Objective To screen hub genes involved in the mitochondria-programmed cell death (PCD) cross-regulation in pulmonary arterial hypertension (PAH) based on bioinformatics and machine-learning strategies, evaluate their diagnostic value, clarify the relevant regulatory mechanisms, and preliminarily identify potential targeted therapeutic drugs. Methods PAH-related data were retrieved from the GEO database, with GSE33463 as the training set while GSE131793 and GSE38267 as independent validation sets. Differential expression analysis was performed on the training set to screen differentially expressed genes (DEGs). The DEGs were intersected with the mitochondria-related gene set and the PCD-related gene set respectively to obtain candidate genes with both differential-expression characteristics and dual biological properties of mitochondria and PCD. Hub genes were identified through protein-protein interaction network analysis, combined with two machine learning algorithms: support vector machine-recursive feature elimination and Boruta. Their expression differences and diagnostic performance were validated in the two independent validation sets. Gene set enrichment analysis (GSEA), construction of miRNA-mRNA regulatory network, and molecular docking were further adopted to explore the regulatory mechanisms of hub genes and potential targeted therapeutic drugs. Results Two hub genes, RPS3 and BCL2L1, were finally identified through multi-layer screening. In the training set and 2 validation sets, RPS3 showed down-regulated expression while BCL2L1 showed up-regulated expression in the PAH group (all P<0.05). The areas under the receiver operating characteristic curve of RPS3 and BCL2L1 were both greater than 0.70, indicating favorable diagnostic efficiency for PAH. GSEA results revealed that RPS3 and BCL2L1 were co-enriched in oxidative phosphorylation, MYC-targets V1, and mammalian target of rapamycin (mTOR) C1 signaling pathways. MiRNA prediction suggested that RPS3 might be regulated by hsa-miR-603, hsa-miR-515-3p, and hsa-miR-33b-3p, and BCL2L1 might be regulated by 13 miRNAs including hsa-miR-4325 and hsa-miR-342-3p. Drug prediction and molecular docking demonstrated that BCL2L1 had strong binding affinity with 4 candidate targeted drugs, apogossypol, gossypol, sangivamycin, and obatoclax (all binding energy<-5.0 kcal/mol), among which obatoclax exhibited the optimal binding affinity (with the lowest binding energy of -9.1 kcal/mol). Conclusion RPS3 and BCL2L1 are hub genes related to mitochondria and PCD in PAH. They may participate in the pathogenesis of PAH through pathways such as mTORC1 and oxidative phosphorylation, and possess favorable diagnostic value and potential for targeted intervention, which provides candidate targets and theoretical basis for mechanism research and precise diagnosis and treatment of PAH.

无