Screening and identification of potential prognostic biomarkers in metastatic skin cutaneous melanoma by bioinformatics analysis

Abstract Skin cutaneous melanoma (SKCM) is a multifactorial disease that presents a poor prognosis due to its rapid progression towards metastasis. This study focused on the identification of prognostic differentially expressed genes (DEGs) between primary and metastatic SKCM. DEGs were obtained using three chip data sets from the Gene Expression Omnibus database. The protein‐protein interaction network was described by STRING and Cytoscape. Kaplan‐Meier curves were implemented to evaluate survival benefits within distinct groups. A total of 258 DEGs were distinguished as possible candidate biomarkers. Besides, survival curves indicated that DSG3, DSC3, PKP1, EVPL, IVL, FLG, SPRR1A and SPRR1B were of significant value to predict the metastatic transformation of melanoma. To further validate our hypotheses, functional enrichment and significant pathways of the hub genes were performed to indicate that the most involved considerable path. In summary, this study identified substantial DEGs participating in melanoma metastasis. DGS3, DSC3, PKP1, EVPL, IVL, FLG, SPRR1A and SPRR1B may be considered as new biomarkers in the therapeutics of metastatic melanoma, which might help us predict the potential metastatic capability of SKCM patients, thus provide earlier precautionary treatments. However, further experiments are still required to support the specific mechanisms of these hub genes.

the Gene Expression Omnibus database. The protein-protein interaction network was described by STRING and Cytoscape. Kaplan-Meier curves were implemented to evaluate survival benefits within distinct groups. A total of 258 DEGs were distinguished as possible candidate biomarkers. Besides, survival curves indicated that DSG3, DSC3, PKP1, EVPL, IVL, FLG, SPRR1A and SPRR1B were of significant value to predict the metastatic transformation of melanoma. To further validate our hypotheses, functional enrichment and significant pathways of the hub genes were performed to indicate that the most involved considerable path. In summary, this study identified substantial DEGs participating in melanoma metastasis. DGS3, DSC3, PKP1, EVPL, IVL, FLG, SPRR1A and SPRR1B may be considered as new biomarkers in the therapeutics of metastatic melanoma, which might help us predict the potential metastatic capability of SKCM patients, thus provide earlier precautionary treatments.
However, further experiments are still required to support the specific mechanisms of these hub genes.

K E Y W O R D S
bioinformatics analysis, biomarker, metastatic melanoma, primary melanoma, prognosis patients with SKCM. 1 The currently known histopathologic features such as tumour thickness and ulceration status have been used for melanoma detection and prognosis prediction. 3 However, the inevitable biases in the measurements of these features affect their application in evaluating melanoma prognosis. In the last few years, there has been a rising interest in the bioinformatics analysis, which can be applied to illustrate large and complicated data sets associated with various cancers. In this study, we screened out differentially expressed genes (DEGs) between primary and metastatic melanoma tissue and utilized bioinformatics analysis to distinguish hub genes and a range of functional enrichment. We are trying to identify the signatures of gene expression that associate with metastasis and survival in melanoma and find out more effective metastasis-associated biomarkers to achieve precision medicine.

| Data collection and DEGs screening
DEGs were obtained from three chip data sets on the Gene Expression Omnibus database by using GEO2R (detailed in the Methods S1).

| Functional enrichment analysis of DEGs
The GO and KEGG pathway analyses of DEGs were performed using Database for Annotation, Visualization and Integrated Discovery (detailed in the Methods S1).

| Construction of PPI network and identification of hub gene
A protein-protein interaction network was drawn with the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) to distinguish the hub genes and explore the interplays among the DEGs (detailed in the Methods S1).

| Validation of hub genes
To further screen the significant hub genes, GraphPad Prism software was utilized to illustrate the differential expression of 369 metastatic melanoma and 103 primary melanoma samples from TCGA database (detailed in the Methods S1).

| Kaplan-Meier survival analysis
Kaplan-Meier analyses were performed in GraphPad Prism software to investigate the correlation between the hub genes expression and the overall survival of patients with SKCM (detailed in the Methods S1).

| Hub genes analysis
See details in the Methods S1.

| Transcription factor network
See details in the Methods S1.

| Hub genes screening between primary and metastatic melanoma tissue
Among the three data sets, 258 genes were overlapped between primary and metastatic SKCM ( Figure 1B,C). The top 10 of the most significant results for functional enrichment were presented, respectively, in Figure 1D. Using the STRING online database and the MCODE plug-in from Cytoscape, a sum of 21 nodes and 209 edges were clustered into the PPI network complex ( Figure 1E-F). Heat map, based on TCGA cohort, showed that potential co-expression relationships between primary and metastatic SKCM might be found in the 21 hub genes ( Figure 1G).

| Clinicopathological statistical analysis and survival outcomes
Based on the TCGA database, the expression of twelve hub genes was higher in primary tissues than in metastatic tissues (P < .05, Figure 1H), and the other nine genes showed no significant difference. After screening the more relevant hub genes, we conducted the survival analysis of the hub gene by using the Kaplan-Meier curves. The outcomes revealed that overexpression of DSG3, DSC3, PKP1, EVPL, IVL, FLG, SPRR1A and SPRR1B genes predicted worse OS (P < .05) in SKCM patients ( Figure 2C).  The previous research demonstrated that these genes performed essential roles in the processes of epidermal development, keratinocyte differentiation, cell-to-cell signalling and cell adhesion. 4 The  11 EVPL is a member of the plakin family of proteins that forms an element of desmosomes and the epidermal cornified envelope. 12 Hu N et al showed that EVPL was lower-expressed in esophageal squamous cell carcinoma (ESCC) compared to healthy tissue, and it may be applied for early detection of ESCC. 13 However, the function of desmosomes in the progression of SKCM remains unclear.

| Co-expression network analysis and transcription factor network
The cornified envelope of the skin is a sizeable insoluble polymer composed by cross-linking of several protein precursors, including IVL, keratolinin, FLG and loricrin. SPRR1 gene encodes a precursor of the keratinocyte cornified envelope, which displays in terminally differentiating human keratinocytes. FLG located on chromosomal locus 1q21.3, which is a reported susceptibility site in SKCM. Other genes on 1q21.3 code for proteins also focus on the terminal differentiation of keratinocytes. 14 They present a crucial role in establishing and maintaining the epidermal barrier. Disruption of the integrity and stability of the epidermal barrier was a hallmark of cancer. Filaggrin can be degraded into free amino acids, which produce the natural moisturizing factor of the epidermis. Decreased variation in FLG was a significant risk factor for atopic dermatitis. Loss of function in FLG was assumed to enhance the susceptibility of skin malignancies due to reduced levels of its degradation products, urocanic acid, which may be protective towards ultraviolet irradiation. 15 However, few findings were focusing on the value of these genes in melanoma metastasis. Our present findings will encourage further investigations of the clinical significance of hub genes in metastatic SKCM.
In summary, this study identified significant DEGs participating in melanoma metastasis. Down-regulated genes, including DGS3, DSC3, PKP1, EVPL, IVL, FLG, SPRR1A and SPRR1B, may be considered as new biomarkers in the therapeutics of metastatic melanoma, which might help us predict the potential metastatic capability of SKCM patients, thus provide earlier precautionary treatments.
However, further experiments are still required to support the specific mechanisms of these hub genes.

ACK N OWLED G EM ENTS
We would like to appreciate the Grammarly (https://www.gramm arly. com/upgrade) for editing the English text of a draft of this manuscript.

CO N FLI C T O F I NTE R E S T
The authors declared that they have no conflicts of interest in this work.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in the Gene Expression Omnibus (GEO) database at https://www. ncbi.nlm.nih.gov/geo/ (reference number GSE46517, GSE15605 and GSE8401) and in The Cancer Genome Atlas (TCGA) database at https://genme-cancer.ucsc.edu/ (cohort: TCGA Melanoma IlluminaHiSeq, n = 474, TCGA Hub).