Single-cell sequencing reveals the functional heterogeneity of melanoma cells and their crosstalk with the tumor microenvironment

Background Cutaneous melanoma is a highly aggressive malignancy characterized by significant heterogeneity, rapid progression, and variable treatment responses. Understanding the functional diversity of melanoma cells and their interactions with the tumor microenvironment (TME) is crucial for developing effective therapeutic strategies and identifying prognostic biomarkers. Methods We performed comprehensive single-cell RNA sequencing (scRNA-seq) analysis of 70,760 cells from 11 melanoma samples. Data processing was conducted using Seurat v4.3.0 with Harmony integration. Cell-cell communication was inferred using CellChat, and pseudotime trajectory analysis was performed using Monocle 2. A prognostic model was constructed by integrating 10 machine learning algorithms within a leave-one-out cross-validation (LOOCV) framework using the TCGA-SKCM cohort. Experimental validation was performed using immunofluorescence analysis on clinical specimens from seven melanoma patients. Results We identified seven major cell types and characterized nine distinct melanoma cell subpopulations with unique molecular signatures. Notably, subpopulations Mela4, Mela6, and Mela9 demonstrated significant associations with favorable patient prognosis and exhibited the highest interaction strength with immune cells in the TME. Cell communication analysis revealed that these subpopulations primarily engaged in signaling through MIF-CD74/CD44/CXCR4 and MHC-I pathways, with CD8 + T cells being the predominant signal recipients. Pseudotime trajectory analysis identified critical genes (CYR61, JUN, RHOC) involved in melanoma cell state transitions. Using an integrative machine learning approach, we developed a melanoma cell-associated signature (MRS) comprising 15 genes that achieved a mean C-index of 0.675 across validation cohorts. Furthermore, High EIF5A expression was significantly associated with poor patient outcomes (p < 0.001), Immunofluorescence analysis showing significantly elevated EIF5A expression in melanoma tissues compared to controls (p < 0.01). Conclusion This study reveals the functional heterogeneity of melanoma cells and their interactions with the immune microenvironment, identifies key subpopulations, prognostic signatures, and EIF5A as a plausible prognostic biomarker candidate and potential therapeutic target that warrants mechanistic validation in melanoma.