Cheng Du,1,* Yuan Tian,2,* Yu Liu,2 Yuanyuan Shi,2 Xingying Chen,1 Hui Chai,2 Yuyan Zhang,2 Hui Shen,1 Lin Zhang2 

1Jiaxing Hospital of Traditional Chinese Medicine, Zhejiang Chinese Medical University, Jiaxing, Zhejiang Province, 314001, People’s Republic of China; 2College of Life Science/Institute of Molecular Medicine, Zhejiang Chinese Medical University, Hangzhou, Zhejiang Province, 310053, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Lin Zhang, College of Life Science/Institute of Molecular Medicine, Zhejiang Chinese Medical University, Hangzhou, 310053, People’s Republic of China, Email 20021047@zcmu.edu.cn Hui Shen, Jiaxing Hospital of Traditional Chinese Medicine, Zhejiang Chinese Medical University, Jiaxing, 314001, People’s Republic of China, Email shenhui19830227@163.com

Introduction: The pathogenesis of retinopathy of prematurity (ROP) is intricate and multifactorial. The current treatments for ROP include laser photocoagulation, cryotherapy, scleral buckling or vitrectomy and anti-VEGF drugs, yet which have numerous adverse effects. Consequently, the objective of this investigation was to mine new biomarkers of ROP and identify potential therapeutics that target these biomarkers.
Methods: GSE130400, the expression profile of ROP, was downloaded from the GEO database and annotated by dplyr package in R. The limma package was employed to identify differentially expressed genes (DEGs) between cases and controls. WGCNA was utilized to obtain oxygen-induced retinopathy (OIR)-related modules, and then key genes were obtained in the intersection of DEGs and the above modules. Biological functions and pathways of key genes were enriched through DAVID, Metascape and Gene set enrichment analysis. Hub genes were screened by three machine learning methods and validated by nomograms, ROC curves and qRT-PCR. CIBERSORT was used to estimate the abundances of immune cells and Pearson analysis revealed immune cells, checkpoints and chemokines associated with hub genes. Finally, we constructed a ceRNA network and a drug–gene interaction network.
Results: Utilizing GSE130400 as a basis, a total of 798 DEGs and 3711 genes from three OIR-related modules in WGCNA were identified, resulting in the identification of 681 key genes. Hub genes Plxnd1, Esm1 and Cd248 were explored using various machine learning methods and proved to be closely related with the occurrence of ROP through nomograms, ROC curves and qRT-PCR. Enrichment analysis revealed a significant enrichment of hub genes in processes related to endothelial cell proliferation, migration, and angiogenesis, among others. Immune analysis displayed that the development of ROP was associated with alterations in M0 macrophages and M2 macrophages, with hub genes participating in the regulation of immune cells M0 macrophages, Th2 cells, and NK resting cells, as well as chemokines lncRNA H19 (H19), CXCR5 (F5), CCL27 (F7) and CCL2 (C2). Furthermore, the drug–gene interaction network displayed 14 possible candidate drugs targeting 3 hub genes.
Conclusion: Plxnd1, Esm1 and Cd248 might play significant roles in the progression of ROP and could become possible diagnostic and therapeutic targets in ROP.

Keywords: retinopathy of prematurity, bioinformatic analysis, Plxnd1, Esm1, Cd248, immunology