Authors: Chien-Wei Peng, Michael Iglesia, Adrienne Visani, Xiang Li, Xiangwei Fang, Reyka G. Jayasinghe, Jingxian Liu, John Herndon, Jacqueline Mudd, Andrew Houston, Preet Lal, Siqi Chen, Ambrose Plante, André Targino da Costa, Kapur B. Dhami, Jennifer Ponce, Robert Fulton, Michael Heinz, Milan Chheda, Feng Chen, Ryan C. Fields, Li Ding
Published: 2025-04-21
DOI: 10.1158/1538-7445.am2025-6257
Source: Full article
Colorectal cancer (CRC) metastasis is highly aggressive and associated with poor outcomes. Recent studies have highlighted greater heterogeneity in gene expression within metastatic tumor cells compared to primary tumor cells. However, the underlying factors driving these differences remain poorly understood. We hypothesize that both genetic alterations and the tumor microenvironment play key roles in shaping this heterogeneity. To investigate these aspects, we established a cohort through the Human Tumor Atlas Network composed of 26 primary CRC samples and 41 metastatic CRC samples from multiple organs. We conducted whole exome sequencing (WES), single-nucleus ATAC/RNA-seq (snATAC/RNA-seq), and spatial transcriptomics (Xenium) assays to comprehensively analyze genetic alterations, gene expression patterns, and their spatial relations. Our findings revealed a higher proportion of TP53 and KRAS mutations in metastatic samples compared to primary tumor samples. Analysis of snRNA-seq data identified distinct tumor cell populations characterized by gene signatures from 12 published gene expression modules. Primary tumor cells were predominantly enriched for tumor intestinal stem cell-like and intestinal modules, while metastatic tumor cells showed higher representation of injury repair, epithelial-mesenchymal transition (EMT), osteoblast-like, and neuroendocrine modules. Notably, a microsatellite instability-high (MSI-high) metastatic sample exhibited relatively higher signatures of absorptive intestinal and squamous cancer modules than other microsatellite-stable samples, suggesting a unique gene expression profile in MSI-high CRC metastasis. To spatially study these cancer modules, we defined spatial tumor microregions as regions of tumor cells separated by stromal tissues. Using the Xenium assay, we mapped the gene modules within these microregions and observed significant enrichment of EMT and injury repair modules at the peripheral tumor microregions in metastatic samples. Conversely, the hypoxia module was more prominently enriched at the core of tumor microregions in metastatic samples compared to primary tumors. These findings suggest a strong association between tumor cell location within the microenvironment and their gene expression profiles, shedding light on the spatial dynamics of tumor heterogeneity in CRC metastasis. In summary, our results suggest that the heterogeneity of gene expression in metastatic CRC is driven by genetic alterations and strongly influenced by the tumor microenvironment.