Characterizing mapk signaling in different cancers Through large public datasets

Melanoma is the most common form of cancer that originates in the skin; decades of intensive investigation generated large datasets that are currently provided to the open public. Large public datasets such as The Cancer Genome Atlas (referred to as TCGA), reveal that the majority of cases for both melanoma and papillary cancer (originating in the thyroid gland) harbor mutations affecting the BRAF gene. The most commonly occurring mutation is missense mutation p.Val600Glu (or p.V600E in one letter code), this particular mutation is associated to hyper activation of downstream protein MAPK1 that causes increased cell proliferation and inhibited apoptosis. Two drugs have been developed that specifically inhibit the mutated BRAF protein, thereby restoring the cell growth signaling to normal levels and effectively inhibiting tumor growth. It has however been observed that tumors can quickly become resistant to these treatment, particularly in colon BRAF mutated cancers. We hypothesize that characterizing alternative signaling for specific subsets, for example those showing normal MAPK1 activity despite BRAF alteration p.V600E, would reveal alternative regulatory mechanisms which could lead to novel targets for drug development. A popular choice to classify complex biological data, is the random forest technique, since it is nonparametric, interpretable and has high prediction accuracies in biologically relevant settings. Using TCGA data we generate four subsets from patients based on MAPK1 activity (high or low phosphorylation values) and presence or absence of BRAF mutation p.V600E. We classify gene measurements with 10.000 trees, iterate 100 times and use the most important (top 10%) measurements to identify signaling routes that bypass BRAF regulation. We perform classification on both melanoma and papillary cancer separately as well as on combined datasets. In the comparison of patients with BRAF mutation p.V600E and decreased MAPK1 activity to other subsets we identified increased BDKRB2, PCLB1, ITPR2 and NOS3 gene expression, and decreased AKT phosphorylation values specifically for this subset. We hypothesize that this is a mechanism that activates cell-proliferation independent of BRAF by NOS3 activation through calcium second messenger systems. Comparison of patients with unaltered BRAF genes and increased MAPK1 activity to other subsets was used to identify signaling that is not specifically BRAF-driven. Classification results show gene measurements to characterize this subset through gene expression and phosphorylation values of PTEN and AKT, MTOR, RPTOR, RBS6KB and EIF4EBP. This suggest circumvention of BRAF regulation through mtor signal transduction regardless of AKT inhibition mediated through PTEN activity.