Exome-Wide Analysis Identifies a Rare EXD3 Missense Variant Associated With Diabetic Kidney Disease

Abstract

Introduction: Diabetic kidney disease (DKD) is a major complication of diabetes, with genetic factors contributing to its progression. Although genome-wide association studies (GWAS) have identified common variants, the role of low-frequency and rare coding variants remains underexplored.

Methods: We performed exome-wide meta-analysis of up to 10,312 individuals with type 1 diabetes (T1D) genotyped using genome arrays with focused exome content. We included 10 DKD definitions based on albuminuria, estimated glomerular filtration rate (eGFR), or both. We analyzed nonsynonymous variants individually and used gene-level analyses for low-frequency (minor allele frequency [MAF] < 5%) and rare (< 1%) variants. Replication was performed in 10,066 participants with T1D and in UK Biobank participants with type 2 diabetes (T2D). Gene expression was assessed in cultured human podocytes.

Results: In addition to the known COL4A3 variant, a novel rare missense variant in EXD3 (p.Asp555Asn, rs200080727, minor allele frequency [MAF] = 0.4%) was associated with DKD (odds ratio [OR] = 8.7, P = 4.5 × 10^-9). The variant was predicted to be deleterious and EXD3 was downregulated in DKD in kidney expression datasets. EXD3 knock-down in a cultured human podocyte cell line reduced nephrin gene expression, suggesting a functional role in podocyte biology. Gene-level analyses identified 7 DKD-associated genes (P < 3.4 × 10^-6), including MUC5B, which harbored multiple low-frequency missense variants and with evidence of replication. Replication in UK Biobank supported the association of EXD3 rs200080727 with albuminuria (P = 0.014).

Conclusion: This study identified a rare EXD3 variant with a strong effect on DKD risk in T1D. Functional data support a role for EXD3 in podocyte integrity and DKD pathogenesis. However, further functional investigations are necessary to understand the underlying molecular mechanisms.

Keywords: EXD3; MUC5B; diabetic kidney disease; exome analysis; missense variants; type 1 diabetes.

Graphical abstract

Figure 1

Study overview. CKD, chronic kidney disease; DKD, diabetic kidney disease; DNCRI, Diabetic Nephropathy Collaborative Research Initiative.; GWAS, genome-wide association study; mRNA, messenger RNA; scRNAseq, single-cell RNA sequencing; siRNA, small interfering RNA; snRNAseq, single-nucleus RNA sequencing; T1D, type 1 diabetes; T2D, type 2 diabetes; THLBB, Finnish THL Biobank; MVP, Million Veteran Program; WES, whole-exome sequencing. Created in BioRender.

Figure 2

Single-variant association analysis discovered an EXD3 variant associated with the CKD + DKD phenotype. (a) Manhattan plot of the association. The horizontal dashed line indicates exome-wide significance, i.e., P-value < 5 × 10⁻⁷. (b) QQ-plot of association. CKD + DKD: cases with albuminuria (albumin excretion rate, albumin excretion rate ≥ 20 μg/min or equivalent) and estimated glomerular filtration rate < 45 ml/min per 1.73 m², controls with normal albumin excretion rate and eGFR ≥ 60 ml/min per 1.73 m². CKD, chronic kidney disease; DKD, diabetic kidney disease.

Figure 3

EXD3 in podocyte biology. (a) Cultured conditionally immortalized human podocytes (hPODs) were differentiated for 14 days at 37 °C and treated with TNF -α (10 ng/ml), PDGF (10 ng/ml) and nephrotoxin PAN (25 μg/ml) for 24 hours. For glucose stimulations, hPODs were stimulated with low glucose (5 mM), high glucose (25 mM) or mannitol (20 mM) for 24 hours. (b) hPODs were transfected with EXD3 siRNA pools or control siRNA (20 nmol/l; 24 hours) for 24 hours. Gene expression analysis of EXD3, NPHS1, PODXL, and CD2AP was performed using TaqMan reagents (Life Technologies) and normalized to GAPDH rRNA using the ΔΔCt method. Data are presented as n = 3 ± SEM. ∗t test P < 0.05. PAN, puromycin aminonucleoside; PDGF, platelet-derived growth factor; siRNA, small interfering RNA; TNF--α, tumor necrosis factor-α.

Figure 4

Gene aggregate analysis identified missense variants in the MUC5B gene associated with the “CKD” phenotype definition in T1D. (a) Manhattan plot of gene aggregate test results for the “CKD” phenotype definition, that is, eGFR < 60 ml/min per 1.73 m². Gene aggregate test results for nonsynonymous variants are drawn with purple color, significant association defined as P-value < 3.4 × 10⁻⁶ (upper dashed red line, corrected for 14,963 genes with nonsynonymous variants). Results for PTVs are drawn with orange color, significant associations defined as P-values < 2.5 × 10⁻⁵ (lower dashed red line, corrected for 2013 genes with protein-truncating variants). (b) Quantile-quantile plot (QQ-plot) for gene aggregate tests of nonsynonymous variants, stratified by aggregate test (burden, SKAT, or variable threshold (VT)) and variant frequency. (c) QQ-plot for gene aggregate tests of protein-truncating variants, stratified by aggregate test and variant frequency. Manhattan and QQ-plots for other diabetic kidney disease phenotype definitions are provided in Supplementary Figure S5. CKD, chronic kidney disease; PTV, protein-truncating variants.