Renal biopsy samples from patients with biopsy-proven DKD (n= 6) (4 males and 2 females; age 42–66 years; mean age 57.0 ± 9.27 years) were obtained from the Division of Nephrology, Renmin Hospital of Wuhan University (Wuhan, China). Para-carcinoma tissues from patients without diabetes who underwent tumor nephrectomies (n = 6) (3 males and 3 females; age 40–63 years; mean age 53.2 ± 8.40 years) were used as controls and were obtained from the Division of Pathology, Renmin Hospital of Wuhan University (Wuhan, China).

Histologic examination and biochemical analysis (urine albumin-to-creatinine ratio <30 mg/g) showed no features of DKD apart from solitary renal cell carcinoma. Investigations were carried out on the principles of the Declaration of Helsinki. Informed consent was obtained from all patients included in this study. Experiments were performed according to the guidelines of Wuhan University, which the Research Ethics Committee of Renmin Hospital of Wuhan University approved.

A total of twelve 8-week-old male Sprague–Dawley rats were randomly divided into the diabetic group (6 rats) and the control group (6 rats). The type I diabetes rat model was established by the intraperitoneal injection of streptozotocin (STZ, 65 mg/kg of body weight; S0130, Sigma-Aldrich, United States). The control rats received 0.1 mol/L citrate buffer (pH = 0.45) (Chen et al., 2020). Blood glucose was measured after administration of STZ. The rats with a blood glucose level exceeding 16.7 mM were considered diabetic. The blood glucose and body weight of rats were recorded at the end of 0, 4, 8, 12, and 16 weeks. In addition, 24-h urine samples were collected at the end of 12 and 16 weeks to evaluate total urine proteins (UTP) at 24 h. All rats were sacrificed at 12 or 16 weeks post-STZ injection, and the kidneys were collected and used for subsequent testing.

Conditionally immortalized human podocytes (HPC) were provided by Dr. Moin A. Saleem (Academic Renal Unit, Southmead Hospital, Bristol, United Kingdom) and cultured in RPMI 1640 medium (SH30809.01, HyClone, United States) supplemented with 10% fetal bovine serum (FBS, 10100; Gibco, United States), penicillin G (100 U/ml) -streptomycin (100 μg/ml) (15140122; Life, United States), 1×insulin transferrin-selenium (ITS, 41400045; Life, United States) at 33°C for proliferation. When the density of HPC reached 70–80%, the cells were transferred to 37°C with an ITS-free medium for differentiation (Fan et al., 2019; Hu et al., 2019; Lin et al., 2020). The differentiated HPC were cultured using different mediums (normal glucose, N = 5 mM d-glucose; mannitol, MA = 5 mM D-glucose+20 mM mannitol; high glucose, HG = 25 mM d-glucose) for 24 h or 36 h; The PERK-inhibited HPC were treated with 20 µM PERK phosphorylation inhibitor (GSK2656157; Selleck, United States).

Cell transfection was conducted using the X-tremeGENE HP DNA Transfection Reagent (Rh-06366 236 001; Roche, Swit). A total of 2 × 10⁵ HPC were seeded into six-well plates, transfected with Mfn2-Myc plasmid (Addgene) or shRNA-Mfn2 plasmid (PIEE248070577; Shanghai GenePharma Co., Ltd.) for Mfn2 overexpression or knockdown, and transfected with pcDNA3.1 plasmid (Addgene) or scrambled-shRNA plasmid (Shanghai GenePharma Co., Ltd.) as control. The HPC were incubated under normal cultured conditions for 36–48 h after adding the complexes of plasmids (1 μg/ml) and transfection reagent (1 μL/ml). The HPC transfected with Mfn2-Myc/pcDNA3.1 plasmid (Addgene) were cultured using HG (25 mM d-glucose) medium for 24 h after transfection 36–48 h.

The renal cortex of rats and HPC were fixed in 2.5% glutaraldehyde to further observe the ultrastructure of glomerular podocyte, mitochondrial morphology, MAMs integrity of glomerular podocyte and HPC (Fan et al., 2019). A transmission electron microscope (TEM; Hitachi, Japan) was used to observe all of the microscopic images.

The aspect ratio of mitochondria and MAMs contact points were assessed using ImageJ software. The percentage of the mitochondrial membrane in contact with ER within a 50 nm range was measured and normalized to the mitochondrial perimeter, as previously described (Tubbs et al., 2018).

The frozen renal biopsies from patients and the frozen kidney sections from rats were stained with Hematoxylin and Eosin (H&E) and Periodic acid-Schiff (PAS) reagents to observe the histopathological changes of the kidney. An upright microscope (Olympus, Japan) was used to observe all of the microscopic images.

For Western blot analysis, total protein from the isolated glomeruli or HPC was extracted with RIPA buffer (P0013; Beyotime, China) mixed with protease inhibitor cocktail (p8340; Sigma-Aldrich, United States), phosphatase inhibitor phosSTOP™ (49068370001; Sigma-Aldrich, United States), and PMSF (ST506; Beyotime, China). The extractions were centrifuged at 13,000 rpm at 4°C for 5 min, then the supernatants were mixed with loading buffer and boiled at 100°C for 10 min. Equal amounts of protein were separated by 8–12% SDS-PAGE and transferred onto PVDF membranes (Millipore, United States). After blocking with 5% skimmed milk at room temperature for 1 h, the membranes were incubated with primary antibodies (Fan et al., 2019). The following primary antibodies were used in this study: Mfn2 (ab123773; 1:1,000; Abcam, United Kingdom); cleaved caspase3 (ab32042; 1:1,000; Abcam, United Kingdom); CHOP (sc-7351; 1:500; Santa Cruz, United States); Bcl2 (ab32124; 1:1,000; Abcam, United Kingdom); Bax (60267-1-lg; 1:1,000; Proteintech, United States); IRE1α (#3294; 1:1,000; CST, United States); ATF6β (15794-1-AP; 1:1,000; Proteintech, United States); p-PERK (#3719; 1:1,000; CST, United States); PERK (sc-377400; 1:500; Santa Cruz, United States); Myc-tag (#562; 1:1,000; Medical & Biological Laboratories CO. LTD., Beijing, China); GAPDH (sc-32233; 1:2000; Santa Cruz, United States). Secondary antibodies conjugated to Horseradish peroxidase (HRP) (ANT019s, ANT020s, 1:10000; Antgene, Wuhan, China) were used at room temperature for 1 h. The membranes were soaked with enhanced chemiluminescence solution (ECL, G2014, Servicebio, Wuhan, China), and the blots were further observed with an X-ray machine (Bio-Rad, United States).

Coimmunoprecipitation was carried out to verify the interaction between Mfn2 and PERK. First, the lysates of HPC cultured in 10 cm² plates were subject to immunoprecipitation with 3–5 µl PERK or Mfn2 antibody. The immunoprecipitates were blotted with Mfn2 (1:1,000; Abcam, United Kingdom) or PERK (1:500; Santa Cruz, United States) antibody at 4°C overnight and then centrifuged at 13,000 rpm at 4°C for 5 min to get the supernatants. After being mixed with loading buffer boiled at 100°C for 10 min, the protein samples were processed for Western blot analysis (Ma et al., 2018).

The frozen renal biopsies from patients and the frozen kidney sections from rats were fixed with 4% paraformaldehyde for 30 min. After block with 5% bovine serum albumin (BSA, ANT101L; Antgene, China) at room temperature for 1 h, the sections were then incubated with the primary antibody (Fan et al., 2019). The Mfn2 (1:200; Abcam, United Kingdom) antibody was used as the primary antibody in this study. All microscopic images were observed using an upright microscope (Olympus, Japan).

The immunofluorescence assay analyses using the frozen renal biopsies from patients, the frozen kidney sections from rats and the cell climbing films were performed as previously described. The following primary antibodies were used in this study: Mfn2 (1:200; Abcam, United Kingdom); WT1 (NB110-60011, 1:50; Novus, GER); Synaptopodin (65194, 1:50; Progen, GER); TOM20 (sc-17764; 1:50; Santa Cruz, United States); Calreticulin (ab92516; 1:100; Abcam, United Kingdom); PERK (1:50; Santa Cruz, United States). The Alex Fluor 488/594 donkey anti-rabbit/Mouse IgG (HL) (ANT023, ANT024, ANT029, ANT030; 1:100; Antgene, Wuhan, China) were used as secondary antibodies. The nuclei were stained with an anti-fluorescence quencher containing DAPI (ANT063; Antgene, Wuhan, China). A confocal microscope (Olympus, Japan) was used to observe all of the microscopic images. The Pearson’s correlation coefficients (PCC) have been calculated in immunofluorescence assay of TOM20 and CRT, which with a value between 1 and −1, was used to quantitative colocalization between two proteins (1 means perfect correlation; −1 means completely excluded; 0 means no relationship) (Schober et al., 2018).

Mitochondrial ROS generation measured using the DCFH-DA fluorescent probe (S0033S; Beyotime, China); Mitochondrial membrane potential (MMP) was evaluated by JC-1 staining (C2003S; mitochondrial membrane potential assay kit with JC-1; Beyotime, China); The ATP Assay Kit (S0027; Beyotime, China) was used to measure the ATP levels.

The glomerular podocyte apoptosis of rats was analyzed by co-staining with TUNEL (C1086; One Step TUNEL Apoptosis Assay Kit, Beyotime, China) and Synaptopodin antibody (1:50; Progen, GER).

The apoptosis rate of HPC was measured by flow cytometry with Annexin V-PE/7AAD double staining (#556547, Annexin V-PE Apoptosis Detection Kit I, BD Pharmingen, United States).

All experiments were performed at least three times independently. All data were summarized as the means ± SD and analyzed using GraphPad Prism5 (GraphPad Software, Inc.). One-way or two-way analysis of variance (ANOVA) was applied to analyze data for comparisons between groups. The p < 0.05 was considered significant.

The STZ-treated rats developed hyperglycemia, and their body weight decreased compared to the control group (Figures 1A,B). Differences in 24-h urine total proteins (UTP) were observed between the two groups, suggesting an impaired renal function in the diabetic rats (Figure 1C). In addition, both HE and PAS staining demonstrated that the diabetic animals presented evident mesangial matrix expansion compared to the controls (Figure 1D). Furthermore, the transmission electron microscopic (TEM) analysis revealed significant fusion of podocyte foot process in diabetic rats (Figure 1E). Collectively, these findings indicated glomerular damage and podocyte injury in diabetic rats.

Even though MAMs dysfunction has been reported to be associated with metabolic syndrome, including DKD (Yang et al., 2021), the in vivo studies on MAMs of podocytes under hyperglycemia conditions are limited. The TEM analysis of glomerular podocytes demonstrated mitochondrial fragmentation and MAMs reduction in the diabetic rats compared with those in the controls, which indicated elongated mitochondria and abundant contact points localized at the interface of the ER and outer mitochondrial membrane (Supplementary Figures 1B–D). In addition, we performed the triple immunofluorescent staining of TOM20 (mitochondrial marker), Calreticulin (CRT, ER marker), and DAPI to evaluate ER-mitochondrial interactions in glomeruli. The results indicated reduced ER-mitochondrial interactions in the diabetic rats (Supplementary Figure 1A).

The expression of Mfn2 was evaluated to clarify the mechanism of mitochondrial fragmentation and MAMs reduction in podocytes. Compared with the controls, the glomeruli from diabetic rats showed decreased Mfn2 levels (Figures 1F–H). The triple immunofluorescent staining of Mfn2, WT1 or Synaptopodin (podocyte marker) and DAPI was conducted to better observe the Mfn2 expression alterations. The results indicated that Mfn2 was located at the cytoplasm, and the Mfn2 expression was reduced in the podocytes of diabetic rats (Figures 1I,J). These results verified that hyperglycemia-induced Mfn2 expression decreases in vivo. Altogether, the results revealed that hyperglycemia-induced reduction of Mfn2 expression was related to glomerular damage and podocyte injury in DKD.

Renal pathological changes such as extracellular matrix deposition were observed in patients with DKD through HE and PAS staining (Figure 2A). As with the tendency of Mfn2 expression in the glomeruli of diabetic rats, Mfn2 was also reduced in the glomeruli of patients with DKD (Figure 2B). The triple immunofluorescent staining of Mfn2, WT1 or Synaptopodin, and DAPI was performed to evaluate Mfn2 expression in glomerular podocytes. A dramatic reduction in the Mfn2 levels was detected in the podocytes of patients with DKD relative to those of healthy individuals (Figures 2C,D). The immunofluorescent staining of TUNEL, Synaptopodin, and DAPI showed increased podocyte apoptosis in patients with DKD (Supplementary Figure 2).

The TEM photomicrographs showed increased mitochondrial fragmentation in vivo. Further, the loss of MAMs contact points in HPC under HG conditions was similar to the alterations in diabetic rats (Figures 3A–C). Additionally, the triple immunofluorescent staining of TOM20, CRT, and DAPI showed reduced ER-mitochondrial interactions in HG-treated HPC (Figures 3J,K).

Reduced expression of Mfn2 was also observed in HG-treated HPC. As shown in (Figures 3D–G), glucose concentration gradients (5, 10, 15, 20, 25, 30 mM) at 24 h and different time points (0, 6, 12, 24, 36, 48 h) in HG conditions (25 mM) were set to evaluate the Mfn2 expression by Western blots. The results suggested that Mfn2 expression was downregulated both in concentration-dependent and time-dependent manners under HG conditions in vitro. In addition, mannitol did not influence Mfn2 expression in HPC compared with normal conditions (Figures 3H,I). Furthermore, HG-treated HPC showed a dramatic decrease of Mfn2 fluorescence in the TOM20-marked mitochondria via immunofluorescent assays (Figure 3L). Therefore, these results suggested that HG alters the expression of Mfn2, thus, resulting in morphological and functional alterations of mitochondria and MAMs reduction in podocytes.

Based on the previous observations linking ER stress to glomeruli damage and podocyte injury (Cybulsky, 2010), we examined the hyperglycemia or HG-induced ER stress in glomeruli from STZ-induced diabetic rats and cultured podocytes. It has been reported that the occurrence of ER stress first activates the UPR pathway, which is then marked by activation of the three sensors of UPR (IRE1, PERK, and ATF6) (Cybulsky, 2017).

Firstly, the protein expression of CHOP and cleaved caspase3 was detected. CHOP is the marker of ER stress-induced apoptosis. Up-regulation of CHOP enhances apoptosis by inducting the pro-apoptotic protein Bax and suppressing the synthesis of anti-apoptotic protein Bcl2 in persistent ER stress (Hata et al., 2015). As shown in (Figures 4A–C), STZ-induced hyperglycemia markedly increased the apoptosis level of glomeruli, which was consistent with our previous observation (Fan et al., 2019; Lin et al., 2020). Decreased Bcl2 expression and increased Bax expression were also observed in glomeruli from diabetic rats (Supplementary Figures 3A–C). Subsequently, we observed the protein levels of three UPR sensors (p-PERK, IRE1α, ATF6β) in glomeruli from STZ-treated rats. Compared with the controls, the diabetic rats had significantly increased expression of p-PERK, IRE1α and ATF6β (Figures 4D–G). These findings suggested that the three UPR signaling pathways were activated in glomeruli from STZ-treated rats. In addition, increased podocyte apoptosis in diabetic rats was observed by the triple immunofluorescent staining of TUNEL, Synaptopodin, and DAPI (Supplementary Figure 3G).

Consistent with the results of the in vivo experiments, the HG-treated HPC showed increased levels of apoptosis-related proteins CHOP and cleaved caspase3 (Figures 4H,I). HG treatment also suppressed Bcl2 expression and induced Bax expression in vitro (Supplementary Figures 3D–F). Different from the results of the in vivo experiments, the increased p-PERK/PERK protein ratio was detected in HG-treated HPC, while no significant changes in the expression of IRE1α and ATF6β (Figures 4J–L). Thus, the results indicated that only the PERK pathway was activated in HG-treated podocytes. Moreover, flow cytometry examined cell apoptosis and showed that HG significantly provokes podocyte apoptosis (Figures 4M, N). These findings demonstrated that ER stress was associated with the progression of glomerular damage and podocyte injury in DKD, and activation of the PERK pathway may be the key factor of podocyte apoptosis in DKD.

JC-1 staining, DCFH-DA fluorescent probe, and ATP assay were conducted to further explore the HG-induced alterations in mitochondrial functions of podocytes. The mitochondrial membrane potential (MMP) is one of the indicators of mitochondrial functions. JC-1 staining revealed that HG exposure markedly decreased MMP (Figures 5A,B). DCFH-DA fluorescent probe was used to detect cellular oxidative stress, as shown in (Figures 5C,D), intracellular ROS generation was markedly increased in the HG-treated HPC. Furthermore, the ATP production decreased in HPC under HG conditions (Figure 5E). These results suggested that HG stimulation negatively affects mitochondrial functions.

Recent studies have revealed that PERK is located at the same membranes (MAMs) as Mfn2 (Verfaillie et al., 2012; Lebeau et al., 2018). Under basal conditions, Mfn2-PERK interaction keeps PERK in an inactive form, which is essential for maintaining ER homeostasis. The triple immunofluorescent staining (Mfn2, PERK, and DAPI) exhibited reduced colocalization of Mfn2 and PERK in glomeruli from diabetic rats and HG-treated HPC (Figures 5F,G). The co-immunoprecipitation assay revealed that Mfn2 physically interacts with PERK in HPC, which was in line with previous studies (Munoz et al., 2013). Additionally, HG treatment induced a dramatic reduction in Mfn2-PERK interaction (Figures 5H,I). These results indicated that HG activated the PERK pathway by downregulating Mfn2 expression; thus, reducing Mfn2-PERK interaction. Therefore, we speculated that the downregulation of Mfn2 expression might result in Mfn2-PERK interaction dissociating and PERK pathway activating in HG-treated podocytes. Subsequently, leading to MAMs reduction and mitochondrial dysfunction.

To assess the regulatory effects of the PERK pathway on mitochondrial functions and MAMs in vitro, the HPC were treated with PERK inhibitor (GSK2656157; 20μM, 24 h) in the presence of HG (25 mM, 24 h). We found that the treatment with GSK blocked PERK phosphorylation and downregulated the CHOP expression in HPC under HG conditions, but exhibited no effects on Mfn2 protein level (Figures 6A–C). GSK treatment also reversed HG-induced Bcl2 and Bax expression in vitro (Supplementary Figures 4A–C). Additionally, as shown in (Figures 6K, 6L), increased co-localization of TOM20 and CRT after GSK treatment was detected under HG conditions. Moreover, GSK attenuated the HG-induced mitochondrial dysfunction and apoptosis in HPC by restoring the MMP, reducing ROS generation, and up-regulating the ATP levels (Figures 6D–J). These results indicated that Mfn2 may lay upstream of PERK to regulate MAMs and mitochondrial functions.

Based on the above data, we identified Mfn2-PERK interaction as a potential site influenced by HG, and Mfn2 may regulate mitochondrial functions and MAMs through the PERK pathway. To further examine the underlying mechanisms, Mfn2-Myc plasmid or shRNA Mfn2 plasmid were transfected into the HPC to overexpress or silence Mfn2. The transfection of Mfn2-Myc plasmid reduced the phosphorylation of PERK and the expression of apoptosis-related proteins (cleaved caspase3 and CHOP), reversed the downregulation of Mfn2 in HG-treated HPC. Conversely, shRNA Mfn2 plasmid transfection increased the phosphorylation of PERK along with the upregulated expression of cleaved caspase3 and CHOP under normal conditions (Figures 7A–D). Consistent with previous speculations, Mfn2 overexpression restored Mfn2-PERK interaction, while Mfn2 silencing reduced the interaction under normal conditions (Supplementary Figure 5C).

The function of podocytes is largely based on maintaining the normal actin-based foot processes structure (He et al., 2013). As shown in (Supplementary Figure 5D), exposure to HG or Mfn2 silencing induced rearrangement of the actin cytoskeleton in HPC, which was prevented by the Mfn2 overexpression. Furthermore, overexpression of Mfn2 attenuated HG-induced mitochondrial dysfunction and apoptosis compared with the HG group. In contrast, Mfn2 silencing triggered mitochondrial dysfunction and apoptosis under normal conditions (Figures 7E–K). Moreover, the immunofluorescence staining showed that the HG + Mfn2-Myc group restored TOM20-CRT interaction while the shRNA Mfn2 group decreased the interaction (Supplementary Figures 5A,B). Taken together, these results demonstrated that Mfn2 plays a protective and anti-apoptotic role in protecting mitochondria and MAMs via suppressing the activation of the PERK pathway.