Multi-omic and multi-species meta-analyses of nicotine consumption

Abstract

Cross-species translational approaches to human genomic analyses are lacking. The present study uses an integrative framework to investigate how genes associated with nicotine use in model organisms contribute to the genetic architecture of human tobacco consumption. First, we created a model organism geneset by collecting results from five animal models of nicotine exposure (RNA expression changes in brain) and then tested the relevance of these genes and flanking genetic variation using genetic data from human cigarettes per day (UK BioBank N = 123,844; all European Ancestry). We tested three hypotheses: (1) DNA variation in, or around, the ‘model organism geneset’ will contribute to the heritability to human tobacco consumption, (2) that the model organism genes will be enriched for genes associated with human tobacco consumption, and (3) that a polygenic score based off our model organism geneset will predict tobacco consumption in the AddHealth sample (N = 1667; all European Ancestry). Our results suggested that: (1) model organism genes accounted for ~5–36% of the observed SNP-heritability in human tobacco consumption (enrichment: 1.60–31.45), (2) model organism genes, but not negative control genes, were enriched for the gene-based associations (MAGMA, H-MAGMA, SMultiXcan) for human cigarettes per day, and (3) polygenic scores based on our model organism geneset predicted cigarettes per day in an independent sample. Altogether, these findings highlight the advantages of using multiple species evidence to isolate genetic factors to better understand the etiological complexity of tobacco and other nicotine consumption.

Integration of evidence across human and model organism studies: A meeting report

“…Many researchers have begun to tackle the issue of variant prioritization by integrating multiple sources of information 72‐74 . Indeed, most GWAS include detailed post‐hoc analyses toward the identification of credible causal variants.…”

“…Studies of alcohol and cannabis use disorders have shown quantitative and qualitative differences between the genetics of consumption quantity and frequency and the genetics of the disorders (e.g., impaired functioning, physical dependence, disruption of social responsibilities) 12,86 . Likewise, a geneset derived from tobacco exposure paradigms in rodents shows modest enrichment for the SNP‐heritability of human tobacco consumption 87 . Notably, inbred strain comparison/selective breeding studies have allowed scientists to examine the effects of genetic background on multiple related traits 88 .…”

“…For example, even just within humans, recent studies suggest that the genetics of human alcohol consumption, particularly frequency of alcohol intake, is only partly related to the genetics of alcohol problems (e.g., impaired functioning, physical dependence, disruption of social responsibilities) 19 . Likewise, a geneset derived from tobacco exposure paradigms in rodents shows modest enrichment for the SNP‐heritability of human tobacco consumption 87 . Therefore, differences in phenotypes and their associated genetic architecture, whether within or across organisms, should be taken into consideration, and leveraged when possible.…”

“…These included understanding (1) the degree of concordance among model organism findings, as well as (2) the extent to which model organism evidence generalizes to humans, (3) the contextual implication of tissue, sex, and ancestry on these effects, and (4) how unifying phenotypic definitions across databases can enhance sample sizes and data integration. To date, several studies have shown enrichment of mouse and rat gene sets (i.e., those that are differentially expressed in the presence of cocaine) in the human brain transcriptome for cocaine use disorder, 89 as well as human GWAS of tobacco/nicotine consumption 87 . Identifying convergent genetic mechanisms between humans and model organisms in SUDs is an exciting challenge but is (relatively) close at hand.…”

“…Many researchers have begun to tackle the issue of variant prioritization by integrating multiple sources of information. 72 , 73 , 74 Indeed, most GWAS include detailed post‐hoc analyses toward the identification of credible causal variants. Network integration is one method that can permit the full illumination of patterns that are shared across gene sets derived from single omics data (e.g., genetic variants, RNA‐seq in bulk tissue, single‐cell RNA‐seq, chromatin immunoprecipitation sequencing [ChIP‐seq], ATAC‐seq, methylome, etc.).…”