Postischemic inactivation of HIF prolyl hydroxylases in endothelium promotes maladaptive kidney repair by inducing glycolysis

Abstract

Ischemic acute kidney injury (AKI) is common in hospitalized patients and increases the risk for chronic kidney disease (CKD). Impaired endothelial cell (EC) functions are thought to contribute in AKI to CKD transition, but the underlying mechanisms remain unclear. Here, we identify a critical role for endothelial oxygen sensing prolyl hydroxylase domain (PHD) enzymes 1-3 in regulating postischemic kidney repair. In renal endothelium, we observed compartment-specific differences in the expression of the 3 PHD isoforms in both mice and humans. Postischemic concurrent inactivation of endothelial PHD1, PHD2, and PHD3 but not PHD2 alone promoted maladaptive kidney repair characterized by exacerbated tissue injury, fibrosis, and inflammation. scRNA-Seq analysis of the postischemic endothelial PHD1, PHD2, and PHD3-deficient (PHDTiEC) kidney revealed an endothelial hypoxia and glycolysis-related gene signature, also observed in human kidneys with severe AKI. This metabolic program was coupled to upregulation of the SLC16A3 gene encoding the lactate exporter monocarboxylate transporter 4 (MCT4). Strikingly, treatment with the MCT4 inhibitor syrosingopine restored adaptive kidney repair in PHDTiEC mice. Mechanistically, MCT4 inhibition suppressed proinflammatory EC activation, reducing monocyte-EC interaction. Our findings suggest avenues for halting AKI to CKD transition based on selectively targeting the endothelial hypoxia-driven glycolysis/MCT4 axis.

Keywords: Chronic kidney disease; Endothelial cells; Hypoxia; Metabolism; Nephrology.

Figure 1. Postischemic inactivation of endothelial PHD2 does not alter postischemic kidney injury.

(A) Experimental schematic illustrates the timing of unilateral renal artery clamping, tamoxifen administration, and analysis. (B) Representative images of H&E- and Picrosirius red– stained sections from day 14 postischemic kidneys of PHD2^iEC mutants and their Cre^– littermates. Right panels show tubular injury score (top) and semiquantitative analysis of Picrosirius red^+ve area in the indicated genotypes. Scale bars: 100 μm (H&E); 200 μm (Picrosirius red). (C) mRNA levels of Loxl2, Tgfb1, and Acta2 in IR and CTL kidneys from PHD2^iEC mice and their Cre^– controls at day 14 after uIRI. All bars show mean ± SEM. For B, unpaired t test with Welch’s correction was used. For C, statistics were determined using 1-way ANOVA with Šidák’s correction for multiple comparisons. n = 7–8. CTL, contralateral; IR, kidney subjected to ulRI; Rel., relative.

Figure 2. scRNA-Seq analysis shows differential expression of PHD1, PHD2, and PHD3 in kidney ECs in mice and humans.

(A–C) scRNA-Seq analysis of RECs extracted from mouse EC Atlas database (https://endotheliomics.shinyapps.io/ec_atlas/). (A) UMAP plot shows 3 EC clusters: cRECs, mRECs, and glomerular RECs (gRECs). Dot plot displays gene expression patterns of cluster-enriched markers. Violin plots (B) and feature plots (C) show the expression of Phd1 (Egln2), Phd2, (Egln1) and Phd3 (Egln3) in murine RECs. (D–F) scRNA-Seq analysis of RECs extracted from normal human kidney biopsies (n = 24). (D) UMAP plot shows arteriolar RECs (aRECs), glomerular RECs, and peritubular RECs (pRECs). Dot plot illustrates gene expression patterns of cluster-enriched markers. Violin plots (E) and feature plots (F) show the expression of PHDs in different REC clusters.

Figure 3. Postischemic simultaneous inactivation of endothelial PHD1, -2, and -3 promotes maladaptive kidney repair.

(A) Schematic illustrating the experimental strategy applied for uIRI studies. (B) Representative images of H&E- and Picrosirius red–stained sections as well as tubular injury score and semiquantitative analysis of Picrosirius red^+ve area on day 14 postischemic kidneys from PHD^TiEC mice and Cre^– littermates. Scale bars: 100 μm (H&E); 200 μm (Picrosirius red). (C) mRNA levels of Loxl2, Tgfb1, and Acta2 in IR and CTL kidneys from PHD^TiEC mice and their Cre^– controls at day 14 after uIRI. (D) Representative images of EMCN immunostaining and semiquantitative analysis of EMCN^+ve peritubular capillary area on day 14 postischemic kidneys from PHD^TiEC mice and Cre^– littermates. (E) Schematic depicting the experimental workflow for bIRI studies. (F) Serum BUN levels at different time points and GFR measurements on day 14 after bIRI. All bars show mean ± SEM of each group. (G) mRNA levels of Acta2 and Loxl2 in IR kidneys from PHD^TiEC mice and their Cre^– controls at day 14 after bIRI. For B, D, F, and G, statistics were determined by unpaired t test with Welch’s correction. For C, 1-way ANOVA with Šidák’s correction for multiple comparisons was used. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. n = 4–8.

Figure 4. scRNA-Seq analysis reveals the cellular landscape of day 14 postischemic kidneys of PHD^TiEC and control mice.

(A) Schematic illustrating the experimental strategy applied for scRNA-Seq analysis. A PHD^TiEC mouse and a Cre^– littermate were subjected to 25 minutes of unilateral renal artery clamping. Treatment with tamoxifen was started on day 1 post uIRI and involved 4 i.p. doses given every other day. Mice were sacrificed for scRNA-Seq analysis on day 14 after uIRI. IR kidneys were isolated and single-cell suspension was prepared and used for scRNA-Seq analysis (n = 1 per genotype). (B) UMAP plot representation of the cell classification in day 14 postischemic kidneys from PHD^TiEC and Cre^– mice. (C) Violin plots display characteristic marker genes for each identified cell population. Right side bar plot shows cell proportions in day 14 postischemic kidneys from Cre^– and PHD^TiEC mice. Highlighted is the increased proportion of FIBs in PHD^TiEC postischemic kidney compared with Cre^– control. PT, Inj-PT, TAL, DCT, CD, pCD, IM-CD, CD-IC, IC, PAR, FIB, PER, EC1-3, URO, Mφ1-4, pMφ, C1q-IM, T, pT, NK, B, DEN, and NEU. (D) Top 2 enriched Hallmark pathways emerged in GSEA hallmark analysis of DEGs for PT, DCT, and CD clusters of PHD^TiEC postischemic kidney as compared with Cre^– control.

Figure 5. Postischemic endothelial PHD inactivation induces a hypoxia and glycolysis gene signature in mRECs.

(A) Dot plot visualization shows the expression of marker genes used to identify cRECs, mRECs, and EndMT-RECs clusters. (B) GSEA in mRECs of PHD^TiEC kidney compared with control. Among the most highly enriched Hallmark pathways were hypoxia and glycolysis. (C) Violin plots show significantly upregulated glycolytic genes in mRECs of PHD^TiEC compared with control. Pathway diagram summarizes the functions of upregulated genes (marked by teal boxes) in glycolysis. (D and E) snRNA-Seq analysis of human kidney tissue from patients with severe AKI and controls (n = 6–8). Analysis was performed on publicly available snRNA-Seq data from Christian Hinze et al. (34). (D) Bubble chart for top 10 enriched Hallmark pathways of upregulated DEGs in kidney ECs from patients with severe AKI compared with controls. (E) Box plots show the expression of glycolytic genes in kidney ECs in controls versus AKI patients. The expression levels of glycolytic genes were extracted from the online interface provided by Christian Hinze et al. (https://shiny.mdc-berlin.de/humAKI). CPM, normalized counts per million.

Figure 6. Postischemic inactivation of endothelial PHDs induces EC-derived proinflammatory responses.

(A) GSEA plots show significant enrichment for GO-biological processes (GOBP) of leukocyte migration and myeloid leukocyte migration in mRECs of PHD^TiEC postischemic kidney compared with control. (B) Violin plots display the expression levels of proinflammatory genes associated with core enrichment in GOBP-leukocyte migration and myeloid leukocyte migration in mRECs of PHD^TiEC compared with control. (C) Shown is the experimental strategy for flow cytometry analysis. Eight days after uIRI, postischemic kidneys from PHD^TiEC and Cre^– control mice were harvested, and flow cytometry analysis of immune cells was performed (n = 4–5). Data are represented as mean ± SEM. Statistics were determined by unpaired t test with Welch’s correction. (D) Representative images of immunofluorescence staining for F4/80 (green) and nuclear DAPI staining (blue) of day 8 postischemic kidneys from PHD^TiEC and Cre^– control mice. Images were captured using a Nikon Ti2 Widefield fluorescence microscope. Scale bar: 100 μm. (E) Shown is NicheNet analysis of mRECs communication with macrophage Mφ1 cluster in day 14 post-IRI kidney of PHD^TiEC mutant compared with control. Top prioritized ligands expressed by mRECs (senders) and target genes that are significantly altered (red, upregulated genes; blue, downregulated genes) in the macrophages Mφ1 (receivers). The interaction pairs were derived from the NicheNet data sources and analysis. (F) Hallmark analysis of significantly upregulated genes in Mφ1 cluster of PHD^TiEC kidney compared with Cre^– control. Top 10 pathways are shown. ^*P < 0.05.

Figure 7. Postischemic inactivation of endothelial ARNT promotes adaptive repair following postischemic kidney injury.

(A) Experimental schematic illustrates the timing of unilateral renal artery clamping, tamoxifen administration, and analysis. (B) Representative images of H&E- and Picrosirius red–stained sections from day 14 postischemic kidneys of ARNT^iEC mutants and their Cre^– littermates. Right panels show tubular injury score (top) and semiquantitative analysis of Picrosirius red^+ve area in the indicated genotypes. Scale bars: 100 μm (H&E); 200 μm (Picrosirius red). (C) mRNA levels of Acta2, Loxl2, Tgfb1, and Havcr1 in IR and CTL kidneys from ARNT^iEC mice and their Cre^– controls at day 14 after uIRI. All bars show mean ± SEM. For B, unpaired t test with Welch’s correction was used. For C, statistics were determined using 1-way ANOVA with Šidák’s correction for multiple comparisons. n = 6–9. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 8. Post-IRI treatment with the MCT4 inhibitor syrosingopine restores adaptive kidney repair in PHD^TiEC mice.

(A) Representative images of immunofluorescence staining for MCT4 (red) and EMCN (green) of contralateral and day 14 postischemic kidneys from PHD^TiEC mice. Zoom-in panels show the increased expression of MCT4 in EMCN^+ve cells in PHD^TiEC postischemic kidney. Images were captured using a Nikon Ti2 Widefield fluorescence microscope. Scale bar: 100 μm. (B) Schematic illustrates the timing of unilateral renal artery clamping, tamoxifen administration, treatment with syrosingopine, and analysis at day 14 after uIRI. (C) Representative images of uninjured kidney compared with day 14 postischemic kidneys treated with vehicle or syrosingopine. All mice are PHD^TiEC mutants. (D) Representative images of H&E- and Picrosirius red–stained day 14 postischemic kidneys from vehicle- versus syrosingopine-treated PHD^TiEC mutants. Right: tubular injury score and semiquantitative analysis of Picrosirius red^+ve area of day 14 postischemic kidneys for the indicated experimental groups. Scale bars: 100 μm (H&E); 200 μm (Picrosirius red). (E) mRNA levels of Loxl2, Tgfb1, and Acta2 in CTL and IR kidneys from vehicle- or syrosyngopine-treated PHD^TiEC mice on day 14 after uIRI. Data are represented as mean ± SEM. For D, statistics were determined by unpaired t test with Welch’s correction. For E, statistics were determined using 1-way ANOVA with Šidák’s correction for multiple comparisons. n = 4–6. ^*P < 0.05; ^***P < 0.001. veh, vehicle; Syro, syrosingopine.

Figure 9. Syrosingopine or MCT4 knockdown suppresses the expression of EC adhesion molecules in HPAECs activated by hypoxia/reoxygenation and IL-1β.

(A) Experimental schematic for HPAECs subjected to 0.5% O₂ for 18 hours in the presence of syrosingopine (5 μM) or MCT4 siRNA followed by reoxygenation for 8 hours in the presence of IL-1β (1 ng/ml). (B) mRNA levels of VCAM1 and ICAM1 in syrosingopine- vs vehicle-treated HPAECs, that were activated by hypoxia/reoxygenation and IL-1β. (C) THP1 monocyte adhesion to inflamed ECs. THP1 monocyte cells, labeled with green CMFDA dye, were introduced on a monolayer of HPAECs that had been subjected to the indicated experimental conditions. Following a 90-minute incubation period, floating cells were washed away and adhered THP1 cells were visualized using a fluorescent microscope and subsequently quantified. Representative images of fluorescent THP1 cells attached to ECs in different experimental groups are presented. Scale bar: 200 μm. (D) mRNA expression of VCAM1 and ICAM1 in HPAECs transfected with control or MCT4 siRNA and subjected to hypoxia/reoxygenation and IL-1β. (E) THP1 monocyte adhesion to inflamed ECs under the same experimental conditions as in D. Scale bar: 200 μm. Data are represented as mean ± SEM. Statistics were determined by 1-way ANOVA with Šidák’s correction for multiple comparisons. n = 3–4. ^*P < 0.05; ^**P <0.01; ^***P <0.001; ^****P < 0.0001. Nx, normoxia; Hx, hypoxia/reoxygenation; C, negative control siRNA; MCT4si, MCT4siRNA.