Allele-specific expression and eQTL analysis in mouse adipose tissue

Background

The simplest definition of cis-eQTLs versus trans, refers to genetic variants that affect expression in an allele specific manner, with implications on underlying mechanism. Yet, due to technical limitations of expression microarrays, the vast majority of eQTL studies performed in the last decade used a genomic distance based definition as a surrogate for cis, therefore exploring local rather than cis-eQTLs.

Results

In this study we use RNAseq to explore allele specific expression (ASE) in adipose tissue of male and female F1 mice, produced from reciprocal crosses of C57BL/6J and DBA/2J strains. Comparison of the identified cis-eQTLs, to local-eQTLs, that were obtained from adipose tissue expression in two previous population based studies in our laboratory, yields poor overlap between the two mapping approaches, while both local-eQTL studies show highly concordant results. Specifically, local-eQTL studies show ~60% overlap between themselves, while only 15-20% of local-eQTLs are identified as cis by ASE, and less than 50% of ASE genes are recovered in local-eQTL studies. Utilizing recently published ENCODE data, we also find that ASE genes show significant bias for SNPs prevalence in DNase I hypersensitive sites that is ASE direction specific.

Conclusions

We suggest a new approach to analysis of allele specific expression that is more sensitive and accurate than the commonly used fisher or chi-square statistics. Our analysis indicates that technical differences between the cis and local-eQTL approaches, such as differences in genomic background or sex specificity, account for relatively small fraction of the discrepancy. Therefore, we suggest that the differences between two eQTL mapping approaches may facilitate sorting of SNP-eQTL interactions into true cis and trans, and that a considerable portion of local-eQTL may actually represent trans interactions.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-471) contains supplementary material, which is available to authorized users.     

Identification of quantitative trait loci and candidate genes for an anxiolytic‐like response to ethanol in BXD recombinant inbred strains

Genetic differences in acute behavioral responses to ethanol contribute to the susceptibility to alcohol use disorder and the reduction of anxiety is a commonly reported motive underlying ethanol consumption among alcoholics. Therefore, we studied the genetic variance in anxiolytic‐like responses to ethanol across the BXD recombinant inbred (RI) mouse panel using the light–dark transition model of anxiety. Strain‐mean genetic mapping and a mixed‐model quantitative trait loci (QTL) analysis replicated several previously published QTL for locomotor activity and identified several novel anxiety‐related loci. Significant loci included a chromosome 11 saline anxiety‐like QTL (Salanq1) and a chromosome 12 locus (Etanq1) influencing the anxiolytic‐like response to ethanol. Etanq1 was successfully validated by studies with BXD advanced intercross strains and fine‐mapped to a region comprising less than 3.5 Mb. Through integration of genome‐wide mRNA expression profiles of the mesocorticolimbic reward circuit (prefrontal cortex, nucleus accumbens and ventral midbrain) across the BXD RI panel, we identified high priority candidate genes within Etanq1, the strongest of which was Ninein (Nin), a Gsk3β‐interacting protein that is highly expressed in the brain.     

Inferred expression regulator activities suggest genes mediating cardiometabolic genetic signals

Expression QTL (eQTL) analyses have suggested many genes mediating genome-wide association study (GWAS) signals but most GWAS signals still lack compelling explanatory genes. We have leveraged an adipose-specific gene regulatory network to infer expression regulator activities and phenotypic master regulators (MRs), which were used to detect activity QTLs (aQTLs) at cardiometabolic trait GWAS loci. Regulator activities were inferred with the VIPER algorithm that integrates enrichment of expected expression changes among a regulator’s target genes with confidence in their regulator-target network interactions and target overlap between different regulators (i.e., pleiotropy). Phenotypic MRs were identified as those regulators whose activities were most important in predicting their respective phenotypes using random forest modeling. While eQTLs were typically more significant than aQTLs in cis, the opposite was true among candidate MRs in trans. Several GWAS loci colocalized with MR trans-eQTLs/aQTLs in the absence of colocalized cis-QTLs. Intriguingly, at the 1p36.1 BMI GWAS locus the EPHB2 cis-aQTL was stronger than its cis-eQTL and colocalized with the GWAS signal and 35 BMI MR trans-aQTLs, suggesting the GWAS signal may be mediated by effects on EPHB2 activity and its downstream effects on a network of BMI MRs. These MR and aQTL analyses represent systems genetic methods that may be broadly applied to supplement standard eQTL analyses for suggesting molecular effects mediating GWAS signals.     

Genomewide association analysis in diverse inbred mice: power and population structure

The discovery of quantitative trait loci (QTL) in model organisms has relied heavily on the ability to perform controlled breeding to generate genotypic and phenotypic diversity. Recently, we and others have demonstrated the use of an existing set of diverse inbred mice (referred to here as the mouse diversity panel, MDP) as a QTL mapping population. The use of the MDP population has many advantages relative to traditional F(2) mapping populations, including increased phenotypic diversity, a higher recombination frequency, and the ability to collect genotype and phenotype data in community databases. However, these methods are complicated by population structure inherent in the MDP and the lack of an analytical framework to assess statistical power. To address these issues, we measured gene expression levels in hypothalamus across the MDP. We then mapped these phenotypes as quantitative traits with our association algorithm, resulting in a large set of expression QTL (eQTL). We utilized these eQTL, and specifically cis-eQTL, to develop a novel nonparametric method for association analysis in structured populations like the MDP. These eQTL data confirmed that the MDP is a suitable mapping population for QTL discovery and that eQTL results can serve as a gold standard for relative measures of statistical power.     

Genetic loci for diet-induced atherosclerotic lesions and plasma lipids in mice

Genetic factors independent of those affecting plasma lipid levels are a major contributor to risk for atherosclerosis in humans, yet the basis for these is poorly understood. This study examined plasma lipids and diet-induced atherosclerosis in 16-month-old female mice of strains C56BL/6J and DBA/2J. Mice of the parental strains, from recombinant inbred strains derived from these (BXD RI), and F₂ progeny were fed an atherogenic diet for 16 weeks, beginning at 1 year of age. This induced atherosclerotic lesion formation in both parental strains, accompanied by increased plasma LDL levels. However, individual BXD RI strains and the BXD F₂ mice demonstrated a range of atherosclerotic lesion formation that was not or at best weakly correlated with plasma lipid levels. Quantitative trait locus (QTL) analysis of the BXD F₂ mice identified a locus with significant linkage (lod 4.5) for aortic lesion size on Chromosome (Chr) 10 that was independent of plasma lipids. Other loci with suggestive or significant linkage for various plasma lipid measures were identified on Chr 2, 3, 4, 5, 6, 7, 11, and 17. In this intercross, the genes primarily influencing atherosclerosis are distinct from those controlling plasma lipid levels.     

Identifying human disease genes through cross-species gene mapping of evolutionary conserved processes

Background

Understanding complex networks that modulate development in humans is hampered by genetic and phenotypic heterogeneity within and between populations. Here we present a method that exploits natural variation in highly diverse mouse genetic reference panels in which genetic and environmental factors can be tightly controlled. The aim of our study is to test a cross-species genetic mapping strategy, which compares data of gene mapping in human patients with functional data obtained by QTL mapping in recombinant inbred mouse strains in order to prioritize human disease candidate genes.

Methodology

We exploit evolutionary conservation of developmental phenotypes to discover gene variants that influence brain development in humans. We studied corpus callosum volume in a recombinant inbred mouse panel (C57BL/6J×DBA/2J, BXD strains) using high-field strength MRI technology. We aligned mouse mapping results for this neuro-anatomical phenotype with genetic data from patients with abnormal corpus callosum (ACC) development.

Principal Findings

From the 61 syndromes which involve an ACC, 51 human candidate genes have been identified. Through interval mapping, we identified a single significant QTL on mouse chromosome 7 for corpus callosum volume with a QTL peak located between 25.5 and 26.7 Mb. Comparing the genes in this mouse QTL region with those associated with human syndromes (involving ACC) and those covered by copy number variations (CNV) yielded a single overlap, namely HNRPU in humans and Hnrpul1 in mice. Further analysis of corpus callosum volume in BXD strains revealed that the corpus callosum was significantly larger in BXD mice with a B genotype at the Hnrpul1 locus than in BXD mice with a D genotype at Hnrpul1 (F = 22.48, p<9.87*10⁻⁵).

Conclusion

This approach that exploits highly diverse mouse strains provides an efficient and effective translational bridge to study the etiology of human developmental disorders, such as autism and schizophrenia.     

Location of theSas-1 locus on mouse chromosome 1

Using three sets of recombinant inbred strains (BXD, BXH, and BXJ), we found the locus controlling an antigenic substance (Sas}-1) in murine serum to be closely linked to the Chromosome-1 marker,Dip-1. This linkage was confirmed by an analysis of backcross linkage. The BXD and backcross data suggest that the gene order isId-1-Dip-1-Sas-1-Mls. Data from the three sets of RI strains and the 32 backcross mice lead to the estimate that the recombination frequency betweenDip-1 andSas-1 is 0.030 ±0.015.     

Expression Quantitative Trait Loci Are Highly Sensitive to Cellular Differentiation State

Genetical genomics is a strategy for mapping gene expression variation to expression quantitative trait loci (eQTLs). We performed a genetical genomics experiment in four functionally distinct but developmentally closely related hematopoietic cell populations isolated from the BXD panel of recombinant inbred mouse strains. This analysis allowed us to analyze eQTL robustness/sensitivity across different cellular differentiation states. Although we identified a large number (365) of “static” eQTLs that were consistently active in all four cell types, we found a much larger number (1,283) of “dynamic” eQTLs showing cell-type–dependence. Of these, 140, 45, 531, and 295 were preferentially active in stem, progenitor, erythroid, and myeloid cells, respectively. A detailed investigation of those dynamic eQTLs showed that in many cases the eQTL specificity was associated with expression changes in the target gene. We found no evidence for target genes that were regulated by distinct eQTLs in different cell types, suggesting that large-scale changes within functional regulatory networks are uncommon. Our results demonstrate that heritable differences in gene expression are highly sensitive to the developmental stage of the cell population under study. Therefore, future genetical genomics studies should aim at studying multiple well-defined and highly purified cell types in order to construct as comprehensive a picture of the changing functional regulatory relationships as possible.