Integrating single-cell and bulk transcriptomics to identify a SUMOylation-related prognostic signature and the oncogenic role of VPS72 in hepatocellular carcinoma.

INTRODUCTION: Hepatocellular carcinoma (HCC) is a highly aggressive malignancy characterized by limited therapeutic options and significant cellular heterogeneity. This study aims to delineate the SUMOylation-mediated transcriptomic landscape and develop a robust prognostic signature to predict survival and immunotherapy response in HCC patients. METHODS: Malignant epithelial cells were classified into SUMOylation-related subtypes using nonnegative matrix factorization. A prognostic signature (SUMOylation.Sig) was constructed via machine learning and validated across multiple cohorts. Immune infiltration, immunotherapy response, and drug sensitivity were evaluated. Furthermore, we performed functional validation of a core signature gene, VPS72, through siRNA-mediated knockdown, colony formation and wound healing assays in HCC cells. RESULTS: We identified 10 prognostic SUMOylation-related genes (SRGs) and stratified malignant epithelial cells into five SUMOylation subtypes. The SUMOylation-C4 subtype was associated with favorable outcomes and enhanced immune infiltration. A 13-gene prognostic SUMOylation.Sig was developed and validated, effectively stratifying patients into low- and high-risk categories with distinct survival and responsiveness to immunotherapy. Low-risk patients exhibited increased immune cell infiltration and superior responses to immune checkpoint inhibitors. Experimental validation confirmed differential expression of signature genes in HCC cell lines. Notably, in vitro experiments demonstrated that VPS72 knockdown significantly suppressed the proliferation and migration of HepG2 and LM3 cells, confirming its oncogenic potential. CONCLUSIONS: Our study establishes a SUMOylation-based cellular taxonomy and a robust 13-gene prognostic tool for HCC. The experimental validation of VPS72 underscores the functional importance of the signature, suggesting its potential involvement in tumor progression and highlighting potential therapeutic targets for personalized clinical management.