Interpretation of Non-coding Mutations Driving Melanoma Risk and Its Comorbidities

Michael Pudjihartono, PhD Candidate, Dr. William Schierding, Senior Research Fellow, Prof Justin O’Sullivan, Liggins Institute

Home

•

Project

•

Interpretation of Non-coding Mutations Driving Melanoma Risk and Its Comorbidities

Background

Melanoma is the deadliest form of skin cancer with increasing worldwide incidence. Understanding the underlying mechanisms driving melanoma is crucial for better treatment and prevention. Over the past two decades, the field of genomics has entered a new golden age due to advancements in genetic sequencing and genotyping technologies, which have made genome-wide assessment of genetic mutations that associate with disease risk (i.e. Genome-wide Association Studies; GWAS) possible. However, the interpretation of GWAS remains a challenge because most Single Nucleotide Polymorphisms (SNPs) found to be associated with diseases, including melanoma, lie in the non-coding regions of the genome. These are regions of the genome which lie outside of genes and thus do not code for any proteins. Rather than affecting protein structure, it is hypothesized that modifying the expression level of distal target genes is one mode of possible mechanism by which these non-coding SNPs affect disease risk. In this study, we integrated publicly available data from melanoma GWAS with tissue-specific markers of the 3D DNA structure (Hi-C) and gene expression data (eQTLs) to identify the genes dysregulated by non-coding risk SNPs in three melanoma-relevant tissues: melanocytes, sun-exposed skin, and not sun-exposed skin. This approach essentially finds the connection from non-coding SNPs to genes, which help to paint a clearer picture on how non-coding SNPs can potentially contribute to the development of melanoma and its comorbidities.

Methods

GWAS data for melanoma was downloaded from the GWAS Catalog¹. Data on 3D genome organization (Hi-C) was downloaded from ENCODE². Expression quantitative trait loci (eQTL) data from melanocyte, sun-exposed skin and not sun-exposed skin was downloaded from GTEx³ and a previous study⁴. These data were then loaded into the CoDeS3D pipeline⁵. The CoDeS3D pipeline takes the GWAS SNPs as input and identify regulatory associations between each SNP and specific target genes that are supported by both physical interaction data (Hi-C) and expression data (eQTL) [Figure 1a]. These identified the final set of melanoma target genes. To identify other comorbid traits associated with melanoma, the melanoma target genes were used as input to identify every other SNPs in the genome that associate with the same set of target genes as melanoma [Figure 1b]. Hypergeometric tests (FDR≤0.05) and bootstrapping (n = 350) were then performed to identify enrichment of the SNPs within GWAS Catalog. This process was repeated separately using expression data from melanocyte, sun-exposed skin and not sun-exposed skin.

Figure 1. Overview of the computational pipeline used in this study. (a) The CoDeS3D pipeline takes GWAS risk SNPs as input and integrates information on the 3D genome organization to identify physical interactions between GWAS SNPs and genes. For each SNP-gene pair, expression quantitative trait loci (eQTL) data was used to assess whether the SNP also associate with changes in the expression level of the interacting gene. (b) The comorbidity analysis uses the final set of target genes as input and consisted of two parts. (1) Tissue-specific gene-regulatory networks (GRN) was generated by identifying every SNP-gene associations in the whole genome using CoDes3D. The GRN was then queried to identify all SNPs targeting the same genes as the melanoma target genes set. Hypergeometric tests (FDR ≤ 0.05) and bootstrapping (n = 350) were then performed to identify traits in the GWAS Catalog whose SNPs are enriched to target the same target genes as melanoma.

Results

Across the three melanoma relevant tissues, a total of 151 genes were identified as being regulated by melanoma GWAS risk SNPs [Figure 2a]. Functional profiling of the 151 target genes showed an enrichment of genes important for various biological processes including developmental processes, apoptosis, and pigmentation. These are consistent with the role of normal cell development, regulation of cell death, and UV damage in the risk of developing melanoma. The comorbidity landscape of melanoma includes a total of 64 significant traits across the three tissues [Figure 2b]. These include 1) Traits with obvious relevance to melanoma (e.g traits relating to cancer, nevus count, pigmentation and telomere length). 2) Traits that are supported by clinical and/or epidemiological observations (e.g increased intraocular pressure in patients with iris melanoma⁶, increased melanoma risk for patients with actinic keratosis⁷). And 3) traits that have not been (or weakly) associated with melanoma (e.g. uterine fibroids and spontaneous coronary artery dissection).

Notably, the comorbidity landscape of melanoma is facilitated by target genes that are tissue specific, with 98 genes (65% of target genes) specific to only one of the three tissue types [Figure 2a]. Nonetheless, the comorbid traits found in all three tissues largely agree with each other (e.g traits relating to cancer, pigmentation traits, telomere length, and nevus) [Figure 2b], suggesting a higher-level convergence at the phenotype and pathway level.

This study provides novel insights into the biological implications of non-coding SNPs associated with melanoma risk and provide a starting point for further experimental validation of functional variants and disease-related genes.

Figure 2. (a) Overlap of the target genes of melanoma risk SNPs found in each of the three tissues. (b) Melanoma comorbidity landscape found in each of the three tissues.

References

Buniello A, Macarthur JAL, Cerezo M, et al. The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics 2019. Nucleic Acids Res. 2019;47(D1):D1005-D1012. doi:10.1093/nar/gky1120
Dunham I, Kundaje A, Aldred SF, et al. An integrated encyclopedia of DNA elements in the human genome. Nature 2012 489:7414. 2012;489(7414):57-74. doi:10.1038/nature11247
Aguet F, Barbeira AN, Bonazzola R, et al. The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science (1979). 2020;369(6509):1318-1330. doi:10.1126/SCIENCE.AAZ1776
Zhang T, Choi J, Kovacs MA, et al. Cell-type-specific eQTL of primary melanocytes facilitates identification of melanoma susceptibility genes. Genome Res. 2018;28(11):1621-1635. doi:10.1101/gr.233304.117
Fadason T, Schierding W, Lumley T, O’Sullivan JM. Chromatin interactions and expression quantitative trait loci reveal genetic drivers of multimorbidities. Nat Commun. Published online 2018. doi:10.1038/s41467-018-07692-y
Shields CL, Materin MA, Shields JA, Gershenbaum E, Singh AD, Smith A. Factors associated with elevated intraocular pressure in eyes with iris melanoma. Br J Ophthalmol. 2001;85(6):666-669. doi:10.1136/BJO.85.6.666
Guorgis G, Anderson CD, Lyth J, Falk M. Actinic Keratosis Diagnosis and Increased Risk of Developing Skin Cancer: A 10-year Cohort Study of 17,651 Patients in Sweden. Acta Derm Venereol. 2020;100(8):1-6. doi:10.2340/00015555-3486