RB was developed by the School of Pharmacy, Chongqing Medical University. It was synthesized as described (Guan et al., 2020). RSG sodium tablets were purchased from Taiji Group Chongqing Fuling Pharmaceutical Factory Co., Ltd. (Chongqing, China). BBR chloride was purchased from West Plant Extraction Factory (Sichuan, China). Sodium carboxymethyl cellulose (CMC) was purchased from Semick Biotechnology Co. Ltd. (Chongqing, China). Primary antibodies against PI3K (p85) (#4257), phospho-AKT (Ser473) (#4060S), AKT (#4691S), and TXNIP (#14715S) were obtained from Cell Signaling Technology (Danvers, MA, United States). Primary antibody against phosphor-PI3K (#4257) was obtained from Affinity Biosciences LTD. (OH, United States). β-Actin (sc-47778) was purchased from Santa Cruz Biotechnology, Inc. (Dallas, TX, United States). Horseradish peroxidase-conjugated secondary antibodies were purchased from Zhongshan Jinqiao Biotechnology Co. Ltd. (Beijing, China). 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino]-2-deoxy-d-glucose (2-NBDG) was purchased from Good Laboratory Practice Bioscience Technology (Montclair, CA, United States). RSG, mannitol, PI3K inhibitor LY294002, and palmitic acid were purchased from Medchem Express (NJ, United States). D-(+)-glucose and insulin were purchased from Beyotime Biotechnology (Beijing, China). 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was purchased from Sangon Biotech Co., Ltd. (Shanghai, China). The Bio-Plex Pro™ Mouse Diabetes Panel 8-Plex Kit was purchased from Bio-Rad Laboratories (Hercules, CA, United States). Dulbecco’s modified Eagle medium (DMEM) and horse serum (HS) were obtained from Gibco (Thermo Fisher Scientific Inc., Waltham, MA, United States). Fetal bovine serum (FBS) was obtained from Procell Life Science and Technology Co., Ltd. (Wuhan, China). A glycogen detection kit was purchased from Solarbio Science and Technology Co., Ltd. (Beijing, China). The glucose test kit was obtained from the Nanjing Jiancheng Bioengineering Institute (Nanjing, China).

Male KKAy mice (4–5 weeks old) and age-matched C57BL/6J mice were purchased from Beijing Huafukang Biology Technology Co. Quality testing of mice was performed by the Experimental Animal Center, Chinese Academy of Medical Science, Beijing, China (Approval number: SCXK Beijing-2019-0008). Unless otherwise stated, all mice were allowed access to food and water ad libitum. Specific pathogen-free (SPF) rooms were maintained at 22°C, 50% humidity, and a 12-h light/dark cycle. The experiment was performed according to the principles and guidelines of the Chinese Council for Animal Care and was approved by the Institutional Animal Care and Use Committee of Chongqing Science and Technology Committee.

C57BL/6J mice were fed a standard chow diet, and KKAy mice were fed a high-sugar high-fat diet (HSHFD) during the experiment. All mice were fed adaptively for 2 weeks. T2DM KKAy mice were monitored for blood glucose levels, and mice with levels >11.1 mM were selected randomly for the experiment. The mice were divided into eight groups: C57BL/6J normal control group (NC), diabetes mellitus control group (DM control), RSG group (RSG, 1.04 mg/kg/day), BBR group (BBR, 195 mg/kg/day), RSG (1.04 mg/kg/day) plus BBR (1.08 mg/kg/day) group (MIX), RB groups (0.7 mg/kg/day, RB-L; 2.11 mg/kg/day, RB-M; 6.33 mg/kg/day, RB-H). All drugs were dissolved in 0.3% CMC. Both the NC and DM control groups were treated with 0.3% CMC (0.1 ml/10 g). Medications were administered intragastrically once daily for eight consecutive weeks. Blood glucose levels were measured weekly using a blood glucose meter (ONETOUCH Ultra Lifecan, United States). Bodyweight was monitored weekly. Food intake was monitored every 2 days. All mice were sacrificed by cervical dislocation after 8 weeks of treatment, and the liver, muscle tissue, white adipose tissue, and pancreas were collected.

Mice were fasted for 12 h, followed by oral glucose administration (2 g/kg body weight, p.o.). Whole venous blood was obtained from the tail vein at 0, 30, 60, 90, and 120 min after administration. Blood glucose levels were measured using an automatic glucometer (ONETOUCH Ultra, Lifecan, United States). The area under the curve (AUC) was calculated by using GraphPad software 8.

Mice were fasted for 12 h and then injected insulin intraperitoneally (0.75 units/kg body weight, i.p.). Whole venous blood was obtained from the tail vein at 0, 15, 30, 60, 90, and 120 min after administration. Blood glucose levels were measured using an automatic glucometer (ONETOUCH Ultra, Lifecan, United States). The area under the curve (AUC) was calculated by using GraphPad software 8.

Orbital venous plexus blood and white fat were collected from mice after the 8-week treatment; blood serum was separated. Serum triglyceride (TG), high-density lipoprotein (HDL), and low-density lipoprotein (LDL) levels were measured using an auto-biochemical analyzer (Mindray BS-220, China). Fasting blood glucose (FBG) levels were measured using a blood glucose meter. Fasting insulin (FIN), gastric inhibitory peptide (GIP), and glucagon-like peptide-1 (GLP-1) levels were determined using Bio-Plex suspension chip system (Bio-Plex 200 System) with Bio-Plex Pro™ Assay kit, and the insulin resistance index (HOMA-IR) and insulin sensitivity index (ISI) were calculated. Following the manufacturer’s protocol, glycogen content in the liver and muscle was measured using respective assay kits. Similarly, glucose consumption and cellular glycogen content were measured using assay kits. HOMA-IR index = FBG (mM) × FIN (mIU/L)/22.5; ISI = 1/[FBG (mM) × FIN (mIU/L)]; white fat index (WFI) = white fat (mg)/body weight of mouse (g).

All mice were sacrificed by cervical dislocation after 8 weeks of treatment, and the liver and pancreas were collected. Mouse tissues were post-fixed in 4% paraformaldehyde for 24 h and sectioned after embedding in paraffin. Sections were prepared and stained with H&E using standard procedure. Slides were examined under Nikon ECLIPSE Ci biological microscope and images were captured using Nikon color digital camera.

C2C12 myoblasts and HepG2 cells were cultured in DMEM supplemented with 10% FBS in a 5% CO₂ incubator at 37°C. For differentiation of C2C12 myoblasts, the cells were cultured in DMEM containing 2% HS for 5 days. C2C12 myotubes and HepG2 cells were transferred to 33.3 mM high-sugar DMEM medium (H-DMEM) containing 0.25 mM palmitate (PA) for 36 h and then treated with the corresponding drugs for 24 h to establish an insulin resistance model. Cells were divided into eight groups: normal control group (NC) treated with the same volume of DMEM, model group (DM) treated with H-DMEM and PA, RSG group treated with H-DMEM and PA + RSG (36 μM), BBR group treated with H-DMEM and PA + BBR (36 μM), MIX group treated with H-DMEM and PA + RSG (36 μM) combined with BBR (36 μM), RB-L group treated with H-DMEM and PA + RB (4 μM), RB-M group treated with H-DMEM and PA + RB (12 μM), and RB-H group treated with H-DMEM and PA + RB (36 μM). In addition, the mannitol group (MAN) was treated with a final concentration of 33.3 mM mannitol for the same duration as the DM control group.

HepG2 cells were cultured in 24-well plates at a density of 1.5 × 10⁴ cells per well and pretreated as described above. After treatment, the cells were washed twice with PBS and then incubated with 0.1 mM 2-NBDG at 37°C for 30 min. Glucose uptake was terminated by three quick washes with ice-cold PBS and recorded using a fluorescence inversion microscope. The average fluorescence intensity was evaluated using the ImageJ software.

Proteins (30–50 μg) were electrophoresed on 8% SDS-PAGE gels and then transferred to PVDF membranes (Millipore, MA, United States). Next, the membranes were blocked for 2 h at room temperature with 5% non-fat dry milk in Tris-buffered saline with Tween-20 (TBST) and then incubated with primary antibodies at a dilution of 1:1,000 for 12 h at 4°C. After washing off the excess primary antibodies with TBST, the membranes were incubated with the appropriate secondary antibodies at a 1:5,000 dilution for 2 h at room temperature. The membranes were washed thrice with TBST for a total of 30 min. An ECL kit was used to enhance chemiluminescence, and a scanner was used to quantify the protein bands. Signal bands were quantified by densitometric analysis using the ImageJ software.

All results are expressed as the mean ± SD. Data were examined using one-way analysis of variance (ANOVA) followed by Tukey’s comparisons. Statistical significance was set at p < 0.05. GraphPad software 8 (San Diego, CA, United States) was used for calculations.

We conducted in vivo experiments using an HSHFD-induced diabetic mouse model KKAy mice to investigate the effects of RB treatment. All selected KKAy mice had blood glucose levels >11.1 mM, which accorded with the characteristics of the diabetic model (Figure 1A). Randomized glucose levels were consistently elevated in the DM control group compared to those in the NC group and peaked at week 6 (>33.3 mM). Compared to the DM control group, the blood glucose levels of the treatment groups were effectively controlled after administration, particularly in the RB-M and RB-H groups. The mean random blood glucose levels in the RB-M group were ameliorated to a greater extent than those in the positive control groups (RSG, BBR, and MIX). However, there were no statistically significant differences among the groups. Additionally, bodyweight of the mice in each group gradually increased throughout the experiment. The weight of mice in the DM control group was always higher than that of mice in the NC group. However, there was little difference between the treatment and DM control groups (Figure 1B). Our results confirmed that the treatment did not affect the body weight of KKAy mice. Moreover, the average food intake in the DM control group was significantly higher than that in the NC group. None of the DM mice showed a difference in food intake until the fourth week. At week 8, the food intake of the RB-M and RB-H groups decreased significantly compared to that of the DM control group (Figure 1C). These data indicated that RB administration could effectively control blood glucose levels in KKAy mice without affecting their body weight.

To determine whether RB affects insulin sensitivity in vivo, T2DM KKAy mice were treated with RB for 8 weeks. OGTT testing showed that the blood glucose levels of DM mice increased significantly after glucose administration (0.5 h) and reached a peak. After 120 min, the blood glucose level remained relatively high (Figure 2A). The basal blood glucose level in the RB groups (RB-L, RB-M, and RB-H) was lower than that in the DM control group, and the blood glucose level after glucose administration was elevated. After 30 min of glucose administration, the blood glucose levels in the RB-H group were lower than those in the RB-M and RB-L groups, with all significantly lower than those in the DM control group. The effects in the RB groups were dose-dependent. The AUC of the RB-L group was lower than that of the BBR group. The AUC of the RB-M group was lower than that of the RSG and BBR groups. The AUC of the RB-H group was lower than that of the RSG, BBR, and MIX groups, all of which were lower than that of the DM control group (Figure 2B). These data indicate that RB effectively improved impaired glucose tolerance in diabetic KKAy mice.

ITT was performed in the eighth week for comparison to determine the effect of RB on insulin tolerance (Figures 2C,D). During ITT, an apparent difference was observed in blood glucose levels between NC and diabetic mice at any time point. In the ITT experiment, mice in the NC group showed rapid decrease followed by slow decrease in blood sugar levels compared with KKAy mice, whereas mice in the DM control group exhibited higher blood sugar and insulin insensitivity, which could be largely relieved by RB treatment. The AUC of the RB group was lower than that of the DM control group, and the blood glucose level of the RB-H group was lower than that of the RB-M and RB-L groups, in a dose-dependent manner. The AUC of the RB-M and RB-H groups was lower than that of the RSG, BBR, and MIX groups, all of which were lower than that of the DM control group. These data demonstrated that RB could effectively improve insulin sensitivity in diabetic KKAy mice.

To investigate changes in insulin resistance, we used the homeostasis model assessment of insulin resistance (HOMA-IR) and insulin sensitivity index (ISI) to evaluate the effect of RB on insulin resistance in KKAy mice. The FIN, FBG, and HOMA-IR indices were higher in the DM control group than in the NC group (Figures 2E–G), indicating the presence of significant systemic insulin resistance in diabetic KKAy mice. The FIN and FBG levels in vivo were reversed in the treated groups (Figures 2E,F). The results showed that RB significantly reversed insulin resistance in KKAy mice. HOMA-IR value and plasma insulin concentration showed a decreasing trend (Figure 2G). ISI values showed an increasing trend (Figure 2H), with the most significant in the RB-H group. Moreover, the HOMA-IR index of the RB-H group was lower than that of the RSG, BBR, and MIX groups. The ISI index of the RB-H group was higher than that of the RSG, BBR, and MIX groups. The HOMA-IR index of the RB-M group was lower than that of the RSG, BBR, and MIX groups. The ISI index of the RB-M group was higher than that of the RSG, BBR, and MIX groups. However, the results were statistically different only when compared with the BBR group. These data indicated that RB treatment effectively ameliorated insulin resistance in diabetic KKAy mice.

When the body is insulin-resistant, the liver excessively releases glucose into the bloodstream due to increased glycogenolysis and gluconeogenesis, causing peripheral tissues, such as muscles, to become insensitive to insulin-regulated glucose metabolism. Eventually, reducing glycogen production and blood glucose consumption lead to hyperglycemia (Petersen et al., 2017; Ormazabal et al., 2018). Our data showed that KKAy mice had significant insulin resistance and low levels of hepatic glycogen and myoglycogen compared with the NC group. After RB treatment, the liver glycogen and myoglycogen contents of KKAy mice significantly increased (Figures 2I,J). Furthermore, RB increased glycogen stores in the liver and muscle in a dose-dependent manner, evidenced by higher muscle glycogen content in the RB-M and RB-H groups than in the RSG and BBR groups and higher liver glycogen content in the RB-M and RB-H groups than in the RSG, BBR, and MIX groups. The data indicate that RB effectively improved glucose utilization and metabolism in diabetic KKAy mice.

Compared with the NC group, significant changes in lipid levels were observed in the DM control group, including elevated TG and LDL levels and decreased HDL in the serum of KKAy mice (Figures 3A–C). After 8 weeks of RB treatment, the serum TG concentration in the RB-M and RB-H groups was lower than that in the positive control groups (RSG, BBR, and MIX). However, the differences were statistically significant only when compared with the RSG and BBR groups (Figure 3A). The serum LDL concentration in the RB-M and RB-H groups was lower than that in the positive control groups and significantly different between the RB-H and BBR groups (Figure 3B). The serum HDL concentration in the RB-M group was higher than that in the positive control groups, but the difference was not statistically significant (Figure 3C). The WFI in the RB-M and RB-H groups was lower than that in the positive control groups; differences were statistically significant between the RB-M group and BBR groups and the RB-H and positive control groups (Figure 3D). These findings suggest that RB administration ameliorates dyslipidemia in T2DM mice, contributing to the improvement of insulin resistance and playing an important role in antidiabetic therapies.

The hormones GIP and GLP-1 are secreted by the gut. GLP-1 is a multifaceted hormone with broad pharmacological potential. GLP-1 receptor agonists have been successfully used clinically for T2DM treatment, whereas GIP levels are increased in individuals with obesity and T2DM (Tasyurek et al., 2014; Holst and Rosenkilde, 2020). The GIP levels in the DM control group were significantly higher than those in the NC group (Figure 3E). The medication intervention significantly decreased the GIP level, particularly in the RB-H group that had GIP level lower than that of the positive control groups (p < 0.01) and close to that of the NC group. Meanwhile, slight difference was observed between the RB-L and RB-M groups and the positive control groups. Additionally, GLP-1 (Figure 3F) in the DM control group was reduced by three-fold compared to that in the NC group. GLP-1 levels tended to increase in response to medical stimulation. Notably, the increase in GLP-1 levels was concentration-relevant rather than concentration-dependent with RB treatment. The RB-M group showed increased level of GLP-1 compared with the NC group in T2DM KKAy mice, and the GLP-1 level in the RB-M and RB-H groups was higher than that in the positive control groups but statistically significant only when compared with the BBR group. These data suggest that RB may improve insulin resistance by increasing serum GLP-1 concentration and decreasing GIP concentration, thereby regulating lipid metabolism, reducing appetite, and protecting islet cells.

We evaluated whether RB treatment had any effect on T2DM-induced hepatocyte alterations and pancreatic islet cells through examining histopathological changes using H&E staining in T2DM KKAy mice after RB intervention. We observed that the hepatocytes were arranged neatly, with no steatosis observed in the NC group (Figure 4A). In contrast, swollen hepatocytes and severe fatty degeneration were observed in the liver cells, with a massive accumulation of fat vesicles in the cytoplasm in the DM control group. Ameliorative effect of steatosis in the RSG and MIX groups was improved compared with that in the DM control group. However, this improvement was not significant in the BBR group. RB treatment was significantly superior to the positive control groups (RSG, BBR, MIX) and reduced liver lipid accumulation in KKAy mice in a dose-dependent manner. Similarly, the pancreatic structure was clear and complete in the NC group, and slightly stained islet cells were arranged in groups (Figure 4B). The β cells were close to each other without obvious cell degeneration or necrosis. In the DM control group, islet cells were hypertrophic and disorganized, with swollen, deformed, and loosely arranged islet β cells. The degree of morphological and structural damage to islet cells was significantly lower in the RB administered groups than in the DM control group. RB treatment significantly repaired pancreatic islet cell damage in a concentration-dependent manner, consistent with hepatic alterations.

To examine the role of RB in regulating glucose metabolism in vitro, high glucose (33.3 mM) and palmitate were used to establish a cellular model of insulin resistance, and the glucose analog 2-NBDG was used as a radiotracer to track glucose uptake. First, to confirm the experimental role of osmolarity induced by high glucose, mannitol (MAN), corresponding to high glucose concentration in the DM control group, was used to assess the effect of a high osmotic environment. The data showed that both the high glucose and mannitol groups (MAN) had decreased cell survival rates in HepG2 and C2C12 cells compared to the normal control group (NC), and there was no statistically significant difference between the MAN and DM control groups (Figure 5A). Moreover, the addition of high glucose significantly reduced glucose uptake, whereas MAN did not alter glucose consumption in HepG2 and C2C12 cells compared to the NC group (Figure 5B). In short, although osmolarity induced by high glucose and the same concentration of mannitol affected cell viability, mannitol had no significant effect on glucose uptake, excluding the effect of a hyperosmotic environment on glucose uptake in HepG2 and C2C12 cells.

Next, we examined whether RB affected glucose uptake in insulin-resistant HepG2 cells (DM) (Figures 5C,D). Insulin resistance in hepatocytes causes impaired insulin signaling and decreased glucose uptake, the major contributors of hyperglycemia. Treatment of HepG2 cells with high glucose and palmitate (PA) triggers a significant reduction in insulin signaling leading to insulin resistance (Wen et al., 2011; Palomer et al., 2018). Our results showed that glucose uptake was significantly lower in DM cells than in NC cells. After treatment with RB (RB-L, RB-M, and RB-H), glucose uptake remarkably increased in a concentration-dependent manner compared to DM cells. Similarly, treatment with 36 μM RSG, BBR, and MIX increased glucose uptake compared to DM cells. The RB-H group showed better glucose uptake than the positive control groups (RSG, BBR, and MIX) at the same concentration. Our data demonstrated that insulin-resistant cells had significantly impaired glucose uptake and that the use of RB increased sugar uptake and further decreased insulin resistance.

Similarly, corresponding effects were observed for cell glucose consumption and glycogen storage. In HepG2 and C2C12 cells, glucose consumption and glycogen stores were significantly reduced in the DM control group compared to the NC group and reversed with RB treatment (Figures 5E–H). Moreover, RB improved cellular glucose consumption and glycogen storage in a dose-dependent manner, which was superior to that of the RSG, BBR, and MIX groups at the same dose (36 μM). Our data suggest that RB effectively restores glucose consumption and glycogen stores and regulates glucose metabolism in insulin-resistant HepG2 and C2C12 cells.

To further test the ability of RB to modulate the insulin signaling pathway, PI3K, AKT, and TXNIP were used as critical messengers. Insulin activates the PI3K/AKT pathway by binding to the insulin receptor that regulates glucose and lipid metabolism (Holland et al., 2007; Saltiel, 2021). Damage to the PI3K/AKT pathway leads to the development of insulin resistance and T2DM (Huang et al., 2018). As an important regulator of cell metabolism and stress, TXNIP, mediated by AKT under glucose stress, is often upregulated in T2DM (Robinson et al., 2013; Hong et al., 2016; Waldhart et al., 2017). To investigate the expression levels of these proteins and confirm the results of the above experiments, western blot analysis was performed. First, high glucose caused noticeably lower PI3K and AKT phosphorylation levels in HepG2 and C2C12 cells (DM control group) and significantly higher TXINP protein levels than that of the NC group (Figure 6), indicating the successful establishment of an insulin-resistant cell model. However, mannitol stimulation with high glucose isotonicity in the DM control group did not alter the expression of p-AKT, p-PI3K, or TXNIP, showing little effect of mannitol on PI3K/AKT/TXNIP signaling regulation (Figure 6A). As shown in Figure 6B, RB treatment increased the expression of p-AKT and p-PI3K. It decreased the expression of TXNIP in a dose-dependent manner in HepG2 cells, with the maximal effect observed at a concentration of 36 μM, superior to that of the RSG, BBR, and MIX groups at the same concentration. The same verification was performed for C2C12 cells. We found a significant increase in p-PI3K and p-AKT expression with RB concentration, superior to RSG, BBR, and MIX groups (Figure 6C). This indicated that RB could activate the PI3K/AKT pathway inhibited by high glucose and palmitate, and mediate the downregulation of TXNIP, consistent with previous studies (Yoshihara et al., 2010; Hong et al., 2016). We then used a specific PI3K inhibitor, LY294002, to investigate whether RB ameliorates insulin resistance through the PI3K/AKT/TXNIP pathway in vitro (Figures 6D,E). HepG2 and C2C12 cells were pretreated with LY294002 (20 μM) for 2 h, followed by treatment with 36 μM RB for 24 h. As shown in Figures 6D and E, the phosphorylation levels of PI3K and AKT were significantly inhibited by LY294002 pretreatment and high glucose + PA stimulation in HepG2 and C2C12 cells. Moreover, high glucose + PA stimulation elevated TXNIP protein expression levels, enhanced after LY294002 pretreatment. In addition, the phosphorylation of PI3K and AKT inhibited by high glucose + PA stimulation was restored and upregulated following treatment with RB. However, LY294002 markedly inhibited RB from restoring PI3K and AKT phosphorylation levels as well as TXNIP levels. As expected, these data confirmed that stimulation with LY294002 significantly attenuated the protective effect of RB on PI3K/AKT/TXNIP signaling.

Next, we assessed the action of RB on PI3K/AKT/TXNIP signaling in vivo. The phosphorylation and total expression levels of PI3K, AKT, and TXNIP were investigated by protein blotting. The data showed that p-PI3K and p-AKT were downregulated, and that TXNIP protein expression was increased in both the liver and muscle of KKAy mice compared to C57BL/6J mice (NC group) (Figures 7A,B). RB treatment restored the levels of p-PI3K and p-AKT in the liver and muscle tissues of KKAy mice. Moreover, RB inhibited TXNIP expression in a concentration-dependent manner and was more effective at higher doses. These results further support that RB treatment may ameliorate peripheral diabetic tissue lesions and insulin resistance via the PI3K/AKT/TXNIP signaling pathway in KKAy diabetic mice.