Integrated miRNA–mRNA network analysis identifies a miR-15a-5p–centered angiogenic program and biomarker-linked regulatory modules in ovarian cancer

Abstract

Ovarian cancer remains the most lethal gynecologic malignancy, in part because it is often diagnosed at an advanced stage, molecular heterogeneity complicates therapeutic selection, and actionable regulatory dependencies remain incompletely defined. MicroRNAs (miRNAs) are well-established post-transcriptional regulators of tumor phenotypes, but most ovarian cancer studies have evaluated either mRNA or miRNA dysregulation in isolation and have therefore provided limited insight into how coordinated miRNA–mRNA programs are organized at the network level. Here we integrated bulk RNA-seq and matched NanoString miRNA profiling from 27 matched cell-line-derived samples, comprising 21 ovarian cancer cell-line samples and 6 control cell-line samples, to define differentially expressed genes and miRNAs, infer regulatory edges, identify community structure, and prioritize biologically coherent modules. RNA-seq identified 7,961 genes at false discovery rate (FDR) < 0.05, of which 2,932 remained strict differentially expressed genes after imposing |log2 fold-change| ≥ 1. miRNA profiling identified 40 significant miRNAs, with a strongly asymmetric landscape dominated by downregulation (35 down, 5 up). Integration of 24 miRNAs with 1,236 mRNAs yielded 4,614 expression-inferred miRNA–mRNA edges after filtering for negative matched-sample Spearman correlation (cor < -0.4; Benjamini–Hochberg adjusted cor_fdr < 0.05), followed by Leiden community detection that resolved five modules. Module numbers represent arbitrary Leiden community identifiers and do not imply biological rank or priority. Two modules showed Benjamini–Hochberg significant Gene Ontology enrichment: an extracellular-matrix/developmental module and an angiogenesis-associated module. The latter was centered on hsa-miR-15a-5p, which was connected by inferred regulatory edges to 109 of 119 genes in Module 5 and was linked to an angiogenic leading-edge set including ADGRG1, ANGPTL4, EREG, HTATIP2, ID1, PTK2, RAPGEF3, SYK, and TNFAIP2. In the discovery dataset, miR-15a-5p was downregulated, whereas the corresponding Module 5 angiogenesis-linked genes were strongly upregulated. In 420 matched TCGA-OV primary tumors, the Module 5 angiogenesis gene set showed a modest positive association with miR-15a-5p, suggesting that the bulk-tumor signal may reflect cellular composition, indirect regulation, or broader angiogenic tumor-state biology rather than simple one-to-one inverse miRNA-mediated repression. Biomarker triage further concentrated priority on epithelial/tumor-associated genes including EPCAM, ESRP1, ST14, TM4SF18, RAB25, and ADRA2C, with hsa-miR-140-5p, hsa-miR-15a-5p, and hsa-miR-381-3p emerging as recurrent regulators. Together, these results define an ovarian-cancer regulatory architecture in which a miR-15a-5p–centered angiogenic program coexists with biomarker-linked epithelial modules, offering a systems-level framework for mechanistic follow-up and translational prioritization.