Single-cell transcriptomics identifies an -driven proliferative tumor subpopulation associated with poor prognosis in hepatocellular carcinoma.

[BACKGROUND] Hepatocellular carcinoma (HCC) is a highly heterogeneous cancer with complex tumor-immune interactions. This heterogeneity, particularly in tumor and immune cells, complicates treatment and prognostic evaluation. Although recent studies have revealed distinct tumor cell states and immune dysfunction in HCC, the molecular basis underlying tumor aggressiveness remains poorly understood. A deeper understanding of the molecular and functional diversity of both tumor and immune cell populations, especially the identification of stem-like tumor subpopulations and immunosuppressive mechanisms, along with the development of robust prognostic biomarkers, is essential for advancing precision oncology and improving clinical outcomes.

[METHODS] We integrated three publicly available single-cell RNA sequencing (scRNA-seq) datasets from GEO to delineate the cellular architecture of the HCC tumor microenvironment. Unsupervised clustering and dimensionality reduction were employed to identify major cell types and tumor subpopulations. Functional annotation was performed using canonical markers, Monocle, CytoTRACE, and AUCell scoring. was identified as a candidate oncogene and validated through knockdown experiments. The interaction between T cell subsets and tumor subpopulations were further characterized. A prognostic risk model was constructed using LASSO regression.

[RESULTS] Six major cell types were identified in HCC TME. Tumor cells were subdivided into three distinct clusters: Tumor_C0, Tumor_C1 and Tumor_C2. Tumor_C2 showed the highest stemness, pro-metastatic activity and immunogenic cell death signatures. was highly expressed in Tumor_C2 and associated with poor prognosis. The knockdown of reduced H2A.Z protein levels, inhibited proliferation, invasion, and induced apoptosis. T cell analysis revealed five subpopulations. It was found that Tumor_C2 interacts with the proliferative and exhausted T cell subpopulations, suggesting a potential functional relationship between them. The prognostic model based on tumor_C2 transcriptomic features effectively stratified patient survival across multiple cohorts, with robust AUCs and Kaplan-Meier survival distinctions.

[CONCLUSION] We identified a proliferative, stem-like tumor cell subpopulation (Tumor_C2) in HCC characterized by high expression, which drives tumor aggressiveness. T cell analysis revealed significant interactions with Tumor_C2. Moreover, a prognostic model based on Tumor_C2 features effectively stratified patient survival across multiple cohorts. Together, these findings highlight potential therapeutic targets for improving patient outcomes.