Berberine hydrochloride was purchased from Shanghai Source Leaf Biological Technology Co., Ltd. (Shanghai, China), the purity of BBR is ≥98%. Oil red O reagent, high-performance liquid chromatography grade formic acid and ammonium acetate, and BA standards including cholic acid (CA), CDCA, DCA, ursodeoxycholic acid (UDCA), α, β and ω muricholic acids (MCA), tauroursodeoxycholic acid (TUDCA), taurochenodeoxycholic acid (TCDCA), tauro α-muricholic acid (TαMCA), tauro β-muricholic acid (TβMCA), tauro ω-muricholic acid (TωMCA), TCA, and taurodeoxycholic acid (TDCA) were obtained from Sigma-Aldrich (St. Louis, Mo, United States) and the purity of the above reagents was higher than 98%.

Male C57BL/6J mice (6–8 weeks old, 22 ± 2 g) were purchased from SLAC Laboratory Animal Co., Ltd. (Shanghai, China). Male FXR knockout (FXR^−/-) mice on a C57BL/6J background were given as a present by Prof. Li Yang from Shanghai University of Traditional Chinese medicine. After 1-week acclimatization, the mice were fed with chow diet (normal control group, n = 8) or high-fat diet (HFD, 60% of calories derived from fat, Research Diets, NJ, United States) supplemented with 1% dextran sulfate sodium (DSS) in drinking water periodically (7-day DSS administration followed by a 10-day interval). After 12 weeks of the diet, the HFD-DSS fed C57BL/6J mice were then randomly divided into NASH group (vehicle control, continuously supply of HFD-DSS, n = 8) and BBR-treated group (with HFD-DSS and BBR treatment, n = 8). The LD50 of BBR is 713.58 mg/kg in mice, and the present study applied the routine dosage of 100 mg/kg/d to the NASH mice. BBR was suspended in 0.5% carboxymethyl cellulose sodium solution (CMC-Na) and administered to the mice by gavage (0.1 ml/10 g body weight) once a day for 4 weeks, the normal control and vehicle control (NASH) mice were given an equal volume of 0.5% CMC-Na solution. For HFD-DSS fed FXR^−/−mice, BBR and vehicle treatment were the same as C57BL/6J mice. All animals were maintained under controlled temperature (25°C ± 2°C) and humidity (60 ± 5%) at a 12 h light-dark cycle, and access to food and water ad libitum. Food consumption and body weight were recorded weekly. The animal experiments were all approved by the Animal Experiment Ethics Committee of Shanghai University of Traditional Chinese Medicine.

At the end of the experiment, all animals were fasted for 12 h, anesthetized intraperitoneally with 2% pentobarbital sodium, and sacrificed. To obtain the serum samples, blood was collected and centrifugated at 4°C, 3,000 rpm for 15 min. Serum lipid profiles including triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-c), and serum enzymes as indicators of liver damage such as alanine transaminase (ALT) and aspartate transaminase (AST) were analyzed using the Hitachi full-automatic system.

For measurement of hepatic TG content, liver tissues were rapidly excised and rinsed with precooled normal saline, the same portion of liver samples were cut and homogenized with 10 volumes of ice-cold ethanol in a homogenizer. The homogenate was extracted overnight at 4°C and centrifuged at 4°C, 3,000 rpm for 15 min. The organic layer was removed and the content of TG and TC were measured using commercial kits from Nanjing Jiancheng Bioengineering Institute (Nanjing, China) according to the manufacturer’s instruction.

Liver histological changes were examined according to the previously described method (Li Q. et al., 2020). Briefly, for hematoxylin and eosin (H&E) staining, liver tissues were fixed in 10% formalin solution for 24 h, dehydrated, paraffin-embedded, sectioned to ∼5 μm thickness, and stained with H&E reagent (Kohypath, Shanghai, China). For Oil Red O (ORO) staining, frozen liver tissues were embedded in Tissue-Tek OCT Compound (Sakura, Tokyo, Japan), cut into ∼8 μm frozen sections, and stained with ORO reagent (Sigma, St. Louis, MO, United States). Images were captured under a Nikon Eclipse 50i microscope (Nikon, Tokyo, Japan) with a magnification of 200×.

Fecal samples were collected from the cecum of the mice, and total DNA in feces was isolated using the Qiagen QIAmp® Fast DNA Stool Mini Kit (Qiagen, CA, United States) according to the recommended protocol. The extracted DNA from each sample was used as a template to amplify the V3-V4 region of bacterial 16S rDNA genes of distinct regions was carried out by PCR, and then further gel-purified and quantified. Purified amplicons were pooled in equimolar and sequenced on the Illumina MiSeq^TM platform (paired-end) by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The raw sequencing reads were generated and quality-filtered by QIIME version 1.9.1 with optimized parameters and analyzed using Mothur version 1.33.0 (http://www.mothur.org/), UPARSE version 7.1 (http://drive5.com/uparse/) and R version 3.6.1 (http://www.R-project.org/). The raw data of 16S rDNA was deposited at https://www.ncbi.nlm.nih.gov/Traces/study/?acc=PRJNA753953.

Serum, liver, and fecal samples were collected, and the BAs were extracted and quantified by ultra-performance liquid chromatography tandem-mass spectrometry (UPLC/MS) according to our previous research (Li Q. et al., 2020).

Liver tissues were lysed with radioimmunoprecipitation assay lysis buffer (Beyotime Institute of Biotechnology, Shanghai, China) with protease and phosphatase inhibitors (Roche, Indiana, United States). The prepared protein was separated by 12% SDS-PAGE, transferred to polyvinylidene difluoride membranes (Millipore, Temecula, CA, United States), and incubated with the primary antibody and secondary antibody, respectively. Anti-FXR, anti-FGF15, and anti-small heterodimer partner (SHP) were obtained from Abcam (Cambridge, United Kingdom), anti-cholesterol 7-alpha hydroxy-lase (CYP7A1) and anti-cytochrome P450, family 8, subfamily B, polypeptide 1 (CYP8B1) were purchased from Santa Cruz Biotechnology (Texas, CA, United States), anti-nuclear factor-κB (NF-κB) p65 and anti-NF-κB P-p65 were purchase from Cell Signaling Technology (Beverly, MA, United States). As an internal control, β-actin was purchased from HuaBio (Hangzhou, China). Anti-rabbit IgG and anti-mouse IgG was obtained from Cell Signaling Technology (Beverly, MA, United States). The bands were visualized by ECL chemiluminescence detection kit (Millipore, MA, United States), and quantified using the Tanon 5200 Chemiluminescent Imaging System (Tanon Science & Technology Inc., Shanghai, China).

Quantitative Real Time-PCR (RT-qPCR) was conducted as described previously (Li Q. et al., 2020). Briefly, total RNA was extracted from liver and intestinal tissues using TRIzol reagent (Thermo Fisher Scientific, United States), reversely transcribed into cDNA by reverse transcription kits (Promega, Madison, WI, United States), and subjected to PCR on a StepOne Applied PCR system (Applied Biosystems, Carlsbad, CA, United States) using PowerUp SYBR Green Master Mix (TOYOBO, Osaka, Japan). The relative gene expression was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) using a 2^−ΔΔCt method. The gene function and primer sequences were listed in Table 1.

The quantitative data were presented with mean ± SEM and analyzed by one-way analysis of variance (ANOVA) and unpaired two-tailed Student’s t-test using SPSS 18.0 statistical software (Chicago, IL, United States) and GraphPad Prism version 8.0 (San Diego, CA, United States). p < 0.05 was considered statistically different.

To address the effects of BBR on NASH, we used HFD-DSS induced NASH model. The treatment of DSS is shown to increase portal liposaccharide (LPS) level and therefore exacerbates the progression of NASH (Gabele et al., 2011; Achiwa et al., 2016). Mice fed with HFD-DSS demonstrated a typical NASH phenotype as evidenced by increased body weight and liver weight, obvious hepatic steatosis and inflammation, as well as increased liver TG content. During the 4-week treatment, all the mice were well-groomed, alert, active and showed good condition. The stool color of BBR-treated mice was a little bit yellow due to the drug administration. BBR treatment significantly reduced body weight and liver weight by 10 and 8%, respectively (Figures 1A,B), improved hepatic steatosis and inflammation (Figures 1C,D), and reduced liver TG content by approximately 30% (Figure 1E). The serum levels of AST and ALT in NASH mice were significantly increased, and BBR treatment could decrease the serum ALT and AST levels by 30% each (Figure 1C). Assembly, serum TG, TC, and LDL-c levels were increased in NASH mice, and BBR treatment partially restored these parameters (Figure 1F). These results implied that BBR exhibits robust efficacy against NASH in HFD-DSS fed mice.

Gut dysbiosis is associated with the pathogenesis of NASH, and modulating gut microbiota is reported to contribute to the efficacy of BBR in various diseases. Here we analyzed the overall structural changes of gut microbiota in response to BBR by 16S rDNA sequencing. A total of 3,274 operational taxonomic units (OTUs) were captured. UniFrac distance-based principal coordinate analysis (PCoA) (Figure 2A) and Bray-Curtis distance (Figure 2B) analysis revealed distinct clustering of microbe communities among control mice, NASH mice, as well as BBR treated mice. To assess the overall composition of the bacterial community in different groups, we analyzed the degree of bacterial taxonomic similarity at the phylum level. Compared to control mice, NASH mice displayed a higher Firmicutes-to-Bacteroidetes ratio due to the significant decrease in the relative abundance of Bacteroidetes, whereas BBR treatment increased the relative abundance of Bacteroidetes (Figure 2C). At the family level, BBR treatment significantly increased the relative abundance of Clostridiales, Lactobacillaceae, and Bacteroidale (Figures 2D–G). Assembly, these increased bacteria are participating in bile salt hydrolase (BSH) activity and secondary BA production, suggesting that BBR reshapes gut microbiota and might affect BA profiles in NASH mice.

The interaction of gut microbiota and BA metabolism has been reported in NASH development and progression (Sydor et al., 2020). Enriched BSH producing microbiomes upon BBR treatment indicated that the BA profiles might alter accordingly. BBR treatment decreased the total BA levels in the liver (Figure 3A) and serum (Figure 3B) and increased the content of fecal BAs (Figure 3C). BA profiling (Supplementary Tables S1–S3) showed that fecal CDCA, UDCA, DCA, and MCA of NASH mice were all significantly increased in response to BBR treatment, whereas the conjugated BAs in feces were unchanged (TDCA, TCA, TωMCA) or decreased (TβMCA) upon BBR treatment (Figure 3D). BBR remarkably decreased the percentage of liver βMCA and TβMCA (Figure 3E), but had no obvious effects on serum BA profiles (Figure 3F). These results suggest that BBR promotes intestinal unconjugated and secondary BA production in NASH mice.

Certain BA species (e.g., DCA, UDCA) that are mediated by BBR are known FXR agonists, so we next investigated the protein expression of intestinal FXR in mice. The protein expression of intestinal FXR showed a 50% decrease in NASH mice, and BBR treatment could reverse the decrease of FXR expression (Figures 4A,B, Supplementary Figure S1). Up-regulation of intestinal FXR promotes the production of FGF15 in epithelium cells, and BBR-treatment prevented the decrease of intestinal FGF15 levels in NASH mice. Intestine-derived FGF15 may enter the liver through the portal vein, and the liver FGF15 expression was also significantly increased upon BBR treatment. We also detected the mRNA expression of liver FXR, SHP, and FGFR4, but did not find any statistical difference between NASH mice and BBR treated mice (Figure 4C). FGF15 is a suppressor of BA synthesis, and CYP7A1 and CYP8B1 are the main enzymes of BA synthesis that could convert cholesterol into CA, CDCA, and β-MCA. We found that the protein expression of CYP7A1 and CYP8B1 of BBR treated mice were all significantly decreased in comparison to untreated NASH mice, whereas the expression of liver FXR and SHP was not statistically different between BBR treated mice and untreated NASH mice (Figure 4D; Supplementary Figure S2). These results indicate that the activation of intestinal but not liver FXR inhibits the BA synthesis.

To confirm the effect of BBR on intestinal FXR activation and FGF15 secretion, we detected the downstream lipogenesis genes in the liver, and found that BBR treatment significantly decreased the mRNA expression of sterol-regulatory element-binding protein 1 (SREBP1), acetyl-CoA carboxylase (ACC), fatty acid synthase (FASN), and stearoyl-CoA desaturase-1 (SCD1) in NASH mice (Figure 5A), suggesting the suppression of de novo lipogenesis upon BBR treatment. In addition, BBR treatment decreased the ratio of phosphorylated p65 to total p65 (Figure 5B; Supplementary Figure S3), indicating the suppression of NF-κB activation in the liver. Accordingly, the mRNA expression of inflammatory cytokines tumor necrosis factor alpha (TNF-α) (Figure 5C) and interleukin-6 (IL-6) (Figure 5D) was also decreased in BBR treated mice. These results indicate that BBR suppresses lipogenesis and inflammatory pathways via intestinal FXR activation.

To confirm the effect of BBR on FXR, we conducted experiments on FXR^−/− mice. The results showed that BBR treatment has no effect on body weight and liver weight in FXR^−/− mice (Figure 6A). The serum levels of ALT and AST were also comparable between BBR treated and untreated FXR^−/− mice (Figure 6B). FXR^−/- mice demonstrated obvious hepatic steatosis and inflammatory cell infiltration in the liver, whereas the effects of BBR on improving hepatic steatosis were negligible in FXR^−/− mice (Figures 6C,D). Consistently, the serum levels of TG, TC, and LDL-c, as well as the mRNA expressions of TNF-α and IL-6 were not statistically different between BBR treated and untreated FXR^−/− mice (Figures 6E,F). These results suggest that the beneficial effects of BBR are abolished in FXR^−/- mice.