Identifying a Comprehensive ceRNA Network to Reveal Novel Targets for the Pathogenesis of Parkinson's Disease

Parkinson's disease (PD) is the second commonest progressive neurodegenerative disease worldwide. Increasing evidence reveals that non-coding RNAs play roles in the pathophysiological process of PD. The notion called competing endogenous RNAs (ceRNAs) network is used to describe the roles of non-coding RNAs. According to this theory, long non-coding RNAs (lncRNAs) act as microRNAs (miRNAs) sponges by miRNA response elements or miRNA binding sites to control the availability of endogenous miRNA for binding to their target mRNAs. This study aimed to construct a ceRNA network in PD, which might have the potential to clarify the pathogenesis of PD. We investigated differential expression (DE) lncRNAs and mRNAs in substantia nigra array data GSE7621 between PD patients and healthy controls from the Gene Expression Omnibus database. And we used starBase 2.0 and miRWalk 2.0 databases to predict miRNAs that have interactions with DElncRNAs and DEmRNAs. Based on DElncRNAs, DEmRNAs and predicted miRNAs, two ceRNA networks were constructed. The first one was based on lncRNA-miRNA interactions and miRNA-mRNA interactions that shared the same miRNAs that we predicted, on which function annotation and PPI analysis were performed to identify hub genes. Hereby the second ceRNA network was generated to explore the core section in the first ceRNA network and was validated in external datasets. As a result, we identified 31 DE lncRNAs and 1,828 DEmRNAs, and finally constructed the first ceRNA network associated with PD, including 9 lncRNAs, 18 miRNAs, and 185 mRNAs. mRNAs in the first ceRNA network focused on autophagy, DNA repair and vesicle transport, which were critical pathological processes in PD. Nineteen hub genes in the first ceRNA network identified through PPI analysis, the second ceRNA network was constructed to annotate the core part of the first one. Moreover, the core subnetwork was validated in external datasets, of which several nodes including FBXL7, PTBP2, and lncRNA NEAT1 were verified. In conclusion, a ceRNA network was constructed based on the differential expression profiles of whole substantia nigra tissues of normal and PD patients, and the network was subsequently identified which revealed its association with autophagy, DNA repair and vesicle transport. The core subnetwork of the ceRNA network was identified and validated in external data. Our findings offered novel insights into the roles of ceRNAs in the pathogenesis of PD and provided promising diagnostic biomarkers.