Deep learning approach identified a gene signature predictive of the severity of renal damage caused by chronic cadmium accumulation

Abstract

Epidemiology studies have indicated that environmental cadmium exposure, even at low levels, will result in chronic cadmium accumulation in the kidney with profound adverse consequences and that the diabetic population is more susceptible. However, the underlying mechanisms are yet not fully understood. In the present study, we applied an animal model to study chronic cadmium exposure-induced renal injury and performed whole transcriptome profiling studies. Repetitive CdCl₂ exposure resulted in cadmium accumulation and remarkable renal injuries in the animals. The diabetic ob/ob mice manifested increased severity of renal injury compared with the wild type C57BL/6 J littermate controls. RNA-Seq data showed that cadmium treatment induced dramatic gene expression changes in a dose-dependent manner. Among the differentially expressed genes include the apoptosis hallmark genes which significantly demarcated the treatment effects. Pathway enrichment and network analyses revealed biological oxidation (mainly glucuronidation) as one of the major stress responses induced by cadmium treatment. We next implemented a deep learning algorithm in conjunction with cloud computing and discovered a gene signature that can predict the degree of renal injury induced by cadmium treatment. The present study provided, for the first time, a comprehensive mechanistic understanding of chronic cadmium-induced nephrotoxicity in normal and diabetic populations at the whole genome level.

Keywords: Cloud computing; Heavy metal; Oxidative stress; RNA-Seq.