Salidroside (purity > 98%) was purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). Compound C was from Selleck (Houston, TX, United States).

All animals were housed in a temperature-controlled room with a 12-h/12-h light/dark cycle and were allowed free access to food and water during the course of experiments. Before the experiment, the mice were kept for 1 week to acclimatize them to the conditions. The 4-week-old male C57BL/6 mice (SLAC Laboratory Animals, Shanghai, China) were fed a high-fat diet (HFD) (p1400f, SLAC Laboratory Animals, Shanghai, China) (n = 16) or normal chow diet (n = 8). After 10 weeks of the HFD, salidroside intervention (100 mg/kg/day) was initiated by gavage once a day for 5 weeks. The control groups were given vehicle (saline). The 4-week-old male C57Bl/KsJ (BKS) mice (wild type, n = 8) and BKS.Cg-Dock7m +/+ Leprdb/J (db/db) mice (n = 16) were ordered from Model Animal Research Center of Nanjing University (Nanjing, China). Salidroside (100 mg/kg/day) was administrated orally by gavage once a day for 5 weeks. The control groups were given vehicle (saline). Fasting blood glucose and body weight of mice were monitored every 5 days. Glucose measurements were performed on blood drawn from the tail vein using a Glucometer (Accu-Chek Active; Roche Inc.). All animal experiments were conducted in accordance with Provisions and General Recommendation of Chinese Experimental Animals Administration Legislation and approved by the Research Animal Care Committee of Nanjing University of Chinese Medicine.

After 5-week salidroside treatment, mice were fasted 12 h overnight for oral glucose tolerance test (OGTT) experiments. In the next morning, mice were taken glucose orally at a dose of 2 mg/g body weight. Then blood samples of mice were obtained at time points 0, 30, 60, 90, and 120 min for blood glucose measurements using a Glucometer (Accu-Chek Active; Roche Inc.).

Blood and tissue samples of mice were collected after 5-week salidroside treatment, and tissue samples were weighed. Serum insulin level was determined using a mouse insulin ELISA kit (RayBiotech, Norcross, GA, United States). Contents of total cholesterol (CHO), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerol (TG) in serum were tested by commercial assay kits (Jiancheng Bioengineering Inc., Nanjing, China). Enzyme activities of Superoxide dismutase (SOD), Glutathione peroxidase (GPx), Catalase (CAT), and content of Malonaldehyde (MDA) were assessed using tissue lysate of adipose and pancreas by commercial assay kits (Jiancheng Bioengineering Inc., Nanjing, China). All measurements were performed and calculated according to the instructions of the assay kits.

Pancreas of mice were weighed, fixed with 4% paraformaldehyde and then embedded in paraffin to prepare 4 μm pancreatic sections. In brief, the paraffin sections of mouse pancreases were deparaffinized, rehydrated, antigen unmasking (Vector Laboratories, Inc., United States), blocking, and incubated overnight at 4°C with anti-insulin antibody (1:200 dilution) (ab7842, Abcam). In the next day, the sections were washed and stained by FITC secondary antibody. Then the β-cell areas and the entire tissue areas were scanned by a Nikon MEA53200 (Nikon, Japan) microscope. The cross-sectional areas of pancreas and β-cells were determined and calculated by NIS-Elements software (Nikon). The average β-cell mass of each mouse was calculated as follows: the islets/pancreas area ratio × the pancreatic weight. Five consecutive sections from each pancreas (6 mice per group) were used for β-cell mass measurements.

Mouse islets were isolated from 10-week-old male C57BL/6 mice or 10-week-old male dbdb mice by common bile duct perfusion using Collagenase type 4 (Sunshine Biotechnology, China). Infuse 2 ml of collagenase solution (2 mg/ml) into a mouse pancreas. Transfer infused pancreas to 2 ml collagenase solution in 50-ml falcon tube and digest at 37°C for 20–25 min in water bath. Shake tubes every 5 min. Stop the reaction by adding 30 ml ice-cold quenching buffer and shake, then centrifuge at 1200 rpm for 3 min at 4°C. Remove the supernatant, and repeat wash step by quenching buffer twice. Resuspend pellets with 13 ml of cold Histopaque-1077 (Sigma) by pipetting. Fill to 25 ml with cold serum (-) RPMI media carefully with transfer pipettes. Add media slowly and gently to make clear 2 layers. Centrifuge 2500 rpm, 24 min, 4°C (Slow acceleration, no brake). Pick up floating islets in the interface between media and Histopaque with transfer pipettes, then wash islets twice using full media. Isolated islets were transferred to 6-well plates (20–30 islets/well) and cultured in RPMI 1640 containing 11.1 mmol/l glucose, 100 U/ml penicillin, 100 mg/ml streptomycin, and 10% FBS (All from Gibco). The Min6 cell line was a kind of gift from Prof. Dongming Su (Nanjing Medical University, China) and maintained in 5 mM glucose DMEM, supplemented with 10% FBS, 50 μmol/L b-mercaptoethanol (Sigma), 100 U/ml penicillin, and 0.1 mg/ml streptomycin in 5% CO₂ at 37°C.

Cultured mouse islets in 6-well plates were fixed with 4% paraformaldehyde followed by permeabilization with 0.5% Triton X-100. The 4-μm paraffin sections of mouse pancreas were deparaffinized, rehydrated, and antigen unmasking. Then islets or sections were incubated with blocking buffer for 1 h at RT followed by incubating overnight at 4°C with antibody of insulin (Abcam), or Ki67 (1:50) (BD Pharmingen), or PDX1 (1:100) (Abcam). In the next day, islets or sections were washed by TBS and incubated by FITC- or Cy3-conjugated secondary antibodies (1:400) (Abcam) for 1 h at RT. β-cell apoptosis of pancreatic setions was detected by the commercial TUNEL staining kit according to the manufacturer’s instructions (TUNEL Brightred Apoptosis Detection Kit,Vazyme Biotech Co.). β-cell apoptosis of cultured islets was analyzed by the commercial TUNEL staining kit (In Situ Cell Death Detection Kit, TMR red; Roche Diagnostics). Slides were mounted with Vectashield Mounting Medium containing 4′6-diamidino-2-phenylindole (DAPI) as nuclear dye (Vector Laboratories, Inc., United States). Fluorescence was detected and analyzed using an Nikon MEA53200 (Nikon) microscope.

Nuclear and cytoplasm extractions of Min6 cells were performed according to the instructions of NE-PER Nuclear and Cytoplasm Extraction Reagents (Pierce biotechnology, United States). For each treatment, 1 × 10⁶ Min6 cells were used for nuclear fractionation. The purity of fractions was analyzed by probing the membranes with anti-GAPDH for cytosolic and anti-PARP for nuclear extracts.

For acute insulin release, cultured mouse islets (20–30 islets/well in 6-well plates) were washed by PBS twice and pre-incubated in 1 ml Krebs-Ringer bicarbonate buffer (KRB) containing 2.8 mM glucose for 30 min. The KRB was removed and replaced by 1 ml KRB containing 2.8 mM glucose for 1 h incubation (collected this 2.8KRB as basal). Then islets were incubated in KRB containing 16.7 mM glucose for 1 h (collected this 16.7KRB as stimulated). For insulin content, cells were extracted with 0.18 N HCl in 70% ethanol. Insulin contents were determined using a mouse insulin ELISA kit (RayBiotech, Norcross, GA, United States). The insulin stimulatory index of islets is calculated as stimulated/basal. Three wells per treatment group were used.

Min6 cells were washed in PBS and lysed (Beyotime Biotech, Shanghai, China). For each treatment, 20 μg proteins were used for western blotting analysis. The following primary antibodies were used at 1:1000 dilution: p-JNK (#9255), JNK (#9258), Actin (#4967), c-Casp3 (#9661), p-FOXO1 (#9461), FOXO1 (#9462), p-AMPKα (#2535), AMPKα (#2532), p-AKT (#9271), PARP (#9542), GAPDH (#2118), (all from Cell Signaling); NOX2 (Abcam, #ab39072), PDX1 (Abcam, #ab47267) followed by incubation with horseradish-peroxidase-linked IgG peroxidase (1:5000, Vazyme Biotech, Nanjing, China). The bands were visualized and densities of the bands were analyzed using Tanon ChemImaging Systems (Tanon Science & Technology Co., Ltd., Shanghai, China).

To detect the ROS production in Min6 cells with different treatments, 1 × 10⁶ Min6 cells were collected, washed by PBS, and re-suspended in 500 μl serum (-) DMEM containing 10 μM DCFDA (Beyotime Biotech, China) for 30 min at 37°C in the dark. Then cells were washed 3 times with PBS and analyzed with flow cytometry (Becton Dickinson) with excitation set at 488 nm and emission at 530 nm.

The ΔΨm changes in Min6 cells with different treatments were measured by uptake of rhodamine 123 (Rh123). The 1 × 10⁶ treated cells were harvested and washed twice with PBS, re-suspended in 500 μl serum (-) DMEM containing 2 μM Rh123 (Beyotime Biotech, China), and incubated at 37°C for 30 min in the dark. The samples were then immediately detected by flow cytometry (Becton Dickinson) with excitation set at 503 nm and emission at 527 nm.

Data are presented as means ± SD and were analyzed by paired, Student’s t-test or by analysis of variance with a Bonferroni correction for multiple group comparisons.

It was reported that salidroside (100 mg/kg/day) showed strong glucose lowering effect on db/db mice which was similar to effect of metformin (200 mg/kg/day) (Zheng et al., 2015). For this reason, the dose of 100 mg/kg/day salidroside was used in our experiments.

To prevent progression of diabetes, we initially investigated effects of salidroside in 4-week db/db mice which are considered to be pre-diabetic. As showed in Figure 1A, salidroside could not significantly alleviate the increase of blood glucose in db/db mice in the first 15 days. However, after 21-day treatment, db/db mice were protected from developing severe hyperglycemia by salidroside and this effect was sustained in following days compared to vehicle db/db group. Moreover, there were no detectable changes in the body weights of salidroside treated db/db mice compared to vehicle db/db groups (Figure 1B). After the 5-week treatment, experiments of OGTT were performed. The response to OGTT was impaired in the db/db mice with a huge increase of blood glucose level (Figure 1C), and db/db mice treated with salidroside were significantly improved their tolerance to glucose compared to vehicle-treated db/db mice.

Another obese mouse model with metabolic syndrome, HFD mice were applied to test the function of salidroside as well, and 10-week HFD feeding induced a marked increase of fasting blood glucose in HFD mice compared with the normal diet (ND) mice (Figure 1E). Similarly, salidroside only showed hypoglycemic effect after 15 days treatment compared to the vehicles and its effect was enhanced with a longer treatment. Also, the body weight gain of HFD mice was unaffected by salidroside (Figure 1F). In addition, the improvement of OGTT in HFD mice with salidroside administration was observed (Figure 1G).

Meanwhile, an increasing tendency of fasting serum insulin levels was noticed in salidroside treated db/db or HFD mice, but no significant changes were found when compared with vehicle groups (Figures 1D,H).

High-fat diet (HFD) and db/db mice are both obese animal models with elevated total CHO and TG contents. Here we noticed that salidroside slightly reduced total CHO or TG levels in HFD and db/db mice but without significant differences compared to vehicle groups (Figure 2A). However, salidroside significantly increased levels of HDL-C, but decreased levels of LDL-C in HFD and db/db mice.

A publication reported the protective effect of salidroside on oxidative stress in diabetic mice induced with a single injection of 200 mg/kg alloxan (Li et al., 2011). In this study, it was obtained that salidroside displayed strong antioxidant ability in pancreas of HFD and db/db mice as well. As presented in Figure 2B, a reduction in activity of antioxidant enzyme GPx, or SOD, or CAT was observed in pancreas of diabetic mice, while activities of these enzymes were enhanced by salidroside. On the other hand, MDA, a product of lipid peroxidation, was increased in diabetic mice, but was decreased by salidroside. Similar results were obtained in subcutaneous white adipose tissue of diabetic mice (Figure 2C). Salidroside significantly enhanced SOD activity and declined MDA content in adipose tissue of db/db and HFD mice.

As we reported before (Yang et al., 2016), here a significant reduction of β-cell mass in diabetic mice was found compared to ND or WT mice (Figure 3A). In contrast, an increase of β-cell mass was observed in salidroside treated group, which indicated the protective effect of salidroside on β-cell survival. Additionally, the positive staining of Ki67 in β-cells (a marker of cell proliferation) can be found in pancreas from diabetic mice treated by salidroside, but was hardly observed in mice treated by vehicle (Figure 3B). Meanwhile, β-cell apoptosis was examined by TUNEL staining (Figure 3B). TUNEL-positive β-cell was barely found in sections from both ND groups and WT group. The β-cell apoptosis was strongly increased in HFD and db/db mice, which was suppressed by salidroside. We postulate that β-cell survival improved by salidroside contributed to the increase of β-cell mass partially.

To further confirm the effect of salidroside on β-cell survival, we detected β-cell proliferation and apoptosis by Ki67/TUNEL co-staining using isolated mouse islets cultured with different diabetic stimuli. As shown in Figure 3C, high glucose (33.3), or the cytokines mixture of IL-1β/IFN-γ (ILIF), or palmitic acid (FFA, free fatty acid), or H₂O₂ (as a positive inducer of ROS) reduced β-cell proliferation, but induced β-cell apoptosis, compared with control group. On the contrary, islets can be protected by salidroside against these deleterious effects. Proliferation of β-cell was significantly increased by salidroside compared to DMSO-treated, and β-cell apoptosis was decreased simultaneously.

Moreover, glucose-stimulated insulin secretion (GSIS) assay revealed that the insulin secretion of cultured islets was strongly impaired by these different diabetic stimuli (Figure 3D). Salidroside, however, improved the function of islets by increasing the stimulatory index significantly. Likewise, the function of insulin secretion of islets isolated from db/db mice was enhanced by salidroside (Figure 3E). Our findings demonstrated salidroside was able to protect β-cell survival and function against diabetic stimuli.

Direct exposure to oxidants such as H₂O₂ or secondary to glucolipotoxicity, inflammatory cytokines could lead to oxidative stress in β-cell. Here increased ROS production was detected in Min6 cells treated by high glucose or H₂O₂, the destroyed ΔΨm (mitochondrial membrane potential) was found in Min6 cells as well. As a natural antioxidant, salidroside effectively suppressed ROS production and recovered ΔΨm in β-cells compared to DMSO-treated cells (Figure 4A).

NADPH oxidase (NOX) is a key regulator of ROS production and mediates ROS triggered cell apoptosis. As presented in Figure 4B, NOX2 was up-regulated by high glucose or H₂O₂ in Min6 cells, followed by increased levels of p-JNK and cleaved caspase 3 which resulted in β-cell apoptosis finally. However, salidroside down-regulated NOX2 expression and inhibited subsequent activations of JNK and caspase 3 to prevent β-cell death.

Studies have shown that inactivation of FOXO1 contributes to β-cell protection under oxidative damage (Ding et al., 2014; Shao et al., 2014). It is known that in β-cells FOXO1 acts as a negative regulator of PDX1 functions and mediate β-cell dysfunction and apoptosis. Phosphorylation of FOXO1 (Ser256) indicated the function loss of it. Interesting, here we noticed that the p-FOXO1 (Ser256) level was up-regulated by salidroside (Figure 4B). Next the impacts of salidroside on FOXO1/PDX1 were investigated.

The PDX1 is an essential transcription factor for maintaining β-cell function and survival. Impairment of PDX1 nuclear localization results in β-cell dysfunction. Here we found in Min6 cells treated by high glucose or H₂O₂, PDX1 predominantly accumulated in the cytoplasm fraction, while most of FOXO1 proteins located in the nucleus (Figure 5A). However, with salidroside treatment, PDX1 translocated into the nucleus, and nuclear exclusion of FOXO1 was observed. The nuclear/cytosolic ratio of PDX1 was significantly elevated by salidroside compared to DMSO-treated cells.

The AKT could suppress the function of FOXO1 by directly phosphorylating it at Ser256. As expected, here we observed that salidroside activated AKT by increasing p-AKT level (Figure 5B). More importantly, we noticed that FOXO1 phosphorylation induced by salidroside was correlated with AMPK activation. Compound C, an inhibitor of AMPK, blocked up-regulations of p-AMPK and p-AKT stimulated by salidroside. The FOXO1 phosphorylation was suppressed by compound C simultaneously. Interesting, here we found the decrease of NOX2 expression induced by salidroside was correlated to AMPK activation and was suppressed by compound C as well (Figure 5B). Quercetin, a bioactive natural compound, was reported could decrease NOX2 expression through activating AMPK as a potent antioxidant (Hung et al., 2015). Our findings provide another evidence to support the association between AMPK and NOX2.

Further investigations confirmed that PDX1 nuclear localization promoted by salidroside was AMPK activation dependent, which could be blocked by compound C (Figure 5C). Additionally, β-cell function of insulin secretion improved by salidroside was diminished by compound C as well (Figure 5D).

Given these findings, we assume that salidroside may protect β-cell survival and function through AMPK mediated inhibitions of NOX2 and FOXO1 simultaneously.