Single-cell and spatial-resolved profiling reveals cancer-associated fibroblast heterogeneity in colorectal cancer...

Colorectal cancer (CRC) presents significant treatment challenges due to its high heterogeneity and complex intercellular interactions. Further exploration of CRC subtypes and interactions among tumor-specific clusters will facilitate the development of personalized treatment strategies.

Single-cell RNA sequencing and bulk RNA sequencing datasets were integrated to determine CRC metabolic subtypes by hierarchical clustering. The analysis was further extended to cellchat, pseudotime, immune infiltration, and clinicopathological relevance to explore the characteristics of secreted frizzled related protein 2 (SFRP2) + cancer-associated fibroblast (CAF) clusters, and validated by spatial transcriptomics (ST), in vivo experiments, and multiple immunohistochemistry (mIHC).

CRC samples were stably classified into three heterogeneous metabolic subtypes, each exhibiting different microenvironment and CAF heterogeneity, particularly in the distribution of SFRP2 + CAF, which was aligned with metabolic activity. SFRP2 + CAF exhibits high extracellular matrix (ECM) activity and is closely involved in cellular communication, not only promoting the malignant progression of cancer cells but also inducing the differentiation of Tregs. Compared to responders of chemotherapy, the proportion of SFRP2 + CAFs is significantly increased in non-responders. Importantly, mIHC and ST analyses confirm that cancer cells with low expression of agmatinase (AGMAT) can recruit SFRP2 + CAFs, and Treg infiltration surrounding SFRP2 + CAFs was observed. AGMAT combined with oxaliplatin showed the best efficacy in vivo, which may be associated with the inhibition of SFRP2 + CAF infiltration.

Our study identified and described the potential protumor biological properties of SFRP2 + CAFs, and AGMAT may be a valuable target for disrupting their properties.