A novel method for identifying nonlinear gene–environment interactions in case–control association studies

The genetic influences on complex disease traits generally depend on the joint effects of multiple genetic variants, environmental factors, as well as their interplays. Gene × environment (G × E) interactions play vital roles in determining an individual’s disease risk, but the underlying genetic machinery is poorly understood. Traditional analysis assuming linear relationship between genetic and environmental factors, along with their interactions, is commonly pursued under the regression-based framework to examine G × E interactions. This assumption, however, could be violated due to nonlinear responses of genetic variants to environmental stimuli. As an extension to our previous work on continuous traits, we proposed a flexible varying-coefficient model for the detection of nonlinear G × E interaction with binary disease traits. Varying coefficients were approximated by a non-parametric regression function through which one can assess the nonlinear response of genetic factors to environmental changes. A group of statistical tests were proposed to elucidate various mechanisms of G × E interaction. The utility of the proposed method was illustrated via simulation and real data analysis with application to type 2 diabetes.     

Characterisation of CpG methylation in the upstream control region of mouse Nat2: Evidence for a gene–environment interaction in a polymorphic gene implicated in folate metabolism

Human arylamine N-acetyltransferase 1 (NAT1), a polymorphic xenobiotic metabolising enzyme, has been investigated in relation to susceptibility and prognosis in certain types of cancer. Both human NAT1 and its murine equivalent NAT2 have previously been shown to play roles in the catabolism of folate, which is required for the synthesis of S-adenosylmethionine, the methyl donor for cellular methylation reactions. We have tested whether the expression of mouse Nat2 is subject to epigenetic regulation, specifically CpG methylation in the promoter region, by determining levels of 5-methylcytosine by bisulphite sequencing and methylation-specific PCR. Under normal conditions, methylation levels of the Nat2 promoter were low, and varied in different tissues. However, CpG methylation was significantly increased by dietary folate supplementation, and increased methylation corresponded to decreased use of the core promoter. Functional deletion of the Nat2 gene gave rise to a significant increase in Nat2 methylation, extending our previous observations that folate catabolism is decreased in Nat2 null mice. Mouse NAT2 is likely to influence epigenetic gene control, particularly of its own locus, and this is consistent with recent evidence associating aberrant mouse Nat2/human NAT1 gene expression with certain developmental malformations and cancers.     

Characterizing maternal isolation-induced ultrasonic vocalizations in a gene–environment interaction rat model for autism

Deficits in social communication and language development belong to the earliest diagnostic criteria of autism spectrum disorders. Of the many risk factors for autism spectrum disorder, the contactin-associated protein-like 2 gene, CNTNAP2, is thought to be important for language development. The present study used a rat model to investigate the potential compounding effects of autism spectrum disorder risk gene mutation and environmental challenges, including breeding conditions or maternal immune activation during pregnancy, on early vocal communication in the offspring. Maternal isolation-induced ultrasonic vocalizations from Cntnap2 wildtype and knockout rats at selected postnatal days were analyzed for their acoustic, temporal and syntax characteristics. Cntnap2 knockout pups from heterozygous breeding showed normal numbers and largely similar temporal structures of ultrasonic vocalizations to wildtype controls, whereas both parameters were affected in homozygously bred knockouts. Homozygous breeding further exacerbated altered pitch and transitioning between call types found in Cntnap2 knockout pups from heterozygous breeding. In contrast, the effect of maternal immune activation on the offspring's vocal communication was confined to call type syntax, but left ultrasonic vocalization acoustic and temporal organization intact. Our results support the “double-hit hypothesis” of autism spectrum disorder risk gene–environment interactions and emphasize that complex features of vocal communication are a useful tool for identifying early autistic-like features in rodent models.     

Analysis of gene × environment interactions in sibships using mixed models

Background

Gene × environment models are widely used to assess genetic and environmental risks and their association with a phenotype of interest for many complex diseases. Mixed generalized linear models were used to assess gene × environment interactions with respect to systolic blood pressure on sibships adjusting for repeated measures and hierarchical nesting structures. A data set containing 410 sibships from the Framingham Heart Study offspring cohort (part of the Genetic Analysis Workshop 13 data) was used for all analyses. Three mixed gene × environment models, all adjusting for repeated measurement and varying levels of nesting, were compared for precision of estimates: 1) all sibships with adjustment for two levels of nesting (sibs within sibships and sibs within pedigrees), 2) all sibships with adjustment for one level of nesting (sibs within sibships), and 3) 100 data sets containing random draws of one sibship per extended pedigree adjusting for one level of nesting.

Results

The main effects were: gender, baseline age, body mass index (BMI), hypertensive treatment, cigarettes per day, grams of alcohol per day, and marker GATA48G07A. The interaction fixed effects were: baseline age by gender, baseline age by cigarettes per day, baseline age by hypertensive treatment, baseline age by BMI, hypertensive treatment by BMI, and baseline age by marker GATA48G07A. The estimates for all three nesting techniques were not widely discrepant, but precision of estimates and determination of significant effects did change with the change in adjustment for nesting.

Conclusion

Our results show the importance of the adjustment for all levels of hierarchical nesting of sibs in the presence of repeated measures.

     

Genome-wide investigation of gene–environment interactions in colorectal cancer

Colorectal cancer (CRC), one of the most frequent neoplasias worldwide, has both genetic and environmental causes. As yet, however, gene–environment (G × E) interactions in CRC have been studied mostly for a small number of candidate genes only. Therefore, we investigated the possible interaction, in CRC etiology, between single-nucleotide polymorphisms (SNPs) on the one hand, and overweight, smoking and alcohol consumption on the other, at a genome-wide level. To this end, we adopted a two-tiered approach comprising a case-only screening stage I (314 cases) and a case–control validation stage II (259 cases, 1,002 controls). Interactions with the smallest p value in stage I were verified in stage II using multiple logistic regression analysis adjusted for sex and age. In addition, we specifically studied known CRC-associated SNPs for possible G × E interactions. Upon adjustment for sex and age, and after allowing for multiple testing, however, only a single SNP (rs1944511) was found to be involved in a statistically significant interaction, namely with overweight (multiplicity-corrected p = 0.042 in stage II). Several other G × E interactions were nominally significant but failed correction for multiple testing, including a previously reported interaction between rs9929218 and alcohol consumption that also emerged in our candidate SNP study (nominal p = 0.008). Notably, none of the interactions identified in our genome-wide analysis was with a previously reported CRC-associated SNP. Our study therefore highlights the potential of an “agnostic” genome-wide approach to G × E analysis.     

Functional genomic studies of aldo–keto reductases

Aldose reductase (AR) is considered a potential mediator of diabetic complications and is a drug target for inhibitors of diabetic retinopathy and neuropathy in clinical trials. However, the physiological role of this enzyme still has not been established. Since effective inhibition of diabetic complications will require early intervention, it is important to delineate whether AR fulfills a physiological role that cannot be compensated by an alternate aldo–keto reductase. Functional genomics provides a variety of powerful new tools to probe the physiological roles of individual genes, especially those comprising gene families. Several eucaryotic genomes have been sequenced and annotated, including yeast, nematode and fly. To probe the function of AR, we have chosen to utilize the budding yeast Saccharomyces cerevisiae as a potential model system. Unlike Caenorhabditis elegans and D. melanogaster, yeast provides a more desirable system for our studies because its genome is manipulated more readily and is able to sustain multiple gene deletions in the presence of either drug or auxotrophic selectable markers. Using BLAST searches against the human AR gene sequence, we identified six genes in the complete S. cerevisiae genome with strong homology to AR. In all cases, amino acids thought to play important catalytic roles in human AR are conserved in the yeast AR-like genes. All six yeast AR-like open reading frames (ORFs) have been cloned into plasmid expression vectors. Substrate and AR inhibitor specificities have been surveyed on four of the enzyme forms to identify, which are the most functionally similar to human AR. Our data reveal that two of the enzymes (YDR368Wp and YHR104Wp) are notable for their similarity to human AR in terms of activity with aldoses and substituted aromatic aldehydes. Ongoing studies are aimed at characterizing the phenotypes of yeast strains containing single and multiple knockouts of the AR-like genes.     

Gene–gene and gene–environment interactions in ulcerative colitis

Genome-wide association studies (GWAS) have identified at least 133 ulcerative colitis (UC) associated loci. The role of genetic factors in clinical practice is not clearly defined. The relevance of genetic variants to disease pathogenesis is still uncertain because of not characterized gene–gene and gene–environment interactions. We examined the predictive value of combining the 133 UC risk loci with genetic interactions in an ongoing inflammatory bowel disease (IBD) GWAS. The Wellcome Trust Case–Control Consortium (WTCCC) IBD GWAS was used as a replication cohort. We applied logic regression (LR), a novel adaptive regression methodology, to search for high-order interactions. Exploratory genotype correlations with UC sub-phenotypes [extent of disease, need of surgery, age of onset, extra-intestinal manifestations and primary sclerosing cholangitis (PSC)] were conducted. The combination of 133 UC loci yielded good UC risk predictability [area under the curve (AUC) of 0.86]. A higher cumulative allele score predicted higher UC risk. Through LR, several lines of evidence for genetic interactions were identified and successfully replicated in the WTCCC cohort. The genetic interactions combined with the gene-smoking interaction significantly improved predictability in the model (AUC, from 0.86 to 0.89, P = 3.26E?05). Explained UC variance increased from 37 to 42 % after adding the interaction terms. A within case analysis found suggested genetic association with PSC. Our study demonstrates that the LR methodology allows the identification and replication of high-order genetic interactions in UC GWAS datasets. UC risk can be predicted by a 133 loci and improved by adding gene–gene and gene–environment interactions.     

Evidence of gene–environment interaction for the IRF6 gene and maternal multivitamin supplementation in controlling the risk of cleft lip with/without cleft palate

Although multiple genes have been identified as genetic risk factors for isolated, non-syndromic cleft lip with/without cleft palate (CL/P), a complex and heterogeneous birth defect, interferon regulatory factor 6 gene (IRF6) is one of the best documented genetic risk factors. In this study, we tested for association between markers in IRF6 and CL/P in 326 Chinese case–parent trios, considering gene–environment interaction for two common maternal exposures, and parent-of-origin effects. CL/P case–parent trios from three sites in mainland China and Taiwan were genotyped for 22 single nucleotide polymorphisms (SNPs) in IRF6. The transmission disequilibrium test was used to test for marginal effects of individual SNPs. We used PBAT to screen the SNPs and haplotypes for gene–environment (G × E) interaction and conditional logistic regression models to quantify effect sizes for SNP–environment interaction. After Bonferroni correction, 14 SNPs showed statistically significant association with CL/P. Evidence of G × E interaction was found for both maternal exposures, multivitamin supplementation and environmental tobacco smoke (ETS). Two SNPs showed evidence of interaction with multivitamin supplementation in conditional logistic regression models (rs2076153 nominal P = 0.019, rs17015218 nominal P = 0.012). In addition, rs1044516 yielded evidence for interaction with maternal ETS (nominal P = 0.041). Haplotype analysis using PBAT also suggested interaction between SNPs in IRF6 and both multivitamin supplementation and ETS. However, no evidence for maternal genotypic effects or significant parent-of-origin effects was seen in these data. These results suggest IRF6 gene may influence risk of CL/P through interaction with multivitamin supplementation and ETS in the Chinese population.     

Assessing the importance of genotype × environment interaction for root traits in rice using a mapping population II: conventional QTL analysis

Modifying plant root systems is considered a means of crop improvement targeted to low-resource environments, particularly low nutrient and drought-prone agriculture. The identification of quantitative trait loci (QTLs) for root traits has stimulated marker-assisted breeding to this end, but different QTLs have been detected in different populations of the same species, and importantly, in the same population when grown in different experimental environments. The presence of QTL × environment interaction is implicated, and this must be characterised if the utility of the target QTLs is to be realised. Previous attempts to do this suffer from a lack of control over replicate environments and inadequate statistical rigour. The Bala × Azucena mapping population was grown in two replicate experiments of four treatment environments, a control, a low light, a low soil nitrogen and a low soil water treatment. After a 4 weeks growth, maximum root length, maximum root thickness, root mass below 50 cm, total plant dry mass, % root mass and shoot length were measured. A summary of the overall results is presented in an accompanying paper. Here, QTL analysis by composite interval mapping is presented. A total of 145 QTLs were detected, mapping to 37 discrete loci on all chromosomes. Superficial evidence of QTL × E (great difference in LOD score) was tested by single-marker analysis which confirmed QTL × E for five loci representing only five individual trait-loci interactions. Some loci appeared to be stable across environments. Some QTLs were clearly more or less active under low light, low nitrogen or drought. A few notable loci on chromosomes 1, 2, 3, 5, 7 and 9 are briefly discussed. Also discussed are some remaining statistical shortcomings that will be addressed in another companion paper.     

Parent–child pair design for detecting gene–environment interactions in complex diseases

It is becoming clear that the etiology of complex diseases involves not only genetic and environmental factors but also gene–environment (GE) interactions. Therefore, it is important to take account of all these factors to improve the power of an epidemiological study design. We propose here a novel parent–child pair (PCP) design for this purpose. In comparison with conventional designs, this approach has the following advantages: (a) PCP is a 4 × 16 design consisting of pairs of parent–child (PC) genotype statuses, PC exposure statuses and PC disease statuses. Therefore, it utilizes more information than the traditional approaches in association studies; (b) It can determine whether findings in studies of association between disease and genetic or environmental factors and their interaction are spurious, arising from Hardy–Weinberg disequilibrium or the other factors; (c) Since the information from both parents and children of the PC pairs are used in this design, it has high power for detecting association of candidate gene, exposure with a complex disease and GE interaction. We also present a set of estimates of relative risks of candidate genes, exposures and GE interactions under the multiplicative model and a method for computing the sample size requirements to test for these relative risks in the context of the PCP design.     

Gene–environment and gene–gene interactions of specific MTHFR,MTR and CBS gene variants in relation to homocysteine in black South Africans

The methylenetetrahydrofolate reductase (MTHFR), cystathione-β-synthase (CBS) and methionine synthase (MTR) genes interact with each other and the environment. These interactions could influence homocysteine (Hcy) and diseases contingent thereon. We determined single nucleotide polymorphisms (SNPs) within these genes, their relationships and interactions with total Hcy concentrations within black South Africans to address the increased prevalence of diseases associated with Hcy. The MTHFR 677 TT and MTR 2756 AA genotypes were associated with higher Hcy concentrations (16.6 and 10.1 μmol/L; p < 0.05) compared to subjects harboring the MTHFR 677 CT/CC and the MTR 2756 AG genotypes (10.5, 9.7 and 9.5 μmol/L, respectively). The investigated CBS genotypes did not influence Hcy. We demonstrated interactions between the area of residence and the CBS T833C/844ins68 genotypes (p = 0.005) so that when harboring the wildtype allele, rural subjects had significantly higher Hcy than their urban counterparts, but when hosting the variant allele the environment made no difference to Hcy. Between the CBS T833C/844ins68 or G9276A and MTHFR C677T genotypes, there were two-way interactions (p = 0.003 and = 0.004, respectively), with regard to Hcy. Subjects harboring the MTHFR 677 TT genotype in combination with the CBS 833 TT/homozygous 844 non-insert or the MTHFR 677 TT genotype in combination with the CBS 9276 GA/GG displayed higher Hcy concentrations.     

Robust Bayesian variable selection for gene–environment interactions

Gene–environment (G× E) interactions have important implications to elucidate the etiology of complex diseases beyond the main genetic and environmental effects. Outliers and data contamination in disease phenotypes of G× E studies have been commonly encountered, leading to the development of a broad spectrum of robust regularization methods. Nevertheless, within the Bayesian framework, the issue has not been taken care of in existing studies. We develop a fully Bayesian robust variable selection method for G× E interaction studies. The proposed Bayesian method can effectively accommodate heavy-tailed errors and outliers in the response variable while conducting variable selection by accounting for structural sparsity. In particular, for the robust sparse group selection, the spike-and-slab priors have been imposed on both individual and group levels to identify important main and interaction effects robustly. An efficient Gibbs sampler has been developed to facilitate fast computation. Extensive simulation studies, analysis of diabetes data with single-nucleotide polymorphism measurements from the Nurses' Health Study, and The Cancer Genome Atlas melanoma data with gene expression measurements demonstrate the superior performance of the proposed method over multiple competing alternatives.     

Utilization of lysine 13C-methylation NMR for protein–protein interaction studies

Chemical modification is an easy way for stable isotope labeling of non-labeled proteins. The reductive ¹³C-methylation of the amino group of the lysine side-chain by ¹³C-formaldehyde is a post-modification and is applicable to most proteins since this chemical modification specifically and quickly proceeds under mild conditions such as 4 °C, pH 6.8, overnight. ¹³C-methylation has been used for NMR to study the interactions between the methylated proteins and various molecules, such as small ligands, nucleic acids and peptides. Here we applied lysine ¹³C-methylation NMR to monitor protein–protein interactions. The affinity and the intermolecular interaction sites of methylated ubiquitin with three ubiquitin-interacting proteins were successfully determined using chemical-shift perturbation experiments via the ¹H–¹³C HSQC spectra of the ¹³C-methylated-lysine methyl groups. The lysine ¹³C-methylation NMR results also emphasized the importance of the usage of side-chain signals to monitor the intermolecular interaction sites, and was applicable to studying samples with concentrations in the low sub-micromolar range.     

Assessing the importance of genotype x environment interaction for root traits in rice using a mapping population. I: a soil-filled box screen

Altering root system architecture is considered a method of improving crop water and soil nutrient capture. The analysis of quantitative trait loci (QTLs) for root traits has revealed inconsistency in the same population evaluated in different environments. It must be clarified if this is due to genotype × environment interaction or considerations of statistics if the value of QTLs for marker-assisted breeding is to be estimated. A modified split-plot design was used where a main plot corresponded to a separate experiment. The main plot factor had four treatments (environments), which were completely randomized among eight trials, so that each treatment was replicated twice. The sub-plot factor consisted of 168 recombinant inbreed lines of the Bala × Azucena rice mapping population, randomly allocated to the seven soil-filled boxes. The aim of the trial was to quantify QTL × environment interaction. The treatments were chosen to alter partitioning to roots; consisting of a control treatment (high-soil nitrogen, high light and high-water content) and further treatments where light, soil nitrogen or soil water was reduced singly. After 4 weeks growth, maximum root length (MRL), maximum root thickness, root mass below 50 cm, total plant dry mass (%), root mass and shoot length were measured. The treatments affected plant growth as predicted; low nitrogen and drought increased relative root partitioning, low-light decreased it. The parental varieties Bala and Azucena differed significantly for all traits. Broad-sense heritability of most traits was high (57–86%). Variation due to treatment was the most important influence on the variance, while genotype was next. Genotype × environment interaction was detected for all traits except MRL, although the proportion of variation due to this interaction was generally small. It is concluded that genotype × environment interaction is present but less important than genotypic variation. A companion paper presents QTL × environment analysis of data.     

Detection of disease-associated deletions in case–control studies using SNP genotypes with application to rheumatoid arthritis

Genomic deletions have long been known to play a causative role in microdeletion syndromes. Recent whole-genome genetic studies have shown that deletions can increase the risk for several psychiatric disorders, suggesting that genomic deletions play an important role in the genetic basis of complex traits. However, the association between genomic deletions and common, complex diseases has not yet been systematically investigated in gene mapping studies. Likelihood-based statistical methods for identifying disease-associated deletions have recently been developed for familial studies of parent-offspring trios. The purpose of this study is to develop statistical approaches for detecting genomic deletions associated with complex disease in case–control studies. Our methods are designed to be used with dense single nucleotide polymorphism (SNP) genotypes to detect deletions in large-scale or whole-genome genetic studies. As more and more SNP genotype data for genome-wide association studies become available, development of sophisticated statistical approaches will be needed that use these data. Our proposed statistical methods are designed to be used in SNP-by-SNP analyses and in cluster analyses based on combined evidence from multiple SNPs. We found that these methods are useful for detecting disease-associated deletions and are robust in the presence of linkage disequilibrium using simulated SNP data sets. Furthermore, we applied the proposed statistical methods to SNP genotype data of chromosome 6p for 868 rheumatoid arthritis patients and 1,197 controls from the North American Rheumatoid Arthritis Consortium. We detected disease-associated deletions within the region of human leukocyte antigen in which genomic deletions were previously discovered in rheumatoid arthritis patients.     

Microbial population index and community structure in saline–alkaline soil using gene targeted metagenomics

Population indices of bacteria and archaea were investigated from saline–alkaline soil and a possible microbe–environment pattern was established using gene targeted metagenomics. Clone libraries were constructed using 16S rRNA and functional gene(s) involved in carbon fixation (cbbL), nitrogen fixation (nifH), ammonia oxidation (amoA) and sulfur metabolism (apsA). Molecular phylogeny revealed the dominance of Actinobacteria, Firmicutes and Proteobacteria along with archaeal members of Halobacteraceae. The library consisted of novel bacterial (20%) and archaeal (38%) genera showing ≤95% similarity to previously retrieved sequences. Phylogenetic analysis indicated ability of inhabitant to survive in stress condition. The 16S rRNA gene libraries contained novel gene sequences and were distantly homologous with cultured bacteria. Functional gene libraries were found unique and most of the clones were distantly related to Proteobacteria, while clones of nifH gene library also showed homology with Cyanobacteria and Firmicutes. Quantitative real-time PCR exhibited that bacterial abundance was two orders of magnitude higher than archaeal. The gene(s) quantification indicated the size of the functional guilds harboring relevant key genes. The study provides insights on microbial ecology and different metabolic interactions occurring in saline–alkaline soil, possessing phylogenetically diverse groups of bacteria and archaea, which may be explored further for gene cataloging and metabolic profiling.