From genomics to signaling networks in triple-wt melanomas

Genome projects have revealed, within a highly altered genome, genetic drivers of human melanoma, whose targeting resulted in first therapeutic breakthroughs. However, a considerable subset of cutaneous melanomas are left without such validated driver alterations (triple-wt melanomas). Genes can be grouped into (signaling) pathways to describe distinct tumor-promoting activities and a better functional knowledge of these pathways and their interplay may pave the way to the identification of novel therapy targets. We here suggest building a broader contextualized view of signaling network activities in triple-wt melanomas and hypothesize that such information will allow prediction of novel therapy targets/responses where genomic analyses have failed so far. We will (i) extend a reference network of signalling pathways in melanoma from public genomic and drug response data with genomic and (phospho)proteomic data from triple-wt patient-derived melanoma xenografts (PDX); (ii) refine these sub-networks through functional drug perturbation data; and (iii) simulate a potential future therapy selection strategy in patients through a preclinical in vivo model system with triple-wt/KIT-wt-PDX. The most innovative aspects of the proposal include the extraction of a systems biology view from own and public genomic, proteomic and drug response data to simulate a future therapy selection strategy with improved analytical tools. That way we expect to cluster tumors along equivalent signaling structures rather than genomic alterations alone. Though this application is a pilot project of show case nature, the joint forces of novel computational approaches relying on data integration, network theory and clustering, together with cutting edge (phospho)proteomic and drug screen technology can help to establish a new paradigm of melanoma as well as cancer cell biology (therapy) studies. We have set up an interdisciplinary team of investigators with complementary expertise that fulfills all the prerequisites required to successfully accomplish the project. We have profound expertise in systems-biology analyses (Jacques Colinge, Univ. of Montpellier); in proteomics (Keiryn L. Bennett, CeMM, Vienna; Markus Hartl, Univ. of Vienna); drug screen (Stefan Kubicek, CeMM, Vienna) technology; and in melanoma genomics, cell biology, 3D cell cultures, preclinical in vivo therapy models (Stephan N. Wagner, Med. Univ. of Vienna).

Human melanoma exhibits considerable molecular heterogeneity. The functional consequences of this diversity are still incompletely understood, this is especially true for the so-called tripleWT melanomas. To identify signaling pathways describing tumor-promoting activities in this melanoma genotype, we compared public datasets of human melanoma tissues (incl. the Cancer Genome Atlas (TCGA) database) for different genotypes of human cutaneous melanoma. While genetic background was not the main determinant of the presence of additional genetic alterations in these tissues, differential gene expression analysis identified genes that are differentially regulated in the triple WT subtype. To enhance these transcript-only data, we developed a bioinformatics tool, ReactomeGSA, that allows researchers to rapidly compare data sets from proteomics, transcriptomics, and microarray experiments at the pathway level, regardless of the species studied. To our surprise, ReactomeGSA revealed that most of the differences between tripleWT and other genetic melanoma types were not in the signaling of the tumor cells themselves, but in the immune cells of the surrounding tumor microenvironment, particularly in signatures indicative of changes in B and T lymphocytes. We therefore focused on characterizing one such regulated cell type in human melanomas, namely B lymphocytes. By integrating data from five Cancer Genome Atlas (TCGA) transcriptomics studies and two Clinical Proteomic Tumor Analysis Consortium (CPTAC) proteomics studies with the ReactomeGSA tool, we identified a B cell subtype associated with unfavorable prognosis in melanoma patients. This analysis of "-omics" data was complemented by characterization of B-cell subtypes directly in human melanoma tissue. We were the first to develop a 7-color multiplex immunofluorescence-based classification for human B cell subtypes directly in tissue for this purpose. Using this technique, we were able to detect significant changes in the composition and function of B cell subtypes associated with melanoma progression. To further characterize these B cell subtypes at the molecular level, we have developed a new R/Bioconductor package, scAnnotatR, that will enable future characterization of B cells in scRNASeq datasets with unprecedented detail. In seven peer-reviewed publications in prestigious journals, we were able to describe the development of novel bioinformatics and immunology tools for improved molecular characterization of human tumor tissue. Based on a genomic analysis of different genetic subtypes of melanoma, we thus found first evidence for significant differences in immune cells, in particular B lymphocytes, of the respective tumor microenvironment. These lymphocytes may serve as markers for tumor progression and presumably as targets for improved individualized therapies for patients.