CerS6 links ceramide metabolism to innate immune responses in diabetic kidney disease

Abstract

Ectopic lipid deposition, mitochondrial injury, and inflammatory responses contribute to the development of diabetic kidney disease (DKD); however, the mechanistic link between these processes remains unclear. In this study, we demonstrate that the ceramide synthase 6 (CerS6) is primarily localized in podocytes of the glomeruli and is upregulated in two different models of diabetic mice. Podocyte-specific CerS6 knockout ameliorates glomerular injury and inflammatory responses in male diabetic mice and in male mice with adriamycin-induced nephropathy. In contrast, podocyte-specific overexpression of CerS6 sufficiently induces proteinuria. Mechanistically, CerS6-derived ceramide (d18:1/16:0) can bind to the mitochondrial channel protein VDAC1 at Glu59 residue, initiating mitochondrial DNA (mtDNA) leakage, activating the cGAS-STING signaling pathway, and ultimately promoting an immune-inflammatory response in the kidney. Importantly, CERS6 expression is increased in podocytes from kidney biopsies of patients with DKD and focal segmental glomerulosclerosis (FSGS), and the expression level of CERS6 is correlated negatively with glomerular filtration rate and positively with proteinuria. Thus, our findings suggest that targeting CerS6 may be a potential therapeutic strategy for proteinuric kidney diseases.

Introduction

Diabetic kidney disease (DKD) is the leading cause of end-stage renal disease (ESRD) worldwide, affecting approximately 40% of diabetic patients and posing a huge burden on public health¹. Current interventions, such as hypoglycemic agents and renin-angiotensin-aldosterone system (RAAS) blockers, are unable to completely prevent DKD progression to ESRD². Therefore, other unknown mechanisms besides hyperglycemia and hemodynamic abnormalities may dominate or coordinate the development of DKD. The clinical use of novel hypoglycemic sodium-glucose cotransporter 2 (SGLT2) inhibitors and glucagon-like peptide-1 (GLP-1) agonists confirm this view, as they exert cardiovascular and renal protection independent of hypoglycemic effects^3,4.

Emerging evidence has established that renal lipid metabolism is disturbed in DKD, which is a pathological hallmark of DKD⁵. However, different kidney cell types exhibit different sensitivities to lipid accumulation. Unlike tubular cells, the induction of considerable triglyceride accumulation does not produce a fibrotic phenotype⁶, and podocytes are particularly sensitive to lipid accumulation, which leads to insulin resistance and apoptosis, a hallmark of proteinuria, and deterioration of renal function^7,8. This may be attributed to the variations in lipid accumulation in different kidney cells. Under diabetic conditions, podocytes are dominated by cholesterol and sphingolipid accumulation in lipid rafts, and lipid overload triggers slit diaphragm (SD) structural damage and aberrant signaling^9,10. Recently, pioneering studies have demonstrated that abnormalities in sphingolipid metabolites, such as ceramide (Cer), sphingosine, and sphingosine-1-phosphate (S1P), are associated with podocyte injury in kidney diseases^10,11. However, the exact mechanism through which sphingolipid accumulation leads to podocyte injury remains poorly understood.

As central intermediates in sphingolipid metabolism, ceramides are involved in the regulation of essential cellular processes, including mitochondrial damage, cell senescence, and apoptosis^12,13,14. Ceramide synthases (CerS) are critical enzymes required for the de-novo synthesis of ceramides¹⁵. To date, six CerS (CerS1-CerS6) have been identified, which show diverse tissue distribution and substrate specificity for different acyl chain lengths (C_14:0-C_30:0)¹⁵. Chew et al. demonstrated that elevated plasma levels of Cer (d18:1/16:0), which is generated by CerS6, are associated with a higher risk of T2DM¹⁶. Additionally, increased CerS6 and its-derived Cer (d18:1/16:0) aggravate obesity-induced mitochondrial fragmentation and insulin resistance by interacting with mitochondrial fission factor (Mff)¹³. Notably, recent single-cell RNA sequencing (scRNA-seq) analyses and spatial metabolomics have identified CerS6 as a podocyte-specific essential gene, while its catalytic production of sphingomyelin (d18:1/16:0) specifically accumulated in the glomerulus region of kidney cortex^17,18. However, whether CerS6 and Cer (d18:1/16:0) are altered in DKD and whether these changes affect podocyte fate remains unclear.

In this study, we revealed that CerS6 was predominantly expressed in the podocytes of the glomeruli and was increased in patients with DKD, focal segmental glomerulosclerosis (FSGS), and mouse models. Podocyte-specific deletion of CerS6 alleviated podocyte injury and proteinuria, which may be attributed to the amelioration of mitochondrial damage and the inflammatory response. Specifically, accumulated CerS6-derived Cer (d18:1/16:0) in the mitochondria induced the leakage of mitochondrial DNA (mtDNA), further activating the intracellular DNA receptor cyclic GMP-AMP (cGAMP) synthase (cGAS), and subsequently triggering an innate immune response by activating the stimulator of interferon genes (STING). Therefore, our results suggest that CerS6 is a key mediator of tandem podocyte lipid accumulation, mitochondrial damage, and inflammatory response and may serve as a promising therapeutic target for DKD.

Results

CerS6 was elevated in podocytes from two different mouse models of glomerular injury and patients with glomerular disease

We first evaluated the expression of CerS6 in various organs of mature mice by Western blotting (WB) and found that CerS6 was present in the liver, spleen, and kidneys (Fig. 1A). To understand the expression profile of CerS6 in the kidney, we reanalyzed previously published single-cell RNA sequencing (scRNA-seq) data of healthy adult human¹⁹ and mouse²⁰ kidneys. The scRNA-seq data from healthy adult human kidneys¹⁹ presented in Fig. 1B indicate that CerS6 is predominantly expressed in podocytes²⁰. And the data depicted in Fig. S1A derived from the kidneys of healthy mice demonstrate that CerS6 is expressed in both podocytes and tubular cells. However, the violin and dot plots further illustrate a significantly higher expression of CerS6 in podocytes compared to tubular cells. We performed trichrome fluorescent staining (CerS6/Synaptopodin/LTL) on both human and mouse kidney sections. The results indicate that in human kidney samples, CERS6 is predominantly expressed in podocytes, with comparatively lower expression observed in the tubules (Fig. S1B). Similarly, staining of mouse kidney sections revealed abundant expression of CerS6 in podocytes, alongside detectable levels in the tubules (Fig. S1C). Podocyte injury was strongly associated with damage to the glomerular filtration barrier and proteinuria. Therefore, we next analyzed the CerS6 expression profile in the kidneys of patients with DKD using scRNA-seq data²¹. Similarly, the data also showed that CerS6 was mainly expressed in podocytes and the gene activity of CerS6 was increased in DKD (Fig. 1C). To validate the sequencing results, we further evaluated the expression of CerS6 in mouse models. Importantly, CerS6 was significantly upregulated in the glomeruli from two different mouse models of glomerular injuries, including the streptozocin plus high-fat diet (STZ/HFD)-induced diabetic mice, which is a model of late-stage type II diabetes (T2D) (Fig. 1D), db/db mice, a spontaneous T2D model (Fig. 1E), and adriamycin (ADR)-induced FSGS (Fig. S2A) by WB and immunohistochemistry (IHC) analysis (Fig. 1F and Fig. S2B). Double immunofluorescence for the podocyte markers synaptopodin and CerS6 showed that CerS6 was localized primarily in podocytes and was markedly increased in db/db and STZ/HFD-induced diabetic mice (Fig. 1G and Fig. S2C). Since CerS6 mainly catalyzes the synthesis of Cer (d18:1/16:0), we next measured ceramide levels in the kidney cortex. The LC-MS results revealed that four ceramides, Cer (d18:1/16:0), Cer (d18:1/18:0), Cer (d18:1/22:0), and Cer (d18:1/24:1), were elevated in the renal cortex of STZ/HFD-induced diabetic mice (Fig. S2D), while three ceramides, Cer (d18:1/16:0), Cer (d18:1/18:0), and Cer (d18:1/22:0), were increased in ADR-treated mice (Fig. S2E). Notably, the elevation of Cer (d18:1/16:0) was the most significant in both models. To further analyze the expression pattern of CerS6 in podocytes, we isolated primary podocytes from mT/mG/NPHS2-Cre mice (Fig. 1H). CerS6 expression was significantly elevated in primary podocytes treated with high glucose (HG) or ADR in a concentration-dependent manner (Fig. 1I and Fig. S2F). Similarly, the CERS6 was markedly increased in cultured human podocytes after exposure to HG or ADR (Fig. 1J and Fig. S2G). Moreover, five ceramides, Cer (d18:1/16:0), Cer (d18:1/18:0), Cer (d18:1/22:0), Cer (d18:1/23:0), and Cer (d18:1/24:1), were increased in podocytes after HG exposure, with Cer (d18:1/16:0) showing the most pronounced increase (Fig. S2H).

Fig. 1: CerS6 was elevated in podocytes from two different mouse models of glomerular injuries and patients with glomerular disease.

A Representative Western blots and quantification of CerS6 relative to GAPDH in mouse organs including heart, liver, spleen, lung, and kidney (n = 5 biologically independent experiments). B scRNA-seq analysis revealed that CERS6 was specifically expressed in podocytes. t-SNE plots showing human kidney cell clusters and the expression of CERS6. C scRNA-seq analysis established that the gene activity of CERS6 was increased in podocytes from DKD patients. D, E Representative Western blots and quantification of CerS6 expression in glomeruli from streptozocin plus high-fat diet (STZ/HFD)-induced diabetic mice (D), and db/db mice (E) (n = 6 mice per group). F Representative immunohistochemistry (IHC) images of CerS6 in the kidney from db/db mice (n = 6 mice per group). Scale bar: 10 μm. G Representative immunofluorescent (IF) images of CerS6 (green) and synaptopodin (red) in glomeruli from db/db mice (n = 6 mice per group). Scale bar: 10 μm. H A schematic diagram showing the isolation of primary podocytes (Created in BioRender. https://BioRender.com/d65a412). I, J Representative Western blots and quantification of CerS6 expression in primary podocyte (I) and human podocyte (J) treated with HG (n = 3 biologically independent experiments). K Representative IF images of CERS6 (green) and Synaptopodin (red) in human renal tissue from control subjects (n = 10), patients with diabetic kidney disease (DKD) (n = 10), or focal segmental glomerulosclerosis (FSGS) (n = 10). Scale bar: 50 μm (white), 25 μm (red). L–N Correlation between glomerular CERS6 expression and estimated glomerular filtration rate (eGFR, n = 20) (L), serum creatinine (SCr, n = 20) (M), or 24 h urine total protein (n = 20) (N) in patients with DKD and FSGS. Data are expressed as mean ± SEM (A, D, E, I, J, and K). Two-tailed Student’s unpaired t-test analysis (D and E). One-way ANOVA followed by Tukey’s post-test (I and J). Kruskal–Wallis test followed by Dunn’s post-test (K). Nonparametric Spearman’s correlation coefficient r with two-tailed p-value (L–N). Source data are provided as a Source Data file.

To corroborate these results, we measured CERS6 expression in human kidney biopsy samples from patients with DKD and FSGS. Double immunofluorescence for synaptopodin and CERS6 revealed that the expression of CerS6 in podocytes was also elevated in subjects with DKD or FSGS compared to that in normal subjects (Fig. 1K). Importantly, the level of CERS6 in podocytes was negatively correlated with the estimated glomerular filtration rate (eGFR) (Fig. 1L) and positively correlated with serum creatinine (Fig. 1M) and proteinuria (Fig. 1N) levels in all subjects. Taken together, these results demonstrate that CerS6 expression is significantly increased in glomerular disease and is highly involved in podocyte injury.

Podocyte-specific CerS6 deletion attenuated glomerular and podocyte injury in diabetic mice

We further established podocyte-specific CerS6 knockout mice (Cre⁺/CerS6^fl/fl mice) using the Cre-Loxp system (Fig. S3A), as confirmed by tail genotyping (Fig. S3B). WB analysis revealed that CerS6 expression in the glomeruli was significantly lower in Cre⁺/CerS6^fl/fl mice than in Cre⁺/CerS6^+/+ mice (Fig. S3C). The knockout of CerS6 in podocytes from Cre⁺/CerS6^fl/fl mice was confirmed using double immunofluorescence staining (Fig. S3D). In addition, the Cer (d18:1/16:0) levels were significantly reduced in the renal cortex of Cre⁺/CerS6^fl/fl mice compared to Cre⁺/CerS6^+/+ mice (Fig. S3E). All Cre⁺/CerS6^fl/fl mice were viable and fertile, and did not show any phenotypic changes under normal conditions (Fig. 2B–D).

Fig. 2: Podocyte-specific CerS6 deletion attenuated glomerular and podocyte injury in diabetic mice.

A A schematic diagram showing the construction of the STZ/HFD-induced diabetic mice model (Created in BioRender. https://BioRender.com/i73m293). B Urine albumin-to-creatinine ratio (ACR) of mice in different groups (n = 6 mice per group). C Representative hematoxylin and eosin (H&E) staining images and quantification of glomerular mesangial matrix expansion in different groups of mice (n = 6 mice per group). Scale bars: 20 μm. D Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) analyses of podocyte ultrastructure in different groups of mice (n = 6 mice per group). Scale bars: 1 μm. E–G Representative IF images and quantifications of expressions of Wilms’ tumor-1(WT-1) (E), synaptopodin (F) and nephrin (G) in glomeruli from mice (n = 6 mice per group). Scale bars: 20 μm. Source data are provided as a Source Data file. Data are expressed as mean ± SEM (B–G). Two-way ANOVA followed by Tukey’s post-test (B–G). Source data are provided as a Source Data file.

To explore the therapeutic effect of targeting CerS6 in mice with DKD, we induced DKD in Cre⁺/CerS6^fl/fl and Cre⁺/CerS6^+/+ mice using STZ/HFD (Fig. 2A). Diabetic Cre⁺/CerS6^fl/fl mice exhibited decreased urinary albumin excretion and less glomerular mesangial expansion than diabetic Cre⁺/CerS6^+/+ mice (Fig. 2B, C). Transmission electron microscopy (TEM) was performed to examine the glomerular ultrastructure and revealed less basement membrane thickening, podocyte foot process broadening, and effacement in diabetic Cre⁺/CerS6^fl/fl mice than in diabetic Cre⁺/CerS6^+/+ mice (Fig. 2D). Scanning electron microscopy (SEM) analysis confirmed these findings (Fig. 2D). In addition, CerS6 deletion reduced diabetes-induced podocyte loss, as evidenced by podocyte-specific marker staining for Wilms tumor-1 (WT-1), synaptopodin, and nephrin (Fig. 2E–G) as well as by WB analysis (Fig. S4). These results demonstrate that CerS6 deletion alleviates diabetes-induced podocyte injury. Glomerular sclerosis and renal tubulointerstitial fibrosis are important clinical indicators of DKD, we further evaluated whether knockdown of CerS6 can alleviate these pathological changes in diabetic mice. Diabetic Cre⁺/CerS6^+/+ mice exhibited glomerular sclerosis (Fig. S5A) and mild tubulointerstitial fibrosis (Fig. S5B). Interestingly, podocyte-specific CerS6 knockout alleviated glomerular sclerosis but did not reduce renal tubulointerstitial fibrosis (Fig. S5B).

Podocyte-specific CerS6 deletion alleviated glomerular and podocyte inflammation under diabetic conditions

To investigate the specific mechanisms of CerS6-mediated podocyte injury, we isolated glomeruli from diabetic mice and performed RNA sequencing²², which revealed significant changes in the pathways related to inflammation and lipid metabolism (Fig. 3A, B). In addition, many studies have confirmed that inflammatory response is closely related to DKD progression^23,24. Therefore, we verified whether the knockout of CerS6 in podocytes could reduce the glomerular inflammatory response. Our results demonstrated that the number of CD68-positive cells and the mRNA levels of inflammatory cytokines, including interleukin-1β (Il-1β), interleukin-6 (Il-6), tumor necrosis factor-α (Tnf-α), and monocyte chemoattractant protein-1 (Mcp-1) were significantly increased in glomeruli with diabetic Cre⁺/CerS6^+/+ mice, and these alternations were rescued by CerS6 deletion (Fig. 3C, D), suggesting that CerS6 was associated with glomerular inflammation. Since the cyclic GMP-AMP (cGAMP) synthase (cGAS)-stimulator of interferon genes (STING) pathway plays an important role in podocyte inflammation and injury^23,24, we further examined whether CerS6 could affect the activation of the cGAS-STING pathway. Interestingly, in the glomeruli of diabetic Cre⁺/CerS6^+/+ mice, the protein levels of cGAS, STING, phosphorylated TANK-binding kinase 1 (TBK1), and p65 were dramatically elevated; however, podocyte-specific deletion of CerS6 blunted these changes (Fig. 3E). To further confirm these results in vitro, cultured human podocytes were treated with HG, and HG-induced increases in inflammatory cytokine mRNA and activation of the cGAS-STING pathway were alleviated by CerS6 knockdown (Fig. S6A, Fig. 3F, G), suggesting that CerS6 could activate the inflammatory response by affecting the cGAS-STING pathway.

Fig. 3: Podocyte-specific CerS6 deletion alleviated glomerular and podocyte inflammation under diabetic conditions.

A GO analyses of the differential expression genes (DEGs) in the glomeruli of the db/m and db/db mice. B KEGG analyses of the DEGs in the glomeruli of the db/m and db/db mice. C Representative IF images and quantifications of CD68 expression in glomeruli from mice (n = 6 mice per group). White outlines indicate glomerulus, and white arrows mark CD68-positive cells. Scale bars: 20 μm. D Relative mRNA levels of inflammatory cytokines (Il-1β, Il-6, Tnf-α, and Mcp-1) in glomeruli from different groups of mice (n = 6 mice per group). *p < 0.05. E Representative Western blots and quantification of expressions of cGAS, STING, phosphorylation of TBK1 and p65 in glomeruli from different groups of mice (n = 6 mice per group). The samples derived from the same experiments but different gels for cGAS, STING, TBK, p-TBK1, p-p65, p65, and Actin were processed in parallel. F Relative mRNA levels of inflammatory cytokines (IL-1β, IL-6, TNF-α, and MCP-1) in podocytes with different treatments (n = 3 biologically independent experiments). G Representative Western blots and quantification of expressions of cGAS, STING, phosphorylation of TBK1 and p65 in podocytes with different treatments (n = 3 biologically independent experiments). The samples derived from the same experiments but different gels for cGAS, STING, TBK, p-TBK1, p-p65, p65, and Actin were processed in parallel. Data are expressed as mean ± SEM (C–G). Two-way ANOVA followed by Tukey’s post-test (C–G). Source data are provided as a Source Data file.

Podocyte-specific overexpression of CerS6 in mice sufficiently induced podocyte injury

To elucidate the role of CerS6 in glomerular and podocyte injury, we generated podocyte-specific CerS6 knock-in mice (Cre⁺/Rosa26-fl-STOP-fl-CerS6, CerS6^po-cKI mice) using the CRISPR-Cas9 system (Fig. S7A), as confirmed by tail genotyping (Fig. S7B). The glomerular CerS6 level was significantly increased in CerS6^po-cKI mice compared to that in control mice (CerS6^ctrl mice) by WB analysis (Fig. S7C), and the overexpression of CerS6 in podocytes from CerS6^po-cKI mice was confirmed by double immunofluorescence staining (Fig. S7D). Additionally, the Cer (d18:1/16:0) levels were markedly increased in the renal cortex of CerS6^po-cKI mice (Fig. S7E). Next, CerS6^po-cKI and CerS6^ctrl mice in age-matched groups were analyzed (Fig. 4A). CerS6^po-cKI mice exhibited increased urinary albumin excretion and glomerular mesangial expansion at 3 months of age, which were further aggravated at 4 months of age compared to those in CerS6^ctrl mice at different ages (Fig. 4B, C). In addition, TEM and SEM results showed that CerS6^po-cKI mice developed podocyte injury at 3 months of age, as evidenced by basement membrane thickening, podocyte foot process broadening, and effacement; these changes were also elevated at 4 months of age (Fig. 4D). Furthermore, the expression of podocyte-specific markers, including WT-1, synaptopodin, and nephrin, decreased with age in CerS6^po-cKI mice (Fig. 4E), suggesting podocyte loss in CerS6^po-cKI mice.

Fig. 4: Podocyte-specific overexpression of CerS6 in mice sufficiently induced podocyte injury.

A A schematic diagram showing the construction of the podocyte-specific CerS6 knock-in mice (CerS6^po-cKI mice) (Created in BioRender. https://BioRender.com/i73m293). B ACR of mice at different age (n = 6 mice per group). C Representative PAS staining images and quantification of glomerular mesangial matrix expansion in mice at different ages (n = 6 mice per group). Scale bars: 20 μm. D SEM and TEM analyses of podocyte ultrastructure in mice at different ages (n = 6 mice per group). Scale bars: 1 μm. E Representative IF images and quantifications of expressions of WT-1, synaptopodin and nephrin in glomeruli from mice at different age (n = 6 mice per group). Scale bars: 20 μm. Data are expressed as mean ± SEM (B–E). Two-tailed Student’s unpaired t-test analysis (B–E). Source data are provided as a Source Data file.

CerS6 contributed to podocyte mtDNA release and mitochondrial dysfunction induced by high glucose

Next, we explored the mechanism underlying the activation of the cGAS-STING pathway in podocytes under diabetic conditions. Because mtDNA leakage into the cytosol is an important stimulator that activates the cGAS-STING pathway, we hypothesized that CerS6 elevation may activate the cGAS-STING pathway by triggering mitochondrial damage and mtDNA leakage. Dysfunctional mitochondria of renal cells also release mtDNA into the urine; therefore, urinary mtDNA content can be used to evaluate the level of mtDNA release in podocytes to some extent. We collected urine samples from patients with DKD and FSGS to evaluate urinary mtDNA levels (Fig. 5A) and showed that it was significantly increased in subjects with DKD and FSGS compared to normal subjects (Fig. 5B). Importantly, urinary mtDNA levels negatively correlated with eGFR (Fig. 5C) and positively correlated with proteinuria (Fig. 5D) in all subjects. In addition, urinary mtDNA levels were increased in diabetic Cre⁺/CerS6^+/+mice, and diabetic Cre⁺/CerS6^fl/fl mice exhibited decreased urinary mtDNA levels (Fig. 5E), suggesting that CerS6 is associated with podocyte mtDNA release. Next, we examined the relationship between CerS6 expression and mtDNA release from the cultured podocytes. Confocal microscopy showed that HG treatment resulted in a marked increase in mtDNA in the cytosol of podocytes, and knockdown of CerS6 significantly attenuated this alteration (Fig. 5F). In addition, mitochondria-free cytosolic fractions of podocytes were isolated, and the amount of mtDNA in the cytosolic fractions was quantified by qRT-PCR. The results revealed that the deletion of CerS6 significantly mitigated the HG-induced increase in cytosolic mtDNA in podocytes (Fig. 5G), confirming that CerS6 deletion prevented HG-induced mtDNA leakage in podocytes. Furthermore, fluorescence staining showed that the colocalization of dsDNA with cGAS in the cytosol was significantly increased in podocytes after exposure to HG, but CerS6 knockdown significantly ameliorated this colocalization (Fig. 5H).

Fig. 5: CerS6 contributed to podocyte mtDNA release and mitochondrial dysfunction induced by high glucose.

A A schematic diagram showing the collection of urine samples from DKD, FSGS patients, and diabetic mice (Created in BioRender. https://BioRender.com/i73m293). B Urine was collected from patients with DKD (n = 10) or FSGS (n = 10), urinary mtDNA levels were quantified by qPCR. C, D Correlation between urinary mtDNA levels and eGFR (n = 20) (C), or 24 h urine total protein (n = 20) (D) in all subjects. E Urine was collected from different groups of mice, urinary mtDNA levels were quantified by qPCR (n = 6 mice per group). F Representative IF images of Tomm20 (red) and dsDNA (green) in cultured podocytes (n = 3 biologically independent experiments). The white arrow marks the mtDNA leakage. Scale bars: 10 μm, 5 μm (Zoom). G Cytosolic translocation of mtDNA in cultured podocytes was quantified by qPCR (n = 3 biologically independent experiments). H Representative IF images of cGAS (red) and dsDNA (green) in cultured podocytes (n = 3 biologically independent experiments). The white arrow marks the binding of the leaked mtDNA to cGAS. Scale bars: 10 μm, 5 μm (Zoom). I, J Representative images and quantification of mitoSOX staining showing the mtROS accumulation (I) and JC-1 staining showing the MMP levels in cultured podocytes (J), (n = 3 biologically independent experiments). Scale bars: 100 μm. K Representative images of mitochondrial morphology visualized by MitoTracker red staining in cultured podocytes, (n = 3 biologically independent experiments). Scale bars: 10 μm. L Relative ATP levels in cultured podocytes (n = 3 biologically independent experiments). Data are expressed as mean ± SEM (B, E, G, H, I, J and L). Kruskal–Wallis test followed by Dunn’s post-test (B). Nonparametric Spearman’s correlation coefficient r with two-tailed p-value (C and D). Two-way ANOVA followed by Tukey’s post-test (E, G, H, I, J and L). Source data are provided as a Source Data file.

To investigate whether CerS6 expression was associated with mitochondrial damage, we examined mitochondrial function and morphology in podocytes. MitoSOX red fluorescence detection demonstrated that CerS6 knockdown significantly mitigated HG-induced mitochondrial ROS (mtROS) production (Fig. 5I). JC-1 staining indicated that the ratio of JC-1 aggregates (red) to JC-1 monomers (green), an indicator of mitochondrial membrane potential (MMP), was markedly decreased in podocytes after HG treatment, and this change was reversed by CerS6 deletion (Fig. 5J). In addition, CerS6 deletion ameliorated HG-induced mitochondrial fission in podocytes (Fig. 5K). Finally, CerS6 knockdown rescued the HG-induced reduction in adenosine triphosphate (ATP) (Fig. 5L), and in basal and maximal mitochondrial respiration (Fig. S8A). These findings indicated that the upregulation of CerS6 was implicated in HG-induced mitochondrial damage and mtDNA release, which could contribute to the activation of the cGAS-STING pathway.

VDAC1 oligomerization was required for CerS6-overexpression or HG-induced mtDNA release

Next, we explored the mechanism by which CerS6 triggered mtDNA leakage from podocytes. Previous studies have reported that CerS6-derived Cer (d18:1/16:0) is accumulated in mitochondria and binds to mitochondrial proteins to regulate their function and morphology¹³. We isolated mitochondria from podocytes and measured the levels of ceramides, and LS-MS results showed that Cer (d18:1/16:0) was significantly increased in mitochondria from podocytes after exposure to HG (Fig. 6A). Moreover, knockdown of CerS6 led to a significant reduction of Cer (d18:1/16:0) in podocyte mitochondria (Fig. S6B). Therefore, it is reasonable to speculate that mito-Cer (d18:1/16:0) interacts with the mitochondrial proteins to trigger mtDNA leakage. In addition, it has been well established that oligomerization of voltage-dependent anion channel 1 (VDAC1) and translocation of BAX from the cytosol to the mitochondria to mediate mitochondrial outer membrane permeabilization (MOMP), which allows proteins and mtDNA, often associated with inflammation, to escape from the mitochondria^25,26. Hence, we examined whether HG could affect the oligomerization of VDAC1 and BAX localization in podocytes, and also evaluated the role of CerS6 in this process. WB analysis demonstrated that HG significantly induced VDAC1 oligomerization and BAX mitochondrial translocation, while knockdown of CerS6 only diminished VDAC1 oligomerization but had no effect on BAX translocation (Fig. 6B, Fig. S9A, B). Furthermore, we used Autodock Vina to dock Cer (d18:1/16:0) to the VDAC1 domain, and found that Cer (d18:1/16:0) has a certain binding effect with VDAC1 (Fig. 6C). Through investigating the inter-molecular hydrogen bonds, the key amino acid residues through which Cer (d18:1/16:0) interacted with VDAC1 were determined, including Thr42, Glu59, Lys61. In addition, molecular dynamics (MD) simulations revealed that VDAC1 has a binding site for Cer (d18:1/16:0) buried in the membrane interior on one side of the barrel wall and that its structure changes after binds with Cer (d18:1/16:0) (Fig. 6D, E, Supplementary Movie 1). To determine the exact binding site of Cer (d18:1/16:0) on VDAC1, the three key amino acid residues Thr42, Glu59, and Lys61 were individually mutated to glutamine. We next introduced plasmids expressing mutant VADC1^T42Q, VADC1^E59Q, or VADC1^K61Q into podocytes, and CerS6 overexpression was achieved by transfection with LV-CerS6 (Fig. S6C). The LC-MS results demonstrated that overexpression of CerS6 markedly elevated the Cer (d18:1/16:0) levels in podocyte mitochondria (Fig. S6D). Confocal microscopy showed that CerS6-overexpression-induced mtDNA release was only rescued by VADC1^E59Q mutant, while the VADC1^T42Q and VADC1^K61Q mutant have no effect on mtDNA release (Fig. 6F, G). Importantly, the WB analysis revealed that the CerS6-overexpression-induced VDAC1 oligomerization (Fig. 6H, Fig. S9C), and the binding of leaked mtDNA to cGAS (Fig. 6I) was notably reduced in the VADC1^E59Q mutant. These data suggested that mito-Cer (d18:1/16:0) could bind to VDAC1 Glu59 residues and induce its oligomerization to trigger mtDNA release.

Fig. 6: VDAC1 oligomerization was required for CerS6-overexpression or HG-induced mtDNA release.

A Ceramide profile of mitochondria of podocytes treated with HG or control (Created in BioRender. https://BioRender.com/b19y621) (n = 5 biologically independent experiments). B Representative Western blot showing VDAC1 oligomerization and BAX mitochondrial translocation in cultured podocytes (n = 3 biologically independent experiments). The samples derived from the same experiments but different gels for VDAC1, BAX, and Actin were processed in parallel. C Docking and virtual screening of VDAC1 with Cer (d18:1/16:0). D Stills from an MD simulation, showing the approach and binding of a Cer (d18:1/16:0) molecule to VDAC1. E A schematic diagram showing VDAC1 oligomerization–dependent release of mtDNA triggered by Cer (d18:1/16:0) (Created in BioRender. https://BioRender.com/b19y621). F Representative IF images of Tomm20 (red) and dsDNA (green) in cultured podocytes (n = 3 biologically independent experiments). The white arrow marks the mtDNA leakage. Scale bars: 10 μm, 5 μm (Zoom). G A schematic diagram showing the effects of different VADC1 mutants on mtDNA leakage (Created in BioRender. https://BioRender.com/b19y621). H Representative Western blot showing VDAC1 oligomerization in cultured podocytes (n = 3 biologically independent experiments). The samples derived from the same experiments but different gels for VDAC1, and Actin were processed in parallel. I Representative IF images of cGAS (red) and dsDNA (green) in cultured podocytes (n = 3 biologically independent experiments). The white arrow marks the binding of the leaked mtDNA to cGAS. Scale bars: 10 μm. Data are expressed as mean ± SEM (A). Two-tailed Student’s unpaired t-test analysis (A). Source data are provided as a Source Data file.

To further confirm that VDAC1 oligomerization is crucial for mtDNA release as a result of Cer (d18:1/16:0) accumulation in mitochondria, CerS6-overexpressed podocytes were co-treated with VBIT-4, an inhibitor of VDAC1 oligomerization. Our results demonstrated that the CerS6-overexpression-induced mtDNA release (Fig. S10A, B), and that the binding of leaked mtDNA to cGAS was rescued by VBIT-4 (Fig. S10C). Moreover, mitochondrial dysfunction, including ATP deficiency, mtROS accumulation, MMP depolarization, (Fig. S10D, E), and impaired basal and maximal mitochondrial respiration caused by overexpression of CerS6 was restored with VBIT-4 treatment (Fig. S8B). In conclusion, these findings demonstrate that VDAC1 oligomerization is indispensable for CerS6-derived Cer (d18:1/16:0) to induce mtDNA release.

CerS6 and VDAC1 cooperatively mediated podocyte injury and inflammation

To further verify that VDAC1-dependent mtDNA release plays an important role in CerS6-overexpression-induced podocyte injury and inflammation, podocyte-specific Vdac1 knockout mice (Cre⁺/Vdac1^fl/fl mice) were injected with intrarenal adeno-associated viral vector serotype 9 (AAV9)-CerS6 to overexpress CerS6 (Fig. 7A). Mice were injected with AAV9-CerS6 at 2 months of age, and the CerS6 level was significantly increased in AAV9-CerS6 mice compared to control mice (AAV9-Crtl) at 4 (Fig. S12A), 5 months of age. Notably, CerS6 expression remained elevated even in the kidneys of 6-month-old mice (Fig. S11A). Additionally, the Cer (d18:1/16:0) levels were significantly increased in the renal cortex of AAV9-CerS6 mice at 6 months of age (Fig. S11B). Next, we evaluated to determine whether overexpression of CerS6 by AAV9 delivery could produce a renal phenotype. The results showed that AAV9-CerS6 mice exhibited increased urinary albumin excretion, glomerular mesangial expansion, and podocyte foot process broadening and effacement at 5 months of age; these changes were further aggravated at 6 months of age (Fig. S12B–D). Although accumulation of Cer (d18:1/16:0) may have occurred in tubular cells, there was no obvious tubular injury and change in urine osmotic pressure in AAV9-CerS6 mice at 6 months of age (Fig. S13A, B). However, it is noteworthy that overexpression of CerS6 in the kidney led to an increase in the number of CD68-positive cells (Fig. S13C) and a slight activation of the cGAS-STING pathway in the tubular area (Fig. S13D, E). We conclude that this may be attributed to the fact that CerS6 is primarily expressed in podocytes, where it likely plays a more crucial role. In addition, we simultaneously measured urinary mtDNA in CerS6^po-cKI and AAV9-CerS6 mice. Our results showed that urinary mtDNA levels in CerS6^po-cKI mice were increased by 121% compared with control mice, while that of AAV9-CerS6 mice were increased by 275% (Fig. S14). These data suggest that leakage of podocyte mtDNA may be a primary contributor to elevated urinary mtDNA in glomerular diseases, and further demonstrated that CerS6-mediated ceramide accumulation plays an important role in mtDNA leakage within kidney cells.

Fig. 7: CerS6 and VDAC1 cooperatively mediated podocyte injury and inflammation.

A A schematic diagram showing overexpression of CerS6 in the kidneys of Cre⁺/VDAC1^fl/fl mice (Created in BioRender. https://BioRender.com/g05h226). B ACR of mice in different groups (n = 6 mice per group). C PAS staining images and quantification of glomerular mesangial matrix expansion in different groups of mice (n = 6 mice per group). Scale bars: 20 μm. D SEM and TEM analyses of podocyte ultrastructure in different groups of mice (n = 6 mice per group). Scale bars: 1 μm. E–G Representative IF images and quantifications of expressions of WT-1 (E), synaptopodin (F) and nephrin (G) in glomeruli from mice (n = 6 mice per group). Scale bars: 20 μm. H Representative IF images and quantifications of CD68 expression in glomeruli from mice (n = 6 mice per group). White outlines indicate glomerulus, and white arrows mark CD68-positive cells. Scale bars: 20 μm. I Relative mRNA levels of inflammatory cytokines (Il-1β, Il-6, Tnf-α, and Mcp-1) in glomeruli from different groups of mice (n = 6 mice per group). J Representative Western blots and quantification of expressions of cGAS, STING, phosphorylation of TBK1, and p65 in glomeruli from different groups of mice (n = 6 mice per group). The samples derived from the same experiments but different gels for cGAS, STING, TBK, p-TBK1, p-p65, p65, and Actin were processed in parallel. Data are expressed as mean ± SEM (B–J). Two-way ANOVA followed by Tukey’s post-test (B–J). Source data are provided as a Source Data file.

The Cre⁺/Vdac1^fl/fl mice were generated using the Cre-Loxp system, which was verified by tail genotyping (Fig. S15A, B). Knockdown of Vdac1 in podocytes was confirmed by WB analysis (Fig. S15C) and double immunofluorescence for synaptopodin and CerS6 (Fig. S15D). Importantly, knockout of Vdac1 in podocytes significantly attenuated CerS6-overexpression-induced proteinuria, mesangial matrix accumulation, basement membrane thickening, and podocyte foot process broadening and effacement (Fig. 7B–D). In addition, the reduced expression of WT-1, synaptopodin, and nephrin caused by CerS6 overexpression was restored by Vdac1 deletion (Fig. 7E–G). CerS6 overexpression increased the numbers of CD68-positive cells and the levels of inflammatory cytokines in the glomeruli, which were suppressed by Vdac1 knockout (Fig. 7H, I). Furthermore, the elevated expression of cGAS, STING, and phosphorylated TBK1 and p65 in glomeruli induced by CerS6 overexpression was ameliorated by Vdac1 knockout (Fig. 7J). Taken together, these results demonstrated that CerS6 and VDAC1 cooperatively mediated podocyte injury and inflammation.

Deletion of CerS6 mitigated podocyte injury and proteinuria in ADR mice

Finally, we used a murine model of ADR-induced nephropathy, which is a model of human FSGS, to confirm the broad implications of CerS6 in podocyte injury (Fig. 8A). ADR-treated Cre⁺/CerS6^fl/fl mice exhibited decreased urinary albumin excretion and lower glomerular mesangial expansion than ADR-treated Cre⁺/CerS6^+/+ mice (Fig. 8B, C). TEM analysis revealed less basement membrane thickening, podocyte foot process broadening, and effacement in ADR-treated Cre⁺/CerS6^fl/fl mice than in ADR-treated Cre⁺/CerS6^+/+ mice (Fig. 8D). SEM analysis confirmed these findings (Fig. 8D). In addition, CerS6 deletion reduced ADR-induced podocyte loss, as evidenced by WT-1, synaptopodin, and nephrin staining (Fig. 8E–G). These results demonstrate that CerS6 deletion alleviates ADR-induced podocyte injury.

Fig. 8: Deletion of CerS6 mitigated podocyte injury and proteinuria in ADR mice.

A A schematic diagram showing the construction of the ADR-induced nephropathy model (Created in BioRender. https://BioRender.com/g05h226). B ACR of mice in different groups (n = 6 mice per group). C PAS staining images and quantification of glomerular mesangial matrix expansion in different groups of mice (n = 6 mice per group). Scale bars: 20 μm. D SEM and TEM analyses of podocyte ultrastructure in different groups of mice (n = 6 mice per group). Scale bars: 1 μm. E–G Representative IF images and quantifications of expressions of WT-1 (E), synaptopodin (F), and nephrin (G) in glomeruli from mice (n = 6 mice per group). Scale bars: 20 μm. Data are expressed as mean ± SEM (B–G). Two-way ANOVA followed by Tukey’s post-test (B–G). Source data are provided as a Source Data file.

Discussion

Although glomerular sphingolipid accumulation, mitochondrial damage, and inflammatory responses are the main pathological features of DKD^27,28,29, whether they synergistically promote DKD or whether there is an interplay between these lesions in DKD remains unclear. In this study, we demonstrated that the ceramide synthase CerS6 was primarily localized in the podocytes of glomeruli and was significantly upregulated in two mouse models of glomerular disease. Importantly, podocyte CerS6 levels were also increased in subjects with DKD and other podocytopathies, suggesting that CerS6 is a potential marker for human glomerular disease. In addition, the podocyte-specific CerS6 knockout significantly mitigated glomerular injury and inflammatory responses in DKD and FSGS mouse models, indicating that CerS6 is a key molecule involved in podocyte injury. Mechanistically, the present study confirmed that CerS6-derived Cer (d18:1/16:0) could bind to the mitochondrial channel protein VDAC1 at Glu59 residue, resulting in increased mitochondrial membrane permeability and the initiation of mtDNA leakage, activating the cGAS-STING signaling pathway, and ultimately promoting an immune-inflammatory response in the kidney (Fig. 9). The findings of this study provide a potential link between sphingolipid accumulation and mtDNA-dependent innate immunity. Targeting CerS6 may represent an effective therapeutic approach for the treatment of DKD.

Fig. 9: Schematic depicting the role of CerS6 in ceramide metabolism and innate immune responses.

In diabetic conditions, CerS6-derived ceramide (d18:1/16:0) can bind to the mitochondrial channel protein VDAC1, initiating mtDNA leakage, activating the cGAS-STING signaling pathway, and ultimately promoting an immune-inflammatory response in the kidney. Collectively, we have identified a mechanism by which podocyte sphingolipid accumulation, mitochondrial damage, mtDNA leakage, and inflammatory responses are interconnected via a CerS6-mediated Cer (d18:1/16:0) synthesis pathway, which contributes to podocyte injury and DKD progression. This figure is Created in BioRender. https://BioRender.com/v99r843.

Over the past 20 years, the lipid nephrotoxicity hypothesis has been widely recognized, and ectopic lipid accumulation in the kidney is closely related to the occurrence and development of various kidney diseases^5,8,30,31. Compared with tubular cells, podocytes are extremely sensitive to lipid accumulation⁸. The main reason for this is the low capacity of lipid metabolism. Our previous studies have confirmed that podocytes have low mitochondrial density, especially during the foot process³². Therefore, a limited number of mitochondria cannot metabolize excess fatty acids (FA) under pathological conditions, resulting in FA accumulation and lipid droplet formation in the podocytes³³. In addition, the specialized lipid raft structure of podocytes contributes to their sensitivity to lipid accumulation. Lipid raft structures contain more sphingolipids than the other plasma membrane domains^11,34. Slight sphingolipid alterations in the lipid raft structure can affect glomerular permselectivity and intercellular signal transduction^33,35. Notably, early scRNA-seq studies and recent research have suggested the importance of CerS6, a key enzyme in ceramide metabolism, in podocytes through bioinformatics, but they have not provided the necessary experimental validation^17,18. The present study showed that CerS6 was mainly expressed in podocytes of the glomeruli. Ceramides are the core intermediate metabolites of the sphingolipid metabolic pathway, and pioneering studies have confirmed that ceramide accumulation is closely related to podocyte injury^34,36,37,38. Especially in Fabry’s disease, which is characterized by deficient activity of α-galactosidase, resulting in the accumulation of globotriaosylceramide in the kidney, and its characteristic pathological change was podocyte detachment and proteinuria, which highlighted the role of ceramides disturbance in podocyte injury^39,40. Our study demonstrated that CerS6 is markedly increased in podocytes from patients with DKD and FSGS. More importantly, the podocyte-specific overexpression of CerS6 in mice induces podocyte foot process effacement and proteinuria. However, the mechanism through which CerS6 mediates podocyte injury remains unclear.

In recent years, the pathogenic role of inflammation in DKD has attracted considerable attention, crucial findings were obtained from the clinical use of novel hypoglycemic agents, which found SGLT2i and GLP-1 receptor agonists significantly reduced the inflammation independent of their glucose-lowering actions^41,42,43. We isolated glomeruli from diabetic mice and performed RNA sequencing, which revealed significant changes in pathways related to inflammation and lipid metabolism. Given that CerS6 plays an important role in podocyte injury, we constructed podocyte-specific CerS6 knockout mice and showed that CerS6 depletion significantly attenuated the glomerular inflammatory response and podocyte injury in diabetic mice, suggesting that elevated CerS6 is strongly associated with inflammation in DKD. CerS6 mainly produces C16 long-chain ceramides, an important constituent of eukaryotic cellular membranes, and is involved in altering mitochondrial membrane permeabilization¹². Most importantly, RNA sequencing analysis revealed that membrane-associated molecules in the glomeruli of diabetic mice were significantly altered. Emerging evidence suggests that oligomerization of the mitochondrial outer membrane protein VDAC1 and the transfer of Bax to the mitochondria mediate mitochondrial membrane permeabilization, which enables mtDNA release into the cytoplasm, activates the intracellular DNA receptor cGAS, and subsequently activates STING. Activated STING recruits and activates TBK1, resulting in the expression of proinflammatory cytokines via downstream-coupled nuclear transcriptional factors NF-κB^25,26. Here, we found that CerS6 knockdown significantly mitigated mitochondrial damage, mtDNA leakage, and activation of the cGAS-STING signaling pathway under HG conditions and that overexpression of CerS6 sufficiently induced mtDNA leakage in podocytes. However, the specific mechanism through which CerS6 mediates mtDNA release remains largely unknown. Our results showed that CerS6-dervied Cer (d18:1/16:0) could bind to VDAC1 to form VDAC pores, ultimately leading to mtDNA release. We also preliminary explored the binding site of Cer (d18:1/16:0) to VDAC1, and it is reasonable to identify the Glu59 residue as the key binding site because the negatively charged Glu residue buried in the membrane of VDAC is critical for ceramide binding¹². Therefore, our findings indicate that Cer (d18:1/16:0) accumulation in the mitochondria is a signal for mtDNA release. Recently, another study highlights the vital role of the acyl chain length ceramide-specific regulation of metabolic homeostasis in hypothalamic neurons. CerS6 emerges as a key player in this process, suggesting its involvement in appetite and metabolic rate control during nutritional switches⁴⁴. The study also suggests an important role for CerS6 in metabolic diseases. Diabetes is another important metabolic disease. In our study, we focus on the unique characteristics of CerS6 mainly expressed in podocytes, and its key role in DKD, one of the most important complications of diabetes. Targeting CerS6-induced metabolic inflammation in podocytes holds promise for DKD therapy, offering potential advantages over complete inhibition of whole cells of the kidney with reduced side effects.

This study has several important clinical implications. Recently, the roles of Cers6 and Cer (d18:1/16:0) in metabolic diseases have received significant attention, as CerS6 inhibition significantly ameliorates obesity-associated insulin resistance¹³. Notably, a recent study confirmed that C16-ceramides generate proinflammatory/pro-apoptotic ceramide-rich platforms on retinal endothelial surfaces, which contributes to diabetic retinopathy progression, suggesting the key role of ceramide accumulation in the inflammatory response of diabetic complications⁴⁵. Besides, mitochondrial damage and mtDNA leakage trigger an inflammatory response through the activation of the cGAS-STING pathway, which not only plays an important role in kidney diseases^23,26,46,47, but is also involved in the progression of lupus-like diseases, non-alcoholic fatty liver disease, and aging^25,48,49. Although the targeted inhibition of STING is a promising therapeutic approach, it may reduce innate immunity owing to off-target effects, leading to viral infections and tumor progression⁵⁰. This study demonstrated that CerS6 is mainly expressed in the podocytes of the glomeruli and that CerS6-derived Cer (d18:1/16:0) is an upstream event for mtDNA leakage and STING activation. Therefore, targeting CerS6 may be more effective than countering a single downstream event with fewer off-target effects. In addition, because CerS6 is specifically expressed in podocytes, drug agents targeting CerS6 can act independently of the podocyte-targeted transport system, which reduces the cost and difficulty of drug development. Despite the significant findings of our study, there are several limitations that should be acknowledged. First, our results demonstrated that CerS6 was primarily localized in podocytes and revealed its therapeutic value in treating proteinuric kidney diseases in mouse models, but we were unable to investigate its effects in human subjects. Second, the mechanism by which CerS6 is increased in the diabetic state has not been elucidated. Additionally, the exact mechanism by which Cer (d18:1/16:0) regulates VDAC1 oligomerization remains unclear. Third, some scRNA-seq databases did not show Cgas/Sing expression in human¹⁹ and mouse²⁰ podocytes. It may be the fact that the inability to detect Cgas/Sing expression in podocytes within scRNA-seq databases may stem from the inherent technical and biological limitations of scRNA-seq, despite its utility in providing gene expression data at the cellular level. These limitations likely contribute to the discrepancies between scRNA-seq results and experimental findings. Fourth, this study evaluated the interaction between cGAS and mtDNA using immunofluorescence, particularly in podocytes. However, it may be challenging to obtain significant binding characteristics using this method. In future studies, we will explore and develop additional research methods to further investigate this interaction. Future research should aim to address these knowledge gaps and provide a more comprehensive understanding of the underlying mechanisms.

Overall, this study demonstrated that CerS6 is an important contributor to sphingolipid accumulation, mitochondrial damage, mtDNA leakage, and inflammatory responses in podocytes. Targeted regulation of CerS6 represents an effective therapeutic approach for the treatment and prevention of proteinuric kidney diseases.

Methods

Experimental model and study participant details

Human renal samples

Renal biopsies were performed as part of a routine clinical diagnostic investigation and collected as described in Table S1. Human kidney biopsy samples were obtained from the Division of Nephrology, Renmin Hospital of Wuhan University, and the Department of Pathology, Renmin Hospital of Wuhan University. Control samples were para-carcinoma tissues from individuals without diabetes or kidney disease who underwent tumor nephrectomy. No sex or gender analysis was performed due to the low sample size. This study was conducted following the principles of the Declaration of Helsinki and approved by the Research Ethics Committee of Renmin Hospital of Wuhan University (Approval number: WDRY2023-K040), and informed consent was obtained from all patients.

Mouse studies

All experimental protocols for animal studies were approved by the Committee on the Ethics of Animal Experiments of Renmin Hospital of Wuhan University and were conducted in accordance with institutional guidelines (Approval number: WDRM-20220507B). All mice were housed in temperature-controlled rooms on a 12 h light–dark cycle and allowed unrestricted access to water and standard rodent chow. All cages, bedding, and water were autoclaved, and the cages were changed three times per week. For all studies, mice were randomly assigned to experimental groups and littermates were used. In all of the experiments, considering that female mice are resistant to obesity and type 2 diabetes, only male mice were used.

Generation of podocyte-specific CerS6 knock-in mice

Podocyte-specific CerS6 knock-in mice (NPHS2-Cre/Rosa26-fl-STOP-fl-CerS6; CerS6^po-cKI mice) were generated by crossbreeding Rosa26-fl-STOP-fl-CerS6 mice (Gempharmatech Co. Ltd., Nanjing, China) with B6. Cg-Tg (NPHS2-Cre) 295 Lbh/J mice were obtained from Jackson Laboratory (stock# 008205). Mice with two WT alleles expressing Cre recombinase (NPHS2-Cre/Rosa26-STOP-CerS6 mice; CerS6^ctrl mice) were used as controls. CerS6^po-cKI and age-matched CerS6^ctrl mice were divided into three groups: 2-month-old group, 3-month-old group, and 4-month-old group (n = 6). DNA from homozygous (Rosa26-fl-STOP-fl-CerS6) produced a 499 bp product, WT (Rosa26-STOP-CerS6) type produced a 420 bp product, and heterozygotes (Rosa26-fl-STOP-CerS6) produced two bands, respectively. Cre-positive (Cre⁺) mice produced a 413 bp product, whereas Cre−negative (Cre⁻) produced no product. The primers used for genotyping are listed Table S2.

Generation of podocyte-specific CerS6 knockout mice

Podocyte-specific CerS6 knockout mice (NPHS2-Cre/CerS6^fl/flmice; Cre⁺/CerS6^fl/fl mice) were generated by crossbreeding CerS6-floxed mice (Gempharmatech Co., Ltd., Nanjing, China) with B6. Cg-Tg (NPHS2-cre) 295 Lbh/J mice were obtained from Jackson Laboratory (stock# 008205). Mice with two WT alleles expressing the Cre recombinase (NPHS2-Cre/CerS6^+/+mice; Cre⁺/ CerS6^+/+ mice) served as controls. DNA from homozygotes (CerS6^fl/fl) produced a 361 bp product, WT (CerS6^+/+) produced a 260 bp product, and heterozygotes (CerS6^fl/+) produced two bands. Cre-positive (Cre⁺) produced a 413 bp product, Cre Cre-negative (Cre⁻) had no product. The primers used for genotyping are listed in Table S2.

Generation of podocyte-specific Vdac1 knockout mice

Podocyte-specific Vdac1 knockout mice (NPHS2-Cre/Vdac1^fl/flmice; Cre⁺/Vdac1^fl/fl mice) were generated by crossbreeding Vdac1-floxed mice (Cyagen Biosciences Inc., Suzhou, China) with the B6. Cg-Tg (NPHS2-cre) 295 Lbh/J mice were obtained from Jackson Laboratory (stock# 008205). Exon 4 ~ 5 was selected as conditional knockout region. Mice with two WT alleles expressing the Cre recombinase (NPHS2-Cre/Vdac1^+/+mice; Cre⁺/Vdac1^+/+ mice) served as controls. DNA from homozygous (Vdac1^fl/fl) produced a 417 bp product, WT (Vdac1^+/+) type produced a 350 bp product, and heterozygotes (Vdac1^fl/+) produced two bands, respectively. Cre-positive (Cre⁺) produced a 413 bp product, Cre-negative (Cre⁻) had no product. The primers used for genotyping are listed in Table S2.

Intrarenal adeno-associated virus vector serotype 9 (AAV9) delivery

AAV9-Ctrl and age-matched AAV9-CerS6 mice were divided into three groups: a 2-month-old group, a 3-month-old group, and a 4-month-old group (n = 6). Eight-week-old male Cre⁺/Vdac1^fl/fl mice and their littermate Cre⁺/Vdac1^+/+ mice were randomly divided into 4 groups: Cre⁺/Vdac1^+/+ + AAV9-Ctrl group, Cre⁺/Vdac1^+/+ + AAV9-CerS6 group, Cre⁺/Vdac1^fl/fl + AAV9-Ctrl group, Cre⁺/Vdac1^fl/fl + AAV9-CerS6 group (n  =  6). In vivo overexpression of CerS6 in Cre⁺/Vdac1^fl/fl mice and Cre⁺/CerS6^+/+ mice was achieved by intrarenal AAV9 injection³². The mice were anesthetized with pentobarbital sodium (30 mg/kg, #P3761, Sigma) by intraperitoneal injection, and were injected with 1 × 10¹² vector genome (vg)/mL AAV9-CerS6 (AAV9-CerS6,) or AAV9-CMV-null (AAV9-Ctrl) genomic particles (Hanbio Biotechnology, China) into six different sites (10 μL at each site) of the renal cortex with a glass micropipette. The coding sequence of CerS6 is listed in Supplementary Table 2.

STZ/HFD-induced diabetic mice

Six-week-old male Cre⁺/CerS6^fl/fl mice and their littermate Cre⁺/CerS6^+/+ mice were randomly divided into 4 groups: Cre⁺/CerS6^+/+ + control group, Cre⁺/CerS6^+/+ + STZ/HFD group, Cre⁺/CerS6^fl/fl + control group, Cre⁺/CerS6^fl/fl +  STZ/HFD group (n = 6). The mice were fed an HFD (60 kcal% from fat, Research Diets, #D12492) or a control diet (Research Diets, #D12450J) for 4 weeks. Mice were fasted for 4 h and then intraperitoneally injected with STZ (50 mg/kg body weight daily for 3 days; #S0130, Sigma, USA) to induce partial insulin deficiency. All the mice were allowed unrestricted access to water and food, followed by continued HFD feeding for an additional 12 weeks. After collecting 24 h urine samples and blood, the mice were euthanized, and the kidney samples were harvested for biochemical and pathological analyses.

Spontaneous type 2 diabetic db/db mice

Seven-week-old male db/db mice (n = 6) and matched db/m mice (n = 6) were purchased from CAVENS Laboratory Animals (Jiangsu, China). Blood glucose and proteinuria levels were measured weekly as indicators of the successful establishment of T2D models.

Adriamycin-induced nephropathy in mice

Ten-week-old male Cre⁺/CerS6^fl/fl mice and their littermate Cre⁺/CerS6^+/+ mice were randomly divided into 4 groups: Cre⁺/CerS6^+/+ + control group, Cre⁺/CerS6^+/+ + ADR group, Cre⁺/CerS6^fl/fl + control group, Cre⁺/CerS6^fl/fl + ADR group (n = 6). The ADR nephropathy model for human FSGS was administered ADR (18 mg/kg, #D1515, Sigma) intravenously by tail vein injection⁵¹. 24 h urine samples were collected weekly for biochemical analyses. The mice were euthanized five weeks after treatment, and kidney tissue samples were harvested for biochemical and pathological analysis.

Cell culture

Conditionally immortalized human podocytes (HPC) were kindly gifted by Dr. Moin A. Saleem (Academic Renal Unit, Southmead Hospital, Bristol, UK). This cell line was developed via transfection with the temperature-sensitive SV40-T gene, allowing the cells to proliferate at a “permissive” temperature (33 °C). Upon transfer to the “non-permissive” temperature (37 °C), the cells undergo growth arrest and differentiation⁵². Podocytes were cultured in RPMI 1640 medium (Gibco, USA) containing 10% fetal bovine serum (FBS; Gibco, USA), 100 μg/mL streptomycin (Invitrogen, USA), and 1× insulin-transferrin-selenium (ITS; Invitrogen, USA) at 33 °C. Differentiation was induced by culturing the cells at 37 °C for 14 days without ITS³². Differentiated podocytes were exposed to high glucose (HG, a final concentration of 20 or 40 mmol/L in culture medium) or ADR (a final concentration of 0.2–0.8 μg/mL in culture medium)^8,51. For the VBIT-4 experiments, podocytes were pretreated with 10 µM VBIT-4 (#T13287, Targetmol, USA). CerS6 siRNA (Sangon, China) was transfected into podocytes using HiPerFect (Qiagen, Germany), according to the manufacturer’s protocol. Overexpression of CerS6 by a CerS6-lentivirus transfection was produced by Hanbio Biotechnology (Shanghai, China). pcDNA3.1-VDAC1, pcDNA3.1-VDAC1^T42Q, pcDNA3.1-VDAC1^E59Q, and pcDNA3.1-VDAC1^K61Q plasmid (Paivibio, China) was transfected into podocytes using X‑tremeGENETM Transfection Reagent (Roche Diagnostics) according to the manufacturer’s instructions. Sequences of oligonucleotides targeting CerS6, Coding sequence of CerS6, and primers used for site-directed mutagenesis are listed in Supplementary Table 2.

Western blotting

Total tissue and cell proteins were extracted from RIPA buffer containing a protease inhibitor cocktail (#P8340, Sigma), phenylmethylsulfonyl fluoride (#ST505, Beyotime, China), and phosphatase inhibitor (P1081, Beyotime, China). Proteins were separated by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride (PVDF) membranes. The PVDF membrane was blocked in 5% milk for one hour and incubated with primary antibody at 4°C overnight. The secondary antibody used was horseradish peroxidase (HRP)-labeled goat anti-rabbit/mouse IgG. The details of the antibodies used are listed in Table S3. The blots were visualized using a Monad imaging system (Wuhan, China) and quantified using ImageJ software.

Single-cell gene expression profiling

Recently published scRNA-seq data have been used to analyze the expression of CerS6 in mouse and human kidneys^19,20,21. Data normalization, dimensionality reduction, clustering, and differential expression analyses were performed using Interactive Kidney Transcriptomics (http://humphreyslab.com/SingleCell/).

Histological analysis of renal tissues

Mouse kidney samples were fixed with 4% paraformaldehyde (PFA) at 4°C overnight, and then embedded in paraffin. 4 μm transverse sections were cut for pathological analysis under the blind method. Hematoxylin and eosin (H&E), Periodic Acid-Schiff (PAS), Masson’s trichrome, and Sirius Red staining were performed according to standard histological protocols. An Olympus microscope (Tokyo, Japan) was used to photograph at least 10 glomeruli per mouse at 40× or 90× magnification. The positive area of the glomeruli was quantified using ImageJ software (National Institutes of Health, USA).

Immunofluorescence staining

Paraffin‐embedded kidney sections (4 μm) were deparaffinized and antigen retrieval, which was performed under high pressure in citrate buffer (0.01 mol/L, pH 6.0, Beyotime, China) for 10 min. In vitro, the cell-climbing pieces were fixed with 4% PFA. In Figs. 5H, 6I and S10C, the fix and permeabilization was performed by incubating podocytes in acetone and methanol at −20°C for 20 min, 1% BSA was added to block the cells for 1 h at room temperature²⁶. Then, the sections were incubated with different primary antibodies at 4 °C overnight, followed by secondary antibodies incubated for 2 h away from light. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI, Beyotime, China). The details of the antibodies used are listed in Table S3. An Olympus microscope (Japan) was used to photograph at least five glomeruli per human subject or 10 glomeruli per mouse at 40× or 90× magnification. The positive area of the glomeruli was quantified using ImageJ software (National Institutes of Health, USA).

Scanning electron microscopy

Renal cortex was fixed with 3% glutaraldehyde plus 2% paraformaldehyde overnight at 4 °C. The renal cortex was washed three times with 0.1 M phosphate buffer (PB, pH 7.4), and then postfixed with 1% osmium tetroxide in 0.1 M PB (pH 7.4) for 1 h at room temperature, and the samples were washed three times again. The samples were dehydrated with a gradually increasing concentration series of ethanol, each time for 5 min. Then, the samples were placed in a critical point dryer (K850, Quorum, UK) for drying. The samples were attached to metallic stubs using carbon stickers and sputter-coated with gold for 30 s. The samples were examined in a Scanning Electron Microscope (SU8100, Hitachi, Japan). Five glomeruli were obtained from each mouse, and ten micrographs were obtained from each glomerulus.

Transmission electron microscopy

The renal cortex was fixed with 2.5% glutaraldehyde overnight at 4°C. Electron microscopy sample handling and detection were performed at the Electron Microscopy Center of Renmin Hospital of Wuhan University. Five glomeruli were obtained from each mouse, and ten micrographs were obtained from each glomerulus. Micrographs were analyzed using the ImageJ software (National Institutes of Health, USA). GBM thickness, foot process width, and number of foot processes per micrometer of GBM were calculated using a curvimeter (SAKURAI Co., LTD, Japan).

Isolation of glomeruli

The glomeruli were isolated using a modified sieving method^32,53. Before removing the kidney, it was perfused with sterile Hanks’ balanced salt solution (HBSS). The kidney was cut into 1mm³ pieces with a scalpel and then digested in a solution containing collagenase II (#10103586001, Roche), proteinase E (#P6911, Sigma), and deoxyribonuclease I (#D4527, Sigma) at 37 °C for 30 min. After digestion, the samples were sequentially pressed through a 100-µm cell strainer (BD, USA), 70-µm cell strainer (BD, USA), and 40-µm cell strainer (Greiner bio-one, German). The glomeruli are collected in the inner layer of the 40-um cell strainer. Finally, the cell suspension was centrifuged at 200 × g for 5 min at 4 °C, and the pellet was re-suspended in a 5 mL culture medium.

Isolation of primary podocytes

Primary podocytes were isolated from a double-fluorescent Cre reporter mouse^32,53. Glomeruli were harvested from mT/mG/NPHS2-Cre mice at two weeks of age, which was generated by crossbreeding with B6 mice. Cg-Tg (NPHS2-Cre) 295 Lbh/J mice (Jackson Laboratory; stock# 008205) and Gt (ROSA)26^{Sortm4(ACTB-tdTomato, -EGFP) Luo/J} mice (Jackson Laboratory; stock# 007676), then cultivated in standard podocyte culture medium (RPMI 1640 medium containing 10% fetal bovine serum), 100 μg/mL streptomycin, and 1× ITS. After 5 days of expansion, flow cytometry (BD, USA) was used to isolate the green fluorescent protein (GFP)–positive cells (primary podocytes) and negative cells (other cells), and cultured at 37 °C in 5% CO₂ condition.

RNA sequencing and transcriptomic analyses

Isolated glomeruli from 24-week-old db/db and db/m mice were collected, immersed in TRIZOL buffer (#15596018CN, Invitrogen, USA), quick-frozen, and stored in liquid nitrogen. RNA-Seq analysis was performed by Myhalic Biotechnological Co., Ltd. (Wuhan, China). Differentially expressed genes (DEGs) were identified using DESeq2 (adjusted p < 0.05). Heatmaps were generated using the TB tools. Gene ontology (GO) enrichment and pathway enrichment analyses were performed using Metascape (https://metascape.org).

Isolation and quantification of urinary mtDNA

Urine samples were collected as a part of routine clinical diagnostic investigations, as shown in Table S1. Urinary mtDNA was quantified by Real-Time PCR⁵⁴. Protease inhibitors were added to urine samples. Urine was centrifuged at 1000 × g to remove intact cells and cellular debris to exclude mtDNA contained in cells released into the urine, and the supernatant was collected and stored at −80 °C until analysis. Urine DNA was isolated and purified using the DNeasy Blood and Tissue KIT (#69506, Qiagen, GER). The mtDNA content in urine was determined by quantitative PCR (qPCR), which was performed using SYBR Green PCR Master Mix (#Q111-02, Vazyme, Nanjing, China) and a Real-Time PCR System (Bio-Rad, USA). The copy number of the mitochondrial gene ND1(mt-ND1) was normalized to that of the nuclear DNA nuclear gene β-actin. Primers used are listed in Table S2.

Detection of mtDNA in cytosolic extracts

Cytosolic mtDNA in the cytosolic fraction was measured by cell compartment fractionation followed by Real-Time PCR⁴⁶. Podocytes were lysed in NP-40 buffer (#C3228, Sigma) and incubated on ice for 15 min, followed by centrifugation at 100,000 × g at 4 °C for 15 min. Cytosolic mitochondrial DNA was isolated from the supernatant cytosolic fraction using the DNeasy Blood & Tissue KIT according to the manufacturer’s instructions. The mtDNA content in the cytosolic fraction was determined by quantitative PCR (qPCR) using the SYBR Green PCR Master Mix and a Real-Time PCR System (Bio-Rad, USA). The copy number of the mitochondrial gene ND1(mt-ND1) was normalized to the nuclear DNA encoding 18S ribosomal RNA (18S rRNA). Primers used are listed in Table S2.

RNA extraction and real-time RT-PCR

The TRIzol reagent was used to isolate Total RNA according to the manufacturer’s instructions. 1 μg of total RNA was reverse transcribed using a cDNA synthesis kit (#G3331-50, Servicebio, China). The mRNA expression levels were detected by qPCR, which was performed using SYBR Green PCR Master Mix and a Real-Time PCR System (Bio-Rad, USA). Primers used are listed in Table S2.

Measurements of oxygen consumption rate (OCR)

The OCR was measured by Seahorse Bioscience XFe24 Extracellular Flux Analyzers according to the manufacturer’s instructions (Seahorse Bioscience, USA). The data was automatically calculated, recorded, and plotted using XF24 software version 1.8 (Seahorse Bioscience).

Lipidomic analysis of ceramides in kidney, podocytes, and mitochondria

The mitochondria of podocytes were extracted using a mitochondrial isolation kit (#C3601, Beyotime, China) according to the manufacturer’s instructions. The kidney, podocytes, and isolated mitochondria were analyzed by lipidomic analysis of ceramide. Analysis of ceramides was conducted at LipidALL Technologies⁵⁵. Internal standards including d7-Cer d18:1/24:0 and d7-Cer d18:1/15:0 were added together with extraction solvent comprising ethyl acetate: isopropanol = 2:8 (v/v) into the samples. Samples were incubated at 1000 × g, 4 °C for 10 min, and centrifuged at 4 °C, 3000 × g for 10 min. clean supernatant was used for LC-MS/MS analysis. Samples were analyzed on a Shimadzu Nexera 30-AD HPLC coupled with Sciex TRIPLE QUAD 6500 PLUS under electrospray ionization mode. A Waters ACQUITY UPLC BEH C18 column (2.1 × 100 mm, 1.7 μm) (Waters, Dublin, Ireland) was used for the chromatographic separation of individual ceramides. Ion source settings were: Curtain gas, 20; positive ion mode ion spray voltage, 4500 V; temperature, 450 °C; ion source gas 1, 80; ion source gas 2, 70. Individual lipids were quantitated by referencing spiked internal standards. Data was analyzed using GraphPad Prism 9.0 (GraphPad Software, USA). n = 6 (Control group) and n = 6 (STZ/HFD group) independent samples. n = 6 (Control group) and n = 6 (ADR group) independent samples. n = 5 (Control group) and n = 5 (HG group) independent samples. n = 6 (Cre⁺/CerS6^fl/fl group) and n = 6 (Cre⁺/CerS6⁺/⁺ group) independent samples. n = 5 (si-NC group) and n = 5 (si-CerS6 group group) independent samples. n = 5 (LV-Ctrl group) and n = 5 (LV-CerS6 group group) independent samples. n = 6 (CerS6^Ctrl group) and n = 6 (CerS6^po-cKI group) independent samples. n = 6 (AAV9-Ctrl group) and n = 6 (AAV-CerS6 group) independent samples.

Mitochondrial membrane potential (MMP), reactive oxygen species (mtROS), and adenosine triphosphate (ATP) analysis

Podocytes were cultured in 6-well plates. According to the manufacturer’s instructions: (1) An enhanced mitochondrial membrane potential assay kit with JC-1 (#C2003S, Beyotime) was used to evaluate the MMP. (2) The MitoSOX Red Mitochondrial Superoxide Indicator (#M36007, Invitrogen, USA) was used to measure mtROS. MMP and mtROS were visualized using an Olympus inverted microscope and analyzed using ImageJ software. (3) An ATP Assay Kit (#S0027, Beyotime, China) was used to measure ATP levels, which were quantified using a microplate reader.

MitoTracker staining

The mitochondrial morphology was visualized by MitoTracker Red CMXRos staining (#M7512, Invitrogen, USA) according to the manufacturer’s instructions. The fixed cell-climbing pieces were incubated with 50 nm MitoTracker Red dye at 37 °C for 30 min. All images were obtained using an Olympus confocal microscope.

Computational docking

Cer (d18:1/16:0) was docked to the VDAC1 protein using Autodock Vina 1.1.2 (Center for Computational Structural Biology, CA) according to the protocol of computational docking provided by the AutoDock suite. PyMol 2.3.0 (Schr Odinger, LLC) was used to visualize the interaction modes of the docking results.

MD simulations

Utilize the GROMACS 2020.6 to undertake a 100 ns molecular dynamics (MD) simulation in order to further verify the rationality and reliability of the docking results. Employ the Amber03 force field and the Amber GAFF force field respectively to generate the parameters and topology files of the protein and small molecule ligands. To render the simulation system electrically neutral, substitute some of the solvent water molecules with Na⁺ and Cl⁻ at a concentration of 0.15 mol/L. Utilize the steepest descent method to minimize the energy consumption of the entire system, and ultimately reduce the unreasonable contacts or atomic overlaps within the entire system. The pre-balance is conducted in two phases, using the NVT system under the conditions of 300 K and 100 ps to simulate the first-phase equilibrium in order to stabilize the temperature of the system, and employing the NPT system under the conditions of 1 bar and 100 ps to simulate the second-phase equilibrium in order to stabilize the pressure of the system. The leapfrog algorithm is utilized for the equilibrium dynamic integration, and all MD simulations are performed under isothermal and isobaric conditions with a temperature of 300 K and a pressure of 1 bar, for a duration of 100 ns. After the dynamic simulation is completed, the trajectory is corrected with the protein as the center. Calculate the average conformation of the protein during the simulation, and draw a comparison diagram of the protein structure before and after the dynamic simulation.

Quantification and statistical analysis

All experiments were performed at least thrice and data were represented as the mean ± SD. For statistical analysis, data distribution was assumed to be normal and was analyzed using GraphPad Prism 9.0 (GraphPad Software, USA). Student’s t-test (unpaired, two-tailed) was used to compare differences between two groups. One-way ANOVA followed by post hoc Tukey’s test was used to analyze differences between multiple groups with one variable. Two-way ANOVA followed by post hoc Tukey’s test was used to compare multiple groups with more than one variable. For data with a non-Gaussian distribution, we performed a nonparametric statistical analysis using the Kruskal–Wallis test followed by Dunn’s post-hot test for multiple comparisons Spearman’s correlation coefficient assay was performed to analyze the correlations between two variables. p < 0.05 was considered to be statistically significant. The sample sizes are provided in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.