SGLT2 inhibition protects kidney function by SAM-dependent epigenetic repression of inflammatory genes under metabolic stress

Abstract

Clinically, blockade of renal glucose resorption by sodium-glucose cotransporter 2 (SGLT2) inhibitors slows progression of kidney disease, yet the underlying mechanisms are not fully understood. We hypothesized that altered renal metabolites underlie observed kidney protection when SGLT2 function is lost. S-adenosylmethionine (SAM) levels were increased in kidneys from mice lacking SGLT2 function on a diabetogenic high-fat diet (SPHFD) compared with WT mice fed HFD. Elevated SAM in SPHFD was associated with improved kidney function and decreased expression of NF-κB pathway-related genes. Injured proximal tubular cells that emerged under HFD conditions in WT mice and humans consistently showed reduction in expression of the SAM synthetase Mat2a/MAT2A, while MAT2A inhibition, which reduces SAM production, abrogated kidney protection in SPHFD mice. Histone H3 lysine 27 (H3K27) repressive trimethylation of NF-κB-related genes was increased in SPHFD, consistent with SAM's role as a methyl donor. Our data support a model whereby SGLT2 loss enhances SAM levels within the kidney, leading to epigenetic repression of inflammatory genes and kidney protection under metabolic stress.

Keywords: Diabetes; Epigenetics; Metabolism; NF-kappaB; Nephrology.

Figure 1. LOF of SGLT2 improved glucose intolerance and kidney injury.

(A) Schematic of experimental protocol. (B) Chronological BW change. (C) Chronological changes in food intake. (D–F) HFD elevated glucose intolerance (D and E) and insulin resistance (F). Glucose intolerance was blunted in SP mice (D and E). GTT, glucose tolerance test; ITT, insulin tolerance test. (G) Representative images of TUNEL, PAS, and Sirius Red staining at 8 or 18 week feeding time points across groups. The arrows indicate TUNEL-positive cells. Right panel, quantification. (H) Representative images of fibronectin immunofluorescence in the kidney cortex at the 18 week feeding time point. Right panel, quantification. (I) uACR at 8 and 17 week feeding time points. (J) Serum creatinine level. (K) Protein level of KIM1 in renal cortex of mice at the 18 week feeding time point. Right panel, quantification. (L) RT-qPCR analysis of Havcr1. Scale bars: 100 μm. Sample numbers: WT-ND 8wks, n = 10; SP-ND 8wks, n = 9; WT-HFD 8wks, n = 14; SP-HFD 8wks, n = 13; WT-ND 18wks, n = 7; WT-HFD 18wks, n = 12; SP-ND 18wks, n = 4; SP-HFD 18wks, n = 8. Data were analyzed by 1-way (G–L) or 2-way (B–F) ANOVA; ***P < 0.001, **P < 0.01, and *P < 0.05 by Tukey’s test. Values are presented as mean ± SEM.

Figure 2. Population of injured PTCs is enriched in WT^HFD mice.

(A) Uniform Manifold Approximation and Projection (UMAP) demonstrating 22 distinct cell types in kidney. (B) Dot plot of canonical cell marker genes (size of the dot indicates the percent positive cells, and color indicates relative expression). (C) UMAP colored by experimental groups. Right panel shows PTC clusters by genotype. (D) Stacked bar plot displaying distribution of relative cell percentage of total cells. PT, proximal tubule; TAL, thick ascending limb of the loop of Henle; DCT, distal convoluted tubule; IMCD, inner medullary collecting duct; CD, collecting duct; IC, intercalated cells; GEC, glomerular endothelial cell; EC, endothelial cell; MC, mesangial cell; Pod, podocyte; DC, dendritic cell.

Figure 3. Inflammatory gene pathways are enriched in the PT-HFD cluster.

(A) Bubble plot showing pathway enrichment of upregulated pathways in PT-HFD (size of the dot indicates the percent positive cells, and color indicates relative expression). (B) Western blot analysis of phosphorylated/total p65 in tissue isolated from the renal cortex across groups at the 18 week feeding time point. (C) RT-qPCR analysis of proinflammatory cytokines and chemokines in tissue isolated from the renal cortex across groups (n = 6–8 per group). (D) Representative images of immunohistochemistry of p65 in the kidney and quantification (right panel). Scale bar: 100 μm. Data were analyzed by 1-way ANOVA; ***P < 0.001, **P < 0.01, and *P < 0.05 by Tukey’s test. Values are presented as mean ± SEM.

Figure 4. Metabolic profiles in the kidney.

(A) Metabolic enrichment pathway analysis in tissue isolated from the renal cortex. Color indicates adjusted P value. (B) Methionine metabolism and network pathways. (C) Relative expression of SAM, MTA, and metabolites related to cysteine and taurine metabolism in renal cortex (n = 6 per group). Boxes show the 25th to 75th percentiles, center lines indicate medians, whiskers extend to min and max, and all data points are shown. **P < 0.01 and *P < 0.05; Student’s t test. (D) Violin plot showing Mat2a expression in PTC versus PT-HFD of the WT-HFD. ***Adjusted P < 0.001. Values are presented as mean ± SEM.

Figure 5. Inhibition of methionine enzyme, MAT2A, abrogates kidney protection in SP^HFD mice.

(A) Schematic protocol. (B) Chronological changes in percentage of BW change. (C–E) Low-dose MAT2Ai does not alter glucose tolerance and insulin secretion capacity, but high dose MAT2Ai does lower them. (F) Serum creatinine level. (G) Representative images of PAS staining. Right panel, quantification. HPF, High-power field (original magnification ×40). Sample numbers: WT-PL, n = 7; WT-MAT2Ai LD, n = 7; WT-MAT2Ai HD, n = 3; SP-PL, n = 8; SP-MAT2Ai LD, n = 8; SP-MAT2Ai HD, n = 5. Scale bar: 50 μm. Data were analyzed by 1-way (C–G) or 2-way (B) ANOVA; ***P < 0.001, **P < 0.01, and *P < 0.05 by Tukey’s test. Values are presented as mean ± SEM.

Figure 6. Inhibition of MAT2A modulates the tubular inflammatory phenotype in SP^HFD mice.

(A) RNA-Seq heatmap for the top 50 differentially regulated genes. (B) Bubble plot of pathway enrichment; upregulated pathways in SP^HFD-LD_MAT2Ai versus SP^HFD placebo (size of the dot indicates the percent positive cells, and color indicates relative expression). (C) Representative images of p65 staining across groups. Lower panel, quantification analysis. HPF, High-power field (original magnification ×40). (D) RT-qPCR analysis of proinflammatory cytokines and chemokines in tissue isolated from the renal cortex across groups. Sample numbers: WT-PL, n = 7; WT-MAT2Ai LD, n = 7; SP-PL, n = 8; SP-MAT2Ai LD, n = 8. Scale bar: 50 μm. One-way ANOVA; ***P < 0.001, **P < 0.01, and *P < 0.05 by Tukey’s test. Values are presented as mean ± SEM.

Figure 7. SAM supplementation inhibits HG-induced pathogenic phenotypes in human PTCs.

(A) Schematic protocol of in vitro study using HK-2 cells. (B and C) SAM supplementation inhibited (B) transcript levels of proinflammatory cytokines (n = 18–24 cultures per group) and increased (C) phosphorylated p65 protein expression upon HG treatment. (D) MAT2Ai exacerbated proinflammatory cytokine expression, which are inhibited by SAM supplementation (n = 12–18 cultures per group). (E) Schematic protocol of in vitro study using RPTECs. (F) SAM supplementation inhibited transcript levels of proinflammatory cytokines in the presence of HG and palmitic acid (n = 6 cultures per group). One-way ANOVA; ***P < 0.001, **P < 0.01, and *P < 0.05 by Tukey’s test. Values are presented as mean ± SEM.

Figure 8. SP^HFD mice exhibit increased repressive histone modification at M-regulated gene promoters.

(A) Tracks for Fos, Junb, and Jun of H3K27me3 CUT&RUN in the indicated mouse renal cortex. Data from 3 mice for each condition are shown. The y axis indicates reference-normalized reads per million (RRPM). (B) Heatmap showing CUT&RUN signal for H3K27me3 and H3K4me3 at M-regulated genes. Signal is centered on the TSS. The scale of signal is shown as RRPM × 10³. n = 90. (C and D) Meta-profiles of CUT&RUN H3K27me3 signal at the M-regulated genes. The y axis indicates the mean H3K27me3 signal (RRPM). n = 90 (C) and 2,924 (D).

Figure 9. Working model for SGLT2 kidney protection under HFD metabolic stress.

Straight arrows indicate the direction of progression. T-shaped arrows indicate inhibition.