Abstract
INTRODUCTION: Renal tubular injury contributes to the decline in kidney function in patients with diabetes. Cell type-specific DNA methylation patterns have been used to calculate proportions of particular cell types. In this study, we developed a method to detect renal tubular injury in patients with diabetes by detecting exfoliated tubular cells shed into the urine based on tubular cell-specific DNA methylation patterns. RESEARCH DESIGN AND METHODS: We identified DNA methylation patterns specific for human renal proximal tubular cells through compartment-specific methylome analysis. We next determined the methylation levels of proximal tubule-specific loci in urine sediment of patients with diabetes and analyzed correlation with clinical variables. RESULTS: We identified genomic loci in SMTNL2 and G6PC to be selectively unmethylated in human proximal tubular cells. The methylation levels of SMTNL2 and G6PC in urine sediment, deemed to reflect the proportion of exfoliated proximal tubular cells due to injury, correlated well with each other. Methylation levels of SMTNL2 in urine sediment significantly correlated with the annual decline in estimated glomerular filtration rate. Moreover, addition of urinary SMTNL2 methylation to a model containing known risk factors significantly improved discrimination of patients with diabetes with faster estimated glomerular filtration rate decline. CONCLUSIONS: This study demonstrates that patients with diabetes with continual loss in kidney function may be stratified by a specific DNA methylation signature through epigenetic urinalysis and provides further evidence at the level of exfoliated cells in the urine that injury of proximal tubular cells may contribute to pathogenesis of diabetic kidney disease.