Bioinformatic-based genetic characterizations of neural regulation in skin cutaneous melanoma

BACKGROUND: Recent discoveries uncovered the complex cancer–nerve interactions in several cancer types including skin cutaneous melanoma (SKCM). However, the genetic characterization of neural regulation in SKCM is unclear. METHODS: Transcriptomic expression data were collected from the TCGA and GTEx portal, and the differences in cancer–nerve crosstalk-associated gene expressions between normal skin and SKCM tissues were analyzed. The cBioPortal dataset was utilized to implement the gene mutation analysis. PPI analysis was performed using the STRING database. Functional enrichment analysis was analyzed by the R package clusterProfiler. K-M plotter, univariate, multivariate, and LASSO regression were used for prognostic analysis and verification. The GEPIA dataset was performed to analyze the association of gene expression with SKCM clinical stage. ssGSEA and GSCA datasets were used for immune cell infiltration analysis. GSEA was used to elucidate the significant function and pathway differences. RESULTS: A total of 66 cancer–nerve crosstalk-associated genes were identified, 60 of which were up- or downregulated in SKCM and KEGG analysis suggested that they are mainly enriched in the calcium signaling pathway, Ras signaling pathway, PI3K-Akt signaling pathway, and so on. A gene prognostic model including eight genes (GRIN3A, CCR2, CHRNA4, CSF1, NTN1, ADRB1, CHRNB4, and CHRNG) was built and verified by independent cohorts GSE59455 and GSE19234. A nomogram was constructed containing clinical characteristics and the above eight genes, and the AUCs of the 1-, 3-, and 5-year ROC were 0.850, 0.811, and 0.792, respectively. Expression of CCR2, GRIN3A, and CSF1 was associated with SKCM clinical stages. There existed broad and strong correlations of the prognostic gene set with immune infiltration and immune checkpoint genes. CHRNA4 and CHRNG were independent poor prognostic genes, and multiple metabolic pathways were enriched in high CHRNA4 expression cells. CONCLUSION: Comprehensive bioinformatics analysis of cancer–nerve crosstalk-associated genes in SKCM was performed, and an effective prognostic model was constructed based on clinical characteristics and eight genes (GRIN3A, CCR2, CHRNA4, CSF1, NTN1, ADRB1, CHRNB4, and CHRNG), which were widely related to clinical stages and immunological features. Our work may be helpful for further investigation in the molecular mechanisms correlated with neural regulation in SKCM, and in searching new therapeutic targets.