AUTHOR=Zhang Peng , Chen Honglin , Xie Bin , Zhao Wenhua , Shang Qi , He Jiahui , Shen Gengyang , Yu Xiang , Zhang Zhida , Zhu Guangye , Chen Guifeng , Yu Fuyong , Liang De , Tang Jingjing , Cui Jianchao , Liu Zhixiang , Ren Hui , Jiang Xiaobing TITLE=Bioinformatics identification and experimental validation of m6A-related diagnostic biomarkers in the subtype classification of blood monocytes from postmenopausal osteoporosis patients JOURNAL=Frontiers in Endocrinology VOLUME=14 YEAR=2023 URL=https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2023.990078 DOI=10.3389/fendo.2023.990078 ISSN=1664-2392 ABSTRACT=Background

Postmenopausal osteoporosis (PMOP) is a common bone disorder. Existing study has confirmed the role of exosome in regulating RNA N6-methyladenosine (m6A) methylation as therapies in osteoporosis. However, it still stays unclear on the roles of m6A modulators derived from serum exosome in PMOP. A comprehensive evaluation on the roles of m6A modulators in the diagnostic biomarkers and subtype identification of PMOP on the basis of GSE56815 and GSE2208 datasets was carried out to investigate the molecular mechanisms of m6A modulators in PMOP.

Methods

We carried out a series of bioinformatics analyses including difference analysis to identify significant m6A modulators, m6A model construction of random forest, support vector machine and nomogram, m6A subtype consensus clustering, GO and KEGG enrichment analysis of differentially expressed genes (DEGs) between different m6A patterns, principal component analysis, and single sample gene set enrichment analysis (ssGSEA) for evaluation of immune cell infiltration, experimental validation of significant m6A modulators by real-time quantitative polymerase chain reaction (RT-qPCR), etc.

Results

In the current study, we authenticated 7 significant m6A modulators via difference analysis between normal and PMOP patients from GSE56815 and GSE2208 datasets. In order to predict the risk of PMOP, we adopted random forest model to identify 7 diagnostic m6A modulators, including FTO, FMR1, YTHDC2, HNRNPC, RBM15, RBM15B and WTAP. Then we selected the 7 diagnostic m6A modulators to construct a nomogram model, which could provide benefit with patients according to our subsequent decision curve analysis. We classified PMOP patients into 2 m6A subtypes (clusterA and clusterB) on the basis of the significant m6A modulators via a consensus clustering approach. In addition, principal component analysis was utilized to evaluate the m6A score of each sample for quantification of the m6A subgroups. The m6A scores of patients in clusterB were higher than those of patients in clusterA. Moreover, we observed that the patients in clusterA had close correlation with immature B cell and gamma delta T cell immunity while clusterB was linked to monocyte, neutrophil, CD56dim natural killer cell, and regulatory T cell immunity, which has close connection with osteoclast differentiation. Notably, m6A modulators detected by RT-qPCR showed generally consistent expression levels with the bioinformatics results.

Conclusion

In general, m6A modulators exert integral function in the pathological process of PMOP. Our study of m6A patterns may provide diagnostic biomarkers and immunotherapeutic strategies for future PMOP treatment.