The herbal formula of TKFS, was provided by Tibet Autonomous Region Institute of Traditional Tibetan Hospital. For each batch, TKFS (0.5 g) were accurately weighed, and extracted with 50 mL of 50% methanol in an ultrasonic water bath for 20 min. Additional 50% methanol was added to adjust the volume, and then the solvent was filtered through a 0.22 μm microfiltration membrane.

The multiple-components of the TFKS were analyzed with the Agilent 1260 HPLC system (Karlsruhe, Germany), including a quaternary solvent delivery system, an on-line degasser, an auto-sampler, a column temperature controller and a photodiode array detector coupled with an analytical workstation. The samples were analyzed using a Waters SunFire C18 column (250 mm × 4.6 mm, 5 μm) at 30°C. The binary gradient elution system consisted of methanol (A) and 0.2% phosphoric acid (B), and separation was achieved using the following gradient program: 0–15 min, 97% B; 15–16 min, 97–85% B; 16–90 min, 85–50% B. The flow rate was set at 1.0 mL/min and the sample injection volume was 10 μL. The detection wavelength was set at 273 nm.

Data analysis was performed by a professional software named Similarity Evaluation System for Chromatographic Fingerprint of Traditional Chinese Medicine composed by Chinese Pharmacopoeia Committee (Version 2009 A), which was recommended by China Food and Drug Administration (CFDA) (Tang et al., 2014; He et al., 2015).

Seven-week-old male diabetes spontaneous diabetic mutation Lepr^db mice (referred as db/db mice, n = 30, body weight, 39.5 ± 1.6 g) and non-diabetic wild type littermates (referred as WT mice, n = 6, body weight, 19.9 ± 0.8 g) were purchased from the Model Animal Research Center of Nanjing University (Nanjing, China). They were housed in SPF animal rooms at a 12 h light–dark cycle with the suitable relative humidity (55 ± 5%) and temperature (22 ± 2°C). Before the experiment all mice were acclimated for 1 week to the environment. All experimental animal procedures followed international guidelines for care and use of laboratory animals and were approved by the Animal Ethics Committee of South-central University For Nationalities. Metformin tablets were purchased from Beijing Jing Feng Pharmaceutical Factory (Beijing, China).

The db/db mice were randomly divided into five groups (n = 6 per group): db/db mice plus 0.9% saline group (model control group), db/db mice plus low dose of TKFS group (TKFS 0.5 g/kg), middle dose of TKFS group (TKFS 1.0 g/kg), and high dose of TKFS group (TKFS 2.0 g/kg), db/db mice plus metformin group (Metformin 200 mg/kg), and WT mice plus 0.9% saline group (n = 6 as normal control group). According to the directions on the label that have used in human beings (6 g/day per adult), dose conversion from human to mice was converted according to body surface area and eventually the doses of TKFS were decided above for the mice.

Both TKFS and metformin were dissolved in 0.9% saline, and mice were treated by oral gavage administration with doses described above once a day for 4 weeks. At the end of the last week, all mice were fasted overnight. After testing FBG, all mice were sacrificed with an overdose of intraperitoneal injection of pentobarbital (90 mg/kg). The mouse blood will be quickly collected and immediately centrifuged (4°C, 300 g for 10 min) to gather the blood serum samples, which were subsequently stored at -20°C for further analysis. The pancreas tissues were dissected and immediately immersed in 4% paraformaldehyde for histological analysis. The skeletal muscles were dissected and immediately frozen in liquid nitrogen or fixed in 4% paraformaldehyde for western blot or immunohistochemical analysis.

Fasting blood glucose and body weight were measured at 08:00 to 08:30 on the first day of each week during the treatment after fasted overnight. Blood samples were obtained by tail-prick and measured by using a standard glucometer (LifeScan, Inc., Milpitas, CA, United States). The body weight was measured with an electronic weighing scale.

In the IPITT test, after the mice were fasted for 6 h at the second week of the experiment, they were i.p. injected with insulin at 0.75 IU/kg as reported previously (Liu et al., 2014; Zhang et al., 2014), and blood glucose levels were monitored at 0, 30, 60, and 120 min after the insulin injection. The total area-under-the-curve (AUC) of glucose from the sampling period from 0 to 120 min was determined as the AUC value (Runtuwene et al., 2016).

The IPGTT was performed on mice at third week of treatment with TKFS after fasted overnight, then D-glucose (0.75 g/kg) was carried out by i.p. injection to each mice (Zhang et al., 2014). The blood glucose levels as well as AUC values were recorded as described above.

Fasting serum insulin level were measured by using an enzyme-linked immunosorbent assay mouse insulin ELISA kit (CSB-E05071m, CUSABIO BIOTECH CO., Ltd, Wuhan, China). IR value was calculated according to a previous study (Dong et al., 2015), by using a formula as following: HOMA-IR = FINS (μU/mL) × FBG (mmol/L) /22.5.

The serum total cholesterol (TC), triglycerides (TG), and low-density lipoprotein cholesterol (LDL-C) levels were detected by using enzymatic colorimetric kits and following the product instructions (Nanjing Jiancheng Bioengineering Institute, Nanjing, China).

The tails of pancreas from each mouse were fixed in 4% paraformaldehyde, followed by processing of paraffin embedding, tissue sectioning, and hematoxylin and eosin (HE) staining. Morphological structure of islet of pancreas was observed and photographed by using an optical microscope (BH-2, Olympus, Japan).

The paraffin sections were deparaffinized in xylene and rehydrated through graded washes of ethanol as described previously (Ma et al., 2015). After antigen retrieval with high-temperature heating in a citrate buffer, the slides were incubated with 3% H₂O₂ buffer for 10 min and washed it out with PBST, then incubated with anti-GLUT4 (ab33780, Abcam, Cambridge, United Kingdom) at 1:100 dilution at 4°C overnight. After washed with PBST, the slides were incubated with biotin-labeled secondary antibody (PV-6001 Zhongshan Jinqiao Co., Ltd, Beijing, China) at room temperature for 2 h. Then the sections were incubated with avidin-biotin-peroxidase complex (1:50, Elite ABC Ki, Vector) at room temperature for 2 h. Finally after washed with PBST, the sections were incubated with a reaction buffer containing 0.02% (w/v) DAB and 0.003% (v/v) H₂O₂ in 0.01 M Tris-HCl (pH 7.4) at room temperature for 5–10 min. After DAB staining, the slides were washed, dehydrated again and mounted, finally observed and photographed under an optical microscope (BH-2, Olympus, Japan).

The total protein from the skeletal muscles were extracted as described previously (Yang et al., 2015). Next the protein was fractionated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto PVDF membrane. After blocking with 5% non-fat milk for 1 h, the membranes were incubated with anti-Akt (#4685), anti-phospho-AktSer473 (#4060), anti-GLUT4 (#2213), anti-AMPKα (#5831), phospho-AMPKα (Thr172) (#2535) (CST, Danvers, MA, United States) and anti-GAPDH (Proteintech, Wuhan, China) primary antibodies at 1:1000 dilution at 4°C overnight. Then the membrane were washed in TBST and incubated with appropriate horseradish peroxidase-conjugated secondary antibodies. The protein bands were visualized by using a BeyoECL Plus (P0018, Beyotime Biotechnology, China), and a densitometry analysis was performed by using ImageJ software (NIH, Bethesda, MD, United States). GAPDH was used as the internal control for semi-quantitative analysis.

All data were expressed as the means ± SEM. GraphPad Prism 5 software (San Diego, CA, United States) was used for the data statistical analysis and graphics. Unpaired t-test was used to analyze statistical comparisons between two groups. Multiple comparisons were compared by one-way analysis of variance (ANOVA) followed by Bonferroni’s post hoc tests. p-value < 0.05 was assumed as statistically significant.

Based on the results of determination, HPLC fingerprints for TKFS were established. Reference chromatographic fingerprint for TKFS was generated based on 10 samples. A good separation and reproducible chromatogram was achieved and 13 peaks were marked as the common peaks (from peak 1 to peak 13) (Figure 1A) in the chromatograms of the 10 batches (Figure 1B). Our developed method successfully determined their features, especially those identities of Gallic acid (the 3rd peak of TKFS sample at 273 nm in Figure 1A), the most important active constituent of TKFS, was verified and chosen (showed by Peak 3) to calculate the relative retention time (RRT) and relative peak area (RPA) of all the other peaks (Table 1). The results from the 10 batches of samples (Figure 1B) indicated that the RPAs of the 13 common peaks (Figure 1A) varied, but the RRT was invariable for the herb. The similarities of the TKFS were calculated, compared with the reference chromatogram, the least similarity value of these samples was 0.90, which indicated that the samples had similar chemical compositions and this reference chromatogram could be applied as a standard HPLC fingerprint.

The FBG levels and body weights in model control group were significantly increased than those in WT group in all weeks we studied (p < 0.01) (Figures 2A,B). Compared to the model control, all groups with TKFS at 0.5 g/kg, 1.0 g/kg and 2.0 g/kg as well as metformin significantly lowered FBG levels than those in model control group (p < 0.05 or p < 0.01) (Figure 2A). However, there was no significant difference in the body weights between all db/db groups treated with TKFS at different doses or metformin to that with saline (p > 0.05) (Figure 2B).

In IPGTT and IPITT test, the TKFS (0.5 g/kg, 1.0 g/kg, and 2.0 g/kg) and metformin obviously lowered the levels of AUC in db/db mice (showed that in model control group), which are significantly increased than those in WT group (normal control group) (p < 0.05, p < 0.01) (Figures 3A,B).

The levels of FINS and the HOMA-IR index were markedly enhanced in db/db diabetic mice compared to those in WT mice. TKFS (1.0 g/kg and 2.0 g/kg) significantly decreased the insulin level compared to the model control group (p < 0.01), and the HOMA-IR was significantly reduced as well (p < 0.01) (Figures 4A,B). There were no detectable changes in the FINS of db/db mice treated with TKFS (0.5 g/kg) and metformin compared with their vehicle controls (Figure 4A), but both of them, decreased the HOMA-IR index significantly (Figure 4B).

The serum concentrations of TC, TG and LDL-C in the model control group were markedly increased than those in the WT group (p < 0.01). TKFS (1.0 g/kg and 2.0 g/kg) and metformin treatments all significantly inhibited the elevation in the levels of serum TC, TG, LDL-C of the model control db/db mice group (p < 0.05, p < 0.01) as shown in Figures 5A–C.

Reduced or impaired β-cell function is one of the typical components in the pathogenesis of T2DM and previous studies clearly supported that it is essential to preserve or rescue the β-cell function as treating the T2DM (Saisho, 2015). As shown in Figure 6, compared to the WT group, the model control group db/db mice showed more frequently pathological changes, such as: atrophy of islets, islet cells necrosis, lacking organization of islet cells, vanished pancreas and pancreatic acini boundaries, or hypertrophy islet cells, etc. These abnormal histological changes were significantly alleviated in the TKFS (1.0 g/kg and 2.0 g/kg) and metformin treatment group compare to the model control group (Figure 6).

To evaluate whether TKFS regulates blood glucose and ameliorates IR via PI3K/Akt or AMPK signaling pathways, levels of p-Akt and p-AMPK in skeletal muscles were determined by western blot (Figures 7A,B), and GLUT4 were determined by western blot (Figure 7C) and immunochemistry analysis (Figure 7D) respectively. As shown in Figures 7A,B, the p-Akt and p-AMPK levels were significantly reduced in model control group compared with that in WT group (p < 0.01). TKFS (0.5 g/kg, 1.0 g/kg, and 2.0 g/kg) and metformin significantly reversed the down-regulation of p-Akt and p-AMPK expression compared with that in model control group (p < 0.05). The GLUT4 protein expression was dramatically lower in model control group than that in WT group (p < 0.01). Treatment with TKFS (0.5 g/kg, 1.0 g/kg, and 2.0 g/kg) or metformin significantly normalized the protein expression of GLUT4 in skeletal muscles down-regulated in the model control group (Figures 7C,D).