Identification and Verification of SLC6A15 Involved in Keloid via Bioinformatics Analysis and Machine Learning

Keloid is a fibroproliferative disorder that poses a challenge in clinical management. This study aims to identify and functionally annotate differentially expressed genes (DEGs) in keloid and explore the potential role of SLC6A15. The data were obtained from GEO (GSE218922 and GSE7890), and the DEGs and module genes were obtained with Limma and WGCNA. KEGG and GO enrichment analysis, and machine learning algorithms (Random Forest, Boruta, and XGBoost) were conducted to explore the keloid-related key genes. Finally, qRT-PCR was carried out to detect SLC6A15 mRNA expression, and CCK-8 and flow cytometry were employed to assess cell proliferation and apoptosis. We obtained 147 DEGs between keloid fibroblasts and normal fibroblasts, and 193 DEGs between keloid stem cells and normal stem cells, followed by acquisition of 40 intersection DEGs. These intersection DEGs were mainly enriched in external encapsulating structure organization, extracellular matrix organization, and were closely related to cytoskeleton in muscle cells and arrhythmogenic right ventricular cardiomyopathy (ARVC). WGCNA analysis identified five modules, with the blue modules showing a significant negative correlation with keloid. Afterwards, three machine learning methods were applied to analyze DEGs in keloid, identifying SLC6A15 as the most important gene. Further validation demonstrated that SLC6A15 was lowly expressed in keloid tissues and fibroblasts, and SLC6A15 overexpression inhibited proliferation and facilitated apoptosis in keloid fibroblasts. This study identified SLC6A15 as a potential biomarker for keloid, providing new research clues for the treatment target of this disorder.