Pathway dysregulation analysis of the nucleotide excision repair mechanisms reveals it is not a common feature of melanomas

Abstract

Dear Editor, Ultraviolet Radiation (UVR) is the major environmental mutagen driving the development of melanoma. The major UVR‐induced DNA lesions, 6–4 photoproducts and cyclobutane pyrimidine dimers are primarily repaired by Nucleotide Excision Repair (NER). The very high levels of UV signature mutations (USM; Signature 7 mutations) that are observed in melanomas are the direct outcome of unrepaired UV‐induced lesions (Alexandrov et al., 2013; Hodis et al., 2012), suggesting that there must be defects in the repair of these lesions, and thus potentially in NER. For further description of signature 7 mutations, please refer to (Alexandrov et al., 2013). To date, there have been contradictory reports on the efficiency of NER in melanomas. Most of these studies have used small numbers of melanoma cell lines (Belanger, Rajotte, & Drobetsky, 2014; Budden et al., 2016; Gaddameedhi et al., 2010). To address this question in a larger dataset, we have examined the relationship between the expression of a comprehensive panel of NER component genes (KEGG NER pathway, 47 genes; Supporting information Table S1) using The Cancer Genome Analysis (TCGA) melanoma dataset (Akbani et al., 2015) and USMs for each sample from (Alexandrov et al., 2013). Our analysis used “Pathifier” (Drier, Sheffer, & Domany, 2013), a validated tool developed to quantify pathway deregulation in gene expression data. Pathifier transforms gene‐level information into pathway‐level information to model a Pathway Deregulation Score (PDS) for each sample (Drier et al., 2013; Huang, Yee, Ching, Yu, & Garmire, 2014; Liu et al., 2016). We quantify the amount of dysregulation of a pathway in a sample by measuring the deviation of the sample from normal behaviour. More specifically, the expression data is used to construct a principal curve using the Hastie and Stuetzle's algorithm, onto which each sample is projected. The PDS represents the projected distance of the sample along the curve, with a larger deviation from normal expression levels indicating a higher PDS, and therefore more dysregulation in the pathway. In our study, the PDS for each sample was related to the number of S7 mutations, to determine strength of association between dysregulation of the pathway and USM load. The TCGA data was filtered to only include Malignant Melanoma NOS and Nodular Melanoma. A total of 353 samples with complete RNAseq and USM data, including 67 primary melanomas was further analysed (Supporting information Table S2).