Contribution of transposable elements to drug resistance in melanoma

The main objective of this thesis was to investigate the involvement of the non- coding genome in drug resistance mechanisms of melanoma cells. Drug resistance is a major clinical problem as it renders currently available targeted therapies inefficient and is rapidly evolving in melanoma patients who are treated with combinations of Mitogen-activated protein kinase (MAPK) pathway and other specific inhibitors. For melanoma patients who carry a mutation in the NRAS gene (25% of patients), no clinically approved targeted therapy combinations are available. Combined cell cycle inhibition (targeting CDK4/6) and MAPK pathway inhibition (targeting MEK) are currently undergoing clinical trials. Here, we used this drug combination to evoke resistance in NRAS mutated cell lines and acquired different omics data sets along the adaptation processes to the treatment (Figure 1). First, we characterized the transcriptomic changes and performed Gene Set Enrichment Analysis to reveal deregulated biological processes. Our data show that all cell lines trigger an interferon and a senescence program. The interferon responses result from the expression of non-coding transcripts which trigger an innate immune response, a phenomenon called viral mimicry. Our data suggests an unforeseen role of interferon gamma in senescence escape. Then, we profiled the kinome of those cell lines and performed network- based enrichment to identify signaling pathways which could support adaptation to the treatment. Overlaying kinase activities on xml models of KEGG pathways allowed us to identify key molecular determinants. Our results associate insulin signaling and decreased activity of ERK5 with persistence of senescence. Moreover, we investigated the contribution of miRNAs to resistance. We acquired both small RNA-seq and miRNA interactions data and assessed their informativeness. We further constructed a miRNA-mRNA network and validated selected interactions by qPCR. Most miRNAs did not seem to exert a strong effect on their targets, as assessed by inverse expression patterns. Extending the analysis to 2 BRAF mutant cell lines, we found miR-211-5p to be upregulated in the proliferative cell lines. Finally, we integrated bulk and matching single-cell data to reveal perturbed processes and predicted novel genes associated with senescence escape. We applied a matrix factorization method to the adjacency matrices of a protein-protein interaction and a co-expression network together with the single- cell expression data and reconstructed condition-specific molecular network. Using graphlet-based topological analysis, we unrevealed perturbed biological processes and predicted a set of novel genes involved in senescence escape. Altogether, our results provide insights on how NRAS mutated melanoma cells adapt to the combination of CDK4/6 and MEK inhibitors. Our data are reminiscent of the Neural Crest Stem Cell (NCSC)-like state, which was further characterized as one of the Drug-Tolerant Persisters (DTPs) states in melanoma. DTPs are now considered important drivers of drug resistance in multiple cancer types. Our data reveal an unforeseen interplay between the expression of non-coding portions of the genomes, the activation of viral mimicry and senescence as processes underlying the DTP state.