Scholay

学术搜索 · AI 审稿 · LaTeX 协作

HDAC2-mediated chromatin remodeling drives hepatocellular carcinoma progression: an integrative analysis of computational pathology and multi-transcriptomics

作者:Shiyang Yin, Xuancheng Zhou, Lai Jiang, Yuheng Gu, Jingxuan Dai, Haiqing Chen, Shengke Zhang, Jiaan Lu, Gang Huang, Yuxuan Jiang, Xun Sang, Kun Yuan, Hua Yang, Guanhu Yang, Shangke Huang, Hao Chi, Ke Xu · 发表于:Journal of Translational Medicine · 年份:2025 · DOI:10.1186/s12967-025-07517-9 · 被引用次数:3 · 研究领域:Ferroptosis and cancer prognosis、Bioinformatics and Genomic Networks、Histone Deacetylase Inhibitors Research

BACKGROUND: Hepatocellular carcinoma (HCC) is the most common primary liver cancer, characterized by high heterogeneity and poor prognosis. Chromatin remodeling regulates chromatin structure and function, and its dysregulation promotes tumor progression. Histone deacetylase 2 (HDAC2) is a key regulator involved in this process. This study links clinicopathological features with molecular mechanisms to elucidate the role of the chromatin-remodeling factor HDAC2 in hepatocellular carcinoma. METHOD: Based on whole-slide images (WSIs) from TCGA-LIHC, pathological features were extracted by combining ResNet-50 and CellProfiler to build a prognostic model. SHAP was subsequently applied to interpret the most influential predictive features. Through integrative multi-omics analysis, HDAC2 was identified as a core gene, and its association with malignant phenotypes was validated using the Human Protein Atlas (HPA), bulk transcriptomic data, and single-cell datasets. Malignant cells were further identified with inferCNVpy, while CellChat and Vector were used to characterize intercellular communication and developmental trajectories. Finally, spatial transcriptomics integrated with the Spotlight and MISTy algorithms revealed the spatial distribution of HDAC2 and its interactions within the tumor microenvironment. The pro-tumor capability of HDAC2 was verified by both cell-based assays and nude mouse tumorigenicity experiments. RESULTS: From WSIs, 630 features were extracted with CellPro...