Identification and validation of a KRAS-macrophage-associated gene signature as prognostic biomarkers and potential therapeutic targets in melanoma

Abstract

Introduction: Skin cutaneous melanoma (SKCM) is a highly aggressive form of cancer with poor prognosis, characterized by significant molecular and immune heterogeneity. The activation of KRAS signaling pathways is implicated in melanoma progression, yet its role in shaping the tumor microenvironment, particularly in macrophage infiltration, remains poorly understood.

Methods: A comprehensive multi-platform approach was employed, analyzing gene expression data from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases. Gene set enrichment analysis (GSEA) was utilized to characterize the molecular pathways associated with KRAS signaling. Single-cell RNA sequencing (scRNA-seq) was leveraged to investigate the cellular heterogeneity within the SKCM tumor microenvironment, and macrophage populations were categorized using the Monocle2 algorithm. A KRAS-Macrophage Prognostic Associated Gene (KMPAG) signature was developed by integrating these findings, followed by validation using a least absolute shrinkage and selection operator (LASSO) regression model. The prognostic value of the KMPAG signature was assessed through its correlation with clinical outcomes, immune cell infiltration patterns, response to therapy, drug sensitivity, and miRNA-gene regulatory interactions. Cell-cell communication within the SKCM microenvironment was explored using the "CellChat" tool. Experimental validation of gene expression was performed via immunohistochemistry (IHC) and functional assays in gene-modified melanoma cell lines.

Results: Twenty-two genes involved in KRAS signaling were identified as critical for patient survival. Single-cell analysis revealed nine distinct cell populations within the SKCM microenvironment, leading to the construction of the KMPAG risk model, which incorporated three key genes-CLEC4A, CXCL10, and LAT2. This signature effectively reclassified macrophage subsets, offering improved diagnostic and prognostic capabilities. Furthermore, the KMPAG signature correlated with a range of clinical parameters, including immune infiltration levels, tumor stage, and therapy response. The model also provided insights into the immune landscape of SKCM, facilitating the prediction of responses to immunotherapy. Functional assays demonstrated that downregulation of CLEC4A significantly promoted melanoma cell proliferation, migration, and invasion.

Conclusion: This study highlights the importance of KRAS signaling and macrophage infiltration in melanoma prognosis. The KMPAG gene signature presents a novel prognostic tool, offering insights into personalized treatment strategies and predictive biomarkers for immunotherapy in SKCM. Further exploration of CLEC4A's role in melanoma progression may provide new therapeutic avenues for targeted intervention.

Keywords: KRAS signaling; immune microenvironment; macrophage infiltration; melanoma; prognostic biomarker; single-cell RNA sequencing.

Figure 1

Investigation of the KRAS Signaling Pathway Enrichment in SKCM Datasets via GSEA. (A) GSEA comparing primary melanomas vs. normal skin (GSE15605). (B, C) GSEA contrasting metastatic vs. primary melanomas (GSE8401, GSE46517), showing increased KRAS pathway activation in advanced disease.

Figure 2

Identification of hub genes related to survival within the KRAS signaling pathway. (A) Univariate Cox regression results for KRAS pathway genes in GSE65904. (B) Parallel analysis in TCGA-SKCM. (C) Venn diagram showing the 22 common survival-associated hub genes.

Figure 3

Single-cell RNA sequencing reveals cellular heterogeneity and KRAS hub gene expression patterns. (A) t-SNE clustering of melanoma single cells. (B) Annotation of principal cell types. (C) Heatmap of KRAS pathway marker expression across clusters. (D, E) Intersection of KRAS hub genes with cell-type markers, highlighting macrophages. (F, G) Monocle2 pseudotime trajectory analysis across all identified cell populations.

Figure 4

Construction and validation of the KMPAG prognostic risk model. (A, B) LASSO regression analysis selecting CLEC4A, CXCL10, and LAT2. (C–E) Risk score distribution, survival status, and Kaplan-Meier OS curves for high/low-risk groups in the TCGA training set. (F-H) Validation in the GSE65904 cohort.

Figure 5

Single-cell analysis of macrophage subpopulations. (A) Heatmap of differentially expressed genes in macrophage subclusters. (B) Pseudotime trajectory reconstruction of macrophage state transitions. (C) Branch plots indicating potential differentiation points. (D, E) Expression trends of KMPAG genes (CLEC4A, CXCL10, LAT2) along the macrophage pseudotime trajectory.

Figure 6

Diagnostic accuracy and functional enrichment analysis of KMPAG hub genes. (A–C) ROC curves evaluating the diagnostic performance of CLEC4A, CXCL10, and LAT2 using TCGA-SKCM vs. GTEx data. (D–F) GSEA plots for representative pathways associated with each hub gene. (G–I) GSEA plots for representative disease-related pathways associated with each hub gene.

Figure 7

Clinical correlations of the KMPAG signature in SKCM. (A) Heatmap associating KMPAG risk scores with clinical factors (T stage, ulceration, etc.). (B–D) Boxplots and statistical tests linking risk scores to T category, Clark level, and pathological stage in TCGA-SKCM.

Figure 8

Association between KMPAG signature and tumor immune infiltration. (A, B) CIBERSORT estimation and comparison of 22 immune cell types in high- vs. low-risk groups within the TCGA-SKCM cohort. (C, D) Corresponding immune infiltration analysis in the GSE65904 cohort. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 9

Evaluation of KMPAG signature association with immunotherapy response biomarkers. (A) Comparison of TIDE scores (T cell dysfunction, exclusion, M2 TAM) between KMPAG high- and low-risk groups in TCGA-SKCM. (B) Corresponding analysis in the GSE65904 cohort. (C, D) Validation attempt using clinical outcomes (PFS, Response Rate) in the external ICB-treated cohort GSE78220. ns: p > 0.05; ***p < 0.001.

Figure 10

Analysis of the cell-cell communication network in the melanoma TME. (A, B) CellChat summary plots showing the frequency and weight of intercellular interactions. (C) Heatmap illustrating cell types as signal senders/receivers. (D, E) Network analysis highlighting the CXCL signaling pathway and key interactions involving macrophages.

Figure 11

Expression analysis and clinical relevance of CLEC4A in melanoma. (A) CLEC4A expression comparison (TCGA-SKCM vs. GTEx). (B) CLEC4A expression across clinical groups in TCGA-SKCM. (C) Correlation of CLEC4A expression with immune infiltration. (D) Correlation of CLEC4A with CXCL10 and LAT2. (E) Representative IHC staining of CLEC4A in benign nevi vs. melanoma. (F) RT-qPCR analysis of CLEC4A mRNA levels in 8 benign nevi vs. 12 melanoma tissues. (G) Western blot analysis of CLEC4A protein expression in benign nevi [N] vs. melanoma tissues [M]. **p < 0.01; ***p < 0.001.

Figure 12

Functional investigation of CLEC4A’s role in melanoma cell behavior. (A) CLEC4A mRNA expression in melanoma cell lines. (B) Verification of CLEC4A knockdown/overexpression in A375 cells. (C-E) Proliferation assays (EdU, colony formation, CCK-8). (F, G) Migration and invasion assays (wound healing, Transwell). *p < 0.05; **p < 0.01; ***p < 0.001.