustalW multiple alignment [ 39 ]. Non-structured linkers poor in hydrophobic residues were predicted with the PONDR algorithm [ 44 ]. Abbreviations GEF: Guanine nucleotide exchange factor. CYH: cytohesins/ARNO; EFA: Exchange Factor for Arf6; FBS: F-Box/Sec7; TBS: TBC/Sec7; GBF: Golgi-associated BFA-resistant guanine nucleotide exchange Factor; BIG: BFA-Inhibited Guanine nucleotide exchange factor; GBG: GBF/BIG Gefs; SYT1: S uppressor of ypt . DCB: Dimerization/Cyclophilin Binding; HUS: Homology Upstream of Sec7; HDS: Homology Downstream of Sec7; TBC: Tre/Bub2/Cdc16; SCF: Skp1/Cull1/F box. Authors' contributions B.M. and V.B. carried out sequence and phylogenetic analysis. A.J. participated in the domain analysis. J.Co. annotated Paramecium sequences. D.S. and P.M. performed splicing pattern analysis. N.K. and G.J. analyzed the distribution of large ArfGEFs in plants. J.Ch. conceived and coordinated the study and wrote the manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC553965.xml |
524253 | Motifs in Brain Networks | Complex brains have evolved a highly efficient network architecture whose structural connectivity is capable of generating a large repertoire of functional states. We detect characteristic network building blocks (structural and functional motifs) in neuroanatomical data sets and identify a small set of structural motifs that occur in significantly increased numbers. Our analysis suggests the hypothesis that brain networks maximize both the number and the diversity of functional motifs, while the repertoire of structural motifs remains small. Using functional motif number as a cost function in an optimization algorithm, we obtain network topologies that resemble real brain networks across a broad spectrum of structural measures, including small-world attributes. These results are consistent with the hypothesis that highly evolved neural architectures are organized to maximize functional repertoires and to support highly efficient integration of information. | Introduction The complex vertebrate brain has evolved from simpler networks of neurons over a time span of many millions of years. Brain networks have increased in size and complexity ( Jerison 1973 ; Butler and Hodos 1996 ; Kaas 2000 ; Krubitzer 2000 ), as have the flexibility of interactions with the environment and the range of potential behaviors that can be generated ( Changizi 2003 ). Most of the rules governing the evolutionary process toward more complex brains are still unknown, although the central roles of modularization ( Kaas 2000 ), conservation of wiring length ( Cherniak 1994 ; Chklovskii et al. 2002 ), and of the elaboration of network connectivity ( Laughlin and Sejnowski 2003 ) are becoming increasingly evident. Systematic investigations of neuronal connectivity in the nematode Caenorhabditis elegans ( White et al. 1986 ) and of large-scale interregional pathways in the mammalian cerebral cortex of rat ( Burns et al. 2000 ), cat ( Scannell et al. 1995 ; Scannell et al. 1999 ; Hilgetag et al. 2000 ; Kötter and Sommer 2000 ), and macaque monkey ( Felleman et al. 1991 ; Young 1993 ; Hilgetag et al. 2000 ; Stephan et al. 2000 ) have demonstrated that the topology of these networks is neither entirely random nor entirely regular. Instead, analysis of structural and functional data has shown ( Hilgetag et al. 2000 ; Sporns et al. 2000 ; Stephan et al. 2000 ; Sporns and Zwi 2004 ) that these networks can be characterized by a high degree of clustering, with short path lengths linking individual components, thus exhibiting small-world properties ( Watts and Strogatz 1998 ; Watts 1999 ) as do many other complex networks ( Strogatz 2001 ; Albert and Barabasi 2002 ). These structural attributes are instrumental in generating functional specialization ( Zeki 1978 ; Passingham et al. 2002 ) and functional integration ( Bressler 1995 ; Tononi et al. 1998 ; McIntosh 2000 ; Varela et al., 2001 ; Friston 2002 ), and they support a large repertoire of complex and metastable dynamical states ( Bressler and Kelso 2001 ; Sporns and Tononi 2002 ; Sporns 2004 ). Fluctuating and distributed patterns of dynamical interactions among functionally specialized areas result in rapid switches in functional and effective connectivity ( McIntosh et al. 1999 ; Büchel and Friston 2000 ; McIntosh et al., 2003 ; Brovelli et al. 2004 ). The structural and functional anatomy of brain networks reflects the dual challenges of extracting specialized information and integrating the information in real time ( Tononi and Sporns 2003 ). What rules underlie the organization of the particular types of networks that we see in complex brains? It is likely that, as networks become more complex, already existing simpler networks are largely preserved, extended, and combined, while it is less likely that complex structures are generated entirely de novo. One hypothesis states that complex and highly evolved networks arise from the addition of network elements in positions where they maximize the overall processing power of the neural architecture. This could be achieved by increasing the number of existing processing configurations or by introducing new processing configurations that add to the robustness or range of cognitive and behavioral repertoires. We may gain insight into the rules governing the structure of complex networks by investigating their composition from smaller network building blocks. Those building blocks are called “motifs” (in analogy to driving elements that are elaborated in a musical theme or composition), and they have been examined in the context of gene regulatory, metabolic, and other biological and artificial networks ( Milo et al. 2002 ; Milo et al. 2004 ). Motifs occur in distinct motif classes that can be distinguished according to the size (M) of the motif, equal to the number of nodes (vertices), and the number and pattern of interconnections. For a more formal definition of motifs and related concepts, see Materials and Methods . While the most common definition of network motifs is based on their structural characteristics ( Milo et al. 2002 ), structural motifs of neuronal networks form the physical substrate for a repertoire of distinct functional modes of information processing. In brain networks, a structural motif may consist of a set of brain areas and pathways that can potentially engage in different patterns of interactions depending on their degree of activation, the surrounding neural context or the behavioral state of the organism. Thus, we propose a distinction between structural and functional motifs. Structural motifs quantify anatomical building blocks, whereas functional motifs represent elementary processing modes of a network ( Figure 1 ). In this paper, functional motifs refer to specific combinations of nodes and connections (contained within structural motifs) that may be selectively recruited or activated in the course of neural information processing. Sorting all possible structural motifs within a network as a function of motif class yields a motif frequency spectrum that records the number of distinct motifs in each structural motif class. Given the motif frequency spectrum, one can easily obtain the motif number, defined as the total number of distinct occurrences of any motif of size M, and the motif diversity, defined as the number of classes that are represented within the network by at least one example. Figure 1 Definition of Structural and Functional Motifs, and Motif Detection (A) From a network, we select a subset of three vertices and their interconnections, representing a candidate structural motif. (B) The candidate motif is matched to the 13 motif classes for motif size M = 3. Numbers refer to the ID. The candidate motif is detected as a motif with ID = 13. In detecting structural motifs, only exact matches of candidate motif and motif class are counted. (C) A single instance of a structural motif contains many instances of functional motifs. Here, a structural motif ( M = 3, ID = 13) is shown to contain, for example, two distinct instances of the functional motif ID = 9, one motif ID = 2, and one motif ID = 7. Many other distinct instances of functional motifs are present that are not shown in the figure. Note that, in order to be counted as a functional motif of size M = 3, all three vertices of the original structural motif must participate. For a very similar distinction between structural and functional motifs (“interlaced circuits”) and an illustration see Ashby (1960) , p. 53. Clearly, the number of vertices (N) and edges (K) within a large network has a strong effect on the motif number and diversity of its constituent structural and functional motifs. But even if N and K are held constant, different connection patterns will result in different repertoires of such network motifs, expressed in terms of both number and diversity. These considerations lead us to formulate hypotheses concerning the rules for brain network organization in terms of network motifs. We hypothesize that neuronal networks have evolved such that their repertoire of potential functional interactions (functional motifs) is both large and highly diverse, while their physical architecture is constructed from structural motifs that are less numerous and less diverse. A large functional repertoire facilitates flexible and dynamic processing, while a small structural repertoire promotes efficient encoding and assembly. We investigate this hypothesis first by performing an analysis of structural and functional motifs in various brain networks. We compare the motif properties of real brain networks with random networks and with networks that follow specific connection rules such as neighborhood connectivity (lattice networks). We identify some motif classes that occur more frequently in real brain networks, as compared to random or lattice topologies. Second, by rewiring random networks and imposing a cost function that maximizes functional motif number, network topologies are generated that resemble real brain networks across a broad spectrum of structural measures, including small-world attributes. The results of our analyses are consistent with the hypothesis that complex brain networks maximize functional motif number and diversity while maintaining relatively low structural motif number and diversity. Results Motif Frequency Analysis We obtained complete structural motif frequency spectra for large-scale connection matrices of macaque visual cortex, macaque cortex, and cat cortex, for motifs sizes of M = 2, 3, 4, and 5 (estimations). In addition, we obtained motif frequency spectra for the matrix of interneuronal connections (“chemical synapses”) of C. elegans , for motif sizes M = 2, 3, and 4 (estimations). For each neural connectivity matrix we generated equivalent (N, K) random and lattice matrices, preserving degree distributions ( n = 100; see Materials and Methods ), and we obtained their structural motif frequency spectra for comparison. Thus, statistical significance of a motif can only be reached if it occurs in significantly increased proportions with respect to both random and lattice reference cases. Table 1 summarizes the data for structural and functional motif number. Large-scale connection matrices exhibit a consistent statistical trend. Their structural motif number is relatively low, and their functional motif number is relatively high, with both measures approaching the corresponding values of lattice networks. All of these brain networks contain a very high proportion of connected motifs (e.g., 53.2% for M = 3 in macaque visual cortex versus 24.6% in corresponding random networks). All neuronal networks (all cortical networks and C. elegans ) showed maximal functional motif diversity for all motif sizes examined (values of 2, 13, 199, and 9,364 for M = 2 to 5). Their structural motif diversity tended to be submaximal. For example, the structural motif diversity of macaque cortex was significantly reduced in comparison to random matrices (168, compared to 198 ± 1 for random networks at M = 4). This tendency was especially pronounced for higher values of M (e.g., 3,697 for macaque visual cortex, compared to 8,887 ± 112 for random networks at M = 5). Table 1 Structural and Functional Motif Number for Cortical Connection Matrices and Corresponding Random and Lattice Matrices Numbers are actual values (for real matrices) and mean and standard deviation (in parentheses, for random and lattice matrices, n = 100) Figure 2 shows motif frequency spectra for structural motifs ( M = 3, M = 4) found within the network of the macaque visual cortex and C. elegans and their corresponding reference cases. Spectra of macaque and C. elegans networks are both less similar to random networks than to lattice networks. For M = 3, in the case of the macaque visual cortex, some motif counts appear decreased over random networks (e.g., motif identity number [ID] = 1,…,6) while other motif counts appear increased (e.g., ID = 9) over both random and lattice networks. Table 2 and Figure 3 A summarize structural motifs whose motif counts were significantly increased in brain networks as compared to both random networks and lattice networks of identical degree distributions, for sizes M = 2, 3, and 4. Given motif frequencies from samples of n = 100 random or lattice networks, we calculated z -scores for the corresponding motifs in neuronal networks. Only structural motifs that were significantly increased ( z > 5.0, p < 0.0001) in real networks as compared to both random and lattice networks are tabulated. Despite variations in size, areal composition, species, and collating authors, specific motif classes consistently emerged across several different cortical networks. Figure 3 displays those structural motifs that were consistently encountered in all three cortical connection matrices. Particularly noteworthy is the consistent appearance of motif ID = 9 ( M = 3) in all cortical matrices examined in this study. The appearance of this motif cannot be explained by a higher proportion of reciprocal (mutual) edges (a motif of size M = 2): While random networks contain fewer such edges, lattice networks contain an equally high proportion of such edges (for example, macaque visual cortex has 69 single edges and 121 double edges, while a sample of 100 comparison lattice networks contains 70.6 ± 4.71 single edges and 120.2 ± 2.36 double edges). No motif of size M = 2 was significantly increased in frequency for any of the connection matrices in this study ( Table 2 ). Furthermore, other motifs containing double edges (e.g., ID = 6, 12, etc.) were not increased. A different set of significantly increased structural motifs was found for C. elegans . Motif ID = 9 was not significantly increased in frequency, while two other non-connected motifs (ID = 4 and 6) occurred more frequently than expected. Figure 2 Comparison of Structural Motif Frequency Spectra for Macaque Visual Cortex and C. elegans (A) Spectra for structural motifs of size M = 3. (B) Spectra for structural motifs of size M = 4. Figure 3 Structural Motifs that Occurred in Significantly Increased Numbers at Motif Sizes M = 3 and M = 4 (A) Structural motifs found in all three large-scale cortical networks analyzed in this study (see Table 2 ). (B) Structural motifs found in networks optimized for functional motif number (see Table 4 ). Numbers refer to the motif's ID. Table 2 Structural Motifs That Are Significantly Increased in Brain Networks See Figure 3 for displays of the significant motifs (shown with their ID). Note that no significant differences are found for any of the networks at M = 2. Numbers are giving actual values (for real matrices) and mean and standard deviation (in brackets, for random and lattice matrices, n = 100). All z -scores > 5.0, with a single exception noted by asterisk (cat, M = 3, lattice) Table 4 Significantly Increased Structural Motifs of Optimized Networks Compare motif ID with those shown in Figure 3 and Table 2 . As in Table 3 , all networks were optimized for high functional motif number ( M = 3, N = 30, K = 311, mean and standard deviation for n = 10 exemplars). Optimizations and comparisons of macaque and cat matrices produce similar results (unpublished data) Table 3 Structural Motif Number and Networks Optimized for Functional Motif Number All networks were optimized for high functional motif number ( M = 3, N = 30, K = 311, mean and standard deviation for n = 10 exemplars) Each vertex (brain area) participates in a subset of the structural motifs that compose the entire network. We asked whether individual brain areas participate in similar or different sets of motifs and whether motif participation might reveal functional relationships. We define the motif fingerprint of a brain area as the number of distinct structural motifs of size M that the area participates in. Motif fingerprints characterize brain areas, as do other structural and functional features ( Passingham et al. 2002 ), and they are directly related to other connectional metastructures forming various kinds of network participation indices (Kötter and Stephan, 2003). Figure 4 shows polar plots of motif fingerprints ( M = 3) for several visual areas of macaque visual cortex. Motif ID = 9 was the only motif found to be significantly increased over both random and lattice networks, but it was increased for only five visual areas (V1, V3, V4, MSTd, and DP). All of these areas showed highly similar motif fingerprints characterized by a specific ratio of motif classes 9, 12, and 13 ( Figure 4 A and 4 C). Other areas, such as V2, V4t, and PITv show very different motif fingerprints ( Figure 4 A), and cluster analysis reveals them as members of clusters of visual areas participating in a different set of motifs. For example, most inferotemporal areas as well as visually related prefrontal areas 46 and FEF belong to a separate cluster with motif fingerprints that differ significantly from those of all other cortical areas ( Figure 4 B). Figure 4 Motif Fingerprints for Motif size M = 3 in Macaque Visual Cortex (A) Motif fingerprints for five areas with significantly increased motif ID = 9 (V1, V3, V4, MSTd, DP, names in bold) as well as areas V2, V4t, and PITv. Polar plots display the motif participation number for 13 motif classes with M = 3 (see Figure 1 ). Note that, despite differences in the absolute motif participation numbers, areas V1, V3, V4, MSTd and DP show highly similar motif fingerprints. (B) Hierarchical cluster analysis of motif fingerprints. The Pearson correlation coefficients between all pairs of motif fingerprints were used in a consecutive linking procedure using Euclidean distances based on the farthest members of each cluster (for details see Kötter and Stephan [2003] ). Areas with more similar motif fingerprints are linked at smaller distances. The five areas with significantly increased motif ID = 9 are indicated in bold typeface. (C) Hierarchical cluster analysis of single area motif frequency spectra using the same procedures on orthogonal data of (B). Motif classes 9, 12, and 13 covary across the 30 visual areas and form a distinct branch of the cluster tree. Optimization of Motif Number We hypothesized that high functional motif number and diversity represent important ingredients in the global organization of cortical networks, and that a selective advantage for these two properties might contribute to the generation of other significant structural properties. To test this hypothesis we applied an evolutionary algorithm ( Sporns et al. 2000 ) that selects for networks with high functional motif number, while rewiring their connectivity. All simulations were carried out with networks of size N = 30, K = 311 (matching macaque visual cortex), in generations of 10 individuals, with a low rewiring rate of one connection per generation and a survivor rate of one network per generation, over 2,000 generations. Convergence was robust and consistent structural features of optimized connection matrices were observed. Figure 3 B, Figure 5 , Table 3 , and Table 4 summarize results obtained from the optimizations. When maximizing functional motif number ( Figure 5 A), we obtained networks that closely resembled real brain networks with respect to their structural and functional motif number, motif diversity (unpublished data), structural motif frequency spectrum, and the specific structural motifs that occurred with significantly increased frequency ( Tables 3 and 4 ). Optimizing functional motif number invariably resulted in a significant decline in the number of structural motifs. Figure 3 B illustrates the set of structural motifs that appeared in significantly increased numbers after optimizing functional motif number. Note the appearance of motifs that are identical or highly similar to those obtained from an analysis of large-scale cortical matrices. These structural similarities are observed for the motif size at which the networks were optimized ( M = 3) as well as at lower and higher motif sizes ( Table 4 ). In contrast, when maximizing structural motif number, we obtained networks with strikingly different structural attributes ( Figure 5 B) that bore no resemblance to real brain networks. We found no overlap with real networks of significantly enhanced motifs at any of the motif sizes we examined. Figure 5 Properties of Networks ( n = 10) Optimized for Structural and Functional Motif Number (A) Maximization of functional motif number ( N = 30, K = 311). Each maximization starts from different random initial conditions, including a different set of 10 random networks. From left to right, each graph shows plots of functional motif number, structural motif number, motif frequency spectrum ( M = 3) of optimized networks, and clustering coefficient. (B) Maximization of structural motif number ( N = 30, K = 311). Graphs are as in (A). Compare the motif frequency spectrum in (A) with the corresponding plot for the macaque visual cortex in Figure 2 A (first row, left bar graph). Initially, random networks in generation 1 exhibited frequency spectra identical to those for random networks in Figure 2 A (first row, middle panel). To further characterize these networks, we calculated their clustering coefficient and their path length to determine if they exhibited small-world properties ( Figure 5 ). We found that networks that maximized functional motif number also had clustering coefficients that were much higher than those of random networks (γ = 0.5288 ± 0.0201 for optimized networks; γ = 0.4323 ± 0.0073 for random networks), while their path lengths remained relatively short (λ = 1.7891 ± 0.0275 for optimized networks; λ = 1.9300 for a nearest-neighbor lattice network). Both measures closely approximated those of macaque visual cortex (γ = 0.5313, λ = 1.7256). In contrast, networks that maximized structural motif number had clustering coefficients that were indistinguishable from those of random networks (γ = 0.4273 ± 0.0029), and were significantly lower than that of macaque visual cortex. Discussion The importance of a large repertoire of functional circuits for flexible and efficient neural processing has long been recognized ( Walter 1953 ; Ashby 1960 ) and has recently received renewed theoretical and experimental attention ( Tononi et al. 1999 ; Tononi and Sporns 2003 ). In this paper we investigate the building blocks of brain networks and how their composition and topological patterning enables flexible neural function. Our hypotheses and analysis rest upon a fundamental distinction between structural and functional motifs. In this work, functional motifs refer to the different patterns or combinations of nodes and connections that could occur within the constraints of a given structural motif. We do not assume anything about their function, or which functional motif is actually selected by physiological mechanisms. We only assume that a particular structural motif is necessary to support a repertoire of functional motifs that may, or may not, be called upon for neuronal computations. Our hypothesis is that the connection patterns of real brain networks maximize functional motif number and diversity, thus ensuring a large repertoire of functional or effective circuits, while they minimize the number and diversity of structural motifs, thus promoting efficient assembly and encoding. We observe that the functional motif number of a variety of real brain networks is very high compared to equivalent random networks, while their structural motif number is comparably low. We then demonstrate that optimization of functional motif number can yield networks that resemble real brain networks in several structural characteristics, including their motif frequency spectra, motifs that occur in significantly increased numbers, and small-world measures. The functional implications of some network structures—such as reciprocal, convergent, and divergent connections or cycles—have been discussed in the context of network participation indices ( Kötter and Stephan 2003 ) and network complexity ( Sporns et al. 2000 ). Various large-scale cortical connection matrices examined in this study and collected by different authors and from different species, exhibit striking commonalities in their global patterning and motif compositions. Particularly interesting is the increased occurrence of a single motif at M = 3 (ID = 9; see Figure 3 ) and its expanded versions at M = 4 (ID = 46, 95, 148, 178). These motifs essentially form of a chain of reciprocally connected units, while pairs of connections linking the ends of the chain are absent. In functional terms, units in these motifs are highly integrated with their neighbors, while some pairs of units remain more segregated from each other and do not communicate directly. Thus, this motif type combines two major principles of cortical functional organization, integration and segregation ( Tononi et al. 1998 ; Friston 2002 ), and it may be associated with a specific type of neural dynamics ( Zhigulin 2003 ). The occurrence of this motif type is not due to an artifact of recording or collating connection pathways, as it also appears in increased proportion in optimized and rewired networks (see Table 4 ). In contrast to large-scale cortical networks, the invertebrate network of C. elegans exhibits very different patterns that are less indicative of high integration and segregation. At M = 3, motif ID = 9 does not occur in higher-than-expected numbers, while other motifs (ID = 4 and ID = 6) are increased. Our results suggest that large-scale cortical connection matrices form a distinct family ( Milo et al. 2004 ) of networks that can be characterized by their motif frequency spectra, while invertebrate neuronal networks do not appear to belong to this family. Optimizing functional motif number yields networks that resemble real brain networks across a broad spectrum of structural measures, including several that did not appear to be linked in trivial ways to the optimized measure. Increasing the functional motif number tends to lead to a concomitant decrease in structural motif number, as individual connections become locally dense, thus increasing the abundance of motifs with more local connections and thus greater functional diversity. We note that maximal numbers of functional motifs are not reached in ideal lattices (nearest-neighbor connectivity); rather, optimized networks routinely exhibit functional motif numbers that exceed those of ideal lattices, and they belong to a general class of networks that maintain a mixture of “local” and “long-range” connectivity. Importantly, even though structural and functional motifs are directly related (each structural motif contains a fixed set and spectrum of functional motifs), optimizing structural and functional motif number yielded strikingly different connection topologies. Optimizing functional (but not structural) motif number produced a tendency toward the emergence of small-world attributes (high clustering coefficient and short path length), a mode of connectivity that promotes functional cooperation, recurrent processing, and efficient information exchange ( Sporns et al. 2004 ). High clustering is due to “locally dense” connectivity promoting fewer, denser, and functionally more potent motifs. An admixture of “long-range” connections, which is compatible with achieving very high functional motif number, serves to maintain short minimal paths throughout the network. Interestingly, networks optimized for complexity ( Tononi et al. 1994 ; Sporns et al. 2000 ) also exhibit small-world attributes, conserve wiring length, and produce motif frequency spectra similar to those of networks optimized for functional motif number (including a significantly increased abundance of motif ID = 9, M = 3; unpublished data). In turn, networks optimized for functional motif number have significantly higher complexity than random networks, while those optimized for structural motif number are much less complex. Thus, it appears that several criteria for optimality (complexity, clustering coefficient, wiring length, functional motif number) favor similar global network architectures that are all characterized by two coexisting organizational principles, functional segregation and functional integration. The functional motif frequency spectrum provides a sophisticated way of characterizing subtypes of such networks geared at more specific functional modes of information processing. Materials and Methods Formal definitions All networks and network motifs in this paper are described as graphs of units (called nodes or vertices) with directed (i.e., nonsymmetrical) connections (called edges). A “motif” is a connected graph or network consisting of M vertices and a set of edges (maximally M 2 – M, for directed graphs, minimally M – 1 with connectedness ensured) forming a subgraph of a larger network. For each M there is a limited set of distinct motif classes. For M = 2, 3, 4, and 5, the corresponding numbers of motif classes are 2, 13, 199, and 9,364 ( Harary and Palmer 1973 ). See Figure 1 B for an illustration of the set of 13 motif classes for motifs of size M = 3. A “structural motif” of size M is composed of a specific set of M vertices that are linked by edges ( Figure 1 A). The resulting network of size M is called a “structural motif” because a larger network could be structurally assembled from a finite set of such motifs. Essentially, structural motifs form the structural building blocks of larger networks. Our definition of structural motifs is identical to the definition of motifs introduced in Milo et al. (2002) . A structural motif provides the complete anatomical substrate for possible functional interactions among its constituent vertices. However, in real neuronal networks, not all structural connections participate in functional interactions at all times. As different edges or connections become functionally engaged, different “functional motifs” emerge within a single structural motif. The former (functional) refers to “processing modes” or “effective circuits,” while the latter (structural) refers to “anatomical elements” or “building blocks.” The existence of different functional motifs greatly enhances the processing power of any neuronal architecture. We then distinguish structural motifs from functional motifs that form a set of subgraphs of the structural motif. All such functional motifs consist of the original M vertices of the structural motif, but contain only a subset of its edges (see Figure 1 C for examples). Note that a fully connected structural motif such as ID = 13 for M = 3 contains the maximal number of functional motifs. For each exemplar of a structural motif of a specific motif class, there is a fixed complement of constituent potential functional motifs (essentially forming a look-up table of potential functional circuits). Thus, the functional motif frequency spectrum is easily obtained from the structural motif frequency spectrum, without the need for additional motif detection. This definition implies that functional motifs are more naturally applied to networks with vertices that contain multiple neurons or neuronal populations. In the present study, our main focus is on motifs of large-scale connection matrices; data for the single neuron network of C. elegans are provided in Table 1 for statistical comparison only. A “connected motif” is a structural motif that forms a strongly connected graph. In a connected motif, all constituent vertices can be reached from all other constituent vertices. Such a motif, in principle, allows all vertices to exert causal effects on each other. For M = 3, motifs with ID = 7, 9, 10, 12, and 13 are connected motifs. A “motif frequency spectrum” records the number of occurrences of each motif of a given class for a size M . The motif frequency spectrum for structural motifs is obtained by motif detection. The motif frequency spectrum for functional motifs can be obtained from the structural spectrum by simple multiplication with the characteristic number of functional motifs for the respective structural motif. “Motif number” is the total number of all motifs of all classes (for a given size M ) encountered in a network. The motif number is obtained as the sum over the motif frequency spectrum, either structural or functional. “Motif diversity” is the total number of all motif classes (for a given size M ) encountered in a network. The motif diversity is obtained as the number of all motif classes for which the frequency spectrum is greater than zero. “Motif participation number” is the number of instances of a given motif class that a particular vertex participates in. For example, if a vertex participates in 12 distinct motifs with M = 3, ID = 13, it has a motif participation number of 12 for this particular motif. The “motif fingerprint” is the spectrum of motif participation numbers for all motifs of a given size M that a particular vertex participates in. The motif fingerprint is equivalent to a motif frequency spectrum for a single vertex of the network. Neurobiological data sets. All datasets used in this study are available in Matlab format at http://www.indiana.edu/~cortex/CCNL.html . Some of the matrices used in this study have been modified to remove areas with few known connections, or areas that are not part of the cerebral cortex. We note, however, that the nature of the data reported in this paper does not critically depend on these small changes, which usually affected only very small subset of the areas and connections. The connection matrix of the macaque visual cortex is based on Felleman et al. (1991) , and was modified as follows. The connections of areas {PITd, PIT, PITv}, {CITd, CIT, CITv}, and {STPp, STP, STPa} were consolidated by eliminating PIT, CIT, and STP and assigning their connections to {PITd, PITv}, {CITd, CITv}, and {STPp, STPa}, respectively. Areas MIP and MDP were eliminated due to lack of connectional information. The modified matrix has N = 30 and K = 311. The connection matrix of the macaque cortex is based on Young (1993) . Two areas, HIPP (the hippocampus) and AMYG (the amygdala) were deleted from the matrix, resulting in N = 71 and K = 746. The connection matrix of cat cortex was transcribed from Scannell et al. (1999) . For the large-scale analysis, density information was discarded and all pathways were encoded as either present or absent. For the analysis of intracortical pathways, we discarded the hippocampus and all thalamocortical pathways. The resulting matrix has N = 52 and K = 820. The connection matrix of C. elegans ( White et al. 1986 ) was retrieved from http://www.wormbase.org and is described at http://elegans.swmed.edu/parts/neurodata_readme.txt . It contains data for the nerve ring and very anterior section of the ventral cord for two individual hermaphrodite worms (JSH, N2U). We used data of all chemical synapses from both individuals, discarding data on gap junctions (electrical synapses), resulting in a matrix of N = 197 neurons and K = 1,974 directed connections. Other studies used matrices with N = 282 (Watts and Strogatz, 1998), N = 280 ( Milo et al., 2004 ), or N = 252 ( Milo et al., 2002 ). Despite these variations, our results on motifs in C. elegans are consistent with those of these earlier studies. Currently available datasets are likely to contain errors or missing connections that have not been investigated and do not take into account possible intersubject variability or rank-ordered or graded connection densities or strengths. While these issues have not been addressed systematically, some exploratory analyses suggest that the results reported in this paper are invariant with respect to small variations in connection patterns. Reference cases: random and lattice networks. A statistical evaluation of motif frequencies depends on a choice of reference cases (“null hypotheses”). Milo et al. (2002) generated random networks with identical structural motif frequencies at level M – 1 in order to perform statistical comparisons at level M . This corrected for the “carrying over” of significant motif components from lower to higher levels and allowed detection of the level of M at which significant structures emerged. The choice of reference cases in this paper reflects the specific question we ask about motifs in brain networks: Independent of the level M, how do the motif number, diversity, and composition of real brain networks compare to other network topologies, specifically to both random and lattice networks? We constrain the comparison by fixing the size of the networks (N and K) and by imposing equal degree distributions on all comparison networks (see also Milo et al. 2004 ). We note that the additional reference case of the lattice network led to the exclusion of motifs that occur in increased numbers simply because of local clustering of connections ( Artzy-Randrup et al. 2004 ; Milo et al. 2004a ). Random and lattice matrices that preserve the in-degree and out-degree for each vertex are generated from the original anatomical connection matrices by a Markov-chain algorithm ( Maslov and Sneppen 2002 ; Milo et al. 2002 ). For random matrices, a pair of vertices ( i 1 , j 1 ) and ( i 2 , j 2 ) is selected for which c i 1 j 1 = 1, c i 2 j 2 = 1, c i 1 j 2 = 0, and c i 2 j 1 = 0. Then we set c i 1 j 1 = 0, c i 2 j 2 = 0, c i 1 j 2 = 1, and c i 2 j 1 = 1. This is repeated until the connection topology of the original matrix is randomized. For lattice matrices, the same Markov procedure is employed but swaps are only carried out if the resulting matrix has nonzero entries that are located closer to the main diagonal (thus approximating a lattice or ring topology). This algorithm is implemented as a probabilistic optimization using a weighted cost function. Numerical methods. All graph theory methods used in this paper—including those for calculating clustering coefficients and path lengths ( Sporns 2002 )—as well as motif detection algorithms are available in Matlab format at http://www.indiana.edu/~cortex/CCNL.html . In some cases, for large networks or high values of M, we employed random sampling to estimate motif frequency spectra and their associated values for motif number and diversity. We selected different sample sizes to ensure convergence of these estimates and performed up to ten separate runs to generate good estimates. The evolutionary algorithms used in this study for optimizing structural and functional motif numbers of networks were similar to the algorithm described in Sporns et al. (2000) . Briefly, motif number was calculated for generations of ten individuals. The single individual with the highest motif number was selected and copied; all other individuals were deleted. The next generation was composed of the single survivor and nine rewired copies (using a rewiring rate of one connection). The first generation was composed of ten random networks. The rewiring procedure typically proceeded for 2,000 generations, changing only the connection pattern or topology. N, K, and the original degree distribution were conserved. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC524253.xml |
524521 | Intracardiac echocardiography to guide transseptal catheterization for radiofrequency catheter ablation of left-sided accessory pathways: two case reports | Intracardiac echocardiography (ICE) is a useful tool for guiding transseptal puncture during electrophysiological mapping and ablation procedures. Left-sided accessory pathways (LSAP) can be ablated by using two different modalities: retrograde approach through the aortic valve and transseptal approach with puncture of the fossa ovalis. We shall report two cases of LSAP where transcatheter radiofrequency ablation (TCRFA) was firstly attempted via transaortic approach with ineffective results. Subsequently, a transseptal approach under ICE guidance has been performed. During atrial septal puncture ICE was able to locate the needle tip position precisely and provided a clear visualization of the "tenting effect" on the fossa ovalis. ICE allowed a better mapping of the mitral ring and a more effective catheter ablation manipulation and tip contact which resulted in a persistent and complete ablation of the accessory pathway with a shorter time of fluoroscopic exposure. ICE-guided transseptal approach might be a promising modality for TCRFA of LSAP. | Background Trans-catheter radiofrequency ablation (TCRFA) has become the treatment of choice for patients suffering from refractory to medical treatment supraventricular tachycardias [ 1 , 2 ]. During percutaneous ablation procedures, catheter location is usually monitored by using fluoroscopy together with the analysis of intracardiac electrograms in order to clarify the mechanisms underlining the arrhythmia and to both locate its origin and record its circuit. This technique may be adequate for several standard ablative procedures, but it has still some limitations concerning the treatment of more complex tachycardia forms. Left-sided accessory pathway (LSAP) may be ablated using two different modalities: conventional approach through the aortic valve, or transseptal puncture of the fossa ovalis. By using the traditional approach, the left atrium is reached by a retrograde way through the left ventricle and crossing the mitral valve [ 3 , 4 ] whereas, with the transseptal puncture, the mitral ring is reached by an anterograde approach. For this reason, this approach has been considered an alternative and complementary technique for the transvenous introduction of catheters into the left cavities of the heart [ 5 - 8 ]. However, transseptal puncture under fluoroscopic guidance alone, might be hampered by some acute and potentially lethal complications that may be challenging even for expert electrophysiologists [ 9 ]. With the technological and miniaturization advances of low frequency transducers capable of enhanced tissue penetration, intracardiac echocardiography (ICE) has become feasible and potentially useful for guiding transseptal puncture and ablation procedures, especially when location of specific anatomic landmarks appears to be more crucial [ 10 , 11 ]. In this manuscript, two cases of ICE-guided catheter ablation of LSAP via transseptal approach have been described. Case Presentation Case 1 A 24 year-old woman with Wolf-Parkinson-White (WPW) syndrome and recurrent episodes of sustained supraventricular tachycardia, refractory to medical therapy, was referred to our Department in order to attempt TCRFA procedure. The electrophysiological study was performed off therapy using three diagnostic catheters (one decapolar and two quadripolars) which have been positioned in the coronary sinus (CS1 → CS5 = distal → proximal coronary sinus), His bundle region, and high right atrium/right ventricular apex. The clinical arrhythmia was diagnosed as an orthodromic atrio-ventricular re-entrant tachycardia (AVRT) due to a LSAP that was repetitively induced by both right atrial and ventricular programmed electrical stimulation. The length of the tachycardia cycle ranged from 310 to 300 msec. A well-localized accessory pathway insertion was detected in the lateral left free wall with atrial-ventricular (AV) and ventricular-atrial (VA) intervals of fusion noticeable only in CS2 (Fig 1 ). During programmed atrial stimulation and AVRT, the transaortic approach was initially attempted by using a 7F-4 mm bidirectional ablation catheter (Blazer HTD, Boston Scientific). Although very short VA and AV intervals (<50 msec without isoelectric line between the two potentials) and satisfactory mean temperature (54–56 C° for 20–30 sec with less than 30 Watts for each radiofrequency application) were obtained, repeated radiofrequency erogations resulted in a transient interruption of the anomalous connection, with recurrence of retrograde conduction along the accessory pathway and atrioventricular re-entrant tachycardia inducibility after a period of 20–30 min. Therefore, by using a mechanical single element ultrasound 9F-9 MHz catheter, an ICE-guided transseptal approach was attempted in order to cranially map the mitral ring (Ultra ICE Boston Scientific/CVIS San José, CA USA). This approach has provided a 360° two-dimensional imaging on a transverse plane perpendicular to the transducer with a radial field of view of approximately 5 cm in depth. ICE catheter was introduced via femoral vein through a 10F long sheath and advanced into the high right atrium. Careful handling of the catheter, along the interatrial septum, provided optimal view of the fossa ovalis that could be easily recognized as a thin area compared to the surrounding atrial structures. The standard approach to transseptal catheterization using a long sheath and dilator over a Brockenbrough needle, was used [ 5 ]. The tip of the needle-dilator-sheath apparatus was positioned facing the middle portion of the fossa ovalis. Careful up and down movements of ICE catheter allowed a clear visualization of the needle as well as its contact with the septal wall causing the typical "tenting" effect (Fig 2 ), which allowed precise location of the puncture of the needle. Figure 1 Orthodromic AV re-entrant tachycardia. Orthodromic AV re-entrant tachycardia induction with programmed atrial stimulation. Notice VA fusion on CS2 (lateral portion of the mitral ring; see arrow). Figure 2 Brockenbrough needle and the fossa ovalis. Brockenbrough needle in right atrium approaching the fossa ovalis (right panel); typical "tenting" of the fossa (see arrow) just before septal puncture (left panel). Note the left atrial free wall close to the interatrial septum. (FO = fossa ovalis; LA = left atrium; RA = right atrium). Accessory pathway was completely and definitively ablated with this approach after two erogations lasting each 40 sec. Although the AV and VA intervals that had been selected to deliver radiofrequency energy were similar to those previously recorded, the only difference was a better AV ratio (Fig 3 ). Such findings might be related to the cranial approach of the atrioventricular ring which has allowed a more stable catheter position on endocardium surface and a lack of lateral sliding, resulting in a higher mean temperature (60° ± 3° vs 54° ± 2°). Radiation exposure was shorter with the transseptal approach (11 vs 19 min) when compared to the retrograde approach. Figure 3 Ablation of the accessory pathway. Radiofrequency erogation during atrial pacing with maximal pre-excitation. Notice AV fusion (major arrow) followed by AV split (two minor arrows) on the ABLd recording, corresponding to lateral mitral annulus. Case 2 A 61 year-old man was referred to our Department for a TCRFA procedure due to a recurrent sustained supraventricular tachycardia refractory to antiarrhytmic drugs. The electrophysiological study has been performed by using the previously described protocol. An atrioventricular re-entrant tachycardia, due to an overt LSAP, was repetitively induced by atrial and ventricular programmed electrical stimulation. The length of the tachycardia cycle ranged from 340 to 320 msec. The accessory pathway insertion resulted well-localized in the left lateral mitral ring during AVRT and atrial pacing with maximal pre-excitation (Fig 4A,4B ). The trans-aortic approach was firstly attempted by using a 7F- 4 mm bidirectional ablation catheter (Blazer HTD, Boston Scientific). Despite an optimal catheter tip temperature for a reasonable period of time (five erogations reaching a mean temperature of 55° lasting 30–40 sec), only a temporary interruption of the anomalous pathway conduction was obtained. Therefore, ICE-guided transseptal approach was performed as previously described (see additional file: Movie 1 transseptal puncture.avi ). The approach resulted in a persistent and complete ablation of the accessory pathway after two successful radiofrequency erogations which have been delivered when the VA interval resulted fused and its ratio was 1,5:1. In this case, ICE allowed a more complete mapping of the mitral ring and a confirmation of effective catheter ablation tip contact. As in the previously described case, the superior approach to the left AV ring resulted in a better manipulation of the ablation catheter in addition to a reduction in fluoroscopic time (10 vs 21 min). Figure 4 Left sided accessory pathway. The shortest AV and VA are recorded in CS2 (see arrows) during programmed atrial stimulation (A) and atrio-ventricular orthodromic re-entrant tachycardia (B) respectively suggesting a left sided posterolateral accessory AV connection. Conclusion Intracardiac echocardiography and catheter ablation procedures During a standard ICE examination, sequential pull-back of the probe from the superior vena cava through the right atrium to the inferior vena cava allows a detailed identification of important structures such as: the crista terminalis; the right atrial appendage; the caval and the coronary sinus orifices; the fossa ovalis; the ascending aorta and its root; the pulmonary artery; the right ventricle and all of the cardiac valves [ 12 - 14 ]. Thus, this technique has the potential to provide a direct visualization of the endocardium and to precisely locate the ablation catheter which can be identified by the highly specific fan-shaped echocardiographic artefact of the large tip of the ablation electrode [ 11 , 12 ]. ICE is a well-recognized tool to guide ablation procedures. Compared to fluoroscopy, which does not provide definition of endocardial structures, ICE gives a highly accurate evaluation of firm and stable tissue contact. This results in a reduced radiofrequency power output. In addition, it is possible to visualize the lesion site and formation, such as swelling and crater formation. Moreover, possible complications such as occurrence of microbubbles, clot formation and pericardial effusion may be promptly detected through this ultrasound technique [ 10 , 15 ]. A recently developed phased-array intracardiac echocardiography device provides two-dimensional and Doppler images of the heart [ 10 ]. Recently, new strategies for ablation of atrial fibrillation (linear atrial ablation and focal ablation of triggers) have been proposed. Phased-Array intracardiac echocardiography enables a direct visualization of the pulmonary veins and allows the assessment of the Doppler flow velocity for all of the pulmonary veins [ 16 ]. This technique has been proved to be helpful in monitoring pulmonary vein isolation in patients with atrial fibrillation by improving the outcome and decreasing the incidence of complications such as pulmonary vein stenosis [ 17 ]. Intracardiac echocardiography and transseptal catheterization Transseptal catheterization is usually performed under fluoroscopic guidance [ 5 ]. However, safe transseptal puncture requires detailed visualization of the fossa ovalis that cannot be obtained through fluoroscopy. Thus, it remains a difficult procedure especially for patients which present anatomic abnormalities such as atrial and aortic root dilatation, musculoskeletal disorders and lipomatous hypertrophy of the atrial septum [ 18 ]. Furthermore, some complications of the transseptal catheterization due to accidental puncture of adjacent structures [ 6 ], such as atrial or aortic perforations, and pericardial tamponade, although rare, can be severe and life threatening. Moreover, in hypovolemic states, the left atrial free wall may be closed to the atrial septum. This potentially dangerous condition can be promptly recognized through the ICE technique, suggesting fluid administration in order to restore an adequate blood volume and to prevent left atrial free wall puncture [ 19 ]. The ability to visualize the anatomy of atrial septum and the localization of the fossa ovalis, may greatly enhance both the safety and the efficacy of the transseptal catheterization without any additional morbidity to the procedure [ 19 , 20 ]. As in the above-described cases, ICE allows a continuous monitoring of this procedure showing the tip of the sheath and Brockenbrough needle placed against the middle of the fossa ovalis immediately prior to the puncture [ 19 , 20 ]. Previous reports suggested the use of both transthoracic and transesophageal echocardiography, but these techniques present some limitations [ 21 , 22 ]. The former fails to display, in detail the fossa ovalis, the latter requires sedation which limits communication with the patient during the procedure and increases the risk of hypoventilation. Intracardiac echocardiography and left-sided accessory pathway LSAP represent the majority (59%) of all accessory pathway locations [ 23 ]. They can be ablated via transaortic approach. However, severe complications have been reported, including aortic dissection, lesion of the aortic valve, late endocarditis, peripheral and cerebral thromboembolic events [ 3 , 4 ]. Some authors, using the transseptal approach, have reported both a decreased procedural duration and radiation exposure, with higher success rate compared to the transaortic technique [ 7 , 24 ]. Two factors have been taken into account: 1) several left accessory pathways exhibit a broad or oblique insertion on the AV ring, so that they require a multisite radiofrequency energy delivery on both the atrial and ventricular sides in order to be ablated; 2) with the transaortic approach left atrio-ventricular ring mapping becomes a cumbersome procedure due to the retrograde crossing of the ablation catheter through the aortic and the mitral valves. In such a situation, the ablation catheter should be carefully manipulated in order to avoid entrapment into the mitral valve apparatus [ 4 ]. The use of ICE during ablation procedures in order to treat LSAP has been previously reported [ 25 ]. The observations pointed out by our case report have shown that ICE provides a precise localization of the needle tip position, a clear visualization of the fossa ovalis, a more complete mapping of the mitral ring and a more effective catheter ablation manipulation as well as tip contact. All of these advantages resulted in a persistent and complete ablation of the LSAP with a shorter time in fluoroscopic exposure in opposition to the transaortic approach. It has also been taken into account that a standard TCRFA procedure involves a radiation burden of about 17–25 mSv which appears to be equivalent to what is usually absorbed from natural radiation exposure in during a time period of 10 years and to about 1000 chest X-ray [ 26 ]. The combined use of ICE allows to reduce to a half the time of fluoroscopic exposure that induces a significant reduction in radiation load and consequent long term oncogenic risk both for patients and physicians [ 27 ]. In conclusion, ICE-guided transseptal approach might be a promising modality for TCRFA of LSAP. However, this report needs to be confirmed by further studies. List of abbreviations AV= atrial-ventricular; AVRNT= atrio-ventricular re-entrant tachycardia; CS= coronary sinus; ICE = intracardiac echocardiography; LSAP= left sided accessory pathway; TCRFA= transcatheter radiofrequency ablation; VA= ventricular-atrial; WPW= Wolf-Parkinson-White. Competing interests The authors declare that they have no competing interests. Authors' contributions RC carried out intracardiac echocardiography and drafted the manuscript, VD carried out electrophysiological study, AS drafted the manuscript, MS carried out electrophysiological study, MS drafted the revision of the manuscript, GG drafted the manuscript. All authors have read and approved the final manuscript. Supplementary Material additional file: Movie 1 transseptal puncture.avi Brockenbrough needle immediately after the puncture of the fossa ovalis can be clearly visualized. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC524521.xml |
549524 | Ancel Keys: a tribute | Ancel Keys, Ph.D., who died in November, 2004, at the age of 100, was among the first scientists to recognize that human atherosclerosis is not an inevitable consequence of aging, and that a high-fat diet can be a major risk factor for coronary heart disease. During World War II, he and a group of talented co-workers at the University of Minnesota conducted a large-scale study of experimentally-induced human starvation. The data generated by this study – which was immediately recognized to be a classic – continue to be of inestimable value to nutrition scientists. In his later years, Keys spent more time at his home in Naples, Italy, where he had the opportunity to continue his personal study of the beneficial effects on health and longevity of a Mediterranean diet. | Ancel Keys, who died in November, 2004, was an excellent testimonial to the health-promoting effects of his beloved Mediterranean diet. He lived to be 100 and, as the New York Times obituary put it, "remained intellectually active through his 97 th year." His latter years were spent mostly at his home in Naples, ItaIy. I never had the privilege of knowing him well, but encountered him occasionally at scientific meetings where we were both speakers. He was friendly but, I thought, reserved. What struck me about Ancel was his remarkable absence from the counsels of the nutrition establishment. Despite his acknowledged expertise and importance in the field, he was not a member of AMA's Council on Foods & Nutrition (at least not during the many years I served on that organization). I never saw him at any of the NIH advisory committees on which I served. He did not play a role in the deliberations of the Food and Nutrition Board of the National Research Council. He was not involved in the American Society for Clinical Nutrition during its heyday. Why was this? Perhaps the fact that he was a physiologist (later an epidemiologist) and not a physician played some role. Also, I think he preferred to go his own way, and – to some extent – he remained aloof from "academic nutrition." Yet he was willing to lecture to many audiences and was not considered to be a scientific eccentric; to the contrary, his epidemiological work was frequently cited and praised, and his monumental study of experimentally-induced semistarvation in human subjects [ 1 ] was immediately recognized to be a classic. Keys and his capable associates conducted careful physiological and psychological studies of 32 initially healthy conscientious objectors (to World War II) through 6 months of experimentally induced semistarvation, followed by a year or more of rehabilitation. These studies generated a cornucopia of data – data that are all the more valuable now because such an experiment would not have a chance of being approved by today's Institutional Review Boards. Protein-calorie malnutrition (PCM) – in effect, famine – remains endemic in many parts of the world; moreover, PCM is the most common nutritional problem encountered in U.S. hospitals and nursing homes. The studies carried out by Keys and his co-workers make it possible for us to distinguish the effects of semistarvation on the body's strength, composition, physiological status, and mood from the confounding effects of such underlying diseases as cancer, intestinal malabsorption, renal insufficiency, emphysema, etc. – illnesses that often give rise to conditioned PCM. The Minnesota group showed clearly that semistarvation can be independently responsible for an array of psychological problems such as anxiety, depression, and hypochondria. From their studies, it is possible to demonstrate a clear relationship between a decline in fat-free mass and PCM-associated morbidity. Keys's major scientific achievements are enumerated in some detail by Jane E. Brody in her New York Times obituary, dated November 23, 2004. For those of us who worked for so many years to call attention to the relationship of serum total cholesterol to risk of coronary heart disease (CHD), and to the cholesterol-raising effects of certain saturated fats, Keys will always be one of the major prophets who provided the early evidence that atherosclerosis is not an inevitable concomitant of aging, and that a diet high in saturated fat content can be a major risk factor for CHD. The practical outcome of the work in this field – to which Ancel contributed so much – is the extraordinary decrease in mortality from coronary heart disease that has occurred during the past half-century. Cancer has finally replaced heart disease as America's number one killer. Ancel had his well-deserved reward – a long, productive life unencumbered by an excess of committee meetings, and the opportunity to contemplate the Tyrrhenian sea while enjoying the benefits of a Mediterranean diet. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549524.xml |
520753 | Harpgophytum procumbens for osteoarthritis and low back pain: A systematic review | Background The objective of this review is to determine the effectiveness of Harpagophytum procumbens preparations in the treatment of various forms of musculoskeletal pain. Methods Several databases and other sources were searched to identify randomized controlled trials, quasi-randomized controlled trials, and controlled clinical trials testing Harpagophytum preparations in adults suffering from pain due to osteoarthritis or low back pain. Results Given the clinical heterogeneity and insufficient data for statistical pooling, trials were described in a narrative way, taking into consideration methodological quality scores. Twelve trials were included with six investigating osteoarthritis (two were identical trials), four low back pain, and three mixed-pain conditions. Conclusions There is limited evidence for an ethanolic Harpagophytum extract containing less than <30 mg harpagoside per day in the treatment of knee and hip osteoarthritis. There is moderate evidence of effectiveness for (1) the use of a Harpagophytum powder at 60 mg harpagoside in the treatment of osteoarthritis of the spine, hip and knee; (2) the use of an aqueous Harpagophytum extract at a daily dose of 100 mg harpagoside in the treatment of acute exacerbations of chronic non-specific low back pain; and (3) the use of an aqueous extract of Harpagophytum procumbens at 60 mg harpagoside being non-inferior to 12.5 mg rofecoxib per day for chronic non-specific low-back pain (NSLBP) in the short term. Strong evidence exists for the use of an aqueous Harpagophytum extract at a daily dose equivalent of 50 mg harpagoside in the treatment of acute exacerbations of chronic NSLBP. | Background Natives in the steppes of South and Southwest Africa use the secondary root tubers of Harpagophytum procumbens (H) for the treatment of various diseases, including musculoskeletal complaints. For more than half a century, various preparations from H have been continuously used in Europe and have become an established traditional treatment for rheumatic complaints. The monograph of the European Scientific Cooperative on Phytotherapy (ESCOP) [ 1 ] recommends H preparations for painful osteoarthritis and relief of low back pain in a dosage equivalent of up to nine grams of crude plant material and over a treatment period of at least two to three months. It has been suggested that the plant material should contain not less than 1.2% of the constituent harpagoside, an iridoid glycoside. Pharmacological studies indicate that in various animal models (e.g. the writhing test in mice) [ 2 ] the extract is more effective than its marker compound harpagoside. However, a number of contradictory findings make it difficult to draw definitive conclusions on the analgesic and anti-inflammatory effect of H preparations [ 3 ]. Recent in-vitro studies indicate that preparations from H may interact with the inflammatory cascade, including the cytokines [ 4 - 6 ]. Moreover, a significant decrease in stimulated production of matrix-degrading enzymes has recently been shown in isolated chondrocytes [ 7 ] and a dose-dependent weak elastase inhibition [ 8 ]. The objective of this review was to determine the effectiveness of Harpagophytum preparations in the treatment of musculoskeletal pain. Methods Searching Two reviewers (JG) and (SC) conducted electronic searches using the following databases: PUBMED (1966 up to September 16, 2003), EMBASE (OVID technologies: 1980 to wk 40 2003), Cochrane Controlled Trials Registry, Cochrane Musculoskeletal specialized register, Dissertation Abstracts, BIDS ISI, and the Cochrane Complementary Medicine Fields Specialized Register. The search strategy was developed by combining a highly sensitive method for isolating controlled clinical trials developed for the Cochrane Collaboration [ 9 ] [see appendix 1] with a variety of indexing and text words specific to the intervention ( H. procumbens) and musculoskeletal conditions. Search strategies were modified for each database [for PUBMED strategy see appendix 2]. One reviewer (SC) contacted experts and acquired relevant citations. In addition, manufacturers of commercial Harpagophytum preparations and content experts were contacted and asked to contribute published and unpublished material. Reference lists in review articles and the retrieved trials were searched for further trials. Study selection Trials that met the criteria outlined in Table 1 were included. Two individuals (JG & SC) independently reviewed titles and abstracts to determine study inclusion. A consensus method was used to resolve disagreements about inclusion of studies. Table 1 Inclusion criteria for considering studies for this review Types of studies Randomized controlled trials (RTCs), quasi-randomized controlled trials, and controlled clinical trials (CCTs) with no language restriction. Types of participants Adults suffering from pain in the musculoskeletal system due to osteoarthritis or low back pain. Types of interventions Studies utilizing preparations of Harpagophytum procumbens were included. Preparations may differ in the solvent (water, alcohol) used to prepare the extract (if not crude powdered plant material is used), the drug extract ratio, and the galenic aplication form. They also differ in the content of the active principles (the sum of active ingredients) and in the quantity of the co-active marker compound harpagoside (Chrubasik et al. 1996, Sporer and Chrubasik 1999). Types of outcome measures Primary outcome: pain (e.g. visual analogue scale, visual rating scale, pain component of the disease-specific Arhus Low Back Pain Index, component pain of the Western Ontario MacMaster (WOMAC) instrument). Secondary outcomes: number of pain-free patients (defined as being pain-free on at least five days in the last treatment week without taking any rescue medication see above), functional indices (e.g. Lequesne index, finger-ground distance), and generic outcome measures [global assessments, health assessment questionnaire (HAQ)] or the consumption of additional analgesic treatment. Data abstraction Two reviewers (SC, JG) extracted data from each trial using a standardized form. The following data were extracted from each study: authors' names; date of publication; country of origin; type of study, including number of study centers; participants (numbers, disease(s), characteristics of the study population (age, size, weight, gender)); duration of acute exacerbation or chronic disease; baseline values with details on pain and previous treatments; additional treatments; types of outcome measures; summary statistics; timing of outcome assessment; withdrawals and drop-outs; and adverse events. Blinding to authors, institution or journal title was not done, given the contradictory evidence relating blinded reviewing and bias [ 10 ]; additionally, two reviewers (JG and SC) were very familiar with the literature. Methodological quality assessment Methodological quality was assessed using the criteria list developed by van Tulder et al (1997; 2003) [ 10 , 11 ]. Specifically, the internal validity criteria A, B, C, E, F, G, H, I, J, K, L1, N, and O were used. Each criterion could be scored as yes (Y), no (N), or don't know (DK). The score of Y reflects the fulfillment of that criterion. The scoring of N reflects lack of fulfillment of that criterion. The scoring of DK reflects the inability to determine whether or not the criterion was fulfilled. High quality studies are defined as those that fulfill more than 50% (>6) of the quality criteria. Sensitivity analyses were carried out to explore the results when the definition of high quality trials was set at 40 % (>5) and 60% (>7) fulfillment of the quality criteria. According to the Van Tulder Scoring [ 10 , 11 ], the levels of evidence were defined as follows: 1. Strong – consistent findings among multiple high quality RCTs 2. Moderate – consistent findings among multiple low quality RCTs and/or CCTs and/or one high quality RCT 3. Limited – one low quality RCT and/or CCT 4. Conflicting – inconsistent findings among multiple trials (RCTs and/or CCTs) 5. No evidence from trials – no RCTs or CCTs Planned subgroup analyses included: (1) pain site, (2) type of pain (acute (≤ 6 weeks duration), sub-acute (6 to 12 weeks duration), and chronic pain (> 12 weeks)), and (3) comparison (botanical medicine (considering preparation form) versus placebo and botanical medicine versus other treatment). Results A total of 130 citations were isolated from electronic searches and all abstracts were retrieved. A total of three additional references were supplied by content experts. A total of 120 papers were excluded because of publication type (reviews or reports) or because of improper trial design. The complete references were retrieved for the remaining 13 trials. All 13 references were included in the current review [ 12 - 24 ]. Of these, two trials were duplicate publications [ 13 , 21 ], leaving 12 trials with unique data. A total of five randomized trials included 385 patients (range 46 to 122) with osteoarthritis of the hip or the knee [ 12 , 13 , 18 , 21 - 23 ]. Three of these trials were placebo controlled [ 12 , 18 , 22 ], and two were compared to standard pharmaceutical treatment forms [ 13 , 23 ]. The H preparations in these trials included a powder of crude plant material [ 13 , 22 ], a 60% ethanolic extract (solvent 60% ethanol)[ 12 , 18 ], and an aqueous extract [ 23 ]. A total of four trials included 505 patients (range 88 to 197) with acute exacerbation's of chronic non-specific low back pain [ 14 - 17 ]. All were randomized controlled trials, with two using placebo control [ 14 , 16 ], one using various conventional treatment controls (e.g. NSAIDs, exercise, massage, nerve blocks, acupuncture, etc.) [ 15 ], and one using a Cox-2 inhibitor control (Vioxx) [ 17 ]. These trials all used aqueous extracts of H. The last three trials included 215 patients with various forms of musculoskeletal pain (range 50 to 100) [ 19 , 20 , 24 ]. Schmelz et al (1997)[ 24 ] included subjects with acute exacerbations of joint arthrosis, chronic low back pain, and rheumatic muscle pain. Guyader (1984) [ 20 ] included subjects with gonarthrosis, poly-arthrosis, coxarthrosis, and arthrosis of the cervical spine, lumbar spine, or the nerve root canal. Gobel et al (2001) [ 19 ] included subjects with pain and/or muscle tension in shoulder, neck, and/or back. All three trials were placebo controlled. Schmelz et al (1997) [ 24 ] used an aqueous extract, Guyader (1984) [ 20 ] dried mother tincture (solvent 45% ethanol), and Gobel et al (2001) [ 19 ] an ethanolic extract (solvent 60% ethanol). All trials, except for two [ 12 , 24 ], reported adverse events for the interventions. Methodological quality and sensitivity analysis Ratings for each trial on each quality criterion are reported in Table 2 . Of the trials including patients with various forms of arthritis, four were considered high quality [ 13 , 18 , 22 , 23 ] and one low quality [ 12 ] with 60% and 50% cutoffs for methodological quality fulfillment; all trials were considered high quality if the cutoff was 40% [ 12 , 13 , 18 , 22 , 23 ]. There were very few instances of inadequate reporting (DK scores) in these trials (5/52) criteria assessed across trials); in cases where inadequate reporting was found, these tended not to be the same criteria across trials. The authors of these trials were not contacted for clarification regarding these items. Table 2 Methodological quality of controlled trials of Harpagophytum procumbens Methodological Quality Criteria Gobel et al, 2001 Schmelz et al, 1999 Guyader, 1984 Chrubasik et al, 1996 Chrubasik et al, 1997 Chrubasik et al, 1999 Chrubasik et al, 2003 Chantre et al, 2000 Frerick et al, 2001 Lecomte & Costa, 1992 Biller et al, 2002 Schruffer, 1980 A Were eligibility criteria specified? n n n y y y y y y n y n B Was randomization appropriate? y y y y y y y y y y y y C Was treatment allocation concealed? y y y y n y y y y y y y E Were groups similar at baseline regarding important prognostic indicators? dk dk n y y y y y n dk dk dk F Were outcome measure(s) and the control interventions explicitly described? y y y y y y y y y y y y G Were co-interventions avoided or comparable? dk dk dk y y y y y y dk y y H Were the outcome measures relevant? y y y y y y y y n y n y I Were adverse events described? y dk y y y y y y y y n y J Were drop-outs described? n y y y y y y y n dk n y K Was the sample size based on a priori power calculation? n n n y n y n y n n n n L1 Did the study include intention-to-treat analysis? and/or n y n n y y y y n y n y N Were point estimates and measures of variability presented for the POM? y n n y y y y y y y n n O Was the timing of outcomes appropriate? y y n y y y y y y y y Y Total 7 7 6 12 11 13 12 13 8 8 6 9 Of the trials utilizing subjects with acute exacerbations of chronic non-specific low back pain, all four were considered to be of high methodological quality regardless of the cutoff [ 14 - 17 ]. Methodological aspects that were unclear in the published reports were clarified by the study author (SC). Of those trials including a mixed sample of subjects [ 19 , 20 , 24 ], two were low quality when the cutoff was 60%, or 50% and all three were high quality when the cutoff was 40%. All trials did not adequately report sufficient information to judge baseline similarity, if co-interventions were avoided or comparable, or if there were any adverse events [ 24 ]. Harpagophytum preparations for osteoarthritis Powdered crude plant material compared to placebo Lecomte & Costa (1992) [ 22 ] utilized a powder of the secondary roots of H in 89 subjects (44 placebo; 45 H) with 98 locations of arthrosis (with two locations in three subjects in the placebo group and six in the H group). Of those with one location, 31 had osteoarthritis of the spine, 18 of the cervical spine, 14 of the hip, and 30 of the knee. Results favoured the H group. Detailed descriptions of each original study included in this review are provided in Table 3 . Table 3 Description of trials included in this review Study Sample Size Condition; mean age (range) Harpagophytum Intervention / control Outcome measures and effects Adverse effects Reviewer's Overall Conclusions Schrüffler 1980 (Germany) 50 Osteoarthritis; 51 years 2500 mg/ day, (harpagoside less than 30 mg per day) / Phenylbutazone for 4 weeks mean pain improvements: H 80%, Phenylbutazone 72%; physical impairment: H n = 1, Phenybutazone n = 5; morning stiffness: H n = 2, Phenybutazone n = 5 0 H vs 4 Phenylbutazone H better than Phenylbutazone Lecomte and Costa 1992 (France) 89 Osteoarthritis; (55–75) years 2000 mg/day (Harpagoside content estimated indirectly as 60 mg per day) / placebo for 60 days mean pain improvement: H 38%, P 25% p < .05; finger-ground distance modified Schober test (cm) mean improvement: H 16%, P 6% p < .05 none for either group H better than placebo Biller 2002 (Germany) 78 Osteoarthritis; not stated 4500 mg/day, (harpagoside content estimated at < 30 mg per day) / placebo for 20 weeks responders: H 90%, P 80% p-value not stated; mean consumption of ibuprofen: H .1, P .5 tablets not stated H better than placebo Chantre et al. 2000 (France) 122 Osteoarthritis; 62 years 4500 mg / day, (57 mg harpagoside per day) / Diacerhein for 16 weeks difference after 16 weeks between groups as measured by Lequesne functional index: less than 10 mm NS (intention-to-treat analysis with not all possible confounders considered) 10 H vs 21 Diacerhein H not worse than diacerhein Frerick et al. 2001 (Germany) 46 Osteoarthritis; 59 years 4500 mg/day, (< 30 mg harpagoside per day) / placebo for 20 weeks responders: H 71%, P 41% p=.041; WOMAC component pain NS (type of statistical analysis not stated) 8 H vs 7 P H better than placebo Chrubasik et al. 1996b (Germany) 118 Back pain; 54 years 4500 mg/day, (50 mg harpagoside per day) / placebo for 4 weeks mean tramadol consumption: H 99 ± 157 mg, P 102 ± 250 mg p =.44; number of pain-free patients at 4th week: H 9 P 1 p=.008; percentage change Arhus component pain: H 34%, P 6% p=.016 (per protocol analysis) 4 H vs 10 P not on primary outcome measure Chrubasik et al. 1997 (Germany) 102 Back pain; 49 years 4500 mg/day, (30 mg harpagoside per day) / conventionally treating physicians administering oral NSAIDs, physical exercises, or paravertaebral injections for 6 weeks number of pain-free patients 4th week: H 16, C12 NS; number of pain free patients 6th week: H 20, C 23 NS; percentage change Arhus component pain after four weeks: H 23%, C 22% p=.95; after 6 weeks H 33%, C 38% p=.38 5 H vs 0 C H not worse than C Chrubasik et al. 1999 (Germany) 197 Back pain; 56 years 4500 and 9000 mg/day, (50 and 100 mg harpagoside per day) / placebo for 4 weeks number of pain-free patients: H-100 18%, H-50 9%, P 5% p=.027); percentage change Arhus component pain: H-100 vs H-50 vs P NS (intention-to-treat analysis) 10 P, 18 H-50, 17 H-100 H better than placebo Chrubasik et al. 2003a (Germany) 88 Back pain; 62 years 4500 mg/day, (60 mg harpagoside per day) / Rofecoxib for 6 weeks number of pain-free patients: H 22%, Rofecoxib 11% NS; percentage change Arhus component pain: H 30%, Rofecoxib 29% (intention-to-treat analysis) 14 H, 14 Rofecoxib H not worse than Rofecoxib Schmelz and Hämmerle 1999 (Germany) 100 Mixed pain; not stated 4500 mg/day, (30 mg harpagoside per day) / placebo for 30 days free of low back pain: H n = 4, P n = 2; free of other pain: H n = 5, P n = 0 (confounders not considered) not stated H better than placebo Guyader 1984 (France) 50 Mixed pain; 64 years Harpagoside content estimated indirectly as <20 mg harpagoside per day / placebo for 1–3 'cycles' of 21 days each mean pain improvements: H 72%, P 65% (confounders not considered) 6 H vs 3 P H better than placebo Goebel et al. 2001 (Germany) 65 Mixed pain; 28 years 4500 mg/day, (< 30 mg harpagoside per day) / placebo for 28 days 4 H vs 2 P H better than placebo Key: NS = not significant; H = harpagophytum; P = placebo; WOMAC = Western Ontario and McMaster Universities Arthritis Index Powdered crude plant material compared to diacerhein Chantre et al (2000) [ 13 ] gave H cryodried drug powder (proprietary product Harpadol R ) or Diacerhein (D), to 122 subjects with acute exacerbations of hip and knee. Groups did not differ significantly in sponanteous pain or the Lequesne index, though differences from baseline were larger for the H group. Subjects in the H group used less diclofenac (mean = 21 tablets) than those in the D group (60 tablets) and also used less acetominophen-caffeine (H = 40 tablets and D = 60 tablets). H extract (solvent 60% ethanol) compared to placebo Biller (2002) [ 12 ] gave knee arthrosis participants a H product named Flexiloges R or placebo in addition to ibuprofen (at 800 mg ibuprofen during weeks 1–8, 400 mg during weeks 9 – 16), and only H or placebo during weeks 17–20. The main outcome measure was the responder rate, which allowed a WOMAC pain score increase of up to 20% and no additional consumption of ibuprofen in weeks 17 to 20. Results favoured the H group. Frerick et al (2001) [ 18 ] gave 46 individuals with acute exacerbations of coxarthrosis Flexiloges R or placebo in addition to ibuprofen (at 800 mg ibuprofen during weeks 1–8, 400 mg during weeks 9–16) and only H or placebo during weeks 17–20. The main outcome was the responder rate, which was defined as the number of patients that required fewer than 4000 mg ibuprofen and had a pain score increase of no more than 20% on the WOMAC component pain during weeks 17 to 20. Results indicated more responders in the H group. Aqueous extract compared to NSAID Schruffler (1980) [ 23 ] compared H (proprietary product, Salus R, ) with phenylbutazone among 40 individuals with acute exacerbations of rheumatic joint and muscle pain and 10 with gouty arthritis. Results favoured the H group. Harpagophytum procumbens for acute exacerbation of chronic non-specific low back pain (NSLBP) Aqueous extract compared to placebo In the Chrubasik study (1996) [ 14 ], 128 patients suffering from pseudo-radiating or non-radiating NSLBP were allocated to receive either a proprietary extract, Doloteffin R, , or placebo. Results favoured the H group (see Table 4). Of the 59 patients in the H group, five dropped out, one of these due to tachycardia. Of the 59 patients in the P group, four dropped out for unknown reasons. A total of four adverse effects occurred in the H group. These consisted of two individuals with nausea/emesis due to the tramadol, one patient with repeated tussive irritation, and the patient with tachycardia mentioned above. A total of 10 adverse events occurred in the P group. These included nausea (N = 2), and one patient each of nausea/vertigo due to tramadol, fatigue/vertigo, vertigo alone, diuresis/normalization of constipation (i.e. intractable constipation), constipation (several times), diuresis (several times), and sleep disturbances (permanent). An additional study by Chrubasik et al (1999) [ 16 ] randomized subjects suffering from pseudo-radiating or non-radiating NSLBP were allocated to receive the proprietary extract WS1531, at a dose equivalent to either 4500 mg H (with 50 mg harpagoside per day, H 50 ) or 9000 mg H /day (with 100 mg harpagoside per day, H 100 ), or placebo (P; N = 66). Participants had acute exacerbations of non-specific low back pain; current pain that was > 5 on a VAS (0–10). The median durations of chronic pain were P 15 years, H 50 15 years, H 100 15 years. The number of patients with acute exacerbations of greater than 3 months for each group was P 54 (82%), H 50 53 (82%), H 100 55 (83%). Current pain and Arhus scores were similar among all groups. The principle outcome measure, the number of patients who were pain-free without the permitted rescue medication for 5 days out of the last treatment week was 3 (P), 6 (H 50 ) and 10 (H 100 ) (p = 0.027, one-tailed Cochrane-Armitage test). The authors found significant improvements in pain in both H groups as compared with the placebo group. A subgroup analysis found differences between groups for those without pain radiating to the legs (H 100 40%, H 50 43% and P 23%; P = 0.017) and for those without a neurological deficit (H 100 40%, H 50 60% and P 20%; P = 0.034). Aqueous extract compared to NSAID Chrubasik et al (2003) [ 17 ] randomized 88 individuals suffering from pseudo-radiating or non-radiating NSLBP recieved either the proprietary extract Doloteffin R (N = 44) or Vioxx R (rofecoxib; R; N = 44). The percentage of subjects with chronic pain of > 6 days for each group was H 84%, R 84% and acute pain of > 90 days in H 91%, R 89%. Results indicated statistically non-significant difference between H and R. Also, a total of 21 (group H) and 13 (group R) patients used tramadol, with the average consumption being 230 mg (H) and 133 mg (R). A total of seven drop-outs occurred due to adverse events (H = 1; R = 6); two others resulted from excessive low back pain (R = 2). There were a total of seven protocol violations. A total of 14 participants in each group had adverse events with gastrointenstinal complaints, equaling eight in the H group and nine in the R group, with more severe events in the R group. Two adverse events were reported to be unrelated to the study medication in the H group. Aqueous extract compared to various conventional treatments In another study by Chrubasik et al (1997) [ 15 ], participants suffering from non-radiating NSLBP were randomized to receive either the proprietary extract Jurcuba R or conventional treatments (NSAIDs, exercise, massage, nerve blocks, acupuncture). Subjects' pain was > 5 (VAS 0–10) on at least two of the following five scores: pain at rest, while sitting, lying, and walking or at night. The median duration of acute pain was six weeks for both groups. The median duration of chronic pain was 120 months for the H group and 72 months for the control group. There were no statistically significant differences between the groups. Harpagophytum preparations for mixed pain conditions Dried mother tincture compared to placebo In a placebo controlled double-blind study, Guyader (1984) [ 20 ] included 50 patients with poly-arthrosis (N = 14), coxarthrosis (N = 2), arthrosis of the cervical spine (N = 11), lumbar spine (N = 2), and arthrosis of the nerve root canal (N = 6). Subjects were given either "Extract G", a dried mother tincture of Devil's claw (H), or placebo (P) for one to three cycles (18 single P, 16 single H, 6 PH, 4 HP, 2 HH, 1 PPP, 1 PHH, 1 PPH, 1 PHP) of 21 days with seven day intervals. Outcome measures included pain at rest, pain during exercise, joint pressure pain, pain while walking (in cases of cox- and gonarthrosis), and pain at night. All outcomes were taken 10 days after each cycle and assessed on a five-point rating scale for amount of pain (no pain = 0, mild = 1, moderate = 2, severe = 3, excrutiating = 4). A total of 70 cycles were analyzed (37 cycles for P and 33 for H) and a mean pain improvement of 72% in the H group and 65% in the placebo group (p < 0.05) was found. Drop-outs included two subjects. During the H cycles, six adverse events occurred, including nausea, gastralgia, diarrhea, severe constipation, pruritic eruptions with erythema, and generalized pruritis. During the P cycles, five adverse events were observed, including gastralgia, sweating, headache, and aerophagy. Aqueous extract compared to placebo In a double-blind, placebo-controlled trial, Schmelz et al (1997) [ 24 ] randomized 100 individuals with acute exacerbations of joint arthrosis (N = 29 (H), N = 27 (P)), chronic low back pain (N = 14 H, N = 17 P), and rheumatic muscle pain (N = 7 H, N = 6 P). Subjects were given either the proprietary extract Arthrotabs R containing extract based on 4500 mg crude plant material (equivalent of 30 mg harpagoside) per day or a placebo for 30 days. Outcome measures included a subjective pain scale (no pain, mild pain, moderate pain, severe pain, excrutiating pain) at baseline and after four weeks of treatment. The number of pain-free individuals with low back pain was H n = 4, P n = 2, and for other pain sites H n = 5, P n = 0. Harpagophytum extract (solvent 60% ethanol) compared to placebo Gobel et al (2001) [ 19 ] conducted a double-blind study in 65 individuals with pain or muscle tenseness in shoulder, neck, and/or back having lasted for 14 days prior to the study. They were randomized to placebo (N = 32) or a proprietary extract (Rivoltan; N = 31). The daily extract dosage was based on 4500 mg crude plant material (equivalent to <30 mg harpagoside per day). Outcomes included a visual analogue scale for pain (0–50 mm) as well as an experimental test battery for pain and muscle tension before and after treatment. In the per-protocol analysis the Hgroup had less pain than the P group. A total of two individuals dropped out of the trial with four adverse events occurring in the H group versus two in the P group. Discussion Osteoarthritis of the knee, hip, and spine as well as non-specific low back pain may be associated with pain, stiffness, limitation of function, and diminished quality of life [ 25 ]. Although treatment guidelines recommend simple analgesics as first-line drugs [ 26 ], surveys indicate that NSAIDs are used in preference to simple analgesics despite the lower safety-margin and the higher cost [ 11 , 27 , 28 ]. Because of the high incidence of NSAID-related adverse events and complications in the gastrointestinal and cardiovascular systems (especially in the elderly), and the high costs related to adverse events (i.e. gastrointestinal bleeding or perforation), additional medical attendances, diagnostic procedures, treatments and admissions to hospital, alternatives to NSAID therapy should be strongly considered [ 29 - 33 ]. This qualitative analysis of the 12 trials suggests that specific preparations and doses of Harpagophytum procumbens may be effective in various types of musculoskeletal pain conditions. Statistical pooling was not possible because of a lack of adequate data and clinical heterogeneity. The sensitivity analysis for methodological quality revealed that the trials on low back pain were of high quality, the trials on osteoarthritis were of high quality except one moderate quality study, and the trials on mixed pain conditions were of moderate quality. The quality of reporting in most of these trials was good. In order to increase transparency, trialists should refer to the CONSORT statement in designing and reporting clinical trials of herbal medicinal products [ 34 ]. One high quality trial indicates that there is moderate evidence of effectiveness for powdered H plant material at a dose equivalent to 60 mg of harpagoside per day for osteoarthritis of the spine, hip, and knee. However, because of the clinical heterogeneity of patients in this trial, a confirmatory study is required to firmly establish efficacy for each location of osteoarthritis. In one high quality study, 4500 mg powder containing 57 mg harpagoside in the daily dosage showed moderate evidence for non-inferiority to diacerhein in patients suffering from acute exacerbations of osteoarthritis in the hip and knee. Two trials employed an ethanolic extract (solvent 60% ethanol) containing less than 30 mg harpagoside per day in patients with osteoarthritis of the knee [ 12 ] and hip [ 18 ]. Both trials showed statisitically significant favourable results for the H group in terms of percentage of responders. However, the definition of responder in these trials may be questioned because of an allowance of pain increase up to 20% and additional rescue medication in one of the studies [ 18 ]. Therefore, given the low methodological quality of the trials and the lack of clinically significant differences between groups, we conclude there is limited evidence for the use of an ethanolic H extract based on 4500 mg crude plant material per day in patients with osteoarthritis of the knee and hip. Additional high quality trials must be done to determine the efficacy of Harpagophytum procumbens in osteoathritis. These trials must include homogenous pain conditions and must test H against standard osteoarthitis medications. Additionally, trialists should consider using symptom severity outcome measures that have proven validity and reliability, such as visual analogue scales [ 35 - 37 ], osteoarthritis specific outcome measures (e.g. WOMAC, Lequesne Index) [ 35 - 38 ], and health-related quality of life instruments (e.g. Medical Outcomes Survey Short-Form 36) [ 35 , 37 , 38 ]. A total of four high quality trials tested various dosages of Hextract in acute exacerbations of chronic non-specific low back pain. Two trials with a total of 325 patients showed that an aqueous extract at the equivalent daily dosage of 50 mg harpagoside appears to reduce pain in patients with acute episodes of chronic NSLBP greater than does placebo [ 14 , 16 ]. Therefore, the 50 mg harpagoside per dose of an aqueous extract of H can be said to have strong evidence for the treatment of acute episodes of chronic NSLBP in the short term. Additionally, a one year survey indicates that the aqueous extract is well tolerated [ 38 ]. One trial with 197 patients showed that an aqueous H extract at the equivalent daily dose of 100 mg harpagoside appears to reduce pain in patients with acute episodes of chronic NSLBP greater than does placebo [ 16 ]. Therefore, the 100 mg harpagoside per dose of an aqueous H extract has moderate evidence for the treatment of acute episodes of chronic NSLBP in the short term. Superiority of the higher dose was seen in the primary outcome (number of pain-free patients) but not in the secondary outcome measure. Therefore, there is moderate evidence for superiority of the 100 mg H dose to the 50 mg H dose. However, additional trials are required to confirm superiority of 100 mg H over 50 mg H. It is possible that a subgroup of individuals with neurological deficits (e.g. radiation into the leg) may respond well to the 100 mg harpagoside dose, yet more research is required to clarify this. An aqueous extract of H at the equivalent daily dose of 60 mg harpagoside appears to be equivalent to 12.5 mg Rofecoxib in improving pain in individuals with acute episodes of chronicNSLBP [ 17 ]. Therefore, a 60 mg daily harpagoside dose in aqueous extract of H has moderate evidence for being not inferior to 12.5 mg rofecoxib per day in the treatment of acute episodes of chronic NSLBP in the short term. Additional high quality trials, especially over longer treatment periods, are mandatory. Furthermore, equivalence trials testing Harpagophytum procumbens against standard treatments will clarify relative efficacy and safety. The final trials (two of moderate [ 19 , 24 ] and of poor quality [ 20 ] included heterogeneous musculoskeletal pain conditions. Therefore, it is difficult to reach any conclusions on the basis of these trials. Future trials should attempt to include homogenous pain conditions. The results obtained with the proprietary Harpagophytum products containing aqueous extracts can neither be transferred to an aqueous extract containing less harpagoside in the daily dosage [ 41 ] or to a product containing an ethanolic extract. This can only be done if the ethanolic extract was shown to be essentially similar to aqueous extract, and if both extracts have the same qualitative and quantitative composition of co-active constituents, same pharmaceutical form, and bioequivalence in terms of safety and efficacy [ 40 ]. With 60% ethanol as solvent, only half the amount of harpagoside (and possibly other co-active constituents) is extracted compared to water as solvent [ 42 ]; therefore, it is of great importance that a confirmatory study provide evidence of effectiveness for the ethanolic extract. Since the "active principle" has not yet been identified for Harpagophytum procumbens, the constituent harpagoside is used as a marker for standardization of Harpagophytum preparations. For harpagoside, the dose-dependent absorption into systemic circulation has been shown and may be related to lipoxygenase inhibition [ 43 ]. However, it remains to be established if the inhibitory effect on leukotriene production corresponds to therapeutic efficacy. Future research should attempt to identify the active constituent or profile of constituents that relate to therapeutic efficacy in order to make extract dosing transparent. There are several drawbacks to the present study. First, this is a qualitative review and as such it does not provide a quantitative summary of results, thus making it difficult to determine the size of effect of each intervention. Secondly, this review includes a small number of trials, often with small sample sizes. This makes it difficult to state definitive conclusions of efficacy and suggests the need for more trials. On the other hand, the trials reviewed were generally of good methodological quality and have several statistically significant and clinically significant effects. Therefore, these trials help us reach some clear conclusions regarding the use of specific preparations and doses of Harpagophytum procumbens for osteoarthritis and non-specific low back pain. Another strength of this study is the comprehensive search strategy, the methodological quality assessment, and the use of an accepted method for a best evidence synthesis. Future reviews may attempt to statistically combine the results of such trials into a meta-analysis. Conclusions There is limited evidence for an ethanolic Harpagophytum extract containing less than <30 mg harpagoside per day in the treatment of knee and hip osteoarthritis. There is moderate evidence of effectiveness for (1) the use of a Harpagophytum powder at 60 mg harpagoside in the treatment of osteoarthritis of the spine, hip and knee; (2) the use of an aqueous Harpagophytum extract at a daily dose of 100 mg harpagoside in the treatment of acute exacerbations of chronic non-specific low back pain; and (3) the use of an aqueous extract of Harpagophytum procumbens at 60 mg harpagoside being non-inferior to 12.5 mg rofecoxib per day for chronic non-specific low back pain (NSLBP) in the short term. Strong evidence exists for the use of an aqueous Harpagophytum extract at a daily dose equivalent of 50 mg harpagoside in the treatment of acute exacerbations of chronic NSLBP. Competing interests One individual (SC) was an author of several original trials included in this systematic review. This did not appear to influence the content of this paper. Authors' contributions JG developed the initial idea for the manuscript, searched for trials, extracted data and wrote and edited the manuscript. SC developed the initial idea for the manuscript, searched for trials, extracted data, and wrote and edited the manuscript. EM extracted data and wrote and edited the manuscript. Appendix 1 Highly sensitive search strategy for randomized controlled trial searches using PUBMED (randomized controlled trial [pt] OR controlled clinical trial [pt] OR randomized controlled trials [mh] OR random allocation [mh] OR double-blind method [mh] OR single-blind method [mh] OR clinical trial [pt] OR clinical trials [mh] OR (clinical trial [tw]) OR ((singl* [tw] OR doubl* [tw] OR trebl* [tw] OR tripl* [tw]) AND (mask* [tw] OR blind* {tw])) OR (latin square [tw]) OR placebos [mh] OR placebo* [tw] OR random* [tw] OR research design [mh:noexp] OR comparative study [mh] OR evaluation studies [mh] OR follow-up studies [mh] OR prospective studies [mh] OR cross-over studies [mh] OR control* [tw] OR prospectiv* [tw] or volunteer* [tw]) NOT (animal [mh] NOT human [mh]) Appendix 2 PUBMED search strategy (randomized controlled trial [pt] OR controlled clinical trial [pt] OR randomized controlled trials [mh] OR random allocation [mh] OR double-blind method [mh] OR single-blind method [mh] OR clinical trial [pt] OR clinical trials [mh] OR (clinical trial [tw]) OR ((singl* [tw] OR doubl* [tw] OR trebl* [tw] OR tripl* [tw]) AND (mask* [tw] OR blind* {tw])) OR (latin square [tw]) OR placebos [mh] OR placebo* [tw] OR random* [tw] OR research design [mh:noexp] OR comparative study [mh] OR evaluation studies [mh] OR follow-up studies [mh] OR prospective studies [mh] OR cross-over studies [mh] OR control* [tw] OR prospectiv* [tw] or volunteer* [tw]) NOT (animal [mh] NOT human [mh]) AND ("harpagophytum procumbens" OR (devil's AND claw)) AND (pain OR "musculoskeletal pain" OR "muscle pain" OR "skeletal pain" OR "bone pain" OR "joint pain" OR "extremity pain" OR myaligia OR osteoarthritis OR "rheumatoid arthritis" OR arthrosis OR "low back pain" OR lumbago OR "back pain") NOT review Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC520753.xml |
554978 | Effects of thrombin, PAR-1 activating peptide and a PAR-1 antagonist on umbilical artery resistance in vitro | Background The non-thrombotic effects of thrombin in cardiovascular tissues, as mediated via the protease activated receptors (PARs), and particularly PAR-1, have been the focus of much recent research. The aims of this study were to evaluate the effects of thrombin, a specific PAR-1 activating peptide (PAR1-AP), and a PAR-1 antagonist on human umbilical artery tone in vitro. Methods Human umbilical artery samples were obtained from 17 women at term. Arterial rings were suspended under physiologic conditions for isometric recording. The in vitro effects of thrombin (0.5 units/mL to 3 units/mL), PAR1-AP TFLLR-NH2 [10(-9) to 10(-6) M], and PAR-1 antagonist (N-trans cinnamoyl- p-fluoroPhe-p-guanidinoPhe-Leu-Arg-Orn-NH2) [10(-9) M to 10(-5) M] on umbilical artery tone were measured. Results Both thrombin and TFLLR-NH2 exerted a potent cumulative vasodilatory effect on human umbilical artery resistance (P < 0.001). The mean net maximal inhibition (MMI) for thrombin was 53.05% (n = 6; SEM = 1.43) at tissue bath concentration of 3 units/mL. The MMI with TFLLR-NH2 was 61.50 % (n = 6; SEM = 1.43) at bath concentration of 10(-6) M. In comparison to vehicle control, the PAR-1 antagonist did not show a significant relaxant or contractile effect (P > 0.05). Conclusion These findings highlight a potential role for thrombin and PAR-1 receptors in vascular regulation of feto-placental blood flow in normal pregnancy, and in association with the vascular lesions associated with IUGR and pre-eclampsia. | Background In disorders resulting in poor fetal growth, and in pre-eclampsia, thrombotic lesions are frequently observed in the maternal and fetal vascular components of the placenta, [ 1 - 3 ] and hence have been implicated in the pathophysiology of these conditions. In addition, it has been reported that in vivo generation of thrombin, in maternal plasma, is higher in patients with small for gestation age fetuses and with pre-eclampsia, than in normal pregnancy [ 4 ]. It is well established, for many years, that thrombin plays a role in blood coagulation, but its effects in many other cell and tissue types (smooth muscle cells, endothelial cells, lymphocytes) [ 5 , 6 ] have been the subject of more recent attention. It is now apparent that thrombin can regulate target cells by cleaving and activating a family of G-protein-coupled protease-activated receptors (PARs)[ 5 - 7 ]. This proteolytic cleavage of PARs is mediated by a family of enzymes that require serine within the active site i.e. serine proteases [ 5 ]. There are 4 major PAR subtypes (PAR1-4) with diverse reported functions in various tissues [ 6 ]. PAR activation has been closely linked to inflammation [ 6 , 8 ], contraction of vascular [ 5 ] and non-vascular [ 9 , 10 ] smooth muscle, and platelet activation [ 6 ]. Although PAR-1, PAR-3 and PAR-4 [ 11 ] are all known to be thrombin receptors, the mechanism of activation by thrombin at these different PARs varies [ 12 , 13 ]. There is growing evidence, from vascular tissue studies in several animal models, that non-thrombotic thrombin-mediated signalling events are central to the response to the disease process typical of vascular lesion formation in atherosclerosis [ 14 ]. The direct effects of thrombin on vascular cells, via the PAR receptors, and particularly PAR-1, have been the main focus of investigation for this hypothesis. PARs 1, 3 and 4 are activated by thrombin [ 11 ], but PAR-1 is activated at low thrombin concentrations and most of what is known about thrombin signalling downstream of the receptors, has been derived from studies of PAR-1 [ 14 ]. While other PAR subtypes are present in human arterial vessels, it appears that PAR-1 is primarily involved in endothelium-dependent relaxation to thrombin and trypsin [ 15 ]. To our knowledge, there are no data outlining the potential effects of thrombin, or specific PAR-1 receptor modulation, on the feto-placental circulation, despite the critical role of thrombin in disorders of this vasculature. The aims of this study were to evaluate the direct effects of thrombin, the specific PAR-1 activating peptide (PAR1-AP), TFLLR-NH 2 (Thr-Phe-Leu-Leu-Arg-NH 2 ), and the PAR-1 specific antagonist (N-trans cinnamoyl -p-fluoroPhe-p-quanidinoPhe-Leu-Arg-Orn-NH 2 ) on human umbilical artery tone in vitro. Methods The study was carried out in the Department of Obstetrics and Gynaecology, University College Hospital Galway, Ireland between May 2002 and April 2003. Sections of human umbilical cord approximately 10 cm in length were excised from the proximal segment of the cord (i.e. closest to the placental attachment) immediately after elective cesarean delivery. Samples were obtained from 17 women after elective cesarean section at term and from one patient after normal vaginal delivery. All pregnancies were uncomplicated and there was no evidence of hypertensive disease or intrauterine fetal growth restriction. The mean maternal age was 32.47 years (range 25–40 years). The median period of gestation was 38 weeks (range 37–42 weeks). The reasons for cesarean section included previous cesarean section (n = 9), breech presentation (n = 6), previous myomectomy (n = 1) and unstable fetal lie (n = 1). At the time of recruitment 4 women were nulliparous and 13 women were parous. Samples were immediately placed in cold buffered Krebs Henseleit physiological salt solution (pH 7.4) of the following composition: potassium chloride 4.7 mmol/L, sodium chloride 118 mmol/L, magnesium sulphate 1.2 mmol/L, calcium chloride 1.2 mmol/L, potassium phosphate 1.2 mmol/L, sodium bicarbonate 25 mmol/L and glucose 11 mmol/L. Maternal written informed consent was obtained prior to tissue collection, and the tissue collection procedure was approved by the Research Ethics Committee at University College Hospital Galway. Umbilical arteries were carefully dissected free of Wharton's Jelly and cut in rings 4–5 mm in axial length. Rings were suspended individually on stainless steel hooks inserted into their lumens and mounted under 2 g (30 mN) of isometric tension, in glass-jacketed tissue baths, as previously described [ 16 , 17 ]. Each bath-contained 10 mL of oxygenated (95% O 2 / 5% CO 2 ) Krebs Henseleit physiological salt solution (PSS) at 37°C and pH 7.4. Rings were allowed to equilibrate for 90 minutes with regular washouts of PSS. During this interval, spontaneous tone developed. After the equilibration period, the vessel rings were challenged with 60 mM potassium chloride (KCl). Three washouts with PSS were carried out once the maximum response had reached a plateau, and a 20 minute recovery period was allowed in order that baseline be attained again. The KCl challenge was performed three times. After the last KCl challenge, 40 minutes recovery was allowed, and contraction was then stimulated by bath exposure of the vessel rings to 5-hydroxytriptamine (5-HT) (10 -7 M). Once maximum contractile response to 5-HT was attained, the rings were allowed to remain at plateau for 20 minutes. Concentration- effect experiments were performed by cumulative additions of thrombin, the PAR1-AP, or the PAR-1 antagonist to the tissue bath. Thrombin was added to the tissue bath at an initial concentration of 0.5 units/mL, and this was increased at 20-minute intervals to 1 unit/mL, 2 units/mL and 3 units/mL respectively. The bath concentrations ranges investigated for PAR1-AP were 1 nanomol/L, 10 nanomol/L, 100 nanomol/L and 1 micromol/L (i.e. 10 -9 - 10 -6 M), and for PAR-1 antagonist were 1 nanomol/L, 10 nanomol/L, 100 nanomol/L, 1 micromol/L and 10 micromol/L (i.e. 10 -9 -10 -5 M), all at 20-minute intervals. The effects of thrombin, PAR1-AP and PAR-1 antagonist were demonstrated by expressing the mean amplitude calculated during the 20-minute period following addition of each drug concentration, as a percentage of the mean amplitude obtained in the 20 minutes prior to any drug addition. This measurement represents percentage contractility or tone, and subtracted from 100%, provides the percentage relaxation value for each bath concentration of vehicle and study compounds. The net relaxant effect of each compound was calculated by subtracting the percentage contractility value calculated for its respective vehicle (for thrombin, PAR1-AP or PAR-1 antagonist), control experiment, at each similar bath concentration. All of the umbilical artery samples used for experimentation for each compound were obtained from different women (i.e. n = 6 for PAR1-AP), for example, was achieved by using umbilical artery samples from 6 different women). The allocation of umbilical artery samples for the different experiments was entirely random. Fresh Krebs Henseleit physiological salt solution was made and buffered daily. KCl solutions were prepared on the day of experimentation. A stock solution of 5-HT (Sigma-Aldrich, Dublin, Ireland) was made up in de-ionised water and diluted with Krebs solution. Thrombin was purchased from Sigma-Aldrich (Dublin, Ireland) and a stock solution of 1100 U/ml prepared in de-ionised water and stored at -20°C. PAR1-AP was purchased from Tocris Cookson Ltd (Bristol, UK) and a stock solution of 1 millimol/L was prepared using deionized water, with subsequent dilutions in PSS. The PAR-1 antagonist was prepared and assayed by methods previously described [ 18 ]. The IC 50 value for antagonist inhibition of platelet aggregation stimulated with 1 micromol/L SFLLRN agonist was determined to be 0.1 micromol/L. It was prepared as a 10 millimol/L stock solution in dimethysulphoxide (DMSO) and stored in room temperature protected from direct light. Final bath concentrations of DMSO, at the highest concentration of PAR-1 antagonist, did not exceed 1%, for both study and vehicle control strips. Serial dilutions of thrombin, PAR1-AP and the PAR-1 antagonist were made using Krebs Henseleit physiological salt solutions. Comparisons of measurements of amplitude for each bath concentration of thrombin, PAR1-AP and the PAR-1 antagonist, or respective control vehicle, were made using a one-way ANOVA. Post-hoc comparisons were made using the Tukey HSD test. A P value <0.05 was accepted as statistically significant. The statistical package SPSS version 10 was used for statistical calculations. Results A recording from control experiments (i.e. without addition of vehicle, thrombin, PAR1-AP or PAR-1 antagonist) demonstrating umbilical artery tone due to bath exposure of the ring to serotonin, for the entire duration of an experiment, is shown in Figure 1A . The mean net spontaneous relaxation of tone observed was 14.39% (SEM = 2.76). Thrombin exerted a potent and cumulative vasodilatory effect on umbilical artery tone in comparison to simultaneous vehicle (PSS) only experiments. A representative recording of the effects of thrombin is shown in Figure 1B . At bath concentrations at or greater than 0.5 units/mL thrombin exerted a significant vasodilatory effect. The net vasodilatory effects of thrombin are provided in Table 1 . The mean net inhibition of tone observed, at maximum thrombin concentration investigated (i.e. 3 units/mL) was 53.5% (SEM = 4.62; n = 6; P < 0.001). Figure 1 Representative recording of A) Serotonin induced contraction of umbilical artery, B) Serotonin induced contraction followed by cumulative additions of Thrombin and C) Serotonin induced contraction followed by cumulative additions of PAR1-AP. Table 1 Net inhibitory effect of thrombin, PAR-1 and PAR1-AP on human umbilical artery tone. Thrombin (n = 6) (% ± SEM) PAR1-AP (n = 6) (% ± SEM) PAR-1 Antagonist (n = 6) (% ± SEM) (P Value) 0.5 U/ml #28.90 ± 2.60 10 -9 M *23.91 ± 5.64 10 -9 M 2.83 ± 3.05 (NS) (0.630) 1 U/ml *35.39 ± 3.91 10 -8 M *37.32 ± 2.29 10 -8 M 3.80 ± 2.33 (NS) (0.471) 2 U/ml *44.72 ± 2.31 10 -7 M *52.39 ± 1.28 10 -7 M 5.60 ± 3.74 (NS) (0.228) 3 U/ml *53.51 ± 4.62 10 -6 M *61.50 ± 1.43 10 -6 M 7.03 ± 5.74 (NS) (0.148) 10 -5 M 6.87 ± 4.48 (NS) (0.815) Values presented represent the mean inhibitory effects on umbilical artery tone i.e. after adjusting for control / vehicle experiments. The values provided represent % inhibition ± standard error of the mean (SEM). The PAR1-AP used was Threonine-Phenylalanine-Leucine-leucine-Arginine-NH 2 . The PAR-1 antagonist used was Ser- pFPhe-pGPe-Leu-Arg-Orn-NH 2 . The values were compared with amplitude measurements observed prior to drug addition (NS not significant, # P < 0.01, * P < 0.001). TFLLR-NH 2 , the PAR1-AP, similarly exerted a cumulative vasodilatory effect on umbilical artery tone. A representative recording of the cumulative effects of PAR1-AP are shown in Figure 1C , with the mean difference in amplitude measurements (i.e. in comparison to simultaneous control experiments) provided in Table 1 . The mean net inhibition of tone observed, at maximum PAR1-AP concentration investigated (i.e. 10 -6 M / 1 micromol/L) was 61.5% (SEM = 1.43; n= 6; P < 0.001). For the PAR-1 antagonist, N-trans cinnamoyl-p-fluoroPhe-p-guanidinoPhe-Leu-Arg-Orn-NH 2 , no relaxation of umbilical artery tone in vitro was observed, in comparison to vehicle only control experiments. Figure 2A demonstrates a representative recording of umbilical artery tone, after exposure to vehicle only (i.e. DMSO added cumulatively). The mean net relaxation of tone observed with vehicle was 70.28% (SEM = 2.98). In Figure 2B , a representative recording of the effects of cumulatively increasing bath exposure of arterial rings to the PAR-1 antagonist is shown. The mean net maximal inhibition exerted was 6.87% (SEM = 1.57; n = 6; P = 0.280). Finally, there was no observed difference in the vasodilatory effects of thrombin, PAR1-AP or the PAR-1 antagonist in relation to parity. Figure 2 Representative recordings of A) umbilical artery tone after serotonin induced contraction with exposure to vehicle only (i.e. DMSO added cumulatively), and B) the effects of cumulatively increasing bath exposure of arterial rings to Ser- pFPhe-pGPe-Leu-Arg-Orn-NH 2 after serotonin induced contraction. Discussion This study demonstrates that thrombin exerts a potent vasodilatory effect on human umbilical artery vasculature in vitro. While it is known that thrombin has an inhibitory effect on vessel tone in other animal and human vascular tissue types [ 19 - 21 ] this is the first report, to our knowledge, of the direct effects of thrombin on human umbilical artery vasculature. Receptors for thrombin, the PAR family of receptors, are present on vascular smooth muscle cells, and on endothelial cells [ 14 , 22 ], with the effects of thrombin mediated mainly via the PAR 1, 3 and 4 receptor subtypes [ 11 ]. The potential physiological function of thrombin in mediating vascular tone in the umbilical circulation in normal pregnancy is unknown. Our findings also raise questions in relation to the role of the non-thrombotic effects of thrombin in the feto-placental vasculature in disorders of pregnancy such as pre-eclampsia and intrauterine growth restriction, which are classically associated with thrombotic lesions or a relative excess of thrombin [ 1 - 3 ]. The possibility that the vasodilatory effect of thrombin may serve to counteract the diminished perfusion associated with the pathophysiology of these conditions, at least in the early stages of disease, is one hypothesis. A further theory is that the much enhanced uterine contractility elicited by thrombin and PAR1-AP [ 13 ] may be concomitantly associated with a feto-placental vasodilatory effect to maintain good utero-placental blood flow. We have not elucidated the exact mechanism, or mechanisms, of the thrombin mediated alteration in vascular tone in human umbilical artery. We have however demonstrated that a similar effect is elicited by bath exposure of the arterial rings to the specific PAR1-AP, TFLLR-NH 2 . PAR1-APs are recently designed small synthetic peptide ligands which mimic the effects of proteases by binding directly to the activation site of the PAR-1 receptor, bypassing the need for proteolytic cleavage of the receptor. Soluble peptide ligands, as PAR-APs are, can vary greatly in potency as agonists in comparison with proteases [ 10 ]. TFLLR-NH 2 is reportedly one of the more selective PAR1-APs and is deemed most preferable for purposes of studying the physiologic role of PAR-1 [ 10 ]. In view of the fact that PAR-1 function is central to a thrombin-mediated effect in other tissues [ 11 , 14 , 15 ] and that a PAR-1 mediated effect in human umbilical artery vasculature is similar to that elicited by thrombin, the findings from this study are suggestive of a major role for PAR-1 in mediating the vasodilatory effect of thrombin in umbilical artery vasculature. Other possible mechanisms, and the potential role of PARs 3 and 4, have not been evaluated in this study. The PAR-1 antagonist, N-trans cinnamoyl-p-fluoroPhe-p-guanidinoPhe-Leu-Arg-Orn-NH 2 , did not alter umbilical artery tone. This peptide is a competitive antagonist i.e., it competes for the agonist binding site and does not appear to have any other activity [ 23 ]. The direct effects of PAR-antagonism had not previously been evaluated on umbilical artery preparations in vitro, and hence their inclusion in this study. We have also previously demonstrated that this specific PAR-1 antagonist exerted a relaxant effect on human myometrium in vitro, the mechanism of which is unknown [ 24 ]. The results observed here serve to confirm that N-trans cinnamoyl-p-fluoroPhe-p-guanidinoPhe-Leu-Arg-Orn-NH 2 is inactive towards the PAR-1 receptor. The true role of PAR antagonists, in such experiments, is for the purpose of investigation of the potential effects of an agonist at a PAR subtype. Pre-incubation of the arterial rings with PAR-1 antagonist, for example, should alter the response elicited by PAR1-AP. Such experiments, with selective blockade of the PAR subtypes, in normal and diseased pregnancies, are a subject of further studies. There are some limitations to this study. All umbilical cord artery samples were obtained at elective cesarean section. This was included in the design of the study in order to maintain uniformity in terms of mode of delivery, and, to avoid using cord samples that may have undergone excessive traction in the third stage of labor at vaginal delivery. It is unknown whether this latter point is valid, or not, in studies using in vitro umbilical preparations. We are currently performing comparative studies from samples obtained at vaginal delivery to further evaluate this matter. Secondly, the effects of thrombin and PAR-1 modulation, on smaller vessels in the feto-placenta circulation (i.e. placental arteries), and in association with disorders of pregnancy, would compliment our knowledge of the importance of this pathway. Finally, there are limitations in extrapolating from in vitro experiments to the in vivo situation, but the experiments conducted here represent a reliable and valid in vitro model for these vascular preparations. Conclusion In conclusion, thrombin exerts a potent vasodilatory effect on umbilical artery preparations in vitro. A similar effect is also observed using a specific PAR1-AP. The potential non-thrombotic role of thrombin, and PAR subtype modulation, in regulation of the feto-placental circulation in normal pregnancy, and in pregnancies complicated by hypertensive disease or intrauterine growth restriction, is highlighted by these findings. Authors' contributions AJO'L performed the experiments and wrote the manuscript. CJO'S and NR performed the experiments. AMF analysed the data and wrote the manuscript. JTE provided the PAR-1 antagonist (N-trans cinnamoyl- p-fluoroPhe-p-guanidinoPhe-Leu-Arg-Orn-NH 2 ) and wrote the manuscript. JJM designed, supervised the study and wrote the manuscript. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC554978.xml |
549518 | Colonic Paneth cell metaplasia is pre-neoplastic condition of colonic cancer or not? | Background The carcinogenesis of colorectal cancer has been accepted by a model for a cascade of genetic alterations, named the adenoma-carcinoma sequence. In order to elucidate the carcinogenesis of the colorectal cancer more clearly, the genetic abnormalies of the non-neoplastic mucosal epithelium of the colon and rectum should be investigated. It has been speculated that colonic Paneth cell metaplasia (PaM) is one of the pre-neoplastic mucosa of colonic cancer. Therefore, we studied the propria mucosa of the right colon with PaM from the standpoints of the frequency of the K-ras codon 12 mutations (K-ras), which is initial genetic abnormality in colorectal cancer, and the loss of heterozygosity of microsatellite markers (LOH-MS), which has a relationship to development of colorectal cancer. Methods Fifty-two regions with PaM histopathologically from 12 surgically resected right colon specimens were studied. DNA extraction of the colonic mucosa with PaM was obtained using a microdissection method, and the frequency of the K-ras of PaM was investigated by enriched polymerase chain reaction-enzyme linked mini-sequence assay, and the frequency of the LOH-MS (D2S123, D17S250 and D5S346) of PaM was examined by high resolution fluorescenced labeled PCR primers. Results K-ras mutation was detected in fifteen regions among 52 PaM (28.9%). All mutations were a single mutation and GGT changed to AGT in eleven and GAT in four. LOH-MS were detected in twenty-one regions among 52 PaM (40.4%) (D2S123: 35.4%, 17/48 regions, D17S250: 13.7%, 7/51 regions, and D5S346: 0%, 0/52 regions). No K-ras mutations and LOH-MS were detected in the controls (Colorectal mucosa with no PaM). Conclusions Colonic mucosa with Paneth cell metaplasia may be one of the pre-neoplastic mucosa in the development of the colonic epithelial neoplasia. | Background Vogelstein et al. [ 1 ] have reported the involvement of multistage genetic abnormalities in the development of colorectal cancers, and pointed out that K-ras mutation is the initial genetic abnormality in the adenoma-carcinoma sequence [ 2 ] of the development of the colorectal cancer. Recently, some reports have pointed out that the replication errors and loss of heterozygosity of the microsatellite markers have the development of the colorectal cancer [ 3 , 4 ]. However, the genetic abnormalities of the non-neoplastic mucosal epithelium of the colon and rectum has not been investigated, except the aberrant crypt foci [ 5 ] and hyperplastic polyp [ 6 ], although the colorectal epithelial neoplasia is derived from the colorectal mucosal epithelium. For the preventive medicine, the genetic abnormalities of the pre-neoplastic mucosa of the colorectal cancer should be known. We have speculated previously that the colorectal Paneth cell metaplasia (PaM) is one of pre-neoplastic mucosa on the development of the colorectal epithelial neoplasias [ 7 ], because PaM were seen very frequently in the adjacent mucosa to the minute-sized colorectal epithelial neoplasias as well as within these neoplasias. The main purpose of the present study was to investigate the frequencies of the K-ras codon 12 mutations (K-ras) and the loss of heterozygosity of dinucleotide microsatellite markers (LOH-MS) in the propria mucosa with PaM of the right colon. Methods The materials were 12 surgically resected specimens of right colon, which had the colonic carcinomas were present, and histological diagnosis was assessed at the Department of Pathology, Juntendo Izunagaoka Hospital. Although we have intended to investigate the PaM in the normal colon of the individuals without cancer, using the biopsies specimens, it was difficult to detect the PaM in these specimens. No inflammatory bowel diseases were included in the materials. Informed consent was obtained from all the patients to investigate the genetic alterations in the current study. The specimens were fixed in 10 % buffered formalin solution and prepared by cutting the non-neoplasitic area into 3 – 5 mm sections. Each section was embedded in paraffin and stained with hematoxylin and eosin (HE), followed by immunohistochemical staining for anti-lysozyme (DAKO, Japan). Immunohistochemical stainings were performed by the avidin-biotin-peroxidase-complex method, at dilution of 1 : 100. Fifty-two colonic mucosal regions with PaM, distant from neoplastic lesion, aberrant crypt foci and hyperplastic polyp, were detected by HE staining and anti-lysozyme antibody staining, and these mucosa were used as the target regions in the current study. All paraffin blocks of the target regions and 12 paraffin blocks with no PaM as controls of each materials were used for the DNA extraction. DNA extraction Paraffin blocks with the target foci mentioned above were prepared for DNA extraction. The target foci were microdissected using a 20-gauge needle, comparing the slide with HE staining in the same position. The extracted DNA was diluted with 5 ml of TaKaRa DEXPAT (for DNA Extraction from Paraffin-embedded Tissue, TaKaRa Biomedical Inc.). Analysis of the K-ras Mutation of K-ras was analyzed and compared by enriched polymerase chain reaction-enzyme linked mini-sequence assay (PCR-ELMA). In PCR-ELMA [ 8 , 9 ], upstream primer for the first and second PCR was 5'-TAAACTTGTGGTAGTTGG-AACT-3', downstream primer for the first PCR was 5'-GTTGGATCA-TATTCGTC-CAC-3', and downstream for primer the second PCR was 5'-CAAATGAT-CTGA-ATTAGCTG-3'. The first PCR reaction was performed containing 1 μL of DNA lysate, 100 μM dNTP, 1.5 mM MgCl 2 , 1 μM each primer, 0.625 U Taq DNA polymerase and 1 × PCR buffer [containing 10 mM Tris-HCl(pH 8.3 at 25 degrees of temperature), 50 mM KCl and 0.001%(w/v) gelatin] in a thermal cycler. Then, 10 μL of the denatured second PCR product was hybridized with probes to detect the K-ras codon 12 wild-type (GGT) and six mutant (GAT, GCT, GTT, AGT, CGT and TGT) DNAs were immobilized, at 55 degrees of temperature for 30 minutes, and 100 μL of biotinylated A and 0.01 U of TdqDNA polypmerase were added and incubation was continued at 55 degrees of temperature for 30 minutes. For development, 100 μL of avidin-horseradish peroxidase conjugate was added and the reaction was performed at room temperature for 30 minutes. Then, 100 μL of tetramethyl-benzidine (TMB) substrate was added and the plates were left to develop in the dark at room temperature for 20 minutes. Finally, 100 μL of stop solution was contained and the light absorbance of each sample was measured by spectrophotometry (Multiskan Multisoft, Labsystems, Tokyo) with a 450 nm filter wavelength (Figure 1 ). Figure 1 Enriched polymerase chain reaction-enzyme linked mini-sequence assay (PCR-ELMA), showing K-ras codon 12 mutation in the colonic Paneth cell metaplasia. AGT-type mutation was seen in line 3. Analysis of the LOH-MS Although we have intended to investigate the frequency of microsatellite instabilities using five microsatellite markers, which has been recommended by National Cancer Institute[ 10 ], the replication error or microsatellite instabilities could not be detected. Although the true reasons for these failures was not know, we may think that the target foci were too small and post-formalin solution condition with long time. Therefore, only LOH was investigated described below. Three dinucleotide microsatellite markers (D2S123, D17S250 and D5S346) were selected for LOH among the microsatellite markers, recommended by National Cancer Institute, because commonly it has been thought to be difficult to study the LOH using mononucleotide microsatellite markers. These LOH-MS were investigated using high resolution fluorescenced labeled PCR primers in proportion to the method of Tsuchida et al. [ 11 ]. The outline of it : 1. PCR was performed containing 1 μL of DNA lysate, 100 μM dNTP, 1.5 mM MgCl 2 , 1 μM each primer marked with fluorescent dye of three colors as blue, green and yellow, 0.625 U Taq DNA polymerase and 1 × PCR buffer [containing 10 mM Tris-HCl(pH 8.3 at 25°C), 50 mM KCl and 0.001%(w/v) gelatin] in a thermal cycler. 2. The electrophoresis was conducted 2 hours by means of an ABI-377 DNA auto-sequencer (PE Biosystems, Inc., Foster City, CA, USA). 3. A comparison was made of peaks of same marker arising from normal tissue and the propria mucosa with PaM, using the Gene Scan TM waveform analyzed softwave, and LOH were assessed. That is to say, LOH (+) was assessed when the ratio of the peak area of (the propria mucosa with PaM/normal tissue) was less than 70 % or was more than 143 %. The data were analyzed statistically with Student's t-test (t-test) and chi-square test ; a p-value of less than 0.05 was considered to be significant. Results (Table 1 ) Table 1 K-ras mutation and the loss of heterozygosity of microsatellitemarkers of colonic Paneth cell metaplasia K-ras* D2S123 D17S250 D5S346 Total LOH* PaM 28.9% 35.4% 13.7% 0% 40.4% (15/52) (17/48) (7/51) (0/52) (21/52) Normal 0% 0% 0% 0% 0% (0/8) (0/8) (0/8) (0/8) (0/8) PaM : Colonic Paneth cell metaplasia, Normal : Normal colonic mucosa, K-ras : K-ras codon 12 mutation, * : between PaM and Normal, p < 0.01, chi-square test K-ras mutation was detected in fifteen regions among 52 PaM (28.9 %). All mutations were a single mutation. For K-ras mutation patterns, 11 showed GGT to AGT, and four showed to GAT. LOH-MS was detected in twenty-one regions among 52 PaM (40.4 %) (D2S123: 35.4 %, 17/48 regions, D17S250: 13.7 %, 7/51 regions, and D5S346: 0 %, 0/52 regions). No K-ras mutations and LOH-MS were detected in the controls (Colorectal mucosa with no PaM, no neoplastic lesion, no aberrant crypt foci and no hyperplastic polyp). Thus, the frequency of both K-ras mutation and LOH-MS in the colonic mucosa with PaM were significantly higher than those of the controls (p < 0.01, chi-square test). Discussion K-ras mutations have been detected in several human neoplasias [ 12 - 14 ], and it has been pointed out that K-ras mutation is the initial genetic abnormality in the development of the colorectal cancers [ 1 ]. Recently, the replication errors of the gene is thought to be important in the development of the colorectal cancer. And the loss of heterozygosity of the microsatellite markers, which are often used as the targets in the investigation for the replication errors of the gene, is also considered to be important in the development of the colorectal cancer [ 3 , 4 ]. Namely, the carcinogenesis of colorectal cancer is almost clarified. In order to conclude the carcinogenesis of the colorectal cancer more clearly, the genetic abnormality of the non-neoplastic mucosal epithelium of the colon and rectum should be investigated, however, because the colorectal cancers are derived from the colorectal mucosa. It is also important for the preventive medicine of the colorectal cancer to know the carcinogenesis of it. However, until now, the genetic abnormalities of the colorectal non-neoplastic mucosa is unclear, except the aberrant crypt foci [ 5 ] and hyperplastic polyp [ 6 ]. We have already reported that the colorectal Paneth cell metaplasia (PaM) is one of pre-neoplastic mucosa on the development of the colorectal epithelial neoplasias [ 7 ], because PaM were seen very frequently in the adjacent mucosa to the minute-sized colorectal epithelial neoplasias and within these neoplasias. Therefore, K-ras mutation and the loss of heterozygosity of microsatellite markers of PaM were investigated in this study, and the current study is thought to be the first report focusing on this. Our results showed that K-ras mutation was detected in fifteen regions among 52 PaM (28.9%), and LOH-MS was detected in twenty-one regions among 52 PaM (40.4%). Namely, K-ras mutation and LOH-MS of PaM were not rare and the frequency of those of PaM were higher than those of the normal colonic mucosa, and it came to light that some PaM had the genetic abnormalities which had a relationship to the development of colorectal cancer. Paneth cells, which are usually situated at the base of the glands of the small intestine, were first identified by Scwalbe [ 15 ] in 1872, were studied in detail morphologically by Paneth [ 16 ] in 1888. Now, these cells have been one of the most famous cells in the gastro-intestinal tract, however, the detail function of these is not clear. Paneth cells are sometimes seen in the colorectal tubules, for example in the proximal colonic mucosa of elderly subjects [ 7 , 17 ], in patients with ulcerative colitis [ 18 ] and colonic epithelial neoplasia [ 1 , 19 , 20 ], although the reasons why Paneth cells appear in the large bowel is still unknown. Described above, many riddles about Paneth cell are still remained. And the gene abnormalities of colonic mucosa with PaM in the current study may be not equal to those of single Paneth cell in the colonic mucosa, because it is difficult to obtain only single Paneth cell in the colonic mucosa, even if microdissection method is used. However, we have been able to investigate the colonic mucosa with PaM, and we think it very interesting that some PaM have K-ras mutation and the loss of heterozygosity of microsatellite markers, and these PaM may be thought to be the pre-neoplastic mucosa in development of the colonic epithelial neoplasia. Further molecular studies concerning Paneth cell metaplasia in the large bowel should be warranted. Conclusions Colonic mucosa with Paneth cell metaplasia may be one of the pre-neoplastic mucosa in the development of the colonic epithelial neoplasia. Abbreviations PaM, colonic Paneth cell metaplasia; K-ras, K-ras codon 12 mutations; LOH-MS, loss of heterozygosity of microsatellite markers; HE, hematoxylin and eosin; PCR, polymerase chain reaction; ELMA, enzyme linked mini-sequence assay | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549518.xml |
314478 | Protein Essential for Malarial Parasite to Reach and Infect Liver Cells | null | Plasmodium, the microscopic parasite that causes malaria, passes through two hosts, two reproductive modes, four habitats, and over half-a-dozen distinct developmental stages in one lifecycle. When a Plasmodium -infected mosquito bites a human, it injects the parasite—sequestered in the mosquito's salivary glands in its sporozoite stage—into the victim's bloodstream. Within hours, the sporozoites invade the liver—a critical stage for establishing infection—and spend the next few weeks asexually dividing inside liver cells, eventually releasing thousands of merozoites into the bloodstream. Merozoites quickly invade red blood cells and begin a second round of asexual proliferation. The infected cells rupture and die, releasing more parasites and toxins. The toxins cause malaria's characteristic fever and chills, and the liberated merozoites initiate another cycle of red blood cell attacks. An unresolved question has been how the circulating sporozoites reach the liver cells in the first place, since liver cells are separated from the bloodstream by a layer of endothelial and Kupffer cells, which form the walls of the liver capillaries. (Kupffer cells project into the bloodstream and remove contaminants.) Having identified a protein required for sporozoite migration through the capillary lining, Tomoko Ishino, Masao Yuda, and their colleagues at Mie University School of Medicine in Japan may have found an answer. Only four of the roughly 150 vertebrate-infecting Plasmodium species affect humans. P. falciparum , the most pathogenic of the human-infecting species, is closely related to avian and rodent species. One rodent species,— P. berghei —shares fundamental aspects of structure, physiology, and lifecycle with P. falciparum and so serves as a model for the human parasites. Since sporozoites must infect mosquito salivary glands before they can infect the mammalian liver, Yuda's team searched for sporozoite genes that are predicted to encode secretory or membrane proteins and are expressed only in mosquito salivary glands. Their search revealed a coding region conserved in several species of Plasmodium . Tracing the gene's activity through the parasite's life cycle, Yuda's team confirmed that it was expressed only in sporozoites in the mosquito salivary gland—not in the mosquito midgut, where sporozoites are produced after mosquitoes feed on the blood of an infected person. The corresponding protein was localized to micronemes, specialized secretory organelles found at the front end of malaria parasites. Because micronemes are known to play a central role in Plasmodium motility and invasion, the researchers predicted this protein would also be important in migrating to or invading liver cells. They named the protein SPECT, for sporozoite microneme protein essential for cell traversal. Yuda's team tested SPECT's function by generating spect -disrupted mutants and observing how the altered parasites affected their hosts. spect disruption did not affect parasite proliferation in rat red blood cells or interfere with parasite development in the mosquito midgut or salivary glands, but it did have an effect on the parasite's ability to infect the liver. Rats injected with spect -disrupted parasites had significantly lower levels of liver infection than rats injected with nonmutant parasites. Since it was unclear whether the spect -disrupted mutants lost their infectivity or simply could not pass through the cell layer, the researchers inoculated human liver cells with the mutants and found that they infected the cells normally. Yuda's team also tested SPECT's impact on sporozoite cell-passage ability; if the mutants couldn't reach the liver cells, they couldn't infect them. spect -disrupted parasites completely lost their ability to pass through cells. Since traversal of the cellular barrier between liver cells and the circulatory system is a crucial step in malarial infection, the authors conclude, SPECT and other proteins involved in shuttling sporozoites into liver cells could be effective targets for malaria treatment and prevention. Sporozoite migration to hepatocytes | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC314478.xml |
547904 | Chlamydia pneumoniae induces aponecrosis in human aortic smooth muscle cells | Background The intracellular bacterium Chlamydia pneumoniae is suspected to play a role in formation and progression of atherosclerosis. Many studies investigated cell death initiation versus inhibition by Chlamydia pneumoniae in established cell lines but nothing is known in primary human aortic smooth muscle cells, a cell type among others known to be involved in the formation of the atherosclerotic plaque. Type of cell death was analyzed by various methods in primary aortic smooth muscle cells after infection with Chlamydia pneumoniae to investigate a possible pathogenic link in atherosclerosis. Results Chlamydiae were found to be localized up to 72 h post infection in aortic smooth muscle cells either as single bacteria or inside of large inclusions. Quantification of host cell death by lactate dehydrogenase release assay revealed strictly dose and time dependent lysis for all tested isolates of Chlamydia pneumoniae . Phosphatidylserine exposure was detected by flow cytometry in Chlamydia pneumoniae infected cells. Ultrastructure of Chlamydia pneumoniae infected human aortic smooth muscle cells showed extensive membrane- and organelle damage, chromatin condensation but no nuclear fragmentation. DNA fragmentation as well as cell membrane permeability was analyzed by TUNEL and NHS-biotin staining and occurred exclusively in cells carrying Chlamydia pneumoniae spots but not in smooth muscle cells with inclusions. These morphological features of cell death were not accompanied by an activation of caspase-3 as revealed by analysis of enzyme activity but involved mitochondrial membrane depolarization as shown by TMRE uptake and release of cytochrome c from mitochondria. Conclusion This study provides evidence that Chlamydia pneumoniae induce a spot like infection in human aortic smooth muscle cells, which results in a chimeric cell death with both apoptotic and necrotic characteristics. This aponecrotic cell death may assist chronic inflammation in atherosclerotic blood vessels. | Background Infection with Chlamydia pneumoniae (Cpn) usually causes acute respiratory tract infections [ 1 ]. Chronical infection with Cpn may also contribute to formation and progression of atherosclerotic lesions apart from the classical risk factors such as hypertension, hypercholesterolemia and hyperlipidemia [ 2 ]. Cpn has been extensively studied in the context of atherosclerosis [ 3 ] because atherosclerosis and cardiovascular disease are the leading causes of death in the United States, Europe and much of Asia [ 4 , 5 ]. The obligate intracellular bacterium has been detected in atherosclerotic lesions by immunohistochemistry, polymerase chain reaction and electron microscopy [ 6 - 8 ] and has also been cultured from atheromatous plaques [ 9 , 10 ]. On the cellular level smooth muscle cells and macrophages in the intima have been found to be infected with Cpn [ 11 , 12 ]. In general terms, if the inflammatory response does not effectively neutralize or remove the offending agents, such as Cpn, it can continue indefinitely resulting in the progression of the disease [ 2 ]. Chlamydiae exhibit a unique developmental cycle with two morphological distinct infectious and reproductive forms: the elementary- and the reticulate body. The life cycle proceeds for 48 – 72 h and ends with pathogen release that may damage the host cells [ 13 ]. Infection with this pathogen is accompanied by cytoplasmic alterations and damage of the host cells [ 12 ]. A balance between pro- and anti-apoptotic influences by Chlamydia can be postulated. On one hand it has been shown that Chlamydia psittaci and Chlamydia trachomatis can induce apoptosis in vitro [ 14 , 15 ]. On the other hand established cell lines infected with Cpn or Chlamydia trachomatis were protected from apoptotic cell death induced by various stimuli [ 16 , 17 ]. Most of the studies investigating pro- and anti-apoptotic activity of Cpn were performed in tumor cells or established cell lines. Since the character of cell death is affected by host cell type and Chlamydia strain it is of relevance to determine cell death in Chlamydia infected primary cells [ 18 ]. As regards the role of Cpn in atherosclerosis we need to understand death promoting and inhibiting capacities of Cpn in primary cultures of vascular cells. Human aortic smooth muscle cells (HASMC) play an important role in the development of atherosclerotic lesions [ 19 ]. Therefore, we sought to clarify the nature of cell death induced by Cpn in HASMC. Apoptosis and necrosis represent two extreme morphologically defined forms of cell death [ 20 - 23 ]. Recently the hybrid term aponecrosis was introduced describing the incomplete execution of the internal apoptotic pathway and the following necrotic degeneration [ 24 ]. Whether smooth muscle cell death is apoptotic, necrotic or even aponecrotic in nature would predictably influence the inflammatory response in the plaque. Necrotic cells releasing their contents provoke inflammation, whereas in apoptosis dead cells are ingested by neighboring cells and, therefore, do not provoke an inflammatory response [ 25 ]. In order to characterize the fate of HASMC following infection with Cpn we analyzed cell death by morphological and biochemical methods. Results Various morphological characteristics of bacterial infection in HASMC In order to examine the morphological characteristics of Cpn infection HASMC as well as the tumor cell line HEp-2 were infected with Cpn-K6 and Cpn-VR1310, respectively, and stained after 24 h, 48 h and 72 h with anti-Cpn-MOMP antibody. According to the staining protocol intra- and extracellular localization of Cpn could be distinguished (Figure 1 ). Already after 24 h, but also after 48 h and 72 h, HASMC as well as HEp-2 cells were infected with numerous bacteria. Round or oval shaped smooth bordered intracellular inclusions were found in tumor cells (Figure 1E ) as well as HASMC (Figure 1B,D ). In contrast to tumor cells in primary HASMC Cpn were found also as single or rarely as irregular shaped, aggregated immunoreactive spots (Figure 1A, C , 4D ). Quantification of morphology in HASMC revealed that after 72 h all cells were infected with at least one chlamydial spot (100% infection rate). Approximately 25% of the cells carried one or multiple inclusions. The different morphological characteristics in HASMC (inclusions, single and aggregated spots) could also be discriminated 48 h post infection. The ratio of cells carrying inclusions versus cells carrying spots was much higher in HEp-2 cells compared to HASMC irrespective of the infectious dose (Figure 2A ). This ratio did not considerably vary between different Cpn strains K6 and VR1310 or as a function of the number of bacteria used for infection. Spots were usually much lower in number and not as prominently observed in HEp-2 cells compared to HASMC (Figure 2C ). In HASMC the number of spots increased with higher infectious doses while at the same time the number of inclusions remained constant (Figure 2B ) in contrast to Cpn infections of HEp-2 cells where more inclusions were formed at higher infectious doses without an increase of spots. Cpn lyse HASMC in a dose-dependent manner To compare the lytic capacity of different Cpn isolates and to identify a possible time course of host cell death during the chlamydial infectious cycle, LDH release of HASMC was analyzed at 24, 48 and 72 h post infection. HASMC were inoculated with two different Cpn isolates: Cpn-VR1310 and Cpn-K6 at serial 2-fold dilutions ranging from 1 to 32 inclusion forming unit (IFU)/cell (Figure 3 ). Both isolates were found to be free of Mycoplasma spec . contamination as assessed by PCR. LDH release of mock treated HASMC or of cells inoculated with heat inactivated bacteria was comparable to the spontaneous LDH release of uninfected cells (data not shown). Both Cpn isolates lysed HASMC in a strictly dose dependent fashion at all time points analyzed. Cpn mediated host cell lysis at 48 h post infection is shown in figure 3A . Data obtained from 24 h and 72 h post infection were similar and consistent with the displayed data subset. The levels of HASMC lysis induced by infection with 8 IFU/cell of Cpn-K6 (0.9% ± 3) and Cpn-VR1310 (18.6% ± 8) indicate large differences in the lytic capabilities of both Cpn isolates. Comparison of Cpn-induced LDH release at different time points (24, 48 and 72 h) demonstrated a constant increase in Chlamydia -specific HASMC lysis indicating continuous triggering of cell death throughout the whole infectious cycle (Figure 3B ). In order to reach the same amount of HASMC lysis (27% ± 2) 4 IFU/cell of Cpn-VR1310 and 32 IFU/cell of Cpn-K6 respectively were required. This suggests that for a given degree of HASMC lysis either high amounts of a lytically less active isolate or lower amounts of a lytically more active isolate are required. Cpn infection of HASMC causes nuclear chromatin condensation as well as extensive damage of organelles and cell membrane In order to discriminate between apoptotic and necrotic cell death of infected HASMC ultrastructural analysis by transmission electron microscopy (TEM) was carried out. As a control apoptosis was induced by treatment with the well described kinase inhibitor staurosporin. Untreated HASMC had an elongated nucleus rich in euchromatin with one or two well-defined nucleoli (Figure 4A ). The cytoplasm was scattered by numerous narrow-spaced profiles of rough endoplasmic reticulum (rER), a small Golgi complex and single elongated mitochondria. In Cpn infected cultures a variable number of HASMC harbored large inclusions containing many infectious elementary bodies and metabolically active reticular bodies (Figure 4B ). The nucleus of these cells was slightly rounded but chromatin structure as well as morphology of organelles was unchanged. A large number of infected HASMC without Cpn inclusions were destroyed (Figure 4C ). The round shaped nuclei showed a distinct condensation of heterochromatin and condensed nucleoli. Cell organelles including mitochondria and rER profiles were dilated and cell membranes were completely disrupted. Detection of single bacteria by TEM was possible only on rare occasions since distorted organelles in dying cells were barely distinguishable from Cpn. Apoptotic HASMC treated with staurosporin showed a distinct condensation of chromatin (Figure 4D ) and nuclei were fragmented frequently; damage of cell membranes and swelling of organelles appeared to a lower extent than in Cpn infected HASMC. In general the overall ultrastructural morphology of HASMC infected with spot-like Cpn infection suggests a hybrid form of cell death: it shares chromatin condensation with apoptosis and damage of organelles and membranes with necrosis. Spot-like infection induces aponecrotic morphology in HASMC The combination of chromatin staining by DAPI with labeling of DNA strand breaks via TUNEL staining and assessment of membrane integrity by NHS-biotin staining allows a distinction between necrosis as well as early and late phases of apoptosis using confocal laser scanning microscopy [ 23 ]. Cells dying by necrosis exhibit NHS-biotin labeling of the cytoplasm but not chromatin condensation or TUNEL staining of nuclei. Cells in early phases of apoptosis display nuclei with condensed or fragmented chromatin, with or without TUNEL staining, and are surrounded by NHS-biotin negative cytoplasm due to intact cell membranes. In late phases of the apoptotic cell death TUNEL positive nuclei occur together with a NHS-biotin positive cytoplasm. HASMC were infected with Cpn for chosen periods of time or were inoculated with mock isolates or treated with staurosporin. Cells infected with Cpn-VR1310 (Figure 5E, F ) showed an identical staining pattern as cells infected with Cpn-K6 (Figure 5A – D ). Mock inoculated cells displayed round or oval shaped nuclei with finely dispersed chromatin visualized by DAPI staining and were always TUNEL negative. NHS-biotin was bound exclusively to the cell surface indicating intact membranes (Figure 5G ). In Cpn infected cultures a variable labeling pattern of HASMC was observed. Cells with Cpn inclusions always displayed normal chromatin staining and no TUNEL or NHS-biotin labeling (Figure 5A ). HASMC with spot-like infection either contained an unaltered nucleus and NHS-biotin negative cytoplasm (Figure 5B ) or showed a TUNEL positive nucleus with condensed chromatin embedded in NHS-biotin stained cytoplasm (Figure 5C, E ). In HASMC displaying both Cpn spots and aggregates the nucleus was always labeled by TUNEL and the cytoplasm by NHS-biotin (Figure 5E, F ). Cpn-infected HASMC never showed membrane damage in combination with an unaltered nucleus, representing necrotic cell death. Cpn-infected HASMC displaying TUNEL positive nuclei and damaged membranes could be detected at 24, 48 and 72 h post infection. 8 h post infection was the earliest time point when TUNEL positive nuclei could be detected, suggesting induction of apoptosis during the whole life cycle of Cpn (not shown). TUNEL-positive HASMC were found even at infectious doses of 1 IFU / cell for Cpn-VR1310 and 8 IFU / cell for Cpn-K6 and the amount of TUNEL-positive HASMC increased strictly dependent on the chlamydial dose used for infection (Figure 5I ). Again, both Cpn strains used in this study elicited the same features of cell death in HASMC according to the infection morphology. Identical to the lytic capacity (Figure 3A ) of these strains, K6 had to be used at higher IFU compared to VR1310 to generate a similar number of cells with TUNEL-positive nuclei and NHS-biotin positive cytoplasm. In contrast to HASMC neither Cpn-K6 nor Cpn-VR1310 induced a concentration dependent increase in TUNEL positive nuclei in HEp-2 cells and numbers of TUNEL positive nuclei remained below 10% (not shown). In staurosporin treated cells the rounded nuclei were fragmented or exhibited chromatin that was condensed marginally, whereas NHS-biotin labeling was predominantly negative (Figure 5H ). Just a minority of the cells revealed a condensed TUNEL-positive nucleus and at the same time were also NHS-biotin positive reflecting late stages of apoptotic cell death. Cpn infection induces phosphatidylserine externalization in HASMC HASMC were infected for 48 h with Cpn-K6 or -VR1310 in 2-fold dilutions ranging from 8 to 386 IFU/cell or from 1 to 32 IFU/cell, respectively. As controls cultures were inoculated with mock isolates or with heat inactivated bacteria (data not shown) and analyzed for annexin-V binding and propidium iodide uptake. Uninfected as well as Na-azide treated HASMC served as negative and staurosporin-treated cells as positive controls for apoptosis (Figure 6C upper right and upper left panel). Control cultures inoculated with mock lysate (Figure 6 ) or heat inactivated bacteria and untreated cells showed always less than 20% annexin-V single positive cells. This proportion of annexin-V positive cells represented background staining, probably due to the experimental procedure. Upon infection with 128 IFU/cell of Cpn-K6 for 72 h the number of single annexin-V positive cells increased to more than 50% (Figure 6B ). Upon Cpn infection annexin-V and propidium iodide double-positive staining reached 10% as compared with 4% in mock-inoculated and 3% in untreated HASMC (not shown). In staurosporin treated cultures the amount of both annexin-V single and double positive cells increased strongly (Figure 6 upper left panel) indicating a continued induction of classical apoptotic cell death. The amount of annexin-V single positive HASMC infected with Cpn increased linearly in a dose- (Figure 6A ) and time- (Figure 6B ) dependent manner. Cpn-VR1310 again induced phosphatidylserine externalization at much lower doses compared to Cpn-K6. Neither mock inoculation (Figure 5A, B 6C lower left panel), nor inoculation with heat inactivated bacteria (data not shown), induced apoptosis. This indicates that infection with living bacteria is required for induction of cell-death. Taken together, the dose- and time dependence between infection and phosphatidylserine externalization were consistent with induction of early apoptotic events. Chlamydia infection in HASMC induces caspase-3 independent cell death by dissipation of mitochondrial membrane potential After induction of apoptosis an enzymatic cascade is activated leading to the disassembly of the cell [ 23 ]. Key enzyme in this cascade is caspase-3. However, in some cases apoptosis can proceed without involvement of caspase-3 [ 26 ]. In order to investigate the role of caspase-3 HASMC were infected with Cpn-K6 128 IFU/cell or Cpn-VR1310 32 IFU/cell, respectively, for 48 h (24 h and 72 h not shown) or treated with 1 μM staurosporin for 12 h to induce apoptosis. Samples were analyzed for caspase-3 activation by measuring enzyme activity after lysing the cells. No activity of caspase-3 in Cpn infected HASMC (Figure 7A ) could be detected. In contrast distinct caspase-3 activity was found in staurosporin-treated cells. A major role in the induction and regulation of cell death is played by mitochondria. Pro-apoptotic factors result at mitochondria in the dissipation of the mitochondrial membrane potential and release of cytochrome c. After infection of HASMC with Cpn-K6 or Cpn-VR1310 cells were loaded with TMRE (tetramethylrhodamineethylester) [ 27 ], a dye, that is only taken up into mitochondria with an intact mitochondrial potential [ 28 ]. Cells with intact membranes excluding propidium iodide were analyzed by flow cytometry. The number of TMRE negative cells strictly increased with the chlamydial dose whereas blocking of chlamydial protein synthesis by addition of chloramphenicol completely abrogated this effect (Figure 7B ). Moreover, labeling of Cpn-K6 infected or mock treated cells for cytochrome c and mitotracker red revealed distinct loss of cytochrome c staining of the mitochondria in the infected population (Figure 7D ) but not in mock treated cells (Figure 7C ). Taken together, the data suggest Chlamydia pneumoniae induce caspase-independent apoptosis-like cell death in infected HASMC by involving mitochondrial membrane dissipation resulting in the release of cytochrome c. Discussion In our in vitro study cell death was only induced in HASMC by infection with viable Cpn but not after incubation with heat inactivated or chloramphenicol treated bacteria. Additionally, different Cpn isolates showed various lytic activities towards host cells that correlated to different reproduction rates in tumor cells as measured by real time PCR [ 29 ]. A reproductive infection would seem to be a prerequisite for cell death induction in HASMC. The observation that heat inactivated or chloramphenicol treated bacteria never caused changes of the host cells underlines that cell death induction is not a bystander effect of a potential cytopathic chlamydial component such as the heat stable LPS. It can be concluded that HASMC death depends on internalization of the Chlamydia and is associated with the suppression of bacterial replication and productive infection. The necessity for a reproductive infection was also shown for Chlamydia psittaci which induced caspase 3 independent apoptosis in macrophages and epithelial cells only after reproductive infection along with bacterial protein synthesis [ 15 ]. Cell death induction was never found in cells harbouring inclusions which normally are viewed as the manifestation of a reproductive infection. HASMC formed inclusions with a relatively low frequency compared to tumor cells. In contrast, spots resulting presumably from single bacteria were found frequently illustrating a different infection morphology of Cpn in these primary vascular cells. Currently no data are available regarding metabolism, protein synthesis and reproduction or functional relation with host cells of these single, spot-like bacteria. From our data it only can be concluded that these single bacteria are sensitive to chloramphenicol which inhibits protein synthesis in Cpn and abrogates cell death induction. Cell death seems to result from an interaction of the host cell with these single bacteria as it is ongoing for extended times. Cpn induced cell death in HASMC did not fulfill all criteria for apoptosis or classical necrosis. Condensation of chromatin, positive TUNEL staining, externalization of phosphatidylserine, inhibition of TMRE labeling and loss of cytochrome c immunoreactivity in mitochondria were clear markers for apoptosis. But early damage of cell membrane and organelles as found by TEM and NHS-biotin labeling indicated necrotic type of cell death in Cpn infected cells. According to previous classifications Cpn mediated cell death belongs to apoptosis-like cell death [ 30 ] or can be termed aponecrosis due to characteristic ultrastructural features [ 24 ]. Cpn induced cell death in HASMC having features of necrosis and apoptosis is in agreement with a study employing mouse embryonic fibroblasts infected with C. trachomatis which died through a combination of necrosis and apoptosis [ 18 ]. If cell death induction by Cpn in HASMC is caspase-3 independent the question arises about alternative pathways that are involved. Mitochondria are important regulators of cell death and loss of mitochondrial permeability transition is an important indication for apoptosis [ 31 ]. Here we show by TMRE labeling dissipation of the mitochondrial potential of Chlamydia infected cells. This effect could be inhibited by blocking bacterial protein synthesis using chloramphenicol. Moreover, cytochrome c was released from mitochondria of infected HASMC as indicated by loss of immunorecactivity. The important role of mitochondria in Chlamydia induced cell death has also been shown in tumor cell lines infected with Chlamydia psittaci [ 14 ]. Pro-apoptotic Bcl-2 family members like Bax were activated and apoptosis induction occurred during the whole chlamydial life cycle. Apoptosis induced by Chlamydia psittaci was blocked in cells over expressing Bax inhibitor and anti-apoptotic Bcl-2 [ 14 ]. Bax -/- mouse embryonic fibroblasts were shown to be resistant to Chlamydia muridarum -induced apoptosis and fewer bacteria were recovered after two infection cycles. This suggests that Bax-dependent apoptosis may be used to initiate a new round of infection [ 32 ]. In a second study, Chlamydia trachomatis infected mouse fibroblasts showed a higher level of apoptosis than Chlamydia infected HeLa cells as measured by annexin V / propidium iodide double labeling [ 18 ]. This not only shows that Bax activation could reflect a stress response in infected cells [ 33 ] but that primary cells show a different stress response to Chlamydia infection than tumor cells. Beside Bax other factors like apoptosis-inducing factor (AIF) may be involved in caspase-independent cell death [ 34 ] but they have never been investigated in Chlamydia -induced cell death. AIF which is upregulated and translocated from the mitochondria to the cytosol may play an important role in induction of caspase-3 independent cell death [ 34 - 36 ]. Injection of AIF into the cytoplasm induces similar chromatin condensation coupled to phosphatidylserine exposure as we found in Cpn infected HASMC [ 37 ]. Future experiments will have to show if AIF, Bax or other Bcl-2 family members are involved in Cpn induced HASMC death. There exist numerous studies showing that Cpn and Chlamydia trachomatis prevent cells in vitro from undergoing apoptosis [ 16 , 17 , 38 - 44 ]. However, these experiments were performed on established cell lines. Our study shows that in primary cultures of cells such as HASMC various types of Cpn infection occur. Spot like infection resulted in aponecrosis whereas cells with Cpn inclusions appeared to be prevented from cell death. Apparently Cpn in inclusions have evolved strategies to inhibit cell death presumably to complete the developmental cycle. About possible mechanisms can be speculated: it has been shown that Cpn inserts components of a type-III secretion system into the inclusion membrane [ 45 ] and releases factors into the host cell [ 46 - 48 ] that affect host cell metabolism. Cpn secretes proteins that translocate from the inclusion to the cytoplasm [ 47 , 49 ]. CADD, a secreted protein from Chlamydia trachomatis , has been shown to interact with death domains and co-localizes with Fas. Recombinant CADD has been shown to induce caspase dependent apoptosis in several tumor cell lines [ 50 ]. Therefore, other yet unidentified proteins from Chlamydia could interact with the host cell death machinery and inhibit or promote cell death. Currently, it is unclear how the observed morphology of Cpn in HASMC and the induced aponecrotic cell death affects the progression of atherosclerosis. Cpn infection of HASMC was accompanied by phosphatidylserine externalization which might result in the ingestion of dying cells by macrophages. Macrophages will then either stop further progress of Cpn mediated damage or in turn become activated and thereby contribute to inflammation in the atherosclerotic vessel [ 51 , 52 ]. Early membrane damage found in the aponecrotic HASMC may provoke or assist inflammation in the atherosclerotic vessel due to release of cellular constituents. Finally damage of vascular smooth muscle cells by Cpn infection might lead to plaque instability [ 19 ]. Conclusions Cpn infection of cultured HASMC results in cell death that can be termed aponecrosis due to the sharing of apoptotic and necrotic features. This form of cell death occurs just in cells bearing single or aggregated bacteria but not in cells with inclusions. Aponecrotic HASMC may in vivo assist chronic infection in the vascular wall supporting the progression of atherosclerosis. Methods Chlamydial organisms, cells, antibodies and reagents Cpn-VR1310 and Cpn isolate Kajaani 6 (Cpn-K6) were kindly provided by G. Christiansen (Institute of Microbiology and Immunology, University Aarhus, Denmark) and M. Puolakkainen (Haartman Institute, University of Helsinki, Finland), respectively. HEp-2 cells obtained from ATCC (Manassas, USA) were cultivated in MEM-Eagle's (Gibco BRL, Invitrogen, Basel, Switzerland) and HASMC purchased from Clonetics (Verviers, Belgium) were maintained in DMEM supplemented with 10% FCS, 2 mM glutamine, all from Gibco BRL, Invitrogen, and 10 μg/ml neomycin (Sigma Chemicals, Buchs Switzerland). Cells and chlamydial organisms were tested for Mycoplasma spec . contamination by PCR [ 53 ] and 4',6-Diamidin-2'-phenylindol-dihydrochloride (DAPI) (Roche, Mannheim, Germany) staining and were mycoplasma free. Anti-Cpn-MOMP monoclonal antibody was from DAKO (Zug, Switzerland); anti-cytochrome c antibody raised in sheep was from Sigma, FITC-labeled monoclonal anti- Chlamydia -LPS antibody from Biorad (Redmont, USA). Donkey anti-mouse polyclonal antibody was obtained from Jackson Immuno Research (Baltimore, USA) and staurosporin from Sigma Chemicals. Chlamydial culture and infection of HASMC Cpn was cultured according to an established protocol [ 54 ]. Briefly, confluent HEp-2 cells were inoculated with Cpn at 1 IFU/cell in medium, centrifuged at 1000 × g for 1 h and grown for 72 h in medium containing 10% FCS, 2 mM glutamine, 10 μg/ml neomycin and 2 μg/ml cycloheximide (Sigma). Subsequently, cells were harvested, disrupted by sonification and bacteria were purified on a discontinuous renografin (Sigma Chemicals) gradient as previously described [ 55 ]. Non infected HEp-2 cells were subjected to the same harvesting / purification procedure and referred to as mock controls. In order to determine the chlamydial titer (IFU/ml), confluent HEp-2 cells grown on cover slips were infected with serial 2-fold dilutions of purified bacteria. After 72 h cells were fixed, stained with anti- Chlamydia -LPS antibody, inclusions were counted under a fluorescence microscope and the titer was calculated. HASMC were seeded into 6-well plates (Nunc, Roskilde, Denmark) at a density of 2.4 × 10 4 cells/cm 2 one day prior to infection. Infection with chlamydial organisms was performed in DMEM using titers between 8 and 386 IFU/cell which were added to HASMC, centrifuged onto the cells for 1 h at 1000 × g. Subsequently, the supernatant was replaced by DMEM containing 10% FCS, 2 mM glutamine and 10 μg/ml neomycin. In some experiments chloramphenicol 0.1 mg/ml was added analysis or cells were treated with 1 μM staurosporin for 14 h. Cells were grown at 37°C in the presence of 5% CO 2 until they were analyzed. Determination of infection morphology, intra- versus extracellular staining of Cpn Live HASMC and HEp-2 cells were incubated with anti-Cpn-MOMP antibody (1:50) in PBS + 2% BSA for 1 h at room temperature (RT). After washing and fixation with 2% paraformaldehyde + 3% sucrose in PBS at RT, cells were incubated for 1 h with donkey anti-mouse-FITC antibody, diluted 1:100 in PBS + 2% BSA. Bacteria located extracellularly are stained exclusively green by this step. Subsequently, cells were lysed with 0.1% Triton-X 100 for 1 min at RT, rinsed with PBS + 2% BSA for 20 min, and incubated with anti-Cpn-MOMP antibody (1:50) followed by Texas Red-labeled donkey anti-mouse antibody (1:200) in 2% BSA in PBS. The second staining step identifies both intracellular and extracellular bacteria. Counting of cells with inclusions or spots as well as the number of inclusions and spots were performed on volume data obtained using a laser scanning microscope (SP2, Leica, Heidelberg, Germany) followed by image analysis using the software package Imaris (Bitplane, Zurich, Switzerland). Determination of cytotoxicity by lactate dehydrogenase release (LDH) LDH release of HASMC was assessed using a cytotoxicity detection kit (Roche) following the manufacturer's instruction. In brief, HASMC were seeded in a 96-well plate (NUNC) at a density of 2.4 × 10 4 cells/cm 2 , infected with Cpn between 2 and 32 IFU/cell and cultivated for 24 h, 48 h and 72 h. For cell death analysis 100 μl of cell-free supernatant was harvested, mixed with dye solution, incubated for 20 min and absorption was measured at 490 nm. Percent Chlamydia -specific lysis was calculated according to the formula: ((experimental value – spontaneous release)/(maximum release – spontaneous release) × 100. Spontaneous release corresponded to untreated HASMC and infected cells lysed additionally with Triton-X 100 showed the maximal release. Transmission electron microscopy (TEM) Cells were fixed in 50 mM sodium cacodylate buffer pH 7.3 containing 2% glutaraldehyde and 0.8% paraformaldehyde and postfixed with 1% OsO 4 in 50 mM sodium cacodylate buffer, pH 7.3 dehydrated in an ethanol series and embedded into epon (Catalys). Ultrathin sections of 50 nm were contrasted with uranyl acetate and lead citrate and studied with a CM 100 transmission electron microscope (Phillips, Leiden, Netherlands). TUNEL-, NHS-biotin-, Cpn- and DAPI-staining for analysis by confocal microscopy Cells were trypsinized, washed in PBS and incubated with 0.1 mg/ml NHS-biotin (Pierce, Rockford, USA) in PBS for 30 min on ice, fixed with 3% paraformaldehyde + 2% sucrose in PBS and cytospinned onto glass slides. Samples were permeabilized with 0.1% Triton-X 100 in PBS for 1 min at RT and stained for DNA strand breaks using terminal transferase (Roche), incorporating dUTP-FITC (Roche) overnight at 37°C as previously described [ 56 ]. Subsequently, cells were incubated with anti-Cpn-MOMP antibody (1:50) and detected with donkey anti-mouse Texas Red antibody (1:200) in 2% BSA in PBS for 1 h, followed for 45 min by 0.5 μg/ml streptavidin-Cy5 (Jackson Immuno Research) detection of NHS-biotin. Nuclei and DNA of bacteria were stained with 1 μg/ml DAPI in PBS for 10 min. Samples were then embedded in fluorescence mounting medium (Dako) and analyzed using a confocal laser scanning microscope (SP2, Leica, Heidelberg, Germany). Annexin-V/propidium iodide staining for flow cytometry Phosphatidylserine exposure and propidium iodide incorporation was assessed using the annexin-V-FLUOS staining kit (Roche). In brief, supernatants of HASMC cultures were collected and mixed with trypsinized cells. Samples were washed in PBS followed by 20 min staining with annexin-V / propidium iodide staining solution and analyzed by flow cytometry (Becton Dickinson, Basel, Switzerland). Determination of caspase-3 activity 0.2 × 10 6 HASMC consisting of floating and adherent cells were lysed at 4°C in cell lysis buffer (20 mM PIPES, pH 7.2, 100 mM NaCl, 1 mM EDTA, 10 mM DTT, 0.1% CHAPS, 10% sucrose), subjected to three freeze-thaw cycles and centrifuged at 14000 × g for 10 min. Protein content of supernatants was measured using the micro BCA protein assay reagent kit (Pierce). Cell lysates containing 40 μg protein were incubated with DEVD- p -nitroanilide (0.8 mM) in lysis buffer and absorption was measured every 5 min in a spectrophotometer at 405 nm for 6 h at 37°C. Caspase-3 activity was calculated as the initial ascending slope of absorption depicted as arbitrary units. TMRE and propidium iodide staining for analysis of mitochondrial membrane potential Cells were stained with TMRE 100 nM and propidium iodide 2 μg/ml in DMEM with 10% FCS for 30 min at 37°C and 5% CO 2 , collected by trypsinization, combined with detached cells at time of trypsinization and analysed by flow cytometry (Becton Dickinson). Cytochrome c and mitotracker red staining for confocal microscopy analysis Cells were incubated in the presence of 50 nM mitotracker red for 30 min, subsequently fixed with 3% paraformaldehyde + 2% sucrose and were cytospinned onto glass slides. Samples were permeabilized with 0.1% Triton-X 100 in PBS for 1 min at RT. Subsequently stained with sheep anti cytochrome c (1 : 2000) in 3 % BSA in PBS for 1 h at 37°C followed by detetion with donkey anti sheep biotin antibody (1:200) in 0.5 % BSA in PBS for 1 h at RT. Cells were labeled with streptavidin Cy-5 (1 : 200) for 1 h at RT and analysed under a confocal laser scanning microscope. Authors' contributions CD performed all infections and analyzed morphological markers for cell death. CFM and MKJS performed FACS experiments. DG initiated the study and performed the first infections of vascular cells. MW analyzed viability experiments and provided input for exeriments. PG provided critical intellectual input to the study and organized financial support. UZ established assays for the detection of caspase 3, morphological cell death markers and was leading the study. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC547904.xml |
520747 | Nestin-positive mesenchymal stem cells favour the astroglial lineage in neural progenitors and stem cells by releasing active BMP4 | Background Spontaneous repair is limited after CNS injury or degeneration because neurogenesis and axonal regrowth rarely occur in the adult brain. As a result, cell transplantation has raised much interest as potential treatment for patients with CNS lesions. Several types of cells have been considered as candidates for such cell transplantation and replacement therapies. Foetal brain tissue has already been shown to have significant effects in patients with Parkinson's disease. Clinical use of the foetal brain tissue is, however, limited by ethical and technical problems as it requires high numbers of grafted foetal cells and immunosuppression. Alternatively, several reports suggested that mesenchymal stem cells, isolated from adult bone marrow, are multipotent cells and could be used in autograft approach for replacement therapies. Results In this study, we addressed the question of the possible influence of mesenchymal stem cells on neural stem cell fate. We have previously reported that adult rat mesenchymal stem cells are able to express nestin in defined culture conditions (in the absence of serum and after 25 cell population doublings) and we report here that nestin-positive (but not nestin-negative) mesenchymal stem cells are able to favour the astroglial lineage in neural progenitors and stem cells cultivated from embryonic striatum. The increase of the number of GFAP-positive cells is associated with a significant decrease of the number of Tuj1- and O4-positive cells. Using quantitative RT-PCR, we demonstrate that mesenchymal stem cells express LIF, CNTF, BMP2 and BMP4 mRNAs, four cytokines known to play a role in astroglial fate decision. In this model, BMP4 is responsible for the astroglial stimulation and oligodendroglial inhibition, as 1) this cytokine is present in a biologically-active form only in nestin-positive mesenchymal stem cells conditioned medium and 2) anti-BMP4 antibodies inhibit the nestin-positive mesenchymal stem cells conditioned medium inducing effect on astrogliogenesis. Conclusions When thinking carefully about mesenchymal stem cells as candidates for cellular therapy in neurological diseases, their effects on resident neural cell fate have to be considered. | Background During development of the central nervous system (CNS), all types of neuronal and macroglial cells derive from neuroepithelial neural stem cells (NSCs) [ 1 , 2 ]. NSCs self-renew and continue to function as source of new cells in adults [ 3 , 4 ]. The fate determination of neural stem cells is regulated by cell-intrinsic programs as well as extrinsic cues from the surrounding environment [ 5 ]. In the adult, the molecular mechanisms that regulate the production of new neurons in the dentate gyrus or in the subventricular zone are still unknown, although extrinsic factors expressed for example by astrocytes could play a role [ 6 ]. BMPs (Bone Morphogenetic Proteins) are secreted members of the TGF-β superfamily. Together with their receptors, they are abundantly expressed in the brain both during embryogenesis and in the adult [ 7 - 9 ]. More specifically, an expression of BMP4 was demonstrated in different cell types like ectodermal cells [ 10 ], radial glia [ 11 ], haematopoietic cells [ 12 ], chondrocytes [ 13 ] and stromal cells [ 14 ]. As for other members of the TGF-β superfamily, BMPs signal transduction is triggered by binding to type 1 and type 2 serine-threonine kinase receptors, inducing their dimerization. In this manner, BMP2 and BMP4 signalling specifically leads to the formation of BMPR-1A/BMPR-II and BMPR-1B/BMPR-II heterodimers [ 15 ]. BMPs have been shown to play a role in patterning and cellular fate determination in many tissues [ 16 ]. For examples, during development, BMPs are involved in the induction of the neuroectoderm, the patterning of the dorsal roof of neural tube, the development of neural crests and of the peripheral nervous system [ 10 , 17 - 20 ]. In postnatal animals, they promote the differentiation of neuronal precursors in the spinal cord [ 21 ] and in the cortex [ 22 , 23 ]. Finally, they facilitate an astroglial lineage commitment of forebrain subventricular zone progenitor cells [ 24 ]. Cellular therapies using stem cells are promising approaches for the treatment of several chronic or acute neurological diseases such as Parkinson's [ 25 ] or Huntington's diseases [ 26 ] or spinal cord injuries [ 27 ]. One main problem relates to the origin and the nature of the cells to be used for such procedures. Foetal brain tissue has already been shown to have significant effects in patients with Parkinson's disease. Clinical use of the foetal brain tissue is, however, limited by ethical and technical problems as it requires high numbers of grafted foetal cells and a possible immunosuppression. Alternatively, somatic stem cells derived from adult tissues seem to be better candidates for cell replacement therapy [ 28 , 29 ]. These observations raise hopes in cell replacement strategies based on an autograft approach. However, the exact mechanism by which mesenchymal stem cells (MSCs) adopt a neural fate is not completely understood as recent in vivo studies demonstrated that MSCs fuse with host neuronal cells [ 30 , 31 ]. These observations impose a better knowledge of the mechanisms underlying the phenotypic plasticity of somatic stem cells and the characterization of their daughter differentiated cells is a prerequisite before considering their use in the treatment of human patients. Recently, we demonstrated that two phenotypes of MSCs could be obtained in culture: nestin-positive MSCs (npMSCs) which are able to integrate some extrinsic signals when co-cultured with neurons leading to a differentiation into astrocyte-like cells and nestin-negative MSCs (nnMSCs) which are unable to adopt a neural phenotype but remain able to differentiate into adipocytes, chondrocytes or osteocytes [ 32 ]. When considering the use of MSCs as a source of material for cell replacement protocols in neurological diseases, one has also to be concerned by a possible effect of these MSCs on the host nervous tissue and more particularly on immature resident neural cells. In several models of neurological diseases, grafted MSCs were shown to favour host CNS regeneration rather than to express themselves a neural phenotype [ 33 , 34 ]. This positive effect of MSC grafts could result from the release of factors acting on resident immature cells in the adult brain. Recently, it has been demonstrated that following ischemia in the adult striatum, intra-ventricular EGF injections are able to stimulate neurogenesis from resident neural stem cells, although EGF is devoid of any effect in non-ischemic striatum [ 35 ]. These observations emphase the importance of the lesion priming in order to respond to extrinsic factors or cytokines. In this paper, we address the question of the influence of MSCs on neural stem cells in vitro and demonstrate that MSCs favour astroglial lineage. We observe that npMSCs are able to stimulate astroglial fate in striatal progenitor cultures and to repress neuronal and oligodendroglial fate through the release of diffusible factor(s). Using BrdU incorporation, we demonstrate that this npMSC conditioned medium has no effect on the astrocytic or oligodendrocytic progenitor proliferation. Propidium iodide incorporations suggest that the npMSCs conditioned medium protect GFAP-positive cells from cell death in comparison to the effect of nnMSCs conditioned medium or to the control condition. Meanwhile, no increase of cell death is observed in neuronal and oligodendroglial populations. We then demonstrated that BMP4 is present in a biologically-active form in the npMSCs but not in nnMSCs conditioned medium and is responsible for both the increase of astroglial numbers and the inhibition of oligodendrocyte differentiation in striatal NSC cultures. Results Nestin-positive MSCs increase astrocytes number in differentiating neural stem cell cultures Neurosphere-derived cells from GFP-positive E16 green mouse striata include neural stem cells and progenitors which are still proliferating but already committed to a given cell fate. Upon transfer on adherent surfaces (poly-ornithine coated dishes), these cells spread and spontaneously differentiate as follows after 5 days of culture: 44.2 ± 2.5% GFAP-positive cells (astroglial cells), 15 ± 2.1% Tuj1-positive cells (neuronal cells) and 5.86 ± 0.6% 04-positive cells (oligodendroglial cells). When co-cultivated with nestin-positive MSCs (npMSCs), phenotypic allocation of neurosphere-derived cells (identified as GFP-positive cells) become strikingly different: GFAP labelling is increased to 78.5 ± 3.9% (Fig. 1A ) (n = 12, ***Student T-test, P < 0.001), while Tuj-1 and O4 are decreased to respectively 3.2 ± 1.1% (Fig. 1B ) (n = 12, ***Student T test, P < 0.0001) and 2.9 ± 0.5% (Fig. 1C ) (n = 8, ***Student T test, P < 0.0001). On the other hand, if neurospheres are co-cultured with nestin-negative MSCs (nnMSCs), the percentage of labelled cells were 52.1 ± 2.9%, 4.7 ± 0.8% and 7.8 ± 1.5 for respectively GFAP, Tuj-1 and O4. These numbers are not significantly different from the numbers obtained in control cultures for GFAP- and O4-positive cells (n = 8, Student T test, P > 0.05), while significantly lower for Tuj-1-postive cells (n = 8, ***Student T test, P < 0.001, Fig. 1G ). MSCs effect on neurospheres-derived astrocytes number is due to a soluble factor To further characterize the mechanism (soluble factor versus membrane-bound factor) responsible for such an increase in astrocytes number in presence of npMSCs, we tried to induce the differentiation of neurospheres with npMSC conditioned medium. The following data were obtained: 77.5 ± 2.5% of the neurosphere-derived cells became GFAP-positive, 4.3 ± 0.9% were Tuj1-positive and 1.5 ± 0.6% were 04-positive (n = 8, Fig. 2A,2B,2C ). Conversely, in nnMSCs conditioned medium, differentiated phenotypes were distributed as follows: 56 ± 3.5%, 3.7 ± 1.3% and 5.9 ± 1.3% (n = 7, Fig. 2G ) and were similar to results obtained in corresponding co-cultures. Moreover, we confirmed those results by absolute count: 1) in npMSCs conditioned medium 392/504 cells were GFAP-positive, 9/504 cells were O4-positive and 32/504 cells were Tuj1-positive ; 2) in nnMSCs conditioned medium 251/515 cells were GFAP-positive, 29/515 were O4-positive and 23/515 were Tuj1-positive; 3) in the control condition (DEM/F12 + B27), 211/531 cells were GFAP-positive, 27/531 cells were O4-positive and 167/531 cells were Tuj1-positive. Hence, regarding the phenotypic allocation, no significant difference is observed between neurosphere differentiated in co-culture and in conditioned medium. (Student T test, P > 0.05). This clearly suggest that npMSCs-derived soluble factor(s) is (are) responsible for an increase in the astrocytes number and a decrease in the neurons and oligodendrocytes numbers. Nestin-positive MSC conditioned medium act on the GFAP-positve cell death As mentioned above, neurosphere derived-cells include neural stem cells and progenitors which are still proliferating but already committed to a given cell fate. We characterised the proportion of each progenitor cell-type present in our neurospheres. After dissociation, cells were allowed to adhere for one hour in coated dish and then fixed, labelled and counted. In those conditions we observed 14.4 ± 7.1% of GFAP-positive cells, 7.1± 4.1% of Tuj1-positive cells and 3.4 ± 1.6% of O4-positive cells (these data were obtained by absolute counts on 1576 cells, Fig. 3A ). In order to define if npMSCs conditioned medium has an instructive or/and a selective effect on neural cells, we analysed the proliferative capacity and the cell death in the differentiating neurospheres. The BrdU incorporation in differentiating neurosphere-derived cells cultivated in npMSCs- or in nnMSCs-conditioned medium or in control (non-conditioned) medium (DEM/F12 + B27) did not show any significant differences (Statistical test ANOVA, P > 0.05, Fig. 3B,3C ). These data rule out a proliferative effect on already committed astrocyte progenitors or an inhibition of cell proliferation in oligodendroglial and neuronal progenitors. Furthermore, using propidium iodide incorporations and counting, we quantified the cell death in GFAP-, Tuj1- and O4-positive cell population during the differentiation culture in the three conditioned media (by npMSCs, by nnMSCs and non-conditioned). After 48 hours, we observed a significant decrease of the number of GFAP-positive cells which have incorporated the propidium iodide in npMSCs-conditioned medium (***Student T test, p < 0.0001) in comparison to the two other culture condition. This result suggests that npMSC conditioned medium partially inhibit the cell death in the GFAP-positive cell population. Nevertheless, no significant increase of cell death is observed in O4- and Tuj1-positive cell population (Student T test, p > 0.05). When propidum iodide incorporations are performed after four days of differentiation without renewing the conditioned media, no significant differences in the cell death could be observed in the three lineages, whatever the culture condition (Fig. 3D,3E ). BMP4 is present in nestin-positive conditioned medium and is responsible for the increase of GFAP-positive cells in differentiating NSCs Previous reports have shown that several secreted growth factors, including BMP2, BMP4, LIF and CNTF stimulate the differentiation of cultured neural precursors into astrocytes [ 36 - 41 ]. The expression of those cytokines was compared in nnMSCs and in npMSCs using quantitative RT-PCR. Compared to nnMSCs, we found that npMSCs slightly increased the expression level of BMP4 mRNA (190 ± 26%) while the BMP2 (36 ± 7%) and LIF (50 ± 10%) expression is slightly decreased (Fig. 4A ). No significant difference was observed concerning the CNTF mRNA expression. Western blotting analyses of npMSCs, nnMSCs and neurosphere-derived cells conditioned media showed that only npMSCs release in their culture medium the mature and biologically-active form of BMP4 (27 kDa), although nnMSCs and neurosphere-derived cells expressed the biologically-inactive precursor form of BMP4 (57 kDa) (Fig. 4B ). Finally, we observed that neurospheres-derived cells cultivated in npMSCs conditioned medium supplemented with an anti-BMP4 blocking antibodies, differentiate into 47.1 ± 1.5% GFAP-positive cells, 9.9 ± 0.9% Tuj1-positive cells and 5.7 ± 0.5% O4-positive cells (Fig. 4C ). Statistical analyses did not show significant differences in GFAP- or O4-positive cells compared to control culture conditions (student T test, p > 0.05, n = 5). However, the number of Tuj1-positive cells remains significantly lower (***student T test, p < 0.001, n = 5). We thus conclude that the major glial effects (increase of astrocytes and decrease of oligodendrocytes) of the npMSCs on the neural progenitor differentiation is due to the release of biologically-active BMP4. The inhibition of neuronal differentiation should be attributed to another yet uncharacterized soluble factor(s). Discussion During the last few years, a number of studies have addressed the phenotypic plasticity of MSCs. Most of them were performed in vivo and demonstrated that environmental factors play important roles in determining the ability of grafted MSCs to adopt a neural-like phenotype. Grafting of a subset of MSCs into the lateral ventricle of neonatal mice resulted in their migration within the forebrain and cerebellum, and their differentiation into astrocytes [ 42 ]. When MSCs were grafted into adult rat cerebellum after an ischemic lesion, they differentiated into cells expressing neuronal markers [ 43 ]. Nakano et al. [ 44 ] demonstrated that murine bone marrow cells differentiated into three distinct glial phenotypes (oligodendrocytes, astrocytes and microglia) when they were directly injected into the striatum of previously irradiated mice. Similarly, systemic injection of MSCs in lethally irradiated mice allowed their differentiation into neuronal and astroglial cells [ 45 ]. Interestingly, the systemic injection of MSCs in non-irradiated but brain-lesioned mice had positive effects on injury repair, but very few MSC-derived cells expressed neural marker in such conditions [ 22 , 33 , 34 , 46 , 47 ]. Given those results, somatic stem cells raise thus hopes in cell replacement strategies based on an autograft approach. Recent in vivo studies [ 30 , 31 ] demonstrated that MSCs adopt neural fate by fusion with host neural cells. At least so far, there is no a clear and conclusive demonstration of a real differentiation of MSCs in neural cells. However, all those data obtained in vivo suggested that somatic stem cells seem to be promising for cellular therapy in neural diseases whatever the mechanism responsible for. It remains that a better knowledge of the molecular and cellular mechanism underlying the neural phenotypic plasticity of MSCs is required before considering those cells in human graft protocols. As we mentioned above, environment is able to modify the cellular differentiation capacity. In this study, we addressed the question of a possible effect of MSCs on neural progenitor cell fate and we choose an in vitro approach. Co-culture experiments demonstrate that MSCs display multiple activities in the regulation of embryonic striatum-derived progenitors and stem cells. Their effect mainly depends of their age in vitro : npMSCs (more than 10 passages in vitro or 25 doubling cell populations) increase the astrocytes numbers while inhibiting neuronal and oligodendroglial numbers. Conversely, nnMSCs (with a maximum of 5 passages in vitro ) only decrease the neuronal differentiation. Since similar results were obtained using MSC conditioned media, we concluded that these effects could be mediated by diffusible factor(s). We then analysed the effect of various conditioned media on the cell proliferation (using BrdU incorporation) as well as the cell death (using propidium iodide incorporation) in each neurosphere-derived cell types and at two different differentiation period (48 or 96 hours). We only observed a significant decrease of propidium iodide incorporation in GFAP-positive cell population at 48 hours of differentiation suggesting that soluble factor(s) select astroglial lineage by protecting those cells from cell death. The activity which drives neurosphere-derived cells into astrocytes only occurs in npMSC conditioned medium while the neuronal differentiation-inhibiting activity is present both in np- and nnMSC conditioned media. We tried to identify the astrogliogenetic factor(s) by measuring the expression by nn- and npMSCs of cytokines known to promote an astroglial fate [ 36 - 41 ]. Our results show that only npMSCs express mature and biologically-active BMP4 while a biologically-inactive BMP4 precursor form is expressed and released by nnMSCs and neurosphere-derived cells in culture. Indeed, BMP4 is synthesized and released as an inactive precursor before being proteolytically activated by cleavage at the amino acid motif -Arg-Ser-Lys-Arg- [ 48 ]. Relatively little is known about the regulatory mechanisms controlling the susceptibility of individual TGF-β family members to proteolytic cleavage. However, recent studies suggest that members of the subtilisin-like proprotein convertase (SPC) family, SPC1 and SPC4, could enhance the cleavage of BMP4 precursor. The availability of biologically active BMPs may therefore be controlled by the released of their precursor followed by the action of the proprotein convertases [ 49 , 50 ]. The release by npMSCs of biologically-active form of BMP4 which promotes astroglial differentiation and inhibits oligodendroglial differentiation is consistent with previous studies demonstrating that BMPs play multiple roles in development [ 8 ]. Other studies have reported that the effects of BMPs are age- and tissue-dependent [ 51 , 52 ] and that BMPs promote astroglial differentiation and inhibit oligodendroglial differentiation when applied to cultures of cortical cells plated at E16 [ 23 ]. Likewise, brief treatment with BMPs induces astroglial fate in cultured neural precursors from embryonic mice [ 37 ] or in oligodendrocytes from newborn rats [ 51 ]. More recently, Rajan et al. [ 41 ] demonstrated that BMP4 induces astroglial differentiation of E14 and adult cortical neural stem cells from the subventricular zone when they are placed in high density culture. In this case, BMP4 has a true instructive role as NSCs cultures were clonal. Beside the instructive effect of BMP4 on glial lineages, some studies explained how BMP4 could act on astrocytic and oligodendrocytic precursors. A study realised on cerebellar primitive neurectodermal tumor cell line demonstrated that a high concentration of BMP2 and BMP4 attenuate apoptosis [ 53 ]. BMP-mediated inhibition of oligodendrogenesis is controlled through the repression of the former transcription factor olig2 known to be essential for the oligodendrocytic development [ 54 , 55 ]. All of these data suggest that BMP4 released by npMSCs selectively act on astrocytic and oligodendrocytic progenitors. In our experiments, we only demonstrate the anti-apoptotic effect of npMSCs-conditioned medium but we didn't test its possible instructive effect in clonal cultures. As BMP4, present in a biologically-active form in npMSCs conditioned medium, has been identified to be responsible of the increase of astrocytes numbers by the immunoblocking experiment and as it has been already demonstrated to be instructive [ 41 ], one could hypothesise that, in our system, the increase of astrocytes in response to npMSCs-derived BMP4 is a consequence of both a anti-apoptotic effect on GFAP-positive cells and also an instructive effect on NSCs. Conclusions When considering the use of MSCs in cell replacement strategies for the treatment of various neurological diseases, it should be taken into account that those cells could also influence the development host neural precursors. Methods Preparation and culture of rat mesenchymal stem cells (MSCs) Adult rat bone marrow was obtained from femoral and tibial bones by aspiration and was resuspended into 5 ml of DEM (Invitrogen, Merelbeke, Belgium) [ 56 ]. Between 100 and 200 × 10 6 marrow cells were plated on 175-cm 2 tissue culture flask in DEM/10% foetal bovine serum (Invitrogen). After 24 hours, the non-adherent cells were removed. When the rMSCs became confluent, they were resuspended with 0.25% Trypsin and 1 mM EDTA and then sub-cultured. Nestin expression by MSCs was induced as described in [ 32 ]. Preparation and culture of mouse striatal neural progenitor and stem cells Green C57BL/6 mice embryos (Jackson Immunoresearch Laboratory, Inc., West Grove, USA) or NMRI mice were used as a source of striatal neural progenitor and stem cells. Green mouse express GFP under control of the β-actin promoter [ 57 ]. The day of conception was determined by the presence of a vaginal plug (embryonic day 0). E16 striata were isolated and triturated in DEM/F12 (Invitrogen) with a sterile Pasteur pipette. The cell suspension was filtered with a 70 μm-mesh and viable cells were estimated by trypan blue exclusion. The cells were plated (1 × 10 6 cells/75-cm 2 tissue culture flask) in DEM/F12 supplemented with EGF (20 ng/ml, Sigma) and B27 (Invitrogen), a multi-component cell culture supplement devoid of any growth factor. When the size of neurospheres reached approximately 50 cells, they were dissociated into a single cell suspension by trituration and replated in fresh culture medium. Neurospheres with a maximum of 3 passages were used in this study. Co-culture of MSCs and neurospheres MSCs and neurospheres were plated on polyornithine coated dishes for 5 days, in DEM/F12 containing only B27 supplement and were then processed for immunocytochemical analysis as described below. Culture of neurospheres in MSCs conditioned medium MSCs were placed in DEM/F12 medium supplemented with B27 (Invitrogen), during 3 days. The conditioned media were then filtered with 0.22 μm-pore filter before being replaced on plated neurospheres during 5 days. Immunocytochemistry The cultures were fixed with 4% (v/v) paraformaldehyde for 15 minutes at room temperature and washed 3 times in TBS buffer. They were then permeabilized in 1% Triton-X100 (v/v) for 15 minutes and washed 3 times in TBS buffer. Non-specific binding was blocked by a 1 hour treatment in TBST (TBS buffer with 0.1% Tween) containing defatted milk powder (30 mg/ml). The cells were then incubated for 1 hour at room temperature with either anti-glial fibrillary acidic protein (GFAP, Dako, mouse IgG, dilution 1:500), or Tuj1 (Molecular Probes, mouse IgG, dilution 1:1000), or O4 (Chemicon, mouse IgM, dilution 1:100) primary antibodies (diluted in blocking buffer). After 3 washes in TBS, cells were then incubated in FITC- or Cy5-conjugated anti-mouse IgG or IgM (Jackson Immunoresearch, 1:500) for 1 hour at room temperature and in the dark. The nuclei were stained with ethidium homodimer (0.2 μM, Sigma). The preparations were then mounted in Fluoprep™ (Biomerieux; France) and observed using a Bio-Rad MRC1024 laser scanning confocal microscope. The fraction of positive cells was determined by counting 10 non-overlapping microscopic fields (±50 cells/field, ±3500 cells/experiments) for each coverslip in at least three separate experiments (n then corresponds to the number of coverslips). BrdU and propidium iodide incorporation After 24 hours or 3 days of culture, BrdU (20 μM, Sigma) which is a S-phase marker, or propidium iodide (400 mg/ml, Sigma) was added to the differentiating neurosphere cultures for 18 hours before fixation and staining. Tuj1, O4 and GFAP immuno-labellings were performed as described above. For BrdU labelling, coverslips were then post-fixed for 10 minutes in 4% (v:v) paraformaldehyde, incubated in HCl 1 N for 20 minutes at 37°C, washed with sodium perborate solution (50 mM, pH 8.5) and finally incubated with an anti-BrdU antibody for 1 hour at room temperature (Oxford, rat IgG, dilution 1:200) and Cy-5-conjugated anti-rat antibody, 1 hour at room temperature. The preparations were analysed as describe above. RNA extraction and quantitative RT-PCR analysis Total RNA was prepared using the RNeasy total RNA purification kit (Qiagen, Westburg). For cDNA synthesis, random hexamer primers (Invitrogen) were used to prime reverse transcriptase reactions. The cDNA synthesis was carried out using Moloney-murine leukemia virus (M-MLV) Superscript II Reverse transcriptase (Invitrogen) following the manufacturer's instructions. Quantitative PCR was carried out using standard protocols with Quantitec SYBR Green PCR Kit (Qiagen). The PCR mix contained SYBR Greeen Mix, 0.5 μM primers (Table 1 ), 1 ng DNA template and nuclease free water to final volume of 25 μl. PCR were performed on RotorGene RG-3000 (Corbett Research) and analyzed with Rotorgene Software (Corbett Research). The percentage of gene expression by npMSCs was normalized in function of GAPDH gene expression and compared to the gene expression by nestin-negative MSCs that was considered as 100%. Each gene analysis was realised on three different samples with two run/sample (we then provided 6 PCR analyses by target gene) Western Blot After 72 hours of culture, the conditioned medium (1.5 ml) from cultures of neural progenitor cells, npMSCs and nnMSCs were incubated with heparin sepharose CL-6B (Amersham Biosciences, Belgium) at 4°C for 24 hours. After centrifugation (700 × g), the supernatant was removed and bound proteins were eluted in loading buffer (glycerol 10% v/v; Tris 0.05 M pH 6.8; SDS 2%, bromophenol blue and 2.5% v/v – mercaptoethanol) by heating at 70°C for 10 minutes. The protein concentrations of various samples were quantified using the "RC DC Protein Assay" (Bio-Rad, Belgium) and equal protein quantities were loaded in each lane in a 15% sodium dodecyl sulfate polyacrylamide gel and electrophoresed. Then the proteins were transferred to PVDF membrane (Amersham Biosciences, Belgium) using a Trans-Blot Semi-Dry Transfer apparatus (Bio-Rad). The membranes were saturated with 3% gelatin (BioRad) during 1 hour at 37°C, then incubated for 1 hour with a monoclonal antibody against BMP4 (R&D, goat IgG, 0.1 μg/ml) or actin, as control for protein loading (Sigma, mouse IgG 1:5 000) at room temperature and then washed several times with PBS-0.1% Tween. The membrane was then incubated in Cy5-conjugated anti-goat IgG or anti-mouse IgG (Jackson Immunoresearch, 1:2500) for 1 hour at room temperature, in the dark. After several washes in PBS, the membranes were scanned using a Typhoon 9200 Scanner (Amersham Biosciences) and subsequent analyses were performed with ImageQuant Software (Amersham Biosciences). BMP4 neutralization Neurospheres were plated on polyornithine-coated dishes and incubated during 5 days in DEM/F12 medium supplemented with B27 (Invitrogen) and previously conditioned by npMSCs during 3 days or not. Anti-BMP4 antibody (R&D, 2 μg/ml) was added on day 1 of this incubation. Authors' contributions SW performed most of the work and wrote a first draft of the manuscript; FB performed quantitative RT-PCR; GH, GM and PL were involved in the writing of the manuscript; BR conceived the study and participated in its design and coordination. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC520747.xml |
506780 | World health system performance revisited: the impact of varying the relative importance of health system goals | Background In 2002, the World Health Organization published a health system performance ranking for 191 member countries. The ranking was based on five indicators, with fixed weights common to all countries. Methods We investigate the feasibility and desirability of using mathematical programming techniques that allow weights to vary across countries to reflect their varying circumstances and objectives. Results By global distributional measures, scores and ranks are found to be not very sensitive to changes in weights, although differences can be large for individual countries. Conclusions Building the flexibility of variable weights into calculation of the performance index is a useful way to respond to the debates and criticisms appearing since publication of the ranking. | Background The World Health Organization recently published a performance ranking of the health systems of its 191 member countries, and intends to update it at regular intervals [ 1 - 4 ]. It was based on a framework outlining a set of social goals to which health systems should contribute [ 5 ]. It was argued that systems should contribute to improving population health, be responsive to the people they serve and be financed fairly. Five outcome indicators were defined – the level of population health, inequalities in health, the level of responsiveness, inequalities in responsiveness and fairness in financial contributions. Estimates of attainment on these five indicators were made for the 191 countries that were members of WHO at that time, and a composite (overall) attainment indicator was constructed for each country as a weighted average of attainment on the five individual outcome indicators. Publication of the analytical framework and the resulting ranking provoked considerable comment, and a variety of issues concerning the methodology and country positions in the ranking have been raised. A central component of the methodology for measuring overall attainment was the use of fixed weights, common to all countries, to aggregate the five indicators. This feature has been controversial, with some arguing that people in different cultural and social settings value individual health system goals in different ways [ 6 - 13 ]. The fixed weights had the virtue of being based on expert opinion, having been derived from the valuations of 1,007 respondents – largely health system professionals – to a WHO survey [ 14 ]. However the weights were common to all countries, regardless of their development status and cultural traditions. In this paper we examine the sensitivity of the attainment scores to alternative weighting schemes that allow weights to vary across countries. These country-specific weights may reveal varying objectives of policy makers or constraints under which they operate. Melyn and Moesen [ 15 ] have referred to such weights as 'benefit of the doubt' weights. Methods WHO used fixed weights (0.25, 0.25, 0.125, 0.125, 0.25) to aggregate five health system outputs (respectively, the level of population health, inequality in the distribution of health, the level of health system responsiveness, inequality in the distribution of responsiveness, and fairness in financial contributions) into a scalar health system attainment index. Overall attainment ranged from 35.7 (Sierra Leone) to 93.4 (Japan) on a [0–100] scale. We propose here an analytical framework that reduces to the WHO fixed-weight methodology as a special case, but that allows varying degrees of freedom for weights to be defined that – in the sense of Melyn and Moesen – implicitly take into account individual country circumstances. For shorthand, we say countries "choose" such weights, which in reality are determined as solution values of a linear program. If the linear program is a fair representation of the objective function of and constraints faced by decision makers, the weights are indeed "chosen", but even if this condition is not necessarily met, the resulting weights may still be of interest. The extent of freedom to choose weights in a linear program can be set by the analyst. The analyst can enforce fixed weights common to all countries, allow countries complete freedom to choose their own weights, or adopt a middle ground in which countries are granted limited freedom to choose weights within bounds thought to be sensible by experts. A generic statement of the performance evaluation problem in primal-dual linear programming format is: In these programs y is a country's output vector, x is its input vector, Y is the sample output matrix and X is the sample input matrix. In the present context y is (5 × 1), Y is (5 × 191), x is (n × 1) and X is (n × 191), with n to be specified below. The primal program seeks the maximum radial expansion of a country's outputs, provided that it not exceed the standards established by a convex combination (λ ≥ 0, ∑λ = 1) of best-practice countries in the sample. The optimal value of φ provides a distance measure (i.e. an indication of how far a country has to go) to match best practice as observed in the sample. Since φ ≥ 1, the attainment of a country is evaluated as φ -1 ≤ 1. Best practice countries have φ -1 = 1, other countries have φ -1 < 1, so φ provides a basis for acomplete ranking of countries on their relative ability to deliver five health system outputs. The dual program seeks a set of nonnegative weights μ,υ attached to a country's outputs and inputs that maximize its attainment. Each country's output weights are normalized by μy = 1, but each is free to select its own set of nonnegative weights. In constructing the overall attainment index, WHO identified five output indicators and no inputs, preferring to treat each country's health system as a "health output management unit". Consequently each country's input vector is represented as a scalar with unit value. Under these circumstances the performance evaluation problem simplifies to: where ω = υ + υ 0 . The modified primal program seeks the maximum feasible radial expansion of a country's outputs consistent with best practice observed in the sample. The modified dual program seeks a set of nonnegative weights μ for a country's outputs that put it in the best light. A country can be expected to assign relatively large weights to those outputs at which it excels relative to best practice, and relatively low, possibly zero, weights to those outputs at which it lags behind best practice, subject to the normalization μy = 1. (See Annex for a graphical explanation.) By complementary slackness, μ m (Y m λ - φy m ) = 0, m = 1, ..., M, so slack in any element of a country's projected output vector φy implies that the country assigns a zero weight to that output. Since it is unreasonable to allow a country to assign a zero weight to any output deemed sufficiently important to have been included in the WHO performance evaluation exercise, it is desirable to restrict weights in some way. This can be accomplished most easily by appending constraints to the dual side of (2) of the form: γ m ≥ μ m y m /μy ≥ β m , m = 1, ..., M. (3) Restrictions (3) place lower and upper bounds on the relative importance of each output (as measured by μ m ) in total output. Implementation of the weight-restricted linear program requires specifying the 2M = 10 parameters γ m , β m . One procedure is to ignore the upper bounds γ m and set the lower bounds β m > 0. This eliminates the possibility of a country assigning zero weights to those outputs at which it lags behind best practice. A less arbitrary procedure is to follow Takamura and Tone [ 16 ] by adapting Saaty's Analytical Hierarchy Process (AHP) [ 17 ]. This procedure exploits expert judgment, that could be provided for example by the above-mentioned survey of health system professionals, to set lower bounds β m > 0 and upper bounds 1 > γ m . Although these bounds are common to all countries, they allow countries limited freedom to select weights appropriate to their circumstances. Results The fixed weights used by WHO)) [ 4 ] to aggregate the five health system indicators gave countries no freedom to choose weights appropriate to their circumstances. We compare the WHO attainment index with three alternative indexes allowing countries varying degrees of freedom to choose weights. The first index is based on the solution to program (2), without weight restrictions, thereby allowing complete freedom to choose. The second index is based on (2), with lower bounds in (3) of β m = 0.10 on all weights, allowing substantial freedom to choose. The third index is based on (2), with lower and upper bounds in (3) set by a modified AHP procedure. In the modified procedure, "expert opinion" was taken to mean the average values of weights arising from population-representative country surveys, each of which included a module on health system goals [ 9 ]. Respondents were queried about their individual preferences on the five stated health system goals in a total of 51 countries, in some of which multiple surveys were performed, and country means were calculated on the basis of these individual responses [ 18 ]. The survey methods, reliability, validity, representativeness, sample size and respondent characteristics are extensively described in Ustün et al. [ 9 ], and are also reported in summary form in Sadana et al. [ 19 ], Mathers et al. [ 20 ], Mathers et al. [ 21 ] and Sadana et al. [ 22 ]. The survey instruments are available at . For each output, the lower bound for calculation of the third index was taken as the minimum of the country average weights, and the upper bound the maximum [ 23 ]. Country mean weights and survey types, as well as survey wide maximum and minimum weights are shown in Table 1 . The lower bounds are accordingly β m = (0.19, 0.17, 0.12, 0.11, 0.22) and the upper bounds γ m = (0.29, 0.25, 0.18, 0.17, 0.30). This specification allows limited freedom to choose. We refer to the four indexes as WHO, LP1, LP2 and LP3, respectively. Summary statistics of the four attainment indexes appear in Table 2 . The three LP distributions have higher means than the WHO distribution, and two of them have lower dispersion. However increasing restrictions on freedom to choose reduce the mean, and increase the dispersion, of the LP attainment indexes toward the mean and dispersion of the WHO attainment index. Rank correlations between pairs of attainment rankings appear in Table 3 . Despite the distributional changes due to freedom to choose, rank correlations are positive, high and statistically significant. The lowest correlations involve LP1, the index allowing complete freedom to choose. With the exception of LP1, there is strong agreement about the identity of countries in the top and bottom quartiles of the distribution. Japan is ranked #1 and Sierra Leone is ranked #191 on all four indexes. Figure 1 shows plots of WHO attainment scores and rankings versus the three LP attainment scores and rankings. The convergence of the distributions of the LP attainment scores and rankings to the WHO scores and rankings is apparent. Results based on LP1 are unattractive. Over one-third of countries (70 of 191) assign a zero weight to four of the five indicators, and the vast majority of countries assign a weight in excess of 0.9 to either responsiveness distribution or fairness in financial contributions. This means they assign a low or zero weight to population health, the defining goal of the health system, which does not have face validity. In LP1, only Japan, Luxembourg and the United States assign positive weights to all five indicators. Consequently the attainment index is tightly distributed about a very high mean value. The ability to discriminate is sacrificed to freedom to choose, with 75% of countries receiving attainment indexes of 99 or above. Results from LP2 are somewhat more attractive. Nevertheless, when weights are bounded below by 0.1, over three-quarters (147 of 191) of countries assign the minimum weight to four of the five indicators and a 0.6 = 1 - (4 × 0.1) weight to either responsiveness distribution or fairness in financial contributions. The attainment scores are again compressed about a high mean value, and the ability to discriminate is not much improved, with 75% of countries receiving attainment scores of 93 or above. However despite this dramatic compression, the LP2 ranking is globally very similar to the WHO ranking. Eighteen of the countries ranked in the top 20 by WHO appear in the LP2 top 20, and 14 of the countries ranked in the bottom 20 by WHO appear in the LP2 bottom 20. Not surprisingly, the distribution of the attainment scores for LP3 looks even more similar to the distribution of the WHO scores, and has a similar mean and standard deviation. Rank correlation is very high, with only one country falling out of the WHO top 20 and only five countries rising out of the WHO bottom 20. Nonetheless, limited freedom to choose has an important impact on individual countries. The USA, given an ability to emphasize the importance of responsiveness level and responsiveness inequality, and to de-emphasize the importance of health level and inequality in the distribution of health, improves its ranking from #15 to #3. Australia improves from #12 to #7 for similar reasons. Italy, on the other hand, falls from #11 to #18, primarily as a result of the movement of other countries. In LP3, the largest positive changes in ranking are for Mauritius (+25) and Saint Vincent (+22), while the largest negative changes in ranking are for Kazakhstan (-39) and Albania (-36). Discussion We began by questioning the appropriateness of the fixed weight approach to aggregating indicators adopted by WHO, on the grounds that fixed weights deny countries at varying stages of development the freedom to choose. We then proposed a sequence of linear programming models that allowed countries varying freedom to choose the weights assigned to their indicators. LP1 allows complete freedom to choose, and generates weights we consider unacceptable, particularly because so many countries give a zero weight to improving health. LP2 allows considerable freedom to choose, but generates many country-specific weights falling outside the range of the within-country means used as bounds in LP3. Clearly, the validity of a procedure that routinely assigns weights out of the range of representative cross-population preferences should be questioned, even without a sophisticated theory of empirical ethics or democratic choice. LP3 applies the AHP procedure to set plausible bounds on weights, and allows limited freedom to choose. It generates a very similar distribution of the attainment index, and a very high linear rank correlation with the WHO ranking. Despite these similarities, we find the LP3 approach intuitively appealing, and are encouraged by its global concordance with the WHO index. However it is fair to ask: if LP3 and WHO generate such similar rankings, why bother? What value does LP3 add? Indeed, individual countries may come to diametrically opposed conclusions about the benefits of LP3 or WHO on the basis of their implied rank changes (e.g. Mauritius vs. Kazakhstan). Howbeit, our first response to the question "why bother" focuses on the distribution of the LP3 weights in comparison to the WHO weights. The WHO weight on responsiveness inequality was 0.125. But the LP3 upper bound of 0.17 is binding on 182 countries, which implicitly want a higher weight on this indicator. The WHO weight on fairness in financial contributions was 0.25. But the LP3 upper bound of 0.30 is binding on 167 countries that want a higher weight. At the other end, the WHO weight on health level is 0.25. But the LP3 lower bound of 0.19 is binding on 170 countries that want a lower weight. The WHO weight on responsiveness level is 0.125. But the LP3 lower bound of 0.12 is binding on 103 countries that want a lower weight. It appears that a majority of countries at all stages of development may implicitly assign greater importance to indicators of health distribution, and less importance to indicators of health level, than the experts whose judgments formed the basis of the original WHO weights. Our second response is more general. By allowing countries limited freedom to choose their weights, LP3 takes a small but nevertheless significant step toward respecting their varying circumstances. While the narrower the bounds on weights, the smaller the step, even the limited freedom embodied in LP3 makes an important difference to some countries. Conclusions Building in the flexibility of varying weights might be a useful way for WHO to respond to the debates and criticisms appearing since publication of the ranking. We conclude by speculating that a variant of LP3 incorporating information regarding which weights are binding, and in which direction, might yield even greater benefits in terms of respecting individual circumstances. Competing interests JAL, CJLM and DBE were part of the team at WHO that developed the methods for the world health system performance rankings published by the Organization in 2000. Abbreviations AHP (Analytical Hierarchy Process) DEA (Data envelopment analysis) LP (Linear program) LP1 (Linear Program 1) LP2 (Linear Program 2) LP3 (Linear Program 3) WHO (World Health Organization) Authors' contributions JAL developed methods inspired by the benefit-of-the-doubt concept and drafted an earlier version of the manuscript. CAKL adapted and applied the DEA methods described here to health system performance and drafted the initial version of the current manuscript. CJLM conceived the methods for measuring health system performance. DBE coordinated the research at WHO. All authors revised and approved the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: Supplementary Material Additional File 1 Annex.Graphical interpretation of DEA and conceptual links with value theory.doc Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC506780.xml |
516020 | Comparison of bovine leukemia virus (BLV) and CMV promoter-driven reporter gene expression in BLV-infected and non-infected cells | Background Viral promoters are used in mammalian expression vectors because they generally have strong activity in a wide variety of cells of differing tissues and species. Methods The utility of the BLV LTR/promoter (BLVp) for use in mammalian expression vectors was investigated through direct comparison to the CMV promoter (CMVp). Promoter activity was measured using luciferase assays of cell lines from different tissues and species stably transduced with BLVp or CMVp driven luciferase vectors including D17, FLK, BL3.1 and primary bovine B cells. Cells were also modified through the addition of BLV Tax expression vectors and/or BLV infection as well as treatment with trichostatin A (TSA). Results Results indicate the BLV promoter, while having low basal activity compared to the CMV promoter, can be induced to high-levels of activity similar to the CMV promoter in all cells tested. Tax or BLV infection specifically enhanced BLVp activity with no effect on CMVp activity. In contrast, the non-specific activator, TSA, enhanced both BLVp and CMVp activity. Conclusion Based on these data, we conclude the BLV promoter could be very useful for transgene expression in mammalian expression vectors. | Background Viral promoters are commonly used as regulatory elements in gene therapy vectors due to their strong activity in various cell lines in vitro. Probably the most widely used promoter in mammalian expression systems is the human cytomegalovirus immediate-early gene (CMV) promoter. The CMV promoter induces high-level constitutive expression in a variety of mammalian cell lines [ 1 ]. In many gene therapy applications, however, an inducible or cell specific promoter would be more appropriate. A regulated transgene expression system in mammalian cells is preferable for effective and safe gene therapy and for the study of gene function in cell biology. The most important features of an inducible promoter would be 1) low basal expression levels; 2) high induced expression; and 3) inducer-specific, modulated expression [ 2 ]. Our need for a mammalian expression vector promoter for preventative gene therapy that would be induced by bovine leukemia virus (BLV) infection or BLV Tax protein expression led us to investigate the use of the BLV promoter for gene therapy. The U3 region of the BLV promoter, located in the 5' long terminal repeat (LTR), contains several important cis -acting elements in addition to the CAAT box, TATA box, and transcription start site [ 3 , 4 ]. The major regulatory elements are three copies of an imperfectly conserved 21-bp sequence called the tax responsive element (TxRE). The TxREs are essential for the promoter's responsiveness to the Tax transactivator protein encoded by the 3' end of the proviral genome [ 5 ]. These cis -elements contain motifs resembling the cyclic AMP-responsive element (CRE) as well as an E box sequence [ 6 ]. Tax does not bind directly to the TxRE but interacts with cellular proteins that recognize the CRE including the transcription factors CREB, ATF-1, and ATF-2 [ 7 - 9 ]. The transcription factor AP4 can potentially bind to the E box sequence and is important in Tax activation [ 10 ]. There is a glucocorticoid responsive element (GRE) that responds to dexamethasone in the presence of glucocorticoid receptors and Tax [ 11 ]. A nuclear factor κB (NFκB) binding site responds to phorbol 12-myristate 13-acetate (PMA) treatment [ 12 ]. Finally, there is a Tax transactivator independent site specific for the B cell transcription factors PU.1 and Spi-B [ 13 ]. This PU.1/Spi-B binding site may be involved in the B lymphocyte tropism of BLV. We hypothesized that the BLV promoter could be used in mammalian expression vectors for regulated high-level gene expression. Our approach was to compare reporter gene expression driven by either the BLV promoter or CMV promoter in different cell types with or without BLV infection or Tax induction. Our results demonstrate that the BLV promoter can be induced to express the reporter gene to levels as great as the constitutive CMV promoter. Methods Cell culture All cells used in these studies were maintained in RPMI 1640 medium (Invitrogen) supplemented with 10% fetal calf serum (FCS), 4.5% dextrose, 1 mM sodium pyruvate, and antibiotic-antimycotic solution (100 μg/ml penicillin G sodium, 100 μg/ml streptomycin sulfate, 0.25 μg/ml amphotericin B). In addition, the following concentrations of drugs were added for each selective media: Blasticidin-S (Invivogen) 10 μg/ml; G418-sulfate (Alexis Biochemical) 400 μg/ml. The following cell lines were used: D17 [dog osteosarcoma; ATCC CCL-183; [ 14 ]], FLK [sheep kidney; BLV expresser; [ 15 ]], BL3.1 [bovine B-lymphosarcoma; BLV expresser; ATCC CRL-2306; [ 16 ]]. Primary bovine B cells were supplemented with 10 ng/ml each of recombinant human interleukin-4 and interleukin-7 (Peprotech, Inc.), and gamma-irradiated (4,000 R) murine CD40L-expressing L cells (J558L; a gift from Philip Griebel) as described elsewhere [ 17 ]. For the TSA experiments, Trichostatin A (Sigma) was supplemented at 500 nM for 48 h. Cells were cultured at 37°C in a 5% CO2 humidified atmosphere. Viable cells were identified by trypan blue dye exclusion, and cell number was counted with a hemacytometer. Primary bovine B cells were purified as follows. Peripheral blood mononuclear cells (PBMC) were isolated from heparinized cow blood through a ficoll density gradient as previously described [ 18 ]. B cells were separated from the PBMCs using the MiniMACS system following the manufacturer's (Miltenyi Biotec) protocol. Briefly, 1 × 10 7 cells were stained for 15 min at 6° – 12°C with 10 μg/100 μl total volume anti-IgM (PIG45A; VMRD, Inc.). After washing, 20 μl/100 μl total volume of MACS rat anti-mouse IgG2a+b microbeads were mixed with the cells and incubated for 15 min at 6° – 12°C. Cells were thoroughly washed, and magnetically separated. These IgM+ cells were considered primary B cells. Microfluorimetry using anti-IgM (PIG45A; VMRD, Inc.) indicated 90% purity. Stably transduced cell lines were generated after one week in selective media. Primary B cells were analyzed after one week in selective media since they began to die out after two weeks in culture. Vector construction The plasmid pBLV913 (a gift from David Derse), coding for an infectious molecular clone of BLV [ 5 ] was used as the source for the BLV promoter and BLV Tax sequences. Briefly, the BLV promoter from the U3 region of 3' LTR of BLV was isolated from plasmid pBLV913 (Derse) as a 345 bp fragment (GenBank LOCUS BLVCG, ACCESSION K02120 bp 8096 – 8440) and cloned in place of the CMV promoter fragment into pLNCX (Clontech; Genebank LOCUS SYNMMLPLN3 ACCESSION M28247 – CMV promoter removed as Bam HI- Hind III fragment) to create the vector pLNBlv. The pLNBlv and pLNCX retrovector plasmids were modified to place the Gateway Rfa cassette (1.7 kb; Invitrogen) downstream of the internal promoters (BLV or CMV) in order to simplify further cloning, to create retrovector plasmids pLNBlv-G or pLNC-G. For enhanced protein expression, the WPRE element (from plasmid BluescriptII SK+ WPRE-B11 (a gift from Tom Hope–the same as bp 2717–3309 of Genbank Locus OHVHEPBA ACCESSION J04514) was cloned downstream of the Gateway Rfa cassette with standard cloning methods to create vectors pLNC-GW and pLNBlv-GW. The source for firefly luciferase encoding sequence was pGEM-luc (Promega). The luciferase coding sequence was subcloned into pENTR1A (Invitrogen) to engineer the Gateway entry vector pENTR1A/luc. The Luciferase gene was recombined into pLNC-GW or pLNBlv-GW using LR Clonase (Invitrogen) per manufacturer's instructions. The promoter-less luciferase expression control vector pLN[]W/luc was engineered by removing the BLV promoter ( Bam HI digest) from pLNBlv/luc. BLV Tax (Genbank Locus AAF97920) was isolated by reverse transcription PCR from FLK cells and subcloned into pENTR1A (Invitrogen) to engineer the Gateway entry vector pENTR1A/Tax. The Tax gene was recombined into pLBC-GW where the neomycin resistance gene of pLNC-GW was replaced with the blasticidin resistance gene. Throughout these studies we assayed expression vectors with and without the WPRE. WPRE enhanced transgene expression in all cell lines used, and in a promoter-independent fashion (about 2-fold greater for BLVp and CMVp in D17 cells). Subsequently, all data shown in this report are only with vectors containing WPRE. Cell transfection and transduction Retrovirus-mediated gene transfer was accomplished using the BD Retro-X System (BD Biosciences Clontech) following the manufacturer's suggested protocol. Briefly, 100 mm × 20 mm tissue culture dishes (Falcon) were seeded with the packaging cell line GP2-293 at 70–90% confluency. Each dish of GP2-293 cells was co-transfected with 5 μg each of retroviral vector and the envelope glycoprotein expression vector pVSV-G using 15 μl/transfection of Lipofectamine 2000 (Invitrogen) cationic lipid reagent for 3 h in a total volume of 5 ml medium/dish. Subsequently, transfection medium was replaced with 10 ml growth medium, and the cells were incubated for 72 h. Retrovirus-containing supernatant was then harvested and passed through a 0.45 μm cellulose acetate filter, then concentrated by ultracentrifugation at 50,000 × g for 30 min at 4°C. Supernatant was carefully poured off and virus was resuspended in the residue (~200 μl) and frozen (-70°C) for future use. Cells for transduction were plated on 6-well tissue culture plates (Falcon) at 50% confluency. Concentrated retrovirus (titer unknown) along with polybrene (8 μg/ml) were added to one ml/well cells (in a 6-well plate) and incubated overnight. Transduction medium was replaced with fresh growth medium, and the following day cells were split into appropriate selective medium. BLV was harvested from supernatant of FLK cells, concentrated, and used to transduce cells in a similar fashion. Luciferase assay Luciferase assays (Promega) were performed using a single-tube luminometer (Pharmingen) to measure relative light units (RLU) on a linear scale. Cells to be assayed were counted using a hemacytometer, and 1 × 10 6 cells were aliquoted to 1.5 ml microcentrifuge tubes. Then, cells were pelleted at 300 × g for 10 min, washed once with PBS, and lysed with 200 μl reporter lysis buffer (Promega). Lysate was stored at -20°C until assayed. Lysate was thawed and pelleted (300 × g for 10 min), and luciferase was measured with the luminometer using 10 μl lysate/50 μl reagent for 10 s. Linear range was under 1 × 10 7 RLU. Statistical analysis Student's t-test was performed for statistical evaluation of the results. Results are expressed as the arithmetic mean with the variance of the mean (mean ± SE). Results The BLV promoter was engineered to drive reporter genes Our studies utilized a commercially available retroviral system with its standard CMV promoter (CMVp) or replaced with the BLV promoter (BLVp). Figure 1 shows a schematic of the BLV promoter used in these studies with its unique regulatory elements. The luciferase reporter gene was used to compare promoter expression strength within different cell lines and treatments. The Woodchuck Hepatitis Virus posttranscriptional regulatory element (WPRE) was also incorporated to enhance transgene expression within these retroviral vectors [ 19 , 20 ]. WPRE has been reported to significantly stimulate expression of transgenes in a promoter-independent fashion [ 19 ]. Retroviral vectors were used because of the ease of stable cell line establishment, and because of their prominent use in transgenics and gene therapy. The commercially available retroviral vector used in these studies contained the CMV IE promoter for transgene expression. We modified this retrovector for comparison studies replacing the CMV IE promoter with the BLV promoter (see methods). Cells of several different tissues and species were used in our studies. Figure 1 Schematic representation of BLV promoter used in comparison studies. The BLV promoter (BLVp) consisting of the U3 region of the 5'LTR of BLV includes the basic elements of transcription start site (+1), CAAT (nt -97/-92) and TATA (nt -43/-37) boxes as shown. Unique to the BLVp are the three imperfectly conserved 21 bp sequences known as the Tax Responsive Elements (TxRE). The numbers following the TxRE designation represent its position relative to the transcription start site. Each TxRE contains a consensus E box-binding motif overlapping an imperfect cyclic AMP responsive element motif (CRE/Ebox). Additionally, the BLVp contains a glucocorticoid responsive element (GRE), Nuclear Factor Kappa Binding motif (NFkB), and B cell specific PU.1 or Spi-B transactivator binding motif (PU.1/Spi-B). The transcription elements are not drawn to scale. The BLV promoter can be as strong as the CMV promoter depending on the host cell In contrast to the constitutive expression of the CMV promoter, the BLV promoter has cis elements that are dependent on BLV Tax for transgene expression [ 5 , 21 , 22 ]. We hypothesized therefore that in a cell line such as D17, the BLV promoter would have little or no activity compared to the CMV promoter. Conversely, in a cell line expressing the BLV Tax transactivator such as the BLV-producing FLK cell line, the BLV promoter would have similar activity compared to the CMV promoter. We tested this assumption with luciferase as the transgene and found indeed, BLV promoter activity was about 50-fold less than CMV promoter activity in D17 cells but was about equal in FLK cells (Fig. 2 ). As shown in Figure 1 , the BLVp also has a cis element that is B cell specific (PU.1/Spi-B). We therefore compared the strengths of BLV and CMV promoters in primary B cells and a BLV infected B cell line hypothesizing that BLVp expression would be comparable to CMVp activity. BLVp activity was still less than CMVp activity in primary B cells but by only about a 5-fold difference (Fig. 2 ). In the BLV infected BL3.1 cell line, BLVp activity was approximately equal to CMVp activity, analogous to results using the BLV infected FLK cell line. Thus, the BLV promoter can be as strong as the CMV promoter within a cell line under specific conditions e.g. BLV infection/Tax expression. Figure 2 BLVp and CMVp activity comparison in D17, FLK, primary cow B cells, and BL3.1. Relative light units (RLU) of luciferase activity driven by either the BLV promoter (BLVp) or CMV promoter (CMVp) of 1 × 10 6 stably transduced cells was measured during a 10 s period. Bars represent the arithmetic mean and variance of 10 experiments. *P < 0.05; **P < 0.001 determined by t-test. BLV infection enhances BLV promoter expression but has no effect on the CMV promoter Since BLV promoter activity was greater than CMV promoter activity in the BLV infected FLK cell line but minimal compared to CMV promoter activity in the non-BLV infected D17 cell line, we set out to determine whether BLV infection of D17 cells would enhance BLVp and/or suppress CMVp expression. The dog derived D17 cell line can be infected with BLV albeit not very efficiently [ 14 ]. D17 cells were infected with concentrated BLV from FLK cells, then clonally selected for BLV expression using pol RT-PCR and BLV reverse transcriptase assay of the supernatant (data not shown). Luciferase assays demonstrated that BLV promoter activity in infected D17 cells was about 10-fold greater than BLV promoter activity in non-infected D17 cells (Fig. 3 ). CMV promoter activity remained unchanged. Figure 3 BLV infection enhances BLVp activity but has no effect on CMVp activity. D17 cells or D17 cells infected with and productively expressing BLV (D17+BLV) were transduced with luciferase expression vectors. Relative light units (RLU) of luciferase activity driven by either the BLV promoter (BLVp) or CMV promoter (CMVp) of 1 × 10 6 stably trasduced cells was measured during a 10 s period. Bars represent the arithmetic mean and variance of 10 experiments. **P < 0.001 determined by t-test. BLV Tax enhances BLV promoter expression but has no effect on the CMV promoter To assess directly the effect of constitutive Tax expression on the BLV promoter and CMV promoter, BLV Tax was provided as a transgene to cells. As expected, Tax significantly enhanced BLV promoter activity but had no effect on CMV promoter activity (Fig. 4 ) inducing BLVp activity about 48-fold in D17 cells and 4-fold in primary B cells. Interestingly, we found that when BLV infected cells were transduced with the Tax transgene, the resulting increase in BLV promoter activity was a greater-than-additive enhancement of BLV infection and Tax transgene (Table 1 ). This effect could likely be caused by Tax expressed from the trangene upregulating expression of the entire BLV provirus, including Tax. The effect on the CMV promoter was not significant. Further, BLVp activity was enhanced in cell lines FLK (2-fold) and BL3.1 (4-fold) actively producing high-levels of BLV (Fig. 5 ). Figure 4 BLV Tax expression significantly enhances BLVp activity but has no effect on CMVp activity. D17 cells and primary bovine B cells (D17; B cells), or D17 cells and primary bovine B cells stably transduced with a BLV Tax expression vector (D17+TAX; B cells+TAX), were assayed. Relative light units (RLU) of luciferase activity driven by either the BLV promoter (BLVp) or CMV promoter (CMVp) of 1 × 10 6 stably transduced cells were measured during a 10 s period. Bars represent the arithmetic mean and variance of 10 experiments. **P < 0.001 determined by t-test. Table 1 Percent of Basal Luciferase Expression Promoter D17+Tax D17+BLV D17+Tax+BLV BLVp 115 ± 7 1226 ± 15 2038 ± 202 CMVp 96 ± 5 130 ± 23 118 ± 14 Figure 5 Tax trans-gene expression significantly enhances BLVp activity in cells producing high levels of BLV. FLK and BL3.1 cells, or FLK and BL3.1 cells stably transduced with a BLV Tax expression vector (FLK+TAX; BL3.1+TAX) were assayed. Relative light units (RLU) of luciferase activity driven by either the BLV promoter (BLVp) or CMV promoter (CMVp) of 1 × 10 6 stably transduced cells were measured during a 10 s period. Bars represent the arithmetic mean and variance of 10 experiments. *P < 0.05; **P < 0.001 determined by t-test. Trichostatin A non-specifically enhances BLV promoter and CMV promoter Activity The deacetylase inhibitor trichostatin A (TSA) has been shown to be the most efficient activator of BLV expression known to date [ 23 ]. To determine whether increased promoter activity due to TSA was a generalized attribute applicable to the CMVp as well, D17, FLK, and BL3.1 cells possessing BLVp or CMVp driven luciferase expression were treated with TSA. However, since CMVp activity was already 50-fold greater than BLVp activity in D17 cells, comparison of TSA induced promoter activities in a cell line where BLV and CMV promoter activities were similar would permit a more effective evaluation of TSA on the two promoters. Further, TSA induced much less death within the 48 h assay period in BL3.1 cells compared to other cell lines tested (< 10%). Using BL3.1 with counts adjusted for live cells, TSA treatment enhanced both BLVp and CMVp activity by about 40-fold (Fig. 6 ) indicating TSA was a non-specific promoter enhancer. Figure 6 Trichostatin A (TSA) enhances BLVp and CMVp activity. Relative light units (RLU) of luciferase activity driven by either the BLV promoter (BLVp) or CMV promoter (CMVp) of 1 × 10 6 stably trasduced cells was measured during a 10 s period. BL3.1 cells were either non-treated or treated with 500 nM TSA for 48 h. Bars represent the arithmetic mean and variance of 10 experiments. **P < 0.001 determined by t-test. Discussion Viral promoters are used in mammalian expression vectors because they can have strong activity in a wide variety of cells of differing tissues and species. Probably the most employed is the CMV promoter because of its proven high-level constitutive expression in a variety of mammalian cell lines [ 24 , 25 ]. While constitutive transgene expression is suitable for certain research or gene therapy applications, a strong regulated transgene expression is preferable in many other applications [ 26 ]. The BLV promoter, consisting of the U3 region of the LTR, is highly dependant upon Tax for activation and transgene expression. In this study, we set out to determine the strength of BLV promoter activity compared to the strength of the CMV promoter to ascertain the utility of the BLV promoter for mammalian expression vectors. Information on the BLV promoter describing the cis -acting elements and the dependence upon Tax using reporter vectors in mammalian cell lines has been published [ 13 , 27 , 28 ]. However, a direct comparison of promoter strength of the BLV promoter and the standard of mammalian expression vectors, the CMV promoter, has not been performed. Several attributes are important in developing a mammalian expression vector. Probably the most important attribute of a mammalian expression vector promoter is its ability to accomplish high-level transcriptional activity in a large variety of cell types of different tissues and species. Our studies showed that the BLV promoter could achieve similar high-level activity to the CMV promoter in cells expressing BLV Tax or infected with BLV. This comparatively high BLV promoter activity was demonstrated in D17 cells which we have found to be the highest expresser of CMV promoter driven transgenes of all cell lines tested in our laboratory. The CMV promoter activity was still about 5-fold greater than BLV promoter activity in the BLV infected D17 cells compared to the relatively equal activity of the CMV promoter versus BLV promoter in BLV infected FLK cells. However, FLK cells contain four copies of the BLV provirus [ 29 ] whereas BLV infected D17 cells contain a single copy of the provirus (data not shown). Thus there may be relatively greater expression of Tax in FLK cells effecting greater activity of the BLV promoter. Quantitative levels of Tax in BLV infected D17 or FLK cells were not measured. In this study, we showed relative to the CMV promoter high levels of induced BLV promoter activity in cell lines of canine osteosarcoma (D17), fetal lamb kidney (FLK), bovine B-lymphosarcoma (BL3.1), and bovine primary B cell origin. We also have data (not shown) demonstrating high BLV promoter driven transcriptional levels in cell lines derived from bat lung (TB1Lu), monkey kidney (Vero), and human kidney (HEK-293). Other researchers have also shown high BLV promoter activity using reporter gene assays in cell lines of various tissues derived from cow, dog, cat, mouse, human, monkey, sheep, and hamster [ 5 , 6 , 13 , 22 , 23 , 27 , 28 ]. Clearly the BLV promoter possesses the significant trait of high-induced expression in a wide variety of cell types. A second important attribute of an inducible promoter apart from high-induced expression is low basal expression. Researchers have reported barely detectable BLV promoter Tax-independent activity through luciferase assays of COS-1, C8, and KU-1 cell transient transfections [ 28 ]. Our results using reporter vector stable D17 cell lines showed low but definite BLV promoter basal activity. Others measuring BLV promoter-driven luciferase activity in transiently transfected D17 cells reported an above background activity of the BLV promoter, but the basal activity seemed much closer to background than we report here [ 6 , 23 ]. The difference could be due to vectors employed (the commercial retrovector we used had weak promoter activity from the 5'LTR (data not shown) and our vectors contained the WPRE), or that we used stable cell lines versus transient transfections. Researchers using B cell lines (Raji, Daudi, DG75, A20) also showed low, but definite BLV promoter activity in transient and stable transfected cells similar to our results using primary B cells [ 6 , 13 , 27 ]. Nevertheless, in all of these studies Tax addition was able to induce expression ranging from 50 to 800-fold over basal expression. Our data showed Tax enhanced BLV promoter activity to levels comparable to the CMV promoter. A low but significant BLV promoter Tax-independent activity is not surprising considering the E boxes, CRE, GRE, NFkB and PU.1/Spi-B binding sites are available for cellular transactivating factors (Fig. 1 ). In fact, mutation studies of these cis -elements have demonstrated significant decreases in basal level activity, as with the mutation of the GRE site [ 30 ], significant increases in basal level activity, as with the mutation of the CRE sites [ 6 ], or either decrease or increase in basal activity, depending on the cell line assayed, as with mutations of the E box [ 30 ]. Still, compared to CMV promoter activity, or Tax-induced activity, BLV promoter basal activity is very low. A third important attribute for an inducible promoter would be a sensitive modulated response to a specific inducer. Enhancement of the BLV promoter can occur independent of Tax by the addition of activating agents. Phorbol esters, phytohemaglutinin, and lipopolysaccharides have all been shown to enhance BLV promoter expression [ 31 ]. However, all of these agents are non-specific activators and upregulate many promoters within the cell [ 32 ]. The most efficient activator of BLV expression is the deacetylase inhibitor, trichostatin A (TSA). Addition of TSA to D17 cells enhanced luciferase expression driven by the BLV promoter 11-fold over basal expression [ 23 ]. In BL3.1 cells, less variability occurred from TSA induced cell death and basal BLVp and CMVp activity was relatively the same. TSA upregulated activity of both BLV and CMV promoters within BL3.1 by about 40-fold. In contrast, the BLV promoter was specific to Tax activation, while CMV promoter expression was not affected by Tax. For example in D17 cells, Tax specifically increased BLVp activity 48-fold. Nevertheless, the transactivating properties of BLV Tax are not limited to activation of the BLV promoter. Tax has been shown to upregulate Bcl-2 and increase nuclear NFkB activity [ 17 ]. Tax expression induces immortalization of primary rat embryo fibroblasts and causes cytokine-independent B cell growth [ 17 , 33 ]. These "side effects" of Tax may deter the use of BLV promoter for mammalian expression vectors. However, studies have demonstrated that the BLV promoter transactivation and immortalization activities of wild-type Tax can be dissociated by mutations within specific regions of the protein [ 9 ]. In fact, phosphorylation of Tax serines 106 and 293 are required for in vitro cell transformation but not BLV LTR transactivation [ 34 ]. Tax transcriptional activity requires an amino-terminal zinc finger and an internal leucine-rich activation domain [ 9 ]. Phosphorylation-deficient Tax mutants have been developed [ 33 ] and could be used in place of wild-type Tax for BLV promoter transactivation. Other mutations of Tax were shown to enhance BLV promoter activity in 293T cells by 10-fold over wild-type Tax [ 22 ]. However this mutant also transactivated the cellular proto-oncogene c- fos . Clearly, there is great potential to magnify the desirable traits of the BLV promoter/Tax system for mammalian expression vectors and minimize undesirable traits. Conclusions To determine whether the BLV promoter could be a useful mammalian expression vector element, we compared its activity with the CMV immediate early promoter in dog osteosarcoma (D17), BLV-infected fetal lamb kidney (FLK), BLV-infected bovine B-lymphosarcoma (BL3.1), and primary bovine B-cells. Without concomitant Tax expression from a transgene or BLV infection, the BLV promoter activity was low compared to CMV promoter activity. In the presence of Tax or BLV expression, the BLV promoter activity became equally as active as the CMV promoter. The CMV promoter was not influenced by Tax or BLV. Tax overexpressed as a transgene in BLV infected cells resulted in BLV promoter expression greater than CMV promoter expression. The deacetylase inhibitor, trichostatin A was a potent upregulator of both BLV and CMV promoters. Our results indicate the BLV promoter has great potential use as an inducible promoter for mammalian expression vectors. Competing interests None declared. Authors' contributions JSH carried out cell culture work including transfection/transduction and luciferase assays, data preparation and analyses, and drafted the manuscript. KAE performed genetic engineering of vectors. LSK did preliminary work to establish study concepts. RDB and GAS participated in the design and coordination of the study. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516020.xml |
548148 | Genome-scale analysis of positional clustering of mouse testis-specific genes | Background Genes are not randomly distributed on a chromosome as they were thought even after removal of tandem repeats. The positional clustering of co-expressed genes is known in prokaryotes and recently reported in several eukaryotic organisms such as Caenorhabditis elegans , Drosophila melanogaster , and Homo sapiens . In order to further investigate the mode of tissue-specific gene clustering in higher eukaryotes, we have performed a genome-scale analysis of positional clustering of the mouse testis-specific genes. Results Our computational analysis shows that a large proportion of testis-specific genes are clustered in groups of 2 to 5 genes in the mouse genome. The number of clusters is much higher than expected by chance even after removal of tandem repeats. Conclusion Our result suggests that testis-specific genes tend to cluster on the mouse chromosomes. This provides another piece of evidence for the hypothesis that clusters of tissue-specific genes do exist. | Background Positional clustering of co-expressed genes is mainly due to operons in prokaryotes. Genes located within the same operon are transcribed together and thus co-regulated. In general, operons are missing from eukaryotes. Eukaryotic genes appear to be transcribed individually and are thought to be scattered on chromosomes without apparent organization by function or positional expression except for tandem duplication. However, positional clustering has recently been reported in several eukaryotes such as Saccharomyces cerevisiae [ 1 , 23 ], Drosophila melanogaster [ 4 , 5 ], Caenorhabditis elegans [ 2 , 3 , 7 ], and Homo sapiens [ 6 , 8 ]. Cho et al . first showed clustering of co-expressed yeast genes on a genome-scale. They found that about 25% of genes with cell-cycle-dependent expression pattern were adjacent to those induced in the same phase of the cell cycle [ 23 ]. This fact was also reported by Cohen et al . [ 1 ]. Cohen et al . computationally analyzed the whole-genome gene expression using the same dataset. They consider two ORFs to be adjacent if they occur on the same chromosome and if there are no other ORFs or functional elements such as Ty element, long terminal repeats, tRNAs, snRNAs between them. They observed that adjacent or nearby non-adjacent pairs of genes showed correlated expression independent of their orientation in the yeast genome using chromosome correlation maps that display correlations between the expression patterns of genes on the same chromosome. In addition, they also showed that genes with similar functions tend to occur in adjacent positions along the chromosomes. Spellman et al . [ 4 ] found that numerous clusters that span 10 to 30 physically adjacent genes share strikingly similar expression profile in Drosophila melanogaster and these clustered genes account for over 20% of the total assayed genes. By mapping Expressed Sequence Tags (EST) back to the Drosophila genome, Boutanaev et al . [ 5 ] observed almost one thirds of 1661 testis-specific genes are clustered on chromosomes. Similarly, positional clustering of co-expressed genes was also reported in Caenorhabditis elegans . Although operon and tandem duplication are major mechanisms for the observed co-expression gene clusters in the worm, co-expression gene clustering is still evident after exclusion of these two causes [ 3 , 7 ]. Clusters of highly co-expressed genes were also revealed in the human and other mammalian genomes [ 6 , 8 , 5 , 12 ]. Most of these studies focused on a set of tissue-specific genes that are highly expressed. For example, by analyzing the expression profiles of genes in carious tissues, Caron et al . [ 8 ] showed that highly expressed carious genes tend to cluster in large domains, called RIDGEs ( Region of IncreaseD Gene Expression ). However, these RIDGES seem to consist mainly of housekeeping genes [ 6 ]. Gabrielsson et al . [ 9 ] performed a microarray analysis of genes expressed in the human adipose tissue. By mapping these genes back to the human genome, they observed clusters of adipose tissue specific genes. Dempsey et al . [ 10 ] investigated genes from chromosomes 21 and 22 expressed in the human cardio-vascular system. Using EST ( Expressed Sequence Tag ), they showed positional clustering of these genes. These studies suggest that clusters of tissue-specific genes do exist, and might be more frequent than initially thought. Here we performed a genome-wide analysis of testis-specific genes in the mouse genome using a method similar to one proposed in [ 5 ]. We used a publicly available EST (Expressed Sequence Tag) database to identify differentially expressed genes. After mapping testis-expressed ESTs back to the genome, we obtained the testis-expressed gene expression profile by assembling overlapped ESTs. In the same way, we got embryo-expressed gene expression profile. Testis-specific genes were generated by subtracting embryo-expressed gene expression profile from testes-expressed one. From the testis-specific gene expression profile, clustered genes were observed. To further investigate testis-specific gene clusters in the mouse genome, we used the full-length cDNA sequences publicly available in the Fantom 2.1.1 database to identify differently expressed genes. After mapping 60,770 cDNA sequences back to the genome, we obtained the gene expression profile by assembling overlapping cDNA sequences into a single gene model. Testis-specific genes were selected by using only sequences from libraries where the tissue type included the keyword "testis" and not the keyword "embryo". Again, our statistical analysis shows the existence of testis-specific gene clusters along the mouse chromosomes even after removal of tandem duplicated genes. This suggests that considerable clusters of tissue-specific genes do exist in higher eukaryotes. Our results provide further support for the hypothesis that chromatin domain model may be involved in the regulation of gene expression in higher eukaryotes Results The testis-expressed profile is a global view of gene expression and provides a foundation for understanding how the tissue functions in molecular level. We assume that most of the testis-specific genes are expressed after embryo phases in mouse. Therefore, we selected only testis genes models that exclude any expression in embryo for the construction of the testis-specific gene profile, which reveals positional clustering along chromosomes. Here, we analyzed the cluster distribution on all the chromosomes except for chromosome Y. (In Y chromosome, we only identified 12 testis-specific genes and 7 other genes.) Based on our study, we obtain the following observations. a. Density of testis-specific genes The testis-specific gene expression profile along the mouse genome shows that testis-specific genes are distributed on all the chromosomes with different gene density. The ratio (R) of the number of the observed testis-specific genes (O) to the expected number (E) is used to measure the gene density, where E is calculated according to the chromosome's size on assumption that the testis-specific genes are uniformly distributed throughout the genome. The lowest gene density appears on chromosomes 1 (R = 0.834), 15 (R = 0.893), 16 (0.855), and 18 (R = 0.767) except for sex chromosomes. But, chromosomes 9 (R = 1.145), 11 (R = 1.374), 14 (R = 1.207) and 19 (R = 1.48) have rather high gene density. Within a chromosome, the distribution of testes-expressed genes also shows diverse density. b. Genomic location is highly correlated with testis-specific gene expression We observed numerous groups of physically neighboring testis-specific genes that shared strikingly similar expression levels. In eukaryotes, it is clear that the activity of a promoter can be modified by transcription factors binding to DNA sequence, which are located from hundreds to thousands of base pairs from the promoter. Recently, increasing evidences show that genes may also be regulated in a group. For example, when transgenes are removed from their local environment and reinserted elsewhere in the same genome, they tend to show variant levels of expression [ 13 ]. More commonly, most eukaryotic chromosomes exhibit transcriptionally active (euchromatin) or inactive (heterochromatin) regions. In animals, heterochromatin is typically found near the centromere and other regions of low sequence complexity. Since RIDGEs do exist and neighboring genes show similar expression level, we conclude that genomic location has impact on gene expression. c. A large proportion of testis-specific genes are in adjacent pairs We consider two genes are adjacent if there are no other genes between them on the same chromosome. We observed that 1170 testis-specific genes appear in adjacent pairs, which account for 31.8% of the testis-specific genes after removal of tandemly duplicated genes. There are 350 adjacent pairs, 86 adjacent triplets, 23 adjacent quadruplets, and 4 quintuplets (see Table 1 for details), significantly more than would be expected by chance. In fact, the number of observed adjacent gene clusters (including singletons) is smaller than the expected number minus the standard deviation for every chromosome other than chromosomes 15 and 19. Table 1 Chromosomal distribution of adjacent testis-specific genes after removal of tandemly duplicated genes. Size Cluster distributions of adjacent testis-specific genes on chromosomes 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 X Total 2 26 22 21 26 29 22 17 20 14 24 19 18 18 20 6 15 13 9 9 12 350 3 2 5 3 6 6 4 4 4 12 4 8 1 2 5 5 4 3 3 1 4 86 4 2 1 1 2 2 3 1 2 1 2 3 2 1 23 5 1 1 1 1 4 The genes in the quintuplets are probably worthy to be investigated biologically. There is a quintuplet on each of chromosomes 6, 12, 14 and X. The quintuplet on chromosome 6 is located in a 400 kb segment starting at 144691608 and ending at 145141155; the one on chromosome 12 is located in a 600 K segment starting at 101700908 and ending at 102297618; the one on chromosome 14 is located in a 700 K segment starting at 62802300 and ending at 63500814. However, the one on chromosome X seems less significant, which is located in a 3719 K segment starting at 113253694. d. Testis-specific genes show obvious clustering We evaluated the positional clustering of testis-specific genes using a model called the neighborhood model. We define clusters using the distance between two neighboring testis-specific genes along the chromosome. Two testis-specific genes are in a cluster if there is a series of testis-specific genes between them such that the distance between two neighboring genes in the series is less than a specified distance (D). We obtained 3679 testis-specific genes on all the chromosomes except for Y after removal of tandemly duplicated genes. We conducted statistical analysis with 4 different values of D: 25 K, 50 K, 75 K, and 100 K (see Table 2 ). Table 2 Distribution of testis-specific gene clusters after removal of tandemly duplicated genes. Here only the numbers of clusters having at least 2 genes on all the chromosomes except for chromosome Y are given. Size Cluster distributions on chromosomes (distance = 25 K) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 X Total 2 18 22 17 22 23 13 17 17 18 22 16 12 15 15 6 8 7 7 7 7 288 3 1 1 2 1 2 3 1 2 1 3 1 21 4 1 1 2 5 6 Size Cluster distributions on chromosomes (distance = 50 K) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 X Total 2 26 30 22 32 34 20 21 22 19 31 22 16 16 20 15 14 12 8 6 6 392 3 2 5 1 2 6 1 8 3 6 3 3 3 4 3 2 1 2 1 56 4 1 5 2 1 2 2 1 1 1 1 2 19 5 1 1 6 Size Cluster distributions on chromosomes (distance = 75 K) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 X Total 2 27 28 23 37 47 29 24 29 17 34 29 17 18 24 16 16 19 11 9 10 454 3 1 4 4 5 4 5 9 2 9 2 7 3 3 6 5 4 1 1 3 1 79 4 2 6 3 1 3 1 2 4 1 1 1 1 1 1 28 5 1 2 1 4 6 1 1 Size Cluster distributions on chromosomes (distance = 100 K) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 X Total 2 32 33 24 31 42 33 28 27 19 39 23 21 21 25 16 15 24 9 9 12 463 3 2 6 4 7 8 6 12 8 12 3 11 2 4 7 4 6 2 3 6 1 114 4 3 6 5 3 1 1 3 1 3 7 2 2 2 2 1 1 2 1 46 5 1 1 1 2 1 1 7 6 1 1 2 When D = 25 K, we obtained 288 two-gene, 21 three-gene, and 2 four-gene clusters. In total, 18% of the testis-specific genes are contained in these clusters. By incorporating the variance of gene density in different regions on a chromosome, we conclude that 45 two-gene, 7 three-gene and 2 four-gene clusters are significant with p-value less than 4.9·10 -2 , 3.1·10 -2 and 1.2·10 -2 respectively. The most significant cluster is the four-gene cluster on chromosome 2 (p-value = 1.3·10 -3 ). When D = 50 K, we obtained 392 two-gene, 56 three-gene, 19 four-gene and 1 five-gene clusters. In total, 28% of the testis-specific genes are in these clusters. By incorporating the variance of gene density in different regions on a same chromosome, we conclude that 14 two-gene, 6 three-gene and 2 four-gene clusters are significant with p-value less than 4.9·10 -2 , 3.1·10 -2 and 1.2·10 -2 respectively. The most significant one is a two-gene cluster on chromosome 18 (p-value = 5.3·10 -3 ). When D = 75 K, we obtained 454 two-gene, 79 three-gene, 28 four-gene, 4 five-gene and 1 six-gene clusters. In total, 35% of the testis-specific genes are in these clusters. By incorporating the variance of gene density in different regions on a chromosome, we conclude that 10 two-gene, 2 three-gene and 3 four-gene clusters are significant with p-value less than 4.1·10 -2 , 3.1·10 -2 and 4.9·10 -2 respectively. The most significant one is a two-gene cluster on chromosome 16 (p-value = 5.2·10 -3 ). When D = 100 K, we obtained 463 two-gene, 114 three-gene, 46 four-gene, 7 five-gene and 2 six-gene clusters. In total, 41% of the testis-specific genes are in these clusters. By incorporating the variance of gene density in different regions even on the same chromosome, we conclude that 12 two-gene, 4 three-gene and 1 four-gene clusters are significant with p-value less than 4.9·10 -2 , 4.8·10 -2 and 3.2·10 -2 respectively. The most significant one is a two-gene cluster on chromosome 16 (p-value = 9.2·10 -3 ). We also observed that there are more significant gene clusters on chromosomes 5, 7, 12 X than any other chromosomes. Discussion It seems that in most eukaryotes, the transcription factor machinery is sufficient for ensuring co-regulation so that co-localization of genes in a genome is not necessary. However, there could be some degree of selection for keeping co-regulated genes in clusters on a chromosome, for instance, to make them available to transcription more efficient as a group. Gene clusters due to tandem duplication Clustering of co-expressed genes could be due to tandem duplication. Tandem duplication and subsequent divergence of amplified copies might result in an array of paralogous genes that may retain common regulatory element. Such a type of clustering is exemplified by the Hox gene family [ 14 ]. From the testis-specific gene representation profile, we can observe that besides co-expressed adjacent gene pairs, there are large-scale clusters on chromosomes. To determine to what degree tandem duplication contributes to the clustering of testis-specific co-expressed genes, we analyzed cluster distribution after removal of duplicated genes. To identify duplicated testis-specific genes, we performed an all-against-all BLAST homology search using default BLASTN settings. (See the method part for details.) In total, 240 duplicated genes were identified and removed from the list of the testis-specific genes (see Table 3 for details). Table 3 Testis-specific gene distribution before and after removal of tandem duplications. Chromosome Number of genes Difference Before removal After removal 1 249 234 15 2 298 280 18 3 223 209 14 4 230 220 10 5 253 232 21 6 210 201 9 7 229 211 18 8 220 205 15 9 217 207 10 10 223 207 16 11 250 239 11 12 171 162 9 13 179 168 11 14 213 199 14 15 142 136 6 16 128 119 9 17 155 145 10 18 104 94 10 19 109 106 3 X 113 105 8 Y 12 9 3 Total 3928 3688 240 In the most of chromosomes except for chromosomes 2, 5, 7, 10, no more than 15 duplicated genes were identified. On chromosomes 2, 5, 7, 10, we identified 18, 21, 19, 16 duplicated genes respectively. Since, for each of distance threshold value (25 K, 50 K, 75 K and 100 K), we obtained more than 300 gene clusters, we conclude that duplication has no obvious effect on the positional clustering of testis-specific genes in the mouse genome. Non-duplicated clusters of genes Eukaryotic genes have not only promoters, but also one or more enhancers. An enhancer is a DNA sequence that influences transcription of neighboring genes. Enhancers are different from promoters in their ability to act over thousands of base pairs upstream or downstream of a gene. While a promoter is responsible for initiating low levels of transcription and determining the transcription start site, enhancers are responsible for increasing or "enhancing" transcription levels, as well as for regulating cell- or tissue-specific transcription. There are presently two models for non-duplicated gene clustering. One is the incidental expression mode. In this model, co-regulation within an expression neighborhood is due to incidental interactions between transcriptional enhancers and promoters [ 16 ]. When transcription factor binds at the appropriate sites and activates nearby genes together with the target gene, a group of nearby genes tend to express together. In this case sites that bind strong transcriptional activators should create new neighboring co-expressed genes. An alternative model of clustering is that gene clusters may correspond to regions of active chromatin. This explanation is called the chromatin domain model. In mammalian genomes, gene clusters are on a much larger scale (Mb rather than 5–20 kb) and may involve long-range chromatin interactions [ 18 , 19 ]. The structure of the chromatin domain model assumes that opening of the chromatin of an entire cluster depends on activation of a target gene within the cluster. Our findings can most be explained as regulation at the level of chromatin structure for the following reasons. First, the regions showing similarities in expression are quite large; each gene presumably has its own core promoter. Second, one or two genes in a group often displays a high level of differential expression. If the chromatin in a region that contains many genes was 'opened' so that a single target gene could be expressed, it might increase the accessibility of the promoters and enhancers of other genes to transcriptional machinery, leading to corresponding increases in their expression. Such an effect could account for the observations we have made. Although the organization of neighborhoods along a chromosome indicates that there must be cis-acting structures, no known structure correlates with the blocks. The structure basis of positional gene clustering is still under investigation. After removal of tandemly duplicated genes, significant clusters of co-expressed testis-specific genes were still found on all the chromosomes. However, the mechanism responsible for this observation is still unknown. Conclusion We revealed the positional clustering of testis-specific genes in the mouse genome by using EST profiling. Our analysis implies that the mouse genome is enriched with clusters of testis-specific genes. A significant trend for large clusters (at least 3 genes) was discernible. Positional clustering of co-expressed, non-homologous genes has also been reported in nematode [ 3 ] and in mammals [ 6 ]. This indicates that positional clustering is common in higher eukaryotes. Clustering of tissue-specific genes suggests that there exists some higher order in regulation of gene expression, most probably due to the chromatin domain. From this fascinating starting point we expect further insight into the significance of gene clustering and the mechanisms that generate them as more genomes are sequenced and more expression patterns are studied in the coming years. Methods Testis expressed gene models A total of 60,770 full-length cDNA sequences are available in the Fantom (Funtional Annotation of Mouse) 2.1.1 database. These sequences were aligned back to the mouse genome using BLAT [ 21 ] (a program that specializes in the alignment of EST/cDNA sequences to genomic sequences). Overlapping alignments were assembled to produce a complete gene model. This resulted in 34,402 genes. The 3,928 testis-specific genes were identified on the basis of having the keyword "testis" but not the keyword "embryo" in the tissue type information associated with the library. Removal of tandem repeats Clusters of co-expressed genes could be due to tandem duplication. To remove effect of local duplication, we did an all-against-all BLAST search on all 3,928 testis-specific genes. As a result, 240 duplicated genes were identified on the basis of an expectancy value less than 10 -50 as proposed by Friedman and Hughes [ 20 ]. This results in the final 3,688 testis-specific genes used for our statistical analysis. Statistic analysis The neighborhood model Recall that we define clusters using the distance between two neighboring testis-specific genes along the chromosome. Two testis-specific genes are in a cluster if and only if there is a series of testis-specific genes between them such that the distance between two consecutive genes in the series is less than a specified threshold ( D ). To incorporate the variance of gene density in different regions on a chromosome, we divide a chromosome into segments of length L (2000 Kbp here) so that genes on such a segment are roughly uniformly distributed and analyze the significance of a cluster within the segment according to the number N of all the genes (not just testis-specific genes) in the segment and the values of parameters L and D . Since we assume the (start) position of a gene is uniformly distributed in a segment of length L, it falls in an interval ( x , x + D ) in the segment with probability D / L . Thus, the number of genes that fall in this interval has a binomial distribution with mean ND / L . In our case, D / L is smaller than 0.1. This distribution is approximately Poisson with mean ND / L . Here we use this approximation in our analysis. The Poisson approximation implies that the number of the genes that falls in a randomly chosen interval of length D has a Poisson distribution with mean m = ND / L , and so the probability that at least one gene falls in this interval is 1 - e - m . Suppose g 1 , g 2 , …, g n form a cluster in the segment. Then, all the distances of two successive genes are less than D . Thus, the number of genes in a cluster minus one has a geometric distribution with p = 1 - e - m (see page 10, [ 22 ]). This implies that the probability that a cluster has n genes is (1 - p ) p n -1 , n = 0, 1, 2, …. Hence, the p-value of a cluster with n genes is the probability that a cluster has more than n genes, which is p n . In our analysis, D = 25 K, 50, 75 K, 100 K. Actually, the distance between two successive testis-specific genes in the most of observed clusters is much smaller than D in each case. Hence, the real p-value for each cluster could be much smaller than one reported in the result section. Adjacent gene clustering To remove the effect of non-uniform distribution of genes on analysis, we adopt an idea from order statistics. Here, we focus on the position rank of a gene rather than its specific position by ordering all the genes according to their positions on a chromosome. We also treat the set of the testis-specific genes and the set of all the other genes as two types of identical objects. In this way, a specific gene distribution on a chromosome is modeled as a 0–1 string in which 0 represents a test-specific gene and 1 a non-testis-specific gene. Assume there are M genes in a chromosome and T of them are testis-specific genes, where T ≤ M . Then, there are possible gene arrangements on the chromosome and the number of arrangements with r adjacent testis-specific gene clusters is . Thus, the probability that a random arrangement has r adjacent testis-specific gene clusters is Using above formula, we can obtain that the mean number of adjacent testis-specific gene clusters in a random arrangement is and the standard deviation is . Take chromosome 4 for example. We have M = 2060, T = 220, we obtain that the mean number is 196.6 and the standard deviation is 4.4. However, we only observed 182 adjacent gene clusters (including singletons) on the chromosome. Authors' contributions BL and QL carried out the software development, sequence analysis, and the design of the study. LXZ conceived of the study, performed the statistical analysis, and participated in its coordination. All authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC548148.xml |
512285 | Outcome after acute myocardial infarction: a comparison of patients seen by cardiologists and general physicians | Background The management of acute myocardial infarction (AMI) has improved over the last 50 years with the more frequent use of effective medicines and procedures. The clinical benefit of the speciality of the attending physician is less clear. The United Kingdom National Service Framework for coronary heart disease (CHD) suggested that patients with CHD are likely to benefit from cardiological supervision. We set out to assess the effect of access to cardiologists on survival among AMI patients admitted in two UK hospitals. Methods The study was conducted in a university hospital and a district general hospital in England. Information was obtained on age, sex, ethnicity, Carstairs socioeconomic deprivation category derived from postcode of residence, comorbidity, distance from hospital and medication from all patients admitted with acute myocardial infarction in two acute trusts between July 1999 and June 2000. Record linkage to subsequent Hospital Episode Statistics and Registrar General's death records provided follow up information on procedures and mortality up to eighteen months after admission. Cox proportional hazard models were used to investigate the main hypothesis controlling for confounding. The main outcome measure was 18-month survival after myocardial infarction. Results Access to a cardiologist was univariately associated with improved survival (hazard ratio 0.16, 95% CI 0.10 to 0.25). This effect remained after controlling for the effect of patient characteristics (hazard ratio 0.22, 95% CI 0.14 to 0.25). The effect disappeared after controlling for access to effective medication (hazard ratio 0.70, 95% CI 0.33 to 1.46). Conclusions Access to a cardiologist is associated with better survival compared to no access to a cardiologist among a cohort of patients already admitted with AMI. This effect is mainly due to the more frequent use of effective medicines by the group referred to cardiologists. Hospitals may improve survival by improving access to effective medicines and by coordinating care between cardiologists and general physicians. | Background The management of acute myocardial infarction (AMI) has improved over the last 50 years with the more frequent use of effective medicines and procedures. The clinical benefit of the speciality of the attending physician is less clear. The effect of the speciality of the attending physician on mortality in AMI has been studied mainly in the United States (US) with conflicting results arising from the studies [ 3 - 6 ]. The explanations considered for the difference in mortality observed in some of the studies include patient's condition (case mix and comorbidity)[ 3 ], volume of workload[ 5 , 7 ] and treatment given[ 3 ]. Previous studies show that knowledge and use of effective medicines is better among cardiologists compared with general physicians in the US [ 8 - 10 ]. Co-ordination of care between cardiologists and non-cardiologists also improves survival of patients seen by non-cardiologists [ 11 ]. The National Service Framework (NSF) for coronary heart disease (CHD) published in March 2000, suggested that patients with CHD are likely to benefit from cardiological supervision. To provide this level of care, all acute hospitals will eventually need a minimum of two cardiologists [ 1 ]. A recent survey by the Royal College of Physicians found an average of 1.7 whole time equivalent (WTE) cardiologists in the 211 hospitals receiving patients with AMI in the UK [ 2 ]. Eight hospitals did not have a cardiologist. There is a need to establish whether the involvement of a cardiologist in the management of AMI patients affects the quality and outcome of care, and if so to identify ways to improve the outcome of care for patients unable to gain access to a cardiologist. This study aims to assess the effect of access to cardiologists on survival among AMI patients accounting for access to effective investigation, medication, procedures and the underlying condition of the patient at presentation. Methods A retrospective cohort design was used. Study population and inclusion criteria All patients admitted to two hospitals in Eastern England between 1 st July 1999 and 31 st June 2000 with a discharge diagnosis of AMI were included in the analysis. A diagnosis of AMI was based on evidence of raised cardiac enzymes and/or other indicators of myocardial necrosis and on a physician's judgement of ECG changes indicative of AMI. Data collection and analysis Hospital Episode Statistics were used to extract records of all patients admitted with the ICD10 diagnostic codes for Acute Myocardial Infarction (I21, I22, I23) in any diagnostic field. The records were restructured to produce one record per patient using the new NHS number. For records with missing NHS number, a unique identifier was created using date of birth, sex and post code. The unique identifier was also used as a check for the records with NHS numbers. Records of all Finished Consultant Episodes with OPCS 4 codes of either Coronary Artery Bypass Grafting (K40–46) or Percutaneous Transluminal Coronary Angiography (K49–50) were extracted. The process was repeated for angiography (K63 – 65). Record linkage with subsequent Hospital Episode Statistics and Registrar General's records provided follow up information on procedures and death up to eighteen months after the index AMI. All cases were reviewed by examining the case notes. Detailed information on variables summarised below was obtained. List of variables extracted from patients' case notes • Hospital where patient was first admitted • Whether the patient was seen by a cardiologist or not, determined by evidence of a cardiology review in a patients case note • The average distance from residence to hospital first seen was determined by mapping the distance in miles using the RAC™ website distance calculator • A history of comorbidity – defined as the presence of diabetes, metastatic malignancy, cerebrovascular accident (CVA), asthma, chronic obstructive airway disease, renal impairment, endocrine disorder, chronic infection, dementia • Physical impairment with poor mobility • Procedures – exercise testing, angiography, CABG, PTCA • Appropriate medication – β blockers, aspirin, ACE inhibitors, thrombolysis (table 2 ) Table 1 Patient characteristics, and whether seen by cardiologist or by general physician Patient characteristic Physician Cardiologist (N = 275) Non Cardiologist (N = 201) Mean age (SD) 64.4 (SD 12.8) 74.8 (SD 11.4) Sex n (%) Male 275 (71.4) 201 (52.8) History of: Congestive heart failure % 11.3 29.6 Angina % 94.9 86.7 Hypertension % 41.8 46.6 Diabetes mellitus % 12.1 17.1 Impaired mobility % 7.8 17.7 Smoking % 36.0 49.6 Impaired left ventricular function % 24.6 25.0 Table 2 Odds ratios for drug and procedure use after AMI by physician (cardiologist relative to non-cardiologist) Intervention Number (Percent) Odds ratio (95% CI) Cardiologist (275) Non Cardiologist (201) Thrombolytic therapy* 140 (51%) 66 (33%) 2.08 (1.35 – 3.23) β blocker on discharge* 206 (75%) 90 (45%) 3.70 (2.38 – 5.56) Aspirin* 264 (96%) 160 (80%) 6.67 (3.23 – 14.29) ACE inhibitors 124 (45%) 80 (40%) 1.27 (0.01 – 1.92) Exercise testing 192 (70%) 46 (23%) 7.70 (4.76 – 12.50) Coronary angiography 151 (55%) 12 (6%) 4.76 (3.22 – 7.14) Coronary angioplasty 74 (27%) 30 (15%) 1.92 (1.28 – 2.94) Coronary artery bypass grafting 66 (24%) 30 (15%) 1.75 (1.15 – 2.70) * See text for eligibility criteria • A history of Smoking Previous MI Hypertension • Impaired left ventricular function based on echocardiography result or angiography • Severity of AMI using the number of vessels affected (vessels were considered affected if a lesion of 50% or more was noted in the angiography report) • Demographic factors: age, sex, ethnicity, socioeconomic status and Carstairs socio-economic deprivation category derived from postcode of residence Baseline patient characteristics were tabulated. Categorical variables were tested for statistical significance using the χ 2 test. Continuous variables were tested using the Student t test comparing patients seen by cardiologists and those seen by other physicians. The relative odds of drug and procedure use by speciality of physician were calculated. Specific variables included the use of thrombolytic therapy, β blockers, aspirin, ACE inhibitors among clinically appropriate groups, and the use of exercise testing, coronary angiography, angioplasty and coronary artery bypass grafting (CABG). Cox proportional hazards models were used to investigate the effect on survival of seeing a cardiologist. Multivariate models were fitted to control for the effect of patient and hospital characteristics. Controlled factors included: age, sex, comorbidity, hospital, distance from patients residence to hospital and Carstairs socio-economic deprivation category in the first instance. The effect of angiography, revascularisation and the use of effective medicines (aspirin and/or β blockers and/or thrombolysis – eligibility criteria listed below) were subsequently introduced into the model to investigate whether these affected outcome. The factors were included in the model either because they are known to confound the association between type of physician and survival or are known to be associated with survival and were not comparable between the two groups. Eligibility criteria for appropriate use of effective medicines [ 13 ]. Thrombolytic therapy • No warfarin therapy on admission • ST-segment elevation on initial ECG • Less than 12 hours since onset of chest pain • Systolic BP less than 180 mmHg and diastolic BP less than 110 mmHg at presentation • No severe CVA, gastrointestinal disease or chronic liver disease Aspirin • No haemorrhagic complication • No severe CVA, gastrointestinal disease or chronic liver disease or renal failure • No warfarin therapy on admission β blockers • No chronic lung disease • No cardiogenic shock, hypotension, complete heart block or decompensated heart failure The characteristics of the two hospitals are summarised figure 1 . The results from the two hospitals were combined in order to make our findings more generalisable. The general characteristics did not show important differences in the organisation of care between the hospitals. It is unlikely there are important unmeasurable factors associated with a particular hospital that might cause confounding, but hospital of care is still included as a variable in the multivariate analysis to control for centre effect. Patients from the two hospitals undergo procedures in the same specialist cardiothoracic hospital. Figure 1 Characteristics of study hospitals. The main outcome was eighteen-month survival defined as time between date of AMI and death. Statistical analysis was carried out using Stata (Version 6.0). In general for Cox regression analysis, the number of events i.e. deaths (and similarly the number of non-events) per variable modelled should be at least 10 [ 12 ]. Ethical approval was obtained from the two local research ethics committees of the hospitals involved in the study. Permission from the director of research and Caldicott guardians of data in each hospital was also obtained. Results Medical records were obtained for 94% of eligible patients (476 of 506 patients). Four patients did not have enzyme changes of myocardial infarction or were clear cases of miscoding. The mean age of the cohort was 69.3 years (SD 13.2) and 34.5% were female. The patients with missing records did not differ from those with records in terms of age or sex using information from routine data (mean age was 70.4 years with SD 11.1; 36.2% were female). There were no significant differences in the proportion of data missing between the two hospitals including those for eligibility to effective medications and co morbidities. General characteristics Patients in our sample seen by a cardiologist were younger, with a higher proportion of males to females. They were more likely to have a history of CHD prior to myocardial infarction, but less likely to have a history of congestive heart failure or hypertension (Table 1 ). Patients with impaired mobility or who smoked were less likely to be seen by a cardiologist. The difference in the proportion of patients with echocardiographic evidence of impaired left ventricular function complicating their myocardial infarction by physician seen was not significant. Drug and procedure use The likelihood of receiving effective medication (thrombolytic therapy, β blockers, aspirin) was higher among patients seen by a cardiologist compared with those not seen by a cardiologist. Furthermore, patients seen by a cardiologist were more likely to have undergone exercise testing, angiography and revascularisation procedures (Table 2 ). The use of ACE inhibitors was not more frequent among cardiologists. Survival Univariate analysis indicated that patients seen by a cardiologist had better survival (hazard ratio 0.16, 95% CI 0.10 to 0.25). Effect of seeing a cardiologist after controlling for patient characteristics The survival of patients seen by a cardiologist was still better than that for those not seen by a cardiologist (hazard ratio 0.22, 95% CI 0.14 to 0.25) after controlling for the confounding effect of pre-existing comorbidity, age, sex, hospital first seen, deprivation and distance (Table 3 ). The effect of seeing a cardiologist remained after excluding patients that died within a week of MI. The hazard ratio was 0.21 (95% CI 0.12 – 0.37). Table 3 Cox regression analysis for eighteen month survival comparing cardiologists and non-cardiologist physicians and controlling for potential confounders. Variable Mortality (%) Adjusted hazard ratio (CI) * Adjusted hazard ratio (CI) # Seen by a Cardiologist Yes 30 (11%) 0.22 (0.14 – 0.38) 0.70 (0.33 – 1.46) No 88 (44%) Comorbidity Yes 15 (25%) 1.11 (0.62 – 1.99) 0.98 (0.49 – 1.98) No 107 (26%) Hospital A 58 (24%) 1.01 (0.67 – 1.51) 1.08 (0.71 – 1.73) B 65 (28%) Age (years) 30 – 40 1 (8%) 1.05 (1.03 – 1.08) 1.03 (0.99 – 1.06) 41 – 60 8 (7%) 61 – 70 18 (17%) 71 – 99 96 (40%) Sex Males 64 (21%) 0.91 (0.62 – 1.35) 0.83 (0.44 – 1.55) Females 58 (34%) Distance 1.0 (0.97 – 1.04) 0.96 (0.88 – 1.03) Angiography Yes 12 (7%) - 0.87 (0.24 – 3.17) No 112 (36%) Revascularisation Yes 14 (13%) - 0.25 (0.04 – 1.52) No 110 (30%) Effective medicines γ Yes 48 (14%) - 0.15 (0.07 – 0.29) No 76 (56%) * Adjusted for patient characteristics, hospital # Adjusted for patient characteristics, hospital and the use of effective medicines and procedures γ Access to aspirin, and/or β blockers and/ or thrombolysis if appropriate Effect of seeing a cardiologist after controlling for patient characteristics and use of effective intervention The effect of seeing a cardiologist on survival became non-significant once access to angiography, revascularisation procedures and effective medicines (aspirin and/or β blockers and/or thrombolysis) had been adjusted for. The hazard ratio was 0.70 (95% CI 0.33 – 1.46). The fully adjusted model indicated that the most important factor affecting survival was access to effective medication. (Table 3 ). Discussion This study improves our understanding of the care of AMI by cardiologists and general physicians in UK hospitals. Access to a cardiologist was univariately associated with better survival. This effect remained after controlling for the effect of patient characteristics, including the presence of comorbidity, but disappeared when the confounding effect of access to effective medicines was controlled for. As noted in the analysis, patients seen by a cardiologist were more likely to have been prescribed these medicines and to have had exercise testing, angiography and revascularisation. This implies that the survival advantage associated with being seen by a cardiologist is due to the more frequent use of effective medicines and is similar to findings by Chen J et al in the United States [ 3 ]. Previous studies have shown a high level of miscoding in routine data for AMI patients (RM Norris cited in [ 2 ]). The use of a case note review improved the findings of this study by assisting in ascertaining the diagnosis of myocardial infarction and ensuring the accuracy and completeness of the data used in the analysis. The possibility remains that some cases not seen through the cardiology department and not recorded in the patient administration system were missed. This is likely to be a small number and unlikely to invalidate the findings. Identifying all patients with infarction is a major shortcoming of many studies that estimate mortality among patients seen by cardiologists [ 2 ]. This study includes all patients irrespective of where they were managed in the hospital; however patients dying from myocardial infarction before arriving at the hospital were not included. All cases diagnosed by the managing clinician as AMI with ECG changes and found to have enzyme changes indicating myocardial necrosis were included in the analysis after the case notes were reviewed. This may have involved some misclassification due to misdiagnosis, but this is likely to occur at random, affecting only a small number of cases, and can only underestimate the effect of each factor. Limitations of this study include potential for bias in the allocation of patients to treatments and physicians (ie cardiologist/non cardiologist) that were not measured. Another possible limitation is bias arising from medical case notes which were unobtainable. Records from routine data were used to examine whether those with missing records differed in any systematic way from those with available records; it was found that they did not differ in age or sex. The study did not use any criteria for judging clinical appropriateness of the procedures used. Incomplete recording of information in the case notes hampered the determination of appropriateness for effective medication. However there was no difference in the proportion of missing data between the two hospitals. Since doctors may selectively be referring younger patients with lower comorbidity to a cardiologist, we controlled for age and comorbidity in the analysis. However, residual confounding from age and comorbidity could still account for some of the observed difference in survival. All patients admitted to the Coronary Care Unit (CCU) in one of the hospitals are seen by a cardiologist and have a higher chance of accessing thrombolysis, angiography and revascularisation. A cardiologist also sees the majority of patients admitted to the CCU in the second hospital. The protocol for the management of AMI in both hospitals stipulates a cardiology review of all AMI patients. Access to a cardiologist may be a proxy measure of access to effective treatment and it may not be the trigger for effective treatment. These limitations mean that the findings should be interpreted with a degree of caution. In addition our inability to use formal appropriateness criteria limits the interpretation of findings. Despite these limitations the results indicate that, in the short term, acute trusts can improve survival of patients by increasing the use of effective medicines among general physicians. Conclusions We observed better survival among patients seen by a cardiologist compared with patients with no access to a cardiologist, among a cohort of patients already admitted with AMI. This effect was entirely explained by the more frequent use of effective medicines by cardiologists in the multivariate analysis. About eight hospitals in the UK Royal College of Physicians survey have no cardiologist and another 30% have a single cardiologist [ 2 ]. It will take time to provide the minimum of two cardiologists per hospital as recommended in the CHD NSF. There are several reasons why a hospital may benefit from appointing a cardiologist, ranging from the treatment of a specific subgroup of patients that will benefit from revascularisation to prompt management of angina patients via rapid access chest clinics. However, in the short term hospitals can improve the survival of patients admitted with AMI by improving access to effective medicines. Coordination of care between cardiologists and general physicians and targeted interventions using feedback from audit, research and peer education are likely to lead to more frequent use of effective cardiovascular medicines by general physicians. Competing interest None declared Authors' contributions All authors contributed in writing the paper, in addition, IA collected the data and conducted the analysis, DK conceived the idea, BA provided advice on statistical analysis, JP provided input in the conduct of the study, PW contributed in refining the research question and discussing the findings of paper. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC512285.xml |
544941 | Assessing the impact of humanitarian assistance in the health sector | There have been significant improvements in the design and management of humanitarian aid responses in the last decade. In particular, a significant body of knowledge has been accumulated about public health interventions in emergencies, following calls for developing the evidence base of humanitarian health interventions. Several factors have prompted this, such as the increased volume of humanitarian assistance with subsequent higher levels of scrutiny on aid spending, and greater pressure for improving humanitarian aid quality and performance. However, documentation of the ability of humanitarian interventions to alleviate suffering and curb mortality remains limited. This paper argues that epidemiological studies can potentially be a useful tool for measuring the impact of health interventions in humanitarian crises. Survey methods or surveillance systems are mainly used for early warning or needs assessment and their potential for assessing the impact of aid programmes is underutilised. | Introduction In the last decade, humanitarian aid agencies have put a great deal of effort into improving the design and management of humanitarian responses in conflict situations: humanitarian agencies have agreed on international standards, numerous technical guidelines have been produced, training programmes are offered on different aspects of humanitarian programming, and aid programmes are evaluated more systematically. Initiatives as distinct as the Code of Conduct, Sphere, Active Learning Network for Accountability and Performance in Humanitarian Action, the Humanitarian Accountability Partnership International, the Quality Project from the French group Urgence, réhabilitation, développement and People in Aid all have in common a concern for the quality, performance and impact of humanitarian assistance [ 1 ]. In particular, a substantial body of knowledge has been accumulated regarding public health interventions in emergencies, in response to calls for developing the evidence base of humanitarian health interventions [ 2 - 4 ]. Most of these developments have been stimulated by the findings of the system-wide Rwanda evaluation, which highlighted serious questions about the performance of some humanitarian organisations and, more importantly, emphasised the responsibility of the international community: humanitarian aid alone cannot substitute for political action [ 5 , 6 ]. Despite the technical progress in the last decade, knowledge of the impact of humanitarian interventions in alleviating suffering and ultimately reducing mortality in the health and other sectors remains limited. There is a real question whether all the developments in accountability and implementation actually improve the overall performance of humanitarian assistance [ 7 , 8 ]. Whereas there is a significant evidence base on the effectiveness of interventions in acute emergencies, especially in refugee settings, the evidence base is much weaker for situations of protracted conflict with longer term programmes in less controlled settings. It is also difficult to determine the relative contributions of humanitarian aid programmes as distinguished from local coping mechanisms and/or the interventions of national governments. This lack of knowledge exists at the programmatic level and more generally at the sector-wide level. This paper examines the current practice of humanitarian agencies for measuring the impact of health interventions. It is based on a review of the literature as well as on the authors' own field experience. The question explored is whether field epidemiology provides a useful set of tools and methods to determine more accurately the impact of health interventions. The impact of humanitarian assistance The issue of impact is particularly high on the current humanitarian agenda. The increasing interest in impact analysis arises from a number of interlinked developments: the rapid increase in the overall volume of humanitarian assistance in the last decade- from 2 billion in 1990 to $5.5 billion by 2000 [ 9 ]- and the resulting scrutiny on how money is spent, the new public management agenda within the public sector and the adoption of results-based management systems by donors and some aid agencies. There is no definition of impact in humanitarian assistance. The most commonly used definition of impact in international aid is the one provided by the OECD/DAC which defines impact very widely as ' [t]he positive and negative, primary and secondary, long-term effects produced by a development intervention, directly or indirectly, intended or unintended' [ 10 ]. This paper concentrates on the intended effects of aid interventions, e.g. whether the original objectives of a programme have been met. The question of negative, unintended impact, which is undoubtedly important, falls out of the scope of this paper. A central question in an impact assessment is 'what would have happened in the absence of the aid programme?' In theory, there are two approaches to answering this question [ 11 ]: 1. To compare the impact with a control group that did not receive aid – a "with/without" comparison. 2. To do a "before/after" comparison for the beneficiaries of an intervention. These two approaches pose different sets of issues. The "with/without" comparison with the use of control groups creates ethical problems: it is difficult to deliberately exclude a group from access to potentially life-saving relief. It may happen that some particular groups do not receive relief, due to problems with access or lack of resources, but, as Hallam warns, comparisons between people who received assistance and those who did not need to be used very carefully [ 12 ]. As a result, very few experimental trials with randomised allocations of services have ever been conducted in emergency or refugee situations [ 13 , 14 ]. The "before/after" comparison generates different problems. In order to be valid, the comparison implies that no other factor influences the impact, so that it can be fully attributed to the intervention. In reality, a multiplicity of other factors have an influence on the impact, such as the presence of other aid programmes, local coping mechanisms, or changes in the social and economic environment. For example, it is very difficult to attribute a reduction of conflict-related excess mortality to humanitarian aid only: variations in the baseline mortality (due to seasonal trends, disease epidemics, HIV/AIDS etc.) can be significant. There is an increasing recognition of the importance of these other factors, and that humanitarian aid, however vital, is only one element of the picture [ 15 - 18 ]. A theoretical model for measuring impact There are three main considerations that are necessary for measuring the impact of health programmes: the strength of the evidence suggesting causation between the intervention and the change in health status, the validity of a baseline of comparison, and the validity of the indicator employed. These conditions also apply to other types of humanitarian assistance. Criteria of causation Over time, a great deal of debate has arisen over what epidemiological evidence constitutes proof that some exposure or input produced an effect versus what evidence simply implied an association. This distinction can be more than academic, as was seen over three decades of debate regarding the effects of smoking on health. Bradford-Hill put forward criteria for attempting to ascribe causation between an exposure and a health outcome [ 19 ]. These criteria are so widely utilised that some introductory text books simply refer to them as the epidemiological criteria of causation [ 20 ]. While his main motive was to attribute causation of a disease due to exposure to a chemical or biological agent, the logic of these criteria also applies to assessing the positive effects of favourable exposures, such as health programmes. Bradford-Hill said that all of the following conditions can contribute to the argument that an exposure induces a health consequence: 1. The greater the strength of the association, the more likely that it is causative 2. There is a dose-response relationship between the exposure and the health outcome 3. Exposure consistently induces the health consequence in different settings at different times 4. The exposure occurs before the health outcome 5. There is a biologically plausible explanation for the exposure resulting in the health outcome 6. There are not more plausible explanations for the health outcome 7. Experimental results add particular weight to the evidence For virtually all cause and effect health relationships, some of these criteria will not apply. For a programme to be shown to have an impact, criteria 4, 5 and 6 should always be met. Of particular concern to programme evaluation is the issue of biological plausibility and the amount of service provided. Programmes need to be evaluated with particular regard to the likelihood that the level of inputs provided could plausibly result in the outcome reported. That is, the number of clinic visits, or the amount of food provided per child etc. need to be sufficient to induce the health effects observed. Criteria of causation are important for interventions and settings where project impacts are usually not, or cannot be, measured. These criteria of causation can be applied to populations and programmes as readily as Bradford-Hill applied them to specific disease agents. For interventions with a vast literature documenting the attributable benefits (e.g. measles vaccination or Vitamin A supplements), the need to show "proof" that the intervention produced a health benefit may be small, but for many other emergency interventions (e.g. HIV prevention through educational efforts or health benefits from shelter) there may be little or no evidence that such programmes produce any health benefits, making the importance of documenting any benefits great. Most humanitarian programmatic efforts fall somewhere in between, employing types of programmes that have produced documented benefits in some settings, but have failed in others, and may or may not be producing benefits in the setting at hand. Table 1 provides an example of how Bradford-Hill's criteria can be used. Table 1 Application of Bradford-Hill criteria: Katana, Democratic Republic of Congo Starting in December of 2000, the International Rescue Committee (IRC) began a general health programme to support existing government services in Katana Health Zone, Democratic Republic of Congo (DRC). The IRC conducted population-based mortality surveys in this area with 345,000 mostly rural residents. The programme consisted of the provision of drugs, supplies, training and medical oversight in the clinics, a water provision and hygiene education programme in villages with the highest rates of cholera in 2000, a measles immunisation and vitamin A provision campaign, and support to the local health committees which included the donation of vouchers for the most indigent community members. Figure 1 below shows the crude mortality rate (CMR) over the period covered by 5 surveys conducted between 1999 and 2002. IRC claims to have reduced the excess CMR by 60% (from 4.9 to 2.8 deaths per 1000 per month where the baseline is assumed to be 1.5) during the period from 6 to 12 months after implementation and by 70% (from 2.8 to 1.9 deaths per 1000 per month) over the period from 12 to 24 months after implementation. In support of the results in figure 1 being a consequence of the health programme, IRC reported that: • attendance at the clinic rose by 147% between 1999 (~7400 visits per month) and 2001 (~18,300 visits per month average) • 70% of treatments were for malaria and diarrhoea, the main reported causes of death in the 1999 and 2000 surveys, and decreased as a cause of death in 2001 & 2002 • CMR in the five eastern provinces of DRC was estimated by IRC to have increased slightly in 2001 compared to 2000 • A survey in November of 2001 found that 60% of residents that had experienced fever in the preceding two weeks had sought treatment at a clinic Employing Bradford-Hill's criteria, this example shows that: 1) there was a considerable drop in CMR associated with the establishment of the intervention, 2) there was no dose-response effect, 3) the fact that IRC's two other areas of health programmes had similar (but somewhat less dramatic) reductions implies repeatability, 4) the benefit occurred after implementation, 5) the findings are biologically plausible (although 1 visit per resident per year seems low), 6) alternative explanations for the reductions cannot be ruled out given the variance over time and the dramatic changes in violent conflict, although IRC reports that the violence did not dramatically subside until 2002, 7) these are not experimental data. Finally, the fact that the CMR was measured by an apparently valid survey method implies that IRC probably did contribute to a reduction in mortality in Katana [26]. Validity of the baseline of evaluation Attempts to analyse the impact of humanitarian interventions are often handicapped by a lack of baseline data and a lack of knowledge about regular seasonal variations in key indicators of impact. For instance, baseline mortality rates are often not known. Countrywide figures are either unreliable, out of date, or not appropriate as they do not capture the regions where the conflict is occurring. A related problem that is continuously faced by humanitarian agencies is the lack of reliable population statistics [ 21 ]. Figure 1 Mortality in Katana, 1998 – 2002 When there are no baseline data, established norms can be used instead. For example, when people are arriving in a new location or are returning home, it is often impossible to determine the baseline before their arrival. In those cases, norms can be applied as an assumed baseline or as a threshold above or below which the indicator should not fall. Programmes lacking baseline data which keep mortality "low" or keep water and food provision high may be successful in terms of meeting their objectives but still not be able to quantify the impact of the intervention. Validity of the indicator employed The identification and use of relevant indicators is a crucial part of determining the impact of an intervention. Although the terminology varies, the literature generally distinguishes between performance (or process) indicators and impact (or outcome) indicators. • Performance (or process) indicators concerns both the outputs of a programme (number of latrines built, number of training conducted, the quantities of food delivered) and the process of implementation (coverage of a programme, equity of distribution, targeting) • Impact (or outcome) indicators are measures of the actual achievements intended by a programme. Mortality and malnutrition rates are the most commonly used impact indicators for humanitarian programmes, although interventions that aim to support livelihoods as well as save lives might require a broader set of indicators For example, in a measles immunisation campaign, the immunisation coverage within a certain age group is a performance indicator, and the incidence rate of measles cases within that group is an impact indicator [ 22 ]. The Standardised Monitoring and Assessment of Relief and Transition (SMART) inter-agency initiative is an attempt to systematise the collection of impact indicators. SMART has recently stipulated that two measures (Crude Mortality Rate and nutritional status of children under five) are the most basic essential indicators for assessing the severity of population stress and for monitoring the overall effort of the humanitarian community. Those involved in the SMART initiative are currently developing standardised survey methodologies with associated reporting formats and software to address some of the challenges to effective evaluation. Part of the purpose of SMART is to enable global comparisons about the extent of humanitarian need in order to enable resources to be focused where they are most needed. A better understanding of needs based on nutrition and mortality might also enable better analysis of impact. Of course, malnutrition and mortality rates will not necessarily enable impact to be attributed to particular projects or agencies and may only be able to demonstrate the extent to which a relief system as a whole is meeting the needs of a population [ 23 ]. There is a tendency for humanitarian agencies to collect performance rather than impact indicators. This is due to various reasons, such as donor requirements that tend to favour the collection of performance indicators or the belief that impact indicators (such as mortality or morbidity) are sometimes difficult to collect. Arguably, it is easier for humanitarian agencies to monitor their own activities than to go out and monitor or assess the effect these activities have on the population they are assisting. Performance indicators may in some cases provide sufficient evidence about the likely impact of interventions and hence be used as a proxy for impact. However, when the validity of the proxy measure is unclear, there are risks in using performance indicators as a measure of success. For example, there is strong evidence that immunising children against measles has a direct effect on reducing mortality from measles; therefore, immunisation coverage can be used as a proxy for impact on mortality [ 24 , 25 ]. Other types of health interventions, such as reproductive health care, require more research on the link between the intervention and the health outcome before performance indicators can be used as a proxy for health outcomes [ 24 ]. Table 2 illustrates some commonly used indicators with regard to their strength of association to health outcomes, and the ease with which they can be monitored. Table 2 Characteristics of indicators commonly used to justify health programmes. Established validity as measure of health impact Indicator General ease of acquiring data to show health effects Highest • Crude Mortality, <5 mortality Difficult in rural/diffuse settings, easier in camps • Case fatality rate High • Nutritional status of children Easy at the clinic data level, difficult but more valid with population surveys • Disease rates • Immunisation status of children • Patient-specific mental health evaluations Logistically easy, requires skill on part of evaluator • Safety of blood supply Moderate • Food-basket evaluations Easy in camps, more difficult in more diffuse populations • Water and sanitation availability • Reduction in measles, mumps and rubella through reproductive health services Very difficult to measure even though benefits are likely to be occurring • Improved patient outcomes via referrals • Impregnated bednets distributed • Comprehensive, timely health information system Nearly impossible. These are difficult to measure, and all require a series of events to induce a health benefit • Good coordination between sectors • Knowledge & attitudes about services available • Population practices Low • People given seeds and tools, shelter, or other materials Easy to measure. Links to health are likely to be mediated via many steps. • Drainage, fly control activities or tasks • Number of clinic visits • Distance to facilities, health workers per capita • Trainings conducted, numbers trained Easy to measure. May produce no effects on health. • Change in knowledge without documented change in behaviour • Messages/curricula developed Current practice of humanitarian agencies in the health sector In the humanitarian sector, assessment of impact has most often been seen as a sub-set of evaluation. Impact is one of the Organisation for Economic Co-operation and Development / Development Assistance Committee (OECD/DAC) evaluation criteria [ 4 , 27 ]. However, current evaluation practice rarely provides sufficient time for proper impact assessments. Most evaluation reports reviewed for this paper do not go beyond making statements about the impact of interventions. Whereas the question of impact is unarguably important in evaluation practice, a detailed analysis of impact requires a different form of investigation. This may be done either through ongoing monitoring of project implementation, or as a separate research exercise (through surveys, operational research, reviews etc.). The tools developed in field epidemiology provide a set of potentially useful methods for measuring the impact of interventions. Existing approaches such as survey methods or surveillance systems, although seldom used by humanitarian agencies, can provide significant insight about the impact of humanitarian aid. So far, these tools remain mainly used in early warning systems or needs assessment [ 21 ]. The two most common approaches are survey methods and surveillance systems. Survey methods Most surveys are an attempt to actively go out and survey a representative sample of the population, although they may take different forms. WHO and others have produced manuals to guide health workers to conduct specific kinds of surveys, with nutritional anthropometry and Expanded Program on Childhood Immunizations (EPI) coverage methodologies being among the most succinctly described. Aid workers often do not have sufficient skills to take a valid sample and analyse the results of a survey. This is why many initiatives to improve the quality of relief programmes have emphasised the importance of training relief workers in survey methodologies. Some organisational headquarters and some groups such as Epicentre have specialists who can be deployed to assist with the conducting of surveys. Uncertainty about population figures creates particular difficulties in constructing sampling frames for use in surveys. Census data may be many years old while the crisis may have had a dramatic impact on demographics and population numbers due to migration and high mortality. Although cluster surveys are a compromise measure, in many situations (especially in conflict situations or where terrain is very difficult) it may prove difficult to gain access to the 30 clusters proscribed. Nomadic groups may also prove difficult to sample [ 23 , 28 , 29 ]. Despite these difficulties, recent experiences have shown that surveillance methods can be successfully carried out in volatile environments [ 26 ]. Surveillance Systems Surveillance is the systematic collection of information over time for decision making. Surveillance systems are often part of general monitoring systems and have been used for analysing impact in both health and nutrition programmes. Aid agencies sometimes evaluate health programs by establishing a surveillance system at the beginning of a funding cycle and contrasting the rate of health events at the beginning and the end. This is valid if either: a) all of the events of interest are captured by the surveillance network, or b) the data from within the system are representative of the health conditions of the entire population and remain consistently so over the course of the project. Neither of these conditions is commonly met for clinic-based surveillance systems in rural and urban areas, although both of these conditions are often met in well-defined settings like refugee camps. If the majority of a population does not have access to formal health care then a clinic or hospital-based surveillance system will be able to tell very little about the health conditions of the broader population. Of course, not all surveillance systems are linked to utilisation of formal health services. Sentinel site surveillance systems for nutrition monitoring, for instance, involve the monitoring of purposively selected communities in order to detect changes in context, programme and outcome variables. Surveillance systems can be less costly than surveys and may reveal more in-depth information on the causes of malnutrition. Problems with the lack of skills A recurrent problem with the use of epidemiological methods such as surveys and monitoring is the lack of appropriate skills for conducting good quality assessments. Reviewers from the Centers for Disease Control and Prevention (CDC) evaluated the monitoring of projects and the measurement of nutritional status and mortality in Somalia from the period 1991–93 [ 30 ]. They developed a set of criteria for evaluating different kinds of information (surveillance and surveys) and systematically reviewed available reports. They found that the range of methodologies employed and outcomes measured were so variable and of such poor quality that they prevented widespread comparisons, and that, regardless of consistency, much of the data were simply not credible due to poor collection methods. Spiegel et al. from CDC reviewed 125 nutritional surveys conducted in Ethiopia in 1999 and 2000 during a time of famine but relative peace and stability [ 31 ]. The surveys were carried out by 14 organisations with a wide range of survey expertise. Only 67 of the 125 surveys attempted to conduct a sample that represented the population served. Only 9 of those 67 surveys assigned clusters to the population in a manner that was proportional to the sub-units of the population and only 6 of those possessed the minimum number of clusters (30) and children (900) suggested by most nutritional manuals. Spiegel concluded that non-governmental organisation (NGO) workers were woefully unprepared to conduct quantitative assessments of this kind, and that most of the surveys were of such poor quality as to be unhelpful toward making sound relief policy decisions [ 31 ]. The measurement of anthropometry is relatively standardised compared to many other health outcomes such as mortality and mental health status. For example, a mortality survey in Kabare Health Zone in the Eastern Democratic Republic of Congo (DRC) in 1999 was conducted simultaneously with an EPI coverage survey and only included households with a child under 5 years of age. The resulting estimate of crude mortality (1.9 per 1000 per month) was far lower than in a later repeat survey (finding 2.7 per 1000 per month) and, in fact, the initial survey missed most of the excess mortality [ 32 ]. For some project objectives, such as the prevention of HIV transmission, there is not even a widely agreed upon outcome to be measured that serves as a proxy for HIV incidence. The difficulty of assessing outcomes such as mortality is a principal reason for the use of process indicators in place of health outcomes. Thus, without improved staff skills and capacity and a significant change in attitude among donors, it is likely that humanitarian agencies will continue to rely heavily on process indicators and not be expected to prove that programmes influenced the health of the targeted beneficiaries. Review of 15 reports of health-related programmes All final reports of health-related programmes funded by the US Department of State, Bureau of Population, Migration, and Refugees (BPRM) and submitted in 2003 were reviewed for the Humanitarian Policy Group study. Proposals that contained objectives of health-related activities (e.g. shelter provision, food transport) but that did not specifically say they would influence health status were excluded. The remaining 15 final reports were evaluated against the following five criteria: • Was there a health-related objective? • Was the baseline rate measured or a comparison group identified? • Was the health-related outcome measured and reported? • Was the societal level of the evaluation appropriate given the intervention? • Were there any major issues supporting or raising concerns about the reported outcome data? The societal level of a health project and evaluation was categorised as being either on the patient level, the household level or the community level. The expectation was that programmes that intervened on a specific level should be evaluated on that level. For instance, a curative health programme might have benefits at the individual level but it may not be possible to evaluate its impact at a wider level. Six of the 15 reports did not attempt to measure or report any health-related rates or status. Proposals corresponding to five of these six reports only contained process indicators as the initial objectives, and thus the lack of documented health benefits was assured before the projects began. An additional three of the 15 reports contained health data-based objectives but did not present any health-status data, instead reporting process indicators such as the numbers of clinics supported, consultations given, or tons of food distributed. Only four of 15 final reports could demonstrate a health benefit, and three others were likely to have produced a population-based benefit although this was not documented. These four were the only projects to measure baseline rates. Nine of the 15 did not have objectives and measures that matched to societal level. The results of this analysis confirm the general conclusion reached at the July 2002 SMART Monitoring and Evaluation Workshop, that while NGO's and agencies often want to monitor health outcomes, they usually monitor process indicators. Problems with process indicators seen in the BPRM review include: • The cited activity may be related to the health outcome, but the significance of this effort depends on the activities being done well and in sufficient numbers (e.g. Eritrea and Sierra Leone wanted to reduce mortality and reported numbers of clinic-based activities) • The health-related objective is only distantly related to the health outcome (e.g. a programme in Uganda wanted to induce "food self-sufficiency" but reported tons of food distributed) • In some cases, the link between the process indicator and the outcome was simply implausible (e.g. a Balkans programme wanted to reduce dependency on aid of chronically "Extremely Vulnerable Individuals" and reported doing this for some by distributing school books) Interestingly, a mental health programme in Guinea, with perhaps the most difficult-to-measure outcomes, had the most rigorous documentation, which included pre-intervention and post-intervention patient evaluations and the use of non-patient controls. Representatives for the other three programmes which documented impacts felt that very little of the project budget (perhaps <2%) was spent on documenting the impacts. Over 20 NGOs were providing general health services in the eastern DRC in 2000 and 2001 with funding from either Office of U.S. Foreign Disaster Assistance (OFDA) or European Commission's Humanitarian Aid Office. According to OFDA, only two of those agencies could show health benefits associated with their programmes [ 33 ]. This seemed plausible at the time given the violent and chaotic circumstances within which the NGOs operated. The short funding cycles and volatile nature of emergencies often prohibit a systematic and rigorous evaluation of either the impact or the monitoring of multiple agencies in the same setting. Wider level of impact analysis There is an increasing interest in impact analysis at higher levels and a 'system-wide approach to performance' [ 7 ]. Several initiatives and mechanisms, from donors and humanitarian agencies, are attempting to move beyond the project level and consider sectoral, multi-sectoral, or system-wide impacts. For example, one of the objectives of the SMART initiative is to enable judgements about the overall impact of the humanitarian effort. Another example is the Inter-agency Health Evaluations in Humanitarian Crises Initiative that proposes to establish inter-agency health programme reviews in order to find new ways of looking at health programme performance and its impact on the health of affected populations [ 34 ]. A number of issues must be considered with wider levels of impact assessment. First, there is no reason to think that the constraints encountered when measuring the impact of particular interventions are erased when looking at a wider level. A particular difficulty is that of aggregation. The wider the level is, the more aggregation impact data require. Clearly, a donor or an aid agency looking at the overall effectiveness of its aid over a number of years needs far more aggregation than the evaluation of the impact of a single project conducted by a single agency. Finally, wider levels of impact assessment also generate new problems such as that of responsibility. Who is responsible for the collective impact of a number of individual humanitarian projects? Projects may have a positive impact taken individually, but the overall humanitarian effort may be insufficient compared with the level of needs. Who will account for the overall success or failure (if that is in fact possible to measure) of the humanitarian enterprise? This is a typical question that came out of the system-wide Rwanda evaluation. There is also a need for consensus in the relief community about the fundamental objective of health programmes. Conclusion Despite existing efforts to improve the quality, accountability and performance of humanitarian aid in the health sector and more broadly, this paper has shown that there is limited knowledge about the health impact of humanitarian aid. The epidemiological tools potentially useful for analysing the impact of aid programmes are seldom used. As a result, humanitarian efforts rest on a limited evidence base. This is in large part due to lack of epidemiological skills found within NGOs working on the ground. Addressing this skills deficit will be essential if the rigour of routine assessments is to be improved. In the current practice, the health impact of programmes is too often assumed rather than demonstrated. This is largely due to the use of performance or process indicators as proxy for impact, without the necessary evidence that the intervention is robustly linked with a health outcome. There needs to be a consensus regarding which types of intervention (measles immunisation, assuring that people have enough food and water) are linked to good health and the levels of service that are sufficient, in order to document that aid money is well spent. Further research on the links between particular interventions and health outcomes is required to build up this evidence base. Efforts to document project impact should be woven into monitoring and surveillance activities, not only to reduce costs, but as a tool to improve program quality. The absence of systematic monitoring and surveillance in the humanitarian sector is a serious obstacle to assessing the impact of humanitarian aid. All too often an assessment of the impact is considered as a separate activity that takes place at the end of a project. The question of impact must be included throughout the project cycle, from the formulation of objectives to the final evaluation. For health impacts to be more widely documented there needs to be adequately trained, experienced, and motivated staff present at the design and evaluation phases of projects. Part of the solution is increased funding for training and retaining staff who can act as a resource, but there must be also an increased collaboration between donors and relief workers that develops a culture rewarding the documentation of programme failures as well as successes as learning opportunities. While initiatives such as SMART provide a potentially useful platform for analysing the global impact of humanitarian aid, there is a risk that the efforts will focus exclusively on technical discussions regardless of the wider political dimension of humanitarian aid. Some agencies also fear that these mechanisms will reinforce the donor control over humanitarian agencies, instead of solely aiming to increase the quality and performance of humanitarian aid. Nonetheless, increasing accountability in all sectors of international aid and increasing expectations for the wellbeing of the world's downtrodden will eventually demand consistent and widespread documentation of humanitarian benefits. Existing epidemiological techniques can adequately do so, if only they were employed. The challenge will be to make this documentation occur through positive self-improvement motives rather than as a reactive response to criticism. Authors contributions LR, a field epidemiologist, wrote the theoretical parts of the paper and conducted the review of reports of 15 health-related programmes; CAH, a researcher and humanitarian worker, wrote the sections that relate to humanitarian assistance and the practice of impact assessment. This paper is based on a research project into the impact of humanitarian aid carried out by the Humanitarian Policy Group at the Overseas Development Institute. Competing interests The author(s) declare that they have no competing interests. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544941.xml |
516034 | Evaluation of safety and efficacy of gefitinib ('iressa', zd1839) as monotherapy in a series of Chinese patients with advanced non-small-cell lung cancer: experience from a compassionate-use programme | Background The gefitinib compassionate-use programme has enabled >39,000 patients worldwide to receive gefitinib ('Iressa', ZD1839) treatment. This paper reports the outcome of gefitinib treatment in Chinese patients who enrolled into the 'Iressa' Expanded Access Programme (EAP) at the Peking Union Medical College Hospital. Methods Thirty-one patients with advanced or metastatic non-small-cell lung cancer (NSCLC) that had progressed after prior systemic chemotherapy were eligible to receive oral gefitinib 250 mg/day as part of the EAP. Treatment was continued until disease progression or unacceptable toxicity occurred. The impact of treatment on disease-related symptoms and quality of life (QoL) was evaluated with the Chinese versions of European Organization for Research and Treatment of Cancer Quality of Life Questionnaires (EORTC QLQ-C30 and QLQ-LC13). Results Gefitinib was well tolerated. Adverse events (AEs) were generally mild (grade1 and 2) and reversible. The most frequent AEs were acneform rash and diarrhoea. Only one patient withdrew from the study due to a drug-related AE. The objective tumour response rate was 35.5% (95% confidence interval [CI]: 18.6–52.3); median progression-free survival was 5.5 months (95% CI, 1.6 to 9.4); median overall survival was 11.5 months (95% CI, 5.6 to 17.3). The QoL response rates for five functioning scales and global QoL varied from 56–88%. The main symptom response rates varied from 44–84%. QoL and symptom response were correlated with objective tumour response. Conclusion Gefitinib demonstrated safety and efficacy as monotherapy in this series of Chinese patients with advanced NSCLC and was also associated with remarkable symptom relief and improvement in QoL. Although clinical trials are needed to confirm these positive findings, the data suggest that treatment with gefitinib may be beneficial for some Chinese patients who do not respond to chemotherapy and have poor prognosis. | Background Platinum-based combination chemotherapy is the standard first-line treatment for patients with advanced non-small-cell lung cancer (NSCLC). Although meta-analysis of clinical trials proved that combination chemotherapy for advanced NSCLC is superior to best supportive care [ 1 ] and new cytotoxic agents have been developed over the past decade, the 5-year survival rate for these patients remains <1% [ 2 ]. The pace of progress is too slow and new therapeutic approaches are required. Based on our increasing understanding of the biology of NSCLC, targeted therapy is providing some promising new agents. Targeted therapy is usually aimed at a key protein implicated in tumour cell proliferation, survival, invasion or resistance to conventional treatments but spares the normal cells, thereby producing less toxicity than conventional therapies [ 3 ]. Many solid tumours express or highly express the epidermal growth factor receptor (EGFR). In lung cancer, deregulation of EGFR is seen mainly in NSCLC: EGFR is highly expressed in NSCLC at levels varying from 32–80% [ 4 - 6 ]. Abnormal signal transduction arising from the receptor is implicated in the growth and proliferation of these tumours. Thus, disruption of the EGFR signal transduction pathway is an ideal target for anticancer therapy. Gefitinib is an orally active small molecule EGFR tyrosine kinase inhibitor (EGFR-TKI). The tolerability and efficacy of gefitinib have been explored extensively in the West and Japan. In a Phase I trial, Nakagawa et al compared the safety profile, pharmacokinetic parameters and antitumour activity of gefitinib in patients with solid malignant tumours in Japan, the USA and Europe, and no significant difference was found for ethnicity [ 7 ]. However, severe interstitial pneumonia related to gefitinib therapy was more common in the Japanese population [ 8 , 9 ]. In IDEAL ('Iressa' Dose Evaluation in Advanced Lung cancer) 1, which included 102 Japanese patients with advanced NSCLC, the response rate of Japanese patients taking gefitinib was significantly higher than that of non-Japanese patients but bias of baseline factors between strata, not ethnicity, was thought to account for the difference [ 10 ]. But Owing to the paucity of current data on gefitinib between defferent races, the possibility of differences in the toxicity and efficacy of gefitinib for ethnicity cannot be excluded. At present, there are few data regarding the tolerability and efficacy of gefitinib in Chinese patients. Patients with advanced NSCLC who had no alternative therapeutic options have been able to receive gefitinib treatment in a worldwide compassionate-use programme. Here we report the outcome of treatment with oral gefitinib 250 mg/day in Chinese patients with advanced NSCLC participating in the compassionate-use programme at Peking Union Medical College Hospital. Methods Patient population Patients aged ≥18 years with histologically or cytologically confirmed advanced or metastatic NSCLC were eligible for enrolment into the compassionate-use programme after providing written, informed consent. They were required to have failed prior chemotherapy, and had no other treatment options available. Patients who had received chemotherapy were included if the treatment ended >28 days prior to the study. Other eligibility criteria included: adequate bone marrow function (white blood cell count >4 × 10 9 /L, absolute neutrophil count >1.5 × 10 9 cells/L, platelet count >100 × 10 9 /L); liver function (total bilirubin <34 μmol/L, aspartate aminotransferase [AST] <40 IU/L, alkanine aminotransferase [ALT] <41 IU/L); renal function (serum creatinine <150 μmol/L, blood urea nitrogen <10.7 mmol/L); PaO 2 >60 mmHg. Exclusion criteria included a serious pre-existing medical condition (eg uncontrolled infection, interstitial pneumonia or pulmonary fibrosis, severe chronic diarrhoea), pregnancy and lactation. Drug administration One oral gefitinib tablet (250 mg) was taken at about the same time each day. Gefitinib was administered every day without interruption unless disease progression or unacceptable toxicity occurred. Evaluation before and during treatment In this study, 28 days of treatment was defined as one treatment cycle. Baseline evaluation included medical history and physical examination, electrocardiogram, chest X-ray, thorax computed-tomography scan and ultrasonography of the upper abdomen. Laboratory investigations included complete blood counts, urinalysis, and renal and liver function tests. Performance status (PS) was evaluated according to Eastern Cooperative Oncology Group (ECOG) criteria. Brain magnetic resonance imaging and radionuclide bone scans were only performed if metastatic disease was suspected according to the clinical manifestations of each patient. Regarding interstitial lung disease, we paid attention to clinical respiratory symptoms (e.g. dyspnea, cough) and radiographic findings of patients, and monitored PaO 2 during therapy. Patients were evaluated after the first and third cycles of therapy, then every three cycles. Tumour response was evaluated according to Response Evaluation Criteria In Solid Tumors [ 11 ]. In line with the criteria, all patients included in the study were assessed for response to treatment. Each patient was assigned to one of the following categories: complete response (CR); partial response (PR); stable disease (SD); progressive disease (PD); early death from lung cancer; early death from toxicity; early death because of other disease; unknown (not assessable or insufficient data). All AEs were recorded and graded according to National Cancer Institute Common Toxicity Criteria version 2.0. Assessment of QoL QoL was assessed using the Chinese version of the European Organization for Research and Treatment of Cancer (EORTC) core questionnaire, the Quality of Life Questionnaire (QLQ)-C30 (version 3.0) and the supplemental lung-cancer-specific module QLQ-LC13 [ 12 ]. Following the scoring procedure recommended by the EORTC, scores were converted into linear transformation ranging from 0–100. For the functional and global health status/QoL, higher scales represent better functioning. For symptoms, a higher score represents worse symptoms [ 13 ]. QoL and disease-related symptoms were assessed before the start of the therapy and then at the end of each cycle in the first 3 months of treatment. Statistical analysis Logistical regression test models were used to identify baseline factors (gender, PS, histology, TNM stage and prior chemotherapy) that might independently predict tumour response. Median progression-free survival (PFS) and overall survival (OS) were calculated using the Kaplan-Meier method and a log-rank test was used to detect differences OS between strata. Changes in symptoms and QoL were assessed in two different ways. Firstly, the mean scores of QoL and disease-related symptoms at baseline were compared with those at the end of the second cycle of treatment using a paired-sample t test. Secondly, the response rates of symptoms and QoL were calculated. For the assessment of symptoms and QoL, the classification system suggested by Stephens et al was used [ 14 ]. The results were presented as response (ie improvement, control or prevention), no response and nonevaluable cases in the form of percentages. The classification system [ 15 , 16 ] has been used successfully to evaluate QoL and symptoms in patients with lung cancer. Response rates for QoL and disease-related symptoms were compared between patients with and without objective tumour response using Pearson's χ 2 test or Fisher's exact test. Results Patient characteristics Thirty-one eligible patients were enrolled into the EAP at Peking Union Medical College Hospital between October 2002 and October 2003. Patient characteristics are listed in Table 1 . The patient series included 18 (58%) men and 13 (42%) women between 28 and 85 years of age (median age 64). All patients had received at least one platinum-based regimen and most had received more than two regimens (different combinations including platinum, taxane, docetaxel and gemcitabine). Ten (32.3%) patients had squamous-cell carcinoma (SCC) and 20 (64.5%) had adenocarcinoma. Most patients (61.3%) had an ECOG PS of 0–1, 25.8% had a PS of 2 and 12.9% had a PS of 3. TNM stages were as follows: stage IIIa, 1 patient; stage IIIb, 4 patients; stage IV, 26 patients. Toxicity All patients were assessed for toxicity. Twenty-three (74%) patients had at least one AE. Treatment-related toxicities are listed in Table 2 . Almost all AEs (but for one grade 3 acneform rash) were mild (grade 1 or 2). The most frequently reported AEs during treatment were acneform rash (67.7%) and diarrhoea (35.5%). Four (12.9%) patients had grade 1 or 2 stomatitis. Other AEs included nausea (6.5%), vomiting (3.2%), increased ALT (3.2%) and increased AST (3.2%). The majority of AEs were transient and reversible. Only one patient withdrew from treatment due to an AE (on the 50th day of treatment with grade 3 skin rash and exacerbation of dysphagia). This patient had SCC, a PS of 3 and mediastinal lymph node metastases before treatment. The patient suffered moderate haemoptysis prior to gefitinib therapy, which reduced to a mild severity during therapy. Fifteen days after withdrawal from treatment, the patient died of massive haemoptysis. Patient response The assessment of patient response is listed in Table 3 . Of the 31 patients, 1 (3.2%) achieved CR and 10 (32.3%) achieved PR with an overall objective response rate of 35.5% (95% confidence interval [CI]: 18.6–52.3). SD was documented in 7 (22.6%) patients and the overall DCR (CR+PR+SD) was 58.1% (95% CI: 49.2–67.0). The response rate of adenocarcinoma was significantly higher than that of squamose carcinoma [50% (10/20) vs. 10% (1/10)]. Multivariate logistic analysis also showed that the odds of objective response were 10 times higher (odds ratio 10; 95% CI: 1.0 to 93.4; p = 0.028) for patients with adenocarcinoma than for patients with other tumour histologies. Of the 31 patients who received treatment, 2 died due to PD before efficacy assessment. Both these patients had SCC and a PS of 3, and no improvement in either patient had been observed during gefitinib treatment. Median PFS was 5.5 months (95% CI, 1.6 to 9.4); median OS was 11.5 months (95% CI, 5.6 to 17.3). The tumor response was associated with improved OS. Median OS of patients with response was significantly higher than that of those with no response (log-rank test p = 0.0058) (17 vs. 4.4 months). QoL and symptom improvement At baseline, 25 (81% [25/31]) patients returned QoL questionnaires. All 25 patients returned the questionnaires at the first cycle of therapy, 23 (93% [23/25] of those patients returning questionnaires and remaining alive) at the second cycle of therapy, and 18 (74% [18/24] of those patinets returning questionnaires and remaining alive) at the third cycle. The most frequently reported general symptoms were fatigue (100%) and appetite loss (68%), while dyspnoea (100%) and coughing (84%) were the most frequently reported respiratory symptoms. The changes in the mean scores for symptom scales, functioning and global QoL scales are presented in Table 4 . A statistically significant increase in mean score was observed for physical functioning, role functioning, emotional functioning, social functioning and global QoL after 8 weeks of treatment. There was also a trend towards a higher score (p = 0.08) for cognitive functioning. Mean scores for two general symptoms (fatigue and appetite loss) and disease-related symptoms (dyspnoea, coughing, pain in chest, pain in arm or shoulder and pain in other parts) decreased statistically. Patient responses for functioning, global QoL and main symptoms are shown in figure 1 . A >50% response rate was observed in all five functioning criteria and global QoL. With regard to QoL, the highest response rate was observed for emotional functioning (88%), followed by cognitive (72%), physical (68%), social (64%) and role functioning (60%), and global QoL (56%). Haemoptysis had the highest response rate (84%), the rate of improvement (63%) being just lower than that of appetite loss among the five symptoms. The response rates for dyspnoea and coughing were 56% and 68%, respectively. For fatigue and appetite, the response rates were 44% and 80%, respectively. Differences in response rates for symptoms and QoL between patients with or without objective tumour response are shown in Table 5 . Response rates of physical functioning, role functioning, social functioning and global QoL were significantly higher in patients with objective tumour response than in those without. Response rates for emotional functioning and cognitive functioning were also higher in patients with tumour response, though the difference was not statistically significant. Differences were also observed in response rates for symptoms. For dyspnoea, coughing, appetite loss and fatigue, the response rates in the objective responders were higher than those in nonresponders. No significant association was found between haemoptysis and objective tumour response. Discussion A daily oral 250 mg dose of gefitinib was well tolerated in this series of Chinese patients with advanced NSCLC, who received gefitinib as part of a compassionate-use programme. AEs were mild and reversible and different to those associated with conventional chemotherapy, such as neutropenia, thrombocytopenia and neuropathy. No haematological or neural toxicity was observed in the study. As with prior clinical trials evaluating the toxicity of gefitinib, the most frequently reported AEs were acneform rash and diarrhoea (grade 1 or 2). Four patients developed grade 1 or 2 stomatitis during treatment, which has previously been associated with gefitinib treatment in 7.8% of patients in IDEAL 1 but was not reported in IDEAL 2 [ 17 ]. Severe acute interstitial pneumonia is the most serious AE that has been linked with gefitinib therapy. Inoue et al recently reported that 4 of 18 patients receiving gefitinib in their clinic developed severe acute interstitial pneumonia [ 8 ]. The authors stated that 291 of 17,500 patients (1.7%) treated with gefitinib in Japan had developed suspected interstitial pneumonia or acute lung injury. In contrast to the higher incidence of interstitial pneumonia, this severe AE was lower in the rest of the world [ 9 ]. The worldwide frequency of interstitial lung disease to date in ~92,750 patients who have received gefitinib is <1.0% [ 18 ]. The ethnicity between Chinese and Japanese people is probably similar, but we saw no evidence of interstitial pneumonia in the series of patients. Although no patients developed acute interstitial pneumonia, the possibility of drug-related interstitial lung disease could not be excluded because of the small number of patients in our series. The response rate (35.5%) was higher than that of recently presented Phase II trials (18.4% and 11.8% in patients receiving gefitinib 250 mg/day in IDEAL 1 and 2, respectively). The DCRs were 54.4% and 42% for the 250 mg/day dose in IDEAL 1 and 2, respectively, in contrast to 58.1% in our study. Unlike IDEAL 2, which was performed in the USA, IDEAL 1 was a global trial that recruited patients from 43 centres across Europe, Australia, South Africa and Japan and included a total of 102 Japanese patients with advanced NSCLC. The response rate for Japanese patients was higher than that of non-Japanese patients (27.5% versus 10.45%; odds ratio 3.72; p = 0.0023). However, the difference in response rate for ethnicity was not verified using multivariate logistical regression analysis. The difference in response rate between Japanese and non-Japanese patients was attributed to bias of baseline predictive factors of patients (gender, PS and histology) [ 19 , 10 ]. QoL is an important endpoint for assessment of gefitinib treatment and has been included in some phase I and phase II trials. In these trials, the Functional Assessment of Cancer Therapy – Lung questionnaire was used to assess QoL and symptoms. Its validation and sensitivity were verified. Significant improvements in symptoms and QoL were observed in IDEAL 1 and 2. In IDEAL 1, symptom and QoL improvement with a dose of 250 mg/day were 40.3% and 23.9%, respectively, and in IDEAL 2 they were 43.1% and 34.3%, respectively. In both trials, improvements in symptoms and QoL correlated well with tumour response [ 20 ]. In our patient series, the EORTC QLQ-C30 and its supplement module QLQ-LC13 were used for assessment of QoL and symptoms. Together, they are thought to be one of the best developed instruments for lung cancer assessment [ 21 ]. The questionnaires have been translated into >20 languages and validation of the standard Chinese version has been verified [ 22 ]. In our series of patients, mean scores of four functioning scales (physical, role, emotional and social functioning) and global QoL increased significantly after 8 weeks of treatment and were associated with high response from 56–88%. Mean score of cognitive functioning also increased, from 66 at baseline to 76 at the end of second cycle with response of 72% (not statistically significant; p = 0.08). General and disease-related symptom scores increased from 44% to 84%. Importantly, improvement of symptoms and QoL correlated with objective tumour response. Patients with objective tumour response had higher rates of symptom and QoL response. The results indicate that placebo effect is unlikely to explain completely the improvement in symptoms and QoL with gefitinib, though there is no placebo control in our patient series or the EAP. Emotional functioning response was observed in all patients with objective tumour response, but this was not statistically significant when compared with nonresponders (100% versus 79%; p = 0.23). As emotional functioning contained more subjective questions, such as "Did you feel tense?", "Did you worry?", "Did you feel irritable?" and "Did you feel depressed?", the limited difference of emotional functioning response between responders and nonresponders with gefitinib cannot completely exclude the possibility of placebo effect. Thus, a placebo control group should be included in future studies of QoL and symptom improvement with gefitinib. Conclusions In conclusion, the data from this patient series show that an oral dose of gefitinib 250 mg/day is well tolerated and has significant antitumour activity in Chinese patients with advanced NSCLC who had failed previous chemotherapy and for whom no other treatment options were available. Although clinical trials are needed to confirm these positive findings, the data suggest that treatment with gefitinib may be beneficial for some Chinese patients who do not respond to chemotherapy and have poor prognosis. 'Iressa' is a trademark of the AstraZeneca group of companies Competing interests None declared. Authors' contributions XLM and LYL designed the experiments and wrote the manuscript. XTZ, SLW and MZW participated in patient's follow-up and tumour response evaluation. Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516034.xml |
544955 | The immunotherapy of Alzheimer's disease | Only a small percentage of patients with Alzheimer's disease benefit from current drug therapy and for only a relatively short time. This is not surprising as the goal of these drugs is to enhance existing cerebral function in Alzheimer patients and not to block the progression of cognitive decline. In contrast, immunotherapy is directed at clearing the neurotoxic amyloid beta peptide from the brain that directly or indirectly leads to cognitive decline in patients with Alzheimer's disease. The single trial of active immunization with the amyloid beta peptide provided suggestive evidence of a reduction in cerebral amyloid plaques and of stabilization in cognitive function of half the patients who developed good antibody responses to the amyloid beta peptide. However, 6% of actively immunized Alzheimer patients developed sterile meningoencephalitis that forced the cessation of the clinical trial. Passive immunotherapy in animal models of Alzheimer's disease has provided similar benefits comparable to those seen with active immunotherapy and has the potential of being effective in the half of Alzheimer's disease patients who do not make a significant anti-amyloid beta peptide antibody response and without inducing T-cell-mediated encephalitis. Published studies of 5 patients with sporadic Alzheimer disease treated with intravenous immunoglobulin containing anti-amyloid beta peptide antibodies showed that amyloid beta peptide was mobilized from the brain and cognitive decline was interrupted. Further studies of passive immunotherapy are urgently required to confirm these observations. | Introduction The headline "Minimal benefit is seen in drugs for Alzheimer's disease" summarized an article in the New York Times concerning drug therapy of Alzheimer's disease [ 1 ]. The inescapable conclusion was that present drug therapy benefits only a very small percentage of the 4.5 million Americans with Alzheimer's disease patients. This is not surprising as the goal of today's drugs is to enhance existing cerebral function in Alzheimer patients and not to attack the basic causes of their progressive cognitive decline. In contrast to today's drug therapy for Alzheimer's disease, immunotherapy is directed at the neurotoxic amyloid beta peptide that directly or indirectly leads to cognitive decline in patients with Alzheimer's disease. Most investigators believe that the accumulation of amyloid beta peptide in the brain of elderly adults is not only a hallmark of Alzheimer's disease but is the primary cause of cognitive decline [ 2 ]. There is secure evidence from animal studies and preliminary evidence from patients with sporadic Alzheimer disease that immunotherapy can block the accumulation of the neurotoxic amyloid beta peptide in the brain and cognitive decline in patients with Alzheimer's disease. Reducing the level of amyloid beta peptide in the brain can be achieved by decreasing the production of the amyloid beta peptide and/or by increasing its clearance from the brain. Decreased production of amyloid beta peptide would follow the inhibition of the beta and gamma secretases, enzymes that cleave the 40/42 amino acid amyloid beta peptides from the 770 amino acid transmembrane amyloid precursor protein (APP) or the augmentation of the activity of the alpha secretase that lowers amyloid beta peptide production [ 3 , 4 ]. Although inhibitors of beta or gamma secretases are known to suppress amyloid beta peptide generation from APP in cultured cells, the available inhibitors are too toxic for clinical use. Furthermore, it is likely that removal of cerebral amyloid beta peptide will be necessary to reverse the accumulated amyloid beta peptide that is already present at the time when Alzheimer's disease is recognized. Induction or infusion of anti-amyloid beta peptide antibodies prevents the accumulation of cerebral amyloid beta peptide from the brain of mouse models of familial Alzheimer's disease and may work in humans with sporadic Alzheimer's disease [ 6 , 7 ]. Even more encouraging are the reports that immunotherapy reverses cognitive decline in the mouse models of familial Alzheimer's disease and may act similarly in elderly humans with sporadic Alzheimer's disease [ 8 - 10 ]. This article will review and place in historical perspective the development of immunotherapy, in general, and its application to experimental murine models of familial Alzheimer's disease and to a small number of elderly patients with sporadic Alzheimer's disease so far studied. Immunotherapy Immunity was first recognized by the fact that people who recovered from an infection were often protected from re-infection. Thus, Thucydides wrote of the Athens epidemic of plaque that occurred in 430 BC "the same man was never attacked twice – at least fatally" [ 11 ]. Twelve hundred years later, Rhazes wrote that smallpox struck children, rarely late in life, and generally only once [ 12 ]. The fact that clinical disease conveyed immunity raised the question whether the induction of a mild smallpox infection might protect individuals from this often fatal disease. Variolization, the inoculation with pustular material from a smallpox patient, was first reported in the 17 th century in China to protect from virulent smallpox infection [ 13 ] and brought to wider attention in 1714 by Timoni's letter to the Royal Society of London reporting on variolization in Turkey [ 14 ]. Despite the clinical efficacy of variolation, the dose of a living, virulent, pathogen was difficult to control, leading, at times, to serious morbidity or mortality. The next advance in immunotherapy followed the realization that exposure to a related, less virulent, infectious organism such as cowpox, protected individuals from the virulent pathogen, smallpox. In England during the second half of the 18 th century the link between the fine complexion (no scars of smallpox) of milkmaids and their exposure to cowpox (vaccinia) was recognized [ 15 ]. But it was Edward Jenner's publication in 1798 that proved that inoculation with the coxpox, "vaccination", protected susceptible individuals from small pox [ 16 ]. In the years since Jenner's publication, many vaccines have been developed to protect humans and animals against infectious diseases [ 17 ]. Most vaccine stimulate an immune response against the infectious microorganisms that cause disease although some vaccines stimulate an immune response to toxins they produce [ 18 ]. The primary use of vaccines had been to prevent human disease. At the end of the 19 th century, von Behring and Kitasato discovered that immunity to diphtheria and tetanus was conveyed by serum antibodies against the exotoxins of these bacteria [ 19 ]. These investigators, knowing that serum antibodies conveyed immunity, showed that serum anti-diphtheria antibodies that had been induced in one animal could be transferred and, thereby, could cure another animal showing symptoms of the disease. Thus, was born the concept of immunotherapy. The clinical impact of this finding would be great as 50,000 German children died each year of diphtheria at the end of the 19 th century. The first successful application of immunotherapy to humans occurred in 1891 when a child suffering from diphtheria was cured by an infusion of horse serum containing anti-diphtheria antibodies. The horse serum containing anti-toxin antibodies could be lifesaving although many patients, receiving large doses of xenogeneic proteins, developed serum sickness resulting from the patients' immune response to the foreign serum. During the first part of the 20 th century passive immunotherapy of other infectious diseases including pneumococal pneumonia entered clinical practice but was replaced when antibiotics became available during the second third of the 20 th century. Today, passive immunotherapy is attracting renewed interest as a new means to treat neurodegenerative, infectious, autoimmune, and neoplastic diseases [ 7 , 20 , 21 ]. Passive immunotherapy has, thanks largely to remarkable progress in monoclonal antibody and recombinant DNA technologies, become one of the hottest fields of therapeutics. Humanized or human monoclonal antibodies have increased the efficacy of passive immunotherapy and eliminated serum sickness following injection of xenogeneic serum. Now passive immunotherapy has been developed for patients with chronic diseases including atherosclerotic, neoplastic, and neurodegenerative diseases. Active Immunotherapy of Alzheimer's Disease Immunotherapy of Alzheimer's disease followed the development of a mouse model of familial Alzheimer's disease. In 1995, an APP-transgenic mouse expressing a mutant, human APP gene isolated from a Swedish family with inherited Alzheimer's disease was developed [ 22 ]. These mice develop cerebral diffuse deposits of amyloid peptide and amyloid plaques by middle age and were, therefore, a useful model of familial Alzheimer's disease. Four years later, Schenk and his colleagues reported that repeated immunization of such mice with amyloid beta peptide prevented or reversed accumulation of amyloid deposits in the brain of these mice [ 6 ]. Despite the dramatic effects of active immunotherapy on cerebral histopathology in APP-transgenic mice, the key question – did immunotherapy prevent cognitive decline – remained to be answered. Eighteen months later it was reported that the cognitive decline seen in the APP-transgenic mice was blocked by active immunization [ 23 - 25 ]. No toxicity was observed following active immunization of APP-transgenic mice, despite some concern that the administration of the neurotoxic amyloid beta peptide might cause untoward effects [ 26 ]. These observations offered great promise for the treatment of patients with familial Alzheimer's disease and, perhaps, elderly patients with sporadic Alzheimer's disease. A clinical trial of active immunization of elderly patients with sporadic Alzheimer's disease was organized and initiated in 2001. However, there were reasons to question whether the benefits seen in a middle-aged mice model of familial Alzheimer's disease could be directly extrapolated to elderly patients with sporadic Alzheimer's disease. First of all, aging is associated with a decreasing immune response [ 27 ]. Thus, active immunization with influenza or tetanus vaccines induces less protective immunity in old than young persons or experimental animals. It was reported that a large percentage of old mice and elderly humans following active immunization with amyloid beta peptide did not generate a robust anti-amyloid beta peptide antibody response [ 9 , 28 ]. Secondly, active immunization of elderly humans and old mice stimulates an increase in autoimmune responses despite the lower immune response to the foreign antigen [ 29 ]. The clinical trial of active immunization of patients with Alzheimer's disease with amyloid beta peptide, started in 2001, was stopped a year later after 4 patients in the actively immunized group developed sterile meningoencephalitis [ 30 ]. Approximately 6% of the 298 actively immunized Alzheimer's disease patients eventually developed sterile encephalitis. One patient with sporadic Alzheimer's disease died 12 months after the last immunization with amyloid beta peptide from a pulmonary embolus. Post-mortem examination of the brain in this patient revealed CD4+ T cells in a perivascular distribution [ 31 ]. While neither the function nor the specificity of the T cells infiltrating the brain was determined, it is possible that the age-associated tendency to generate autoimmune reactions led these patients to generate autoreactive CD4+T cells that entered the brain and contributed to encephalitis. No comprehensive report of the clinical study of active immunization has yet been published but oral presentations and published results from a subset of actively immunized Alzheimer's disease patients have provided some information [ 9 ]. It appears that: (i) all patients with sterile encephalitis had been actively immunized with amyloid beta peptide; (ii) there was no correlation between the level of serum anti-amyloid beta peptide antibodies and risk of sterile encephalitis; (iii) certain patients with sterile encephalitis had no detectable anti-amyloid beta peptide antibodies in serum; (iv) one-half of the elderly patients with sporadic Alzheimer's disease who were immunized with amyloid beta peptide did not generate significant titers of anti-amyloid beta peptide antibodies; and finally, (v) in a subset of patients with sporadic Alzheimer's disease, those patients who generated significant levels of serum anti-amyloid beta peptide antibodies had little or no cognitive decline during the year of observation following active immunotherapy. In summary, active immunization with amyloid beta peptide in elderly patients with Alzheimer's disease appears to be less effective and more toxic than in the middle-aged APP-transgenic mouse model of familial Alzheimer's disease. These differences may reflect the greater age of the patients with sporadic Alzheimer's disease and the decreased antibody responses to vaccines and the paradoxical increase in autoimmune responses in the elderly. Passive Immunotherapy of Alzheimer's Disease Administration of anti-amyloid beta peptide antibodies would bypass immune senescence and would not be expected to lead to T cell-mediated encephalitis. Furthermore, anti-amyloid beta peptide antibodies not only dissolved aggregates of amyloid beta peptide in vitro but also inhibited aggregated amyloid beta peptide-mediated cytotoxicity in vitro [ 32 ]. In vivo, passive immunotherapy of APP-transgenic mice with anti-amyloid beta peptide antibodies prevented or reversed cerebral amyloid deposition depending on whether treatment was begun before or after cerebral amyloid deposition had occurred [ 7 ]. It has been reported that passive immunotherapy of APP-transgenic mice prevented age-associated cognitive decline on in the APP-transgenic mice after a 6 week course of treatment with anti-amyloid beta peptide antibodies even before there was any detectable decrease in cerebral amyloid plaque number [ 33 ]. Preliminary clinical studies showed that infusing a preparation of human intravenous immunoglobulin (IVIg) containing anti-amyloid beta peptide antibodies into 6 elderly patients with sporadic Alzheimer's disease showed significant cognitive improvement during the 6 months of therapy [ 10 ]. Elan announced that a phase 1 clinical study of passive immunotherapy with humanized monoclonal anti-amyloid beta peptide antibody in patients with mild to moderate Alzheimer's disease had been started at the end of 2003 . The potential benefit of anti-amyloid beta peptide antibodies in humans with sporadic Alzheimer's disease was also inferred from studies of cognitive function in a subset of actively immunized patients with Alzheimer's disease [ 9 ]. In these patients, there was a direct correlation between the level of serum anti-amyloid beta peptide antibodies and cognitive function one year after active immunization. Thus, the patients with the highest serum levels of anti-amyloid beta peptide had little or no cognitive decline while cognitive function declined markedly in the nearly 50% of patients who generated little or no detectable serum anti-amyloid beta peptide antibodies after active immunization. Passive immunotherapy of Alzheimer's disease would require repeated administration of anti-amyloid beta peptide antibodies. For this reason, human anti-amyloid beta peptide antibodies should be used to prevent an immune response to the currently available murine monoclonal immunoglobulins. Several methods are known to obtain human anti-amyloid beta peptide antibodies: (i) purification of specific antibodies from human IVIg; (ii) humanization of murine anti-amyloid beta peptide antibodies by replacing framework portions of the murine anti-amyloid beta peptide antibodies with human framework sequences using recombinant DNA technology [ 34 ]; (iii) generation of human monoclonal anti-amyloid beta peptide antibodies in vitro by human immunoglobulin phage library display techniques or in vivo by immunization of mice whose immunoglobulin loci have been replaced by human Ig genes [ 34 ]. Human IVIg, purified from human plasma, was initially developed as replacement therapy for immunodeficient patients but IVIg has also been shown to be effective therapy in patients with a variety of autoimmune diseases [ 35 ]. Recently, we and others have demonstrated that human serum and IVIg have a significant quantity of human anti-amyloid beta peptide antibodies [ 36 , 37 ]. Such polyclonal human anti-amyloid beta peptide antibody preparations inhibit amyloid beta peptide-induced neurotoxicity in vitro. There is no evidence that anti-amyloid beta peptide antibodies induce sterile encephalitis, observed following passive immunization of several strains of APP-transgenic mice or in the small number of elderly patients with Alzheimer's disease. However, it may be premature to conclude that passive immunotherapy with anti-amyloid beta peptide antibodies does not induce cerebral pathology. It should be remembered that most of the APP-transgenic mice strains tested do not develop amyloid vascular deposits (congophilic angiopathy) that occurs in elderly patients with Alzheimer's disease. It was reported that infusion of murine anti-amyloid beta peptide monoclonal antibodies specific for the N-terminal epitope of the amyloid beta peptide into a strain of APP-transgenic mice, which develop congophilic angiopathy, cerebral hemorrhage was observed [ 38 ]. This untoward effect appears to depend upon the epitope specificity of the anti-amyloid beta peptide antibodies. The Choice of Anti-amyloid Beta Peptide Antibodies for Alzheimer's Disease Pre-clinical data and inferences drawn from immunotherapy in patients with sporadic Alzheimer's disease suggest that passive immunotherapy with anti-amyloid beta peptide antibodies is preferable to active immunotherapy for the treatment of elderly patients with sporadic Alzheimer's disease. However, which anti-amyloid beta peptide antibodies would have greatest therapeutic efficacy and least risk of untoward effects for patients with Alzheimer's disease remains to be determined. However, there is general agreement that anti-amyloid beta peptide antibodies to be administered repeatedly to patients should not stimulate an antibody response to the infused immunoglobulin. There is evidence that humans not only can make immune response to therapeutic antibodies but that such immune responses compromise the action of the therapeutic antibodies [ 39 ]. To date, the only human anti-amyloid beta peptide antibodies reported to improve cognitive function in elderly patients with sporadic Alzheimer's disease are those contained in human IVIg [ 11 ]. In contrast to preparations containing polyclonal human anti-amyloid beta antibodies, several laboratories have produced humanized anti-amyloid beta peptide antibodies including the Elan preparation now in phase I trial. It has not been reported whether this humanized anti-amyloid beta peptide antibody is active in APP-transgenic mice. The therapeutic efficacy of candidate human anti-amyloid beta peptide antibodies can be compared by measuring their capacity to decrease or reverse cerebral amyloid beta peptide accumulation and cognitive decline in RAG-2-deficient, APP-transgenic mice. We have bred these mice that lack lymphocytes and are incapable of generating an immune response to the human anti-amyloid beta peptide antibodies to test antibodies considered for immunotherapy of Alzheimer's disease. If the therapeutic benefit of polyclonal human anti-amyloid beta peptide antibodies is confirmed, it remains to be determined which antibody or antibodies are responsible for the therapeutic effect. Whether a single monoclonal antibody will be effective in humans as it has been in APP-transgenic mice is not certain. In some infectious diseases, a single monoclonal antibody has not been less protective than a mixture of several monoclonal antibodies [ 40 ]. Thus, while each antibody specificity in polyclonal human anti-amyloid beta peptide antibodies may be at a lower concentration than that of a monoclonal antibody, the synergistic effect of multiple antibody specificities may have advantage. Monoclonal anti-amyloid beta peptide antibodies are known to differ in their fine specificity: isotype, affinity, as well as epitope, aggregate, and Fc specificity [ 41 ]. Whether it will be possible to choose an anti-amyloid beta peptide monoclonal antibody based on these characteristics is far from certain. It is likely that testing in immune deficient APP-transgenic mouse would be performed prior to the treatment of patients with Alzheimer's disease. Monoclonal anti-amyloid beta peptide have different specificities. There are three major epitopes on the amyloid beta peptide: (i) antibodies to the N-terminal epitope (amino acids 1–6) of the amyloid peptide bind to aggregated amyloid beta peptide in vitro as well as cerebral and vascular deposits in vivo and APP (ii) antibodies specific for the central region (amino acids 15–25) of the amyloid peptide bind to APP but not to aggregated amyloid beta peptide in vitro, amyloid plaques or vascular amyloid deposits (iii) antibodies specific for the C-terminal region have been less well studied but reported to lack a therapeutic effect in APP-transgenic mice. This may be the reason why the N-terminal-specific anti-amyloid beta peptide antibodies but not central region-specific antibodies cause cerebral hemorrhage presumably from vessels with amyloid beta deposits [ 42 ]. Anti-amyloid beta peptide antibodies that differ in epitope and Fc specificities, dissolve cerebral amyloid plaques and block cognitive decline in APP-transgenic mice [ 41 ]. Anti-amyloid beta peptide antibodies, specific for the N-terminal region of the amyloid peptide, are reported to enter the brain, bind to cerebral amyloid plaques, dissolve the plaque, and mediate Fc-mediated endocytosis followed by catabolism of the amyloid beta peptide within glial cells [ 7 ]. However, the Fc-mediated pathway is not the only route to the dissolution of amyloid plaques. Direct application of anti-amyloid beta peptide antibodies that bind to amyloid plaques but do not express the Fc region of the molecule dissolve cerebral amyloid plaques [ 43 ]. Furthermore, anti-amyloid peptide antibodies dissolve amyloid plaques in APP-transgenic mice that do not express Fc receptors. Surprisingly, treatment of APP-transgenic mice with anti-amyloid beta peptide antibodies specific for the central region of the amyloid peptide that do not stain cerebral amyloid plaques ex vivo and were not detectable within the brain also cleared cerebral amyloid peptide and plaques [ 33 ]. A novel explanation for the mechanism of action of anti-amyloid beta peptide antibodies that do not enter the brain. The "peripheral sink" hypothesis suggests that cerebral amyloid beta peptide in all its forms (monomer, oligomer, fibrils) are in equilibrium with amyloid beta peptide in the blood and, that anti-amyloid peptide antibodies, which cannot cross the blood-brain barrier, deplete cerebral amyloid beta peptide by its mobilization into the blood. It was shown that within a few hours of administering central region-specific anti-amyloid beta peptide antibodies, which do not enter the brain, the level of amyloid beta peptide in the blood increases as much as 1000 fold. Furthermore, the magnitude of the increase in total amyloid beta peptide levels in the blood following a single injection of anti-amyloid beta peptide antibody is a surrogate marker of cerebral amyloid beta peptide load [ 44 ]. Thus, it appears that similar effects – decreased cerebral amyloid load and cognitive loss can occur following treatment of APP-transgenic mice with different anti-amyloid beta peptide monoclonal antibodies by central (entry into the brain) or peripheral (entry into the blood) mechanisms. Both N-terminal-specific and central region-specific anti-amyloid beta peptide antibodies but not C-terminal-specific anti-amyloid beta peptide antibodies also bind to APP the ubiquitous transmembrane cellular protein. C-terminal-specific anti-amyloid beta peptide antibodies can distinguish between amyloid beta 1–40 and 1–42 peptide. As amyloid beta 1–42 peptide is more neurotoxic peptide and forms the nidus of cerebral amyloid plaques antibodies to this amyloid beta peptide might be the most effective antibody for passive immunotherapy as they target the most pathogenic form of the amyloid beta peptide without binding to APP or the less pathogenic amyloid beta 1–40 peptide. However, the C-terminal specific anti-amyloid beta peptide antibodies have been reported not to clear cerebral amyloid beta peptide [ 41 ]. Recently, it has been reported that a C-terminal specific anti-amyloid beta peptide antibody does clear amyloid plaques [ 45 ]. Finally, there is considerable interest in the greater neurotoxicity of soluble amyloid beta peptide oligomers than either the amyloid beta monomers or fibrils [ 46 ]. If this proves to be case, it would be important to test the therapeutic efficacy of monoclonal antibodies to amyloid beta peptide oligomers in APP-transgenic mice [ 47 ]. Conclusion Passive immunotherapy of sporadic Alzheimer's disease offers the potential of reversing the pathologic accumulation of cerebral amyloid beta peptide. To date the only preparation of human anti-amyloid beta peptide antibodies that have been reported to reverse cognitive defects in patients with sporadic Alzheimer's disease are polyclonal anti-amyloid beta peptide antibodies contained in human IVIg. Selection of human anti-amyloid beta peptide antibodies for clinical trial can be tested for therapeutic effect in vivo by their treatment of immunodeficient APP-transgenic mice. Competing Interests The authors declare that they have no competing interests. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC544955.xml |
549530 | Circulating γδ T cells in young/adult and old patients with cutaneous primary melanoma | Background In a previous study we demonstrated the existence of numerical and functional alterations of γδ T cells in healthy elderly. Recently, we analysed the involvement of γδ T lymphocytes in malignant melanoma, describing a lower frequency of circulating γδ T cells, an altered pattern of cytokine production, and an impaired in vitro expansion of these cells in primary cutaneous melanoma patients. Methods In this study we investigated the existence of numerical and functional alterations of circulating γδ T cells in young/adult and old melanoma patients, comparing the data obtained with age-matched healthy subjects. Results We demonstrated that the number of circulating γδ + T cells was significantly and similarly reduced in young/adult and old melanoma patients and in old healthy subjects in comparison with young healthy donors. The decrease was due to a reduction of Vδ2 T cells whereas the number of Vδ1 T cells was not affected. A higher percentage of γδ + T cells producing TNF-α was found in old healthy donors, whereas a reduced number of TNF-α-producing γδ + T cells was present in melanoma patients independently by age. No significant difference was observed in IFN-γ production. After a 10-day in vitro culture, both the percentage and the expansion index of γδ T cells, and in particular of Vδ2 subset, were significantly and similarly reduced both in young/adult and old melanoma patients, and in healthy aged people, in comparison with young/adult healthy subjects. Conclusions Our study demonstrates that the numerical and functional impairment of γδ T cells found in melanoma patients is not correlated with age and that it has characteristics very similar to the alterations of γδ T cells found in old healthy subjects. We suggest that a similar impairment of γδ T cell population may be related to the increased susceptibility to tumors present in the elderly as well as in the pathogenesis of malignant melanoma. | Introduction T lymphocytes bearing the γδ T cell receptor (TCR) represent a minor population of human peripheral lymphocytes (1–10%), the majority of them expressing the CD3 + CD4 - CD8 - phenotype [ 1 - 4 ]. The ability of γδ T cells to respond to nonprocessed and nonpeptidic phosphoantigens in a major histocompatibility complex (MHC)-unrestricted manner is an important feature distinguishing them from αβ T cells [ 5 - 9 ]. In human peripheral blood two main populations of γδ T cells have been identified based on the TCR composition. The predominant subset expresses the Vδ2 chain associated with Vγ9 and represent 70% of the circulating γδ T cells in adults, while a minor subset (approximately 30%) expresses a Vδ1-Jδ1 chain linked to a chain different from Vγ9. At birth the Vδ1 population predominates, while in adults there is a shift towards Vδ2 T lymphocytes, probably due to a selective response to environmental stimuli such as commonly encountered bacteria. [ 10 ]. Although little is known about the physiologic significance of γδ T cells, their marked reactivity toward mycobacterial and parasitic antigens as well as tumor cells suggests that γδ T cells play a role in the anti-infectious and anti-tumoral immune surveillance [ 11 , 4 ]. Few data are available about the number and the function of γδ T cells and of Vδ1 and Vδ2 subsets in aging. The complexity of the gamma delta T cell repertoire has been found to decrease with age as a consequence of the expansion of a few T cell clones [ 12 ]. In our previous paper [ 11 ], we have evaluated the role of γδ T cells from young, old, and centenarian subjects, demonstrating an age-dependent alteration of γδ T lymphocytes, with a lower frequency of circulating γδ T cells, an altered pattern of cytokine production, and an impaired in vitro expansion of these cells. We suggested a role of the γδ T cell impairment in the age-related increase of infections and tumor diseases. Other studies have showed the involvement of γδ T cells in the immune defence against cancer either through a direct reactivity against tumor cells, or because of their regulatory interactions with αβ T cells [ 13 ]. Recently, we described an impairment of γδ T cell population in patients with cutaneous primary melanomas, with a decrease of their absolute number and percentage, an altered cytokine production, and a reduced expansion of γδ T cells, and particularly of the Vδ2 subset [ 14 ]. On the basis of the pivotal role that γδ T cells may have in the elderly and in the immune response against melanoma we tried to find out a possible correlation in the alteration of γδ T cells between aged people and melanoma patients. In this study we evaluated the peripheral representation, the in vitro expansion, and cytokine production γδ T cells from young/adult and old melanoma patients, comparing the results with those obtained in age-matched healthy controls. Materials and Methods Cell preparation and stimulation Human peripheral blood was obtained from 9 young (mean age ± SD, 42.3 ± 9.4 years; median: 41.0 years, range 30–59), 12 old (71.8 ± 5.4 years; median 72.5 years, range 60–80) melanoma patients, and 10 young (39.0 ± 5.7 years; median: 38.5, range 30–55), and 13 old (74.0 ± 2.0 years; median: 74.0 years, range 60–80) healthy donors. Healthy subjects were volunteers in good and stable clinical conditions, and had laboratory parameters in the physiological range. We excluded subjects in poor health with degenerative diseases or in therapy with drugs interfering with the immune system. Melanoma patients have been admitted to the Dermatology Unit of the I.N.R.C.A. Hospital of Ancona. Melanoma patients were in good health other than for the existence of melanoma as checked on the basis of clinical and laboratory parameters. The investigations were performed after approval by a local institutional review board. A written informed consent was obtained from each subject. Diagnosis of melanoma was histologically confirmed. All patients brought cutaneous primary non-metastatic melanoma and were staged according to the new American Joint Committee on Cancer staging system for cutaneous melanoma [ 15 ]. A blood drawing was taken before the surgical excision. Each donor was tested once and all the tests were carried out with a single blood sample. Fresh peripheral blood mononuclear cells (PBMC) were fractionated on Ficoll-Paque (Pharmacia, Uppsala, Sweden) and separated by density gradient centrifugation (400 g, 30 min). Cells from the interface of the gradients were washed twice with Ca 2+ and Mg 2+ - free phosphate buffered saline (PBS, Gibco/Life Technologies, Gaithersburg, MD, USA) and resuspended in RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum, penicillin (100 U/ml) and streptomycin (100 μg/ml) (all from Life Technologies, complete medium) at a concentration of 1.5 × 10 6 /ml. Mononuclear cells were cultured in the complete medium supplemented with 100 U/ml of IL-2 (Chiron Italia, Milan, Italy). Phosphoantigen-specific stimulation of γδ T cells was performed using the nonpeptidic antigen isopentenylpyrophosphate (30 μg/ml, IPP, Sigma Chemical Co., St. Louis, MO, USA). The cells were incubated at 37°C in an atmosphere of 95% air, 5% carbon dioxide, at 90% relative humidity in 24 well plates. Monoclonal antibodies and FACS analysis PBMCs were analysed for cell phenotype through double staining with the following monoclonal antibodies (mAbs): anti-CD3 (PE) and anti-pan γδ T cells (FITC), or anti-TCR Vδ1 or anti-TCR Vδ2. The phycoerythrin (PE) -conjugated monoclonal antibody anti-CD3 was purchased from EuroClone (Devon, UK). The fluorescein isothiocyanate (FITC) -conjugated anti-pan TCR γδ, anti-TCR Vδ1, and anti-TCR Vδ2 were purchased from Endogen (Boston, MA, USA). IgG1 (Becton Dickinson) was used as isotype control. 0.5 × 10 6 PBMCs were washed in PBS containing 0,1% NaN 3 plus 5% FBS and labelled with 5 μl of anti-CD3 or anti-TCR Vδ1 MoAbs or 2.5 μl of anti-pan TCR γδ or anti-TCR Vδ2 for 30 min in ice. At the end of the incubation, cells were washed in PBS containing 0,1% NaN 3 , resuspended in PBS (Gibco) and immediately analysed with a Coulter XL flow cytometer. Intracellular detection of IFN-γ and TNF-α Mononuclear cells were stimulated with IPP and IL-2 for 18 h, and GolgiPlug (a protein transport inhibitor containing brefeldin A, PharMingen, Milton Keynos, England) was added during the last 12 h of culture to block intracellular transport processes and allow cytokine accumulation. 0.5 × 10 6 stimulated cells were stained with the anti-pan TCR γδ mAb for 30 min at 4°C. Fixation-permeabilization of cells was performed in PBS/2% paraformaldehyde for 15 min at 4°C, followed by incubation for 30 min at room temperature in the dark with PE-conjugated anti-human IFNγ mAb or anti-human TNF-α mAb diluted in PBS, 1% BSA, and 0.05% saponin. Cells were finally washed twice in PBS, 1% BSA, and 0.01 % saponin and analysed on a XL flow cytometer (Coulter). Expansion assay PBMC were cultured for up to 10 days in the complete medium supplemented with 100 U/ml of IL-2 and 30 μg/ml of IPP to determine a phosphoantigen-specific stimulation of γδ T cells. After 1 wk of culture, the volume corresponding to half the culture medium was replaced by fresh medium. On day 10 of culture viable cells were determined by trypan blue exclusion and used for FACS analysis. The viability was always greater than 98% as determined by trypan blue exclusion. The expansion of γδ T cells was followed by cytometric analysis through double staining of stimulated cells with anti-CD3 (PE) and anti-pan γδ, or anti Vδ1, or anti Vδ2 T (FITC) mAbs. The absolute number of γδ T cells in each culture was calculated as follow: (percentage of γδ T cells among total cells) × (total cell count)/100. The γδ T cell expansion index was then calculated by dividing the absolute number of γδ T cells in stimulated cultures by the absolute number of γδ T cells before culture [ 16 ]. Statistical analysis Data were analysed for statistical significance by using parametric or nonparametric tests according to the distribution of the data. Comparisons of variables among groups were made by one-way analysis of variance (ANOVA) or Kruskal-Wallis one-way ANOVA on ranks. When significant differences were found, the differences among groups were made by the Student-Newman-Keuls method or Dunn's method. Difference between means was considered significant at the 5% level ( P < 0.05). The statistical analysis was performed with SigmaStat software version 1.03 (Jandel Scientific, Germany). Results Ex vivo analysis of γδ T lymphocytes Peripheral blood lymphocytes from 9 young/adult and 12 old melanoma patients and 10 young and 13 old healthy subjects were analysed for the percentage and the absolute number of γδ T cells through double staining with anti-CD3 and anti-γδ mAbs. As shown in Table 1 the absolute number of γδ T cells was significantly reduced in both groups of melanoma patients and in healthy aged people in comparison with young/adult healthy subjects (p < .01). As shown in Fig. 1A , the ex vivo percentage of CD3 + γδ + T cells in the peripheral blood was significantly lower in melanoma patients than in healthy donors (p < .01). As shown in Table 1 , the absolute number of Vδ1 T cells did not show significant difference in the four groups of donors. Differently, the absolute number of Vδ2 T cells was significantly reduced in both groups of melanoma patients and in old healthy people in comparison with young/adult healthy subjects (p < .0001). The Vδ2 and Vδ1 subsets were differently represented in the four groups: in young/adult healthy controls the Vδ2 subset was predominant (Vδ2/Vδ1 ratio = 2.2) whereas in old healthy donors and in young/adult and old melanoma patients the Vδ2/Vδ1 ratio was progressively decreased. Table 1 Absolute number of γδ T cells, Vδ1 T cells, and Vδ2 T cells, in young/adult and old healthy subjects and melanoma patients. Absolute number Donors γδ T cells Vδ1 T cells Vδ2 T cells Vδ2/Vδ1 ratio Healthy Young/adult 115.2 ± 39.3 a 38.0 ± 11.9 82.6 ± 34.0 2.2 Old 62.1 ± 28.7* 35.9 ± 11.7 37.7 ± 24.6* 1.0 Melanoma Young/adult 74.5 ± 25.3* 31.9 ± 18.1 42.8 ± 9.8* 1.3 Old 52.9 ± 38.0* 20.6 ± 3.7 33.5 ± 31.2* 1.6 a Data are expressed as mean ± S.D. of the number of cells per mm 3 in the peripheral blood. *p at least <0.01 versus young/adult healthy subjects Figure 1 Percentage of γδ T cells in young/adult and old melanoma patients and age-matched healthy subjects. Freshly isolated (A) or 10-day cultured (B) PBMC from young/adult and old melanoma patients and young/adult and old healthy subjects were double stained with MoAb anti pan- γδ (FITC) and anti-CD3 (PE) and analyzed by flow cytometry. Statistical analyses was performed as reported in Mat. and Methods. Cytokine production by γδ T lymphocytes Since it has been demonstrated that activated γδ T cells produce TNF-α and IFN-γ, we studied the intracellular production of these cytokines in one-day stimulated γδ T cells from healthy subjects and melanoma patients. As shown in Fig. 2 the percentage of γδ T cells producing TNF-α was significantly higher in old healthy controls in comparison with young/adult healthy, and young/adult and old melanoma patients (p < .05). The percentage of γδ T cells producing IFN-γ was similar in young/adult and old healthy subjects (mean ± SD, 17.9 ± 10.0 and 14.7 ± 8.9). In a similar way the percentage of γδ T cells producing IFN-γ did not show differences between young/adult and old melanoma subjects (8.5 ± 4.9 and 8.1 ± 1.3) (data not shown). Figure 2 Analysis of TNF-α production by γδ T cells in melanoma patients and age-matched healthy subjects. PBMC young/adult and old melanoma patients and young/adult and old healthy subjects were stimulated for 18 h in the presence of IPP (30 μg per ml) and IL-2 (100 U per ml). The last 12 h of culture were performed in the presence of GolgiPlug, a protein transport inhibitor containing brefeldin. Single-cell analysis of TNF-α synthesis in γδ T cells from a representative subject for each group was performed following dual staining with cell surface anti- γδ T (FITC) MoAb and intracellular anti- TNF-α (PE) MoAb. Number in brackets indicate the percentages of γδ T cells synthesizing TNF-α among total γδ T lymphocytes. Expansion of γδ T lymphocytes The expansion of γδ T cells was evaluated after 10 days of culture in the presence of IPP and low dose interleukin-2 (IL-2). Both the proportion of γδ T cells, evidenced by double staining FACS analysis, and their relative increase in comparison with the γδ T cell number found on day 0 (expansion index), were evaluated. As shown in Fig. 1B , the proportion of γδ T cells reached on day 10 was significantly lower in both groups of melanoma patients and in old healthy subjects than in young/adult healthy donors (p < .01). In a similar way, the expansion index of γδ T cells after 10 days of in vitro culture was significantly reduced in the same three groups mentioned above (p < .05, Table 2 ). As shown in the same Table 2 , the expansion index of the Vδ2 subset was significantly lower in all melanoma patients and in old healthy donors than in young healthy donors (p < .03). Table 2 Expansion index of γδ T cells and Vδ2 T cells in healthy subjects and melanoma patients Expansion Index a Donors γδ T cells Vδ2 T cells Healthy Young/adult 13.1 ± 8.8 b 7.1 ± 4.5 Old 4.6 ± 3.5* 3.1 ± 2.0* Melanoma Young/adult 3.8 ± 2.6* 4.6 ± 0.2* Old 2.0 ± 1.8* 0.9 ± 0.4* a The expansion index was calculated by dividing the absolute number of γδ T cells in stimulated cultures by the absolute number of γδ T cells before culture. b Data are expressed as mean ± S.D. *p at least <0.05 versus young/adult healthy subjects Discussion We and others have demonstrated an impaired potential of γδ T cells in aged people, as evidenced by the reduction of the absolute number of circulating γδ T cells, in particular of the Vδ2 T subset, an altered pattern of cytokine production, an impaired in vitro expansion of these cells, and an increased expression of the early activation marker CD69, in aged people in comparison with young subjects [ 11 , 17 , 18 ]. Recently, studying a group of melanoma patients ranging from young to old age (32–80 yr), we described numerical and functional alterations of γδ T cells from these subjects, once compared to healthy age-matched donors [ 14 ]. In this study we have investigated on whether the age-related impairment of circulating γδ T cells is similar to the alteration found in melanoma patients and if melanoma patients of advanced age have a greater impairment of γδ T cells in comparison with melanoma patients of younger age or with old healthy donors. With these premises, we studied the peripheral representation, in vitro expansion, and cytokine production of γδ T lymphocytes from young/adult and old patients with cutaneous primary melanoma comparing the data obtained with age-matched healthy subjects. We demonstrated that both the number of circulating γδ T cells and their in vitro expansion were decreased in melanoma patients and that the impairment did not correlate with the age of patients. Young/adult and old melanoma patients had a similar derangement of γδ T cells, and this impairment had numerical and functional characteristics like to those found in old healthy subjects. This evidence stresses the relevant role that this lymphocyte population may exert, either directly or through the regulation of T cell-mediated specific responses [ 11 ], both in the elderly and in melanoma patients. The reduction of γδ T cell number well correlated with the decrease of the Vδ2 T cell subset, i.e., the most frequent subset of circulating γδ T cells [ 2 , 4 ]. The Vδ2 population is involved in the reactivity toward microbial antigens and tumor cell antigens [ 4 , 19 ]. The role of Vδ2 T cells in the immune defence against cancer has been demonstrated on the basis of their reactivity against certain lymphoma cells, such as Daudi cells [ 20 ], and for their presence among tumor infiltrating lymphocytes in various cancer types [ 21 ]. Not only the number but also the function of γδ T cells was altered in melanoma patients as well as in old healthy subjects. The in vitro expansion of γδ T cells, that represent one of the most relevant functional parameters for γδ T cells, was significantly reduced in young/adult and old melanoma patients, and in old healthy donors. Under normal conditions, γδ T cells respond to antigen challenge by secreting large quantities of TNF-α and IFN-γ [ 16 , 21 ] which contribute to the activation of both specific and aspecific immune responses. In aged subjects we found an increased production of TNF-α by γδ T cells [ 11 ]. In this study, we show that the percentage of γδ T cells producing TNF-α was significantly reduced in young/adult and old melanoma patients in comparison with age-matched healthy subjects. Probably, the pro-inflammatory state which has been described in old ages [ 22 ], may represent a stimulus for the production of TNF-α in γδ T cells from aged subjects, differently with what occurs in old melanoma patients. In conclusion, we demonstrate that the numerical and functional derangement of γδ T cells which we have found in melanoma patients, is not correlated with age of donors, and that old patients with cutaneous primary melanoma have an impairment of γδ T cells similar to that found in old healthy subjects. This evidence suggests a link between γδ T cell deterioration and the low protection against infections and tumor diseases present in the elderly, as well as the inefficacious immune defense against melanoma, both in young/adult and old ages. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC549530.xml |
518970 | Feasibility of a patient decision aid regarding disclosure of personal health information: qualitative evaluation of the Health Care Information Directive | Background Concerns regarding the privacy of health information are escalating owing both to the growing use of information technology to store and exchange data and to the increasing demand on the part of patients to control the use of their medical records. The objective of this study was to evaluate the Health Care Information Directive (HCID), a recently-developed patient decision aid that aims to delineate the level of health information an individual is willing to share. Methods We convened a series of four focus group meetings with several communities in a large Canadian city. A total of 28 men and women participated, representing health care consumer advocates, urban professionals, senior citizens, and immigrants who speak English as a second language. Data were analysed using qualitative methods. Results Participants lacked substantial knowledge regarding the fate and uses of personal health information. They expressed mistrust concerning how their information will be used and protected. Several suggestions were made towards customizing the use of data according to specific needs rather than broad and full access to their charts. Furthermore, despite concern regarding the implementation of a tool like the HCID, participants were hopeful that a refined instrument could contribute to the improved regulation of health information. Conclusion This study indicated poor knowledge concerning the uses of personal health information, distrust concerning security provisions, and cautious support for a patient decision aid such as the HCID to improve control over health data. | Background Health information privacy has come to the forefront of ethical concern in the early 21 st century [ 1 ]. The advent of electronic health records, information technology, and large databases (such as administrative and genetic) with the potential for extensive linkages have raised concerns about the security of health information [ 2 , 3 ]. The details of where such information flows, who has access, and for what purposes have assumed paramount importance [ 4 - 6 ]. Health information is valuable for numerous purposes, first and foremost for patient care, but also for secondary uses including hospital administration and health services research [ 7 , 8 ]. Such non-clinical care uses are required for system performance and evaluation, as well as to answer questions about disease trends and health outcomes. In general, there is a prima facie obligation for the protection of such intimate and privileged information [ 4 , 8 ]. Scientific surveys and public opinion polls have shown that access to medical records is generally considered appropriate after consent has been obtained for a specific use [ 9 - 12 ]. Yet, this obligation admits to several exceptions and there is lack of clarity as to whether express consent is required for each and every use. Some have argued that requiring unique individual informed consent for each use of health information would be burdensome [ 13 - 15 ]; however, there are public opinion data suggesting that legislative initiatives to require such consent would be viewed favourably [ 9 ]. Internationally, legal initiatives have proposed solutions to the dilemma posed by health information. In Europe and North America, such initiatives have led to restrictive legislation that, according to some commentators, may endanger public health goods [ 4 , 16 , 17 ]. A privacy paradox thwarting further progress has been identified: individuals want both the guaranteed privacy of their personal health information (PHI) and the public benefits that accrue from the use of medical records [ 18 ]. Two distinct avenues have been proposed for the solution of this paradox; the first avenue concentrates on governance issues, whereas the second promotes the development of tools, programs, and systems to enhance the lay understanding of and control over the uses of health data and thereby facilitate informed consent for secondary uses, as advocated by Mandl and his colleagues [ 19 ]. This paper pursues the second avenue. In 2001, Upshur and Goel first proposed the Health Care Information Directive (HCID), a patient decision aid analogous to an advance directive in end-of-life care [ 20 ]. The underlying logic was to combine ethical appropriateness of use of PHI with the sensitivity of the data. As shown in Table 1 , the HCID clearly presents the various permutations and combinations of sensitivity and usage in the form of a matrix. The goal of the tool is to allow individuals to make informed choices about the specific types of health information they are willing to disclose (if any) for a number of specified purposes. In the original proposal, it was stated that pilot testing of the tool, and revision on the basis of such testing, is required [ 20 ]. In order to assess the feasibility of implementing the HCID, two segments of the population were studied: the general lay public and privacy experts such as ethicists, academics, and provincial privacy commissioners. In this paper, we report the results of the study conducted with the general public. Specifically, our objective was to investigate lay knowledge of the uses and accesses of health information and to solicit feedback on the prototype of the HCID. Methods Participants and setting The study was set in Toronto, Canada, a large, culturally-diverse urban centre. In order to sample multiple views and perspectives, a series of focus group meetings was convened with the following four groups: senior citizens, urban professionals, immigrants with English as a second language, and consumer advocates. The latter group was comprised of volunteer members of a well-known national consumer advocacy association for which health care is an issue of primary interest and activity. We opted to employ focus group methodology as opposed to individual interviews or a questionnaire survey in order to capitalize on the effects of group interaction. These particular communities or target groups were selected in order to maximize the variability within our sample on important demographic characteristics such as age, gender, occupation, education, and native language, etc. (available funding allowed for a total of four focus groups). Potential participants were recruited using various methods: the seniors were recruited through a local community centre offering programs for senior citizens; participants in the immigrant/ESL group were recruited through an immigrants group at another community centre; the urban professionals were recruited by way of posters and fliers distributed at a number of hospitals and university sites; and the advocates were recruited through a national consumer advocacy association. In all cases, initial contact with potential participants was made either by telephone or e-mail and then followed up with formal letters of invitation describing the project. The focus group meetings were held in a convenient location for the participants and lasted for up to two hours. We developed a topic guide for the focus group meetings in order to address the main issues related to the feasibility of the HCID (see Appendix 1). The topic guide was developed according to the principles of formative evaluation (also known as 'developmental' evaluation) [ 21 ]. Focus group participants were initially questioned regarding their understanding of health information and its uses. Participants were then introduced to the HCID, which was presented as a patient decision aid currently under development by "university researchers." Participants were informed that their feedback and suggestions would guide the continuing refinement of the tool and were asked to be as candid as possible. Following a brief period of approximately 5–10 minutes in which participants examined the HCID and jotted down any questions/concerns/suggestions, participants were asked to share their thoughts on its relative strengths and limitations. As per the topic guide, issues of content, utility, and feasibility were addressed in turn, followed by a discussion of perceived benefits and burdens. At the end of each meeting, participants were provided an opportunity to raise any issues or concerns that had not been previously addressed. Two of the authors (GCD and CST) moderated the four focus group meetings, which lasted 90 minutes on average. A total of 28 participants took part (see Table 1 for a description of the sample). The meetings were audio-recorded with participants' consent and transcribed verbatim by a professional transcriptionist. To ensure accuracy and to clarify any muffled passages, all transcripts were verified by one of the two group moderators. Participants received an honorarium of $50 for their time, in addition to transportation costs. The study was approved by the Research Ethics Board at Sunnybrook and Women's College Health Sciences Centre and the Office of Research Services at the University of Toronto. All participants signed and returned a consent form. Data analysis The analytic process was one of thematic content analysis. The topic guide developed for the focus group meetings served as the basis for the data analysis process. Two of the authors (GCD and CST) independently read each of the transcripts and identified passages of text relating to each of the various key issues from the topic guide (e.g., content, utility, benefits, etc.) which, for the purposes of coding and analysis, served as the macro-codes. Following this step, lists of themes were constructed and then compared. The transcripts were then independently coded according to an agreed-upon coding scheme. Tests for inter-coder reliability indicated a high level of agreement among the two coders; instances of disagreement were resolved through a process of discussion and negotiation. To strengthen the validity of the findings, the analytic processes of coding and interpretation were reviewed by the senior author (REGU). The results of our analysis are reported according to five key themes: participant knowledge and understanding of health information and its uses; control of access to health data; mistrust of data security provisions; need and utility for a tool such as the HCID; and, finally, perceived implementation barriers. Consenus statements reported below are not a reflection of any explicit consensus development techniques, but rather are summary statements of the research team's observation and interpretation of the focus group discussions. For each quotation, a specific code is provided to identify the speaker as a participant in one of the four focus groups (CA = consumer advocate; IM = immigrant; SC = senior citizen; and UP = urban professional). Results Knowledge and understanding The majority of participants possessed extremely limited knowledge of how their PHI is collected, used, and disclosed. Many confessed to having given little or no thought to the issues involved in the health privacy debate. This was particularly true for recent immigrants: "I think the truth is that I don't know. I've never thought of that before, who has my information."(IM-1). The level of understanding was low among participants in other groups as well, as a number of comments betrayed basic misperceptions of how PHI is currently managed within the Canadian health care system: "It [personal health information] goes in the computer and then it's available to every medical professional in Ontario." (SC-3) "Health providers have access to your file, to your information... but everybody in the financial department, too, because they have to bill OHIP [Ontario Health Insurance Program] so they have to know everything about you." (IM-2) One participant perceptively noted that the general population has limited knowledge of the issue of health privacy: "That's another thing. What do people know about what they can get access to, what they can ask for, and what they can expect? I think the majority of the population have no idea of what they can ask for and expect to get." (CA-4) Control of access Participants' accounts clearly suggested an absence of patient control over the collection, use, and disclosure of PHI. No participants recalled having ever been consulted about how their information was to be used. A great deal of concern was voiced about the extent to which health data appears to be freely accessible to a wide variety of users: "Lawyers, psychologists, social workers, researchers, pharmaceutical companies. Where does it stop?" (UP-4). In the course of describing how they feel about the issue of health privacy, participants repeatedly used terms such as "scary" and "horrifying": "I'm scared to guess who has [access to my health information]. It looks so easy for a lot of people to have access. That's the scary part of it. Maybe your employer can have access to your files, too. I don't know, that's just a guess." (IM-6) The majority of participants expressed concern that their PHI is not adequately safe-guarded and that the implementation of a tool such as the Health Care Information Directive would not result in significantly enhanced privacy or increased security. There was a widespread view that too much data is currently made available when only specific details are required. Doubts were raised about how consent for one specific use only would be managed: "What's going to prevent any leaking from one of these [uses] into the others? ... It just seems to me that if there's information on-line, things are going to be compromised. You know, people make a living doing that stuff. The more they find out about you, the more you can be exploited. It's that simple. These systems, they're not secure yet, and I don't know if they can ever be secure." (UP-3) Health privacy concerns related to the security of electronic databases and the Internet were shared by others: "We've all heard stories where there's been stolen identities. How difficult is it for the victim to get his or her own identity back? Same idea. Where does it end? Where does it stop? Who's got what information? How am I going to protect myself?" (UP-6) Participants suggested a number of other mechanisms that could work in conjunction with the HCID to enhance security and facilitate individual control over PHI. One such mechanism would be an online real-time audit system in which the details of all accesses to an individual's PHI are recorded and made available to those wishing to track access to their PHI over time. Also, the idea of a health data ombud was raised in several groups and received a great deal of support. Mistrust Issues related to trust were raised in each of the four groups. Participants of all ages and socioeconomic status expressed feelings of mistrust in relation to the protection of their privacy and the security of their PHI. A great many participants spoke of how their past experiences with the health care system have fostered significant mistrust and suspicion where their right to privacy is concerned. These accounts revealed a growing distress that large corporations have too much access to and influence on government programs, especially in contrast to the access and influence accorded to patients: "What about the rights of the patient? Let's say I'm the patient. What kind of power do I have? Let's say this [the HCID] was created next year. What power does the patient have to make sure any of this is happening? To me, a pharmaceutical company is way more powerful than the patient." (CA-2) Others were even more sceptical, questioning the trustworthiness of the basic tenets of the model upon which the HCID is based: "By filling this out, I'm buying into the concept of sharing information, but I don't have any faith that it can be kept private.... It will spill, it will bleed, it will flow. So I'm distrustful of the whole thing. This just sets up more spilling and more flowing. If I fill out a form like this, then I'm validating the process, which I don't really trust." (UP-8) Need and utility While the majority were sceptical that the HCID would prevent all breaches of privacy, there was a general consensus that it would serve to enhance significantly the security of health data. One participant noted that the proposed decision aid may also serve a useful purpose as "a sort of consciousness raising" tool. Other impressions varied from "it has some potential" to "it is a great step forward." Reactions were mixed in response to the question of whether the HCID will be successful in empowering individuals and increasing the amount of control over PHI: "I'm very dubious as to whether this matrix will be useful because of the difficulty people will have filling it out. In spite of that, I think the idea has merit and principle. There's merit in what you're trying to do, but I don't think that this is going to succeed." (CA-5) "I guess the reality is our information will be shared, so we might as well get on the bandwagon with regulating it and controlling it... You can't stop it from being shared, so maybe you can influence how it will be shared." (UP-2) Despite the weaknesses and limitations of the present version of the directive, one participant neatly summarized the view of the majority of participants regarding the utility of the HCID or some such tool: "Not having it allows total absence of control, therefore it is a necessary evil." (CA-4). Implementation barriers Participants provided numerous suggestions regarding the formatting of the HCID in order to facilitate implementation. Ideas ranged from simplifying the language and providing definitions of technical terms to modifying the layout and shading those areas where there is no discretion (i.e., for physician payment): "Maybe you've got too many columns.... Well, maybe you're trying to do too many different things at once." (CA-4) "This is too busy, it's too much. If I'm sick, I friggin' don't want to be bothered with it.... Look at this. English is my first language. How would somebody whose mother tongue is something other than English? It's too complicated." (SC-1) "I think people tend to say 'no' for things that are not clear. I would say 'yes' if I knew what it means exactly, but I don't know, so I don't want to take a chance." (IM-1) To address the complexity issue, suggestions were made concerning the need to provide a customer service representative either in a health clinic or via a toll-free helpline for assistance with completing the HCID. Across the four groups, there was great variability in the preferred mode of implementation. The preference among participants in the health advocates and urban professional groups was for an on-line implementation format. In contrast, the majority of the senior citizens and immigrants preferred other options, the former favouring a postal format and the latter the primary care setting. As one senior citizen remarked: "I prefer the doctor's office. I wouldn't fill it and send it back through the mail, no." (SC-3) Discussion A recent editorial in the British Medical Journal suggested that perhaps patients should be asked whether certain items of their medical chart should only be shared with specified individuals or organizations or only for pre-determined purposes [ 16 ]. This paper reports the evaluation of the feasibility of a tool that seeks to accomplish exactly that purpose, namely, greater patient control over how personal health information is used and disclosed. Study participants lacked substantial knowledge regarding the fate and uses of PHI within a publicly-funded health care system. Participants expressed mistrust concerning how their PHI is used and safe-guarded. Several suggestions were made towards customizing the use of data according to specific needs rather than broad and full access to their charts. Furthermore, concerns were expressed regarding the implementation of a tool such as the HCID. Nevertheless, there was hope that a refined instrument could contribute to improved data management and regulation and enhanced privacy protection. Although this study reports on a small sample from a single large urban centre, the focus group participants were drawn from various different niches of Canadian society. This sampling strategy allowed us to explore a broad range of experiences and perspectives; however, further testing and evaluation are required. Ultimately, it will be necessary to evaluate the tool using a representative sample of patients who complete the HCID in a 'live' test of its feasibility. Our findings underscore the difficulties involved in accessing health care data for research and other secondary purposes. Participants acknowledge the myriad benefits derived from the use of health data; however, distrust, lack of respect, and insufficient patient control of the process threaten to undermine these very benefits. This finding has been previously reported by Willison and associates in Canada [ 22 ] and by Robling and colleagues in the UK [ 12 ]. The present results also suggest that the education and information needs of diverse groups such as seniors and immigrants who speak English as a second language should be taken into account when considering strategies to enhance individual control over PHI and minimize the problem of authorization bias when utilizing health information for secondary purposes. Participants appreciated the benefits accorded by a tool such as the HCID. As opposed to forms of blanket consent or other opt in/opt out models, the possibility of exerting greater control over one's PHI was attractive. The participants provided concrete suggestions for improving the format and content of the HCID. It is evident that any method to enhance control of health information via explicit consent requires description of the various forms the data may take, the specific purposes for which the data would be used, and the various channels of the health care system through which the data might flow. To our knowledge, such detailed data flowmaps for PHI do not exist in Canada, although they have been laid out in Great Britain [ 23 ]. A model has been proposed by Schoenberg and Safran [ 24 ]. The creation of such maps is of high priority. An intriguing finding was the appeal of an online data audit system. Possessing the ability to monitor who has accessed their PHI and for what purposes raises the possibility of additional strategies that could empower individuals to control the fate of their health information. This finding has also been verified by Pyper and colleagues [ 25 ] and was previously highlighted by MacDonald [ 26 ]. As well, the concept of a data ombudsperson was considered attractive to a number of participants, indicating that an improved governance framework would be acceptable to some segments of the population. Finally, there is a distinction to be made between using the HCID to enforce the will of the patient versus its use as a documentation tool. Our vision is that the HCID will, ultimately, serve both of these important functions; further follow-up evaluation of a revised model of the HCID using a larger sample (comprised of patients as well as providers) is needed to address this distinction. As with any patient decision aid or empowerment tool, documenting the preference of the patient is only meaningful and useful to the extent that the documented preferences are known and ultimately acted upon by those providing care. We believe the present data illustrate the critical problem of mistrust that currently exists. Indeed, this is one of the greatest challenges to be overcome in the continuing development and validation of this tool. Conclusion This study indicated poor knowledge concerning the uses of health data, distrust concerning current security provisions, and qualified support for a tool such as the HCID to improve patient control over health information. On the basis of this evaluation, the HCID will be revised significantly, including the addition of an educational component, and then submitted to further evaluation. The creation of data flowmaps and the exploration of audit functions and governance structures are strongly recommended as avenues for future research. Competing interests None declared. Authors' contributions CST participated in the collection, analysis, and interpretation of the data and is the primary author of the paper. GCD recruited the participants, moderated the focus group meetings, participated in the data analysis, and contributed to the editing and revising of the paper. REGU initiated and designed the study, participated as a reliability check in the process of data analysis, and contributed to the editing and revising of the paper. As principal investigator, he will act as guarantor. All authors have read and approved the final version of the paper. Appendix 1 FOCUS GROUP TOPIC GUIDE: EVALUATION OF THE 'HEALTH CARE INFORMATION DIRECTIVE' A. Personal Health Information 1. What is your understanding of the term 'personal health information'? 2. Who do you believe has access to your personal health information? 3. Do you believe that consent should be required to access your personal health information? B. Content 1. What is your first impression of the Health Care Information Directive? 2. Is it clear? 3. Is it self-explanatory? C. Utility 1. How useful do you believe the Health Care Information Directive would be in practice? 2. Is it user-friendly? 3. What kind of changes would you suggest? D. Feasibility 1. How feasible is the application of the Health Care Information Directive? 2. Who should present it to the patient? 3. When should it be presented to the patient? E. Benefits and Burdens 1. What are the potential gains of the Health Care Information Directive? 2. What are the potential harms? 3. Does it adequately protect privacy and confidentiality? F. Additional Comments Pre-publication history The pre-publication history for this paper can be accessed here: | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC518970.xml |
340954 | A New Trade Framework for Global Healthcare R&D | Current business models for drug development are inefficient and ineffective - drugs are not reaching all who need them. Hubbard and Love contend that it is time to explore some alternatives | The AIDS crisis has brought to public notice what has always been generally true—that the existing business model for drug development leads to high prices and unequal access. There is now widespread dissatisfaction with drug prices in both the developed ( Families USA 2003 ) and developing world ( Correa 2000 ). Governments and health insurers are finding ways to deny access to the newest and priciest products. In the United States and other countries without a universal public health system, the uninsured simply cannot afford the newest medicines. In developing countries, life-saving medicines are priced beyond the reach of most people, a morally offensive outcome ( TrueVisionTV 2003 ). Huge publicity surrounds negotiated price reductions for specific drugs in specific developing countries, yet the effect on the overall access problem is tiny. Today's high drug prices are a direct consequence of a business model that uses a single payment to cover both the cost of manufacture of a drug and the cost of the research and development (R&D) carried out by manufacturers to discover it. A 20-year patent-based marketing monopoly is then granted to the drug's developers to prevent their prices being undercut by ‘generic’ copies produced by manufacturers who do not have R&D costs to recover. Preventing such ‘free riding’ on R&D has become a global trade issue at the World Trade Organisation (WTO) ( Drahos and Braithwaite 2002 ). The implementation of the TRIPS (Trade-Related Aspects of Intellectual Property Rights) agreement and a growing number of regional and bilateral agreements on intellectual property require most countries to implement tough patent systems that discourage or eliminate competition from manufacturers of generic medicines ( Box 1 ). Unfortunately, monopoly-based business models have unpleasant side effects. Since the primary responsibility of any company is to maximise return on investment, it is unsurprising that there is pressure on pharmaceutical companies to set drug prices to whatever level gives the highest return, excluding those individuals who cannot afford to pay, rather than maximising the number of patients treated. There is also pressure to misuse the power given by patents, using them as anticompetitive weapons to block innovation and extend marketing monopolies. And there are growing fears that the huge growth in the use of patents is in itself starting to inhibit research ( CIPR 2002 ; Anonymous 2003 ; Royal Society 2003 ). Something that is less well recognised is that this system is an enormously inefficient way of purchasing R&D. There is a considerable lack of transparency in pharmaceutical R&D investment, but the available data indicate that only about 10% of drug sales go towards R&D on new products. Only about one-quarter of new drug approvals are rated by the United States Food and Drug Administration (FDA) to have therapeutic benefit over existing treatments ( NIHCM 2002 ; see Figure 1 ). Measured by investment, only about one-fifth of the 10% is invested in innovative products ( Love 2003a ). There is also very little research for diseases that primarily afflict the poor ( Trouiller et al. 2001 ; WHO 2003 ). Figure 1 A Breakdown of the 1,035 New Drugs Approved by the FDA between 1989 and 2000 More than three-fourths are classed as having no therapeutic benefit over existing products, so-called ‘me too’ drugs ( NIHCM 2002 ). Less than 1% address diseases that primarily afflict the poor, for which new treatments would have the greatest effect on world healthcare ( WHO 2003 ). Industry trade associations, reports to investors, and data from income tax returns suggest somewhere between 10% and 15% of the $430 billion revenues (reported in 2002) are spent on R&D, but data from regulatory bodies imply that only approximately 2%–3% is actually spent on R&D that leads to new medicines with therapeutic benefits over existing ones, and even this is inflated by research primarily designed to achieve marketing outcomes ( Love 2003a ; WHO 2003 ). Propping up the present structure for financing R&D ( Figure 2A ) is the widely held belief that the private sector plays a key role in the development of new medicines and that it is necessary to grant patents to incentivise private-sector financing. If this were true, it would make sense to tolerate all sorts of bad outcomes, because the fruits of R&D eventually benefit everyone. But granting a 20-year marketing monopoly on a patented invention is only one way to finance R&D, and the shortcomings of the present system are increasingly hard to ignore. Suggestions for alternatives are beginning to come from many quarters ( Baker and Chatani 2002 ; CGSD 2003 ; Hubbard and Love 2003 ; Weisbrod 2003 ). In this essay, we present practical proposals to modify trade rules based solely on intellectual property so that alternative policy instruments can be used to encourage innovation. Figure 2 Funding Healthcare R&D (A) A schematic of the way the public currently funds healthcare R&D. Academic research funded by government research agencies is paid for via taxes. This is a mixture of pure research into fundamentals and directed research, including clinical trials. Despite this, there is a dogma that academic R&D cannot produce drugs since it does not have the required commercial pressures to turn ideas into products. Patents ensure the public pays for commercial R&D via their purchase of medicines at high prices, compared to those of generic copies. The distortion of research priorities (too much spent on ‘me too’ drugs and too little on neglected diseases) has been recognised by governments for some time, and a variety of push-and-pull mechanisms have been introduced (or are being considered) to encourage research that more closely reflects public priorities. Examples of push incentives are tax breaks for R&D and other incentives such as special marketing monopolies for products as a reward for investing in research on orphan drugs or testing with pediatric patients. Pull incentives currently being discussed are advance-purchase commitments, with which governments guarantee to buy a certain amount of a drug if one is developed, or prize models. Some of these schemes are thought to be inefficient, particularly those that are indiscriminate and provide expensive subsidies relative to the amount of new R&D they ‘encourage’. (B) A schematic of the way funding of healthcare R&D could work if separate competitive markets for sales and R&D were created. A crucial difference is the absence of monopolies on final products, enabling competition between generic producers and greatly reduced prices. Incentives to develop new drugs would be provided by a new virtual market in R&D. ‘Nationally directed R&D funds’ could represent anything from rewards for innovation using market based mechanisms such as prize models (see text) to centralised funding agencies, similar to the NIH model, or multiple R&D investment funding bodies that compete for new resources. Contributions to R&D could be via taxation or as a legal obligation when paying for private healthcare plans (see text). The ability to design what would be rewarded in the virtual market would allow governments to set R&D priorities and build up local capacity within their own countries. Countries could choose weaker patent protection and create an environment in which all research groups could build on each other's work. A New Trade Framework Analysis of worldwide drug expenditure shows that spending varies, but is close to 1% of the gross domestic product (GDP) in most developed and developing countries ( Love 2003b ). Assuming that about a tenth of the revenue from the sale of drugs is ploughed back into R&D on new products, that means that countries already indirectly contribute about 0.1% GDP to support this. This contribution is enforced by trade agreements, which require the granting of patents to prevent ‘free riding’ via the purchase of generic drugs (see Box 1 ). Suppose the World Health Organisation (WHO) developed an R&D contribution ‘norm’ based upon this or a more appropriate figure and that there was international agreement that countries evaluated as meeting this norm would no longer be regarded as ‘free riding’. Trade rules could then be modified to allow countries to meet this norm by any means , not just by the implementation of strict TRIPS intellectual property rules, as at present. Countries that met the norm would then be free to decide whether they wanted to follow a strictly patent-based system as at present, with high drug prices for 20 years, or experiment with new models based on the creation of separate competitive markets for sales and R&D ( Figure 2B ). Countries adopting the latter system would remove patents on final drug compounds, placing them in the public domain. This would allow them to become a freely traded commodity, creating a competitive manufacture and sales market with low generic prices. At the same time, in order to meet the required R&D contribution norm, they would have to create an efficient R&D virtual market alongside. However, the costs of this would be more than offset by the reduction in drugs prices, making substantial savings for that country overall. Business Models for an Effective Virtual R&D Market The existing system ( Figure 2A ), despite its failings, does lead to the development of new drugs. The challenge in creating a virtual R&D market is to find viable business models for successful drug development in the absence of marketing monopoly incentives. One obvious approach is direct funding of drug development. For example, the National Institutes of Health (NIH), the national agency in the United States, already spends $27 billion per year on research, a substantial amount of which is directed towards drug development, including clinical trials. The NIH already has a track record in developing important drugs for severe illnesses, such as cancer or AIDS, showing that this is a viable model. It is also widely recognised that much of the research carried out across the world by similar agencies underpins the existing commercial research that leads to new drugs. Governments could expand direct funding for drug development, either through the existing structures in academia or through funding R&D arms of existing companies to carry out specific drug R&D. Such directed drug development funding could be similar to existing nonprofit development projects, such as those currently resourced to address treatments for neglected diseases like malaria and tuberculosis (TB). Examples of such projects are the Medicines for Malaria Venture ( www.mmv.org ), the Global Alliance for TB Drug Development ( www.tballiance.org ), the International AIDS Vaccine Initiative ( www.iavi.org ), the Drugs for Neglected Diseases Initiative ( Butler 2003b ) ( www.dndi.org ), and the Institute for One World Health ( www.oneworldhealth.org ). Many are doubtful that increased direct funding would generate sufficient incentives or be managed efficiently enough. An alternative market-based approach is one in which R&D organisations compete for rewards for specific R&D output, referred to by economists as a prize model ( Wright 1983 ; Kremer 1998 ; Shavell and van Ypersele 2001 ). In a simple formulation, governments would place large sums into a fund that would be allocated every year to firms that bring new products to market. This could work with or without patents. If products were protected by patents or other intellectual property claims, the government could grant compulsory licenses (a procedure allowed by trade agreements to override monopoly rights on a patent, in return for compensation to rights owners; see Box 1 ) and permit rapid introduction of generic competition. The reward system could be a lump-sum payment, eliminating any incentive to continue to market the product, or a long-term payout structure, which would depend upon evidence of both usage and efficacy. Prize systems could be designed to be fairly similar to the current system, with big payoffs for successful entrepreneurs, but even with this approach, there would be huge opportunities to improve welfare. The reward system could be more rational than the existing system, allocating greater rewards for innovative products and less for ‘me too’ products that do not work better than existing products. Premiums could be given for therapies that address treatment gaps or for inventions that pave the way to new classes of drugs. Organisations competing for prizes might be expected to behave secretly to ensure that they are the ones to obtain ‘credit’ for the fruits of their work. However, progress in research is also driven by free exchange of information. It may be possible to design models that both reward R&D outputs and at the same time encourage complete and continuous openness with intermediate research outputs. There are now a number of examples of open collaborative public goods models ( Cukier 2003 ), such as those used for the Human Genome Project. The proponents of such models point to the success of GNU/Linux in the software field as evidence that major projects can be undertaken with radically different business models. One of the benefits of complete openness is that it allows independent and open evaluation of R&D outputs, which helps in the allocation of ‘credit’ whether in the form or prizes or new research grants. The open-access publishing movement ( Brown et al. 2003 ) has the potential to help in this process by allowing independent analysis of published science, which will help research funding agencies measure research outputs. Competitive Intermediators An R&D contribution norm, established by treaty, would ensure that the amount of money being spent on R&D is maintained. However, new mechanisms would be needed to collect the money to finance the R&D, as it would no longer come via drug sales. This could be via general taxation, although in countries with a private health insurance system this may be anathema. Many will also worry that a centralised national drug development agency taking decisions on R&D priorities and allocation of funds (via prizes or grants as discussed above) could easily become bureaucratic and inefficient. As a possible alternative, we propose a competitive financing scheme that would work through R&D investment intermediators. These R&D funds would be licensed and regulated (like pension funds). Their role would be to manage R&D assets on behalf of consumers. Individuals (or employers) would be required to make minimum contributions into R&D funds, much as there are mandatory contributions to social security or health insurance or to pension funds. Government would set the required contribution, but the individual (or employer) would be free to choose the particular intermediator that received their contributions. Intermediators would compete to attract funds to invest in R&D on the basis of their prowess for drug development and upon their priorities. Different business models for financing R&D could be tested in such a market, with intermediators experimenting with prize systems, direct investments in profit or nonprofit entities, open collaborative public good models, or other approaches. A Change for the Common Good We believe the economics of a change in the paradigm for funding R&D are highly favourable. Taken together, the two core steps of changing the trade framework and moving away from marketing monopolies can change the world in a positive way. We can raise global R&D levels as a matter of policy and ensure that resources flow into the areas of the greatest need, and we can do so knowing that the poor and the rich will have access to new inventions at marginal cost. Policy-makers will be weaned from their current unhealthy addiction to ever-higher levels of intellectual property rights as the only instrument to raise R&D levels, a path that has increasingly reached diminishing returns or become counterproductive. With new instruments to address the overall levels of R&D investment, policy-makers can more constructively address the well-known inefficiencies in the patent system without the fear that global R&D levels will suffer and explore alternative models ( Butler 2003a ). At the same time, the system of prescribing medicines will be transformed by a substantial reduction in the distorting influences of the current multibillion-dollar industry of marketing medicines to doctors and (increasingly) directly to the public. Similarly, without marketing monopolies to protect, there will be far less spent to influence the governments that set the rules that regulate such monopolies. If implemented worldwide, one of our most vexing ethical dilemmas can be resolved in a manner that actually promotes the Doha Declaration on TRIPS and Public Health mandate to encourage access to medicine for all. Box 1. Trade Agreements on Intellectual Property The most important is the World Trade Organisation (WTO) agreement on Trade-Related Aspects of Intellectual Property (TRIPS), which requires member countries to issue 20-year patents on all fields of technology. All but the least-developed countries must comply by 2005. Going much further than the TRIPS are a plethora of regional and bilateral ‘TRIPS-Plus’ trade agreements, pushed in particular by the United States, which require even higher levels of intellectual property protection, such as limitations on the use of compulsory licensing, a tool used by governments to override the strong exclusive rights of a patent in return for compensation to patent owners. In 2001 the WTO adopted the Doha Declaration on TRIPS and Public Health, which said that ‘the Agreement can and should be interpreted and implemented in a manner supportive of WTO Members’ right to protect public health and, in particular, to promote access to medicines for all'. In order to promote ‘access to medicines for all’, countries have to find new ways of financing R&D. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC340954.xml |
535809 | cDNA microarray analysis of bovine embryo gene expression profiles during the pre-implantation period | Background After fertilization, embryo development involves differentiation, as well as development of the fetal body and extra-embryonic tissues until the moment of implantation. During this period various cellular and molecular changes take place with a genetic origin, e.g. the elongation of embryonic tissues, cell-cell contact between the mother and the embryo and placentation. To identify genetic profiles and search for new candidate molecules involved during this period, embryonic gene expression was analyzed with a custom designed utero-placental complementary DNA (cDNA) microarray. Methods Bovine embryos on days 7, 14 and 21, extra-embryonic membranes on day 28 and fetuses on days 28 were collected to represent early embryo, elongating embryo, pre-implantation embryo, post-implantation extra-embryonic membrane and fetus, respectively. Gene expression at these different time points was analyzed using our cDNA microarray. Two clustering algorithms such as k -means and hierarchical clustering methods identified the expression patterns of differentially expressed genes across pre-implantation period. Novel candidate genes were confirmed by real-time RT-PCR. Results In total, 1,773 individual genes were analyzed by complete k -means clustering. Comparison of day 7 and day 14 revealed most genes increased during this period, and a small number of genes exhibiting altered expression decreased as gestation progressed. Clustering analysis demonstrated that trophoblast-cell-specific molecules such as placental lactogens (PLs), prolactin-related proteins (PRPs), interferon-tau, and adhesion molecules apparently all play pivotal roles in the preparation needed for implantation, since their expression was remarkably enhanced during the pre-implantation period. The hierarchical clustering analysis and RT-PCR data revealed new functional roles for certain known genes (dickkopf-1, NPM, etc) as well as novel candidate genes (AW464053, AW465434, AW462349, AW485575) related to already established trophoblast-specific genes such as PLs and PRPs. Conclusions A large number of genes in extra-embryonic membrane increased up to implantation and these profiles provide information fundamental to an understanding of extra-embryonic membrane differentiation and development. Genes in significant expression suggest novel molecules in trophoblast differentiation. | Background Embryogenesis in the period from fertilization to implantation involves various morphological, cellular, and biochemical changes related to genomic activity [ 1 ]. These changes include the elongation of embryonic tissues, cell-cell contact between the mother and the embryo, and placentation. The embryo begins to form the placenta around day 20 of gestation in the bovine [ 2 , 3 ], while embryonic trophoblast and endometrial cells tightly unite to form placentomes on day 30 [ 4 , 5 ]. The bovine embryos at the blastocyst stage are 100 to 200 μm in diameter, but become 20 to 30 cm long by the time of implantation. Embryonic cells undergo both proliferation and differentiation to form the fetus and placenta throughout early embryogenesis. Reprogramming of the genome may be completed and reset during these steps, with embryonic development progressing on to temporal and spatial gene expression [ 6 , 7 ]. The literature data indicate that a number of specific genes are expressed during the period from blastogenesis through implantation, such as interferon-τ (IFN-τ) [ 8 , 9 ], matrix metalloproteinases (MMPs) [ 10 ], heparanase [ 11 ], and retinoid X receptors [ 12 ]. Vast numbers of genes change their expression levels during this period to support the complex mechanisms of embryogenesis and implantation. Although numerous molecules participate in trophoblast differentiation and placentation, the precise molecular and genetic pathways which lead to the formation of placenta remain difficult to clarify. Nevertheless, recent data from animal cloning technology where a somatic or stem cell nucleus is transplanted to an enucleated unfertilized egg, strongly suggested that inappropriate expression of genes is possibly the main cause of early embryo loss [ 13 , 14 ]. These studies have shown that aberrant gene expression patterns were associated with placental abnormalities in several species. Thus, it is clear that the differentiation of trophoblastic cells and the formation of placenta are tightly controlled by mechanisms which precisely govern switch on and off of the appropriate set of genes during embryogenesis. Therefore, a detailed gene expression profile in the pre-implantation embryo provides insights into the molecular mechanisms that are vital to farther our knowledge of the embryogenesis and implantation processes. Unlike traditional molecular methods that have been used to look at a single gene at one time point, a complementary DNA (cDNA) microarray is an efficient tool for analyzing tens of thousands of genes in a single experiment and to identify groups of genes that are critical during early development. We have developed a bovine utero-placental specific cDNA microarray that contains various genes known to play key roles in extra-embryonic membrane development in the bovine [ 15 ]. To our knowledge, there has been no report to date that has evaluated the differential gene expression profile in the embryo as well as in extra-embryonic cells of the pre-implantation embryo in the bovine. Thus, we applied our cDNA microarray to detect gene expressions comprehensively during embryogenesis and implantation in the bovine. Furthermore we attempted to address the expression changes of the key genes that are involved in trophoblastic cell proliferation and differentiation. We first compared the gene expression profiles among blastocyst-stage embryos, elongated-stage embryos, and the trophoblastic tissue at the peri-implantation stage. We further characterized gene expression profiles between the trophoblastic tissue and embryo just after implantation. Comparison of expression differences of specific gene(s) provides important information on the potential role of the particular gene at specific time points of trophoblast development. Approaches such as this help identifying specific genes that are up- or down-regulated during trophoblast development. Although there is no defined standard method that has been developed for the analysis of microarray data, application of k -means [ 16 , 17 ] and hierarchical clustering [ 18 ] methods has been shown to yield reliable data. These two methods have been successfully used to obtain information from the embryo, cytotrophoblast, and uterine tissue during the early stages of pregnancy in humans [ 19 - 22 ] and mice [ 6 , 23 , 24 ]. We therefore applied k -means and hierarchical clustering methods to identify genes involved in trophoblastic differentiation. Materials and Methods Animals and sample collection Totally five Japanese black cows were slaughtered on Days 19, 21, 27 (two) and 28 of gestation (the day of artificial insemination was designated as Day 0), and collected embryonic membrane and fetus. Embryos were collected non-surgically on Days 7 and 14 of gestation from superovulated Japanese black cows. Namely, a follicle-stimulating hormone (FSH; Antrin-R10, Kawasaki Pharmaceutical, Kawasaki, Japan) was injected twice daily at doses of 5 mg, 3 mg, and 2 mg on the first, second, and third day of treatment, respectively. Luteolysis was induced by injecting prostaglandin F 2 α (cloprostenol 750 μg/cow; Fujita Pharmaceutical, Tokyo, Japan) on three days after the initial injection of FSH. Artificial insemination was performed both on the day of estrus (Day 0 of gestation) and the next morning, using frozen semen from Japanese black bulls. Twelve blastocysts collected on Day 7 were pooled to represent the Day 7 embryo (abbreviated as Day 7E), two expanded and elongated embryos collected on Day 14 were pooled to represent Day 14E, two embryos collected on Days 19 and 21 were used individually but designated as Day 21E (n = 2). Extra-embryonic membranes collected on Days 27 (two samples) and 28 (one sample) were used as Day 28 extra-embryonic membranes (28EEM) (n = 3), and two fetuses, excluding the extra-embryonic membranes, collected on Days 27 and 28 were used as the Day 28 fetus (28F) (n = 2). All samples were snap frozen in liquid nitrogen immediately after collection and stored at -80°C until RNA extraction. The days of collection were selected depending on the macro morphological changes of the embryos during the peri-implantation periods; bovine embryo passes into the uterus around a morula stage and develop to a blastocyst stage by Day 7. Embryo begins to elongate on Day 14 and form a long embryo of 20 to 30 cm in length on Day 20 just before implantation occurs. The embryo is easily identifiable around Day 30 around which placentome formation is visible [ 3 , 5 ]. All procedures for these animal experiments were carried out in accordance with the guidelines and ethics approved by the Animal Ethics Committee of the National Institute of Agrobiological Sciences for the use of animals. Sample RNA preparation RNA extraction The total RNA was isolated from the embryos using ISOGEN (NipponGene, Toyama, Japan) according to the manufacturer's instructions. Briefly, Samples were mixed and homogenized with ISOGEN. The aqueous phase was collected after the centrifugation. This phase was washed by chloroform for the RNA purification. The aqueous phase was collected after the centrifugation again. The total RNA pellet in the aqueous phase was obtained using the isopropanol sedimentation. Total RNA was directly used for the T7-based linear amplification for the cDNA microarray analysis. Messenger RNA (mRNA) were reverse transcribed by using the T7-oligo dT primer in the T7-based linear amplification system. T7-based linear amplification Linear-amplified antisense RNA (aRNA) was used for the cDNA microarray experiment. Because the amount of total RNA extracted was small from Day 7 blastocysts for the microarray analysis, mRNA from Day 7 blastocysts was amplified along with all other mRNA samples that were collected from Days 14 to 28 of gestation to eliminate technical variations and to efficiently compare the array data. We used a MessageAmp aRNA kit (Ambion, Austin, TX, USA) for the T7-based linear amplification of mRNA. Synthesis and purification of cDNA and aRNA were carried out according to the manufacturer's instructions. The aRNA amplification procedure consisted of (i) first-strand cDNA synthesis (reverse transcription of mRNA), (ii) second-strand cDNA synthesis, (iii) cDNA purification, (iv) in vitro transcription, and (v) aRNA purification. The aRNA amplification for Day 7 blastocysts was done with two steps purification. The comparison between first and second round amplification was r = 0.96 (data not shown). cDNA microarray Bovine utero-placental cDNA microarray For the series of present experiments we used a utero-placental custom cDNA microarray developed in our laboratory, as described previously [ 13 , 15 ]. Briefly, a cDNA library was constructed from mRNA isolated from endometrial (caruncular and intercaruncular endometrium) and placental tissues (cotyledonary and intercotyledonary fetal membrane) of Japanese Black cows on days 0 and 10 of the estrous cycle, and days 30, 60, 100 and 245 of gestation, respectively. PCR products of about 4,000 clones from the cDNA library which was normalized by hit-picking method [ 15 ] were spotted onto glass slides robotically. Simultaneously, the clones were sequenced by using the MegaBACE 1000 DNA Sequencing System (Amersham Pharmacia Biotech, Piscataway, NJ). The array contained 3,955 spots that were clustered into 1,738 unique genes on the basis of sequence analysis. The sequenced clones were compiled and annotated by basic local alignment search tool (BLASTn), and the clone that was positioned at the top of the hierarchical tree of genes by GenBank database matching was selected. The 1,738 annotated unique genes represented 816 known genes, 530 expressed sequencing tags (ESTs) and 392 unknown novel genes. The unknown sequences were submitted to DDBJ as ESTs. The DDBJ and GenBank accession numbers of genes on this custom microarray are BP106801 to BP113049 and AB098745 to AB099150, respectively. An additional 35 genes that were not included in the cDNA library but also spotted onto the cDNA microarray were used for analysis since these genes have been shown to be characteristically expressed during gestation in bovines and humans [ 9 , 25 - 29 ]. cDNA microarray hybridization We have previously reported our microarray hybridization procedures [ 15 ]. Briefly, two μg of amplified aRNA was reverse transcribed in the presence of cyanine 3 (Cy3) or Cy5 fluorescence dye (Amersham Biosciences, Buckinghamshire, UK) using SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA, USA) to make the hybridization probes. The reaction mixture was incubated for 2 hr at 42°C. The labeled probes were concentrated in a Microcon filter device (Millipore, Bendford, MA, USA), diluted in 15 μl hybridization solution (3.4 × SSC, 0.3% SDS, 20 μg poly (A), and 20 μg yeast tRNA), and applied to the microarray. Identical samples were labeled separately with either Cy3- or Cy5-dye. Thus, two hybridization reactions could be carried out with the same sample. The arrays were sequentially washed with 2 × SSC/0.5% SDS, 0.2 × SSC/0.5% SDS, and 0.2 × SSC solutions after 16 hr incubation at 65°C. The arrays were dried by centrifugation at 1,000 × g. Hybridization images were immediately scanned by a GenePix 4000B laser scanner (Axon Instrument, Union City, CA, USA) and analyzed with GenePix Pro 4.0 computer software. The data were viewed as a scatter plot between Cy3 and Cy5 intensities. Each color intensity value was individually normalized, and the average intensity value of each spot (gene) obtained from different hybridization reactions of the same sample was used for data analysis. Data normalization of the cDNA microarray Data normalization was performed by following the procedures described previously [ 30 , 31 ]. The local background intensity of each array spot was smoothed by local weight regression (Lowess) and subtracted from the spot intensity data. The subtracted intensity data were subjected to non-parametric regression and local variance normalization since non-parametric regression can reduce intensity-dependent biases. The accuracy is improved over that of linear regression if the points in the scatter plot of Cy3 versus (vs.) Cy5 are not distributed around a straight line. Minimum Information About a Microarray Experiment (MIAME; ) compliance was met by depositing all the data in the Gene Expression Omnibus (GEO) repository . The GEO accession numbers are as follows. Platform: GPL1221; Samples: GSM23324, GSM26510, GSM26511, GSM23327, GSM23328, GSM23329, GSM23330, GSM23331, and GSM23332; Series: GSE1414. Cluster analysis of the cDNA microarray data Data for individual genes were obtained by averaging the intensity values of analogous spots on the microarray. Data were log 2 transformed and used for cluster analysis. The TIGR (The Institute for Genome Research) MultiExperiment Viewer (MeV) program was used to derive the k -means and for hierarchical tree clustering analysis [ 32 ]. The general expression patterns of the 1,773 individual genes (including 35 manually spotted genes) were investigated by k -means algorithm. Data for each gene were represented by an eight-dimensional vector. K -means clustering was performed by division into 12 centroid centers. The distance between gene vectors was calculated by the cosine coefficient (vector angle). Hierarchical tree clustering was performed on the annotated and EST genes that demonstrated up-regulation when the comparison was made between Day 28EEM vs. Day 28F. Genes that displayed two-fold (and greater) differences were selected and the normalized expression of each sample was utilized for the cluster analysis. The distance between gene vectors was also calculated by the cosine coefficient. We adopted average linkage clustering in the hierarchical tree clustering method. Data used for the cluster analyses were as follows: Day 21E (n = 2; r = 0.84; P < 0.05); Day 28EEM (n = 3; r ≥ 0.90; P < 0.05); and Day 28F (n = 2; r = 0.90; P < 0.05). Other samples were used independently for cluster analysis. Real-time RT-PCR analysis Gene expression profiles derived from microarray analyses were confirmed quantitatively by real-time RT-PCR analysis. The selected genes for RT-PCR included two key known genes (placental lactogen-Ala (PL-Ala) and prolactin-related protein-1 (PRP-I)) and four apparently up-regulated ESTs, and compared their expression from Day 7 through Day 28. The ESTs selected have the GenBank accession numbers AW464053, AW465434, AW462349, and AW485575. The details of real-time RT-PCR procedures have been described in previous reports [ 11 , 33 ]. Briefly, fifty ng total RNA was reverse transcribed into cDNA for 30 min at 48°C by MultiScribe™ reverse transcriptase with an Oligo dT primer, dNTP mixture, MgCl 2 and an RNase inhibitor. Primer pairs and oligonucleotide probes labeled with a reporter fluorescence dye at the 5' end and a quencher fluorescence dye at the 3' end were designed using Primer Express computer software (Applied Biosystems, Foster City, CA, USA). The primers and probes for the selected genes are listed in Table 1 . The thermal cycling conditions included an initial incubation of samples at 50°C for two minutes and at 95°C for ten minutes, followed by forty cycles with each cycle at 95°C for fifteen seconds and at 60°C for one minute. The cycle threshold value (C T ) is related to the initial quantity of the target gene in each sample determined in real time using an ABI Prism 7700 sequence detector (Applied Biosystems). The relative difference in the initial amount of each mRNA species (or cDNA) was determined by comparing the C T values between the samples at different stages. Bovine GAPDH was used as endogenous control. The standard curve for each gene was generated by serial dilution of plasmid containing PL-Ala, PRP-I, four ESTs, or GAPDH cDNA to quantify mRNA concentrations. A ratio of PL-Ala, PRP-I and four ESTs mRNA to GAPDH mRNA was calculated to adjust for any variation in the RT-PCR reaction. All values are presented as mean ± SD. Data were analyzed initially by ANOVA and followed by either Tukey-Kramer multiple comparison test. P -values of <0.05 were considered significant. Table 1 Oligonucleotide primers and TaqMan probes used for real-time RT-PCR analysis Gene Primer or TaqMan probe Sequence Position PL-Ala (J02840) Forward 5' GCAACATTGGTGGCTAGCAA 3' 262-281 Reverse 5' GCCCTCGCCAAACTGTTTATTA 3' 339-317 Probe 5' CTATAGGCTCGCCAGGGAAATGTTCAATGA 3' 285-314 PRP-I (J02944) Forward 5' CAGACAGGTTTATGAATGCCGC 3' 458-479 Reverse 5' CGCAGGCAGTAGAACAGGTTAT 3' 541-520 Probe 5' TCCTCTGCATCATCTAGTCACGGAGCTG 3' 483-511 AW464053 Forward 5' AATATGCCCAGGGCAAACTG 3' 296-315 Reverse 5' TCGGGAGTTTGGAGGGAATT 3' 368-349 Probe 5' TCAATGCCATCAAGAGCTGCCACAC 3' 323-347 AW465434 Forward 5' ACATCTCCCTGAAAGTGAACCC 3' 287-308 Reverse 5' TCCATCCTTGCAGAAGTCTCCT 3' 369-348 Probe 5' CCCTGGAAGCTCATCTGCAATGTAAAGC 3' 316-343 AW462349 Forward 5' GCGTGGATGGTGTCCTACTTCTA 3' 216-238 Reverse 5' GCCACAACGAGAAACAGGAAA 3' 301-281 Probe 5' TGTCTGTTTGCCTTTACTGGTGAGCCCT 3' 240-267 AW485575 Forward 5' CCTCTGATGAAAGATTGGGAACAG 3' 195-218 Reverse 5' AAGTGCCAGAGATCTTGGCCT 3' 285-265 Probe 5' TTCTCCAAACCAACCACCACCAGCTG 3' 230-255 GAPDH (U85042) Forward 5' AAGGCCATCACCATCTTCCA 3' 178-197 Reverse 5' CCACTACATACTCAGCACCAGCAT 3' 253-230 Probe 5' AGCGAGATCCTGCCAACATCAAGTGG 3' 200-225 Results T7-based linear amplification We confirmed the reliability of this amplification method by comparing unamplified mRNA vs. amplified mRNA (aRNA) derived from the same bovine placental tissue after hybridization into the utero-placental cDNA microarray. We observed a significant correlation (r ≥ 0.81, P < 0.05; data not shown) between the unamplified mRNA and amplified aRNA. Confirmation experiments were conducted twice with reverse labeling (for a total of four times). Previous reports have also described substantial cDNA microarray population correlation between unamplified mRNA and amplified aRNA [ 34 , 35 ]. The correlation of our amplification was equivalent to previous studies [ 34 , 35 ]. Thus, amplified aRNAs can be used as representative materials in cDNA microarray experiments. Differential gene expression profiles determined by cluster analysis Sixty-six genes out of a total of 1,773 were exempted from cluster analysis due to low expression intensity values. Consequently, 1,707 genes, including 833 previously annotated, were classified into 12 categories by k -means clustering, as illustrated in Fig. 1 . Twelve k -means cluster profiles were summarized into three types, in which the gene-expression intensity (i) increased only from Day 7 to Day 14 and thereafter either remained constant or decreased slightly (clusters 2, 3, 12); (ii) Continued to increase until Day 21 (clusters 1, 6, 7, 8, 9, and 10); and (iii) progressively increased until Day 21 (clusters 4, 5, and 11). Clusters 1 and 7 contained more than 200 genes that included annotated and ESTs, whereas clusters 4, 5, and 11 contained fewer than 100 genes. Clusters 5 and 11 contained mostly members of the cytokine family, and clusters 4 and 5 contained ECM-related genes. In particular, trophoblast-specific genes, such as placental lactogens (PLs), prolactin-related proteins (PRPs), and pregnancy-associated glycoproteins (PAGs), were found in cluster 11. Figure 1 K -means clusters of the expression patterns of 1707 genes in a bovine embryo. Lines refer to the k -means of gene expression on Days 7E to 21E, Day 28EEM and Day 28F. The solid line refers to the k -means center of gene expression on Days 7E to 21E and Day 28EEM. The dotted line refers to the k -means center of Day 21E to Day 28F. Temporal specific gene expression in extra-embryonic membranes Pair-wise comparisons were made to identify genes displaying differential expression; the results are given in Fig. 2 . All data considered significant exhibited an increase (up-regulation; yellow) or decrease (down-regulation; blue) of at least two-fold. Top 20 known genes of the most or less expression over two-fold differences were listed in Tables 2 to 5 . All data of individual gene changes are available in an additional file on same web site of this paper or as supplemental tables (Suppl. Tables 1–4 in Additional files – see ). Figure 2 Gene expression correlations of bovine peri-implantation embryo between Day 7E vs. Day 14E, Day 7E vs. Day 21E, Day 7E vs. Day 28EEM, Day 7E vs. Day 28F, Day 14E vs. Day 21E, Day 14E vs. Day 28EEM, Day 14E vs. Day 28F, Day 21E vs. Day 28EEM, Day 21E vs. Day 28F, and Day 28EEM vs. Day 28F. The yellow areas highlight a greater than two-fold gene expression difference (up-regulated) between the X-axis and Y-axis samples. The blue areas highlight a greater than two-fold gene expression difference (down-regulated) between the X-axis and Y-axis samples. The gray areas highlight a 0.5- to 2-fold gene expression difference between the X-axis and Y-axis samples. Table 2 Top 20 known genes of the most or less expression over two-fold differences between Day 7E vs. Day 14E – see also additional file 1 Accession No. Gene name D14E/D7E K -means Classification Day 14E/Day 7E down-regulated genes (<0.5) L07872 Homo sapiens Jk-recombination signal binding protein 0.19 8 Apoptosis & Cell cycle NM_001294 Homo sapiens CLPTM1 0.23 4 Cytokine family NM_001021 Homo sapiens ribosomal protein S17 (RPS17) 0.25 8 Ribosomal M20866 Sus scrofa cofilin 0.25 1 Cytoskeleton M21683 Sus scrofa nonhistone protein HMG1 0.30 8 DNA binding protein AF020508 Bos taurus PAG-6 0.36 11 Cytokine family AF020507 Bos taurus PAG-5 0.39 11 Cytokine family X91755 Bos taurus cathepsin L 0.42 8 Oncogene & Tumor inhibitor NM_005722 Homo sapiens ARP2 actin-related protein 2 homolog (yeast) 0.43 8 Cytoskeleton AF210381 Bos taurus DDVit1 0.44 4 ECM & related AF020506 Bos taurus PAG-4 0.46 1 Cytokine family U21660 Bos taurus phosphatidylcholine transfer protein 0.47 4 L02897 Dog nonerythroid beta-spectrin 0.47 9 AF166124 Homo sapiens selenoprotein X 0.49 8 Day 14E/Day 7E up-regulated genes (2<) X14926 Mus musculus calreticulin 24.00 12 AB009282 Homo sapiens cytochrome b5 16.48 12 NM_004494 Homo sapiens hepatoma-derived growth factor (HDGF) 13.52 12 Oncogene & Tumor inhibitor AF000137 Bos taurus connective tissue growth factor precursor (CTGF) 12.18 12 Cytokine family NM_000365 Homo sapiens triosephosphate isomerase 1 (TPI1) 10.52 12 X89984 Homo sapiens BCL7A protein 10.31 12 Cytoskeleton AF217197 Homo sapiens FBP interacting repressor (FIR) 9.80 12 Transcriptional regulator NM_005720 Homo sapiens ARPC1B 9.45 3 Cytoskeleton NM_001404 Homo sapiens EEF1G 9.43 12 Transcriptional regulator AB003094 Bos taurus ferritin L subunit 9.38 12 X56503 Sus scrofa casein kinase II beta subunit (CKII beta) 8.55 12 Enzyme X13684 Bos taurus glutathione peroxidase (gpx1) 8.39 3 Enzyme NM_002436 Homo sapiens membrane protein, palmitoylated 1 (55 kD) 8.33 12 Membrane protein NM_005022 Homo sapiens profilin 1 (PFN1) 8.09 12 Cytoskelton X01809 Bos taurus cathepsin 3' terminus 8.08 12 Oncogene & Tumor inhibitor AF207664 Homo sapiens matrix metalloprotease (ADAMTS1) 8.03 3 ECM & related X56597 Homo sapiens humFib fibrillarin 7.97 12 AF182001 Bos taurus D4-GDP-dissociation inhibitor (D4-GDI) 7.82 3 NM_007317 Homo sapiens kinesin-like 4 (KNSL4) 7.66 2 M21044 Bos taurus MHC class I BoLA 7.65 2 Cytokine family All data of individual gene changes are available in Suppl. Table 1 of the Additional file. Table 3 Top 20 known genes of the most or less expression over two-fold differences between Day 14E vs. Day 21E. Accession No. Gene name D21E/D14E K -means Classification Day 21E/Day 14E down-regulated genes (<0.5) U21661 Rattus norvegicus myotrophin 0.28 2 Cytokine family Day 21E/Day 14E up-regulated genes (2<) AF004877 Homo sapiens pro-alpha 2(I) collagen (COL1A2) 16.08 4 ECM & related U21660 Bos taurus phosphatidylcholine transfer protein 10.13 4 X15112 Bos taurus cytochrome c oxidase subunit VIb (AED) 10.01 4 X59504 Bos taurus prolactin-like protein (PRP-VI) 9.62 11 Cytokine family AB008683 Bos taurus alpha2(I) collagen (COL1A2) 9.27 4 ECM & related AF105429 Ovis aries H19 7.81 4 X65210 Bos taurus microsatellite DNA 7.62 11 X15975 Bos taurus PRP-V 7.59 11 Cytokine family S74761 Bos taurus water channel protein CHIP29 7.44 9 Membrane protein AF196320 Bos taurus Interferon-tau1C 7.39 2 Cytokine family NM_001613 Homo sapiens actin, alpha 2 (ACTA2) 6.98 4 Cytoskeleton L06151 Bos taurus PAG-2 6.86 11 Cytokine family M32303 Bos taurus metalloproteinase inhibitor 6.82 11 ECM & related NM_001553 Homo sapiens IGFBP7 6.62 9 Cytokine family M21683 Sus scrofa nonhistone protein HMG1 6.58 8 DNA binding protein AF020508 Bos taurus PAG-6 6.54 11 Cytokine family AF125041 Ovis aries decorin 6.42 4 J02944 Bos taurus PRP-1 6.40 11 Cytokine family AB004800 Sus scrofa S100C protein 5.68 9 Cytokine family AF192336 Bos taurus PAG-19 5.47 11 Cytokine family All data of individual gene changes are available in Suppl. Table 2 of the Additional file. Table 4 Top 20 known genes of the most or less expression over two-fold differences between Day 21E vs. Day 28EEM. Accession No. Gene name D28EEM/D21E K -means Classification Day 28EEM/Day 21E down-regulated genes (<0.5) AF196320 Bos taurus Interferon-tau1C 0.07 2 Cytokine family AF033096 Avena sativa nonphototropic hypocotyl 1 (NPH1-1) 0.22 12 L10240 Homo sapiens EMMPRIN 0.24 7 ECM & related U21660 Bos taurus phosphatidylcholine transfer protein 0.27 4 X15112 Bos taurus cytochrome c oxidase subunit VIb (AED) 0.28 4 L34261 Bos taurus palmitoyl-protein thioesterase 0.29 6 Enzyme M26576 Homo sapiens alpha-1 collagen type IV 0.29 6 ECM & related NM_001747 Homo sapiens capping protein, gelsolin-like (CAPG) 0.32 2 Cytoskeleton U46064 Sus scrofa aldehyde reductase (ALR1) 0.33 3 Enzyme AF020513 Bos taurus PAG-11 0.33 12 Cytokine family M11120 Rat 28S rRNA 0.33 2 NM_004718 Homo sapiens COX7A2L 0.34 8 Enzyme AF144763 Bos taurus TIMP-1 0.35 3 ECM & related AF034607 Homo sapiens chloride channel ABP 0.36 8 Membrane protein NM_005556 Homo sapiens keratin 7 (KRT7) 0.36 2 Cytoskeleton AJ243656 Methanobacterium thermoautotrophicum ehbA-Q 0.37 2 M83104 Bos taurus cytochrome b5 reductase 0.37 2 Enzyme X59693 Bos taurus ubiquinol-cytrochrome-c reductase (subunit II) 0.38 6 Enzyme D84557 Homo sapiens HsMcm6 0.38 12 M77234 Homo sapiens ribosomal protein S3a 0.38 6 Ribosomal Day 28EEM/Day 21E up-regulated genes (2 <) AF020508 Bos taurus PAG-6 3.45 11 Cytokine family X01912 Goat epsilon I beta-globin 3.00 9 AF004133 Sus scrofa adipocyte membrane protein 2.66 1 Membrane protein M73961 Ovis aries PAG-1 2.35 6 Cytokine family J02840 Bos taurus placental lactogen (PL-Ala) 2.21 5 Cytokine family AF192336 Bos taurus PAG-19 2.17 11 Cytokine family AB005148 Bos taurus interleukin 1 (IL-1) receptor antagonist 2.13 1 Cytokine family D10989 Bos taurus endothelin ETB receptor 2.12 1 X59504 Bos taurus PRP-VI 2.08 11 Cytokine family M80328 Bos taurus PL-Val 2.01 5 Cytokine family Z11742 Bos taurus annexin XI 2.00 5 Apoptosis & Cell cycle All data of individual gene changes are available in Suppl. Table 3 of the Additional file. Table 5 Top 20 known genes of the most or less expression over two-fold differences between Day 28EEM vs. Day 28F. Accession No. Gene name D28F/D28EEM K -means Classification Day 28F/Day 28EEM down-regulated genes (<0.5) M73961 Ovis aries PAG-1 0.14 6 Cytokine family X89984 Homo sapiens BCL7A protein 0.14 12 Cytoskeleton X15975 Bos taurus PRP-V 0.15 11 Cytokine family AF079545 Ovis aries placental lactogen precursor (PL) 0.16 8 Cytokine family AF020509 Bos taurus PAG-7 0.19 11 Cytokine family J02840 Bos taurus placental lactogen (bPL-Ala) 0.20 5 Cytokine family X59504 Bos taurus PRP-VI 0.20 11 Cytokine family S72871 Homo sapiens GATA-2 transcription factor 0.20 8 Transcriptional regulator AF192336 Bos taurus PAG-19 0.22 11 Cytokine family AF020508 Bos taurus PAG-6 0.23 11 Cytokine family J02944 Bos taurus PRP-I 0.24 11 Cytokine family L06151 Bos taurus PAG-2 0.25 11 Cytokine family AF020512 Bos taurus PAG-10 0.27 5 Cytokine family AF020514 Bos taurus PAG-12 0.28 11 Cytokine family AF192334 Bos taurus PAG-17 0.30 11 Cytokine family AL133034 Homo sapiens clone DKFZp727K171 0.31 11 Z11742 Bos taurus annexin XI 0.31 5 Apoptosis & Cell cycle U89321 Homo sapiens nucleophosmin phosphoprotein (NPM) 0.32 5 X17614 Bos taurus 3 beta hydroxy-5-ene steroid dehydrogenase/delta 5-delta4 isomerase 0.32 5 Enzyme AF192333 Bos taurus PAG-16 0.32 11 Cytokine family Day 28F/Day 28EEM up-regulated genes (2<) L34261 Bos taurus palmitoyl-protein thioesterase 4.32 6 Enzyme AF000137 Bos taurus connective tissue growth factor precursor (CTGF) 2.85 12 Cytokine family NM_004458 Homo sapiens FACL4 transcript variant 1 2.84 7 Enzyme NM_006491 Homo sapiens NOVA1 transcript variant 3 2.82 1 Oncogene & Tumor inhibitor NM_006838 Homo sapiens methionine methionyl aminopeptidase 2 2.72 4 Enzyme AB028449 Homo sapiens helicase-MOI 2.65 9 DNA binding protein AL080102 Homo sapiens clone DKFZp564N1916 2.64 9 AF057300 Homo sapiens truncated RAD50 protein 2.63 3 AB043994 Bos taurus MMP-2 2.56 7 ECM & related NM_006719 Homo sapiens transcript variant ABLIM-m 2.53 1 Cytoskeleton AF195417 Homo sapiens DEAD-box protein abstrakt (ABS) 2.51 2 Z25531 Bos taurus repeat region DNA 2.41 3 AF113682 Homo sapiens clone FLB3436 PRO0868 2.40 4 S76474 Homo sapiens trkB 2.39 1 Cytokine family Y16533 Ovis aries IGF-II 2.39 7 Cytokine family M86739 Bos taurus neuropeptide Y receptor 2.38 7 Cytokine family AF043937 Homo sapiens DHAPAT 2.37 9 Enzyme AF144763 Bos taurus TIMP-1 protein 2.36 3 ECM & related NM_005563 Homo sapiens stathmin 1/oncoprotein 18 (STMN1) 2.36 4 Oncogene & Tumor inhibitor AF198487 Homo sapiens transcription factor LBP-1b 2.34 4 Transcriptional regulator All data of individual gene changes are available in Suppl. Table 4 of the Additional file. Day 7 to Day 14 A comparison between Day 7E and Day 14E clearly indicates that only 26 genes were down-regulated (14 annotated and 12 ESTs), with many genes (680) being up-regulated from Day 7 through Day 14E. The most significant genes during this period are listed in Table 2 and Suppl. Table 1 (see Additional files). They included 22 cytokine-related molecules, 44 enzymes, 21 transcriptional regulators, 16 oncogenes or tumor suppressor genes, 12 apoptosis or cell-cycle molecules, 10 heat-shock proteins (HSP), 9 cell-adhesion molecules, and 8 membrane proteins. A striking expression was found for several genes during this period, i.e. calrecticulin (X14926), which is a 55 kd calcium binding protein of the ER lumen, and caldesmon (X89984), an actin-binding protein, were two of the most significantly expressed genes. PAGs and IFN were also strongly expressed. Other cytokine molecules, such as HDGF (NM_00494), had their expression accented. Cell-function-related genes such as integrin, mucin, ADAMTS, keratin, and cytokeratin were found to be significantly increased in their expression. Day 14 to Day 21 A total of 452 genes were significantly up-regulated from Day 14 through Day 21, and half of these (226 genes) were annotated by BLASTn (Fig. 2 ). Only two genes were down-regulated, and they contained one annotated gene, myotrophin. The expression of various PL-related genes, including PRPs, was up-regulated during this period. The expression of extra-matrix-related genes, such as collagens, proteoglycans, MMPs, extracellular MMP inducer (EMMPRIN), and heparanase, and cell adhesion molecules, such as mucin, integrins, ezrin, and certain cytokines such as insulin-like growth factors (IGFs) and epidermal growth factor receptor (EGFR), were up-regulated. IFN-τ expression was emphasized. The individual gene expression data is provided in Table 3 and Suppl. Table 2 (see Additional files). Day 21 to Day 28 A comparison between Day 21E and Day 28EEM revealed that the expression of 14 genes increased slightly in Day 28EEM and 11 of these were annotated, which included PAGs, PLs, and IL-1. A total of 173 genes were significantly down-regulated in Day 28EEM, including 109 annotated genes; these genes were found mainly in k -means clusters 2, 3, 6, 8, and 12. The most significantly down-regulated gene during this critical period for implantation was IFN- . Down-regulation of certain MMP or matrix-related genes was also found (see Table 4 and Suppl. Table 3 in Additional files) in the top 50 genes of lowest expression. Comparison between extra-embryonic membranes on Day 28 and the fetus on Day 28 A total of 119 genes, including ESTs, decreased significantly in expression, with only 52 annotated genes among them in Day 28EEM (Fig. 2 ). However, in Day 28EEM, 74 up-regulated genes contained members of the cytokine family and molecules that play a role in cell-to-cell interactions, such as PLs, PRPs, PAGs, MHCs, and mucin, as indicated in Fig. 3 . The up-regulated genes included 35 ESTs, which exhibited expression profiles similar to the cytokines during the pre-implantation period in extra-embryonic membranes. In addition to these trophoblast-specific genes, which are known to be specific to extra-embryonic membranes, other interesting genes, such as dickkopf-1, which is a Wnt signal regulation molecule, grancalcin, which is a calcium binding protein, two actin-related proteins (NM_005720, X89984), and cell proliferation related genes (NM_006429, U89321), were specifically found in extra-embryonic membranes when compared to the fetus. The actual roles of these genes in extra-embryonic cell development are as yet unknown. Individual data are provided in Table 5 and Suppl. Table 4 in Additional file 1 . Figure 3 Hierarchical tree cluster of differentially expressed genes in the bovine embryo during the implantation period (Days 7 to 28 of gestation). Two-fold and greater differences in gene expression between Day 28EEM vs. Day 28F are selected in the expression profiles from all samples of Day 7E, 14E, 21E, 28EEM and 28F. Normalized and log 2 transformed expression data were used for the clustering analysis. The yellow cells indicate up-regulated genes in the median of the total value; the blue cells indicate down-regulated genes in the median of the total value. The black cells indicate no changes in expression. Candidate extra-embryonic membrane specific genes Genes related to trophoblast cell differentiation were anticipated to change their expression when the comparison was made between Day 28EEM and Day 28F. Most of the annotated genes of the 74 genes that increased over two-fold in Day 28EEM were PL-, PRP-, and PAG-related genes. The expression of 35 ESTs showed profiles similar to those of the annotated genes during the pre-implantation period, as determined by three hierarchical analyses (Fig. 3 ). An attempt was made to find new genes related to extra-embryonic cell lineages. Four ESTs (AW464053, AW465434, AW462349, and AW485575) were selected and analyzed by real-time RT-PCR. The quantitative data coincided with the microarray data as shown in Figs. 3 and 4 . The relative intensities of these ESTs, PL-Ala and PRP-I were similar to those of real-time RT-PCR. Figure 4 Real-time RT-PCR analysis of PL-Ala, PRP-I, and ESTs (GenBank accession No. AW464053, AW465434, AW462349, and AW485575) mRNA in bovine embryos. Gene expression on Days 7E to 21E, and 28EEM or 28F is provided. The expression of each mRNA was normalized to the expression of GAPDH mesured in same RNA preparation. The expression refers "means ± SD". Values with different letters are significantly different (P < 0.05). Discussion The major reproductive wastage in farm animals is early embryo loss, i.e. the anomalous development of embryos and/or an aberration of placentation [ 36 ]. Various technologies, such as artificial insemination, embryo transfer, and cloning, have been applied to bovine reproduction [ 13 , 37 ]. Precise knowledge of the gene expression profile during peri-implantation is necessary to reduce early losses and to improve the reproductive efficiency of these new technologies [ 6 , 38 - 40 ]. However, little is known about the complex molecular regulation of embryos and extra-embryonic membrane development in cattle. Thus, the genes to be profiled include new, functional gene candidates. We suggest an assessment method or key gene to help clarify the complex mechanisms in early embryo and trophoblast cell proliferation and differentiation. Various genes changes during embryonic development, particularly in extra-embryonic membranes, with specific morphological changes occur in bovine: a remarkable elongation of the embryonic membrane, fusion between fetal membrane and caruncular epithelium, endometrial reorganization [ 3 , 41 , 42 ]. Specific genes, like IFN-τ, PLs, PRPs, PAGs, IGFs, and IGFBPs have been studied as they take important roles around implantation in bovine [ 3 , 43 - 48 ]. However these molecules are essential for development of the embryo and the formation of extra-embryonic membranes, determination of their specific roles are still difficult to prove. Microarray analyses provided the time-dependant genes profiles in accord with the progress of gestation during the peri-implantation period. Most genes that expressed from the blastocyst stage to early placentation were up-regulated, while only 14 of the identified genes exhibited down-regulation (Fig 1 ). For example, selenoprotein X (AF166124) exerts a stimulative function on cell proliferation, particularly in the blastocyst, since its expression decreased on Day 14 compared to Day 7. This result coincided with that in silico work [ 49 ]. We selected genes related to heat-shock proteins (HSP) to interpret comprehensive genes expression, because they were frequently found in the list of significant genes and previous data had revealed their importance in early embryo development [ 50 ]. Ten HSP-related genes out of 333 annotated genes were up-regulated over two-fold on Day 14E in comparison to Day 7E. Similar genes were detected, i.e. seven on Day 21E and three on Day 28EEM. These results suggest a specific importance for HSPs in the early development of embryos. HSPs may disturb the coordination between the conceptus and endometrium in bovines and be related to the induction of early embryo loss [ 51 ]. Aberration of HSP90 expression in mice causes a placental abnormality that is closely related to trophoblast cell differentiation and proliferation [ 52 ]. Intensive IFN-τ expression was found during the implantation period (Days 14E to 21E). The trophoblast elongates and grows into a thread-like structure during this period and produces IFN- which inhibits luteolysis and plays a role in embryonic survival by mediating embryo-maternal crosstalk [ 8 , 43 , 53 - 56 ]. IFN-τ not only has an anti-luteolytic function enabling the establishment of gestation in cows, but apparently also participates in various other functions necessary for embryo survival [ 56 , 57 ]. It has been reported that IFN-τ has an immunoregulative effect [ 58 ] and also interferon modulates the expression of MHC class I antigen in mouse trophoblast cell cultures [ 59 ]. It is interesting that MHC I and its related genes, MHC II and β-microglobulin, exhibited increased expression in the present study only during the implantation period. Cell-adhesion molecules and cytokines, such as the integrin family, have been emphasized their importance for initiation of implantation [ 60 - 62 ]. Recently, Ezrin, a cytoskeletal-membrane linker molecule belonging to the ezrin-radixin-moesin (ERM) family, proved its importance for implantation in mice [ 63 ], exhibiting greater expression than in the pre-implantation stage in bovine blastocysts in the present study. The importance of adhesion molecule involvement via the adhesion-regulating molecules (ARM-1), ICAM-1, LECAM-1, Lu-ECAM-1, calcium, and integrin-binding protein (CIB) was suggested by their expression increasing around the expected implantation starting day (Day 21E). Numerous PAG, PL, and PRP genes, involved in the differentiation of trophoblastic cell lineage, were found among the 39 annotated genes of the 74 that exhibited over a two-fold difference between extra-embryonic membranes and the fetus on Day 28. The expression of most PAG family members increased towards Day 21E. Other binucleate cell-specific genes, such as PL and PRP family members, exhibited a coordinated expression with the PAG family of genes. These results confirm previous reports from this laboratory [ 3 , 13 , 33 , 64 ]. PAG-5, -15, -16, -17, and -19 were distributed to k -means cluster 11 in the present study, whereas PAG-8 was distributed to k -means cluster 3 (Fig. 1 ). The former group of PAG family members is apparently produced by binucleate cells, while those in the latter group may be produced by other cells of the trophectoderm [ 46 ]. Microarray data suggested various new candidate genes for extra-embryonic development even they have known function. Dickkopf-1 may inhibit Wnt proteins, which influence many aspects of embryonic development [ 65 ]. Annexins are a family of structurally related calcium-dependent phospholipid binding proteins [ 66 ]. Grancalcin is a Ca (2+)-binding protein [ 67 ]. Actin-related protein 2/3 (Arp2/3) complex may be related to the actin cytoskeleton [ 68 ]. Chaperonin containing t-complex polypeptide 1 (CCT complex) is essential for the maturation of cyclin E [ 69 ]. Nucleophosmin phosphoprotein (NPM) (U89321), is a major nucleolar protein that is 20 times more abundant in tumors or proliferating cells [ 70 ]. Finally, BCL7A exhibits homology with the actin-binding protein caldesmon [ 71 ]. The most straightforward explanation for present study is that EST genes that display similar expression patterns are functionally related [ 72 ]. AW464053, AW465434, AW462349, and AW485575 gene expression patterns were similar to the expression patterns of PL and PRP-I (Figs. 3 and 4 ). These ESTs were submitted from the cDNA library of Soares normalized bovine placenta (Lewin et al., unpublished). The EST of AW485575 was submitted from a library made from pooled tissue of Day 20 and Day 40 bovine embryos [ 73 ]. However, the function of all four ESTs remains obscure. The AW465434 gene has a 495 bp sequence and was found to be common for 21/21 bp (e-value = 1.1) in Homo sapiens in BAC clone RP11-1246C19 (AC102953). The AW462349 gene has 545 bp sequences and was found to be common for a slight 43/47 bp (e-value = 3 × 10 -7 ) in Sus scrofa clone RP44-519O7 (AC096884). The AW485575 gene has 436 bp sequences and was found to be common for a slight 157/181 bp (e-value = 6 × 10 -39 ) in Homo sapiens hypothetical protein MGC39389 (BC003531). The AW464053 gene expression patterns were similar to the expression pattern of PRP-I, and the gene sequence has a portion in common with an e-value = 1 × 10 -102 and 4 × 10 -31 (common sequence = 326/371, 87% and 105/116, 90%) for PRP-VI (X59504), or an e-value = 1 × 10 -95 (common sequence = 351/408, 86%) for PRP-III (M27240 or NM_174160). We suggest this EST was new member of bPRP. Conclusions This study provides developmental expression changes of a large number of genes in the bovine embryo during the peri-implantation period, with particular focus on genes that express in extra-embryonic membranes. The new gene candidates that were discussed here may address a new set of annotated genes and ESTs for embryonic differentiation and development. Participation of several known genes like Ezrin, CIB, Dickkopf-1, Grancalcin and EST (AW464053, AW465434 etc) in extra-embryonic membrane differentiation and development is elucidated by this microarray analysis. Fundamental investigations of this sort contribute significantly to a better understanding of the effects of various cultural conditions and cellular and genetic manipulation of embryos, including in vitro fertilization, in vitro maturation, and embryo cloning technology. Supplementary Material Additional File 1 Supplement tables list the whole genes which produced the expression difference in Fig. 2 and complement Tables 2–5 beneath (see text for details). The legends of supplemental tables are as follows: Supplement Table 1; Two-fold differentially expressed genes between Day 7E vs. Day 14E. Supplement Table 2; Two-fold differentially expressed genes between Day 14E vs. Day 21E. Supplement Table 3; Two-fold differentially expressed genes between Day 21E vs. Day 28EEM. Supplement Table 4; Two-fold differentially expressed genes between Day 28EEM vs. Day 28F. Click here for file | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC535809.xml |
516785 | Sodium nitroprusside and peroxynitrite effect on hepatic DNases: an in vitro and in vivo study | Background It has been documented that nitric oxide (NO) donor sodium nitroprusside (SNP) and authentic peroxynitrite are capable of promoting apoptosis in a number of different cell types. Various endonucleases have been proposed as candidates responsible for the internucleosomal cleavage of the genomic DNA observed during apoptosis, but the main effect is attributed to the alkaline-DNases (Mg 2+ - and caspase-dependent) and acid-DNase. The aim of this study was to examine an in vivo and in vitro possibility for alkaline- and acid-DNases to be activated by SNP and peroxynitrite. Results The effect on liver tissue alkaline and acid DNase activity together with the markers of tissue and plasma oxidative and nitrosative stress (lipid peroxidation, SH group content, carbonyl groups and nitrotyrosine formation) was investigated in plasma and liver tissue. The activity of liver alkaline DNase increased and that of acid DNase decreased after in vivo treatment with either SNP or peroxynitrite. A difference observed between the in vivo and in vitro effect of oxide donor (i.e., SNP) or peroxynitrite upon alkaline DNase activity existed, and it may be due to the existence of the "inducible" endonuclease. After a spectrophotometric scan analysis of purified DNA, it was documented that both SNP and peroxynitrite induce various DNA modifications (nitroguanine formation being the most important one) whereas DNA fragmentation was not significantly increased. Conclusion Alkaline DNase activation seems to be associated with the programmed destruction of the genome, leading to the fragmentation of damaged DNA sites. Thus, the elimination of damaged cells appears to be a likely factor in prevention against mutation and carcinogenesis. | Background In its response to tissue damage and inflammation induced by a variety of xenobiotics, endotoxins and disease states (such as viral hepatitis), post-ischemic and regenerative injury, the liver produces a large quantity of nitric oxide (NO). Nearly all cell types in liver tissue, including hepatocytes, Kupffer cells, stellate cells and endothelial cells, have the capacity for generating NO. It has been documented that NO is capable of promoting apoptosis in a number of different cell types, generally classified as cGMP-dependent or cGMP-independent [ 1 - 4 ]. The potential of chemical NO donor sodium nitroprusside (SNP) to induce apoptosis directly from NO liberation has been established in vitro [ 5 ]. The fact that NO is capable of triggering apoptosis is consistent with its ability to induce DNA damage, the inhibition of DNA synthesis and cell cycle arrest [ 6 , 7 ]. The reaction product formed between NO • and superoxide [i.e., peroxynitrite (ONOO - )] plays a critical role in the induction of inflammatory reaction and apoptosis, but is also associated with tumor promotion and/or progression. Potentially toxic levels of peroxynitrite can be achieved whenever NO • and O 2 .- production is stimulated, due to the fact that a 100-fold increase in the rate of peroxynitrite formation occurs for every 10-fold increase in NO • and O 2 .- concentration [ 8 ]. Apoptosis, frequently termed "programmed cell death", is the form of cell death that occurs in normal liver in the course of its development and organogenesis, and in adult liver during the renewal of hepatocytes. In addition, apoptosis can be triggered by several hepatotropic viruses and toxic drugs, as well as in various liver diseases and experimental liver conditions such as hepatic allograft rejection. Degradation of the nuclear DNA, a common phenomenon observed in many organisms throughout the evolutionary scale, is one of the best-characterized biochemical features of apoptotic cell death. It has been established that the cell undergoes epigenetic reprogramming in the D 1 phase of programmed cell death, the result of which is the activation of double-stranded DNA fragmentation in the F phase during which the nuclear morphology dramatically changes [ 9 ]. The cleavage of DNA may have a protective function in that it reduces the likelihood for genes in a potentially active site to be transferred from dying cells to the nuclei of viable neighboring cells. It is possible that various endonucleases exert a DNA degrading activity, as well as that many proteins can receive DNA degrading properties upon change of pH conditions [ 11 ]. The most important aspect of apoptosis is the universal property of some proteins to exert a dual function: the protection against proteolysis and the maintenance of the structure and function of normal cells. Being free from the inhibitory complex, however, these proteins may also contribute to protein or chromatin cleavage during apoptosis [ 12 , 13 ]. Changes in DNA degradation may lead to the pathogenesis in various disorders, such as liver cancer [ 14 , 15 ]. On the basis of the pharmacological data supporting the critical role of NO and peroxynitrite in apoptosis, current research studies have evaluated the activity of alkaline and acid DNase during the administration of SNP or of peroxynitrite, as well as the changes in numerous susceptible parameters of nitrosative stress, including SH group oxidation, carbonyl group formation, lipid peroxidation and DNA modification. An assay of enzyme activity was performed using liver tissue after in vivo administration and in vitro treatment of isolated rat hepatocytes or purified commercial enzymes either with SNP or authentic peroxynitrite. Results There are few data concerning the in vivo susceptibility of liver tissue to NO donor SNP and authentic peroxynitrite. A considerable attention has been paid to the establishment of in vivo tolerability and to the markers of apoptotic effects. Both NO and peroxynitrite can directly react with aromatic and sulfhydryl nucleophiles and nitrate aromatic residues. Sulfhydryl groups oxidation was documented in the plasma and, almost equally, in liver tissue. Peroxynitrite administration led to a more pronounced decrease in the concentration of other plasma free radical scavengers such as uric acid, and was followed by an increase in plasma nitrate concentration (Tables 1 and 2 ). According to the data suggesting that peroxynitrite decomposes rapidly to OH • and NO 2 • -like species at physiological pH, it was assumed that the carbonyl groups and the lipid peroxidation product [i.e., malondyaldehyde (MDA)] may play a significant role in liver cell toxicity. Neither plasma nor liver protein carbonyls showed any significant increase. This may be a likely consequence of the significant increase in aromatic amino acids nitration, presumably tyrosine the spectral contribution of which was substracted from the samples treated with 2,4-dinitrophenylhidrazine. Plasma and liver MDA concentrations were not significantly changed, either. The obtained results do not support the data suggesting that oxygen radicals, probably generated during cellular SNP metabolism, may mediate cell toxicity and apoptosis, but do confirm previous in vitro observations [ 16 ]. Plasma alanine aminotranferase (ALT) activity, used as the standard liver functional test, decreased in the peroxynitrite-treated group (Table 1 ). The activity of liver alkaline DNase increased and that of acid DNase decreased after in vivo treatment with either SNP or peroxynitrite (Table 2 ). Table 1 Plasma levels of investigated parameters of Sprague-Dawley rats after in vivo treatment with nitric oxide donor (SNP) and peroxynitrite. Data expressed as Mean (SD); n = 8 per group. Parameters Control SNP Peroxynitrite Lipid peroxidation (MDA) (μmol/l) 7.10 (0.86) 6.78 (2.11) 6.64 (2.85) SH groups (μmol/l) 180.47 (36.08) 150.84 (4.78) 110.40 (11.96)* NO 2 +NO 3 (μmol/l) 45.47 (18.80) 66.82 (4.12) 81.94 (10.18) Uric acid (μmol/l) 64.82 (4.18) 22.34 (8.32) 17.50 (3.21)* Carbonyl groups (μmol/g protein) 179.59 (28.50) 144.90 (67.40) 169.06 (14.88) Nitrotyrosine (nmol/g protein) 96.02 (26.53) 136.17 (27.47) 135.11 (4.64) ALT (U/l) 31.9 (4.15) 34.44 (4.96) 28.5 (5.15) Male Sprague-Dawley rats three months old were allocated into three different groups. Either peroxynitrite (0.5 ml/kg BW of 30 mmol solution) or SNP (10 mg/kg BW) in a volume of 100 μl were administrated in bolus in systemic circulation. The control group received physiological saline solution at the same volume. All procedures were carried out as described in Methods. Asterisk: significantly different from the control (p < 0.05). Table 2 Liver levels of investigated parameters of Sprague-Dawley rats after in vivo treatment with nitric oxide donor (SNP) and peroxynitrite. Data expressed as Mean (SD); n = 8 per group. Parameters Control SNP Peroxynitrite Lipid peroxidation (MDA) (μmol/g protein) 54.42 (8.68) 52.5 (4.34) 58.2 (6.11) SH groups (μmol/g protein) 636.04 (74.58) 538.67 (82.87) 505.52 (41.43) NO 2 +NO 3 (μmol/g protein) 6.15 (2.91) 6.04 (2.61) 5.74 (1.03) Carbonyl groups (μmol/g protein) 4.14 (0.13) 4.25 (0.12) 4.20 (0.26) Nitrotyrosine (nmol/g protein) 10.26 (2.62) 11.40 (1.24) 16.62 (3.58) DNA (mg/g tissue) 10.71 (1.75) 12.65 (1.01) 12.15 (1.25) DNA fragmentation (%) 11.99 (0.33) 13.21 (1.43) 13.0 (0.89) Alkaline-DNase (U/g protein) 11.95 (0.57) 15.30 (0.72) 16.45 (1.04) Acid-DNase (U/g protein) 15.28 (3.42) 12.34 (0.35) 9.34 (0.76) Male Sprague-Dawley rats three months old were allocated into three different groups. Either peroxynitrite (0.5 ml/kg BW of 30 mmol solution) or SNP (10 mg/kg BW) in a volume of 100 μl were administrated in bolus in systemic circulation. The control group received physiological saline solution at the same volume. All procedures were carried out as described in Methods. The ultraviolet spectra of DNA were obtained by spectrophotometric scanning between 230–500 nm on a scan detecting system. According to the data by Yermilov et al. [ 17 ], the appearance of a peak between 375 and 405 nm (depending on pH) corresponds to 8-nitroguanine. In the scan analysis (Figure 1 ), the peak was between 390 and 410 nm with a maximum absorbance at 405 nm in alkaline conditions (DNA extract was adjusted to pH 10). This peak may correspond to the formation of nitro-derivatives, most probably of 8-nitroguanine. The nitroguanine peak at 405 nm was particularly apparent in peroxynitrite-treated samples. Figure 1 The peak appearance of isolated liver DNA . The extraction oftissue DNA was performed according to the method of Wannemacher et al. [50], modified by Setaro & Morley [51], with the protein and nucleic acid precipitation by using ice-cold trichloroacetic acid after lipid extraction. DNA was separated from proteins by hydrolisis of resulting pellet at 96 ± 1°C for 45 min. Samples were analyzed for DNA concentration by ultraviolet absorption difference at 260 and 290 nm. Purified DNA was employed for spectral changes, monitored by using Beckman spectrophotometer. On the basis of the data obtained by Yermilov et al. [13], the appearance of a peak between 375 and 405 nm (depending on pH) corresponds to 8-nitroguanine. The peak appearance was between 390 and 410 nm with the maximum absorbance at 405 nm, obtained in alkaline conditions (DNA extract was adjusted to pH 10). In vivo administration of SNP or peroxynitrite tended to increase the rate of DNA fragmentation, but it was not statistically significant. The rate was estimated according to the percentage of DNA resisting centrifugation at 27 000 g (Table 2 ). After in vitro exposure of isolated hepatocytes to SNP or peroxynitrite, the activity of both alkaline and acid DNase decreased in a dose-dependent fashion (Figure 2 ). During in vitro incubation of purified enzymes DNase I and DNase II with SNP or peroxynitrite, a dose-dependent decrease of enzyme activity was also documented (Figure 3 ). Figure 2 The activity of alkaline and acid DNase after in vitro treatment of isolated hepatocytes with NO donor (SNP) or peroxynitrite . The isolation of hepatocytes was done according to the method already published [42], by using a 1% collagenase dissolved in RPMI 1640 medium. Hepatocytes, isolated from 8 Male Sprague-Dawley rats, were dissolved in a physiological saline solution in a concentration of approximately 10 8 cells/ml. They were divided into seven groups (each comprising 8 samples), exposed to either SNP (0.1, 1 and 10 mmol) or peroxynitrite (0.03, 0.3 and 3 mmol) for a period of 1 hour at 37°C. Given in vitro concentrations were calculated according to the literature data [38]. The activity of alkaline and acid-DNase was measured by the methods of Bartholeyns et al. [43] and acid soluble nucleotides were determined spectrophotometrically at 260 nm. The enzyme activity was expressed as U/g protein. Data (n = 8) in graph is putted as: Mean + SD. Figure 3 The activity of commercial DNase I and DNase II after in vitro treatment with NO donor (SNP) or peroxynitrite. Purified enzymes DNase I (E.C. 3.1.21.1) and DNase II (E.C. 3.1.22.1) were dissolved in physiological saline solution. Hepatocytes, isolated from 8 Male Sprague-Dawley rats, were dissolved in a physiological saline solution in a concentration of approximately 10 8 cells/ml. They were divided into seven groups (each comprising 8 samples) exposed to either SNP (0.1, 1 and 10 mmol) or peroxynitrite (0.03, 0.3 and 3 mmol). The reducing agent (cysteine 1 mmol) was added to SNP to induce in vitro NO release [39]. The activity of alkaline and acid-DNase was measured by the methods of Bartholeyns et al. [43] and acid soluble nucleotides were determined spectrophotometrically at 260 nm. The defined units for purified DNase I and DNase II (increase in absorbance of 0.001/min in a sample containing 0.132 mg DNA, pH 7.4 or pH 5 and 3 ml of reaction mixture) were obtained from the Sigma catalogue label. Data (n = 8) in graph is putted as: Mean + SD. Discussion NO • , a free radical gaseous molecule is one of the simplest compounds found to be continuously produced in humans and animals. It can be derived from L-arginine through the enzyme nitric oxide synthase (NOS) and by different NO donors, including SNP. NO has been shown to play an unprecedented range of roles in biological systems, acting as a universal intracellular and transcellular signaling molecule and the regulator of vascular tone, cell proliferation and apoptosis [ 18 - 20 ]. Peroxynitrite is a strong, relatively long-lived oxidant with a half-life of approximately 0.5–1 s under physiological conditions. Our study confirmed that in both plasma and liver tissue peroxynitrite causes a rapid oxidation of sulfhydryl groups and thioethers, as well as the nitration and hydroxylation of aromatic compounds (Tables 1 and 2 ). A chronic exposure of hepatocytes to reactive nitrogen species exhibits a cytotoxic and cytostatic activity leading to functional and morphological alterations [ 8 , 21 ]. Cell death after exposure to different NO-donors such as SNP has been to date established through the expression of tumor suppressor gene p53 and pro-apoptotic genes such as bax, cyclin-dependent kinase inhibitor p21, the inhibited expression of anti-apoptotic protein bcl-2, the inhibited NF-κB binding activity, ERK and p-38-dependent cytochrome c release, and caspase-3 activation [ 22 - 24 ]. In contrast, the anti-apoptotic effects of NO may be mediated through the mechanisms such as blockade of the recruitment of pro-caspase-9 to the Apaf-1 apoptosome, stimulation of c-GMP-dependent protein kinase, control of mitochondrial permeability transition, induction of the heat shock protein HSP 70, and interaction with the ceramide pathway [ 25 , 26 ]. The prolonged damage of p53 gene by peroxynitrite has been associated with tumor formation. Recent results by Vincent and Maiese [ 3 ] indicate that NO donor SNP (at 300 μmol concentration) is capable of inducing strong apoptotic effects via DNA fragmentation and induction of Mg 2+ -dependent endonuclease activity in the culture of neuronal cells. In our in vivo study (Table 2 ), the activity of alkaline DNase increased within 24 h after exposure to SNP (achieving approximately a similar blood concentration of about 250 μmol) or to authentic peroxynitrite. Several molecules involved in nuclear DNA fragmentation have been detected and characterized based on their ionic sensitivity. Besides the presence of constitutive Ca 2+ /Mg 2+ -dependent endonucleases, a great deal of endonuclease activity within a 7.2–8.0 pH range most probably represents the inducible form of DNase. The molecular weights of the constitutive (NO-independent) and inducible (NO-dependent) endonuclease are similar, as well as their optimum pH range (7.5–8.0). A likely conclusion is that Mg 2+ -dependent endonuclease seems to be a result of de novo synthesized or the pre-existing Ca 2+ /Mg 2+ -dependent endonuclease activation. Up to now, several Mg 2+ - or Ca 2+ /Mg 2+ -dependent alkaline DNases (DNase I) with an optimum activity within the range of 7.5–9.5 have been purified. Some of them, including specific caspase3-activated DNase (CAD), are active upon release of the specific inhibitor ICAD [ 27 , 28 ]. DNase gamma has been documented as a critical component of apoptotic machinery, in that it cleaves the chromosomal DNA into nucleosomal units, thus leading to DNA ladder formation [ 29 ]. The alkaline DNase, active only during apoptosis, has been documented to be inherent to cyclophilins (A, B and C) as well, irrespective of their protein folding (peptidylprolyl cis-trans-isomerase) activity. All of them have the ability to degrade the supercoiled, single stranded and double stranded DNA [ 30 , 31 ]. Besides alkaline DNases, the cation-independent endonuclease with an optimum activity at pH 5, known as acid or DNase II, was identified. One leucocyte elastase inhibitor (LEI) can also exert an acid DNase activity after post-translational modification through the proteolytic cleavage [ 32 ]. The specific involvement of DNase II in physiological nuclear degradation during apoptosis could not be excluded upon decrease of intracellular pH values below 7 with a proton ionophore. Three potential N -nitrosylation sites are important for DNase II regulation [ 32 , 33 ]. Since our experimental data indicated a decrease in acid DNase activity 24 h after exposure to SNP or peroxynitrite (Table 2 ), the inhibition of DNase II may be explained by the nitrosylation of its susceptible sites. Indeed, when isolated hepatocytes were exposed to SNP or peroxynitrite for 1 h, a dose-dependent inhibition of DNase II was also documented (Fig. 2 ). The same result was obtained after exposure of purified enzyme to SNP (in the presence of the reducing agent cysteine 1 mmol) or peroxynitrite (Fig. 3 ). The formation of 8-nitroguanine, 8-oxo-deoxyguanine and oxazolone and the oxidative modification of 2'-deoxyribose into TBA-responsive compounds are the most prominent nucleotide modifications after reactive nitrogen species attack [ 34 , 35 ]. A highly potential mutagenic product 8-nitroguanine can be depurinated yielding apurinic sites capable of inducing GC→TA transversions, GC→CG transversions and deletions [ 17 , 36 ]. The appearance of the nitroguanine peak during the scan analysis of purified DNA at 405 nm was documented in our study (Fig. 1 ). The rate of DNA fragmentation tended to be increased, but the difference was not significant (Table 2 ). Conclusions In vivo administrated SNP and peroxynitrite increase the activity of alkaline DNase. They also induced DNA modifications, such as nitroguanine formation. The obtained DNase activation seems to be associated with the programmed destruction of the genome and cell death. Given the above results and observations, the elimination of damaged hepatic cells appears to be a likely factor in prevention against mutation and carcinogenesis. Methods Chemicals SNP, DNA, DNase I (E.C. 3.1.21.1.) and DNase II (E.C. 3.1.22.1) were obtained from Sigma-Aldrich Company. RPMI-1640, fetal calf serum (FCS) and collagenase were purchased from ICN (Costa Mesa, CA). Authentic peroxynitrite was freshly synthesized by the quench-flow technique [ 37 ] and its concentration was monitored in alkaline solution before use in each experiment by measuring the extinction coefficient at 302 nm [ 38 ]. All other chemicals were of the highest purity range. In vivo study Twenty-four male Sprague-Dawley rats, three months old, were divided into three different groups, each comprising 8 animals. Either SNP (10 mg/kg BW) or peroxynitrite (0.5 ml/kg BW of 30 mmol solution) in a volume of 100 μl were administrated in bolus in systemic circulation by intraventricular injection under penthobarbital sodium anesthesia. The concentrations were calculated according to the literature data concerning their in vivo tolerability and in vitro ability to induce apoptotic effects [ 39 , 40 ]. The calculation of peroxynitrite intra-arterial concentration (6 nmol) was done according to its biological half-life of about 0.6 s, cardiac output of 40 ml/min/100 g and circulating volume of 20 ml and 250 g of rat BW [ 41 ]. The corresponding control group received physiological saline solution in the same volume. The rats were killed 24 h afterwards, under the same anesthesia. Blood was collected from the abdominal aorta and livers were quickly removed, frozen and homogenised on ice. Isolation of hepatocytes The isolation of hepatocytes was done according to a method already published [ 42 ], by using a 1% collagenase dissolved in RPMI 1640 medium. Collagenase was inhibited by using 10% FCS and cells were washed twice in physiological saline solution. Hepatocytes were isolated from 8 Male Sprague-Dawley rats. They were dissolved in a physiological saline solution in a concentration approximately 10 8 cells/ml. They were divided into seven groups (each comprising 8 samples), exposed to either SNP (0.1, 1 and 10 mmol) or peroxynitrite (0.03, 0.3 and 3 mmol) for a period of 1 hour at 37°C. Given in vitro concentrations were calculated according to the literature data [ 38 ]. Purified enzymes DNase I and DNase II were dissolved in physiological saline solution, exposed to the same concentrations of SNP and peroxynitrite, except that the reducing agent (cysteine 1 mmol) was added to SNP to induce in vitro NO release [ 39 ]. Methods for alkaline and acid-DNase The activity of alkaline and acid-DNase was measured by the methods of Bartholeyns et al. [ 43 ] and acid soluble nucleotides were determined spectrophotometrically at 260 nm. The enzyme activity was expressed as U/g protein, for tissue and cell samples. The defined units for purified DNase I and DNase II (increase in absorbance of 0.001 / min in a sample containing 0.132 mg DNA, pH 7.4 or pH 5 and 3 ml of reaction mixture) were obtained from the Sigma catalogue label. Extraction of DNA and proteins The extraction of tissue DNA and proteins was performed according to the method of Wannemacher et al. [ 44 ] modified by Setaro & Morley [ 45 ] by protein and nucleic acid precipitation using ice-cold trichloroacetic acid (TCA), 0.6 N, after lipid extraction. RNA and DNA were isolated by using cold 60% perchloric acid (PCA). DNA was separated from proteins by hydrolysis of resulting pellet at 96 ± 1°C for 45 min after adding 0.5 N PCA. Tissue protein content was measured according to the Lowry et al. procedure [ 46 ]. Samples were analysed for DNA concentration by an ultraviolet absorption difference at 260 and 290 nm. Purified DNA was employed for spectral changes monitored by using Beckman DU 530 spectrophotometer. Protein carbonyls and protein nitrotyrosine were measured in plasma proteins and the remaining protein pellet according to the method of Oliver et al. [ 47 ] modified by Tien et al. [ 48 ]. DNA fragmentation assay was performed according to the method of Jones et al. [ 49 ] based on the percentage of DNA resisting centrifugation at 27 000 g for 20 min. The proportion is expressed as percentage of the total DNA in the uncentrifugated sample. Protein carbonyls were quantified by spectrophotometric measurement of their 2,4 dinitrophenylhydrazone derivatives (ε 370 nm = 22000 M-1 cm -1 ). The difference between the spectrum of the DNPH-treated sample and that of the HCl control was determined and expressed as μmol DNPH/g protein. As nitrotyrosine also absorbs at 370 nm, it was measured according to its spectral contribution at 370 nm. Plasma and tissue SH groups were measured by using DTMB according to the Elman method [ 50 ]. Plasma and tissue lipid peroxidation product MDA was measured according to the method of Ohkava et al. [ 51 ]. Nitrates were measured according to the method of Navarro-Gonzales et al. [ 52 ]. Plasma uric acid and ALT were measured using the Synchron analyzer. Statistics Statistical analysis was made with the software SPSS. The effect of treatments was firstly evaluated by one-way ANOVA. If there was a significant effect, experimental data sets were compared against the control group by the Dunnett post hoc test. Significance level was set at α = 0.05. Data were normally distributed with equal variances among groups. Authors' contributions GK carried out the in vivo and in vitro experiments, culture experiments and wrote the paper. RP and GN carried out DNA spectral analysis. TC performed measurement of SH groups and lipid peroxides. IS performed the measurement of nitrates and nitrites. TJ assisted during in vivo and in vitro experiments and did the graphical presentation. DS performed statistical analysis and assisted during in vivo experiments. DP and RK assisted during in vitro experiments and participated in the design of the study. All the authors read and approved the final manuscript. | /Users/keerthanasridhar/biomedlm/data/PMC000xxxxxx/PMC516785.xml |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.