Analysis of experimental colorectal liver metastases identifies distinct proliferation-independent transcriptional programs in micro- versus macrometastases

Distant relapse following surgical removal of primary stage I-III colorectal cancer is caused by the outgrowth of micrometastases that were undetectable at first diagnosis. Adjuvant chemotherapy lowers the risk of distant metastasis formation by only ~15%, indicating that more effective metastasis-prevention strategies are urgently needed. Research aiming to develop such strategies is challenging, due to the low abundance of micrometastases, their small size, and the paucity of model systems. We applied a p27-based fluorescent reporter to identify quiescent tumor cells in vitro and in vivo. Colon cancer patient-derived organoids (PDOs) were transplanted into the mouse caecum, which led to spontaneous formation of micro- and macrometastases. Liver metastases were analyzed by immunohistochemistry. In addition, a liver perfusion and collagenase-based homogenization protocol was developed which allows isolation of micro- and macrometastases. Subsequently, p27-positive and p27-negative tumor cells were isolated and analyzed by bulk and single cell RNA sequencing. In vitro and in vivo analyses validated the p27-reporter as a robust tool for separating quiescent from cycling tumor cells. In vivo analyses revealed that micrometastases can be either cycling or quiescent in our metastasis model. In depth analyses of quiescent cells obtained from differently sized metastases showed that macrometastases are characterized by TNFα signaling and epithelial-to-mesenchymal transition. By contrast, micrometastases are characterized by upregulation of metabolic pathways including oxidative phosphorylation. Macrometastases, but not micrometastases, are characterized by the presence of cancer-associated fibroblasts and collagen deposition. We developed a strategy to analyze spontaneously formed quiescent micrometastases at single-cell resolution and demonstrate that these lesions form a transcriptionally distinct tumor entity. The data provides leads for developing alternative relapse-prevention strategies that target quiescent micrometastases.