All chemicals and solvents in the experiments were of analytical grade. AOS, which were extracted from the A. squarrosum (L.) Moq. (the number of Virtual Herbarium in Chinese is HIMC0039313) and Metformin tablets (National Drug Approval no. H12020797), were supplied by the Research Institute of Traditional Mongolian Medicine Engineering Technology (Inner Mongolia, China) and Tianjin Pacific Chemical and Pharmaceutical Co., Ltd. (Tianjin, China), respectively. A free fatty acid (FFA) determination kit (Lot no. 20151201A), an insulin determination kit (Lot no. 20151101A), and glycosylated hemoglobin determination kit (hemoglobin A1c [HBA1c]; Lot no. 20151129A) were provided by the Beijing Kainuo Spring Biotechnology Co., Ltd. (Beijing, China). Rabbit monoclonal antibodies for IRS-1 (Cat no. ab52167), IRS-2 (Cat No. ab3690), insulin receptor (INS-R) (Cat no. ab60946), and Glut4 (Cat no. ab654) were provided by Abcam Inc. (Burlingame, CA, United States). Rabbit monoclonal antibodies for GAPDH antibody (Cat no. GTX627408) were provided by GeneTex (Irvine, CA, United States). Goat anti-rabbit IgG (Cat no. 100995) was provided by Sunrise Services Inc. (Gaithersburg, MD, United States). The Kits of Revert Aid First Strand cDNA Synthesis Kit (Batch no. 00285300) and Platinum SYBR Green qPCR Super Mix-UDG with ROX (Batch no. C11744-100) were provided by the Thermo Fisher Scientific (Waltham, MA, United States). RPMI-1640 cell culture medium (Lot no. 8116118) and fetal bovine serum (FBS) (Lot no. A79E00G) were supplied by the Gibco Company (Grand Island, NY, United States). Cell Counting Kit-8 (CCK-8) was supplied by the Dojindo Laboratories (Kyushu Island, Japan). Reference standards (purity >98%) of glucose and 3-amino-9-ethylcarbazole (AEC) were provided by the Shanghai Tauto Bio-Technology Co., Ltd. (Shanghai, China). High-performance liquid chromatography-grade acetonitrile was purchased from Fisher Scientific (Fair Lawn, NJ, United States). The experiments used purified water produced from the Milli-Q Water Purification System (Millipore, Milford, MA, United States).

The extraction and purification of the AOS were performed as described in our previous research (Bao et al., 2019; Bao et al., 2020).

RBG was estimated by the tail snipping method. The blood glucose levels were measured (30 min after dosing) using a glucose meter at 9, 11, 13, 15, and 17 weeks.

The fasting plasma glucose (FPG) level from 12-h fasted mice was measured from the eyeball 17 weeks after administering the injection. The fasting serum insulin levels (FINS) were quantified by using an enzyme-linked immunoassay (ELISA) kit (Keno Spring Biotechnology Co., Ltd., Beijing, China). The homeostasis model assessment-IR (HOMA-IR) index was calculated as previously described (Hsu et al., 2013) using the following equation: FPG (mmol/L) × FINS (mU/L)/22.5.

The levels of advanced glycation end products (AGEs) and FFA were estimated with an ELISA kit (RENGEN Bioengineering Technology Co., Ltd., Beijing, China), and the glycated hemoglobin (HbA1c) level was detected in the blood drawn via a tail prick using HbA1c monitors (NycoCardREADER2; Axis-Shield Diagnostics Co., Ltd., Scotland, United Kingdom) at 17 weeks of injection.

After the animals were sacrificed, their excised pancreas was first rapidly fixed in 2.5% glutaraldehyde for 4 h and then fixed in 1% osmic acid for 1 h. After dehydration of the tissues with ethanol, the tissues were covered with epoxy resin and cut into ultrathin sections, followed by staining with lead citrate. Then, transmission electron microscopy (Hitachi, Japan) was used to observe the pancreas’ ultrastructural changes. The remaining tissues were fixed in 4% paraformaldehyde, embedded in paraffin, and then cut into 4-μm-thick sections and stained with HE or Masson trichrome staining. Finally, the stained section was examined for its morphological structure by a microscope (OLYMPUS, Tokyo, Japan).

The pancreatic tissues embedded in paraffin were dewaxed in xylene and rehydrated in gradient alcohol. After antigen retrieval and hydrogen peroxide (H₂O₂) blocking, the sections were next incubated with a primary antibody against INS-R, IRS-1, IRS-2, and Glut4. The positive expression was evaluated by using diaminobenzidine (ZSGB-BIO, Beijing, China) after incubation with the secondary antibody. Fluorescent images were observed and photographed by using the Bio-Rad Radiance 2,100 Confocal Microscope (Bio-Rad, Hercules, CA, United States).

MIN6 cells of 80% confluency in 48-well plates were cultured in RPMI-1640 medium for 24 h. The growth medium was removed on the day of the trial, after which a 5.5- or 33.3-mM glucose-containing medium was added to each well. After 24 h, the supernatant was collected, and the cells were stimulated with AOS (16 or 32 μg/ml) or metformin (32 μg/ml) for 60 min. The level of insulin secretion in the supernatant and cells were estimated using the Mouse Ultrasensitive Insulin ELISA Kit (Keno Spring Biotechnology Co., Ltd., Beijing, China).

Annexin V-FITC binding and propidium iodide (PI) staining were performed using flow cytometry (Becton Dickinson) according to the manufacturer’s protocol (Becton Dickinson Franklin Lakes, NJ, United States). Briefly, MIN6 cells were seeded in six-well plates and cultured for 24 h, pretreated with or without glucose (33.3 mmol/L) for 24 h, and then exposed to AOS (16, 32, and 64 μg/ml) for 24 h. Then, the supernatant was collected, and the attached cells were trypsinized and collected. The cells were washed twice with ice-cold PBS, resuspended in binding buffer, stained with annexin V-FITC and PI for 15 min, and incubated in the dark. Finally, a flow cytometer was used to analyze the samples. The total percentage of apoptotic cells was determined as the sum of the percentages of both early and late apoptotic subpopulations.

Total RNA was extracted from MIN6 cells using the TRIzol reagent (Invitrogen, Carlsbad, CA, United States) and used to synthesize the first-strand cDNA using the RevertAid First-Strand cDNA Synthesis Kit (Thermo Fisher Scientific). Quantitative real-time PCR (qRT-PCR) was performed using the Real-Time PCR Detection System (Bio-Rad). The specific primer sequences are summarized in Table 1. The results of three independent experiments performed for determining INS-R, IRS-1, IRS-2, and Glut4 mRNA levels were normalized against β-actin levels for each sample.

Total protein was extracted from MIN6 cells with cold RIPA lysis buffer containing proteases and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, followed by electrotransfer onto a PVDF membrane (Millipore, Billerica, MA, United States) and immunoblotting. Then, the membrane was blocked in 5% skim milk in Tris-buffered saline containing 0.1% Tween-20 for 1 h and incubated overnight at 4°C with anti-Glut4, anti-INS-R, anti-IRS-1, anti-IRS-2, and anti-GAPDH primary antibodies (1:1,000; Abcam, Cambridge, MA, United States). The membranes were washed three times in Tris-buffered saline and incubated with goat anti-rabbit horseradish peroxidase-conjugated secondary antibody (1:5,000; Gaithersburg, MD, United States) for 1 h. The expression of the specific protein was visualized using a commercial electrochemiluminescence kit (Invitrogen).

Ultra-performance liquid chromatography was coupled with time-of-flight mass spectrometry analysis (UPCL-TOF/MS^E).

Stock standard solutions of glucose (0.1 mol/L) and AOS (10 mg/ml) were prepared. The diluted solutions were separately prepared by mixing 200 μL of stock internal standard solution and AOS, and 200 μL of 0.4 mol/L AEC/NaBH3CN and 40 μL acetic acid. Each homogenized sample was weighed in a 15 ml plastic centrifuge tube and then heated in a water bath (70°C, 60 min). Before the analysis, a few amounts of the extract were diluted in a solution of methanol for preparing the sample solution. Subsequently, a membrane with a pore size of 0.22 μm was used to filter the solution. The volume of the injection was 10 μL.

The Xevo G2S TOF (Waters MS Technologies, Manchester, United Kingdom) mass spectrometer was used to perform mass spectrometric (MS) analysis and electrospray ionization was performed in the positive ion mode. The capillary voltage reached 3.0 kV. The low and high collision energies were 6 eV and 15–45 eV, respectively. The temperature of the ion source was 100°C. Atomization was performed at 400°C. The flow rate of desolvation reached 600 L/h. Leucine-enkephalin (leu-enk) was used as the lock mass, and the mass axis was corrected with sodium formate.

Spectral data analysis and quantification were completed using Mass Lynx V4.1 and Target Lynx software, respectively. Data acquisition and analyses or quantification were performed using the MassLynx version 4.1 software and the TargetLynx software, which adopt the same standard and calibration model as that of UPLC-QTOF/MS^E. A calibration curve of R ² > 0.99 met the data acceptance criteria.

The data are presented as the mean ± SD from at least three independent experiments. SPSS 17.0 (SPSS, Inc., Chicago, IL, United States) was used for statistical analysis. Significance was tested by a one-way analysis of variance. Differences between groups were considered statistically significant if the p-value was <0.05.

The model group showed significantly higher baseline RBG levels than the control group. The RBG levels in the AOS-treated group were significantly lower than those in the model group. After 8 weeks of AOS administration, the RBG levels in the HAOS and LAOS groups markedly decreased (Figure 2A). It was worth noting that the dose of AOS (380 mg/kg/day or 750 mg/kg/day) was determined in our previous study (Bao et al., 2020). As expected, the body weight of the mouse in each group increased gradually (Figure 2B). There was no obvious change in the growth trend during AOS administration, and the growth curve was similar to that of the model group.

As shown in Figure 2C–E, the levels of FPG and FINS and the HOMA-IR index were markedly increased in the db/db diabetic mice. Compared with the control group, the FPG and FINS levels and the HOMA-IR index in the AOS-treated groups were significantly lower. The HAOS group particularly showed a notable reduction in the FPG levels (Figure 2C–E). Although, there were no significant differences in FPG and FINS levels between the model and AOS-treated groups, AOS- and MET-treated groups showed a lower HOMA-IR index compared to the model group (Figure 2C,D).

As expected, the levels of HbA1c, FFA, and AGEs increased in the model group. The HAOS- and MET-treated groups showed lower HbA1c and FFA levels compared to the model group (p > 0.05, Figure 2F,G). Especially, the AGE level in the HAOS group was not different compared to that in the control group (Figure 2H). The levels of HbA1c and FFA in the LAOS-treated group were significantly higher than those in the HAOS- and MET-treated groups (Figure 2F,G).

We investigated the effects of AOS treatment on the pathological changes in the pancreas of all the groups by performing H&E staining and Masson trichrome staining. H&E staining showed that pancreatic islets in the control group exhibited normal characteristics such as the regular elliptical shape and a clear boundary. Compared to the control group, the number of islets in the model group was markedly lower. The pancreas was severely shrunken and the islets had no clear boundaries. AOS treatment reversed this pathological damage by improving the morphological arrangement of islets and increasing their number, in addition to restoring a clearer structure of the surrounding tissues. In particular, there were more pathological changes in the pancreas of the HAOS group than in the model group (Figure 3A). Next, the results of the Masson trichrome staining indicated that there was no blue collagen fiber in the pancreatic tissues of the control group, the cell contours were clear, the nucleus was centrally located, and the cell boundary was clear. In the model group, the blue-stained collagen fiber was observed between the pancreatic tissues. The blue collagen fiber in the AOS-treated groups was significantly less than that in the model group (Figure 3B).

We performed immunohistochemical staining by using antibodies against INS-R, insulin receptor substrates (IRS-1 and IRS-2), and Glut4 to determine the expression of INS-R, IRS-1, IRS-2, and Glut4 in the pancreas (Figure 4). Light staining of INS-R (Figure 4A), IRS-1 (Figure 4B), IRS-2 (Figure 4C), and Glut4 (Figure 4D) was observed in the pancreas of hyperglycemic db/db mice. On the other hand, INS-R, IRS-1, IRS-2, and Glut4 were strongly expressed in the pancreatic tissues of the control group than in the remaining groups.

Cell viability values of MIN6 cells treated with AOS were higher than those of cells without AOS treatment at concentrations of 0–2000 μg/ml (Figure 5A). It was linear in the range of 10–100 μg/ml, and the proliferation activity gradually decreased at 500 μg/ml and was toxic to cells at 4,000 μg/ml. Cell viability was better in the 16, 32, and 64 μg/ml concentration range (Figure 5B). Compared with the control group, high glucose (33.3 mmol/L) concentration significantly decreased the viability of MIN6 cells. AOS addition enhanced the viability of high glucose-treated cells (Figure 5C). In addition, in all the AOS-treated groups, there was no significant effect on insulin secretion by non-hyperglycemic-stimulated MIN6 cells, and there was no difference compared with that in the control group (Figure 5D). The concentrations of 16 and 32 μg/ml AOS promoted the insulin secretion by MIN6 cells stimulated by high glucose concentrations, and the concentration of 32 μg/ml AOS had the most obvious effect. However, there was no statistically significant difference among the groups when the MIN6 cells were incubated under high-glucose or AOS conditions (Figure 5E).

The apoptotic rate of the MIN6 cells at high glucose concentrations was significantly higher than that in the control groups, whereas, after the AOS treatment, the apoptotic rate was lower than that in the high-glucose-treated group. The 32 μg/ml concentration of AOS resulted in a greater reduction in apoptosis than treatment with 64 μg/ml AOS (Figure 6).

As shown in Figure 7A, the protein levels of INS-R, IRS-1, IRS-2, and Glut4 were significantly lower in the high-glucose group than in the control group. However, AOS treatment increased the levels of these proteins. Especially, the AOS-treated (64 μg/ml) group resulted in the highest increase in the protein levels. To further investigate the potential mechanisms of AOS, the mRNA levels of INS-R, IRS-1, IRS-2, and Glut4 in MIN6 cells were determined. Compared with the high glucose-treated group, we found that AOS did not significantly reduce the IRS-1 and Glut4 mRNA levels (Figure 7F,I). High glucose concentration induced a reduction in INS-R and IRS-2 mRNA levels, but AOS reversed these effects by markedly upregulating INS-R and IRS-2 mRNA expression (Figure 7G,H).

UPLC-Q-TOF/MS^E analysis of AOS was performed with precolumn derivatization with AEC. The reaction mechanism of AEC with sugars in AOS is shown in Figure 8. The enamine was generated by the reaction of the reducing end of AOS and the primary amine of AEC and then reduced to secondary amines by NaBH3CN so that sugars could be labeled by the AEC. The AEC-derived AOS solution was directly separated and analyzed by UPLC-QTOP/MS^E to obtain five types of sugars including glucose, lactose, rutinose, glucuronic acid, and maltotriose. These sugars were inferred by the molecular formula by obtaining a chromatographic peak and a fragment ion peak, as well as by referring to databases such as MassBank, Scifinder, and ChemSpider and general literature (Figure 9). Moreover, the contents of these five sugars of AOS were as follows: glucose, 1.38 ± 0.02 mg/g; lactose, 8.08 ± 0.48 mg/g; rutinose, 14.9 ± 0.5 mg/g; maltotriose, 0.12 ± 0.08 mg/g; and glucuronic acid, 15.35 ± 0.49 mg/g (Table 2).