Integrated analysis of multi-omics data reveals new-functional signature to predict intestinal Cancer Stem Cells

Cancer Stem Cells (CSCs) are a low abundant subpopulation of cancer cells with self-renewal and differentiation abilities, responsible for resistance to therapies. Even though CSCs have been widely studied and characterized from a genomics, transcriptomics and functional perspective, the identification of specific functional CSC markers remains challenging.

Here, we aim to uncover key functional features of intestinal CSCs by developing a novel in silico bioinformatics pipeline mining muti-omics datasets.

To this end, we generated a set of multi-omics datasets (i.e.: RNA-seq, proteomics, tRNA-seq and Ribo-seq) on FACS-sorted murine intestinal healthy stem cells (SCs) and CSCs, or of whole healthy intestinal and cancer cell population from organoids. Interestingly, the use of gold-standard CSCs markers and pathways failed to discriminate CSCs among other cells. Therefore, we developed a bioinformatics novel approach to systematically extract CSCs functional relevant and specific signatures rather than signatures shared among different cell types.

Our novel approach takes advantage of renowned unsupervised clustering approaches to identify unique contributing features of CSCs from multi-omics extracted features (i.e.: codons frequencies, genes expressions or proteins intensities). Strikingly, we identify a new proteomics signature able to cluster CSCs among all other intestinal cells. Importantly, our signature is more performant than all other tested transcriptomics signatures in clustering CSCs and is also able to predict CSCs among all other cell types. In fact, through the use of logistic regression machine learning model, we demonstrated that our signature outperforms all the other ones in predicting CSCs. The functional impact of our signature on CSC fitness is being currently tested by a genome wide CRISPR-Cas9 drop out approach.

Taken together, our methodology uncovered a new predictive signature of CSCs in the intestine and has the potential to identify low abundant cell populations in other relevant biological contexts.