Male Wistar rats with weight between 160 and 180 g were purchased from Hua Fukang Co., Ltd and bred in SPF circumstances in the experimental animal center of Huazhong University of Science and Technology. After acclimatization for 2 weeks, the rats were randomly divided into normal group (NO group) and HGFD group (HGFD group). The rats in NO group was fed with standard laboratory rat chow (the formula contains 35% flour, 20% corn meal, 20% soy meal, 15.5% bran, 5% fish meal, 1% dusty yeast, 0.5% bean oil, 2.5% bone meal, and 0.5% salt) while the diet in HGFD group contained 67.5% standard laboratory rat chow, 12% lard, 2% cholesterol, 0.5% bile salts and 20% sugar. The dietary regimens of the rats in different groups were unchanged until the end of the experiment. One month later, intravenous injection of small dose streptozotocin (STZ, 24mg/kg) was conducted in the rats of HGFD groups (Duca et al., 2015). Then the normal 95% confidence intervals at various time points in OGTT were achieved according to the blood glucose of rats in NO group. If the blood glucose was greater than 20% of the upper limit of confidence interval at any time point, the rats were chosen for further studies as successful models of diabetic rats. Diabetic rats were further randomly divided into model group (MO group) and treatment groups, with 12 rats in each group. All experiments were approved by the animal ethics committee of Huazhong University of Science and Technology (NO.2015S508).

The treatment groups included BBR treatment groups with BBR at high (BH group), middle (BM group), or low dose (BL group) and metformin group (ME group). BBR was purchased from Yabang Pharmaceutical Co. Ltd. in China (H20053180) and dissolved with 0.5% carboxymethyl cellulose sodium glue. High, medium and low dosages in BBR treatment groups were 375, 187.5 and 93.75 mg⋅kg^-1⋅day^-1 respectively (Leng et al., 2004). Daily dose of metformin was 184 mg/kg for the rats in ME treatment group. Daily lavage volume of each rat was 1 ml/100 g weight, and the solution concentration was calculated according to the lavage capacity. In addition, 0.5% sodium carboxymethyl cellulose solution was intragastric administrated to the rats in NO and MO groups. Gastric irrigation doses were adjusted weekly based on rats’ weights for 9 weeks.

When the interventions were over, six rats in each group were used for the detections of intestinal permeability and the other six rats in each group were employed for metabolic studies and sampling. When performing OGTT, rats were overnight fasting for 12 h and 2 g/kg glucose was administrated. The blood from tail veins of each rat were collected at 0, 30, 60, and 120 min after the intragastric administration and the blood glucose levels were detected using a glucose monitor (Roche, German).

At the end of 9 weeks’ treatment, all experimental animals were killed. Anesthesia was performed by intraperitoneal injection with 50 mg/kg pentobarbital sodium. Abdominal aorta blood sampling, intestines and MLN were collected under sterile conditions. Samples were fixed with 4% formaldehyde or preserved in -80°C.

The concentration of serum insulin was detected by using iodine radioimmunoassay kit (Jinding, China). For the detection of triglyceride (TG) content, colorimetric assay (Mingdian, China) was used.

Intestine sections were stained using hematoxylin and eosin (H&E) to assess the histological morphology. Transmission electron microscope (Hitachi, Japan) was used to evaluate the ultrastructural changes. Tissue blocks were fixed with glutaraldehyde. After rinsing and dehydration in ethanol, the tissues were embedded and sliced with ultramicrotome. Following uranyl acetate and lead citrate double staining, transmission electron microscope was used for the morphology observation and taking pictures.

FITC-dextran (Sigma, FD-4, USA) perfusion assay was used to detect intestinal epithelial permeability in the rats. After anesthesia, abdominal middle incision and ileum incision 5 cm distance from the appendix was performed and 10-cm-long ileal segment was ligated with 2.0 silk thread. Subsequently, 1ml FITC-dextran (10 mg/ml) was affused into the ligated intestinal cavity. Intestines were returned back to the abdominal cavity and kept wet by wrapping moist gauze and a plastic film. Thirty minutes later, blood was collected from the abdominal aorta. Serum was separated and fluorescence was detected with a spectrophotometer (Synergy2, USA).

Mesenteric lymph nodes was grinded with strainer mesh and glass rods in the 1640 culture medium containing 2% FBS. Erythrocyte was lysed and the supernatant was dismissed after centrifugation. If the markers were expressed in the cytoplasm and nucleus, FIX/PERM liquid was added to break the cell membrane. Otherwise, the cells were incubated with different antibodies for 30 min away from light and fixed with 1% paraformaldehyde. T cells were stained with CD3-FITC, CD4-APC and CD8-PE (BD Bioscience, USA); Macrophages were with CD11b-APC (BD Bioscience, USA) and CD68-FITC (AbD Serotec, UK), Tregs were with CD4-APC and Foxp3-PE (eBioscience, USA); dendritic cells (DC) were with OX62-FITC (Abcam, UK). Flow cytometry (BD FACSCalibur, USA) was applied for the detection of cells’ population.

Intestinal tissues were separated, rinsed, fixed and cut following paraffin embedding. After using the spontaneous fluorescence quenching agent, the sections were immunostained overnight with primary antibodies against NF-κB (CST, USA) with triton damaging cell membranes. Then fluorescence second antibodies were co-incubated at room temperature for 30 min. Following the rinse, the sections were stained with DAPI (Sigma, USA) for 3 min and sealed. Finally, observation and taking photos were performed under a Nikon fluorescence microscope.

After the extract of intestinal RNA with Trizol reagent (Takara, China) and concentration determination of RNA, reverse transcriptase kit (Takara, China) was used to synthetize the cDNA. Then, amplification was conducted using SYBR premix EX Taq^TM kit (Takara, China) in StepOne PCR detetor (Stepone, USA). Relative expressions of mRNA were calculated by 2^-ΔΔCT. The primer sequences are listed in Table 1.

Proteins of intestines (different sampling volumes for various antibodies) were separated through gel electrophoresis, transferred onto NC membranes, and incubated with antibodies against TLR4 (NOVUS, USA), Occludin-1 (OCLN, Gene Tex, USA), ZO-1 (Invitrogen, USA), Claudin-1 (Abcam, UK), MyD88 (CST, USA), p-IKKβ (Gene Tex, USA),IKKβ (Gene Tex, USA) and β-actin (Affinity, USA). The membranes were then co-incubated with fluorescent second antibodies. Oddesy imaging system was implemented to analyze the protein expressions.

Amylin, GLP-1, GIP and Ghrelin (Ray Biotech, USA) levels in serum were tested using ELISA Kit in line with the competitive enzyme immunoassay principle. In brief, 100 μL serum was added to each well and OD values were detected at 450 nm using a spectrophotometer (Synergy2, USA).

Data were analyzed by SPSS 20.0 software. Statistical significance was assessed by ANOVA following K-S normality test. With homogeneity of variance, significance between different groups was evaluated by Bonferroni test; alternatively, Dunnett’s T3 test was used. P-values < 0.05 suggested that significant differences existed.

Firstly, we detected the effects of BBR on glucose and lipid metabolism in diabetic rats (Figure 1). Consistent with previous study, diabetic rats underwent the weight loss after modeling (Wang et al., 2016), and BBR interventions did not significantly improve the decreased body weights. With 2 g/kg glucose lavage in OGTT, the blood glucose at each time-point increased significantly in diabetic rats, whereas BBR at middle (187.5 mg/kg) and high dosage (375 mg/kg) reversed elevated fasting and postprandial blood glucose. The area under the concentration-time curve (AUC) in OGTT and serum TG levels also decreased in BBR-treated rats. Although no significant difference was shown with regards to serum insulin concentration in different groups, BBR decreased insulin resistance index (IRI) as well as TG in diabetic rats of BM and BH groups. These data indicated that BBR had specific beneficial effects on metabolic parameters. Similar hypoglycemic and lipid-lowering effects were observed in ME group that metformin was administrated as a positive control.

Flow cytometry was used to detect the proportions of T cells, macrophages and DC in MLN. As shown in Figure 2, the percentage of Tregs in MLN of diabetic rats was decreased while it was increased in BBR- and metformin-treated rats. Because macrophages are the sources of many proinflammatory factors, we studied the proportion differences of macrophages with BBR administration. Significantly higher numbers of CD11b + CD68+ MLN cells were observed in MO group while less numbers were observed in BBR treatment and ME groups. As far as CD4 + T cells, CD8+ T cells, the ratio of CD4+ to CD8+ cells and DC were concerned, no significant changes were found.

Intestinal immune factors involved in the maintenance of intestinal immune tolerance and the intestinal barrier integrity. Since BBR had the beneficial effect on immune cells of intestinal immune system, it might also influence the expressions of intestinal immune factors. The results showed that IL-1β, TNF-α and MIF mRNA expressions were elevated in comparison with NO group, while IL-4 and IL-10 mRNA levels were reduced in intestinal tissue of diabetic rats (Figures 3A–E). However, BBR, especially at high doses, showed significant inhibitory effects on the expressions of IL-1β, MIF and TNF-α mRNA. Moreover, BBR treatment also exhibited the increase in IL-4 and IL-10 mRNA expressions in intestinal tissue of diabetic rats. Despite increased tendency for IL-6 mRNA expression and decreased trend for TGF-β mRNA expression in MO group, heterogeneity of variance was large and no statistical difference was found in the treatment groups (Data not shown).

Since gastrointestinal hormones participate in glycemic control, we next examined the effects of BBR on plasma gastrointestinal hormones levels (Figures 3F–I). After HGFD and the administration of STZ, GLP-1 and GIP secretion reduced while pancreatic beta cells-derived Amylin increased in diabetic rats (Figures 3F,I). The administration of BBR at different concentrations increased the secretion of GLP-1, but only BBR at a high dose influenced the production of GIP and Amylin. In the analysis regarding Ghrelin, BBR at a high dose decreased Ghrelin production compared with MO group, and no differences were found between other different groups.

High glucose and fat diet and intestinal inflammation could disrupt the intestinal mucosal barriers, leading to endotoxemia and the development of DM. Thus, we studied the effects of BBR on intestinal tight junction proteins expressions and mucosal barrier permeability. Tight junction protein OCLN, ZO-1 and Claudin-1 expressions were detected using western blot. Compared with the rats in NO group, these tight junction proteins expressions were reduced in the intestine of untreated diabetic rats. BBR treatment could partially increase OCLN and ZO-1 expressions but it showed no effects on Claudin-1 levels (Figures 4A–C). After the intestinal perfusion of FITC-dextran, BBR could reduce serum dextran concentration, namely the intestinal barrier permeability (Figure 4D). Although HE staining illustrated no significant change in intestinal structure between these groups (Supplementary Figure S1), the amount and distribution of filtered FITC-dextran were different in intestinal mucosa and submucosa tissues after intestinal dextran perfusion (Figure 5), which also suggested differences of the intestinal permeability in the rats of various groups. With regard to the ultrastructure of intestinal epithelial cells, intercellular adhesions were damaged, and the adherent junction in diabetic rats was sparse and expanded. These changes were improved after BBR treatment (Figure 6).

Previous studies confirmed that BBR could protect islet beta cells in DM, and also prevent intestinal inflammation in bowel disease through depressing TLR4 signaling pathways (Zhang et al., 2011). To ascertain whether the hypoglycemic and gut-protective effects of BBR were mediated through anti-inflammatory signal transduction pathways, we detected the expressions of TLR4 and downstream MyD88 and p-IKKβ in the intestines, along with p65 NF-κB distributions in epithelial cytoplasm and nucleus. The results revealed BBR, especially at a high dose, reduced intestinal TLR4 and MyD88 protein expressions as well as the phosphorylation of IKKβ (Figures 7A–C). As a main ligand for TLR4, LPS combined with LPS binding protein (LBP) and CD14 to form the trimer; then the complex binds to the TLR4 and activates inflammatory signal transduction pathways. Data of RT-PCR illustrated intestinal LBP and CD14 mRNA levels decreased in BBR-treated rats (Figures 7D,E). Accordingly, Figure 8 showed p65 NF-κB distributions were different in intestinal epithelial cells among different groups, and BBR exerted a potential ability to blunt p65 NF-κB translocation to the nuclei.