AbstractColorectal cancer is a heterogeneous disease that develops through a stepwise accumulation, yet the underlying mechanisms at single-cell resolution remain unclear. In this study, we profiled 751,531 single-cell transcriptomes, spatial transcriptomics, and snMultiomes from 142 multistage samples, revealing the cellular and molecular alterations and dynamic intercellular cross-talk during colorectal cancer development. Additionally, we created a colorectal cancer single-cell expression quantitative trait locus (sc-eQTL) map identifying 16,833 significant pairs across 28 cell subtypes, with more than 76% of sc-eQTLs being cell type–specific and fewer than 15% detectable in bulk datasets. A polygenic risk score derived from sc-eQTLs substantially improved colorectal cancer risk prediction. We prioritized rs4794979 that is associated with an increased colorectal cancer risk (OR = 1.11, P = 2.04 × 10−12) by promoting LGALS9 expression mediated by ELK1. Elevated LGALS9 in epithelia interacts with SLC1A5 on fibroblasts, promoting transformation into cancer-associated fibroblasts and simultaneously inducing CD8+ T-cell exhaustion via the LGALS9–TIM3 axis, thereby facilitating colorectal cancer development. Blocking the LGALS9–TIM3 axis enhanced anti–PD-1 therapy to inhibit colorectal cancer progression.Significance:Our study provides a valuable resource, including a dynamic single-cell landscape and a robust colorectal cancer sc-eQTL atlas, and elucidates the cell type–specific regulation and important cross-talk mechanisms between cell types in the tumor microenvironment, offering deep insights into colorectal cancer tumorigenesis and targeted therapies.
Chen et al. (Mon,) studied this question.