Statistical modeling of interlocus interactions in a complex disease: rejection of the multiplicative model of epistasis in type 1 diabetes

Learning epistatic polygenic phenotypes with Boolean interactions

Detecting epistatic drivers of human phenotypes is a considerable challenge. Traditional approaches use regression to sequentially test multiplicative interaction terms involving pairs of genetic variants. For higher-order interactions and genome-wide large-scale data, this strategy is computationally intractable. Moreover, multiplicative terms used in regression modeling may not capture the form of biological interactions. Building on the Predictability, Computability, Stability (PCS) framework, we introduce the epiTree pipeline to extract higher-order interactions from genomic data using tree-based models. The epiTree pipeline first selects a set of variants derived from tissue-specific estimates of gene expression. Next, it uses iterative random forests (iRF) to search training data for candidate Boolean interactions (pairwise and higher-order). We derive significance tests for interactions, based on a stabilized likelihood ratio test, by simulating Boolean tree-structured null (no epistasis) and alternative (epistasis) distributions on hold-out test data. Finally, our pipeline computes PCS epistasis p-values that probabilisticly quantify improvement in prediction accuracy via bootstrap sampling on the test set. We validate the epiTree pipeline in two case studies using data from the UK Biobank: predicting red hair and multiple sclerosis (MS). In the case of predicting red hair, epiTree recovers known epistatic interactions surrounding MC1R and novel interactions, representing non-linearities not captured by logistic regression models. In the case of predicting MS, a more complex phenotype than red hair, epiTree rankings prioritize novel interactions surrounding HLA-DRB1, a variant previously associated with MS in several populations. Taken together, these results highlight the potential for epiTree rankings to help reduce the design space for follow up experiments.

Targeting Neuroplasticity, Cardiovascular, and Cognitive-Associated Genomic Variants in Familial Alzheimer's Disease

The identification of novel genetic variants contributing to the widespread in the age of onset (AOO) of Alzheimer’s disease (AD) could aid in the prognosis and/or development of new therapeutic strategies focused on early interventions. We recruited 78 individuals with AD from the Paisa genetic isolate in Antioquia, Colombia. These individuals belong to the world largest multigenerational and extended pedigree segregating AD as a consequence of a dominant fully penetrant mutation in the PSEN1 gene and exhibit an AOO ranging from the early 1930s to the late 1970s. To shed light on the genetic underpinning that could explain the large spread of the age of onset (AOO) of AD, 64 single nucleotide polymorphisms (SNP) associated with neuroanatomical, cardiovascular, and cognitive measures in AD were genotyped. Standard quality control and filtering procedures were applied, and single- and multi-locus linear mixed-effects models were used to identify AOO-associated SNPs. A full two-locus interaction model was fitted to define how identified SNPs interact to modulate AOO. We identified two key epistatic interactions between the APOE*E2 allele and SNPs ASTN2-rs7852878 and SNTG1-rs16914781 that delay AOO by up to ~ 8 years (95% CI 3.2–12.7, P = 1.83 × 10⁻³) and ~ 7.6 years (95% CI 3.3–11.8, P = 8.69 × 10⁻⁴), respectively, and validated our previous finding indicating that APOE*E2 delays AOO of AD in PSEN1 E280 mutation carriers. This new evidence involving APOE*E2 as an AOO delayer could be used for developing precision medicine approaches and predictive genomics models to potentially determine AOO in individuals genetically predisposed to AD.

Neuroimaging genomics in psychiatry-a translational approach

Neuroimaging genomics is a relatively new field focused on integrating genomic and imaging data in order to investigate the mechanisms underlying brain phenotypes and neuropsychiatric disorders. While early work in neuroimaging genomics focused on mapping the associations of candidate gene variants with neuroimaging measures in small cohorts, the lack of reproducible results inspired better-powered and unbiased large-scale approaches. Notably, genome-wide association studies (GWAS) of brain imaging in thousands of individuals around the world have led to a range of promising findings. Extensions of such approaches are now addressing epigenetics, gene–gene epistasis, and gene–environment interactions, not only in brain structure, but also in brain function. Complementary developments in systems biology might facilitate the translation of findings from basic neuroscience and neuroimaging genomics to clinical practice. Here, we review recent approaches in neuroimaging genomics—we highlight the latest discoveries, discuss advantages and limitations of current approaches, and consider directions by which the field can move forward to shed light on brain disorders.

Ptpn22 and Cd2 Variations Are Associated with Altered Protein Expression and Susceptibility to Type 1 Diabetes in Nonobese Diabetic Mice

By congenic strain mapping using autoimmune NOD.C57BL/6J congenic mice, we demonstrated previously that the type 1 diabetes (T1D) protection associated with the insulin-dependent diabetes (Idd)10 locus on chromosome 3, originally identified by linkage analysis, was in fact due to three closely linked Idd loci: Idd10, Idd18.1, and Idd18.3. In this study, we define two additional Idd loci—Idd18.2 and Idd18.4—within the boundaries of this cluster of disease-associated genes. Idd18.2 is 1.31 Mb and contains 18 genes, including Ptpn22, which encodes a phosphatase that negatively regulates T and B cell signaling. The human ortholog of Ptpn22, PTPN22, is associated with numerous autoimmune diseases, including T1D. We, therefore, assessed Ptpn22 as a candidate for Idd18.2; resequencing of the NOD Ptpn22 allele revealed 183 single nucleotide polymorphisms with the C57BL/6J (B6) allele—6 exonic and 177 intronic. Functional studies showed higher expression of full-length Ptpn22 RNA and protein, and decreased TCR signaling in congenic strains with B6-derived Idd18.2 susceptibility alleles. The 953-kb Idd18.4 locus contains eight genes, including the candidate Cd2. The CD2 pathway is associated with the human autoimmune disease, multiple sclerosis, and mice with NOD-derived susceptibility alleles at Idd18.4 have lower CD2 expression on B cells. Furthermore, we observed that susceptibility alleles at Idd18.2 can mask the protection provided by Idd10/Cd101 or Idd18.1/Vav3 and Idd18.3. In summary, we describe two new T1D loci, Idd18.2 and Idd18.4, candidate genes within each region, and demonstrate the complex nature of genetic interactions underlying the development of T1D in the NOD mouse model.

Genome-wide interaction analysis reveals replicated epistatic effects on brain structure

The discovery of several genes that affect the risk for Alzheimer's disease ignited a worldwide search for single-nucleotide polymorphisms (SNPs), common genetic variants that affect the brain. Genome-wide search of all possible SNP-SNP interactions is challenging and rarely attempted because of the complexity of conducting approximately 10(11) pairwise statistical tests. However, recent advances in machine learning, for example, iterative sure independence screening, make it possible to analyze data sets with vastly more predictors than observations. Using an implementation of the sure independence screening algorithm (called EPISIS), we performed a genome-wide interaction analysis testing all possible SNP-SNP interactions affecting regional brain volumes measured on magnetic resonance imaging and mapped using tensor-based morphometry. We identified a significant SNP-SNP interaction between rs1345203 and rs1213205 that explains 1.9% of the variance in temporal lobe volume. We mapped the whole brain, voxelwise effects of the interaction in the Alzheimer's Disease Neuroimaging Initiative data set and separately in an independent replication data set of healthy twins (Queensland Twin Imaging). Each additional loading in the interaction effect was associated with approximately 5% greater brain regional brain volume (a protective effect) in both Alzheimer's Disease Neuroimaging Initiative and Queensland Twin Imaging samples.

Polymorphism in the innate immune receptor SIRPα controls CD47 binding and autoimmunity in the nonobese diabetic mouse

The signal regulatory protein (SIRP) locus encodes a family of paired receptors that mediate both activating and inhibitory signals and is associated with type 1 diabetes (T1D) risk. The NOD mouse model recapitulates multiple features of human T1D and enables mechanistic analysis of the impact of genetic variations on disease. In this study, we identify Sirpa encoding an inhibitory receptor on myeloid cells as a gene in the insulin-dependent diabetes locus 13.2 (Idd13.2) that drives islet inflammation and T1D. Compared to T1D-resistant strains, the NOD variant of SIRPα displayed greater binding to its ligand CD47, as well as enhanced T cell proliferation and diabetogenic potency. Myeloid cell-restricted expression of a Sirpa transgene accelerated disease in a dose-dependent manner and displayed genetic and functional interaction with the Idd5 locus to potentiate insulitis progression. Our study demonstrates that variations in both SIRPα sequence and expression level modulate T1D immunopathogenesis. Thus, we identify Sirpa as a T1D risk gene and provide insight into the complex mechanisms by which disease-associated variants act in concert to drive defined stages in disease progression.

An omnibus permutation test on ensembles of two-locus analyses can detect pure epistasis and genetic heterogeneity in genome-wide association studies

This article presents the ability of an omnibus permutation test on ensembles of two-locus analyses (2LOmb) to detect pure epistasis in the presence of genetic heterogeneity. The performance of 2LOmb is evaluated in various simulation scenarios covering two independent causes of complex disease where each cause is governed by a purely epistatic interaction. Different scenarios are set up by varying the number of available single nucleotide polymorphisms (SNPs) in data, number of causative SNPs and ratio of case samples from two affected groups. The simulation results indicate that 2LOmb outperforms multifactor dimensionality reduction (MDR) and random forest (RF) techniques in terms of a low number of output SNPs and a high number of correctly-identified causative SNPs. Moreover, 2LOmb is capable of identifying the number of independent interactions in tractable computational time and can be used in genome-wide association studies. 2LOmb is subsequently applied to a type 1 diabetes mellitus (T1D) data set, which is collected from a UK population by the Wellcome Trust Case Control Consortium (WTCCC). After screening for SNPs that locate within or near genes and exhibit no marginal single-locus effects, the T1D data set is reduced to 95,991 SNPs from 12,146 genes. The 2LOmb search in the reduced T1D data set reveals that 12 SNPs, which can be divided into two independent sets, are associated with the disease. The first SNP set consists of three SNPs from MUC21 (mucin 21, cell surface associated), three SNPs from MUC22 (mucin 22), two SNPs from PSORS1C1 (psoriasis susceptibility 1 candidate 1) and one SNP from TCF19 (transcription factor 19). A four-locus interaction between these four genes is also detected. The second SNP set consists of three SNPs from ATAD1 (ATPase family, AAA domain containing 1). Overall, the findings indicate the detection of pure epistasis in the presence of genetic heterogeneity and provide an alternative explanation for the aetiology of T1D in the UK population.

Genetic interactions among Idd3, Idd5.1, Idd5.2, and Idd5.3 protective loci in the nonobese diabetic mouse model of type 1 diabetes

In the NOD mouse model of type 1 diabetes, insulin-dependent diabetes (Idd) loci control the development of insulitis and diabetes. Independently, protective alleles of Idd3/Il2 or Idd5 are able to partially protect congenic NOD mice from insulitis and diabetes, and to partially tolerize islet-specific CD8(+) T cells. However, when the two regions are combined, mice are almost completely protected, strongly suggesting the existence of genetic interactions between the two loci. Idd5 contains at least three protective subregions/causative gene candidates, Idd5.1/Ctla4, Idd5.2/Slc11a1, and Idd5.3/Acadl, yet it is unknown which of them interacts with Idd3/Il2. Through the use of a series of novel congenic strains containing the Idd3/Il2 region and different combinations of Idd5 subregion(s), we defined these genetic interactions. The combination of Idd3/Il2 and Idd5.3/Acadl was able to provide nearly complete protection from type 1 diabetes, but all three Idd5 subregions were required to protect from insulitis and fully restore self-tolerance. By backcrossing a Slc11a1 knockout allele onto the NOD genetic background, we have demonstrated that Slc11a1 is responsible for the diabetes protection resulting from Idd5.2. We also used Slc11a1 knockout-SCID and Idd5.2-SCID mice to show that both loss-of-function alleles provide protection from insulitis when expressed on the SCID host alone. These results lend further support to the hypothesis that Slc11a1 is Idd5.2.

Exhaustive search of the SNP-sNP interactome identifies epistatic effects on brain volume in two cohorts

The SNP-SNP interactome has rarely been explored in the context of neuroimaging genetics mainly due to the complexity of conducting approximately 10(11) pairwise statistical tests. However, recent advances in machine learning, specifically the iterative sure independence screening (SIS) method, have enabled the analysis of datasets where the number of predictors is much larger than the number of observations. Using an implementation of the SIS algorithm (called EPISIS), we used exhaustive search of the genome-wide, SNP-SNP interactome to identify and prioritize SNPs for interaction analysis. We identified a significant SNP pair, rs1345203 and rs1213205, associated with temporal lobe volume. We further examined the full-brain, voxelwise effects of the interaction in the ADNI dataset and separately in an independent dataset of healthy twins (QTIM). We found that each additional loading in the epistatic effect was associated with approximately 5% greater brain regional brain volume (a protective effect) in both the ADNI and QTIM samples.

A cooperative interaction between LPHN3 and 11q doubles the risk for ADHD

In previous studies of a genetic isolate, we identified significant linkage of attention deficit hyperactivity disorder (ADHD) to 4q, 5q, 8q, 11q and 17p. The existence of unique large size families linked to multiple regions, and the fact that these families came from an isolated population, we hypothesized that two-locus interaction contributions to ADHD were plausible. Several analytical models converged to show significant interaction between 4q and 11q (P<1 × 10(-8)) and 11q and 17p (P<1 × 10(-6)). As we have identified that common variants of the LPHN3 gene were responsible for the 4q linkage signal, we focused on 4q-11q interaction to determine that single-nucleotide polymorphisms (SNPs) harbored in the LPHN3 gene interact with SNPs spanning the 11q region that contains DRD2 and NCAM1 genes, to double the risk of developing ADHD. This interaction not only explains genetic effects much better than taking each of these loci effects by separated but also differences in brain metabolism as depicted by proton magnetic resonance spectroscopy data and pharmacogenetic response to stimulant medication. These findings not only add information about how high order genetic interactions might be implicated in conferring susceptibility to develop ADHD but also show that future studies of the effects of genetic interactions on ADHD clinical information will help to shape predictive models of individual outcome.

Cellular mechanisms of restored β-cell tolerance mediated by protective alleles of Idd3 and Idd5

Type 1 diabetes genes within the interleukin (IL)-2, cytotoxic T-lymphocyte--associated protein 4 (CTLA-4), and natural resistance-associated macrophage protein (NRAMP1) pathways influence development of autoimmune diabetes in humans and NOD mice. In NOD mice, when present together, protective alleles encoding IL-2, Idd3 candidate gene, CTLA-4, NRAMP1, and acetyl-coenzyme A dehydrogenase, long-chain (ACADL) (candidate genes for the Idd5.1, Idd5.2, and Idd5.3 subregions) provide nearly complete diabetes protection. To define where the protective alleles of Idd3 and the Idd5 subregions must be present to protect from diabetes and tolerize islet-specific CD8(+) T cells, SCID mice were reconstituted so that the host and lymphocytes expressed various combinations of protective and susceptibility alleles at Idd3 and Idd5. Although protective Idd3 alleles in the lymphocytes and protective Idd5 alleles in the SCID host contributed most significantly to CD8 tolerance, both were required together in both lymphocyte and nonlymphocyte cells to recapitulate the potent diabetes protection observed in intact Idd3/5 mice. We conclude that genetic regions involved in autoimmune disease are not restricted in their influence to individual cell types. Even a single protective gene product, such as IL-2, must be expressed in both the lymphocytes and dendritic cells to exert its full extent of disease protection. These studies highlight the pleiotropic effects of genes that determine autoimmune disease susceptibility.

A novel method to detect gene-gene interactions in structured populations: MDR-SP

Complex diseases are presumed to be the result of multiple genes and environmental factors, which emphasize the importance of gene - gene and gene - environment interactions. Traditional parametric approaches are limited in their ability to detect high-order interactions and handle sparse data, and standard stepwise procedures may miss interactions with undetectable main effects. To address these limitations, the multifactor dimensionality reduction (MDR) method was developed. MDR is well suited for examining high-order interactions and detecting interactions without main effects. Like most statistical methods in genetic association studies, MDR may also lead to a false positive in the presence of population stratification. Although many statistical methods have been proposed to detect main effects and control for population stratification using genomic markers, not many methods are available to detect interactions and control for population stratification at the same time. In this article, we developed a novel test, MDR in structured populations (MDR-SP), to detect the interactions and control for population stratification. MDR-SP is applicable to both quantitative and qualitative traits and can incorporate covariates. We present simulation studies to demonstrate the validity of the test and to evaluate its power.

Association of FcgR2a, but not FcgR3a, with inflammatory bowel diseases across three Caucasian populations

BACKGROUND

The Fc receptors II and III (FcgR2a, and FcgR3a) play a crucial role in the regulation of the immune response. The FcgR2a*519GG and FcgR3a*559CC genotypes have been associated with several autoimmune diseases including systemic lupus erythematosus, rheumatoid arthritis, nephritis, and possibly to type I diabetes, and celiac disease. In a large multicenter, two-stage study of 6570 people, we tested whether the FcgR2a and FcgR3a genes were also involved in inflammatory bowel disease (IBD), which includes Crohn's disease (CD) and ulcerative colitis (UC).

METHODS

We genotyped the FcgR2a*A519G and FcgR3a*A559C functional variants in 4205 IBD patients in six well-phenotyped Caucasian IBD cohorts and 2365 ethnically matched controls recruited from the Netherlands, Spain, and New Zealand.

RESULTS

In the initial Dutch study we found a significant association of FcgR2a genotypes with IBD (P-genotype = 0.02); while the FcgR2a*519GG was more common in controls (23%) than in IBD patients (18%; odds ratio [OR] = 0.75; 95% confidence interval [CI] 0.61-0.92; P = 0.004). This association was corroborated by a combined analysis across all the study populations (Mantel-Haenszel [MH] OR = 0.84; 0.74-0.95; P = 0.005) in the next stage. The Fcgr2a*GG genotype was associated with both UC (MH-OR = 0.84; 0.72-0.97; P = 0.01) and CD (MH-OR = 0.84; 0.73-0.97; P = 0.01), suggesting that this genotype confers a protective effect against IBD. There was no association of FcgR3a*A559C genotypes with IBD, CD, or UC in any of the three studied populations.

CONCLUSIONS

The FcgR2a*519G functional variant was associated with IBD and reduced susceptibility to UC and to CD in Caucasians. There was no association between FcgR3a*5A559C and IBD, CD or UC.

Nonobese diabetic congenic strain analysis of autoimmune diabetes reveals genetic complexity of the Idd18 locus and identifies Vav3 as a candidate gene

We have used the public sequencing and annotation of the mouse genome to delimit the previously resolved type 1 diabetes (T1D) insulin-dependent diabetes (Idd)18 interval to a region on chromosome 3 that includes the immunologically relevant candidate gene, Vav3. To test the candidacy of Vav3, we developed a novel congenic strain that enabled the resolution of Idd18 to a 604-kb interval, designated Idd18.1, which contains only two annotated genes: the complete sequence of Vav3 and the last exon of the gene encoding NETRIN G1, Ntng1. Targeted sequencing of Idd18.1 in the NOD mouse strain revealed that allelic variation between NOD and C57BL/6J (B6) occurs in noncoding regions with 138 single nucleotide polymorphisms concentrated in the introns between exons 20 and 27 and immediately after the 3' untranslated region. We observed differential expression of VAV3 RNA transcripts in thymocytes when comparing congenic mouse strains with B6 or NOD alleles at Idd18.1. The T1D protection associated with B6 alleles of Idd18.1/Vav3 requires the presence of B6 protective alleles at Idd3, which are correlated with increased IL-2 production and regulatory T cell function. In the absence of B6 protective alleles at Idd3, we detected a second T1D protective B6 locus, Idd18.3, which is closely linked to, but distinct from, Idd18.1. Therefore, genetic mapping, sequencing, and gene expression evidence indicate that alteration of VAV3 expression is an etiological factor in the development of autoimmune beta-cell destruction in NOD mice. This study also demonstrates that a congenic strain mapping approach can isolate closely linked susceptibility genes.

Epistasis and its implications for personal genetics

The widespread availability of high-throughput genotyping technology has opened the door to the era of personal genetics, which brings to consumers the promise of using genetic variations to predict individual susceptibility to common diseases. Despite easy access to commercial personal genetics services, our knowledge of the genetic architecture of common diseases is still very limited and has not yet fulfilled the promise of accurately predicting most people at risk. This is partly because of the complexity of the mapping relationship between genotype and phenotype that is a consequence of epistasis (gene-gene interaction) and other phenomena such as gene-environment interaction and locus heterogeneity. Unfortunately, these aspects of genetic architecture have not been addressed in most of the genetic association studies that provide the knowledge base for interpreting large-scale genetic association results. We provide here an introductory review of how epistasis can affect human health and disease and how it can be detected in population-based studies. We provide some thoughts on the implications of epistasis for personal genetics and some recommendations for improving personal genetics in light of this complexity.

Gene-gene interactions in the NOD mouse model of type 1 diabetes

Human genome wide association studies (GWAS) have recently identified at least four new, non-MHC-linked candidate genes or gene regions causing type one diabetes (T1D), highlighting the need for functional models to investigate how susceptibility alleles at multiple common genes interact to mediate disease. Progress in localizing genes in congenic strains of the nonobese diabetic (NOD) mouse has allowed the reproducible testing of gene functions and gene-gene interactions that can be reflected biologically as intrapathway interactions, for example, IL-2 and its receptor CD25, pathway-pathway interactions such as two signaling pathways within a cell, or cell-cell interactions. Recent studies have identified likely causal genes in two congenic intervals associated with T1D, Idd3, and Idd5, and have documented the occurrence of gene-gene interactions, including "genetic masking", involving the genes encoding the critical immune molecules IL-2 and CTLA-4. The demonstration of gene-gene interactions in congenic mouse models of T1D has major implications for the understanding of human T1D since such biological interactions are highly likely to exist for human T1D genes. Although it is difficult to detect most gene-gene interactions in a population in which susceptibility and protective alleles at many loci are randomly segregating, their existence as revealed in congenic mice reinforces the hypothesis that T1D alleles can have strong biological effects and that such genes highlight pathways to consider as targets for immune intervention.

An ensemble learning approach jointly modeling main and interaction effects in genetic association studies

Complex diseases are presumed to be the results of interactions of several genes and environmental factors, with each gene only having a small effect on the disease. Thus, the methods that can account for gene-gene interactions to search for a set of marker loci in different genes or across genome and to analyze these loci jointly are critical. In this article, we propose an ensemble learning approach (ELA) to detect a set of loci whose main and interaction effects jointly have a significant association with the trait. In the ELA, we first search for "base learners" and then combine the effects of the base learners by a linear model. Each base learner represents a main effect or an interaction effect. The result of the ELA is easy to interpret. When the ELA is applied to analyze a data set, we can get a final model, an overall P-value of the association test between the set of loci involved in the final model and the trait, and an importance measure for each base learner and each marker involved in the final model. The final model is a linear combination of some base learners. We know which base learner represents a main effect and which one represents an interaction effect. The importance measure of each base learner or marker can tell us the relative importance of the base learner or marker in the final model. We used intensive simulation studies as well as a real data set to evaluate the performance of the ELA. Our simulation studies demonstrated that the ELA is more powerful than the single-marker test in all the simulation scenarios. The ELA also outperformed the other three existing multi-locus methods in almost all cases. In an application to a large-scale case-control study for Type 2 diabetes, the ELA identified 11 single nucleotide polymorphisms that have a significant multi-locus effect (P-value=0.01), while none of the single nucleotide polymorphisms showed significant marginal effects and none of the two-locus combinations showed significant two-locus interaction effects.

Genome-wide microarray expression analysis of CD4+ T Cells from nonobese diabetic congenic mice identifies Cd55 (Daf1) and Acadl as candidate genes for type 1 diabetes

NOD.Idd3/5 congenic mice have insulin-dependent diabetes (Idd) regions on chromosomes 1 (Idd5) and 3 (Idd3) derived from the nondiabetic strains B10 and B6, respectively. NOD.Idd3/5 mice are almost completely protected from type 1 diabetes (T1D) but the genes within Idd3 and Idd5 responsible for the disease-altering phenotype have been only partially characterized. To test the hypothesis that candidate Idd genes can be identified by differential gene expression between activated CD4+ T cells from the diabetes-susceptible NOD strain and the diabetes-resistant NOD.Idd3/5 congenic strain, genome-wide microarray expression analysis was performed using an empirical Bayes method. Remarkably, 16 of the 20 most differentially expressed genes were located in the introgressed regions on chromosomes 1 and 3, validating our initial hypothesis. The two genes with the greatest differential RNA expression on chromosome 1 were those encoding decay-accelerating factor (DAF, also known as CD55) and acyl-coenzyme A dehydrogenase, long chain, which are located in the Idd5.4 and Idd5.3 regions, respectively. Neither gene has been implicated previously in the pathogenesis of T1D. In the case of DAF, differential expression of mRNA was extended to the protein level; NOD CD4+ T cells expressed higher levels of cell surface DAF compared with NOD.Idd3/5 CD4+ T cells following activation with anti-CD3 and -CD28. DAF up-regulation was IL-4 dependent and blocked under Th1 conditions. These results validate the approach of using congenic mice together with genome-wide analysis of tissue-specific gene expression to identify novel candidate genes in T1D.

Interactions between Idd5.1/Ctla4 and other type 1 diabetes genes

Two loci, Idd5.1 and Idd5.2, that determine susceptibility to type 1 diabetes (T1D) in the NOD mouse are on chromosome 1. Idd5.1 is likely accounted for by a synonymous single nucleotide polymorphism in exon 2 of Ctla4: the B10-derived T1D-resistant allele increases the expression of the ligand-independent isoform of CTLA-4 (liCTLA-4), a molecule that mediates negative signaling in T cells. Idd5.2 is probably Nramp1 (Slc11a1), which encodes a phagosomal membrane protein that is a metal efflux pump and is important for host defense and Ag presentation. In this study, two additional loci, Idd5.3 and Idd5.4, have been defined to 3.553 and 78 Mb regions, respectively, on linked regions of chromosome 1. The most striking findings, however, concern the evidence we have obtained for strong interactions between these four disease loci that help explain the association of human CTLA4 with T1D. In the presence of a susceptibility allele at Idd5.4, the CTLA-4 resistance allele causes an 80% reduction in T1D, whereas in the presence of a protective allele at Idd5.4, the effects of the resistance allele at Ctla4 are modest or, as in the case in which resistance alleles at Idd5.2 and Idd5.3 are present, completely masked. This masking of CTLA-4 alleles by different genetic backgrounds provides an explanation for our observation that the human CTLA-4 gene is only associated with T1D in the subgroup of human T1D patients with anti-thyroid autoimmunity.

Detecting association using epistatic information

Genetic association studies have been less successful than expected in detecting causal genetic variants, with frequent non-replication when such variants are claimed. Numerous possible reasons have been postulated, including inadequate sample size and possible unobserved stratification. Another possibility, and the focus of this paper, is that of epistasis, or gene-gene interaction. Although unlikely that we may glean information about disease mechanism, based purely upon the data, it may be possible to increase our power to detect an effect by allowing for epistasis within our test statistic. This paper derives an appropriate "omnibus" test for detecting causal loci whist allowing for numerous possible interactions and compares the power of such a test with that of the usual main effects test. This approach differs from that commonly used, for example by Marchini et al. [2005], in that it tests simultaneously for main effects and interactions, rather than interactions alone. The alternative hypothesis being tested by the "omnibus" test is whether a particular locus of interest has an effect on disease status, either marginally or epistatically and is therefore directly comparable to the main effects test at that locus. The paper begins by considering the direct case, in which the putative causal variants are observed and then extends these ideas to the indirect case in which the causal variants are unobserved and we have a set of tag single nucleotide polymorphisms (tag SNPs) representing the regions of interest. In passing, the derivation of the indirect omnibus test statistic leads to a novel "indirect case-only test for interaction".

Retrospective analysis of main and interaction effects in genetic association studies of human complex traits

Background

The etiology of multifactorial human diseases involves complex interactions between numerous environmental factors and alleles of many genes. Efficient statistical tools are demanded in identifying the genetic and environmental variants that affect the risk of disease development. This paper introduces a retrospective polytomous logistic regression model to measure both the main and interaction effects in genetic association studies of human discrete and continuous complex traits. In this model, combinations of genotypes at two interacting loci or of environmental exposure and genotypes at one locus are treated as nominal outcomes of which the proportions are modeled as a function of the disease trait assigning both main and interaction effects and with no assumption of normality in the trait distribution. Performance of our method in detecting interaction effect is compared with that of the case-only model.

Results

Results from our simulation study indicate that our retrospective model exhibits high power in capturing even relatively small effect with reasonable sample sizes. Application of our method to data from an association study on the catalase -262C/T promoter polymorphism and aging phenotypes detected significant main and interaction effects for age-group and allele T on individual's cognitive functioning and produced consistent results in estimating the interaction effect as compared with the popular case-only model.

Conclusion

The retrospective polytomous logistic regression model can be used as a convenient tool for assessing both main and interaction effects in genetic association studies of human multifactorial diseases involving genetic and non-genetic factors as well as categorical or continuous traits.

Application of multi-locus analytical methods to identify interacting loci in case-control studies

To identify interacting loci in genetic epidemiological studies the application of multi-locus methods of analysis is warranted. Several more advanced classification methods have been developed in the past years, including multiple logistic regression, sum statistics, logic regression, and the multifactor dimensionality reduction method. The objective of our study was to apply these four multi-locus methods to simulated case-control datasets that included a variety of underlying statistical two-locus interaction models, in order to compare the methods and evaluate their strengths and weaknesses. The results showed that the ability to identify the interacting loci was generally good for the sum statistic method, the logic regression and MDR. The performance of the logistic regression was more dependent on the underlying model and multiple comparison adjustment procedure. However, identification of the interacting loci in a model with two two-locus interactions of common disease alleles with relatively small effects was impaired in all methods. Several practical and methodological issues that can be considered in the application of these methods, and that may warrant further research, are identified and discussed.

Allelic variation in GAD1 (GAD67) is associated with schizophrenia and influences cortical function and gene expression

Cortical GABAergic dysfunction has been implicated as a key component of the pathophysiology of schizophrenia and decreased expression of the gamma-aminobutyric acid (GABA) synthetic enzyme glutamic acid decarboxylase 67 (GAD(67)), encoded by GAD1, is found in schizophrenic post-mortem brain. We report evidence of distorted transmission of single-nucleotide polymorphism (SNP) alleles in two independent schizophrenia family-based samples. In both samples, allelic association was dependent on the gender of the affected offspring, and in the Clinical Brain Disorders Branch/National Institute of Mental Health (CBDB/NIMH) sample it was also dependent on catechol-O-methyltransferase (COMT) Val158Met genotype. Quantitative transmission disequilibrium test analyses revealed that variation in GAD1 influenced multiple domains of cognition, including declarative memory, attention and working memory. A 5' flanking SNP affecting cognition in the families was also associated in unrelated healthy individuals with inefficient BOLD functional magnetic resonance imaging activation of dorsal prefrontal cortex (PFC) during a working memory task, a physiologic phenotype associated with schizophrenia and altered cortical inhibition. In addition, a SNP in the 5' untranslated (and predicted promoter) region that also influenced cognition was associated with decreased expression of GAD1 mRNA in the PFC of schizophrenic brain. Finally, we observed evidence of statistical epistasis between two SNPs in COMT and SNPs in GAD1, suggesting a potential biological synergism leading to increased risk. These coincident results implicate GAD1 in the etiology of schizophrenia and suggest that the mechanism involves altered cortical GABA inhibitory activity, perhaps modulated by dopaminergic function.

Both additivity and epistasis control the genetic variation for fruit quality traits in tomato

The effect of a gene involved in the variation of a quantitative trait may change due to epistatic interactions with the overall genetic background or with other genes through digenic interactions. The classical populations used to map quantitative trait loci (QTL) are poorly efficient to detect epistasis. To assess the importance of epistasis in the genetic control of fruit quality traits, we compared 13 tomato lines having the same genetic background except for one to five chromosome fragments introgressed from a distant line. Six traits were assessed: fruit soluble solid content, sugar content and titratable acidity, fruit weight, locule number and fruit firmness. Except for firmness, a large part of the variation of the six traits was under additive control, but interactions between QTL leading to epistasis effects were common. In the lines cumulating several QTL regions, all the significant epistatic interactions had a sign opposite to the additive effects, suggesting less than additive epistasis. Finally the re-examination of the segregating population initially used to map the QTL confirmed the extent of epistasis, which frequently involved a region where main effect QTL have been detected in this progeny or in other studies.

Modeling gene-gene interactions using graphical chain models

OBJECTIVE

To investigate whether graphical chain models are suitable to detect gene-gene interaction under different biological models.

METHODS

We conducted a simulation study comparing graphical chain models with logistic regression models regarding their ability to detect underlying biological interaction models. For both methods, we attempted to capture simulation data following 12 different biological models. We used 10 statistical models for both methods. Of the 12 different biological models, four contained no interaction effects, two were multiplicative, and six were epistasis models. For each situation, the choice for a statistical model was based on global model fit as judged by two different information criteria, the BIC and the AIC.

RESULTS

Both methods failed in most of the scenarios to capture the gene-gene interaction present in the simulation data. Only in very specific cases, when disease risk was high and both genes had a dominant effect, present gene-gene interaction was detected.

CONCLUSIONS

Graphical chain models are, similar to logistic regression models, not able to capture gene-gene interactions for arbitrary biological models underlying the data.

Allelic penetrance approach as a tool to model two-locus interaction in complex binary traits

Many binary phenotypes do not follow a classical Mendelian inheritance pattern. Interaction between genetic and environmental factors is thought to contribute to the incomplete penetrance phenomena often observed in these complex binary traits. Several two-locus models for penetrance have been proposed to aid the genetic dissection of binary traits. Such models assume linear genetic effects of both loci in different mathematical scales of penetrance, resembling the analytical framework of quantitative traits. However, changes in phenotypic scale are difficult to envisage in binary traits and limited genetic interpretation is extractable from current modeling of penetrance. To overcome this limitation, we derived an allelic penetrance approach that attributes incomplete penetrance to the stochastic expression of the alleles controlling the phenotype, the genetic background and environmental factors. We applied this approach to formulate dominance and recessiveness in a single diallelic locus and to model different genetic mechanisms for the joint action of two diallelic loci. We fit the models to data on the genetic susceptibility of mice following infections with Listeria monocytogenes and Plasmodium berghei. These models gain in genetic interpretation, because they specify the alleles that are responsible for the genetic (inter)action and their genetic nature (dominant or recessive), and predict genotypic combinations determining the phenotype. Further, we show via computer simulations that the proposed models produce penetrance patterns not captured by traditional two-locus models. This approach provides a new analysis framework for dissecting mechanisms of interlocus joint action in binary traits using genetic crosses.

Design and analysis in genetic studies of human ageing and longevity

With the success of the Human Genome Project and taking advantage of the recent developments in high-throughput genotyping techniques as well as in functional genomics, it is now feasible to collect vast quantities of genetic data with the aim of deciphering the genetics of human complex traits. As a result, the amount of research on human ageing and longevity has been growing rapidly in recent years. The situation raises questions concerning efficient choice of study population, sampling schemes, and methods of data analysis. In this article, we summarize the key issues in genetic studies of human ageing and longevity ranging from research design to statistical analyses. We discuss the virtues and drawbacks of the multidisciplinary approaches including the population-based cross-sectional and cohort studies, family-based linkage analysis, and functional genomics studies. Different analytical approaches are illustrated with their performances compared. In addition, important research topics are highlighted together with experiment design and data analyzing issues to serve as references for future studies.

A combinatorial searching method for detecting a set of interacting loci associated with complex traits

Complex diseases are presumed to be the results of the interaction of several genes and environmental factors, with each gene only having a small effect on the disease. Mapping complex disease genes therefore becomes one of the greatest challenges facing geneticists. Most current approaches of association studies essentially evaluate one marker or one gene (haplotype approach) at a time. These approaches ignore the possibility that effects of multilocus functional genetic units may play a larger role than a single-locus effect in determining trait variability. In this article, we propose a Combinatorial Searching Method (CSM) to detect a set of interacting loci (may be unlinked) that predicts the complex trait. In the application of the CSM, a simple filter is used to filter all the possible locus-sets and retain the candidate locus-sets, then a new objective function based on the cross-validation and partitions of the multi-locus genotypes is proposed to evaluate the retained locus-sets. The locus-set with the largest value of the objective function is the final locus-set and a permutation procedure is performed to evaluate the overall p-value of the test for association between the final locus-set and the trait. The performance of the method is evaluated by simulation studies as well as by being applied to a real data set. The simulation studies show that the CSM has reasonable power to detect high-order interactions. When the CSM is applied to a real data set to detect the locus-set (among the 13 loci in the ACE gene) that predicts systolic blood pressure (SBP) or diastolic blood pressure (DBP), we found that a four-locus gene-gene interaction model best predicts SBP with an overall p-value = 0.033, and similarly a two-locus gene-gene interaction model best predicts DBP with an overall p-value = 0.045.

A flexible computational framework for detecting, characterizing, and interpreting statistical patterns of epistasis in genetic studies of human disease susceptibility

Detecting, characterizing, and interpreting gene-gene interactions or epistasis in studies of human disease susceptibility is both a mathematical and a computational challenge. To address this problem, we have previously developed a multifactor dimensionality reduction (MDR) method for collapsing high-dimensional genetic data into a single dimension (i.e. constructive induction) thus permitting interactions to be detected in relatively small sample sizes. In this paper, we describe a comprehensive and flexible framework for detecting and interpreting gene-gene interactions that utilizes advances in information theory for selecting interesting single-nucleotide polymorphisms (SNPs), MDR for constructive induction, machine learning methods for classification, and finally graphical models for interpretation. We illustrate the usefulness of this strategy using artificial datasets simulated from several different two-locus and three-locus epistasis models. We show that the accuracy, sensitivity, specificity, and precision of a naïve Bayes classifier are significantly improved when SNPs are selected based on their information gain (i.e. class entropy removed) and reduced to a single attribute using MDR. We then apply this strategy to detecting, characterizing, and interpreting epistatic models in a genetic study (n = 500) of atrial fibrillation and show that both classification and model interpretation are significantly improved.

Evidence for association of OCTN genes and IBD5 with ulcerative colitis

BACKGROUND AND AIMS

Genetic association between Crohn's disease (CD) and OCTN1 (SLC22A4) C1672T/OCTN2 (SLC22A5) G-207C variants in IBD5 has recently been reported. These genes encode solute carriers and the association was suggested to be distinct from the background IBD5 risk haplotype. There have been conflicting reports of the association between markers in the IBD5 region and ulcerative colitis (UC) and interaction (epistasis) between this locus and CARD15. Our aim was to ascertain the contribution of OCTN variants to UC and CD in a large independent UK dataset, to seek genetic evidence that the OCTN association is distinct from the IBD5 risk haplotype and to identify interactions between the IBD5 and CARD15 loci.

METHODS

A total of 1104 unrelated Caucasian subjects with inflammatory bowel disease (IBD) (496 CD, 512 UC, 96 indeterminate) and 750 ethnically matched controls were genotyped for three single nucleotide polymorphisms (SNPs) in the CD associated genes (OCTN1+1672, OCTN2-207, and IGR2230), and two flanking IBD5 tagging SNPs, IGR2096 and IGR3096. Data were analysed by logistic regression methods within STATA.

RESULTS

OCTN variants were as strongly associated with UC and IBD overall as they were with CD (p = 0.0001; OR 1.3 (95% confidence interval 1.1-1.5)). OCTN variants were in tight linkage disequilibrium with the extended IBD5 risk haplotype D' 0.79 and 0.88, and r2 = 0.62 and 0.72 for IGR2096 and 3096, respectively. There was no deviation from a multiplicative model of interaction between CARD15 and IBD5 on the penetrance scale.

CONCLUSIONS

The OCTN variants were associated with susceptibility to IBD overall. The effect was equally strong in UC and CD. Although OCTN variants may account for the increased risk of IBD associated with IBD5, a role for other candidate genes within this extended haplotype was not excluded. There was no statistical evidence of interaction between CARD15 and either OCTN or IBD5 variants in susceptibility to IBD.

The stationary distribution of allele frequencies when selection acts at unlinked loci

We consider population genetics models where selection acts at a set of unlinked loci. It is known that if the fitness of an individual is multiplicative across loci, then these loci are independent. We consider general selection models, but assume parent-independent mutation at each locus. For such a model, the joint stationary distribution of allele frequencies is proportional to the stationary distribution under neutrality multiplied by a known function of the mean fitness of the population. We further show how knowledge of this stationary distribution enables direct simulation of the genealogy of a sample at a single-locus. For a specific selection model appropriate for complex disease genes, we use simulation to determine what features of the genealogy differ between our general selection model and a multiplicative model.

Candidate gene association studies and evidence for gene-by-gene interactions

Candidate gene studies in asthma are a powerful and valuable tool in asthma genetics. Although the quality of small-scale, freely associating studies has been questionable, increasingly serious efforts are made to establish, replicate, and verify association results. Association studies may help us to better understand the mechanisms underlying asthma. They may create hypotheses and help to direct functional studies to targets that are likely to give valuable results. However, they should not be over-interpreted; only biologic proof can verify associations between genetic variations and a certain disease outcome. The insight that gene-by-gene and gene-by-environment interactions may be crucial for understanding and pinpoint the complex mechanisms of genetic regulation of multifactorial diseases has gained momentum in the last years when technical improvement allowed for the effective genotyping and analysis of great numbers of polymorphisms in large populations. It can be expected that from this area of research new and exciting results will follow soon.

Genetic association studies

We review the rationale behind and discuss methods of design and analysis of genetic association studies. There are similarities between genetic association studies and classic epidemiological studies of environmental risk factors but there are also issues that are specific to studies of genetic risk factors such as the use of particular family-based designs, the need to account for different underlying genetic mechanisms, and the effect of population history. Association differs from linkage (covered elsewhere in this series) in that the alleles of interest will be the same across the whole population. As with other types of genetic epidemiological study, issues of design, statistical analysis, and interpretation are very important.

Comparison of population- and family-based methods for genetic association analysis in the presence of interacting loci

We compared different ascertainment schemes for genetic association analysis: affected sib-pairs (ASPs), case-parent trios, and unrelated cases and controls. We found, with empirical type 1 diabetes data at four known disease loci, that studies based on case-parent trios and on unmatched cases and controls often gave higher odds ratio estimates and stronger significance test values than ASP designs. We used simulations and a simplified disease model involving two interacting loci, one of large effect and one smaller, to examine interaction models that could cause such an effect. The different ascertainment schemes were compared for power to detect an effect when only the locus of smaller effect was genotyped. ASPs showed the greatest power for association testing under most models of interaction except under additive and certain epistatic crossover models, for which case/controls and case-parent trios did better. All ascertainment schemes gave an unbiased estimation of log genotype relative risks (GRRs) under a multiplicative model. Under nonmultiplicative interactions, GRRs at the minor locus as estimated from ASPs could be biased upwards or downwards, resulting in either an increase or decrease in power compared to the case/control or trio design. For the four known type 1 diabetes loci, we observed decreased risks with ASPs, which could be due to additive interactions with the remaining susceptibility loci. Thus, the optimal ascertainment strategy in genetic association studies depends on the unknown underlying multilocus genetic model, and on whether the goal of the study is to detect an effect or to accurately estimate the resulting disease risks.

Replication in genetic studies of complex traits

Disappointments in replicating initial findings in gene mapping for complex traits are often attributed to small sample sizes and inadequate techniques to determine the threshold value. This is clearly not the whole truth. More fundamental reasons lie in the inherent heterogeneity related to disease, including genetic heterogeneity, differences in allele frequencies, and context-dependency in genetic architecture. There are also other reasons related to the data collection and analysis. Replication may remain a source of frustration unless more emphasis is put on controlling these sources of heterogeneity between studies.

Genome-wide association studies for common diseases and complex traits

Genetic factors strongly affect susceptibility to common diseases and also influence disease-related quantitative traits. Identifying the relevant genes has been difficult, in part because each causal gene only makes a small contribution to overall heritability. Genetic association studies offer a potentially powerful approach for mapping causal genes with modest effects, but are limited because only a small number of genes can be studied at a time. Genome-wide association studies will soon become possible, and could open new frontiers in our understanding and treatment of disease. However, the execution and analysis of such studies will require great care.

Traversing the conceptual divide between biological and statistical epistasis: systems biology and a more modern synthesis

Epistasis plays an important role in the genetic architecture of common human diseases and can be viewed from two perspectives, biological and statistical, each derived from and leading to different assumptions and research strategies. Biological epistasis is the result of physical interactions among biomolecules within gene regulatory networks and biochemical pathways in an individual such that the effect of a gene on a phenotype is dependent on one or more other genes. In contrast, statistical epistasis is defined as deviation from additivity in a mathematical model summarizing the relationship between multilocus genotypes and phenotypic variation in a population. The goal of this essay is to review definitions and examples of biological and statistical epistasis and to explore the relationship between the two. Specifically, we present and discuss the following two questions in the context of human health and disease. First, when does statistical evidence of epistasis in human populations imply underlying biomolecular interactions in the etiology of disease? Second, when do biomolecular interactions produce patterns of statistical epistasis in human populations? Answers to these two reciprocal questions will provide an important framework for using genetic information to improve our ability to diagnose, prevent and treat common human diseases. We propose that systems biology will provide the necessary information for addressing these questions and that model systems such as bacteria, yeast and digital organisms will be a useful place to start.

Heterogeneity in the magnitude of the insulin gene effect on HLA risk in type 1 diabetes

There is still uncertainty concerning the joint action of the two established type 1 diabetes susceptibility loci, the HLA class II DQB1 and DRB1 genes (IDDM1) and the insulin gene (INS) promoter (IDDM2). Some previous studies reported independence, whereas others suggested heterogeneity in the relative effects of the genotypes at these disease loci. In this study, we have assessed the combined effects of the HLA-DQB1/DRB1 and INS genotypes in 944 type 1 diabetic patients and 1,023 control subjects, all from Sardinia. Genotype variation at INS significantly influenced disease susceptibility in all HLA genotype risk categories. However, there was a significant heterogeneity (P = 2.4 x 10(-4)) in the distribution of the INS genotypes in patients with different HLA genotypes. The INS predisposing genotype was less frequent (74.9%) in high-risk HLA genotype-positive patients than in those with HLA intermediate-risk (86.1%) and low-risk (84.8%) categories. Gene-gene interaction modeling led to rejection of the additive model, whereas a multiplicative model showed a better, albeit still partial, fit to the observed data. These genetic results are consistent with an interaction between the protein products of the HLA and INS alleles, in which both the affinity of the various HLA class II molecules for a preproinsulin-derived peptide and the levels of this peptide in the thymus act jointly as key regulators of type 1 diabetes autoimmunity.

Replication of an association between the lymphoid tyrosine phosphatase locus (LYP/PTPN22) with type 1 diabetes, and evidence for its role as a general autoimmunity locus

In the genetic analysis of common, multifactorial diseases, such as type 1 diabetes, true positive irrefutable linkage and association results have been rare to date. Recently, it has been reported that a single nucleotide polymorphism (SNP), 1858C>T, in the gene PTPN22, encoding Arg620Trp in the lymphoid protein tyrosine phosphatase (LYP), which has been shown to be a negative regulator of T-cell activation, is associated with an increased risk of type 1 diabetes. Here, we have replicated these findings in 1,388 type 1 diabetic families and in a collection of 1,599 case and 1,718 control subjects, confirming the association of the PTPN22 locus with type 1 diabetes (family-based relative risk (RR) 1.67 [95% CI 1.46-1.91], and case-control odds ratio (OR) 1.78 [95% CI 1.54-2.06]; overall P = 6.02 x 10(-27)). We also report evidence for an association of Trp(620) with another autoimmune disorder, Graves' disease, in 1,734 case and control subjects (P = 6.24 x 10(-4); OR 1.43 [95% CI 1.17-1.76]). Taken together, these results indicate a more general association of the PTPN22 locus with autoimmune disease.

Potential mechanisms of interferon-alpha induced autoimmunity

The type I interferons (IFN) are cytokines encoded by a multigene family comprising 13 closely related IFN-A genes, and a single IFN-B gene. These factors are rapidly induced upon viral infection, and have pleiotropic effects. Historically, the induction of a cell-autonomous state of antiviral resistance, the inhibition of cell growth, and the regulation of apoptosis were appreciated first. More recently, it became generally accepted that they can regulate immune effector functions. This latter feature led them to be reconsidered as signals linking innate and adaptive immunity, and potentially orchestrating autoimmunity associated with viral infection and IFN-alpha therapy. Common to almost all autoimmune diseases is their polygenic inheritance, incomplete penetrance, and evidence for the role of environmental factors, particularly viral infection. In addition, they are characterized by increased numbers of circulating autoreactive T- and B-cells. Endogenously produced or therapeutically applied IFN-alpha can tilt the usually tightly controlled balance towards activation of these autoreactive cells via a vast array of mechanisms. The genetic susceptibility factors determine which type of autoimmunity will develop. IFN-alpha induces numerous target genes in antigen presenting cells (APC), such that APC are stimulated and enhance humoral autoimmunity, promote isotype switching, and potently activate autoreactive T cells. Moreover, IFN-alpha can synergistically amplify T cell autoreactivity by directly promoting T cell activation and keeping activated T cells alive. In essence, type I IFNs may constitute one example of genes that have been conserved because they confer dominant disease resistance, but at the same time they can trigger autoimmunity in genetically susceptible individuals.

Genetic interactions between susceptibility loci reveal epistatic pathogenic networks in murine lupus

Interactions between Sle1 and other susceptibility loci were required for disease development in the NZM2410 model of lupus. Sle1 corresponds to at least three subloci, Sle1a, Sle1b, and Sle1c, each of which independently causes loss of tolerance to chromatin, but displays a distinctive immune profile. We have used congenic strains to analyze the interactions between the Sle1 subloci and other lupus susceptibility loci using Y autoimmunity accelerator (Yaa) and Faslpr as sensitizing mutations. Sle1 coexpressed with either one of these single susceptibility alleles resulted in a highly penetrant nephritis, splenomegaly, production of nephrophilic antibodies, and increased expression of B- and T-cell activation markers. Here, we show that only Sle1b interacted with Yaa to produce these phenotypes, suggesting that Sle1b and Yaa belong to the same functional pathway. Interactions between the three Sle1 loci and lpr resulted in lymphocyte activation and lupus nephritis, but a significant mortality was observed only for the Sle1a.lpr combination. This suggests a major role for the FAS pathway in keeping in check the loss of tolerance mediated by the Sle1 loci, especially for Sle1a. Our results illustrate the complexity of interactions between susceptibility loci in polygenic diseases such as lupus and may explain the clinical heterogeneity of the disease.

The use of animal models in the study of complex disease: all else is never equal or why do so many human studies fail to replicate animal findings?

The study of the genetics of complex human disease has met with limited success. Many findings with candidate genes fail to replicate despite seemingly overwhelming physiological data implicating the genes. In contrast, animal model studies of the same genes and disease models usually have more consistent results. We propose that one important reason for this is the ability to control genetic background in animal studies. The fact that controlling genetic background can produce more consistent results suggests that the failure to replicate human findings in the same diseases is due to variation in interacting genes. Hence, the contrasting nature of the findings from the different study designs indicates the importance of non-additive genetic effects on human disease. We discuss these issues and some methodological approaches that can detect multilocus effects, using hypertension as a model disease. This article contains supplementary material, which may be viewed at the BioEssays website at http://www.interscience.wiley.com/jpages/0265-9247/suppmat/index.html.

Search for haplotype interactions that influence susceptibility to type 1 diabetes, through use of unphased genotype data

Type 1 diabetes is a T-cell-mediated chronic disease characterized by the autoimmune destruction of pancreatic insulin-producing beta cells and complete insulin deficiency. It is the result of a complex interrelation of genetic and environmental factors, most of which have yet to be identified. Simultaneous identification of these genetic factors, through use of unphased genotype data, has received increasing attention in the past few years. Several approaches have been described, such as the modified transmission/disequilibrium test procedure, the conditional extended transmission/disequilibrium test, and the stepwise logistic-regression procedure. These approaches are limited either by being restricted to family data or by ignoring so-called "haplotype interactions" between alleles. To overcome this limit, the present study provides a general method to identify, on the basis of unphased genotype data, the haplotype blocks that interact to define the risk for a complex disease. The principle underpinning the proposal is minimal entropy. The performance of our procedure is illustrated for both simulated and real data. In particular, for a set of Dutch type 1 diabetes data, our procedure suggests some novel evidence of the interactions between and within haplotype blocks that are across chromosomes 1, 2, 3, 4, 5, 6, 7, 8, 11, 12, 15, 16, 17, 19, and 21. The results demonstrate that, by considering interactions between potential disease haplotype blocks, we may succeed in identifying disease-predisposing genetic variants that might otherwise have remained undetected.

Ancestral processes for non-neutral models of complex diseases

We consider non-neutral models for unlinked loci, where the fitness of a chromosome or individual is not multiplicative across loci. Such models are suitable for many complex diseases, where there are gene-interactions. We derive a genealogical process for such models, called the complex selection graph (CSG). This coalescent-type process is related to the ancestral selection graph, and is derived from the ancestral influence graph by considering the limit as the recombination rate between loci gets large. We analyse the CSG both theoretically and via simulation. The main results are that the gene-interactions do not produce linkage disequilibrium, but do produce dependencies in allele frequencies between loci. For small selection rates, the distributions of the genealogy and the allele frequencies at a single locus are well-approximated by their distributions under a single locus model, where the fitness of each allele is the average of the true fitnesses of that allele with respect to the distribution of alleles at other loci.

Voluntary ethanol consumption by mice: genome-wide analysis of quantitative trait loci and their interactions in a C57BL/6ByJ x 129P3/J F2 intercross

Consumption of ethanol solutions by rodents in two-bottle choice tests is a model to study human alcohol intake. Mice of the C57BL/6ByJ strain have higher ethanol preferences and intakes than do mice of the 129P3/J strain. F2 hybrids between these two strains were phenotyped using two-bottle tests involving a choice between water and either 3% or 10% ethanol. High ethanol preferences and intakes of the B6 mice were inherited as additive or dominant traits in the F2 generation. A genome screen using these F2 mice identified three significant linkages. Two loci, on distal chromosome 4 (Ap3q) and proximal chromosome 7 (Ap7q), strongly affected 10% ethanol intake and weakly affected 3% ethanol intake. A male-specific locus on proximal chromosome 8 (Ap8q) affected 3% ethanol preference, but not indexes of 10% ethanol consumption. In addition, six suggestive linkages (on chromosomes 2, 9, 12, 13, 17, and 18) affecting indexes of 3% and/or 10% ethanol consumption were detected. The loci with significant and suggestive linkages accounted for 35-44% of the genetic variation in ethanol consumption phenotypes. No additive-by-additive epistatic interactions were detected for the primary loci with significant and suggestive linkages. However, there were a few modifiers of the primary linkages and a number of interactions among unlinked loci. This demonstrates a significant role of the genetic background in the variation of ethanol consumption.

Identification of two cerebral malaria resistance loci using an inbred wild-derived mouse strain

Malaria is a complex infectious disease in which the host/parasite interaction is strongly influenced by host genetic factors. The consequences of plasmodial infections range from asymptomatic to severe complications like the neurological syndrome cerebral malaria induced by Plasmodium falciparum in humans and Plasmodium berghei ANKA in rodents. Mice infected with P. berghei ANKA show marked differences in disease manifestation and either die from experimental cerebral malaria (ECM) or from hemolytic anemia caused by hyperparasitemia (HP). A majority of laboratory mouse strains so far investigated are susceptible to ECM; however, a number of wild-derived inbred strains show resistance. To evaluate the genetic basis of this difference, we crossed a uniquely ECM-resistant, wild-derived inbred strain (WLA) with an ECM susceptible laboratory strain (C57BL/6J). All of the (WLA x C57BL/6J) F(1) and 97% of the F(2) progeny displayed ECM resistance similar to the WLA strain. To screen for loci contributing to ECM resistance, we analyzed a cohort of mice backcrossed to the C57BL/6J parental strain. A genome wide screening of this cohort provided significant linkage of ECM resistance to marker loci in two genetic regions on chromosome 1 (chi(2) = 18.98, P = 1.3 x 10(-5)) and on chromosome 11 (chi(2) = 16.51, P = 4.8 x 10(-5)), being designated Berr1 and Berr2, respectively. These data provide the first evidence of loci associated with resistance to murine cerebral malaria, which may have important implications for the search for genetic factors controlling cerebral malaria in humans.