Identification of a Treg-related gene signature for predicting prognosis and immunosuppression in skin cutaneous melanoma

Chao Lian, Rong Jin, Xuanfen ZhangClinical 2025 · DOI

TL;DR

Researchers have identified a 10-gene signature specifically linked to Regulatory T cells (Tregs) that accurately predicts survival outcomes and immune suppression levels in skin cutaneous melanoma (SKCM). By combining big-data bioinformatics with laboratory validation, the study offers a new "risk score" tool that could help clinicians personalize immunotherapy and identifies 54 potential drugs to overcome tumor-induced immune resistance.

What Problem Does This Paper Address?

Skin cutaneous melanoma (SKCM) is a highly aggressive cancer. While modern immunotherapies like PD-1 inhibitors have revolutionized treatment, nearly 40–60% of patients do not respond or eventually develop resistance.

A major culprit behind this resistance is the "immunosuppressive tumor microenvironment." Regulatory T cells (Tregs) act as the immune system's brakes; in a healthy body, they prevent autoimmunity, but in a tumor, they shield cancer cells from attack. The authors sought to move beyond general cancer markers and identify a specific genetic "fingerprint" related to Treg activity. Identifying these genes helps explain why some patients fail therapy and provides new targets to "release the brakes" on the immune system.

Methods

The study employed a multi-stage computational and experimental pipeline:

  1. WGCNA (Weighted Correlation Network Analysis): Instead of looking at genes in isolation, the authors used WGCNA to find clusters (modules) of genes that co-vary specifically with the abundance of Tregs (estimated via the CIBERSORT algorithm).
  2. Signature Construction: They intersected these Treg-related modules with Differentially Expressed Genes (DEGs) found in melanoma vs. normal skin. They then used LASSO regression (to prevent overfitting) and Cox proportional hazards regression to whittle the list down to the 10 most prognostic genes.
  3. Regulatory Modeling: The team built a ceRNA (competing endogenous RNA) network, mapping how long non-coding RNAs (lncRNAs) might act as "sponges" for microRNAs to regulate the 10 signature genes.
  4. Experimental Validation: Unlike many purely computational papers, the authors validated their findings in the lab using qRT-PCR (to measure mRNA) and Western blots (to measure protein levels) across three melanoma cell lines (A375, SK-MEL-28, SK-MEL-5) compared to a normal skin cell line (HaCaT).

Datasets

The authors utilized several well-known public repositories:

  • TCGA-SKCM: Used as the primary training set (458 samples with sequencing and clinical data).
  • GSE46517 (GEO): Used to identify baseline differences between melanoma (104 samples) and healthy skin (7 samples).
  • GSE65904 (GEO): Used as an independent validation set (210 samples) to ensure the model works on different patient populations.

Key Insights and Ideas

  • The 10-Gene Signature: The final model relies on PTPRF, ULK1, TGM3, CRABP2, SV2A, HLA-DQB2, KHDRBS3, VWA5A, CRIP1, and TFAP2C.
  • Biological Divergence: High-risk patients (determined by the signature) showed enrichment in keratinization pathways, suggesting their tumors undergo structural changes that might physically shield them. Low-risk patients showed high activity in autoimmune and infection-related pathways, indicating a "hot" immune environment ready to fight the cancer.
  • The ceRNA Axis: The study highlights a specific regulatory pathway—NEAT1/miR-375/CRABP2—suggesting that the lncRNA NEAT1 might drive the expression of the high-risk gene CRABP2 by sequestering its inhibitory microRNA.
  • Metabolic Links: Several signature genes are linked to metabolism (e.g., ULK1 in autophagy and CRABP2 in retinoic acid signaling), suggesting that Tregs maintain their suppressive power by out-competing other cells for nutrients or altering the chemical environment of the tumor.

Key Results

  • Prognostic Power: The 10-gene risk score was a robust independent predictor of survival. The Area Under the Curve (AUC) for predicting 1-, 3-, and 5-year survival was consistently above 0.6 in both training and validation cohorts.
  • Survival Gap: Patients classified as "high-risk" had significantly shorter overall survival compared to "low-risk" patients (p < 0.0001).
  • Lab Confirmation: Four key "risk" genes (PTPRF, ULK1, TGM3, and CRABP2) were confirmed to be significantly upregulated at both the mRNA and protein levels in melanoma cells compared to healthy controls.
  • Drug Repurposing: The study identified 54 potential drugs that could target this signature, including resveratrol and metronidazole, providing a shortlist for future drug-repurposing trials.

Limitations and Future Work

The authors acknowledge several areas for improvement:

  • Retrospective Nature: The study relies on existing databases; prospective clinical trials are needed to see if the risk score can guide treatment in real-time.
  • Missing Clinical Variables: The datasets lacked detailed information on common melanoma mutations (like BRAF or NRAS) and specific details on the types of immunotherapy patients received.
  • Functional Gap: While the genes are correlated with survival, the study did not perform "loss-of-function" experiments (like CRISPR knockouts) to prove that silencing these genes directly reduces tumor growth or Treg infiltration.
  • Validation Scale: Future work should involve larger, multi-center clinical cohorts and more diverse ethnic backgrounds to ensure the signature is universally applicable.

Summary updated

Abstract

Skin cutaneous melanoma (SKCM) is an aggressive malignancy where regulatory T cells (Tregs) drive an immunosuppressive tumor microenvironment, resulting in poor prognosis. Thus, there was an urgent need to identify Treg-related molecular biomarkers to optimize SKCM's prognostic assessment and therapeutic strategies. In this study, SKCM-related datasets were acquired from public databases. First, gene modules associated with Tregs screened using Weighted Correlation Network Analysis were intersected with differentially expressed genes to obtain Treg-DEGs. Subsequently, univariate Cox proportional hazards regression, LASSO regression, and multivariate Cox proportional hazards regression were employed to construct a prognostic biomarker signature. Furthermore, the biological functions of the prognostic biomarkers were explored by integrating functional enrichment analysis, molecular regulatory network construction, and drug prediction analysis. Finally, in cellular experiments, the mRNA and protein expression levels of the biomarkers were validated using qRT-PCR and Western blot. The risk model constructed based on the 10 prognostic biomarkers (PTPRF, ULK1, TGM3, CRABP2, SV2A, HLA-DQB2, KHDRBS3, VWA5A, CRIP1, and TFAP2C) could well predict the overall survival of SKCM patients. Functional enrichment analyses indicated that high-risk patients were enriched in keratinization pathways, whereas low-risk patients showed activation of autoimmune and infection-related pathways. NEAT1 might have regulated CRABP2 via miR-375. Additionally, 54 potential drugs, including resveratrol and metronidazole, were predicted for targeted therapy. qRT-PCR and Western blot confirmed PTPRF, ULK1, TGM3, and CRABP2 were upregulated at both the mRNA and protein levels. These findings indicate that the Treg-related signature serves as robust prognostic biomarkers and may guide personalized immunotherapy in SCKM.
BibTeX
@inproceedings{lian2025identification,
  author = {Chao Lian and Rong Jin and Xuanfen Zhang},
  title = {{Identification of a Treg-related gene signature for predicting prognosis and immunosuppression in skin cutaneous melanoma}},
  booktitle = {Clinical},
  year = {2025},
  doi = {10.1007/s10238-025-01941-3},
}
Figure