A KDM6A–KLF10 reinforcing feedback mechanism aggravates diabetic podocyte dysfunction

Abstract Diabetic nephropathy is the leading cause of end‐stage renal disease. Although dysfunction of podocytes, also termed glomerular visceral epithelial cells, is critically associated with diabetic nephropathy, the mechanism underlying podocyte dysfunction still remains obscure. Here, we identify that KDM6A, a histone lysine demethylase, reinforces diabetic podocyte dysfunction by creating a positive feedback loop through up‐regulation of its downstream target KLF10. Overexpression of KLF10 in podocytes not only represses multiple podocyte‐specific markers including nephrin, but also conversely increases KDM6A expression. We further show that KLF10 inhibits nephrin expression by directly binding to the gene promoter together with the recruitment of methyltransferase Dnmt1. Importantly, inactivation or knockout of either KDM6A or KLF10 in mice significantly suppresses diabetes‐induced proteinuria and kidney injury. Consistent with the notion, we also show that levels of both KDM6A and KLF10 proteins or mRNAs are substantially elevated in kidney tissues or in urinary exosomes of human diabetic nephropathy patients as compared with control subjects. Our findings therefore suggest that targeting the KDM6A–KLF10 feedback loop may be beneficial to attenuate diabetes‐induced kidney injury.

A Schematic diagram of mouse nephrin gene promoter. The transcriptional start site of mouse nephrin gene promoter (GenBank: AF190638.1) is located 257-bp upstream of the ATG codon. Two evolutionarily conserved promoter regions essential for podocyte-specific expression are indicated by yellow (from À188 to À270; 83-bp region) and red (from À1,870 to À2,106; 237-bp region) boxes, respectively. The 83-bp nephrin promoter element contains a WT-1 binding site, whereas the 237-bp promoter element contains multiple potential Sp1 binding sites. B DNA sequence of the promoter region encompassing the 237-bp promoter region (from À1,870 to À2,106; red color) and the localization of five non-overlapping probes (P-A, P-B, P-C, P-D, and P-E) used in EMSAs. C EMSAs of KLF10 binding to nephrin promoter elements. EMSA experiments were performed by using nephrin promoter elements (including P-A, P-B, P-C, P-D, and P-E) and protein lysates of 293T cells that were transfected with an empty vector or KLF10-expressing plasmid. Arrows indicated the potential KLF10/DNA complexes in EMSA experiments. D Supershift analysis of the KLF10/DNA complex in EMSA. Different antibodies against KLF10 were used to supershift the formed complex in EMSA experiments using the P-E element as a probe. Anti-KLF10 antibody #1 was a kind gift from Dr. Vincent H.S. Chang, whereas anti-KLF10 antibody #2 (ab73537; Abcam) was obtained commercially.
▸ Figure EV3. In vivo analysis of wild-type and KLF10-knockout (KLF10-KO) mice with or without STZ treatment.
A Urinary albumin excretion in wild-type and KLF10-KO mice with or without STZ treatment. Urinary albumin levels were measured with a turbidimetric immunoassay (Autokit Micro Albumin, Wako, Osaka, Japan) at 8 weeks after diabetic induction. *P < 0.05 versus untreated wild-type controls, # P < 0.05 versus STZ-treated wildtype mice (parametric ANOVA and a Bonferroni post hoc test; n = 8). B Systolic blood pressure of wild-type and KLF10-KO mice with or without STZ treatment. Systolic blood pressure was measured by tail-cuff plethysmography (BP-2000 Series II Blood Pressure Analysis System, Visitech Systems, Apex, NC, USA) at 8 weeks after diabetic induction. No statistical differences in mean systolic blood pressures were found between groups (parametric ANOVA and a Bonferroni post hoc test; n = 8). C Levels of urinary cystatin C of wild-type and KLF10-KO mice with or without STZ treatment. Urine cystatin C was measured using an ELISA kit (MSCTC0, R&D Systems) at 8 weeks after diabetic induction. No statistical differences in levels of urinary cystatin C were found between groups (parametric ANOVA and a Bonferroni post hoc test; n = 8). D Detection of glomerular cell apoptosis in wild-type and KLF10-KO mice with or without STZ treatment. TUNEL assay for apoptosis was carried out using an assay kit according to the manufacturer's instruction (#TAAP01D, BioTnA Biotech., Kaohsiung Taiwan). Scale bars, 20 lm. *P < 0.05 versus untreated wild-type controls, # P < 0.05 versus STZ-treated wild-type mice (parametric ANOVA and a Bonferroni post hoc test; n = 3). E Representative electron micrographs of GBM (glomerular basement membrane) thickening and foot process effacement in wild-type and KLF10-KO mice with or without STZ treatment. Kidney specimens for electron microscopy were prepared as described previously (White & Bilous, 2000;Advani et al, 2007), and electron micrographs were taken with a FEI Tecnai G2 F20 S-TWIN Transmission Electro Microscope (TEM). TEM images were processed and analyzed with DigitalMicrograph (Gatan Inc.). Scale bars, 0.5 lm. *P < 0.05 versus untreated wild-type controls, # P < 0.05 versus STZ-treated wild-type mice (parametric ANOVA and a Bonferroni post hoc test; n = 3). F Representative photographs of periodic acid-Schiff (PAS) staining in kidney tissues of wild-type and KLF10-KO mice with or without STZ treatment. Scale bars, 20 lm.
IOD: integrated optical density. *P < 0.05 versus untreated wild-type controls, # P < 0.05 versus STZ-treated wild-type mice (parametric ANOVA and a Bonferroni post hoc test; n = 3). Notably, diabetic KLF10-KO mice substantially displayed reduced PAS-staining intensities in glomeruli and in renal tubules as compared to diabetic wild-type mice.
Data information: Data are expressed as mean AE SEM. See the exact P-values for comparison tests in Appendix Table S9. A Effects of KDM6A overexpression in combination with KLF10 knockdown on the expression of podocyte-specific markers. Relative levels of KDM6A, KLF10, nephrin, and WT-1 expressed in podocytes that were transfected with the indicated plasmids, and siRNAs were determined by Western blot analysis. *P < 0.05 versus vectortransfected podocytes, # P < 0.05 versus podocytes transfected with KDM6A and control siRNA (parametric ANOVA and a Bonferroni post hoc test; n = 3). B Effects of KLF10 overexpression in combination with KDM6A knockdown on the expression of podocyte-specific markers. Relative levels of KDM6A, KLF10, nephrin, and WT-1 expressed in podocytes that were transfected with the indicated plasmids and siRNAs were determined by Western blot analysis. *P < 0.05 versus vectortransfected podocytes, # P < 0.05 versus podocytes transfected with KLF10 and control siRNA (parametric ANOVA and a Bonferroni post hoc test; n = 3).
Data information: Data are expressed as mean AE SEM. See the exact P-values for comparison tests in Appendix Table S10.
Source data are available online for this figure.  Figure EV5. Addition of a TGF-b1-neutralizing antibody has no significant effects on the positive inter-regulation between KDM6A and KLF10 in podocytes.
A Effects of TGF-b1 neutralization on high glucose-mediated gene expression. Podocytes cultured in normal or high glucose were treated with different doses (3 and 10 lg/ml) of TGF-b1-neutralizing antibody (MAB240, R&D Systems) for 48 h. Relative protein levels of KDM6A, KLF10, nephrin, and WT-1 in these treated podocytes were analyzed by Western blot analysis. *P < 0.05 versus normal controls, # P < 0.05 versus podocytes in high glucose (parametric ANOVA and a Bonferroni post hoc test; n = 3). B Effects of TGF-b1 neutralization on KDM6A-mediated gene expression. Podocytes transfected with an empty vector or KDM6A-expressing plasmid were treated with different doses (3 and 10 lg/ml) of TGF-b1 neutralizing antibody (MAB240, R&D Systems) for 48 h. Relative expression levels of KDM6A, KLF10, nephrin, and WT-1 in these treated podocytes were analyzed by Western blot analysis. *P < 0.05 versus podocytes transfected with an empty vector (parametric ANOVA and a Bonferroni post hoc test; n = 3). C Effects of TGF-b1 neutralization on KLF10-mediated gene expression. Podocytes transfected with an empty vector or KLF10-expressing plasmid were treated with different concentrations (3 and 10 lg/ml) of TGF-b1 neutralizing antibody (MAB240, R&D Systems) for 48 h. Relative expression levels of KDM6A, KLF10, nephrin, and WT-1 in these treated podocytes were analyzed by Western blot analysis. *P < 0.05 versus podocytes transfected with an empty vector (parametric ANOVA and a Bonferroni post hoc test; n = 3). D A model for the relationship between TGF-b1 signaling and the KDM6A-KLF10 positive feedback loop in podocytes under diabetic conditions. Under hyperglycemic conditions, increased TGF-b1 production may directly or indirectly trigger activation of the KDM6A-KLF10 positive feedback loop. However, TGF-b1 is not essential for the positive inter-regulation between KDM6A or KLF10.
Data information: Data are expressed as mean AE SEM. See the exact P-values for comparison tests in Appendix Table S11.
Source data are available online for this figure. ◀