Ji-Ni-De-Xie was obtained from Dege County Tibetan Hospital (Institute of Tibetan Medicine) (Sichuan, China). Metformin hydrochloride tablets were purchased from Sino American Shanghai Shiguibao Pharmaceutical Co., Ltd (Shanghai, China) (Lot number: ABS9845); STZ (Lot number: WXBD7077V) was purchased from Sigma-Aldrich Co., St (Louis, United States). The enzyme-linked immunosorbent assay (ELISA) kits of insulin (Ins) (No. MM-0587R1), glycosylated serum protein (GSP) (No. MM-0735R1), tumor necrosis factor-α (TNF-α) (No. MM-0180R1), lipopolysaccharide (LPS) (No. MM-0647R1), glucagon-like peptide 1 (GLP-1) (No. MM-0033R2), and FGF15 (No. MM-0938R1) were obtained from Jiangsu Meimian Industrial Co., Ltd (Jiangsu, China). The ELISA kits of interleukin-6 (IL-6) (No. EK306/3-96) and IL-1β (No. EK301B/3-96) were provided by MULTISCIENCES (LIANKE) Biotech, Co., Ltd (Zhejiang, China). Triglyceride (TG) (No. C061-a), total cholesterol (TC) (No. C063-a), low-density lipoprotein cholesterol (LDL-C, No. C070-a), and high-density lipoprotein cholesterol (HDL-C) (No. C069-a) were purchased from Changchun Huili Biotech CO., LTD. Primary antibodies FXR/NR1H4 Rabbit mAb (1:1500, A24015, ABclonal Technology, China), Rabbit anti-CYP7A1 (1:1000, bs-21430R, Bioss, China), and anti-FGF15 (1:1000, sc-398338, Santa Cruz Biotechnology Inc., United States). Secondary antibodies goat anti-rabbit (1:10000, bs-0295G-HRP, Bioss, China) and HRP goat anti-mouse IgG (H + L) (1:10000, AS003, ABclonal Technology, China). β-actin (1:1000, GB15003, Servicebio, China).

JNDX was ground and sifted through No. 3 sieve, and then accurately weighed 0.1 g into 100 mL conical bottle. 50 mL 50% methanol was added into bottle, and then ultrasound was performed for 30min. After cooling, make up the weight with 50% methanol, and filter the supernatant with 0.22 μm microporous filter membrane.

The chemical constituents of JNDX were analyzed by HPLC-QQQ-MS (Agilent 1260, Agilent Technologies Inc., United States). The analysis was performed using a WondaSil C₁₈-WR chromatographic column (4.6 × 250 mm, 5 μm). Then, the mobile phase was water containing 0.1% formic acid A)–methanol B). The column temperature was 25°C and the sample volume was 5.0 μL. The gradient program’s specific condition was set as follows: 0–1 min, 10%–45% B; 1–3 min, 45%–70% B; 3–4 min, 70%−70% B; 4–5 min, 70%–75% B; 5–7 min, 75%–75% B; 7–10 min, 75%–95% B; 10–20 min, 95% B. The mass spectrum conditions were as follows: electrospray ion source (ESI), high purity nitrogen (dry gas) as a collision gas, with a flow rate of 11 L∙min^-1; capillary voltage was set at 4000 V (+) and 2500 V (−); temperature was maintained at 300°C with a pressure of 15 psi. Simultaneous determination of positive and negative ions was conducted in multiple reaction monitoring (MRM) mode.

Male Sprague-Dawley rats (weight 160–180 g, SPF level) and HFD were purchased from Chengdu Dashuo Experimental Animal Co., LTd. [Laboratory animal license number SCXK (Sichuan) 2020-030, Sichuan, China] and then adaptively housed in 25°C ± 2°C and relative humidity 50%–60% for 1 week. The rats were held in the Experimental Animal Center of Chengdu University of Traditional Chinese Medicine, and the water and diet were unrestricted supplied. The experimental unit license number is SYXK (Sichuan) 2020-124. Chengdu University of Traditional Chinese Medicine’s Animal Ethics Committee consented its approval to this research (the ethical approval number: 2018–15). The study was conducted in compliance with the guidelines set out by the National Health Institutes of China.

According to JNDX instruction, the adult clinical oral administration dose of JNDX is 1 to 2 pills/day (mass is 1.1537 g/pill). Thus, the maximum dose for adults is 0.0330 g/kg/d (assuming a body weight of 70 kg), and the high dose for rats in the present study is 0.330 g/kg/d, the medium dose is 0.165 g/kg/d, and the low dose is 0.083 g/kg/d (Chen, 2006). The specific groups and dosages are shown in Table 2. The JNDX powder was sieved through the 4^th screen, and 0.05% sodium carboxymethyl cellulose (CMC-Na) solvent was used to dissolve the powder and metformin (positive control group) respectively.

After grinding, 50 g of JNDX was weighed and mixed with 10 times the volume of 50% ethanol. It was then heated and reflux-extracted twice for 2 h each time. The filtrate was mixed and concentrated under reduced pressure, resulting in a JNDX extract with a concentration of 0.625 g∙mL^-1. Sprague-Dawley rats were randomly divided into normal group and JNDX group. The JNDX group received an oral gavage of 12.5 g∙kg^-1 of JNDX extract, while the normal group received an equivalent volume of distilled water for three consecutive days. Oral gavage was administered twice daily, with the rats fasting but not restricted from water intake for 12 h before the final administration. Blood samples (0.5 mL) were collected from the orbital vein at 0.25 h, 0.5 h, 1 h, 1.5 h, 2 h, 3 h, 5 h, and 8 h after the final gavage, centrifuged at 3500 rpm for 10 min, and the serum was collected and stored at −80°C. For the JNDX group, 1 mL of serum was collected and mixed with 3 mL of acetonitrile to precipitate proteins. After centrifuging at 11000 rpm for 10 min, the supernatant was dried with nitrogen gas at 37°C. The residue was reconstituted with 200 μL of methanol, centrifuged at 11000 rpm for 10 min, and the supernatant was filtered through a 0.22 μm microporous membrane. Finally, we analyzed the filtrate using ultra high performance liquid chromatography with four stage rod/electrostatic field orbital trap high resolution mass spectrometry (UPLC-Q-Exactive Orbitrap MS) (Thermo-Fisher, United States).

The mass spectrometry conditions were as follows: using HESI ion source, positive ion detection mode with a spray voltage of +3500 V; negative ion detection mode with a spray voltage of −3000 V; sheath gas flow rate of 35.0 Arb, auxiliary gas flow rate of 10.0 Arb, and capillary temperature of 320°C. Full MS/dd-MS² scan mode was employed with a Full MS resolution of 70,000 and dd-MS² resolution of 17,500. The scan range was from m/z 100 to 1500.

High-performance liquid chromatography analysis was performed using a ZORBAX SB-C₁₈ chromatographic column (150 mm × 2.1 mm, 1.8 μm). Then, the mobile phase was methanol A)–water containing 0.1% formic acid B). The column temperature was 30°C and the sample volume was 5.0 μL. The gradient program’s specific condition was set as follows: 0–20 min, 5%–20% A; 20–30 min, 50%–80% A; 30–35 min, 80%–95% A; 35–45 min, 95% A; 45–50 min, 95%–5% A (Luo et al., 2022a).

Sprague-Dawley male rats were randomly divided into 2 groups, 8 were normal group, and the rest were model screening group for T2DM rats. The rats in the model group were fed with HFD for 4 weeks and then fasted for 12 h. On the following day, 1% STZ buffer solution (45 mg∙kg^-1) was prepared for intraperitoneal injection in the model screening group. Besides, the rats in the normal group were injected with an equivalent amount of 0.1 mol∙L^-1 citric acid-sodium citrate buffer solution (pH = 4.1). After STZ injection, the rats had free access to food and water, and fasting blood-glucose (FBG) was measured after fasting for 12 h at 72 h, 2 weeks, and 4 weeks, respectively. The FBG of each rat was measured three times by blood glucose meter (Anwen + code, Sinocare Inc.). The rats with three FBG indices greater than or equal to 11.0 mmol∙L^-1 and the index of FBG steadily were considered as T2DM model rats.

A total of 40 model rats were selected and randomly divided into 5 groups: model group (model), metformin group (metformin), JNDX high-dose group (JNDX-H), JNDX medium-dose group (JNDX-M), and JNDX low-dose group (JNDX-L), with 8 rats in each group. All model rats were raised together with 8 rats in the normal group (normal). During the experiment, the normal group was fed with ordinary diet, and the model group and the JNDX group were fed with HFD.

After the successful modeling of STZ in Sprague-Dawley rats, the rats were given continuous administration of JNDX for 40 days. Subsequently, the rats were fasted for 12 h and then FBG was measured by glucose meter. Blood samples were collected from abdominal aorta, and then the whole blood was left for 2 h, and centrifuged at 3500 rpm at 4°C for 15 min to obtain the serum. The serum was stored at − 80°C for detection. The contents of TG, TC, LDL-C, and HDL-C were measured by an automatic biochemical analyzer (BK-200, Shandong Biobase Scientific Instrument Co., LTD.). The ELISA kits were performed to detect the levels of GSP and Ins and then calculated the value of ho-meostasis model assessment of insulin resistance (HOMA-IR) and insulin sensitivity index (ISI). All experimental procedures were followed strictly with manufacturer’s instructions. HOMA-IR and ISI were calculated according to the following formulas: ISI = Ln   1 / ( F B G   m m o l / m L × I n s   μ U / m L HOMA − IR = F B G   m m o l / L × I n s   μ U / m L / 22.5

(Xu et al., 2021).

Fecal sample DNA of rats was extracted from Magnetic Soil and Stool DNA Kit (TIANGEN BIOTECH Co., LTD., Beijing, China) and metagenomic sequencing was performed. The DNA samples were fragmented into 350 bp fragments using ultrasonication for Illumina sequencing and subsequent PCR amplification. Following clustering, the library preparations were sequenced using the NovaSeq 6000 sequencing platform with PE150 read length.

The relative abundance information of gut microbiota was imported into R software (Version 4.2.2) for statistical analysis. The R package phyloseq was used to calculate α-diversity. α-diversity was studied to explore the complexity of species diversity using Shannon index (Lkhagva et al., 2021). Principal coordinates analysis (PCoA) based on Bray Curtis distance was conducted to explore the characteristics of gut microbiota. Anosim analysis was used to compare the similarities and differences in the composition of gut microbiota between the model and JNDX groups. Functional predictions of gut microbiota were performed using Kyoto Encyclopedia of Genes and Genomes (KEGG) and DIAMOND software (Kanehisa, et al., 2006; 2014; Buchfink, et al., 2015) based on metagenomic sequencing data. The Linear Discriminant Analysis Effect Size (LEfSe) software was employed to calculate the differences in functional pathways among groups at the family, genus, and species levels (LDA score >4.0, p < 0.05) (Segata, et al., 2011).

The cryopreserved serum samples were thawed at room temperature and swirled evenly. Then, 100 μL serum sample was precisely measured and 10 μL internal standard (1.00 μg∙mL^-1) (Cholic acid-d₄, Lot number: ISO-13098, Nanjing Beiyu Biotechnology Co., LTD.) and 200 μL chromatograph-grade acetonitrile were added. The mixture swirled at room temperature for 2 min, stood for 10 min, and centrifuged at 11 000 rpm at 4°C for 10 min. The supernatant was processed through 0.22 μm microporous filter membrane. The serum samples of rats in each group were analyzed by a triple quadrupole mass spectrometer (Agilent G6420A, Agilent Technologies Inc., United States). The analysis method was established on the basis of our previous research work (Yi, 2022).

The levels of LPS, IL-6, IL-1β, TNF-α, GLP-1, and FGF15 were estimated using ELISA kits. The experiment process was carried out in strict accordance with the manufacturer’s instructions.

100 mg of the tissue was taken and fully grinded in 1 mL of pre-cooled lysis buffer. Total RNA was extracted from tissue using Trizol (Lot number: 9109, TAKARA). The concentration of total RNA was measured by ultramicro quantitative nucleic acid protein detector (EVA 3100, Monad Biotechnology Co. LTD., Jiangsu, China). Subsequently, 1 μg RNA template was prepared for reverse transcription using a M-MLV Reverse Transcriptase kit (M170A, Promega Biotech Co., Ltd., Beijing, China) and iScript ™ cDNA Synthesis Kit (1708890, Bio-Rad Laboratories, Inc., United States) was used to synthesize cDNA. Finally, RT-qPCR was performed via fluorescent quantitative PCR instrument (CFX Connect, Bio-Rad Laboratories, Inc., United States). The total reaction system of fluorescence quantitative detection was 10 μL. In addition, the amplification steps were as following: 95°C, 5 min for initial denaturation step, and 95°C, 10 s to 60°C, 30 s for 40 cycles of denaturation, and anneal at 65°C–95°C, 0.5°C/5 s. The primer sequences in this study were presented in Table 3. The 96-well plate was put into the fluorescence quantitative PCR instrument, and the detection was set on the Bio-Rad CFX Manager. The expression multiple was calculated using the ΔΔCT method, and the differences in the samples of relative mRNA expression were normalization by reference genes (GAPDH): ΔCT = CT target gene − CT reference gene, ΔΔCT = ΔCT treatment sample − ΔCT control sample, the analyzed formula = 2^−ΔΔCT (Su et al., 2023).

Animal tissues were homogenized and lysed using freezing grinder (JXFSTPRP-CL, Shanghai Jingxin Industrial Development Co., Ltd., Shanghai, China) and RIPA lysis buffer (AR0102S, BOSTER, China). Then, the samples were centrifuged at 10,000 g for 10 min at 4°C. The supernatant of each sample was obtained and the protein concentration of the sample was measured with BCA kit (AR0146, BOSTER, China). Whereafter, the SDS-PAGE sample loading buffer (AR1112-10, BOSTER, China) was added in total protein and denatured at 100°C for 10 min to obtain WB samples. 10% PAGE high resolution color gel was prepared to separate proteins of different molecular weight, and the isolated proteins were transferred to polyvinylidene difluoride (PVDF) membranes, which were then blocked with 5% BSA at room temperature for 90 min. The membranes were placed in incubator box containing primary antibody at 4°C overnight: FXR (1:1500), CYP7A1 (1:1000), and FGF15 (1:1000). On the second day, after the recovery of the primary antibody, membranes were washed three times with TBST. Next, membranes covered with the second antibody (1:10000) and incubated for 90 min at room temperature. After incubation, the membranes were washed three times and exposed to ultra-sensitive ECL chemiluminescent substrate kit (BL520A, Biosharp, China). Finally, visualizer (ChampChemi 610 Plus, Beijing Sage Creation, China) was employed to visualize the protein bands and ImageJ software (version 1.37V, NIH, United States) was performed to quantify the relative gray value of bands.

The data were presented as means ± standard deviation (SD). GraphPad Prism 8 (San Diego, CA, United States) was performed for the statistical analysis. One-way analysis of variance (ANOVA) was conducted to determine the p-value, with statistical significance established at a p-value <0.05.

The chemical constituents of JNDX were analyzed by HPLC-QQQ-MS technique. A total of 12 constituents (magnoflorine, berberine, curcumin, jateorrhizine, demethoxycurcumin, gallic acid, chebulagic acid, ferulic acid 4-O-β-D-glucopyranoside, hydroxysafflor yellow A, rutin, ferulic acid, and ellagic acid) were identified in JNDX by comparing their retention time and mass spectrum information with standard compounds. In addition, the contents of the 12 constituents in JNDX were calculated according to their calibration curves established by the corresponding standard compounds. The results showed that gallic acid had the highest content (18.11 mg∙g^-1), followed by chebulagic acid (4.98 mg∙g^-1), magnoflorine (3.27 mg∙g^-1), ellagic acid (3.17 mg∙g^-1), ferulic acid 4-O-β-D-glucopyranoside (3.07 mg∙g^-1), and ferulic acid (1.83 mg∙g^-1). The content determination results, mass spectrometry information and chromatogram of JNDX are shown in Supplementary Table S1, S2 and Supplementary Figure S1.

A total of 45 chemical compounds in rat serum after JNDX intervention were identified using UPLC-Q-Exactive Orbitrap MS technology, including 22 prototype components and 23 metabolites. The detailed chemical compounds in rat serum and the diagram of positive and negative ion flow are shown in Supplementary Figure S2 and Supplementary Table S3. These components included berberine, magnoflorine, ferulic acid, ellagic acid, jateorrhizine, demethoxycurcumin, and curcumin. They mainly originated from the combination of B. kansuensis C.K.Schneid., C. longa L., P. emblica L., and T. chebula Retz., suggesting that these four herbs contributed significantly to the treatment of T2DM with JNDX. In conclusion, the results indicate that berberine, magnoflorine, ferulic acid, ellagic acid, jateorrhizine, demethoxycurcumin, and curcumin may be the effective components of JNDX treating T2DM.

After the successful modeling of T2DM rats, the typical diabetes symptoms of eating more, drinking more water, excreting more, while the weight loss were obvious. Figure 1 depicts the specific experimental procedure. The effect of JNDX on T2DM rats was evaluated by measuring relevant blood glucose and insulin-related indices. Figure 2A shows that FBG is significantly higher in model group than in normal group (p ˂ 0.001). After 40 days of administration, both the high dose and medium dose groups of JNDX could significantly reduce FBG (p ˂ 0.01), while the effect of FBG reduction was not significantly different in the low dose group. These results indicate that JNDX could significantly reduce FBG in T2DM rats. In addition, GSP was similar to glycosylated hemoglobin and reflected the average blood glucose concentration over the past 1–3 weeks. The results indicate that the GSP level of the T2DM rats in the model group is significantly higher than that in the normal group (p ˂ 0.001). After 40 days of administration, GSP content was obviously decreased at JNDX-H and JNDX-L group compared with model group (p ˂ 0.05) (Figure 2B).

Ins value of rats in each group were measured by kit. HOMA-IR and ISI were calculated according to FBG and ins of rats after the last administration. Results are shown in Figures 2C, D. Compared with the normal group, the HOMA-IR of the model group was markedly increased (p ˂ 0.001), and the ISI was significantly decreased (p ˂ 0.001). Compared with the model group, HOMA-IR was obviously decreased (p ˂ 0.05), while the ISI was increased observably (p ˂ 0.05) in JNDX-H and JNDX-M groups. The results suggest that JNDX can improve insulin resistance and insulin sensitivity in T2DM rats.

To further investigate the effect of JNDX on inflammation in T2DM rats. We found that the serum levels of LPS, IL-1β, TNF-α, and IL-6 were significantly increased in model group (Figures 3A–D). Both high- and medium-dose JNDX markedly decreased the level of LPS, IL-1β, TNF-α, and IL-6 (p ˂ 0.05), whereas the low-dose JNDX decrease mildly in these indexes. These results demonstrate that JNDX can improve inflammation in T2DM rats.

Through the detection of blood lipid indexes of rats in each group, we found that TG, TC, and LDL-C of T2DM rats in model group were upregulated compared with normal rats (p ˂ 0.001). Nevertheless, after JNDX intervention, these indicator levels were reversed (Figures 3E–G). Besides, the expression of HDL-C was lower in the model group than in the normal group, and JNDX-H could noteworthily increase the level of HDL-C (p ˂ 0.05) (Figure 3H). These results suggest that JNDX can ameliorate the blood lipid in T2DM rats.

Metagenomic analysis of fecal samples was performed to assess the effect of JNDX on gut microbiota in T2DM rats. The Shannon index was significantly different between the normal and model groups (p = 0.0112). There was no significant difference between the model and the JNDX group (Figure 4A). The results showed that JNDX had little effect on the α diversity of gut microbiota. However, PCoA analysis showed a clear separation among the three groups, suggesting that JNDX significantly affected the composition of the gut microbiota (PCoA1, 37.42%, PCoA2, 20.68%, p = 0.001) (Figure 4B). The relative abundance of gut microbiota in normal, model, and JNDX groups at the family, genus, and species levels are presented in Supplementary Figure S3 and Supplementary Table S4. Figures 4C, D show significant enrichment of gut microbiota in model and JNDX rats at the family, genus, and species levels based on LEfSe analysis. The results suggest that JNDX can improve the gut microbiota disorder in T2DM rats by increasing the relative abundance of some bacteria (Coriobacteriaceae, Streptococcaceae, Butyricicoccus, Streptococcus, Subdoligranulum, Butyricicoccus porcorum, Faecalibacterium prausnitzii, and Subdoligranulum variabile) and decreasing the relative abundance of some bacteria (Erysipelotrichaceae, Lactobacillaceae, Allobaculum, Bacteroides, Ligilactobacillus, Allobaculum stercoricanis, Dorea sp CAG317, and Ligilactobacillus animalis). Among them, Butyricicoccus, B. porcorum, and F. prausnitzii have been proven to be beneficial bacteria, promoting improvement in T2DM (Heinken et al., 2014; Xu et al., 2015). Additionally, Allobaculum and Bacteroides have been demonstrated to be harmful bacteria, detrimental to the treatment of T2DM (Wang et al., 2020; Chen et al., 2023). Notably, some bacteria such as Streptococcus and Bacteroides are closely associated with BAs metabolism, indicating that JNDX may further influence BAs profile in T2DM rats.

HPLC-QQQ-MS analyzed 18 BAs, including cholic acid (CA), dihydrocaffeic acid (DHCA), glycocholic acid (GCA), and glycoursodeoxycholic acid (GUDCA), etc. in the serum of rats. The data of each group were analyzed by GraphPad Prism software, and the results are shown in Figure 5. The total BAs, total primary BAs, and total secondary BAs are illustrated in Supplementary Figure S4. Compared to the normal group, the contents of 6 primary BAs [CA, tauroursodeoxycholic acid (TUDCA), taurochenodeoxycholic acid (TCDCA), tauro-alpha-muricholic acid (T-α-MCA), tauro-beta-muricholic acid (T-β-MCA), and taurocholic acid (TCA)] in the model group were significantly decreased (p ˂ 0.05), and the secondary BAs such as deoxycholic acid (DCA) and taurodeoxycholic acid (TDCA) were also markedly reduced (p ˂ 0.001) compared to the normal group. Besides, the primary BAs ursodeoxycholic acid (UDCA) was significantly increased (p ˂ 0.001) and CA was remarkable decreased (p ˂ 0.01) in comparison to those in the normal group. Nevertheless, compared with the model group, UDCA level was significantly decreased (p ˂ 0.001) and CA level was significantly increased (p ˂ 0.001) observably in JNDX group. The results indicate that JNDX can affect BAs metabolism.