Background
Diabetic nephropathy (DN) is a primary driver of end-stage renal disease. Given the heterogeneity of renal lesions and the complex mechanisms of DN, the present-day diagnostic approach remains highly controversial. We aimed to design a diagnostic model by bioinformatics
Conclusion
We confirmed that differentially expressed hub genes may be novel diagnostic biomarkers in DN.
Methods
In this study, transcriptome sequencing was performed on 6 clinical samples (3 from DN patients and 3 from healthy volunteers) from the Second Affiliated Hospital of Kunming Medical University. Construction of a competing endogenous RNA (ceRNA) network based on differentially expressed (DE)-mRNAs and -long noncoding RNAs (lncRNAs). Subsequently, the CytoHubba plugin was used to identify hub genes from DE-mRNAs in the ceRNA network and to perform functional enrichment analysis on them. The least absolute shrinkage and selection operator (LASSO) regression analysis was responsible for screening the diagnostic biomarkers from hub genes and assessing their diagnostic power using ROC curves. The pathways involved in hub genes were revealed by single-gene Gene Set Enrichment Analysis (GSEA). Moreover, we verified the expression levels of diagnostic biomarkers by quantitative real-time polymerase chain reaction (qRT-PCR) and Western blot.
Results
A total of 10 hub genes were screened from the ceRNA network, which appeared to be associated with the viral infection, kidney development, and regulation of immune and inflammatory responses. Subsequently, LASSO regression analysis established a diagnostic model consisting of DDX58, SAMD9L, and TLR6 with a robust diagnostic potency (AUC = 1). Similarly, single-gene GSEA showed a strong association of these diagnostic biomarkers with the viral infection. Furthermore, PCR and Western blot demonstrated showed that DDX58, SAMD9L, and TLR6 were upregulated in DN patients at both transcriptome and protein levels compared to healthy controls.