Approval for this study was obtained from the Ethics Committee of the Metabolic Disease Hospital of Tianjin Medical University, Tianjin, China and informed consent was obtained from all participants prior to orthopedic surgery. This trial has been verified by the Chinese Clinical Trial Registry, and its registration number is ChiCTR-COC-15007025. All participants were patients with lumbar disc herniation who required orthopedic surgery. Among these patients, 10 patients suffered from T2DM and the another 10 were non-diabetic patients. The patients with diabetes had been previously diagnosed at the hospital. Patients with cardiovascular disease, neoplastic disease, neurodegenerative disease and acute inflammation were excluded from the study.

Blood samples were collected from the antecubital vein at the first day after admission. All blood was drawn from the patients at the same time in the morning (between 6 and 8 a.m.). Laboratory test results were evaluated, including fasting plasma glucose (FPG), blood urea nitrogen (BUN), creatinine, aspartate aminotransferase (AST), alanine aminotransferase (ALT) and glycated haemoglobin A1c (HbA1c), levels, and lipid profiles, including total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) and triglyceride (TG) levels. All the laboratory parameters were assessed using conventional laboratory methods. The blood samples were centrifuged for 15 min at 3,000 rpm (AU5400; Olympus, Tokyo, Japan), and plasma was assayed for the laboratory parameters using an automated analyzer (7600A-020; Hitachi, Tokyo, Japan)

Between 300 and 500 mg of paravertebral skeletal muscle was obtained from patients undergoing lumbar vertebral disc decompression discectomy. Dissections of skeletal muscle were obtained within 10 min of delivery and snap-frozen in liquid nitrogen and stored at −80°C until further analysis. Tissues were also embedded in paraffin for histological analysis by hemotoxylin and eosin (H&E) staining. For all surgerical procedures, a spinal anesthesia and/or epidural were used. All muscle samples were obtained at the time of surgery (between 8:30 a.m. and 13:00 p.m.). All patients were fasted overnight.

Male Sprague-Dawley (SD) rats (n=90, 6 weeks old), weighing 161±9 g, were purchased from Beijing Hua Fu Kang Biotechnology Co., Ltd. (Beijing, China). Rats were given free access to food and tap water and were caged individually under a 12-h light-dark cycle at a temperature of 22±3°C and humidity of 55±5%. All animal experiments were conducted in accordance with the Principles of Laboratory Care, and approved by the Institutional Animal Care and Use Committee, Metabolic Disease Hospital of Tianjin Medical University.

The rats were randomly divided into the control (NC) and the high energy diet (HED) groups. In the control group, the animals received a standard chow diet, while the rats in the HED group were fed with an additional high energy emulsion, as previously described (22,23). After 8 weeks on their respective diets, streptozotocin (STZ; 45 mg/kg; Sigma, St. Louis, MO, USA) dissolved in 0.1 mol/l citrate buffer (pH 4.5) was injected into the caudal vein of the rats in the HED group to establish a model of T2DM, while the rats in the control group were injected with sodium citrate buffer. The rats with blood glucose levels ≥16.7 mmol/l at 7 days after the STZ injection were selected as the model of diabetes. On average, 80% of the rats injected with STZ met these criteria. At 1 week following the injection of STZ, the rats with successfully-induced diabetes were randomly divided into the diabetes model (DM) group, the celastrol low-dose group (1 mg/kg/day), the celastrol middle-dose group (3 mg/kg/day) and the celastrol high-dose group (6 mg/kg/day) (n=15 rats per group). The rats in the treatment groups were administered celastrol by gavage, whereas the rats in the NC and DM groups were administered an equal amount of distilled water (2 ml). Following 8 weeks of the respective treatments, rats were anesthetized with an intraperitoneal injection of sodium pentobarbital (30 mg/kg body weight) and tissue samples were collected for analysis. The paravertebral muscle was excised from the rat bodies, and was cut perpendicularly along the longitudinal axis and fixed in phosphate-buffered 20% formaldehyde. Histological paraffin-embedded sections (5 µm) were then prepared for H&E staining. The sections of paravertebral muscle were snap-frozen in liquid nitrogen and stored at −80°C until further analysis.

Body weight and fasting blood glucose (FBG) levels were measured each week. The overnight-fasted rats were weighed by electronic balance (AM1100; Mettler-Toledo AG, Schwerzenbach, Switzerland) at the same time in the morning (between 6 and 8 a.m.). Then blood samples were obtained from the caudal vein and FBG levels were measured using a Accu-Chek Aviva glucometer (Roche Diagnostics, Mannheim, Germany). After 8 weeks of treatment, the rats were euthanized. Blood samples were obtained from the retroorbital venous plexus by using plain microhematocrit capillary tubes tubes (VWR, West Chester, PA, USA) and collected into tubes containing EDTA at the time of sacrifice and were centrifuged at 3,000 × g/min for 15 min. Plasma was separated for measuring FPG, BUN, serum creatinine (SCr), AST, ALT, TC and TG using an automatic biochemistry analyzer (CD-1600CS; Abbott Labs, North Chicago, IL, USA).

Mitochondrial-enriched supernatants were prepared from frozen skeletal muscle samples as previously described (24). The content of malondialdehyde (MDA) and GSH was measured by enzyme-linked immunosorbent assay (ELISA) using respective kits following the manufacturer's instructions (Jiancheng Co. Ltd., Nanjing, China). MnSOD activity was determined by spectrophotometry. Briefly, reaction buffer (50 mM sodium phosphate, 0.1 mM EDTA, 0.01 mM xanthine, and 0.01 mM cytochrome c, pH 7.8) was mixed with 0.005 U/ml xanthine oxidase and 2 mM sodium cyanide, and the absorbance change at 550 nm was followed. Then sample was added stepwise (in 20 µl increments at a concentration of 5–10 mg/ml), and the concentration of sample required to decrease the rate of reaction by 50% (defined as 1 unit of MnSOD activity) was calculated. Enzyme values are presented as units per milligram of protein.

Mitochondrial-enriched supernatants were prepared from frozen skeletal muscle samples as described (24). SIRT3 enzyme activity in the gastrocnemius mitochondria was assayed using a fluorometric kit (Biomoles, Inc., Shoreline, WA, USA) as per the manufacturer's instructions. Briefly, 25 µl distilled water, 5 µl buffer, 5 µl fluorosubstrate peptide 5 µl NAD and 5 µl developer were added to each microtiter plate wells and mixed well. Reactions were initiated by adding 5 µl samples or buffer of samples or recombinant SIRT3 to matching well and mixing thoroughly. SIRT3 activity was measured using a fluorimetric microplate reader at 450 nm with an excitation of 350 nm. We measured and calculated the rate of reaction, while the reaction velocity remained constant. The enzyme activity was normalized to the total protein content of each sample and the results are expressed relative to the mean for the NC group.

Protein lysates were obtained by homogenizing paravertebral muscle with lysis buffer. The protein concentration was measured using the Bio-Rad protein assay (Bio-Rad, Richmond, CA, USA). Approximately 50 µg protein was subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto nitrocellulose membranes. After blocking with 5% dried milk in Tris-buffered saline (TBS) containing 0.1% Tween-20 for 2 h, the membranes were subsequently incubated overnight with primary antibodies diluted in 5% dried milk-TBS containing 0.1% Tween-20. The primary antibodies were as follows: anti-AMPK (#2532, 1:1,000), anti-phosphorylated (p-)AMPK (:#2531, 1:1,000) (both from Cell Signaling Technology, Danvers, MA, USA), anti-α-tubulin (sc-8035, 1;10,000), anti-β-actin (sc-47778, 1:10,000) (both from Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), anti-SIRT3 (ab86671, 1:500), anti-MnSOD (ab13534, 1:2,000) and anti-PGC1α (ab54481, 1:1,000) (all from Abcam Cambridge, MA, USA) antibodies. The membranes were then incubated with appropriate horseradish-peroxidase-conjugated secondary antibodies (sc-2005 and sc-2003, Santa Cruz Biotechnology, Inc.). An enhanced chemiluminescence system was used to visualize the bands. Densitometric analysis was performed using a gel image analysis system (UVP Inc., San Gabriel, CA, USA).

The MnSOD, Sirt3 and PGC1α mRNA levels were quantified by SYBR-Green Real-Time PCR (Invitrogen, Carlsbad, CA, USA). RT-qPCR was performed as previously described (25). RNA was extracted from paravertebral muscle using TRIzol reagent (Invitrogen). Reverse transcription was performed using the RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific, Inc., Waltham, MA, USA). The quantitative PCR (qPCR) measurement of individual cDNAs was carried out using SYBR-Green dye to measure duplex DNA formation with the LightCycler System (Roche Diagnostics). The data were analyzed using Bio-Rad CFX manager software 1.6. The quantified results were normalized to those of GADPH, using the 2-(ΔΔCT) method. The nucleotide sequences of the PCR primers used were as follows: MnSOD forward, 5′-ACTGAAGTTCAATGGTGGGG-3′ and reverse, 5′-GCTTGATAGCCTCCAGCAAC-3′; Sirt3 forward, 5′-TACAGAAATCAGTGCCCCGA-3′ and reverse, 5′-GGTGGACACAAGAACTGCTG-3′; PGC1α forward, 5′-ATGAGAAGCGGGAGTCTGAA-3′ and reverse, 5′-GCGGTCTCTCAGTTCTGTCC-3′; GAPDH forward, 5′-TGCCACTCAGAAGACTGTGG-3′ and reverse, 5′-TTCAGCTCTGGGATGACCTT-3′.

All data are presented as the means ± SD. Differences between groups were analyzed using the Student's t-test and one-way analysis of variance (ANOVA) test using the statistical software. SPSS 18.0 (SPSS Inc., Chicago, IL, USA). Values of p<0.05 were considered to indicate statistically significant differences.

There were no significant differences in gender, age and body weight between the normal control and diabetic patients. The FPG, HbA1c, TG, TC and LDL-C levels were significantly higher in the diabetic patients compared to the normal controls (all P<0.01). Although the HDL-C levels were decreased in the diabetic patients, there were no significant differences between the 2 groups (Table I).

H&E staining of the paraspinal muscle of non-diabetic patients was presented as normal (Fig. 1A and C), whereas the skeletal muscle of the diabetic patients exhibited an irregular fiber structure, with wide gaps, nuclei of diverse sizes, an increased number of nuclei and some with an abnormal position, partly inserted in the muscle fibers (Fig. 1B and D).

Both an increase in ROS production and a decline in ROS clearance can lead to increased mitochondrial oxidative stress (26). In the present study, we focused on MnSOD, the primary antioxidative enzyme in mitochondria. Compared to the control group, MnSOD enzyme activity in the diabetes group was decreased by approximately 50% (Fig. 2A). We also found that the diabetic patients had lower mRNA levels of MnSOD by approximately 50% (Fig. 2B) and a decreased protein content of MnSOD (Fig. 2C) compared to the non-diabetic control group.

It has been demonstrated that SIRT3 and MnSOD protein abundance is regulated in a signaling axis involving both AMPK and PGC1α (27). Thus, in this study, we compared the expression of the AMPK-PGC1α-SIRT3 signaling pathway between diabetic and non-diabetic patients. As shown in Fig. 3, compared with the non-diabetic patients, the PGC1α and SIRT3 mRNA levels were decreased by almost 63 and 47%, respectively in the diabetic patients (Fig. 3A and B). The protein levels of PGC1α (Fig. 3D) and SIRT3 (Fig. 3E) were similar to those of the mRNA levels, and were decreased by 60 and 50% in the diabetic patients, respectively. Western blot analysis revealed an approximate 60% decrease in AMPK phosphorylation levels in the diabetic patients, which was determined by the p-AMPK levels normalized to the total AMPK levels (Fig. 3C).

According to the results of our clinical research, we found that the levels of MnSOD were reduced in the skeletal muscle of diabetic patients, and that the AMPK-PGC1α-SIRT3 pathway was downregulated in the skeletal muscle of diabetic patients. In view of the antioxidant effects of celastrol and that it is widely used in the treatment of diabetic nephropathy, we examined the antioxidant effects of this drug in the skeletal muscle of diabetic rats. We preliminarily investigated whether the antioxidant effects of celastrol are mediated through the AMPK-PGC1α-SIRT3 signaling pathway, and compared the effects of different doses of celastrol on this pathway.

Table II shows the biochemical results for each group. As expected, the FPG concentrations were significantly higher, while body weight was lower in the diabetic rats compared to the normal control rats throughout the experimental period. However, there were no significant differences in the plasma levels of ALT, AST, BUN and SCr between the normal control rats and the diabetic rats. Of note, the TG and TC content increased significantly in the diabetic rats compared with the normal control rats due to the high fat diet (HED). Although it has been reported that treatment with celastrol significantly decreases body weight, blood glucose levels, and plasma TC and TG levels (28), 8 weeks of treatment with various doses of celastrol had no significant effect on body weight, the FPG level, and plasma TC and TG levels in the present study.

H&E staining of the paraspinal muscle of the rats in the NC group exhibited a normal morphology (Fig. 4A and F). However, while the skeletal muscle of the rats in the DM group exhibited an irregular fiber structure, with wide gaps, nuclei of diverse sizes, an increased number of nuclei and some with an abnormal position, partly inserted in the muscle fibers (Fig. 4B and G). Following treatment with celastrol, pathological damage was attenuated in varying degrees compared with the rats in the DM group in a dose-dependent manner (Fig. 4C–E and H–J).

Markers of oxidative stress had been found to be altered in mice with type 2 diabetes and in individuals with diabetes (29,30). Thus, in this study, we analyzed mitochondrial oxidative stress by measuring the levels of MDA, GSH and MnSOD.

The level of MDA in the DM group was 2-fold higher than that of the NC group. Following treatment with 3 and 6 mg/kg celastrol, the levels of MDA were significantly decreased by 35.2 and 36.7% (P<0.05), respectively; while there was no significant difference in the levels of MDA between the DM group and the group treated with celastrol 1 mg/kg (Fig. 5A). By contrast, the level of GSH was decreased by 59.2% due to the onset of diabetes. Treatment with 3 and 6 mg/kg celastrol markedly restored the GSH level (P<0.05) to almost normal levels. However, treatment with 1 mg/kg celastrol had no significant effect compared with the DM group (P>0.05; Fig. 5B).

As an antioxidant defense mechanism, increased MnSOD can partly indicate the attenuation of oxidative stress (31). In this study, we found that the diabetic animals had 64.2 and 52.5% lower mRNA and protein levels of MnSOD, respectively compared to the NC group. Treatment with 3 and 6 mg/kg celastrol significantly increased these levels. The results of western blot analysis also indicated that the MnSOD protein level was upregulated in the paravertebral muscle of rats treated with 3 and 6 mg/kg celastrol (Fig. 6A and B). Furthermore, we analyzed the enzyme activity of MnSOD. We also found that treatment with 3 and 6 mg/kg celastrol markedly enhanced the enzyme activity of MnSOD (Fig. 6C). However, neither the expression nor the enzyme activity of MnSOD exhibited a significant increase following treatment with 1 mg/kg celastrol. Taken together, these findings indicate that celastrol attenuates oxidative stress in a dose-dependent manner.

To determine whether celastrol ameliorates oxidative stress by regulating SIRT3, we assessed the enzyme activity and content of SIRT3 in the paravertebral muscle of experiment rats. As expected, the mRNA and protein level of SIRT3 in the DM group was decreased by 68.4 and 65.3%, respectively, compared to the NC group. Treatment with 3 and 6 mg/kg celastrol significantly increased both the mRNA and protein level of SIRT3 (Fig. 7A and D). Furthermore, we analyzed the enzyme activity of SIRT3. A reduction of 41.9% in SIRT3 enzyme activity was observed in the DM group compared with the NC group. Treatment with 3 and 6 mg/kg celastrol markedly enhanced the enzyme activity of SIRT3 (Fig. 7C). Consistent with the results obtained for MnSOD, neither the expression nor the enzyme activity of SIRT3 exhibited a marked improvement following treatment with 1 mg/kg celastrol.

It has been found that SIRT3 functions as a downstream target of PGC1α, which is directly regulated by AMPK (14,15). Therefore, we examined the possible role of the AMPK-PGC1α signaling pathway as a modulator of the regulatory effects of celastrol on SIRT3. The mRNA and protein levels of PGC1α were decreased in the DM group by 66.3 and 64.9%, respectively compared to the NC group (Fig. 7B and E). Treatment with 3 and 6 mg/kg celastrol significantly increased the levels PGC1α. In accordance with this upregulation, we also observed increased p-AMPK (Fig. 7F) levels in the paravertebral muscle of rats treated with 3 and 6 mg/kg celastrol. Similarly, there was no significant increase in the levels of PGC1α and p-AMPK in the group treated with 1 mg/kg celastrol. Taken together, our results indicated that celastrol increased the expression of p-AMPK and PGC1α in a dose-dependent manner in the skeletal muscle of rats with diabetes. Celastrol activated the AMPK-PGC1α signaling pathway in vivo.