GRP75 Regulates Mitochondrial-Supercomplex Turnover to Modulate Insulin Sensitivity

Insulin resistance usually occurs in obesity and metabolic diseases, including type 2 diabetes (T2D), mainly due to impairment of insulin signaling pathways (1). Besides, low-grade inflammation accompanied by obesity has been considered as a vital factor to insulin resistance (2). As the key regulator of energy metabolism, mitochondria maintain cellular homeostasis by generating ATP through oxidative phosphorylation (OXPHOS). However, this makes it more susceptible to oxidative stress and other factors caused by mitochondrial dysfunction. Damage to mtDNA has been reported to increase insulin resistance in skeletal muscle cells of OGG1-knockout mice, and fragments of mtDNA may be mediators of inflammation by releasing to outside of mitochondria (3). However, the roles of mitochondrial dysfunction in insulin sensitivity and inflammation activation remain unclear.

Mitochondrial quality-control (MQC) systems play a central role in maintaining functional mitochondria. A major MQC-related chaperone is 75-kDa glucose-regulated protein (GRP75; also named mtHSP70/mortalin/PBP74), which is associated with multiple physiological disorders, such as myeloid leukemia and Parkinson disease (4,5). Loss of GRP75 function impairs mitochondrial function by inducing the aggregation of both mitochondria and mitochondrial proteins (4) as well as by increasing mitochondrial oxidative stress (6). In addition to an important member of MQC, GRP75 has a major role in the communication of endoplasmic reticulum (ER) and mitochondria by interaction with voltage-dependent anion channel 1 (VDAC1) at the mitochondrial membrane and inositol-1,4,5-trisphosphate receptor (IP3R) on the ER membrane. This multiprotein complex onto the mitochondria-associated membrane (MAM) is necessary for mitochondrial Ca²⁺ transport (7). Past research has documented that MAM integrity is important in the regulation of hepatic insulin action and resistance (8,9), and emerging evidence indicates that GRP75 might affect mitochondrial oxidative stress by modulating MAM integrity, but the roles of GRP75-containing MAMs in insulin resistance remain debated (9–11).

As the important targets of insulin, liver and adipose participated in many insulin-related pathways, and mitochondrial dysfunction in liver or adipose is strongly associated with insulin resistance and metabolic disorder (12–14). Recently, GRP75 expression was shown to be unaffected in the brain of diabetic rats (15) but increased and decreased, respectively, in the heart of diabetic rats (16) and the glomeruli of T2D KKAy mice (17). However, it is rarely known how the GRP75 expression changes in liver or adipose tissues of diabetic mice or other metabolic stress-induced mice. In this study, we assessed GRP75 function in insulin resistance by using two classic in vitro cell models, 3T3-L1 and AML12, derived from adipose tissue and liver, respectively. Our data indicated that GRP75 itself, rather than GRP75-associated MAM remodeling, is essential for safeguarding mitochondrial function through modulation of mitochondrial-supercomplex turnover. We further demonstrated that mitochondrial malfunction increases mitochondrial fragmentation, thus triggering the cytosolic mtDNA-sensing cGAS-STING–dependent inflammatory response. Furthermore, induction of Grp75 in vivo could prevent high-fat diet (HFD)-induced obesity and insulin resistance, suggesting that GRP75 can serve as potential therapeutic target of insulin resistance-related diabetes and other metabolic disease.

We obtained 5-week-old C57BL/6 mice from Shanghai Laboratory Animal Center (Shanghai, China) and housed them in the animal center of Wenzhou Medical University under a 12/12-h light/dark cycle in a controlled environment of temperature (22 ± 3°C). The animal-use protocol was approved by the animal center of Wenzhou Medical University. For Grp75 overexpression, mice were injected at 5 weeks and 10 weeks of age through the tail vein with adeno-associated viral 9 (AAV9) carrying a reverse transcript of full-length Grp75 cDNA from ViGene Biosciences (Shandong, China) to ensure the virus effect. Control mice were injected with AAV9 carrying empty vector at the same time. One week after the initial injection, all mice were fed a 60% HFD (Research Diet, New Brunswick, NJ) for 8 weeks. Body weight was measured weekly at the same time of the day during the feeding period. The liver, adipose, and muscle tissues of mice fed a normal diet or HFD for 12 weeks were derived from another study in the laboratory (18).

Glucose tolerance tests were performed after a 12-h fast. Glucose levels were measured using a glucometer at 0, 15, 30, 60, 90, and 120 min after the mice were injected with glucose (1.5 g/kg). Before the insulin tolerance test, mice were fasted 6 h. Glucose levels were measured at 0, 15, 30, 45, and 60 min after the mice were injected with regular human insulin (0.75 IU/kg).

During the last week of feeding, the mice were maintained in PhenoMaster metabolic cages (TSE Systems, Bad Homburg, Germany) for 7 days. Indirect calorimetry was used to measure VO₂, VCO₂, and energy expenditure of mice. Food intake was also evaluated on these days.

Retro-orbital blood of mice was collected, and serum was separated via centrifugation at 2,000g for 20 min at 4°C to measure insulin, triglycerides (TGs) (H203-1-1/A110-1-1; Nanjing Jiancheng Bioengineering Institute, Nanjing, China), and FGF21 (KE10042; Proteintech, Hubei, China) levels using ELISA kits according to the manufacturer’s instructions.

Isolated liver or fat tissues were fixed in 4% PBS-buffered paraformaldehyde for 24 h. The fixed tissues were dehydrated and processed for paraffin embedding, and 5-μm sections were cut, followed by staining with hematoxylin and eosin solutions (C0105-1 and C0105-2, respectively; Beyotime, Shanghai, China). For Oil Red O (A600395-0050; Sangon Biotech, Shanghai, China) staining, paraffin-embedded liver sections (10-μm thickness) were stained using a Cryostat Microm HM 525 (Thermo Fisher Scientific, Waltham, MA) prior to dehydration with 30% sucrose (V900116; Sigma-Aldrich, St. Louis, MO) for 48 h.

Mouse 3T3-L1 and AML12 cell lines were purchased from Cell Resource Center, Chinese Academy of Sciences (Shanghai, China), in 2015 and 2019, respectively. The cells were authenticated, by Genetic Testing Biotechnology (Suzhou, China), using short tandem repeat profiling analysis. 3T3-L1 cells were cultured in high-glucose DMEM (Sigma-Aldrich) containing 10% cosmic calf serum (Sigma-Aldrich) at 37°C in a humidified CO₂ incubator. AML12 cells were cultured in 90% 1:1 mixture of DMEM and Ham’s F12 medium (Corning, Corning, NY) with 5 μg/mL insulin, 5 μg/mL transferrin (Sigma-Aldrich), 5 ng/mL selenium (Sigma-Aldrich), and 40 ng/mL dexamethasone (Sangon Biotech) and supplemented with 10% FBS (Clark Bioscience, Shanghai, China) at 37°C in a humidified CO₂ incubator. The 3T3-L1 adipocyte differentiation was performed as previously reported (19).

Two Grp75 shRNAs (both against the ORF region) were used: shGrp75-1, 5′-GAGGCGTCTTTACCAAACTTA-3′ and shGrp75-2, 5′-GCCATACCTTACCATGGATGC-3′. The luciferase shRNA sequence was described previously (5). Two Grp75 stable-knockdown (KD) 3T3-L1 cell lines were generated by transfecting cells with shGrp75-1 and shGrp75-2 plasmids. Grp75 stable-KD AML12 cells were generated using shGrp75-1. For Grp75 overexpression, a reverse transcript of full-length Grp75 cDNA from 3T3-L1 cells was cloned into pcDNA 3.0 (from Prof. Danhui Liu). Control cells were transfected with empty vector. Stable Grp75-KD and -overexpressing (OE) cells were selected using 1,000 μg/mL G418 (Sangon Biotech) and 4 μg/mL puromycin (Sangon Biotech), respectively.

Cells were serum starved for 6 h and incubated with 0.1 mmol/L insulin plus 2-deoxy-2-[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]-d-glucose (2-NBDG, 100 μmol/L; Thermo Fisher Scientific) at 37°C in a CO₂ incubator for 30 min, and the fluorescence signal was measured after 30 min by using a Varioskan Flash Multimode Reader (Thermo Fisher Scientific) (excitation/emission: 485/540 nm). Basal glucose uptake was determined in cells exposed to 2-NBDG but not insulin.

Endogenous VO₂ in intact cells was measured using a Clark-type oxygen electrode (Oroboros Instruments, Innsbruck, Austria), as described (20). Approximately 0.8–1.2 million cells were added into the 2-mL chamber, and 2 μL 100 mmol/L oligomycin or 2 μL 0.1 mmol/L carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) was injected into the chamber to evaluate the mitochondrial proton leak and maximal respiration. All of the final values were normalized based on the cell numbers counted from the chamber when the detection was completed. For mitochondrial complex-dependent respiration, 6 million cells were collected and treated with 0.2% digitonin for 45 s, 25 μL of a 5:1 mixture of 1 mol/L glutamate and 1 mol/L malate, 2 μL of 1 mmol/L rotenone, 10 μL of 1 mol/L succinate, and 3 μL of 10 μmol/L antimycin A were sequentially added for the detection of complex-dependent oxygen respiration. For tissues, 80 μg of mitochondria, isolated using differential centrifugation from liver tissues, was placed into a respirometry chamber. Two separate protocols were applied at 37°C, and the oxygen flux rates were adjusted for mitochondrial concentrations. First protocol is for complex-dependent respiration similar with the cells before. The second protocol involved the consecutive addition of a mixture of complex I/II substrates, including 25 μL of 1 mol/L glutamate and 1 mol/L malate (5:1), and 10 μL of 1 mol/L succinate, 2 μL of 100 mmol/L oligomycin, and 2 μL of 1 mmol/L FCCP.

ATP, mitochondrial membrane potential (MMP), reactive oxygen species (ROS), and NAD-to-NADH ratio were measured using an ATP measurement kit (Thermo Fisher Scientific), the cationic fluorescent redistribution dye, tetramethylrhodaminemethyl ester (TMRM, Thermo Fisher Scientific), the MitoSOX and Carboxy-DCFDA dyes (both from Thermo Fisher Scientific), and NAD/NADH assay kit (Abcam, Cambridge, MA), respectively, as described previously (21). The relative levels of ATP, MMP, and ROS were normalized by protein concentration.

Mitochondrial Ca²⁺ was measured in cells by using Rhod-2-AM (Abcam). For imaging analysis, cells were cultured on coverslips and incubated in Hanks’ balanced salt solution containing 4.5 μmol/L Rhod-2-AM for 30 min. Images (magnification ×200) were captured using a confocal microscope (Nikon). At least three regions were quantitatively analyzed for each cell to generate the average fluorescence intensity by using ImageJ (National Institutes of Health, Bethesda, MD).

Mitochondria from cultured cells were isolated as described (22), with minor modifications. Briefly, 5–6 × 10⁷ pelleted cells were incubated with cold hypotonic homogenization buffer (3.5 mmol/L Tris-HCl, pH 7.8, 2.5 mmol/L NaCl, and 0.5 mmol/L MgCl₂ [all from Sigma-Aldrich]) for 1 min, and then homogenized using a glass homogenizer (Wheaton, Millville, NJ) until >80% of the cells were stained with trypan blue (Thermo Fisher Scientific). A 1:10 volume of hypertonic buffer (0.35 mol/L Tris-HCl, pH 7.8, 0.25 mol/L NaCl, and 50 mmol/L MgCl₂) was added into the homogenate, and crude mitochondria were isolated using a differential centrifugation method. Enzyme activity of four respiratory chain complexes and citrate synthase were measured in the mitochondria of 3T3-L1 cells as described (23); the activity of each complex was normalized against that of citrate synthase, a mitochondrial-matrix marker enzyme.

Whole-cell protein extracts were prepared using radioimmunoprecipitation assay lysis buffer (Cell Signaling Technology, Danvers, MA) supplemented with a protease-inhibitor cocktail (Sigma-Aldrich). Native mitochondrial-membrane proteins were solubilized in digitonin (Sigma-Aldrich) at a ratio of 20 mg wet-weight cells/2 mg digitonin in 40 μL of solubilization buffer (50 mmol/L NaCl, 50 mmol/L imidazole, 2 mmol/L 6-aminohexanoic acid, and 1 mmol/L EDTA [all from Sigma-Aldrich], pH 7.0 at 4°C), and supernatants containing membrane proteins were retained after centrifugation at 20,000g for 30 min. For analyzing mitochondrial OXPHOS complexes, 60 μg of native mitochondrial membrane protein samples containing 0.5% Blue G-250 (Sigma-Aldrich) and 5% glycerol were separated using Blue Native (BN)-PAGE (3–11% gels) as described (24). Proteins separated using BN-PAGE or SDS-PAGE were transferred to 0.22-μm polyvinylidene fluoride membranes (Bio-Rad, Hercules, CA) by using a semidry transfer system (Bio-Rad) and probed with primary antibodies: anti-GRP75 (sc-133137; 1:1,000), anti-GAPDH (sc-32233; 1:2,000), anti–β-actin (sc-47778; 1:5,000), Santa Cruz Biotechnology (Dallas, TX); anti-AKT1 (no. 2938; 1:1,000), anti–phosphorylated (p)-AKT Ser473 (no. 12694; 1:1,000), anti–p-AKT Thr308 (no. 5106; 1:1,000), anti-VDAC (no. 4661; 1:1,000), anti-AMPKα (no. 5832, 1:1,000), anti–p-AMPK (no. 2535, 1:1,000), anti-cGAS (no. 31659; 1:1,000), anti–nuclear factor (NF)-κB (no. 8242; 1:1,000), anti–p-NF-κB (no. 3033; 1:1,000), Cell Signaling Technology; anti-Grim19 (ab110240; 1:1,000), anti-SDHA (ab14715; 1:1,000), anti-UQCRC2 (ab14745; 1:1,000), anti–ATP synthase subunit-α (ab14748; 1:1,000), anti-MTCOI (ab14705; 1:1,000), anti-SOD1 (ab16831; 1:1,000), anti-DRP1 (ab184247; 1:1,000), anti-OPA1 (ab42364; 1:1,000), anti-MFN1 (ab57602; 1:1,000), anti-MFN2 (ab56889; 1:1,000), anti-TOMM20 (ab186734; 1:1,000), anti–MT-CYB (ab219823; 1:1,000), anti-COX IV (ab14744; 1:1,000), anti-ND1 (ab222892; 1:1,000), Abcam; and anti-TOMM70 (no. 14528-1-AP; 1:2,000), anti-IP3R (no. 19962-1-AP; 1:1,000), anti-STING (no. 19851-1-AP; 1:1,000), and anti–tumor necrosis factor (TNF)-α (no. 17590-1-AP; 1:1000), Proteintech (Rosemont, IL). The secondary antibodies used were alkaline phosphatase-conjugated anti-mouse IgG (no. 7056; 1:2,000) or horseradish peroxidase-conjugated anti-rabbit/mouse IgG (no. 7074/no. 7076; 1:2,000), from Cell Signaling Technology. Signals were detected using Clarity Western ECL Substrate (Bio-Rad) or 5-bromo-4-chloro-3′ indolyphosphate p-toluidine (20 mg/mL, Thermo Fisher Scientific)/nitrotetrazolium blue chloride (10 mg/mL, Sigma-Aldrich). Integrated optical density was quantified using a Gel-Pro Analyzer 4.0 (Media Cybernetics, Warrendale, PA).

In immunoprecipitation experiments, proteins from total cell lysates were solubilized with digitonin at a ratio of 20 mg wet-weight cells to 2 mg digitonin in 40 μL of solubilization buffer. GRP75 and IP3R1 was immunoprecipitated from 500 μg of digitonin-solubilized proteins by using Protein A/G PLUS-Agarose (Santa Cruz Biotechnology), according to the manufacturer’s instructions, and the immunocaptured proteins were subject to SDS-PAGE/immunoblotting for analyzing protein-protein interactions.

Cultured cells were fixed by 4% paraformaldehyde/PBS on coverslips, and in situ proximity ligation assay (PLA) (Sigma-Aldrich) experiments were performed at 37°C in a humidified incubator. Briefly, after 1 h blocking with blocking solution, anti-VDAC (1:250) and anti-IP3R1 (1:250) were incubated for 1 h. Subsequently, Duolink PLA probes were applied for 1 h, and ligation reaction and signal amplification were performed for 30 min and 150 min, respectively. Each fluorescent dot represents an interaction between VDAC1 and IP3R1. Quantification of signals (number of red dots per cell) was measured by ImageJ.

Cytosolic mtDNA content was measured as described in a previous report (25). Cells were each divided into two aliquots of equal volume. One was for total DNA extraction and the other for cytosolic fractions. DNA from pure cytosolic fractions was extracted using QIAquick Nucleotide Removal Columns (Qiagen, Hilden, Germany). Quantitative PCR was performed using mtDNA primers (Dloop1 to 3), and the cytosolic mtDNA content was normalized using the values obtained from whole-cell extracts.

Cells in 100-mm dishes were fixed with 2.5% glutaraldehyde (Shanghai Lanji Technology Development Company, Shanghai, China) for 2 h at 4°C, washed thrice with 0.1 mol/L PBS, postfixed in 1% tannic acid (SPI Supplies, West Chester, PA) for 1 h in the dark, washed thrice with 0.1 mol/L PBS and twice with double-distilled water, stained with 1% uranyl acetate (The 404 Company Limited, China National Nuclear Corporation, Lanzhou, China) for 1 h, and, lastly, dehydrated using alcohol (50%, 70%, 90%, 100%, 100%; 10 min each). Pretreated samples were sent to the Laboratory of Electron Microscope of Wenzhou Medical University for analysis. For MAM quantification, the distance from mitochondria to ER <100 nm on the image was recognized as MAM structure, and the ratio between MAM structure and perimeter of mitochondria and ER was evaluated. Five fields of each cell were quantified.

Cells were cultured on coverslips and fixed with 4% paraformaldehyde/PBS (20 min), permeabilized with PBS containing 0.2% Triton X-100 (3 min, room temperature), and incubated overnight with anti-HSP60 (1:300; Santa Cruz Biotechnology) or anti-LC3B (1:250; Cell Signaling Technology) at 4°C in a darkroom. The cells were washed with PBS and incubated with a fluorophore-conjugated secondary antibody, either IgG-Alexa Fluor 594 or IgG-Alexa Fluor 488 (1:300; Cell Signaling Technology), for 1 h at room temperature in the dark, and then stained with DAPI (Beyotime, Hangzhou, China) for 15 min at room temperature. Nine cells from three images (three cells/image) were used for each cell line (26). Fragmented mitochondria-containing cells were treated and incubated with anti-HSP60 as mentioned before, and the quantification methods were defined according to published criteria (27).

To perform transcriptome profiles of 3T3-L1 and AML12 Grp75-KD cells, total RNA were isolated and submitted to Novogene (Beijing, China) for sequencing. With the criterion of an adjusted P value <0.05, differentially expression genes were selected out from 3T3-L1 and AML12 Grp75-KD cells compared with corresponding control cells, respectively. All of the differentially expressed genes (DEGs) were shown by heat map drawing through the pheatmap package in R software. Pathway analysis was based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database using common DEGs in both KD cells.

All experiments were performed in triplicate and independently at least thrice. All images were captured (at magnification ×600) using a confocal laser-scanning microscope (Nikon, Telford, U.K.) and analyzed using ImageJ. Data are presented as means ± SD. All statistical analyses were performed using SPSS 21.0 software (IBM, Armonk, NY). Significance was estimated using the independent Student t test. A null hypothesis was rejected when P < 0.05.

Data that have been archived or can be obtained, upon reasonable request, from the corresponding author.

To understand the correlations between GRP75 and insulin sensitivity, GRP75 level was determined in an insulin-resistant mice model fed the HFD for 12 weeks. We found that GRP75 levels in liver and subcutaneous fat were significantly lower in C57BL/6J mice fed the HFD than in mice fed the standard diet (Fig. 1A and B); however, the other two MAM components, VDAC and IP3R, were largely not affected by the HFD (Fig. 1A). A consistent result was also found in insulin-resistant 3T3-L1 cells (Fig. 1C). These results suggested that GRP75-related MAM remodeling or change of MAM-independent GRP75 function may be associated with insulin sensitivity. Notably, the correlations between GRP75 and insulin sensitivity are likely tissue specific, while GRP75 level was not affected by the HFD in mice muscle (Supplementary Fig. 1). To further clarify the effect of GRP75 in insulin sensitivity, two and one Grp75-KD cell models were generated by using 3T3-L1 and AML12 cells, respectively, and one Grp75-OE cell model was established in 3T3-L1 cells (Supplementary Fig. 2). Our results showed that KD of GRP75 did not affect differentiation of 3T3-L1 cells (Supplementary Fig. 3) but led to decreased sensitivity to insulin-stimulated glucose uptake in 3T3-L1 adipocytes (Fig. 1D) compared with control cells. In 3T3-L1 adipocytes with Grp75 OE, a trend toward higher insulin-stimulated glucose uptake was detected compared with control cells (Fig. 1E). At the molecular level, insulin-stimulated pAkt Thr308/Ser473 levels were lower in Grp75-KD cells (Fig. 1F for 3T3-L1 adipocytes and Fig. 1G for AML12 cells) than in paired control cells, and 3T3-L1 adipocytes with Grp75 OE had a higher level of insulin-stimulated Akt phosphorylation than control 3T3-L1 adipocytes (Fig. 1H). Besides, pAMPK levels were higher in Grp75-KD cells (Fig. 1I for 3T3-L1 adipocytes and Fig. 1J for AML12 cells) than in paired control cells, and 3T3-L1 adipocytes with Grp75 OE had a lower level of AMPK phosphorylation than control 3T3-L1 adipocytes (Fig. 1K). The activation of AMPK is usually accompanied with the decreased insulin secretion. Thus, these findings provide both in vitro and in vivo evidences that GRP75 modulates insulin action.

GRP75 was linked to HFD-induced insulin resistance through its pivotal role in MAM integrity and MAM-associated mitochondrial calcium loading (9,10); however, neither KD nor OE of Grp75 affected the structure or formation of MAM in 3T3-L1 cells (Fig. 2A and B). GRP75 maintained MAM structure by interacting with ER protein IP3R and mitochondrial outer membrane protein VDAC; thus, we next asked whether the amount of the IP3R-GRP75-VDAC complex was altered due to artificial modulation of the GRP75 level. Coimmunoprecipitation experiments showed that interactions between IP3R and GRP75 (Fig. 2C), as well as GRP75 and VDAC (Fig. 2D), were not affected by KD of Grp75 in 3T3-L1 cells. Moreover, in situ PLA strongly indicated that MAM structures were not affected in our 3T3-L1 model with Grp75 KD since the number of IP3R-VDAC complexes were similar between control and Grp75-KD cells (Fig. 2E), although there seems to be a slight effect on ER stress along with the changes of expression of GRP75 in cells (Supplementary Fig. 4). The fact that mitochondrial calcium loading was not significantly altered in both 3T3-L1 (Fig. 2F) and AML12 cells (Fig. 2G) with KD of Grp75 compared with their matched control cells indicated the minimal effect of Grp75 KD in MAM. Together, these results suggest that KD of Grp75 minimally affected the structure and function of MAM. Other GRP75-related functions may be involved in the regulation of cellular insulin action.

In addition to its regulatory role in MAM formation, GRP75 is responsible for maintaining mitochondrial protein stability through mitochondrial protein folding and assembly machinery (28), partially of those proteins with complex structures. We therefore speculate that functions of mitochondrial OXPHOS complexes, the most complex super molecules in mammalian cells, may be regulated by GRP75. Notably, levels of both mitochondrial- and nuclear-encoded OXPHOS complex subunits were significantly lower in 3T3-L1 Grp75-KD than in control cells (Fig. 3A). Accordingly, the steady-state levels of mitochondrial complex I-, III-, and IV-containing respiratory chain supercomplexes were lower in 3T3-L1 (Fig. 3B) and AML12 (Supplementary Fig. 5) Grp75-KD in than control cells. Moreover, Grp75-KD 3T3-L1 cells have lower activities of respiratory chain complexes I and III than control cells, whereas activities of complex II and IV remain unaffected (Fig. 3C).

Inconsistent with the above results, mitochondrial complex I-dependent respiration showed a similar decline in Grp75-KD 3T3-L1 cells compared with control cells, while complex II-/III-dependent respiration had no obvious change (Fig. 3D). To ask further how GRP75 regulates the activity of OXPHOS complexes, we analyzed the mitochondrial-supercomplex turnover, which is known to be tightly controlled by the mitochondrial protein quality control system (29). To profile supercomplex-machinery assembly, we cultured cells in the presence of chloramphenicol (CAP, 40 μg/mL), a reversible mitochondrial translation inhibitor, for 7 days to exhaust existing supercomplex machinery by blocking the synthesis of mtDNA-encoded subunits of respiratory chain complexes; subsequently, we collected cells at various time points after CAP removal and analyzed respiratory complexes by using BN-PAGE. As shown in Fig. 3E, the assembling of mitochondrial supercomplexes was slower in Grp75-KD 3T3-L1 cells than in control cells. To examine whether GRP75 influences respiratory chain complex stability, cells were cultured with CAP for 24 h. In 3T3-L1 Grp75-KD cells, the percentage of degraded supercomplexes was significantly higher in Grp75-KD than in control cells (P = 0.056) (Fig. 3F), indicating that mitochondrial supercomplex is less stable in Grp75-KD cells compared with control cells. Altogether, our data indicate that GRP75 modulates mitochondrial supercomplex turnover.

Given that GRP75 regulates turnover of mitochondrial respiratory chain supercomplexes, we asked whether bioenergetic function was altered in cells with altered GRP75 levels. As shown in Fig. 4A and B, intracellular basal mitochondrial respiration, proton leakage associated respiration, and maximal respiration were significantly lower in Grp75-KD 3T3-L1 cells and AML12 cells than in respective control cells. Conversely, basal mitochondrial respiration, proton leakage, and maximal respiration were all significantly higher (relative to control) in 3T3-L1 cells stably overexpressing Grp75 (Fig. 4C). Consistently, ATP levels were lower in 3T3-L1 and AML12 cells after Grp75 KD relative to that in paired control cells (Fig. 4D and E), whereas ATP levels were higher in Grp75-OE cells than in control cells (Fig. 4F). Moreover, while both mitochondrion-derived and total ROS levels were higher in Grp75-KD 3T3-L1 (Fig. 4G and H) and AML12 cells (Fig. 4I and J) than in their paired control cells, ROS levels were significantly lower in Grp75-OE cells than in control cells (Fig. 4K and L). In parallel, the MMP showed a significant decline despite the expression change of Grp75 in 3T3-L1 and AML12 cells (Fig. 4M and O). However, no NAD-to-NADH ratio was detected in 3T3-L1 and AML12 cells after Grp75 KD or OE relative to that in paired control cells (Fig. 4P and R), suggesting the reverse electron transport is not present now. These results indicated that mitochondrial function was negatively positively regulated by GRP75.

Mitochondrial morphology and its turnover may be affected by an altered mitochondrial function. To test whether modulation of the GRP75 level affects mitochondrial morphology, we used transmission electron microscopy to examine changes in mitochondrial morphology, which revealed both swollen mitochondria and mitochondria featuring ruptured cristae in Grp75-KD 3T3-L1 cells compared with control cells (Fig. 5A); moreover, immunofluorescence confocal microscopy showed increased fragmented mitochondria-containing cells in Grp75-KD 3T3-L1 cells than control cells (Fig. 5B), both of which suggested Grp75 KD in 3T3-L1 cells led to abnormal mitochondrial morphology. Moreover, the level of mitofusin 1/2 (MFN1/2), a mitochondrial fusion marker, was lower in Grp75-KD 3T3-L1 cells (Fig. 5C) and AML12 cells (Supplementary Fig. 6A) than in respective control cells, suggesting a diminished activation of mitochondrial fusion in Grp75-KD cells. Although mitochondrial quality is lower in Grp75-KD cells, mitophagy is likely not affected in Grp75-KD cells with the fact that both mitochondrial-containing autophagosomes and autolysosomes did not differ between Grp75-KD cells and control cells (Fig. 5D and E), which was confirmed further by the fact that treatment of the cells with chloroquine did not change the decreased insulin-stimulated pAkt Thr308/Ser473 levels in Grp75-KD 3T3-L1 cells (Fig. 5F) and increased insulin-stimulated pAkt Thr308/Ser473 levels in Grp75-OE cells (Supplementary Fig. 6B). Moreover, mtDNA copy number (Fig. 5G) and mitochondrial mass (Fig. 5H and I and Supplementary Fig. 6C) were not affected by genetic modulation of Grp75 expression. These results indicate that KD of Grp75 impaired mitochondrial morphology without inducing mitophagy.

Change of mitochondrial function is closely associated with activation or repression of mitochondrial retrograde signaling (30), we therefore performed gene expression profiling in one Grp75-KD 3T3-L1 cell and AML12 cell with respective control cells. There were totally 12,010 and 4,028 DEGs in 3T3-L1– and AML12-KD cells compared with control cells, respectively (Supplementary Fig. 6A). To identify the signaling pathways related to Grp75 gene expression, we performed pathway analysis based on the KEGG database using the common DEGs in both KD cells, and found 640 upregulated and 890 downregulated, respectively (Supplementary Fig. 6B). Despite no significant signaling pathway enriched by downregulated genes (Supplementary Table 1), there were multiple pathways related to inflammation cytokines that were enriched by upregulated genes, such as the interleukin-17 signaling pathway, TNF signaling pathway, and JAK-STAT signaling pathway, demonstrating the activation of inflammatory response in Grp75-KD cells (Fig. 6A). While mtDNA release into cytosol is one of the key events to induce an inflammation response (31), we determined the release of mtDNA in 3T3-L1 and AML12 cells with KD of Grp75. As shown in Fig. 6B and C, increased mtDNA release was confirmed by the higher cytosolic–to–whole-cell DNA ratio in Grp75-KD 3T3-L1 (Fig. 6B) and AML12 cells (Fig. 6C) compared with paired control cells through separating the cytoplasm (Supplementary Fig. 6C). Cytosolic mtDNA has been considered as a cytosolic DNA sensor to activate the cGAS-STING pathway (25), thus promoting a TNF-mediated immune response (32). Therefore, we assessed the protein expression associated with the cGAS-STING pathway and TNF signaling pathway. The expression of cGAS, STING, and pNF-κB were obviously increased in 3T3-L1 and AML12 Grp75-KD cells compared with control cells (Fig. 6D and E). Moreover, the expression of TNF-α was also increased in 3T3-L1 and AML12 Grp75-KD cells than in control cells, demonstrating the inflammatory response activation. To mimic the impairment of mitochondrial function caused by Grp75 KD, we treated the 3T3-L1 cells with rotenone or antimycin A to inhibit the function of mitochondrial respiratory chain complex I and III, respectively. Notably, neither complex I nor complex III inhibition led to the increased fragmented mitochondrial (Fig. 6F), which was mimicked by Grp75 KD (Fig. 5B). Both treatments increased the cytosolic–to–whole-cell DNA ratio (Fig. 6G) and thus activated the cGAS-STING pathway (Fig. 6H) compared with control cells. We then asked whether Grp75 KD can induce the activation the cGAS-STING pathway via increasing mitochondrial fragmentation. By KD of Mfn2 in 3T3-L1 cells, we found that KD of Mfn2 induced mitochondrial fragmentation (Fig. 6I), mtDNA release (Fig. 6J), and activation of the cGAS-STING pathway (Fig. 6K). These results indicate that KD of Grp75 increased mitochondrial fragmentation via impairing mitochondrial function, which further promoted mtDNA release into cytosol and triggered a cGAS-STING–dependent proinflammatory response.

To further evaluate the role of Grp75 in insulin action in vivo, mice were administered adenovirus-Grp75 or adenovirus-Vector by tail vein injection using a recombinant AAV system and fed the HFD at the age of 6 weeks old. Grp75 OE was verified in mice liver while the expression of GRP75 in subcutaneous white adipose tissue (sWAT) was unchanged (Supplementary Fig. 8A and B); notably, we found that induction of Grp75 did not alter the VDAC level (Supplementary Fig. 8A). Mice with Grp75 OE had lower weight gain (Fig. 7A) and smaller body size (Fig. 7B) than control mice, mainly due to the decreased fat mass (Fig. 7C). The improved lipid metabolism was confirmed, while Grp75-OE mice have a lower level of TG and weight of sWAT and gonadal WAT (gWAT) compared with control mice fed the HFD (Fig. 7D and E). Moreover, the cell sizes of sWAT and gWAT (Fig. 7F) were significantly smaller in Grp75-OE mice than in control vector mice. Besides, lipid accumulation in the liver was also decreased in Grp75-OE mice more than in control vector mice (Fig. 7G). All these results revealed that induction of Grp75 could prevent HFD-induced whole-body lipid accumulation. Next, we assessed the whole-body metabolic rate of mice using indirect calorimetry. Grp75-OE mice displayed higher energy expenditure, especially in the dark period (Fig. 7H–J), while food intake remained unaffected (Fig. 7K). Furthermore, Grp75-OE mice displayed better glucose tolerance (Fig. 7L) and enhanced insulin sensitivity (Fig. 7M) even with a lower serum insulin level (Fig. 7N) compared with control mice. By focus on the liver, we found that Grp75-OE mice have an equal mitochondrial respiration (Supplementary Fig. 8C) but higher ATP content than control mice (Supplementary Fig. 8D). Besides, HFD-induced inflammatory stress was alleviated by Grp75 OE despite unchanged cGAS and STING expression (Supplementary Fig. 8E). Moreover, the serum FGF21 levels in the Grp75-OE mice are higher than those in control mice, although there is no obvious statistical difference (Supplementary Fig. 8F). Altogether, our results suggested that induction of Grp75 prevents HFD-induced obesity and insulin resistance via suppressing inflammatory stress and promoting liver-fat communication during HFD feeding.

GRP75 is a multifunctional protein: In mitochondria, GRP75 is an MQC chaperone that helps with the folding and subsequent assembly of newly synthesized proteins and the disassembly of damaged proteins, whereas in MAMs, GRP75 associates with VDAC1 and IP3-receptor type1 (IP3R1) to facilitate Ca²⁺ exchange between mitochondria and ER (33). Moreover, GRP75 binds to cytoplasmic P53 and promotes tumorigenesis (34) and to complement C9 and protects cells against complement-dependent cytotoxicity (35). Although the molecular mechanisms of GRP75 action have been extensively studied using cell models, the regulatory roles of GRP75 in human ailments other than cancer are poorly understood. Recently, genetic inhibition of GRP75 was shown to disrupt MAM integrity and induce insulin resistance (9). However, overenrichment of MAM contacts was shown to increase mitochondrial Ca²⁺ load and ROS generation (10,36) and contribute to insulin resistance and T2D. Notably, we found no data on GRP75-level changes in ob/ob mice or mice fed a HFD or a high-fat and high-sucrose diet for 16 weeks. Here, we found that GRP75 levels decreased in both the liver and fat from HFD-fed (12 weeks) mice. We did not examine GRP75 levels in ob/ob mice, and the GRP75 level was unaffected in muscle from HFD-fed mice, but improved glucose metabolism and increased GRP75 levels from vastus lateralis muscle were observed in patients with impaired glucose metabolism after a long-term exercise and dietary intervention (37). Accordingly, GRP75 levels were unaltered in MAM fractions from skeletal muscle of obese and diabetic mice, but genetic modulation of GRP75 expression altered the insulin effect in skeletal muscle (8). In this scenario, GRP75-level increase in patient muscle likely exerts an enhancer rather than normalization effect on insulin action, because GRP75 level is unaffected during the development of insulin resistance and T2D. We further showed that genetic modulation of GRP75 altered insulin-stimulated glucose uptake in vitro (3T3-L1 adipocytes) and in vivo (HFD-fed mice), which suggests that GRP75 levels contribute to increased insulin sensitivity.

Conflicting reports have been published of the MAM effect on insulin action, but they all agree that MAM might affect insulin action through alteration of mitochondrial function (8–10,38). Because GRP75 is predominantly localized in mitochondria (39), we believe that genetic modulation of GRP75 could directly affect mitochondrial functions unrelated to MAM. Furthermore, our preliminary data showed that mitochondrial maximal respiration decreased in live mitochondria from mice fed a high-fat and high-sucrose diet (4 weeks) (18), which argues that mitochondrial dysfunction might represent an independent early event of MAM miscommunication in specific diet-induced mouse models. The major function of the MAM component VDAC1/GRP75/IP3R1 complex is calcium exchange between ER reticulum and mitochondria, and MAM disruption and overenrichment lead to calcium decrease and overloading in mitochondria, respectively. Unexpectedly, we detected no change in calcium loading in mitochondria from Grp75-KD 3T3-L1 and AML12 cells when we used a mitochondrion-specific calcium probe, Rhod-2-AM (40). Furthermore, we performed classic electron microscopy as well as coimmunoprecipitation assays to test MAM integrity in Grp75-KD cells, and both results suggested that MAM integrity was unaffected by Grp75 KD. Therefore, mitochondrial function alteration due to genetic modulation of Grp75 in 3T3-L1 and AML12 cells, rather than MAM miscommunication, is likely the mechanism underlying the observed effect on insulin action.

The role of Grp75 in maintaining mitochondrial function has been extensively studied, and changes in intracellular ATP content and ROS production in Grp75-KD or -OE cells have been frequently reported (41,42), but how GRP75 regulates mitochondrial function is poorly understood. Here, we confirmed that genetic modulation of Grp75 affected intracellular ATP content and ROS production, which might be caused by changes in mitochondrial respiratory function. We further found that Grp75 KD in 3T3-L1 cells impaired the activity of respiratory chain complexes I and III. GRP75 was also shown to affect the activity of mitochondrial complexes IV and V (43,44). MQC system impairment by Grp75 KD might play a central role in the regulation of mitochondrial complex activity (4), but how functional mitochondrial complexes are maintained by GRP75 is unknown.

Here, we found that GRP75 is critical for maintaining mitochondrial respiratory chain supercomplex stability: Grp75 KD increased the degradation rate of supercomplexes per cell. However, how GRP75 stabilizes extremely high-molecular-weight supercomplexes is unclear. One possibility is that increased ROS products in mitochondria destabilize mitochondrial supercomplexes. Therefore, additional methods, such as complexome profiling, are required to completely elucidate the mechanism by which GRP75 affects mitochondrial complex/supercomplex turnover (45).

Mitochondrial dysfunction and turnover (fission/fusion, mitophagy) are highly interconnected, but no generally accepted causal relationships between these processes are recognized, and further investigation is also required to address the causal relationships between Grp75-KD–induced mitochondrial dysfunction and mitochondrial turnover (46,47). However, our results support the notion that mitochondrial dysfunction inhibits mitochondrial fusion (48,49). We speculate that Grp75-KD–induced mitochondrial malfunction leads to abnormal mitochondrial turnover, which might further impair mitochondrial function through a negative feedback loop.

The casual role of mitochondrial dysfunction in insulin resistance and T2D has been comprehensively characterized (50), and an increase in mitochondrial oxidative stress and ROS level has been implicated in insulin resistance (51). Here, we found that Grp75 KD increased both mitochondrial and cellular ROS products in AML12 and 3T3-L1 cells, whereas Grp75 OE limited ROS generation in cells. Increasing ROS levels in cells might form an attacking force of mitochondria, disturb the mtDNA stability, and make the mtDNA release to cytosol (52). In this study, the significantly increased cytosolic–to–whole-cell mtDNA ratio in Grp75-KD cells compared with control cells demonstrated there exists mtDNA damage and cytosolic mtDNA release. Cytosolic double-stranded DNA was found related to the activation of the cGAS-STING pathway and downstream proinflammatory response (53). Thus, GRP75 could be expected to regulate mitochondrial function and induce mtDNA release to cytosol and subsequently activate the inflammatory response pathways. Unfortunately, we did not find the same performance of GRP75 regulation in vivo. Although the OE of GRP75 in vivo showed significantly decreased body weight, hepatic lipid accumulation, and increased insulin sensitivity compared with the control mice fed the HFD, there was no obvious changed hepatic mitochondrial oxidative phosphorylation and turnover. We do not rule out that the increased insulin sensitivity is due to the body weight differences between GRP75-OE and control mice. Perhaps there exists other unknown mechanisms, such as a secondary endocrine (54) or centrally mediated effect (55) in vivo, that regulate the body phenotype and insulin actions, which requires more research in the future.

In summary, we have presented a MAM-unrelated role of GRP75 in insulin sensitivity. Our findings support a positive association between GRP75 and insulin action in vitro and demonstrate that GRP75 affects mitochondrial function by modulating mitochondrial-complex/supercomplex turnover, which further regulates mitochondrial fragmentation and contributes to insulin sensitivity through the cytosolic mtDNA-sensing cGAS-STING dependent proinflammatory response.

Acknowledgments. The authors thank all participants for their involvement in the study.

Funding. This work was supported by the General Program of the National Natural Science Foundation of China (82072338 to H.Z. and 82072366 to J.L.), Key Program of the National Natural Science Foundation of China (81830071 to J.L.), the Key Discipline of Zhejiang Province in Medical Technology (First Class, Category A) and Wenzhou Municipal Science and Technology Bureau (Y20210107 to H.Z.).

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

Author Contributions. Q.Z., T.L., F.G., Y.F., B.L., X.S., H.C., L.J., and D.C. performed the in vitro assays. Q.Z. and H.F. wrote the manuscript. T.L., Y.F., B.L., X.S., H.C., Z.Z., and S.G., performed the mice work. F.G. and L.S. performed the imaging work. H.Z., J.L., and H.F. conceived the study and designed the experiments. Q.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.