A better prognosis for genetic association studies in mice

Pisani L, Yoo B, Bejerano G, Peltz G. Analysis of structural variation among inbred mouse strains

BACKGROUND

'Long read' sequencing methods have been used to identify previously uncharacterized structural variants that cause human genetic diseases. Therefore, we investigated whether long read sequencing could facilitate genetic analysis of murine models for human diseases.

RESULTS

The genomes of six inbred strains (BTBR T + Itpr3tf/J, 129Sv1/J, C57BL/6/J, Balb/c/J, A/J, SJL/J) were analyzed using long read sequencing. Our results revealed that (i) Structural variants are very abundant within the genome of inbred strains (4.8 per gene) and (ii) that we cannot accurately infer whether structural variants are present using conventional short read genomic sequence data, even when nearby SNP alleles are known. The advantage of having a more complete map was demonstrated by analyzing the genomic sequence of BTBR mice. Based upon this analysis, knockin mice were generated and used to characterize a BTBR-unique 8-bp deletion within Draxin that contributes to the BTBR neuroanatomic abnormalities, which resemble human autism spectrum disorder.

CONCLUSION

A more complete map of the pattern of genetic variation among inbred strains, which is produced by long read genomic sequencing of the genomes of additional inbred strains, could facilitate genetic discovery when murine models of human diseases are analyzed.

An automated multi-modal graph-based pipeline for mouse genetic discovery

MOTIVATION

Our ability to identify causative genetic factors for mouse genetic models of human diseases and biomedical traits has been limited by the difficulties associated with identifying true causative factors, which are often obscured by the many false positive genetic associations produced by a GWAS.

RESULTS

To accelerate the pace of genetic discovery, we developed a graph neural network (GNN)-based automated pipeline (GNNHap) that could rapidly analyze mouse genetic model data and identify high probability causal genetic factors for analyzed traits. After assessing the strength of allelic associations with the strain response pattern; this pipeline analyzes 29M published papers to assess candidate gene-phenotype relationships; and incorporates the information obtained from a protein-protein interaction network and protein sequence features into the analysis. The GNN model produces markedly improved results relative to that of a simple linear neural network. We demonstrate that GNNHap can identify novel causative genetic factors for murine models of diabetes/obesity and for cataract formation, which were validated by the phenotypes appearing in previously analyzed gene knockout mice. The diabetes/obesity results indicate how characterization of the underlying genetic architecture enables new therapies to be discovered and tested by applying 'precision medicine' principles to murine models.

AVAILABILITY AND IMPLEMENTATION

The GNNHap source code is freely available at https://github.com/zqfang/gnnhap, and the new version of the HBCGM program is available at https://github.com/zqfang/haplomap.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

What Have We Learned (or Expect to) From Analysis of Murine Genetic Models Related to Substance Use Disorders?

The tremendous public health problem created by substance use disorders (SUDs) presents a major opportunity for mouse genetics. Inbred mouse strains exhibit substantial and heritable differences in their responses to drugs of abuse (DOA) and in many of the behaviors associated with susceptibility to SUD. Therefore, genetic discoveries emerging from analysis of murine genetic models can provide critically needed insight into the neurobiological effects of DOA, and they can reveal how genetic factors affect susceptibility drug addiction. There are already indications, emerging from our prior analyses of murine genetic models of responses related to SUDs that mouse genetic models of SUD can provide actionable information, which can lead to new approaches for alleviating SUDs. Lastly, we consider the features of murine genetic models that enable causative genetic factors to be successfully identified; and the methodologies that facilitate genetic discovery.

The Effect of Population Structure on Murine Genome-Wide Association Studies

The ability to use genome-wide association studies (GWAS) for genetic discovery depends upon our ability to distinguish true causative from false positive association signals. Population structure (PS) has been shown to cause false positive signals in GWAS. PS correction is routinely used for analysis of human GWAS results, and it has been assumed that it also should be utilized for murine GWAS using inbred strains. Nevertheless, there are fundamental differences between murine and human GWAS, and the impact of PS on murine GWAS results has not been carefully investigated. To assess the impact of PS on murine GWAS, we examined 8223 datasets that characterized biomedical responses in panels of inbred mouse strains. Rather than treat PS as a confounding variable, we examined it as a response variable. Surprisingly, we found that PS had a minimal impact on datasets measuring responses in ≤20 strains; and had surprisingly little impact on most datasets characterizing 21 - 40 inbred strains. Moreover, we show that true positive association signals arising from haplotype blocks, SNPs or indels, which were experimentally demonstrated to be causative for trait differences, would be rejected if PS correction were applied to them. Our results indicate because of the special conditions created by GWAS (the use of inbred strains, small sample sizes) PS assessment results should be carefully evaluated in conjunction with other criteria, when murine GWAS results are evaluated.

In silico candidate variant and gene identification using inbred mouse strains

Mice are the most widely used animal model to study genotype to phenotype relationships. Inbred mice are genetically identical, which eliminates genetic heterogeneity and makes them particularly useful for genetic studies. Many different strains have been bred over decades and a vast amount of phenotypic data has been generated. In addition, recently whole genome sequencing-based genome-wide genotype data for many widely used inbred strains has been released. Here, we present an approach for in silico fine-mapping that uses genotypic data of 37 inbred mouse strains together with phenotypic data provided by the user to propose candidate variants and genes for the phenotype under study. Public genome-wide genotype data covering more than 74 million variant sites is queried efficiently in real-time to provide those variants that are compatible with the observed phenotype differences between strains. Variants can be filtered by molecular consequences and by corresponding molecular impact. Candidate gene lists can be generated from variant lists on the fly. Fine-mapping together with annotation or filtering of results is provided in a Bioconductor package called MouseFM. In order to characterize candidate variant lists under various settings, MouseFM was applied to two expression data sets across 20 inbred mouse strains, one from neutrophils and one from CD4⁺ T cells. Fine-mapping was assessed for about 10,000 genes, respectively, and identified candidate variants and haplotypes for many expression quantitative trait loci (eQTLs) reported previously based on these data. For albinism, MouseFM reports only one variant allele of moderate or high molecular impact that only albino mice share: a missense variant in the Tyr gene, reported previously to be causal for this phenotype. Performing in silico fine-mapping for interfrontal bone formation in mice using four strains with and five strains without interfrontal bone results in 12 genes. Of these, three are related to skull shaping abnormality. Finally performing fine-mapping for dystrophic cardiac calcification by comparing 9 strains showing the phenotype with eight strains lacking it, we identify only one moderate impact variant in the known causal gene Abcc6. In summary, this illustrates the benefit of using MouseFM for candidate variant and gene identification.

Transcription factor p73 regulates Th1 differentiation

Inter-individual differences in T helper (Th) cell responses affect susceptibility to infectious, allergic and autoimmune diseases. To identify factors contributing to these response differences, here we analyze in vitro differentiated Th1 cells from 16 inbred mouse strains. Haplotype-based computational genetic analysis indicates that the p53 family protein, p73, affects Th1 differentiation. In cells differentiated under Th1 conditions in vitro, p73 negatively regulates IFNγ production. p73 binds within, or upstream of, and modulates the expression of Th1 differentiation-related genes such as Ifng and Il12rb2. Furthermore, in mouse experimental autoimmune encephalitis, p73-deficient mice have increased IFNγ production and less disease severity, whereas in an adoptive transfer model of inflammatory bowel disease, transfer of p73-deficient naïve CD4⁺ T cells increases Th1 responses and augments disease severity. Our results thus identify p73 as a negative regulator of the Th1 immune response, suggesting that p73 dysregulation may contribute to susceptibility to autoimmune disease.

Heterogeneous helper T (Th) cell responses contribute to differential susceptibility to immunological disorders. Here the authors perform haplotype-based computational screens of 16 inbred mouse strains to identify a transcription factor, p73, as an important negative regulator of Th1 differentiation, with p73 deficient mice manifesting alterations in two inflammatory disease models.

Evidence of CNIH3 involvement in opioid dependence

Opioid dependence, a severe addictive disorder and major societal problem, has been demonstrated to be moderately heritable. We conducted a genome-wide association study in Comorbidity and Trauma Study data comparing opioid-dependent daily injectors (N=1167) with opioid misusers who never progressed to daily injection (N=161). The strongest associations, observed for CNIH3 single-nucleotide polymorphisms (SNPs), were confirmed in two independent samples, the Yale-Penn genetic studies of opioid, cocaine and alcohol dependence and the Study of Addiction: Genetics and Environment, which both contain non-dependent opioid misusers and opioid-dependent individuals. Meta-analyses found five genome-wide significant CNIH3 SNPs. The A allele of rs10799590, the most highly associated SNP, was robustly protective (P=4.30E-9; odds ratio 0.64 (95% confidence interval 0.55-0.74)). Epigenetic annotation predicts that this SNP is functional in fetal brain. Neuroimaging data from the Duke Neurogenetics Study (N=312) provide evidence of this SNP's in vivo functionality; rs10799590 A allele carriers displayed significantly greater right amygdala habituation to threat-related facial expressions, a phenotype associated with resilience to psychopathology. Computational genetic analyses of physical dependence on morphine across 23 mouse strains yielded significant correlations for haplotypes in CNIH3 and functionally related genes. These convergent findings support CNIH3 involvement in the pathophysiology of opioid dependence, complementing prior studies implicating the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) glutamate system.

The role of Abcb5 alleles in susceptibility to haloperidol-induced toxicity in mice and humans

BACKGROUND

We know very little about the genetic factors affecting susceptibility to drug-induced central nervous system (CNS) toxicities, and this has limited our ability to optimally utilize existing drugs or to develop new drugs for CNS disorders. For example, haloperidol is a potent dopamine antagonist that is used to treat psychotic disorders, but 50% of treated patients develop characteristic extrapyramidal symptoms caused by haloperidol-induced toxicity (HIT), which limits its clinical utility. We do not have any information about the genetic factors affecting this drug-induced toxicity. HIT in humans is directly mirrored in a murine genetic model, where inbred mouse strains are differentially susceptible to HIT. Therefore, we genetically analyzed this murine model and performed a translational human genetic association study.

METHODS AND FINDINGS

A whole genome SNP database and computational genetic mapping were used to analyze the murine genetic model of HIT. Guided by the mouse genetic analysis, we demonstrate that genetic variation within an ABC-drug efflux transporter (Abcb5) affected susceptibility to HIT. In situ hybridization results reveal that Abcb5 is expressed in brain capillaries, and by cerebellar Purkinje cells. We also analyzed chromosome substitution strains, imaged haloperidol abundance in brain tissue sections and directly measured haloperidol (and its metabolite) levels in brain, and characterized Abcb5 knockout mice. Our results demonstrate that Abcb5 is part of the blood-brain barrier; it affects susceptibility to HIT by altering the brain concentration of haloperidol. Moreover, a genetic association study in a haloperidol-treated human cohort indicates that human ABCB5 alleles had a time-dependent effect on susceptibility to individual and combined measures of HIT. Abcb5 alleles are pharmacogenetic factors that affect susceptibility to HIT, but it is likely that additional pharmacogenetic susceptibility factors will be discovered.

CONCLUSIONS

ABCB5 alleles alter susceptibility to HIT in mouse and humans. This discovery leads to a new model that (at least in part) explains inter-individual differences in susceptibility to a drug-induced CNS toxicity.

Genomic profiling of collaborative cross founder mice infected with respiratory viruses reveals novel transcripts and infection-related strain-specific gene and isoform expression

Genetic variation between diverse mouse species is well-characterized, yet existing knowledge of the mouse transcriptome comes largely from one mouse strain (C57BL/6J). As such, it is unlikely to reflect the transcriptional complexity of the mouse species. Gene transcription is dynamic and condition-specific; therefore, to better understand the mouse transcriptional response to respiratory virus infection, we infected the eight founder strains of the Collaborative Cross with either influenza A virus or severe acute respiratory syndrome coronavirus and sequenced lung RNA samples at 2 and 4 days after infection. We found numerous instances of transcripts that were not present in the C57BL/6J reference annotation, indicating that a nontrivial proportion of the mouse genome is transcribed but poorly annotated. Of these novel transcripts, 2150 could be aligned to human or rat genomes, but not to existing mouse genomes, suggesting functionally conserved sequences not yet recorded in mouse genomes. We also found that respiratory virus infection induced differential expression of 4287 splicing junctions, resulting in strain-specific isoform expression. Of these, 59 were influenced by strain-specific mutations within 2 base pairs of key intron–exon boundaries, suggesting cis-regulated expression. Our results reveal the complexity of the transcriptional response to viral infection, previously undocumented genomic elements, and extensive diversity in the response across mouse strains. These findings identify hitherto unexplored transcriptional patterns and undocumented transcripts in genetically diverse mice. Host genetic variation drives the complexity and diversity of the host response by eliciting starkly different transcriptional profiles in response to a viral infection.

The Netrin-1 receptor DCC is a regulator of maladaptive responses to chronic morphine administration

BACKGROUND

Opioids are the cornerstone of treatment for moderate to severe pain, but chronic use leads to maladaptations that include: tolerance, dependence and opioid-induced hyperalgesia (OIH). These responses limit the utility of opioids, as well as our ability to control chronic pain. Despite decades of research, we have no therapies or proven strategies to overcome this problem. However, murine haplotype based computational genetic mapping and a SNP data base generated from analysis of whole-genome sequence data (whole-genome HBCGM), provides a hypothesis-free method for discovering novel genes affecting opioid maladaptive responses.

RESULTS

Whole genome-HBCGM was used to analyze phenotypic data on morphine-induced tolerance, dependence and hyperalgesia obtained from 23 inbred strains. The robustness of the genetic mapping results was analyzed using strain subsets. In addition, the results of analyzing all of the opioid-related traits together were examined. To characterize the functional role of the leading candidate gene, we analyzed transgenic animals, mRNA and protein expression in behaviorally divergent mouse strains, and immunohistochemistry in spinal cord tissue. Our mapping procedure identified the allelic pattern within the netrin-1 receptor gene (Dcc) as most robustly associated with OIH, and it was also strongly associated with the combination of the other maladaptive opioid traits analyzed. Adult mice heterozygous for the Dcc gene had significantly less tendency to develop OIH, become tolerant or show evidence of dependence after chronic exposure to morphine. The difference in opiate responses was shown not to be due to basal or morphine-stimulated differences in the level of Dcc expression in spinal cord tissue, and was not associated with nociceptive neurochemical or anatomical alterations in the spinal cord or dorsal root ganglia in adult animals.

CONCLUSIONS

Whole-genome HBCGM is a powerful tool for identifying genes affecting biomedical traits such as opioid maladaptations. We demonstrate that Dcc affects tolerance, dependence and OIH after chronic opioid exposure, though not through simple differences in expression in the adult spinal cord.

Identification of exercise capacity QTL using association mapping in inbred mice

There are large interindividual differences in exercise capacity. It is well established that there is a genetic basis for these differences. However, the genetic factors underlying this variation are undefined. Therefore, the purpose of this study was to identify novel putative quantitative trait loci (QTL) for exercise capacity by measuring exercise capacity in inbred mice and performing genome-wide association mapping. Exercise capacity, defined as run time and work, was assessed in male mice (n = 6) from 34 strains of classical and wild-derived inbred mice performing a graded treadmill test. Genome-wide association mapping was performed with an efficient mixed-model association (EMMA) algorithm to identify QTL. Exercise capacity was significantly different across strains. Run time varied by 2.7-fold between the highest running strain (C58/J) and the lowest running strain (A/J). These same strains showed a 16.5-fold difference in work. Significant associations were identified for exercise time on chromosomes 1, 2, 7, 11, and 13. The QTL interval on chromosome 2 (~168 Mb) contains one gene, Nfatc2, and overlaps with a suggestive QTL for training responsiveness in humans. These results provide phenotype data on the widest range of inbred strains tested thus far and indicate that genetic background significantly influences exercise capacity. Furthermore, the novel QTLs identified in the current study provide new targets for investigating the underlying mechanisms for variation in exercise capacity.

Computational genetic discoveries that could improve perioperative medicine

PURPOSE OF REVIEW

The review examines the rationale and translational utility of computational genetic studies using murine models of biomedical traits.

RECENT FINDINGS

Computational genetic mapping studies have identified the genetic basis for biomedical trait differences in 16 different murine models, including several that are of importance to perioperative medicine.

SUMMARY

The results have generated new treatments for alleviating incisional pain and narcotic drug withdrawal symptoms, which are now in clinical trials. A recent study identified allelic differences affecting chronic pain responses in mice and humans, which may enable a new 'personalized' approach to treating chronic pain.