Neutrophil Extracellular Traps Induce Glomerular Endothelial Cell Dysfunction and Pyroptosis in Diabetic Kidney Disease Free

Diabetic kidney disease (DKD) is the primary cause of end-stage renal disease (1). Current therapies, such as hypoglycemic agents, renin-angiotensin-aldosterone system blockers, and sodium–glucose cotransporter 2 inhibitors, cannot completely prevent DKD progression (2,3). Therefore, a better understanding of the underlying mechanisms of DKD and identification of potential effective therapeutic targets is urgently needed.

Glomerular endothelial cells (GECs) are a key component of the glomerular filtration barrier (4). Endothelial tight junctions and the glycocalyx with negative charge covering endothelial cells are crucial for maintaining glomerular filtration (5,6). Increasing data suggest that endothelial injury plays an important role in the pathogenesis of DKD (7).

Neutrophils are one of the cellular components of innate immunity. When stimulated by pathogens, neutrophils release a network structure composed of loose chromatin, histones, and a variety of neutrophil granule proteins, such as myeloperoxidase (MPO) and neutrophil elastase (NE), and the integrated complexes are termed neutrophil extracellular traps (NETs) (8). The main processes of NETs formation consist of peptidylarginine deiminase 4 (PAD4) converting the arginine residue of histones into citrullinate, citrullinated histones loosening chromatin, and neutrophil granule proteins adhering to the network structure (9,10). Recently, NETs were shown to be involved in a variety of chronic diseases associated with metabolic disorders, such as atherosclerosis, gout, and nonalcoholic fatty liver disease (11,12). The serum level of NETs was also shown to be significantly increased among patients with diabetes compared with people without diabetes (13,14). However, whether NETs are involved in glomerular endothelial injury and further contribute to DKD progression remains unclear.

Here, we investigated the association between NETs and DKD in patients with diabetes and diabetic mouse models (streptozotocin [STZ]-induced and db/db mice). To explore whether eliminating NETs could ameliorate GECs injury and alleviate DKD, we used DNase I or PAD4 knockout in these mouse models. The mechanisms by which NETs promote GECs injury were further explored.

The study recruited 140 patients with type 2 diabetes from the endocrinology department of the First Affiliated Hospital of Chongqing Medical University from February 2014 to February 2017. All patients were divided into DKD (urinary albumin-to-creatinine ratio [UACR] >30 mg/g or estimated glomerular filtration rate [eGFR] <60 mL/min/1.73 m²) or non-DKD (UACR <30 mg/g and eGFR ≥60 mL/min/1.73 m²) groups. Clinical data, including age, sex, BMI, and UACR, are shown in Supplementary Table 1. No differences in age, sex, BMI, or blood glucose were found between the two groups. The study was approved by the Chongqing Medical University Ethics Committee. All subjects gave written informed consent.

Kidney tissue was obtained from patients with or without diabetes who had undergone nephrectomy for conventional renal carcinoma at the First Affiliated Hospital of Chongqing Medical University, Chongqing, China, and was examined from regions of the kidney unaffected by tumors. The clinical biochemical data of these patients were extracted from the electronic medical record system.

Animal care and experimental procedures were performed with approval from the Chongqing Medical University Animal Care Committees. Mice were housed in a specific pathogen-free facility with free access to chow and water and a 12-h day/night cycle.

Male C57BL/6J mice were purchased from Chongqing Medical University. Diabetes was induced with low-dose STZ (Sigma-Aldrich) intraperitoneal injection (50 mg/kg) at 8 weeks of age for 5 consecutive days. Meanwhile, the control mice were injected with citrate buffer. Body weight and fasting blood glucose levels were monitored biweekly. Diabetes was confirmed by a fasting blood glucose level of 15 mmol/L. Diabetic and control mice were both treated with 200 units of DNase I (Bioss Antibodies, D13004) twice a week via vein injection for 16 weeks after treatment with STZ. These mice were sacrificed at 24 weeks of age.

Male db/db and db/m mice were purchased from the Nanjing University Model Animal Research Center. The db/db mice and db/m mice were both treated with 200 units of DNase I twice a week via vein injection for 14 weeks at 8 weeks of age. Similarly, body weight and fasting blood glucose levels were monitored biweekly. These mice were sacrificed at 22 weeks of age.

PAD4^−/− mice (C57BL/6J background) were purchased from The Jackson Laboratory. Age- and sex-matched wild-type (WT) C57BL/6J mice served as controls. Diabetes was induced with an intraperitoneal injection of STZ for 5 consecutive days (50 mg/kg) at 8 weeks of age. These mice were sacrificed after 20 weeks of diabetes.

The levels of circulating NETs in the population were measured by quantifying the amount of serum MPO-DNA complexes (Cell Death Detection ELISA PLUS kit, Roche, 11774425001). ELISA kits were used in mice to measure serum double-stranded (ds)DNA (Invitrogen, P11496), serum NE (R&D Systems, MELA20), urinary albumin (Abcam, ab108792), and creatinine (Cayman Chemical, 500701). All ELISAs were performed in accordance with the manufacturers’ instructions.

The tissues were harvested and fixed with 4% paraformaldehyde overnight, and then the paraformaldehyde soaked in the tissues was replaced with 70% ethanol. The tissues were soaked in gradient ethanol from 70 to 100% for dehydration, followed by clearing agent, and finally embedded in paraffin. Sections were stained with periodic acid Schiff (Servicebio, Wuhan, China, G1008) after deparaffinization by standard methods. Glomerular vascular cell adhesion molecule 1 (VCAM-1) was determined in paraffin-embedded kidney sections with a VCAM-1 antibody (Santa Cruz Biotechnology, sc-13160). The quantification of VCAM1 is represented as the folding change relative to WT-control mice, as described in a previous study (15). Quantitative analysis was performed using ImageJ software.

For immunofluorescence staining, paraffin sections of kidneys were deparaffinized, antigen-repaired using sodium citrate buffer, and antigen-blocked with immunofluorescence blocking buffer (Cell Signaling Technology, 12411). Next, sections were incubated overnight at 4°C with the following primary antibodies: MPO (Abcam, ab9535), NE (Abcam, ab68672), citrullinated histone H3 (CitH3; Abcam, ab5103), lymphocyte antigen 6 complex locus G6D (Ly6G; Abcam, ab25024), platelet endothelial cell adhesion molecule 1 (CD31; Servicebio, GB11063), gasdermin D (GSDMD; Santa Cruz Biotechnology, sc-393581), apoptosis-related speckle-like protein (ASC; Santa Cruz Biotechnology, sc-514414), and nucleotide-binding oligomerization domain (NOD)-like receptor family pyrin domain-containing 3 (NLRP3; Santa Cruz Biotechnology, sc-134306). Human renal GECs (HRGECs; ScienCell, 4000) were fixed with 4% paraformaldehyde for 20 min, followed by antigen blockade using immunofluorescence blocking buffer for 1 h. Samples were stained with antibodies against vascular endothelial (VE)-cadherin (Abcam, ab33168), CD31 (Cell Signaling Technology, 3528), and endothelin 1 (ET1; Abcam, ab2786). Raw images were captured using a fluorescence microscope (Leica, DM4000 132 LHD) or a confocal microscope (Leica, TCS SP8). Quantitative analysis was performed using ImageJ software.

Mouse kidneys or HRGECs were lysed in radioimmunoprecipitation assay buffer (Thermo Fisher Scientific, 89900) supplemented with protease inhibitor cocktail (Thermo Fisher Scientific, 78430). The lysates were separated by 12% SDS-PAGE, and the proteins were transferred to a polyvinylidene difluoride membrane. After being blocked with 5% skim milk, the membranes were incubated with primary antibodies against CitH3, VCAM-1, ET1, ASC, GSDMD (Cell Signaling Technology, 97558), and β-actin (Cell Signaling Technology, 4970S) at 4°C overnight. Then, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature.

Human blood neutrophils were isolated from human blood using Polymorphprep (Axi-Shield) following the manufacturer’s instructions. Human blood neutrophils (1 × 10⁶) were seeded in RPMI Medium 1640 (Gibco) supplemented with 3% FBS and stimulated with 100 nmol/L phorbol 12-myristate 13-acetate (Selleck, S7791) or high glucose (25 mmol/L) for 4 h. Afterward, the culture medium was removed, and NETs were incubated with the restriction enzymes BseRI, PacI, NdeI, and AfII (New England Biolabs) for 1 h. Supernatants containing the NETs fragments were collected and centrifuged 10 min at 1,000g to remove remaining cell debris. NETs were quantified using a Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, P11496) according to the manufacturer’s instructions. NETs were verified by staining with Sytox green (Thermo Fisher Scientific, S7020).

HRGECs were purchased from ScienCell. HRGECs were cultured in endothelial cell medium (ScienCell) with 1% endothelial cell growth supplement, 1% antibiotic solution, and 5% FBS. Cells were treated with 100 nmol/L oleylamine (MilliporeSigma, O7805) for 30 min and then washed and treated with 200 ng/mL or 300 ng/mL NETs for 24 h.

HRGECs were seeded onto polyester membranes (0.4-µm pore size, Corning) and then treated with NETs for 24 h. The medium in the insert was replaced by serum-free medium containing 0.5 mg/mL FITC-labeled BSA (Solarbio, SF063), and the medium in the well was replaced with serum-free medium containing 0.5 mg/mL unlabeled BSA (Sigma-Aldrich). After 1 h, aliquots were removed from the wells, and fluorescence was determined with excitation at 495 nm and emission at 520 nm in a fluorescence microplate reader (Synergy H1).

Transwell migration assays were performed as described previously (16). A total of 5 × 10⁵ neutrophils were resuspended in 1% penicillin/streptomycin serum-free RPMI 1640 medium (Gibco). Then, neutrophils were seeded in the upper wells of the Transwell inserts (3-µm pore, polycarbonate membrane, Corning), and the supernatant of NET-activated HRGECs was added to the lower wells. Following an incubation period (30 min at 37°C), the cells that migrated through the membrane were stained with crystal violet and counted.

The incidence of cell death was detected by flow cytometry using annexin V/propidium iodide (PI) double staining according to the manufacturer’s protocol. HRGECs were harvested, washed, and incubated with annexin V and PI at 37°C for 30 min. Finally, the samples were analyzed by flow cytometry (BD Biosciences, Franklin Lakes, NJ).

HRGECs were treated with NETs or vehicle control, and the experiments were repeated five times to obtain five samples for each group. Then, 3 μg RNA for each sample was used as input material for RNA library preparation. Sequencing libraries were generated using a NEBNext Multiplex RNA Library Prep Set for Illumina (San Diego, CA). Raw data in FASTQ format were processed through custom Perl and Python scripts. Transcripts with P values of <0.05 and absolute values of logtwofold change of >1 were assigned as differentially expressed transcripts. Pathway analyses of the differentially expressed genes were performed using Metascape. The mRNA sequencing data were deposited at the Gene Expression Omnibus under accession number GSE189875 (secure token: kdgrsiqondupdkp).

Cell death in kidney sections was detected using a 3,3′-diaminobenzidine (activated streptavidin-horseradish peroxidase) TUNEL Detection Kit (Servicebio, G1507) or a DeadEnd Fluorometric TUNEL System (Promega, G3250) in accordance with the manufacturer’s protocol.

Statistical analyses were performed using GraphPad Prism 8. Results are expressed as the mean ± SD or median (interquartile range). If the data showed a normal distribution, t tests were performed for comparisons between two groups, one-way ANOVA was applied for comparisons of multiple groups, and the Pearson correlation coefficient was used for correlation analyses. Data in abnormal distribution were analyzed by nonparametric test. The results were considered significant if P < 0.05.

All data generated or analyzed during this study are included in the published article and in the Supplementary Material.

We measured serum levels of MPO-DNA complexes, a well-established marker of NETs (17), among patients who had type 2 diabetes, with or without DKD. A significant elevation of serum MPO-DNA complexes in patients with DKD was observed compared with those without DKD (Fig. 1A). In addition, a higher level of serum MPO-DNA complexes was significantly associated with a higher UACR (Fig. 1B).

To further explore whether NETs were deposited in the glomerulus, the NETs markers MPO and NE were stained in the renal cortex by immunofluorescence. Confocal microscopy revealed the colocalization of MPO and NE in the glomeruli of DKD patients, which indicated NETs deposition. We did not observe any positive immunostaining of MPO or NE in the glomeruli of volunteers without diabetes (Fig. 1C). In addition, immunofluorescence colocalization of NETs and endothelial cell marker CD31 in human kidney tissue was conducted. Immunofluorescence staining showed NETs in the glomeruli of DKD patients were close to GECs (Supplementary Fig. 1A).

To inhibit NETs by degrading NET-associated DNA, DNase I was administered to STZ-induced diabetic mice (a model of type 1 diabetes) and to db/db mice (a model of type 2 diabetes) at 8 weeks of age (Supplementary Figs. 2A and 3A). Elevated NETs (as measured by dsDNA and NE) in STZ and db/db mice were significantly reduced after treatment with DNase I (Figs. 2A and B and 3A and B). We also observed decreased albuminuria (Figs. 2C and 3C), ameliorated glomerular hypertrophy and mesangial matrix expansion, as manifested by periodic acid Schiff staining (Figs. 2D and E and 3D and E), and attenuated podocyte foot process effacement and glomerular basement membrane thickening (imaged by electron microscopy, Supplementary Figs. 2D and 3D) after treating these diabetic mice with DNase I. Blood glucose concentration and body weight were not influenced by DNase I (Supplementary Figs. 2B and C and 3B and C).

In addition, the expression of VCAM-1, an indicator of GECs injury, was obviously increased in the glomeruli of STZ and db/db mice, and treatment with DNase I significantly decreased VCAM-1 expression (Figs. 2F and 3F). Furthermore, TUNEL-positive cells were also increased in the glomeruli of STZ and db/db mice and were diminished after treatment with DNase I (Figs. 2G and 3G). DNase I significantly reduced the death of GECs in diabetic mice (Supplementary Figs. 2E and 3E).

PAD4 regulates NETs formation via citrullination of histones. To further explore the role of NETs in glomerulopathy and GECs injury, PAD4^−/− mice were generated. Diabetes was induced by low-dose STZ injections in PAD4^−/− mice and WT control mice. Mice injected with citrate buffer vehicle served as nondiabetic controls. Serum NETs markers (dsDNA) and glomerular NETs markers (CitH3 and Ly6G) were dramatically decreased in diabetic PAD4^−/− mice (Fig. 4A–C). Compared with diabetic WT mice, diabetic PAD4^−/− mice exhibited a lower kidney-to-body weight ratio (Fig. 4D) and a lower UACR (Fig. 4E). We observed blunted glomerular hypertrophy and mesangial expansion (Fig. 4F and G), less podocyte foot process effacement, and diminished glomerular basement membrane thickening (Supplementary Fig. 4C). In addition, VCAM-1 expression was significantly reduced in diabetic PAD4^−/− mice compared with diabetic WT mice (Fig. 4H). We also observed a diminished CD31/TUNEL double-staining–positive area in diabetic PAD4^−/− mice (Fig. 4I), which suggested attenuated GECs injury. Hyperglycemia and body weight loss did not differ between diabetic PAD4^−/− and WT mice (Supplementary Fig. 4A and B).

Fluorescence of Sytox green and scanning electron microscopy analysis suggested that high glucose could trigger NETs formation (Fig. 5A and Supplementary Fig. 5A). Furthermore, high glucose increased neutrophil reactive oxygen species (ROS) production (Supplementary Fig. 5B), and treatment with the ROS inhibitor diphenyleneiodonium or the ROS scavenger N-acetylcysteine significantly inhibited high glucose–induced NETs formation (Fig. 5A).

When high glucose–induced NETs were used to treat HRGECs, the endothelial integrity markers VE-cadherin and CD31 (18,19) were decreased, the endothelial injury markers VCAM-1 and ET1 were increased (Fig. 5B and C), and the transendothelial passage of albumin was elevated (Fig. 5D). It was reported that injuries to HRGECs promote the recruitment and adhesion of circulating leukocytes (20). Supernatants from NET-treated HRGECs were also able to evoke the chemotactic attraction of neutrophils (Fig. 5E) and trigger NETs formation (Fig. 5F), which suggested a vicious cycle of NETs-GECs-NETs.

NETs were shown to be cytotoxic to endothelial cells (21), while the specific type of death remains controversial (22–25). We confirmed that high glucose–induced NETs exerted a dose-dependent cytotoxic effect on HRGECs (Fig. 6A). Then, we performed transcriptome sequencing to examine the gene expression profiles in control- and NET-treated HRGECs. Transcriptome profiling revealed 438 differentially expressed genes between control- and NET-treated HRGECs (Fig. 6B), and most of them were located in the cell membrane. The functional enrichment and pathway analyses revealed an enrichment of genes related to transmembrane transport (Fig. 6C). Scanning electron microscopy was performed to determine the effect of NETs on the membranes of HRGECs. Interestingly, compared with controls, NET-treated HRGECs exhibited ball-like bulge and pore formation of the cell membrane under a scanning electron microscope (Fig. 6D). The morphological features of HRGECs incubated with NETs were consistent with cell pyroptosis (26). Flow cytometry analysis of PI indicated HRGEC death induced by NETs (Fig. 6E). Furthermore, we examined markers of pyroptosis (27) and found higher mRNA levels of ASC, NLRP3, interleukin (IL)-β and IL-18 (Fig. 6F) and up-regulated protein levels of cleaved GSDMD and ASC in NET-treated HRGECs (Fig. 6G). In vivo, immunofluorescence staining indicated increased colocalization of pyroptosis-related markers (GSDMD, ASC, or NLRP3) with a GECs marker (CD31) in the glomeruli of diabetic WT mice, while diabetic PAD4^−/− mice showed decreased pyroptosis-related markers compared with diabetic WT mice (Fig. 6H). Among patients with DKD, immunofluorescence staining indicated an increased staining intensity for the colocalization of CD31 with GSDMD or ASC in glomeruli compared with volunteers who did not have diabetes (Fig. 6I). Consistent with previous reports on pyroptosis (26,28), the changes in morphological features and biochemical markers in vitro and in vivo indicated that NETs induced pyroptosis in GECs.

Bioelectricity was shown to be involved in histone (a component of NETs)-induced cytotoxicity and pore formation of the cell membrane (24,29,30). As histones are positively charged and GECs surfaces are negatively charged and HRGEC membrane potential-related molecules were interfered with NETs in transcriptome profiling (Fig. 6C), we tested whether NET-induced HRGEC pyroptosis was mediated by surface charge. Increasing the HRGEC surface potential by oleylamine significantly alleviated the death of HRGECs induced by NETs (Fig. 6E). Furthermore, NET-induced expression levels of ASC, NLRP3, IL-1β, IL-18, and cleaved GSDMD in HRGECs were also decreased after oleylamine treatment (Fig. 6F and G), which suggested that NETs may induce GECs pyroptosis via the surface charge of the cell membrane.

NETs have previously been linked to diabetes and autoimmune renal diseases (12,31), but their relationship to DKD has been poorly addressed. Here, we identified that DKD patients and diabetic mice both exhibited higher levels of NETs in serum and glomeruli, which correlated with more severe GECs injury and worsened renal function. Treatment with DNase I or PAD4 knockout significantly alleviated GECs injury and reversed diabetes-induced histopathological glomerulopathy. Furthermore, NET-induced GECs injury manifested as pyroptosis, which could be attenuated by oleylamine to increase the surface charge in vitro. Targeting NETs might be useful to prevent DKD progression.

NETs were initially considered a new type of innate immunity (32), and previous studies showed that NETs were involved in autoimmune renal diseases such as lupus nephritis, anti-neutrophil cytoplasmic antibody–associated glomerulonephritis, and anti-glomerular basement membrane nephritis (8,12). The role of NETs in metabolic renal diseases has not been completely addressed. In the current study, we not only observed increased NETs deposition in the glomeruli of DKD patients and diabetic mice but also further proved that treatment with DNase I or deletion of the PAD4 gene efficiently alleviated DKD. For the first time, these data confirmed an important role of NETs in metabolic renal disease beyond previous autoimmune renal diseases.

Increasing evidence suggests that GECs injury plays a major role in DKD progression. As one of the major components of the filtration barrier, GECs have a unique feature of fenestrations and can handle many filtrations. Meanwhile, GECs are vulnerable to many biochemical damage factors, as they are directly exposed to blood circulation. Previously, NETs were shown to be able to promote vascular leakage and endothelial-to-mesenchymal transition in a disease model of lupus nephritis (33), and anti-neutrophil cytoplasmic antibody-mediated endothelial damage was also proven to be mediated by NETs (34). Beyond GECs injury in autoimmune kidney diseases, we demonstrated that in diabetic mice, treatment with DNase I and knockout of PAD4 significantly reduced the expression of VCAM1 in glomeruli and alleviated diabetes-induced GECs injury. In vitro, incubation of human glomerular primary endothelial cells with high glucose–induced NETs resulted in higher expression levels of VCAM-1 and ET1, which suggested aggressive GECs injury.

We further explored the type of GECs injury induced by NETs. Previous studies reported that NETs promoted cell death in endothelial cells, podocytes, renal tubular cells, epithelial cells, and cardiomyocytes (21,23,35,36). Some mechanisms of NET-induced cell death were also investigated. Histones, which are major components of NETs, were proven to mediate cytotoxicity through Toll-like receptor 2/4 (TLR2/4) in glomerulonephritis (23) and acute kidney injury (37). Another study showed that histones and myeloperoxidase of NETs were responsible for NET-mediated cytotoxicity to epithelial cells and endothelial cells (21). Furthermore, complement dependence (38) and NET-externalized matrix metalloproteinases are involved in the process by which NETs promote endothelial cell death (39). Our transcriptome profiling results suggested that NETs caused changes in multiple genes related to the cell membrane of HRGECs. Scanning electron microscopy of NET-incubated GECs revealed ball-like bulge and pore formation in the cell membrane, which was consistent with the morphological characteristics of pyroptosis. The expression levels of cleaved GSDMD and NLRP3, the main pathway proteins of pyroptosis, were also increased when GECs were incubated with NETs. In vivo, we demonstrated that GECs pyroptosis induced by diabetes was significantly alleviated by PAD4 knockout. These results suggested that the injury type of GECs induced by NETs was pyroptosis.

Accumulating data have shown that pyroptosis-mediated cell death is an important pathophysiological change in DKD, but the exact mechanism has not been completely uncovered (40). GSDMD, caspase-1/4/11, NLRP3 inflammasome, thioredoxin-interacting protein, and some circular RNAs were shown to be involved in pyroptosis-related signaling pathways and to promote DKD progression (41). In other disease models, NETs attract monocytes in a surface charge–dependent fashion (42), and histones of NETs induce pore formation in smooth muscle cell membranes through surface charge, resulting in lytic death (24). In addition, histones can also recognize and bind to the cell membrane receptor Clec2d through charge (43). Of note, Clec2d carries histone-DNA complexes into endosomes to stimulate TLR9, and the activation of TLRs can induce pyroptosis (44). In our study, transcriptome profiling analysis uncovered multiple molecules referring to the membrane potential of HRGECs. Considering that NETs contain a variety of positively charged proteins and that GECs are characterized as negatively charged, we used oleylamine to increase the cell surface charge of HRGECs. Interestingly, oleylamine not only reversed the NET-induced death of HRGECs but also inhibited pyroptosis-related signaling pathways in GECs. These data indicated that NETs induced GECs pyroptosis via cell surface charge.

The main strength of our study is that we demonstrate the crucial role of NETs in DKD progression in different mouse models and humans. We further uncovered the potential mechanism of NET-induced GECs pyroptosis from the perspective of cell surface charge.

The major limitation of our study is the mouse model of global PAD4 knockout rather than a bone marrow–specific PAD4 knockout. Although PAD4 is mainly expressed in bone marrow, we could not completely rule out that the altered GECs susceptibility to cell death is attributed to PAD4 knockout in other tissues beyond bone marrow. Another limitation is that our observations do not rule out the effect of other potential pathways, such as TLRs, in mediating GECs pyroptosis. In addition, a charge-dependent manner in which NETs induce endothelial cell pyroptosis was not proven in vivo. Future studies are warranted to confirm which components of NETs induce pyroptosis in GECs.

In conclusion, GECs pyroptosis mediated by charge is involved in the pathogenesis of DKD, which is promoted by NETs, while elimination of NETs significantly attenuates GECs injury and alleviates the progression of DKD. Further studies may be needed to explore whether NETs are a potential therapeutic target for improving DKD.

Funding. This work was supported by the National Natural Science Foundation of China, Major Joint Project (U21A20355), National Natural Science Foundation of China (81800731, 81870567, 81970720, 82000810, 82170825, and 82270878), the China Postdoctoral Science Foundation (2019M663499), the Chongqing Postdoctoral Innovative Talents Support Project (2020[379]), Joint Medical Research Project of Chongqing Science and Technology Commission & Chongqing Health and Family Planning Commission (Major Project, 2021ZDXM002), the Natural Science Foundation of Chongqing (cstc2020jcyj-bshX0081), Outstanding Talents of the First Affiliated Hospital of Chongqing Medical University 2019 (2019-4-22 for J.H.), Program for Youth Innovation in Future Medicine, Chongqing Medical University, Chongqing Outstanding Youth Funds (cstc2019jcyjjq0006), and Bethune Merck Diabetes Research Foundation (G2018030).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. F.Z., L.M., and X.L. performed the experiments, analyzed the data, and wrote the manuscript. R.G., C.P., and B.K. provided technical support and critical discussions of the manuscript. Y.W. and T.L. managed clinical data and samples. J.W., Y.Y., and L.G. provided technical advice on mouse studies. Q.L. and Z.W. supervised the project and critical discussions of the manuscript. S.Y. and J.H. designed the experiments. S.Y. and J.H are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.