Single-cell and spatial transcriptomics reveal correlation between RNA methylation-related miRNA risk model and immune infiltration in hepatocellular carcinoma.

Introduction: Increasing evidence highlights the pivotal role of RNA methylation and miRNAs in hepatocellular carcinoma (HCC). However, the risk associated with RNA methylation-related miRNAs (RMRMs) in the HCC immune microenvironment remains largely unknown. Here, we predicted the correlation between RMRM risk and immune cell infiltration in HCC using machine learning. Methods: MiRNA sequencing data was used to identify RMRMs. A risk score model of HCC was developed utilizing four RMRMs, including miR-551a, miR-4739, miR-326, and miR-210-3p. Results: Patients with high-risk scores exhibited poorer prognoses. Single-cell RNA sequencing (scRNA-seq) analysis revealed the high-risk group exhibited increased infiltration levels of several immune cell subtypes, including myeloid-derived suppressor cell (MDSC), macrophage, and T cells. The data integration of scRNA-seq and bulk RNA-seq showed the decreased TIDE score in the high-risk patients and the elevated levels of Macro-secreted phosphoprotein 1 (SPP1), MDSC-meiotic nuclear divisions 1 (MND1), γδ T cells, and Macro-complement C1q C chain (C1QC) predicted adverse prognosis. ScRNA-seq and spatial transcriptomics data integration unveiled the spatial distribution of RMRMs risk scores and their correlation with immune cell subtype localization. Risk model-based clustering of HCC samples revealed that cluster 2, characterized by a higher risk score, correlated with a poorer prognosis and reduced immune and stromal scores. In vitro, the overexpression of miR-4739 in Huh-7 cells significantly induced SPP1 Discussion: Our study reveals that the RMRM risk model could effectively predict the prognosis of HCC, and SPP1