Integration of single-cell and bulk RNA sequencing data to identify liquid–liquid phase separation-related prognostic biomarkers in hepatocellular carcinoma

Liquid–liquid phase separation (LLPS) has been linked to the initiation and progression of cancers. This study aimed to investigate the LLPS-related molecular features and develop a prognostic signature for hepatocellular carcinoma (HCC). Single-cell RNA sequencing (scRNA-seq) and bulk RNA-seq data were obtained from open-source repositories. The Seurat package was used for processing and analyzing scRNA-seq data. LLPS-related genes (LLPSRGs) were identified by integrating scRNA-seq and bulk RNA-seq data. Univariate Cox and LASSO regression analyses were employed to construct an LLPS-related prognostic risk model. A predictive nomogram was also developed based on the risk score and clinicopathologic features. Additionally, correlations between clinicopathological features, biological functions, and tumor mutations with prognostic risk were explored. Experimental validation included cell transfection, wound healing, and transwell assays. Fourteen distinct cell clusters were identified, with malignant hepatocytes exhibiting the highest LLPS score. High-LLPS hepatocytes strongly interacted with other cells, showing elevated expression of EGFR-ERGF, EGFR-AREG, MIF-CD44, and MIF-CXCR4 interactions. The LLPS score was associated with malignant differentiation of hepatocytes. Ten LLPSRGs were identified as part of the LLPS-related risk signature, and potential agents were predicted: Olitigaltin and Lactose for LGALS3, and Sitamaquine, Phenazopyridine, Sulfanilamide, Pamaquine, Sodium ascorbate, and Co-trimoxazole for G6PD. Knockdown of LGALS3 inhibited HCC cell migration and invasion. LLPS-related molecular characteristics were identified, and an LLPS-related prognostic model was developed, offering novel insights into HCC research and providing potential therapeutic targets.