|
|
Identification of diagnostic biomarkers related to osteoporosis exosomes and preliminary drug screening
Liang Zhou, Pan Chengzhen, Chen Feng, Zhang Chi, Yang Bo, Wei Zongbo, Meng Jianhua, Zhou Zhu
2026, 30 (13):
3458-3473.
doi: 10.12307/2026.324
BACKGROUND: In recent years, exosomes, as key mediators of intercellular communication, have played important roles in the occurrence, progression, and treatment of osteoporosis. The active molecules they carry, such as miRNAs and proteins, can regulate the functions of osteoblasts and osteoclasts and affect bone metabolic balance. However, the specific mechanisms still require further research.
OBJECTIVE: To identify and validate core exosomal genes in osteoporosis, explore their immune regulatory mechanisms, and predict potential targeted drugs using 4D-DIA proteomics, multiple machine learning algorithms, and Mendelian randomization analysis, providing new insights for mechanistic research and precision treatment of osteoporosis.
METHODS: Twelve Sprague-Dawley rats were divided into two groups: sham surgery group and osteoporosis model group, with 6 rats in each group. After model establishment by ovariectomy, femoral tissue samples of rats were collected for 4D-DIA proteomic analysis to identify differentially expressed genes, along with Weighted Gene Co-expression Network Analysis. Expression profiles GSE56815 and GSE7158 were collected from the GEO database as validation datasets. Exosome-related genes were downloaded from the GeneCards database. The intersections of these genes with Weighted Gene Co-expression Network Analysis module genes from proteomics and differentially expressed genes from validation datasets were used to obtain osteoporosis-exosome related genes. Functional enrichment analysis was performed. Subsequently, three machine learning algorithms - Random Forest, LASSO, and support vector machine - were used to screen feature genes separately, and their intersection was taken to obtain osteoporosis-exosome core genes. A prediction model was further established and the receiver operating characteristic curve was verified. CIBERSORT was used for immune infiltration analysis to examine the differential expression of immune cell subpopulations in osteoporosis. Single-sample gene set enrichment analysis was used to analyze the correlation between osteoporosis-exosome core genes and immune cell subpopulations, and simultaneously analyze the relevant biological pathways of the core genes. StarBase database was applied to predict the RNA Binding Protein regulatory network of osteoporosis-exosome core genes. Finally, two-sample Mendelian randomization was employed to verify the causal relationship between exosome core genes and osteoporosis. Drug enrichment analysis was conducted through DSigDB database. Molecular docking visualization was performed using the CB-DOCK2 website.
RESULTS AND CONCLUSION: (1) 4D-DIA proteomics identified 1 322 differential proteins related to osteoporosis. Through Weighted Gene Co-expression Network Analysis, two characteristic modules containing 402 genes were identified. The Gene Cards database curated 878 exosome-related genes. Differential analysis of the GEO validation dataset identified 4 447 differential proteins. The intersection of these three gene sets yielded 31 genes associated with osteoporosis and exosomes. (2) Functional enrichment analysis of the related genes indicated that they were primarily associated with neutrophil extracellular trap formation, the Rap1 signaling pathway, and focal adhesions. (3) Three machine learning algorithms identified four core genes related to osteoporosis and exosomes, among which two genes (ITGB3 and SERPINA1) exhibited consistent differential expression in both animal models and the GEO validation dataset. The receiver operating characteristic curve showed that ITGB3 and SERPINA1 exhibited high area under the curve values in both the animal models and the GEO validation dataset. Models constructed with either individual genes or a combination of the two genes achieved area under the curve values greater than 0.9. (4) Immune infiltration gene set enrichment analysis revealed that the high expression of ITGB3 and SERPINA1 was positively correlated with M1 macrophages. High expression of ITGB3 and SERPINA1 was also associated with the NOD-like receptor signaling pathway. (5) The RNA Binding Protein regulatory network showed that ITGB3 and SERPINA1 jointly regulated 10 RNA-binding proteins, including HNRNPC, G3BP1, EIF3D, CTCF, U2AF2, and MDTH. (6) Two-sample Mendelian randomization analysis indicated that SERPINA1 exerted an inhibitory effect on osteoporosis, making it a protective factor against osteoporosis. (7) Drug enrichment analysis of SERPINA1 identified 36 drugs with a P-value < 0.05. Molecular docking revealed that 9 of these drugs had binding energies lower than -29.4 kJ/mol, with β-carotene exhibiting the strongest binding energy to SERPINA1 (-35.28 kJ/mol). The above findings confirm that ITGB3 and SERPINA1 are core genes associated with osteoporosis-exosomes. Both genes play a critical role in disease progression by participating in specific immune processes and regulating the NOD-like receptor signaling pathway. These genes exhibit precise predictive potential for the diagnosis of osteoporosis.
Figures and Tables |
References |
Related Articles |
Metrics
|