CpG Methylation-Driven Pleiotropic Spliced Isoforms of PRSS3 Facilitate Gastric Intratumor Heterogeneity and Metastasis.

Investigating the phenotypic contribution of gene-spliced isoforms to tumor cellular heterogeneity can facilitate the development of innovative strategies for precision medicine. Serine protease 3 (PRSS3), a trypsin-like protease with four spliced variants (PRSS3-SVs: PRSS3-V1 to V4), plays diverse biological roles in cancer progression. Herein, we systematically analyzed the expression patterns and functional implications of PRSS3-SVs in gastric cancer (GC) using an integrative approach that combined bioinformatic analyses, CpG site-specific methylation detection, splice-specific qPCR, isoform-based methodologies, and multiple functional assays. Our findings revealed that differentially expressed PRSS3 isoforms, predominantly PRSS3-V1 and PRSS3-V2, are regulated by intragenic methylation and exert pleiotropic roles in GC. Overexpression of PRSS3 transcripts suppressed GC cell proliferation via the NF-κB signaling pathway, while exerting distinct effects on matrix metalloproteinase-associated cell migration and invasion. Clinically, patients with low PRSS3-V1 or high PRSS3-V2 expression exhibited poorer survival outcomes, and the expression difference between these two transcripts was identified as an independent prognostic indicator for GC patients. Epigenetically, differential methylation patterns within PRSS3 enabled stratification of GC patients into subgroups characterized by either high-methylation with low-expressed PRSS3-SVs or low-methylation with high-expressed PRSS3-SVs. The UHRF1/DNMT1 complex was found to mediate CpG site methylation and regulate PRSS3 transcripts, particularly silencing PRSS3-V1 through intragenic CpG methylation. This methylation pattern was associated with reduced survival rates and further validated its correlation with tumor metastasis in an independent cohort (n = 243). Our study elucidates that methylation-regulated alternative splicing contributes to phenotypic heterogeneity in GC, highlighting its potential advantage over differentially expressed genes in improving stratification strategies for precision oncology.