Genetic Determinants of Cell-Type-Specific Responses to Cardiotoxic Chemotherapeutics

Cardiovascular toxicity (CT) is a major side effect of the anticancer drugs doxorubicin (DOX) and 5-fluorouracil (5-FU), with up to 48% of patients treated with DOX and 35% of patients treated with 5-FU experiencing CT. Inter-individual differences in susceptibility to CT are influenced in part by genetic variants that modulate gene regulation in a drug-dependent manner, known as response expression quantitative trait loci (response eQTLs). These variants and their regulatory effects are difficult to characterize because they are highly context-dependent, emerging only under specific combinations of genotype, cell type, and environmental exposure. Moreover, detecting response eQTLs requires profiling dozens of individuals to achieve sufficient statistical power. As a result, their discovery demands a scalable in vitro model that captures diverse CT-relevant cell types. To address this, we developed a guided differentiation protocol using induced pluripotent stem cells to produce heterogeneous differentiating cultures of cardiac cells (cardiac HDCs). In under two weeks, guided differentiation reproducibly generates cardiac HDCs that harbor 14 different cell types, including cardiomyocytes, epicardial cells, fibroblasts, endothelial cells, hematopoietic cells, hepatic endoderm, and neural crest cells. To identify drug response eQTLs, we independently treated cardiac HDCs from 46 individuals with 1 µM DOX, 5 µM 5-FU, or a vehicle control for 24 hours. We then performed scRNA-seq to capture transcriptional responses and identified 3,014 unique genes (eGenes), each associated with at least one cis-eQTL. We also identified 912 distinct response eGenes, where the corresponding eQTLs exert regulatory effects in response to drug treatment. Several of the response eGenes we identified are associated with drug-induced CT, including HAS3, ABCB4, RBL2, TNFRSF1A, ZNF521, and LAMC1. Over a third (35.7%) of drug-response eGenes were shared between the DOX and 5-FU treatments. We also performed colocalization of our eQTLs with genome-wide association studies (GWASs) of CT-relevant diseases. We found probable colocalizations for both DOX and 5-FU-response eQTLs. Notably, in epithelial cells, we discovered a locus that likely colocalized with ischemic heart disease and the DOX-response eGene interleukin-6 receptor (IL6R). The IL6R cytokine receptor has an established role in heart disease where it is a druggable target. Further, IL6R binds IL6 as its ligand, and IL6 is a candidate biomarker for anthracycline-induced CT. Results produced using cardiac HDCs from control and drug-treated conditions demonstrate the importance of studying gene regulation in a variety of cell-type-condition contexts. Further, our findings enable discovery of genetic variants underlying CT and support development of genotype-based predictors for treating cancer patients.