A common glomerular transcriptomic signature distinguishes diabetic kidney disease from other kidney diseases in humans and mice

Background: Current uncertainties about the similarity between human diseases and their experimental models are hampering the development of new therapies. This is especially the case for diabetic kidney disease (DKD), the most common cause of end-stage kidney disease. To better understand the nature of the commonality between humans and their mouse models, we posed the question: in diabetic kidney disease are transcriptional profiles primarily disease-specific or species-specific? Methods: We performed a meta-comparison of the glomerular transcriptomic characteristics of 133 human and 66 mouse samples including five human kidney diseases and five mouse models, validating expression patterns of a central node by immunohistochemistry. Findings: Principal component analysis controlled for mouse background, revealed that gene expression changes in glomeruli from humans with DKD are more similar to those of diabetic mice than they are to other human glomerular diseases. This similarity enabled the construction of a discriminatory classifier that distinguishes diabetic glomeruli from other glomerular phenotypes regardless of their species of origin. To identify where the commonality between mice and humans with diabetes lies, networks of maximally perturbed protein interactions were examined, identifying a central role for the epidermal growth factor receptor (EGFR). By immunohistochemical staining, we found EGFR to be approximately doubled in its glomerular expression in both humans and mice. Interpretation: These findings indicate that diabetic mouse models do mimic some of the features of human kidney disease, at least with respect to their glomerular transcriptomic signatures, and they identify EGFR as being a central player in this inter-species overlap.