Differential cellular origins of the extracellular matrix of tumor and normal tissues according to colorectal cancer subtypes.

Background: Understanding the proteomic-level heterogeneity of the tumor microenvironment (TME) in colorectal cancer (CRC) is crucial due to its well-known heterogeneity. While heterogenous CRC has been extensively characterized at the molecular subtype level, research into the functional heterogeneity of fibroblasts, particularly their relationship with extracellular matrix (ECM) alterations, remains limited. Addressing this gap is essential for a comprehensive understanding of CRC progression and the development of targeted therapies.

Methods: 24 tissue samples from 21 CRC patients, along with adjacent normal tissues (NAT), were collected and decellularized using a detergent-based method to enrich the ECM component. Proteomic analysis of ECM-enriched samples was performed using tandem mass tag (TMT) spectrometry, followed by statistical analysis including differential expression protein (DEP) analysis. Single-cell RNA sequencing (scRNA-Seq) data from public datasets were integrated and analyzed to delineate cell states within the TME. Bulk tissue RNA-Seq and bioinformatics analysis, including consensus molecular subtype (CMS) classification and single-cell level deconvolution of TCGA bulk RNA-seq data, were conducted to further explore gene expression patterns and TME composition.

Results: Differential cellular origin of the NAT and tumorous ECM proteins were identified, revealing 110 ECM proteins enriched in NAT and 28 ECM proteins in tumor tissues. Desmoplastic and WNT5A inflammatory fibroblasts were indicated as the sources of tumor-enriched ECM proteins, while ADAMDEC1 expressing fibroblasts and PI16 expressing fibroblast were identified as the sources of NAT-enriched ECM proteins. Deconvolution of bulk RNA-seq of CRC tissues discriminated CMS-specific fibroblast state, reflecting the biological traits of each CMS subtype. Specially, seven ECM genes specific to mesenchymal subtype (CMS4), including PI16 fibroblast-related 4 genes (SFRP2, PRELP, OGN, SRPX) and desmoplastic fibroblast-related 3 genes (THBS2, CTHRC1, BGN), showed a significant association with poorer survival in patient with CRC.

Conclusion: We conducted an extracellular matrix (ECM)-focused profiling of the TME by integrating quantitative proteomics with single-cell RNA sequencing (scRNA-seq) data from CRC patients. We identified the ECM proteins of NAT and tumor tissue, and established a cell-matrisome database. We defined mesenchymal subtype-specific molecules associated with specific fibroblast subtypes showing a significant association with poorer survival in patients with CRC. Our ECM-focused profiling of tumor stroma provides new insights as indicators for biological processes and clinical endpoints.