Multiomics Analyses Identify AKR1A1 as a Biomarker for Diabetic Kidney Disease

Abstract

Diabetic kidney disease (DKD) is the leading cause of end-stage kidney disease. Because many genes associate with DKD, multiomics approaches were used to narrow the list of functional genes, gene products, and related pathways providing insights into the pathophysiological mechanisms of DKD. The Kidney Precision Medicine Project human kidney single-cell RNA-sequencing (scRNA-seq) data set and Mendeley Data on human kidney cortex biopsy proteomics were used. The R package Seurat was used to analyze scRNA-seq data and data from a subset of proximal tubule cells. PathfindR was applied for pathway analysis in cell type-specific differentially expressed genes and the R limma package was used to analyze differential protein expression in kidney cortex. A total of 790 differentially expressed genes were identified in proximal tubule cells, including 530 upregulated and 260 downregulated transcripts. Compared with differentially expressed proteins, 24 genes or proteins were in common. An integrated analysis combining protein quantitative trait loci, genome-wide association study hits (namely, estimated glomerular filtration rate), and a plasma metabolomics analysis was performed using baseline metabolites predictive of DKD progression in our longitudinal Diabetes Heart Study samples. The aldo-keto reductase family 1 member A1 gene (AKR1A1) was revealed as a potential molecular hub for DKD cellular dysfunction in several cross-linked pathways featured by deficiency of this enzyme.

Graphical abstract

Figure 1

Study design and workflow. KPMP, Kidney Precision Medicine Project.

Figure 2

Differential expression of AKR1A1 and its involved pathways in PT cells. A) AKR1A1 differential expression in kidney cells from the Kidney Precision Medicine Project (KPMP) project using scRNA-seq. AKR1A1 is enriched in PT cells, where the expression level was lower in patients with DKD than in living donor (LD) control participants. DTL, descending thin limb of loop of Henle; ATL, ascending thin limb of loop of Henle; TAL, thick ascending limb cell; EC, endothelial cell; PC, principal cell; IC, intercalate cell; CNT, connecting tubule cell; DCT, distal convoluted tubule; PEC, parietal epithelial cells; POD, podocyte. B) Validation of AKR1A1 protein differential expression in kidney PT cells by immunofluorescence on kidney cryosections. AKR1A1 protein is predominantly enriched in PT cells, with a lower level in glomerular cells and distal tubule cells. CD13 is a PT cell marker, the expression level of which is comparable between patients with DKD and negative control participants (CTL). The relative level of AKR1A1 protein in PTs was normalized by CD13 in corresponding tubules. The representative images reflected the median AKR1A1 protein levels in PT cells in patients with DKD (n = 4) versus control participants (n = 5). Each individual AKR1A1/CD13 protein level was averaged from four different microscopy areas. C) Violin plot with individual sample dots of normalized AKR1A1 protein levels in PT cells. AKR1A1 protein levels in PT cells from patients with DKD were significantly lower than those in control participants (P = 0.006, t test with unequal variance; adjusted P = 0.015 after adjusting for age and sex). D) Kidney PT cell–specific pathways involved in DKD. The R package pathfindR was used to identify KEGG pathways on the basis of differential gene expression in PT cells that were subsets from scRNA-seq gene expression profiles. AKR1A1-related pathways fell into two of the top three clusters (red, blue, and green) in terms of overall significance ranking. These pathways include chemical carcinogenesis-reactive oxygen species (hsa05208 in red cluster), ascorbate and aldarate metabolism (hsa00053 in green cluster), and pentose and glucuronate interconversion (hsa00040 in green cluster).

Figure 2

Differential expression of AKR1A1 and its involved pathways in PT cells. A) AKR1A1 differential expression in kidney cells from the Kidney Precision Medicine Project (KPMP) project using scRNA-seq. AKR1A1 is enriched in PT cells, where the expression level was lower in patients with DKD than in living donor (LD) control participants. DTL, descending thin limb of loop of Henle; ATL, ascending thin limb of loop of Henle; TAL, thick ascending limb cell; EC, endothelial cell; PC, principal cell; IC, intercalate cell; CNT, connecting tubule cell; DCT, distal convoluted tubule; PEC, parietal epithelial cells; POD, podocyte. B) Validation of AKR1A1 protein differential expression in kidney PT cells by immunofluorescence on kidney cryosections. AKR1A1 protein is predominantly enriched in PT cells, with a lower level in glomerular cells and distal tubule cells. CD13 is a PT cell marker, the expression level of which is comparable between patients with DKD and negative control participants (CTL). The relative level of AKR1A1 protein in PTs was normalized by CD13 in corresponding tubules. The representative images reflected the median AKR1A1 protein levels in PT cells in patients with DKD (n = 4) versus control participants (n = 5). Each individual AKR1A1/CD13 protein level was averaged from four different microscopy areas. C) Violin plot with individual sample dots of normalized AKR1A1 protein levels in PT cells. AKR1A1 protein levels in PT cells from patients with DKD were significantly lower than those in control participants (P = 0.006, t test with unequal variance; adjusted P = 0.015 after adjusting for age and sex). D) Kidney PT cell–specific pathways involved in DKD. The R package pathfindR was used to identify KEGG pathways on the basis of differential gene expression in PT cells that were subsets from scRNA-seq gene expression profiles. AKR1A1-related pathways fell into two of the top three clusters (red, blue, and green) in terms of overall significance ranking. These pathways include chemical carcinogenesis-reactive oxygen species (hsa05208 in red cluster), ascorbate and aldarate metabolism (hsa00053 in green cluster), and pentose and glucuronate interconversion (hsa00040 in green cluster).

Figure 3

Top serum metabolites that predicted development of DKD in the meta-analysis. A) The covariates in model 1 included duration between visits, age, sex, BMI, diabetes duration, African ancestry, and HbA_1c. The covariates in model 2 include those in model 1 plus systolic blood pressure and ACE or angiotensin receptor blocker use. Bubble sizes represent the value of −log10(FDR). * Compounds that have not been officially confirmed based on a standard, but identified by virtue of their recurrent chromatographic and spectral nature. FDR_p: false discovery rate–adjusted P value. B) Potential role of AKR1A1 in detoxification and glucuronate metabolism. Insufficiency of AKR1A1 may lead to accumulation of 3,4-dihyroxybutyrate and glucuronate, the main metabolites that predict DKD in patients with diabetes in the AA-DHS study from the metabolomics analysis.

Figure 4

Summary diagram.