Characterization and clustering of kinase isoform expression in metastatic melanoma

Abstract

Mutations to the human kinome are known to play causal roles in cancer. The kinome regulates numerous cell processes including growth, proliferation, differentiation, and apoptosis. In addition to aberrant expression, aberrant alternative splicing of cancer-driver genes is receiving increased attention as it could lead to loss or gain of functional domains, altering a kinase’s downstream impact. The present study quantifies changes in gene expression and isoform ratios in the kinome of metastatic melanoma cells relative to primary tumors. We contrast 538 total kinases and 3,040 known kinase isoforms between 103 primary tumor and 367 metastatic samples from The Cancer Genome Atlas (TCGA). We find strong evidence of differential expression (DE) at the gene level in 123 kinases (23%). Additionally, of the 468 kinases with alternative isoforms, 60 (13%) had significant difference in isoform ratios (DIR). Notably, DE and DIR have little correlation; for instance, although DE highlights enrichment in receptor tyrosine kinases (RTKs), DIR identifies altered splicing in non-receptor tyrosine kinases (nRTKs). Using exon junction mapping, we identify five examples of splicing events favored in metastatic samples. We demonstrate differential apoptosis and protein localization between SLK isoforms in metastatic melanoma. We cluster isoform expression data and identify subgroups that correlate with genomic subtypes and anatomic tumor locations. Notably, distinct DE and DIR patterns separate samples with BRAF hotspot mutations and (N/K/H)RAS hotspot mutations, the latter of which lacks effective kinase inhibitor treatments. DE in RAS mutants concentrates in CMGC kinases (a group including cell cycle and splicing regulators) rather than RTKs as in BRAF mutants. Furthermore, isoforms in the RAS kinase subgroup show enrichment for cancer-related processes such as angiogenesis and cell migration. Our results reveal a new approach to therapeutic target identification and demonstrate how different mutational subtypes may respond differently to treatments highlighting possible new driver events in cancer.

SoMAS: Finding somatic mutations associated with alternative splicing in human cancers

“…Comparing to other TCGA-based clustering schemes with variation of information (VoI) analysis (see Methods), we observed the strongest concordance of our Isoform clustering scheme with cancer type and mRNA, followed by miRNA and iCluster (Figure 1G). The VoI analysis further shows that the gene isoform expression provides higher resolution of cancer type classification compared to mRNA at the gene level (1.6 vs. 2.1 in VoI), which is consistent with our previous finding that the differential isoform ratio has little correlation with the differential expression at gene level in the human kinome in SKCM cancer [4].…”

“…Aberrant AS is common in human cancers and affects cancer progression in many aspects known as the 'hallmarks of cancer' [2][3][4]. Somatic disease-causing mutations, or driver mutations, are commonly assumed to exert their effects by altering amino acid sequences, which eventually impairs the normal function of a protein [5][6][7].…”

SEPepQuant enables comprehensive protein isoform characterization in shotgun proteomics

“…Therefore, this exon skipping may lead to reduced SLK activity and decreased apoptosis to facilitate tumor progression. Consistently, a recent analysis of RNASeq data from melanoma tumors has shown that expression of the long isoform is decreased whereas the short isoform is increased in metastatic tumors compared with primary tumors, suggesting a role of the exon skipping in facilitating metastasis 31 . Transferrin is another gene found to be correlated with poor prognosis at the SEPEP level but not gene level.…”

SEPepQuant enhances the detection of possible isoform regulations in shotgun proteomics

“…Therefore, this exon skipping may lead to reduced SLK activity and decreased apoptosis to facilitate tumor progression. Consistently, a recent analysis of RNASeq data from melanoma tumors has shown that expression of the long isoform is decreased, whereas the short isoform is increased in metastatic tumors compared with primary tumors, suggesting a role of the exon skipping in facilitating metastasis 38 . Transferrin is another gene found to be correlated with poor prognosis at the SEPEP level but not gene level.…”