Abstract
BACKGROUND: Advances in diagnosis and treatment of premature ovarian insufficiency (POI) are urgently needed. Mitochondrial dysfunction and PANoptosis are associated with ovarian damage. In this study, we aimed to identify mitochondrial and PANoptosis-related biomarkers for POI. METHODS: Integrative bioinformatics analysis was performed based on RNA sequencing datasets from the Gene Expression Omnibus (GEO) database, combined with the MitoCarta3.0 database and PANoptosis-related genes. The feature differentially expressed genes (DEGs) were screened using machine learning methods. Based on the expression levels of mitochondrial characteristic genes (ACSM3, ALDH1L1, MAOB, OSBPL1A) and pan-apoptotic genes (NOS2, UCHL1) screened out by the above machine learning, the mitochondrial and PANoptosis phenotypic scores were calculated through the GSVA algorithm. Differential expression of biomarkers was validated in clinical samples, and the role of nitric oxide synthase 2 (NOS2) was explored in POI mouse models. RESULTS: Four mitochondrial-related genes (ACSM3, ALDH1L1, MAOB, and OSBPL1A) and two PANoptosis-related genes (NOS2 and UCHL1) were screened. The mitochondrial phenotype scores of the samples were positively correlated with PANoptosis scores. All biomarkers were downregulated in human granulosa cells. Notably, NOS2, identified as the most significant biomarker, was downregulated in the POI model. Moreover, the overexpression of NOS2 markedly suppressed cell death and reactive oxygen species (ROS) production in the ovarian tissues of POI mice. CONCLUSIONS: The identified biomarkers may play key roles in the pathology of POI and serve as candidate biomarkers for its diagnosis and treatment. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13048-025-01839-4.