Deficiency of Mitochondrial Glycerol 3-Phosphate Dehydrogenase Exacerbates Podocyte Injury and the Progression of Diabetic Kidney Disease

Diabetic kidney disease (DKD; also known as diabetic nephropathy) is the leading cause of end-stage renal disease worldwide, as more than half of all patients requiring dialysis have diabetes (1). Compared with other vascular complications, DKD exhibits the strongest correlation with mortality in patients with diabetes (2), leading to a large and growing challenge for patients, health systems, and governments (3). Although early intervention with optimal glucose and blood pressure control to prevent major adverse kidney outcomes is currently recommended for DKD treatment, the residual risk of adverse renal and cardiovascular outcomes remains high (4); therefore, it is critically important that the pathogenesis of DKD be understood to identify reliable biomarkers to stratify high-risk patients before onset and enable the development of novel therapeutic approaches to either stop or reverse DKD progression.

DKD is characterized by a compromised glomerular filtration barrier (GFB), leading to proteinuric kidney diseases associated with significant amounts of protein in the urine (5,6), mainly resulting from the injury or functional impairment of any of three components of the GFB (i.e., glomerular endothelial cells, basement membrane, or podocytes) (7). Podocytes have received special interest because they are terminally differentiated epithelial cells and have a limited ability to repair and regenerate, indicating that their loss and injury represent an irreversible step. Podocytes physiologically form interdigitated foot processes constituting the crucial component of the GFB for protein leakage; thus, podocyte loss and damage are common features of many glomerular diseases and are highly correlated with proteinuria (8), making podocyte dysfunction one of the earliest glomerular histological changes that occur in DKD.

Glomerular cells contain abundant mitochondria due to their continuously high levels of energy consumption (9). However, metabolic alterations, such as mitochondrial dysfunction and compromised energy metabolism, usually occur during the early stages of renal diseases, as they disrupt mitochondrial homeostasis (10). This is accompanied by a range of changes in mitochondria, including fragmented morphology, elevated production of reactive oxygen species (ROS), and loss of mitochondrial membrane potential (MMP), which contribute to podocyte loss and detachment and foot process effacement, resulting in GFB destruction and proteinuria (11).

Mitochondrial glycerol 3-phosphate dehydrogenase (mGPDH; encoded by the GPD2 gene) is an integral component of the mammalian respiratory chain located in the inner mitochondrial membrane. mGPDH is abundantly expressed in multiple tissues and organs, and previous studies have demonstrated its regulatory effects on pancreatic β-cells (12) and thyroid cancer cells (13), as well as its involvement in inflammatory responses (14). Recently, we reported a cardinal role for mGPDH in hepatic lipid homeostasis (15) and muscle repair (16). During our study on skeletal muscle regeneration in the context of diabetic conditions, a marked change in the urea albumin-to-creatinine ratio (UACR) was observed in response to mGPDH gene alterations, indicating possible relationships between mGPDH and DKD. Currently, few reports have examined the role of mGPDH in the kidney; therefore, the physiological role of mGPDH in renal cell function, as well as in the development and progression of glomerular disease, is of great interest.

In this study, we revealed a novel role for mGPDH in initiating pleiotropic protective actions by dampening the receptor for advanced glycation end product (RAGE) signaling pathway in podocytes. Podocyte-specific mGPDH-knockout (KO) mice with diabetes or adriamycin (ADR) nephrotoxicity developed exacerbated albuminuria and glomerular pathology. Conversely, restoration of mGPDH expression in diseased glomeruli resulted in renoprotection and attenuation of mitochondrial dysfunction, indicating that mGPDH may be a potential therapeutic target for ameliorating DKD.

Human renal tissues were collected by renal biopsy according to protocols approved by the Ethics Committee of the Second Affiliated Hospital of Army Medical University (Chongqing, China; Institutional Review Board–approved protocol number 2016-056-01; clinical trial register number ChiCTR-ROC-17010719). Written consent was obtained from each patient, and their clinical parameters were extracted from medical records. Kidney tissue was formalin fixed and paraffin embedded, and sections were then cut at 3 μm and stained with periodic acid Schiff (PAS) for histopathology assessment according to the Renal Pathology Society’s Pathologic Classification of Diabetic Nephropathy. All procedures were consistent with the principles of the Declaration of Helsinki. The clinical characteristics of patients with DKD and healthy subjects are presented in Supplementary Table 1.

All mouse studies were conducted according to protocols approved by the Laboratory Animal Welfare and Ethics Committee of the Army Medical University (AMUWEC2020073). Male C57BL/6, db/m, db/db (C57BLKs/J-leprdb/leprdb), BALB/c, and mGPDH^flox/flox mice were purchased from the Model Animal Research Center of Nanjing University. NPHS2-Cre mice were purchased from Cyagen Biosciences. Mice were housed under a 12-h dark/light cycle at 22°C with ad libitum access to food and water. Type 1 diabetes in mice was induced as previously described, and mice were euthanized after 24 weeks (17,18). To evaluate the effects of mGPDH rescue, two diabetic mouse models and ADR mice were transduced with AAV9-NPHS1-mGPDH and control vectors AAV9-NPHS1-GFP (2 × 10¹¹ vector genomes; Genomeditech) via tail vein injection. Mice were sacrificed 4 weeks later.

Urea albumin and creatinine were detected by commercial assay kits (Jiancheng Biochemical Co.) according to the manufacturer’s protocol. UACR was calculated as follows: urea albumin/urea creatinine.

Immunohistochemistry, PAS, and Masson staining were performed as previously described (17). A semiquantitative score (sclerosis index) was assessed on PAS-stained tissue sections and calculated to assess the extent of glomerular sclerosis as previously described (18), and 20 fields and 51–94 glomeruli/group were assessed to obtain the average sclerosis index score. The interstitial fibrosis score was assessed on Masson-stained sections according to previously reported methods (19).

Kidney cortical tissue samples were fixed in 2.5% glutaraldehyde, and transmission electron microscopy (TEM) was performed using a Hitachi TEM system and analyzed using ImageJ (National Institutes of Health). Corresponding parameters were calculated as previously described (20).

Immunofluorescence staining was performed as described previously (16). The following were used as primary antibodies: mGPDH (Abcam), synaptopodin (Proteintech), CD31 (Abcam), α-smooth muscle actin (α-SMA) (Abcam), and Wilm tumor-1 (WT1) (Abcam).

Mouse glomeruli were isolated using magnetic Dynabeads (14013; Invitrogen) as previously described (18). Primary podocytes were further isolated by culturing glomeruli on precoated collagen I dishes for 5 days, trypsinized, and filtered through a 40-μm cell strainer.

Quantitative real-time PCR (qRT-PCR) was performed according to our previous protocol (15). Primer sequences are presented in Supplementary Table 2.

ROS generation was assessed by MitoSOX Red staining or fluorescence of a dichlorofluorescein (DCF) probe (Invitrogen) according to our previous descriptions (15).

Total RNA of isolated glomeruli samples from podocyte-specific mGPDH KO and littermate wild-type (WT) mice was harvested. RNA integrity and genomic DNA contamination were assessed by denaturing agarose gel electrophoresis and an Agilent 2100 Nano (Agilent Technologies). RNA sequencing (RNA-Seq) was performed by Majorbio Bio-pharm Technology Co., Ltd. (Shanghai, China) using the Illumina platform. Data were analyzed using the free online platform I-Sanger.

Serum concentrations of AGEs, advanced oxidation end products (AOPPs), and S100 were measured using commercial ELISA kits (Abcam, Nanjing Shenbeijia Biotechnology Co, Ltd, and Westang Bio-Tech Co Ltd., respectively) according to the manufacturer’s instructions.

Mouse podocytes were purchased from BNBIO (BNCC342021), and their identification was confirmed by the same company. Podocytes were cultured in RPMI 1640 (Gibco) with 10% FBS (ExCell Bio) in a 5% CO₂ incubator at 33°C and were then thermoshifted to 37°C for 10 days before experiments. Passages 12 to 18 were used for experiments. Mycoplasma determination was performed by Shanghai Biowing Applied Biotechnology Co., Ltd. For gene knockdown and overexpression experiments, podocytes were seeded into six-well plates and transfected with siRNA oligonucleotides or plasmid. mGPDH-, RAGE-, s100A10-, sirtuin 5 (SIRT5)–, CPT1A-, and Annexin A2–specific and corresponding negative control siRNAs were purchased from Qiagen or RiboBio. mGPDH and control vector plasmids were generated by GenScript.

The oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of podocytes were assessed by an Agilent Seahorse XF extracellular flux analyzer (Agilent Technologies) according to our previous description (16). Briefly, podocytes were seeded into cell culture plates (Agilent Technologies). Data were calculated from three independent measurements obtained prior to or after compound injection.

MMP was measured using the JC-1 mitochondrial potential sensor (Invitrogen) according to the manufacturer’s protocols. Podocytes were harvested and incubated with JC-1 solution, and the fluorescence of JC-1 monomers and J-aggregates was analyzed by flow cytometry. NAD⁺ content was measured using an NAD⁺ Assay Kit (Beyotime Biotechnology) according to the manufacturer’s recommendation.

For Western blotting, isolated glomeruli or culture podocytes were harvested and lysed as previously described (15). The following antibodies were used: mGPDH (ab188585), fibronectin (Fn) (ab2413), high-mobility group box 1 (HMGB1) (ab18256), S100 (ab4066), and RAGE (ab3611) from Abcam; and S100A4 (16105), S100A10 (11250), S100B (15146), tubulin (11224), SIRT5 (15122), Annexin A2 (11256), and CPT1A (15184) from Proteintech. For immunoprecipitation, lysates were incubated with anti-S100A10 antibody and analyzed by immunoblotting with an antibody against succinyl lysine (PTM BIO).

Data are presented as the means ± SEM unless otherwise indicated. Statistical analysis was performed using GraphPad Prism 8 for Macintosh. For comparisons between two groups, unpaired Student t tests with no assumption of equal variance were used. For comparisons of more than two groups, one-way or two-way ANOVA was used. When overall F tests were significant (P < 0.05), Tukey post hoc testing was conducted for adjustment. For OCR and ECAR experiments, two-way repeated-measures ANOVA with Tukey post hoc testing was used. A two-sided P value <0.05 was considered significant.

All data related to our conclusions in this article are presented in the main text and/or the supplemental materials. Additional data related to this manuscript can be requested from the authors.

In an initial histochemical survey, we observed high levels of mGPDH conserved in the glomerular area in the mouse kidney (Fig. 1A), and renal tubular expression was less prominent. To investigate the possible role of mGPDH in diabetic kidneys, we assessed mGPDH expression in mouse models of type 1 diabetes (streptozotocin [STZ]–induced diabetic mice) and type 2 diabetes (db/db mice). qRT-PCR analysis and Western blot analysis revealed repressed mGPDH expression in glomeruli isolated from both diabetic mouse models relative to the corresponding control, with an inverse correlation with induction of Fn, a marker of renal fibrosis (21) (Fig. 1B and C). Moreover, to reveal which glomerular cell types are responsible for the downregulation of mGPDH, we performed double immunofluorescence staining of mGPDH and the podocyte marker synaptopodin, endothelial cell marker CD31, and mesangial cell marker α-SMA. There was a significant reduction in mGPDH fluorescence that overlapped well with the signal from the synaptopodin in the glomeruli from both DKD mouse models (assessment of the PAS-positive area) compared with those of control mice (Fig. 1D). In contrast, little mGPDH fluorescence colocalized with CD31 or α-SMA fluorescence (Supplementary Fig. 1). Remarkably, diminished staining of mGPDH was confirmed in the podocytes of patients with DKD when compared with normal kidney subjects (ChiCTR-ROC-17010719, Fig. 1E and Supplementary Table 1). Taken together, the podocyte-dominated downregulation of mGPDH observed in mice and patients with DKD indicates the potential involvement of mGPDH in podocyte function and diabetic kidney progression.

To explore the impact of downregulated mGPDH in podocytes during diabetes, we generated podocyte-specific mGPDH KO mice by breeding previously described mGPDH conditional allele (mGPDH^fl/fl) mice (15) with podocin-Cre mice that express Cre recombinase in podocytes under the control of the human podocin NPHS2 promoter (18,22) (hereafter referred to as KO mice) (Supplementary Fig. 2A). Excision of mGPDH from podocytes of KO mice was confirmed by immunofluorescence staining of kidney sections (Supplementary Fig. 2B) and by qRT-PCR and immunoblotting analysis of isolated glomeruli (Supplementary Fig. 2C and D). KO mice did not differ in fasting blood glucose compared with littermate WT mice in either nondiabetic or STZ-induced diabetic (17,18) conditions (Fig. 2A), whereas UACR was significantly increased in KO mice with diabetes compared with WT mice with diabetes (Fig. 2B). Histological analysis showed greater mesangial expansion and interstitial fibrosis (Fig. 2C) as well as increased extracellular matrix (ECM)–related gene expression in diabetic KO mice than in diabetic WT mice (Fig. 2D). Likewise, quantification of podocytes by WT1 protein expression demonstrated a notable loss of podocytes in KO mice compared with WT mice under diabetic conditions (Fig. 2E). Consistent with these observations, TEM images confirmed that KO mice experienced more severe podocyte injury, as evidenced by glomerular basement membrane (GBM) thickening, podocyte foot process broadening, and effacement in the context of DKD (Fig. 2F).

Mounting evidence suggests that experimental and clinical models of DKD are characterized by altered mitochondrial morphology, biogenesis, and ROS generation (23). We next examined whether the significant histopathologic consequences elicited by disruption of podocyte-specific mGPDH expression under diabetes were associated with mitochondrial function. As shown in Fig. 2G, mitochondrial ultrastructure and the corresponding quantitative analysis, including aspect ratio and form factor with smaller values, indicated less elongated, small punctate, and rounded mitochondria from KO mice with diabetes. In keeping with this, reduced mtDNA content, MitoTracker Green fluorescence, MMP, repressed gene expression of oxidative phosphorylation (OXPHOS) and mitochondrial biogenesis manipulator Ppargc1a, as well as induction of mitochondrial ROS as measured by MitoSOX staining were observed in KO mice with diabetes compared with WT mice with diabetes (Fig. 2H–M). Collectively, these results suggest that genetic ablation of mGPDH from podocytes sensitizes mice to mitochondrial dysfunction and glomerular disease under diabetes.

To confirm the role of mGPDH in maintaining podocyte function, we also exposed mice to the podocyte toxin ADR, a model of nondiabetic proteinuria that is commonly used to mimic focal segmental glomerulosclerosis (24). Immunofluorescence staining was performed, and mRNA and protein levels of mGPDH were reduced in kidney sections and glomeruli isolated from ADR-treated mice, respectively (Supplementary Fig. 3), consistent with the observation under diabetic circumstances. Together, these findings suggest that decreased mGPDH expression may contribute to podocyte injury in proteinuric kidney diseases. We next induced ADR nephropathy in mGPDH KO and WT mice and observed a similar phenotype: KO mice exhibited significantly increased UACR, GBM thickening, glomerulosclerosis, and podocyte injury, as well as impaired mitochondrial function, compared with ADR-treated WT mice (Fig. 3). These results suggest that inhibition of mGPDH also aggravates proteinuria, podocyte damage, and glomerular disease in ADR mice.

We next evaluated the molecular mechanism underlying the glomerular injury by podocyte deletion of mGPDH through RNA-Seq analysis. Comparison of glomeruli isolated from diabetic mGPDH KO and WT groups identified 1,327 statistically significant differentially expressed genes (769 upregulated genes and 558 downregulated genes; log₂ fold change >2.0; P < 0.05) (Supplementary Table 3). Hierarchical clustering of the 30 most highly upregulated and 30 downregulated genes showed distinct patterns between these two groups (Fig. 4A). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed that fatty acid degradation and peroxisomes were the top two items downregulated by mGPDH deficiency (Fig. 4B and Supplementary Table 4). The primary process of the former is fatty acid β-oxidation, which facilitates OXPHOS in mitochondria and ATP production (25), while the latter is involved in regulating redox signaling pathways in mitochondrial homeostasis (26). Importantly, a subset of genes involved in the AGE–RAGE signaling pathway in diabetic complications exhibited the highest significance in the upregulated gene set in response to mGPDH deficiency (Fig. 4C and Supplementary Table 4). RAGE is a signal transduction receptor, and its activation via multiple ligands is proinflammatory, especially in the kidney, as RAGE is expressed in renal cells, such as podocytes and endothelial cells (27). Our biological function analysis supported this concept by showing that gene ontology terms associated with “inflammatory response” and “response to cytokine” were notably altered (Supplementary Fig. 4 and Supplementary Table 5). In this case, we tested and confirmed that RAGE is overexpressed in podocytes in diabetic mGPDH KO mice compared with diabetic WT mice (Fig. 4D and Supplementary Fig. 5). Additionally, mGPDH regulated RAGE and its downstream inflammatory markers (p65, CCL2, interleukin-1β [IL-1β], IL-6, and tumor necrosis factor-α) in vivo and in cultured mouse podocytes with transient mGPDH transfection (Fig. 4E and F).

Next, we addressed the relationship between mGPDH and RAGE in the regulation of podocyte function. Consistent with our in vivo results, under high glucose (HG) treatment, mGPDH knockdown in cultured mouse podocytes resulted in increased DNA fragmentation and decreased podocyte-specific protein synaptopodin and nephrin compared with the siRNA control (Si-ctrl) group (Fig. 4G and H). We also observed the induction of genes encoding inflammatory responses (CCL2 and IL-1β) and ECM proteins (Fn and Col1) in response to siRNA targeting mGPDH (Fig. 4I). In addition, for the mitochondrial functional assay, reduced OCR, ECAR, mitochondrial mass, MMP, mRNA expression of OXPHOS and Ppargc1a, and induced intracellular ROS generation, as measured by staining with DCF fluorescence dye, were assessed in the Si-mGPDH group (Fig. 4J–P and Supplementary Fig. 6). RAGE siRNA negated the above effects induced by mGPDH knockdown (Fig. 4G–P and Supplementary Fig. 6). These bioinformatics analyses indicate that inhibition of the RAGE pathway rescues the functional defect in mGPDH-deficient podocytes.

Given that previous studies have reported the beneficial effects of RAGE deletion in delaying the progression of diabetic renal disease (28), we further confirmed the hypothesis that the RAGE pathway might be involved in mGPDH-regulated effects in vivo. Diabetic mGPDH KO and WT mice were injected in the tail vein with a recombinant adeno-associated virus 9 vector containing the podocyte marker NPHS1 promoter (i.e., AAV9-NPHS1-shRAGE), which is a gene therapy approach to primarily transduce cells within the glomerulus (29,30). In accordance with the in vitro data, knockdown of RAGE abolished the enhanced albuminuria, glomerular pathology, podocyte injury, and mitochondrial dysfunction observed in mGPDH KO mice (Fig. 5), suggesting that mGPDH deletion accelerates the progression of glomerular disease via RAGE activation.

Subsequently, we further elucidated how mGPDH regulates RAGE signaling. RAGE and its ligands, AGEs, AOPPs, proinflammatory S100/calgranulins, and HMGB1 are well recognized in functional studies of DKD (29,31). Nevertheless, we did not detect obvious alterations in AGE or AOPP concentrations or HMGB1 expression between mGPDH KO and WT mice under diabetic conditions (Supplementary Fig. 7). However, serum concentrations of S100 were increased in diabetic KO mice (Fig. 6A), and this induction was further confirmed by S100 protein expression in isolated glomeruli (Supplementary Fig. 7C). The S100 protein family is composed of 21 members with high degrees of structural similarity but independent functions (32), and previous studies have shown that they participate in biological processes, such as inflammation, differentiation, and energy metabolism (33). Although several members of the S100 protein have been implicated in the pathophysiology of DKD, only three were proven to be expressed in podocytes (i.e., S100A4, S100A10, and S100B) (34–36). Therefore, we first assessed whether mGPDH influences expression of these three proteins, and our results indicated that only S100A10 responded to mGPDH loss- and gain-of-function manipulations in cultured podocytes (Fig. 6B).

To further determine whether mGPDH exerts an effect through S100A10, siRNAs targeting mGPDH and S100A10 were cotransfected into cultured mouse podocytes. S100A10 elimination prevented augmented podocyte loss, genes encoding RAGE, the inflammatory response, and ECM production caused by mGPDH deletion in response to HG treatment (Fig. 6C–E). Moreover, the reduced OCR, mitochondrial mass, MMP, OXPHOS genes, and enhanced ROS were abolished by S100A10 knockdown (Fig. 6F–L and Supplementary Fig. 8). These results indicate that S100A10 is necessary for the downstream effects of mGPDH deletion.

As a member of the S100 protein family and a RAGE ligand (37,38), S100A10 was previously reported to be involved in the inflammatory response (39), mitochondrial energetics (40), and epithelial cell differentiation (41). Levels of the S100A10 transcript were not affected by mGPDH depletion (Fig. 6E and Supplementary Fig. 9), indicating that regulation of S100A10 by mGPDH occurs at the translational or posttranslational level. A recent study reported that the S100A10 protein was destabilized due to lysine succinylation regulated by mitochondrial SIRT5 desuccinylase (42). SIRT5 belongs to the NAD⁺-dependent sirtuin family and is identified as the only protein among sirtuins that influences succinyl modifications (43). Since mGPDH was previously reported to induce NAD⁺ content in mouse myoblasts by our group (16), and this alteration has also been found in macrophages (14), we wanted to examine the changes in NAD⁺ content and S100A10 succinylation under mGPDH manipulation in cultured podocytes. Results revealed decreased NAD⁺ content in response to siRNA targeting mGPDH and increased NAD⁺ content when mGPDH was overexpressed (Fig. 6M). Accordingly, knocking down mGPDH increased the succinylation level of S100A10 (Fig. 6N), while overexpressing mGPDH dramatically reduced S100A10 succinylation levels (Fig. 6O). Moreover, the in vivo data confirmed an increase in succinylation of S100A10, as well as S100A10 and RAGE protein levels in the glomeruli of diabetic KO mice (Fig. 6P and Q). Taken together, our data demonstrate that mGPDH diminishes RAGE signaling by increasing the NAD⁺ content to reduce S100A10 succinylation.

We next explored whether mGPDH regulation of S100A10 succinylation status is mediated via desuccinylation by SIRT5. SIRT5 siRNA was applied along with mGPDH deletion. Results showed that induction of S100A10 succinylation in the absence of mGPDH was abolished by Si-SIRT5 (Fig. 6R). The role of CPT1A was also assessed since it is reported to modulate S100A10 succinylation (42). However, CPT1A siRNA did not hinder the regulation of S100A10 succinylation by mGPDH (Supplementary Fig. 10A), indicating that although both SIRT5 and CPT1A influence succinylation of S100A10, SIRT5 may be the primary mechanism for mGPDH regulation of S100A10 succinylation status. Moreover, a pulse-chase analysis showed that cycloheximide dramatically decreased the half-life of endogenous S100A10 in the control compared with podocytes treated with Si-mGPDH (Supplementary Fig. 10B). Additionally, there are other methods of posttranslational modification of S100A10, such as Annexin A2 stabilizing S100A10 by inhibiting its protein ubiquitination and degradation (44). Therefore, we performed an experiment to examine whether mGPDH-regulated S100A10 protein expression is related to Annexin A2. The results showed that although knockdown of Annexin A2 decreased S100A10 protein levels, it might not be involved in the effect of mGPDH on S100A10 (Supplementary Fig. 10C).

Next, we evaluated the effects of rescuing mGPDH expression to reverse the observations during disease, and mGPDH renal-specific overexpression was induced by injection of AAV9-NPHS1-mGPDH. mGPDH overexpression in podocytes was confirmed by immunofluorescence staining of kidney sections in each model (Supplementary Fig. 11). In line with the observation in mGPDH KO mice, AAV9-NPHS1-mGPDH did not affect blood glucose in either diabetic model (Fig. 7A and K). However, restoration of mGPDH significantly attenuated UACR, glomerulosclerosis, and podocyte damage; meanwhile, AAV9-mGPDH–treated mice showed a marked improvement in mitochondrial morphology, with clearly elongated and less rounded mitochondria, enhanced mitochondrial biogenesis and metabolism, and decreased ROS content under diabetic and ADR conditions (Figs. 7B–I and L–S and 8A–F). Furthermore, we found that expression of S100A10 and RAGE was decreased by overexpressing mGPDH (Figs. 7J and T and 8G), consistent with our in vitro mechanistic observations. These results all indicate that targeting mGPDH might represent a promising therapy for podocyte injury and glomerular disease.

Our observations uncovered an important role for mGPDH deficiency in promoting podocyte injury and glomerulosclerosis during DKD and ADR nephropathy, two independent models of proteinuric kidney disease. In our study, analyses of both human and mouse disease samples provided clinical and basic pathological insights into an association between decreased podocyte mGPDH and glomerular damage, a correlation that was functionally validated by podocyte mGPDH ablation in vivo and in vitro. We did not observe obviously different morphological phenotypes between nondiabetic mGPDH KO and WT mice. Although several parameters showed alterative trends in nondiabetic KO mice, such as increased UACR and ECM-related genes and decreased podocyte numbers, they were not statistically significant, except for decreased Uqcrc1 gene of OXPHOS. However, after 24 weeks of STZ challenge, diabetic KO mice exhibited exacerbated proteinuria, glomerular disease, podocyte damage, and mitochondrial dysfunction compared with diabetic WT mice, suggesting that mGPDH deficiency might be a genetic background contributor to DKD that accelerates hyperglycemia-induced mitochondrial dysfunction and podocyte injury. This pattern is similar to that reported for other crucial genetic determinants identified in DKD (21,45,46), which also showed no significant changes under nondiabetic conditions, but its loss reinforced the disease phenotype under lifestyle challenges. Notably, we demonstrated molecular and functional evidence for mitochondrial oxidative stress and dysfunction contributing to podocyte injury, glomerular damage, and proteinuria in diabetic mGPDH KO mice. This was in accordance with a previously identified role for mGPDH in promoting mitochondrial biogenesis in muscle regeneration (16) as well as with clinical and experimental evidence showing that podocytes with abnormal mitochondria and increased ROS production lead to glomerular disease (47). One limitation while exploring mitochondrial function is that we used two-dimensional TEM to examine mitochondrial ultrastructure, which must be interpreted with caution due to its nature and the difficulty in quantifying the structures without making assumptions about their shape. Certainly, three-dimensional TEM is a proven approach to solve this problem and obtain robust data for analysis of mitochondrial morphology (48,49), which should be used for analysis in our future study.

We found that mGPDH insufficiency induced RAGE by inhibiting S100A10 desuccinylation, which may be a previously unrecognized mechanism for the initiation of RAGE signaling during DKD. RAGE has been critically linked to podocyte functional perturbation (50). Our RNA-Seq analysis suggested that mGPDH regulates the RAGE signaling pathway in diabetic complications, and mice lacking mGPDH were more susceptible to inducers of both diabetic and nondiabetic proteinuria models. Knocking down RAGE abrogated mGPDH deficiency–induced podocyte injury and glomerular disease, suggesting that the effects of mGPDH are exerted through the regulation of RAGE signaling. Moreover, mGPDH regulates RAGE signaling by repressing S100A10. As a RAGE ligand (37,38), S100A10 was previously described to be involved in the podocyte-mediated immunity (35). Our study suggested that inhibition of S100A10 in cultured podocytes blocked mGPDH knockdown-induced podocyte loss and mitochondrial dysfunction, providing further evidence that S100A10 plays a regulatory role and indicating that mGPDH might be an upstream regulator of S100A10 during cellular processes in podocytes. Furthermore, we demonstrated that mGPDH-mediated NAD⁺ content repressed S100A10 expression at least partly due to increased S100A10 desuccinylation, a recently identified form of protein posttranslational modification, and this is similar to the modulatory mechanism exerted by SIRT5 on S100A10 in gastric cancer (42). We next confirmed that induction of S100A10 succinylation in the absence of mGPDH was abolished by Si-SIRT5 but not by siRNA targeting CPT1A, another mediator of S100A10 succinylation, indicating that although both SIRT5 and CPT1A influence the succinylation of S100A10, SIRT5 may be the primary mechanism for mGPDH’s regulation of S100A10 succinylation status. We also demonstrated that Annexin A2, which was reported to stabilize S100A10 by inhibiting its protein ubiquitination and degradation (44), is not involved in the effect of mGPDH on S100A10. Other possible posttranslational modifications of S100A10 in podocytes will be the subject of further investigation. Additionally, the capability of mGPDH to produce NAD⁺ was evidenced by earlier studies (14,16) as well as by KEGG analyses showing that mGPDH might regulate metabolism of tryptophan, an essential amino acid involved in the metabolic pathways for NAD⁺ (51).

Our findings propose mGPDH as a potential therapeutic target for glomerular disease. Restoration of mGPDH expression from the reduced levels found in podocytes from patients and mice with DKD or ADR nephropathy significantly attenuated podocyte dysfunction and glomerular injury in both models of proteinuric kidney disease. Studies by our colleagues and others have suggested that several antioxidants (sulforaphane, cinnamaldehyde, and n-3 polyunsaturated fatty acids) (17,52) and NAD⁺ supplementation (resveratrol, nicotinamide riboside, and nicotinamide mononucleotide) (53–55) may have potential benefits in enhancing the antioxidant capacity to improve kidney diseases. Moreover, recent studies also reported that antidiabetic agents (thiazolidinediones, dipeptidyl peptidase 4 inhibitors, and sodium–glucose cotransporter inhibitors) may offer potential alternatives for the treatment of podocyte damage (56–58). Therefore, further investigations are required to explore the potential relationships and effects of mGPDH and these therapies on renal diseases.

In conclusion, our study revealed a role for mGPDH deficiency in contributing to podocyte injury and proteinuria, at least in part by activating the S100A10/RAGE pathway. Moreover, rescuing renal mGPDH expression might have therapeutic potential for the treatment of DKD.

Acknowledgments. The authors thank Prof. Junli Liu (Shanghai Jiao Tong University School of Medicine) for help with comments on the manuscript before submission.

Funding. This work was supported by the National Science Fund for Distinguished Young Scholars (grant 81925007), the National Natural Science Foundation of China (grants 82070836, 82070881, 82000769, and 81970752), and the “Talent Project” of Army Medical University (2017R013, 2019R047, and 2019XQYYYJ003-2).

Duality of Interest. No potential conflicts of interest were reported.

Author Contributions. H.Q., X.G., X.L., R.Z., and L.Z. were responsible for acquisition of data and statistical analysis. X.G., Y.W., and B.H. performed analysis and interpretation of data. H.Q. and Y.Z. drafted the manuscript. H.Z. and Y.Z. performed critical revision of the manuscript for important intellectual content. H.Z. and Y.Z. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.