Machine Learning Identifies Robust Matrisome Markers and Regulatory Mechanisms in Cancer

Abstract

The expression and regulation of matrisome genes—the ensemble of extracellular matrix, ECM, ECM-associated proteins and regulators as well as cytokines, chemokines and growth factors—is of paramount importance for many biological processes and signals within the tumor microenvironment. The availability of large and diverse multi-omics data enables mapping and understanding of the regulatory circuitry governing the tumor matrisome to an unprecedented level, though such a volume of information requires robust approaches to data analysis and integration. In this study, we show that combining Pan-Cancer expression data from The Cancer Genome Atlas (TCGA) with genomics, epigenomics and microenvironmental features from TCGA and other sources enables the identification of “landmark” matrisome genes and machine learning-based reconstruction of their regulatory networks in 74 clinical and molecular subtypes of human cancers and approx. 6700 patients. These results, enriched for prognostic genes and cross-validated markers at the protein level, unravel the role of genetic and epigenetic programs in governing the tumor matrisome and allow the prioritization of tumor-specific matrisome genes (and their regulators) for the development of novel therapeutic approaches.

A guide to the composition and functions of the extracellular matrix

“…The matrisome has been experimentally characterized by mass spectrometry in various healthy and diseased tissues including skin [93], the cerebrovascular system [94], normal and fibrotic human liver [95–97], and normal and fibrotic human lung [98]. The matrisome has also been investigated in numerous cancer types [99–103], leading to the definition of a tumor matrisome index measuring deregulated matrisome associated with tumor progression [104], to determine cancer‐induced changes and cancer markers [105]. Proteomic data from 17 studies on the ECM of 15 normal tissue types, six cancer types, and other diseases including vascular defects and lung and liver fibrosis have been curated and compiled in MatrisomeDB, the ECM‐protein knowledge database (http://www.pepchem.org/matrisomedb) [106].…”

The Burden of Post-Translational Modification (PTM)—Disrupting Mutations in the Tumor Matrisome

“…On the other hand, approximate time-series of colon and lung cancer show that mutational events that affect structural matrisome components such as fibronectin 1 and collagen 1 ( FN1 and COL1A1 ) as well as proteinases ( MMP2 and ADAM10 ) and other functional moieties ( NTN4 , PCSK6 , and SVEP1 ) are likely tumor driver [ 70 ]—though, again, these data are on a different scale than that of this manuscript, so comparison is only approximate and conceptual; it is worth noticing, however, that we found PTM mut for all these genes but NTN4 , that FN1 , COL1A1 , MMP2 , and ADAM10 all harbor PTM mut in functional domains and interact reciprocally and that, in the network of interactions between PTM mut -affected matrisome elements, these genes (especially FN1 and MMP2 ) are major hubs, all these evidence suggesting that matrisome PTM mut might significantly contribute to altered TME dynamics. Further along this line, we notice that 143 PTM mut are not found at all in healthy samples from TCGA irrespective of tumor type ( Table S10 ) and that 9 PTM mut affect “landmark” matrisome genes that deeply characterize the given tumors [ 71 ], possibly candidating the carrier genes to a more prominent role in cancer.…”

Recreating the extracellular matrix: novel 3D cell culture platforms in cancer research

“…The main challenge right now remains the creation of novel platforms, including matrix‐free and matrix‐based 3D culturing, 3D bioprinting and PDOs, which recapitulate microenvironmental cues that mirror the in vivo pathology of the disease. In this context, combining knowledge from bioengineering with the availability of multi‐omics data could offer revolutionary possibilities in understanding the role of genetic/epigenetic programs governing the tumour matrisome and allow the design of biomimetic models through targeting specific ECM biomarkers [56]. The stochastic evaluation of all the microenvironmental interactions will shed light on the high‐importance factors that drive specific phenotypes and in turn will allow the design of scaffolds that most accurately mimic these conditions.…”