Genome-wide association study of cocaine self-administration behavior in Heterogeneous Stock rats

Article https://doi.org/10.1038/s41467-026-73694-w Genome-wide association study of cocaine self-administration behavior in Heterogeneous Stock rats Received: 13 August 2025 A list of authors and their affiliations appears at the end of the paper 1234567890():,; 1234567890():,; Accepted: 13 May 2026 Check for updates Cocaine use disorder (CUD) is a major public health crisis. The specific genes mediating CUD remain largely unknown. We conducted a genome-wide association study (GWAS) using outbred N/NIH Heterogeneous Stock (HS; n = 836, female = 415, male = 421) rats. We examined CUD-related phenotypes including acquisition of self-administration, escalation of intake, and compulsive-like responding. These traits were phenotypically correlated and exhibited modest SNP heritability (h2 = 0.07 – 0.16). We identified six genomewide significant associations (>-log10(p)=5.58; α = 0.05 by permutation). One locus on chromosome 19 was associated with variable time between cocaine infusions (post infusion interval) and contains several carboxylesterase genes that are orthologous to the human CES1 gene. Notably, carboxylesterases metabolize cocaine. Three non-synonymous coding variants in Ces1c and Ces1d were in perfect linkage disequilibrium with this locus. The other five loci contained promising coding and expression variants, including Trak2, a gene previously associated with CUD in human GWAS and Slc10a7, Plcl1, and Satb2 which have been associated with alcohol and tobacco use disorder. This is the largest genetic study of cocaine self-administration ever conducted in rats. Our results replicate previous loci associated with CUD in humans and provide several novel biological insights including the potential of pharmacological strategies targeting carboxylesterases. Cocaine use disorder (CUD) is a pressing public health issue. The 2022 National Survey on Drug Use and Health found that approximately 5 million individuals in the US reported using cocaine in the past year1. Notably, 1.4 million people meet the criteria for CUD. The complexity of CUD is further amplified by its frequent comorbidity with other neuropsychiatric disorders, underscoring the challenges in developing effective therapeutic strategies. CUD is a highly heritable and genetically complex trait. Twin studies estimate the heritability of cocaine dependence to be as high as 70%2–4, a finding supported by recent comprehensive reviews5,6. Complementing these findings, genome-wide association studies (GWAS) have also revealed a significant heritable component, with single nucleotide polymorphism (SNP)-based heritability estimated at 27–30%7,8. Despite the strong evidence of the genetic etiology of CUD, few genome-wide significant associations have been reported: FAM53B, NCOR29, LINC01411, TRAK2, LPHN210, and FAM78B11. The paucity of significant and replicated associations for CUD limits our understanding of this disorder, hampering our ability to identify novel pharmacological targets. Rodent models provide a complementary resource for exploring the genetic basis of CUD-like behaviors under tightly controlled experimental conditions. There are several advantages to using rodent models. First, the greater linkage disequilibrium (LD) found in rodents provides greater power albeit at the expense of mapping precision12. In addition, rodent models allow controlled exposure of naïve animals to addictive drugs under tightly-regulated experimental conditions that e-mail: ; Nature Communications | (2026)17:4876 1 Article still capture the extremes of drug-taking behavior. Controlled drug exposure allows examination of precise drug-taking patterns (e.g. the post infusion intervals of drug self-administration) and measures of demand (e.g. through behavior economics procedures such as the progressive ratio paradigms used in this study). Furthermore, rodents are exposed to uniform environmental conditions prior to drug exposure. It is also possible to avoid genetically influenced environmental exposures (e.g. higher rates of drug experimentation due to genetically-influenced risk-taking behavior13–16). Finally, rodents offer the possibility of experimental manipulations, either using experimental therapeutic agents, for which safety has not yet been established, and by genetic manipulations using CRISPR, viral vectors and opto- and chemo-genetic manipulations. Thus, the cellular and circuitlevel effects of identified polymorphisms can be experimentally investigated, which is advantageous for understanding mechanism and development of therapeutic agents. For these reasons, we and others have pursued rodent studies of CUD-like behaviors. Inbred mouse panels such as the Collaborative Cross (CC) recombinant inbred strains exhibit widely varying locomotor response to cocaine17,18. Recent studies using the hybrid mouse diversity panel (HMDP) and BXD recombinant inbred mouse population have identified candidate genes associated with cocaine self-administration19,20, including Fam53b, an ortholog to the human gene (FAM53B) implicated in CUD21. In addition to inbred strains, outbred rodent models offer a valuable tool for dissecting the genetic basis of CUD-like behaviors. Outbred populations such as the Heterogeneous Stock (HS) rats provide a high degree of genetic heterogeneity and low levels of linkage disequilibrium, enabling greater mapping precision than inbred or traditional F2 intercross populations. Prior work has established the phenotypic diversity of HS rats across a broad range of addictionrelevant behaviors, including cocaine self-administration12,22–28. Importantly, GWAS conducted in HS rats have identified numerous genetic variants associated with clinically relevant phenotypes related to a broad spectrum of human disorders29–36. Building on this framework, we expanded the sample used for a phenotypic analysis in de Guglielmo et al. (2024) to include nearly 900 HS rats that had undergone extended access to cocaine intravenous self-administration. This procedure captures key stages of the addiction process, such as initial acquisition of cocaine self-administration, escalation of cocaine intake, motivation to seek cocaine, and compulsive-like cocaine use despite adverse consequences37–39. By dissecting both the macrostructure, such as patterns of intake over multiple sessions, and the microstructure of cocaine self-administration, such as the differences in time between drug infusions, we can examine both correlated and orthogonal aspects of cocaine selfadministration. This level of detail is unobtainable in human studies, highlighting a major advantage of performing this work in rats. We then used 5,446,333 SNP genotypes to perform a GWAS analysis. In addition to the GWAS results, we used various secondary analysis strategies, such as integration with coding variants and eQTLs to uncover novel genetic drivers of cocaine self-administration behaviors. Results HS rats exhibit wide variation in cocaine self-administrati (...truncated)