2.1 Chemicals Reagents

Chenodeoxycholic acid, CA, DCA, LCA, Ursodeoxycholic acid (UDCA), Tauro-chenodeoxycholic acid (T-CDCA), Glyco-chenodeoxycholic acid (G-CDCA), Tauro-cholic acid (T-CA), Glyco-cholic acid (G-CA), Tauro- deoxycholic acid (T-DCA), Tauro-lithocholic acid (T-LCA), Tauro-β-muricholic acid (T-βMCA), Tauro-ursodeoxycholic acid (T-UDCA), and Glyco-deoxycholic acid (G-DCA) were purchased from Dalian Meilun Biotechnology Co. LTD. Hyocholic acid (HCA) was purchased from ChemCruzTM Biochemical company. Hyodexycholic acid (HDCA) and the internal standard (IS, DCA-d5) were purchased from Sigma-Aldrich (St Louis, MO). α-Muricholic acid (αMCA), β-muricholic acid (βMCA), ω-muricholic acid (ωMCA), Tauro-α-muricholic acid (T-αMCA), Tauro-hyodexycholic acid (T-HDCA), and Glyco-ursodeoxycholic acid (G-UDCA) were purchased from Toronto Research Chemicals Inc. (TRC). Kae and K-3-G were purchased from Chengdu Must Bio-Technology Co. Ltd. and Extrasynthese (Lyon, Rhône, French), respectively. K-7-G and K-7-S were obtained from Karebay Biochem (Shanghai, China). Phenylmethanesulfonyl fluoride and dithiothreitol were bought from Sigma-Aldrich Co. (St. Louis, MO, USA). The sequencing grade modified trypsin was obtained from Promega (Madison, WI). The peptides of DMEs and internal standard (IS, purity > 95%) were purchased from Your R&D Partner. HPLC-grade formic acid, acetonitrile, and methanol were acquired from Merck (Darmstadt, Germany). Coomassie brilliant blue for protein measurement was purchased from Bio-Rad (Hercules, California, USA).

2.2 Mouse Treatment

Male C57BL/6 mice were purchased from the Experimental Animal Center of Sun Yat-sen University. All animal experiments were conducted in strict accordance with the guidelines approved by the Institutional Animal Care and Use Committee of the International Institute for Translational Chinese Medicine (IITCM_20180140). The animals were housed in a controlled environment with a relative humidity of 50%–60% and an ambient temperature of 24°C–26°C, under a regulated 12-h light–dark cycle to ensure their well-being and the validity of the experimental conditions. Upon completion of the experimental procedures, the animals were subjected to a 12-h fasting period, during which they continued to have unlimited access to water. The mice were euthanized by cervical dislocation after anesthesia via inhaling isoflurane.

2.3 Experimental Design

Male C57BL/6 mice, aged 6 weeks, were randomly assigned to three experimental groups. The first group served as the control, receiving a diet supplemented with 2.5% Na₂CO₃ (Ctr, n = 8). The other two groups were treated with Kae at a dosage of 50 mg/kg (K50, n = 8) and 100 mg/kg (K100, n = 8), respectively. Firstly (0-day), blood samples (approximately 15 μL) were collected through snipping the mouse tail near its tip at 3, 6, 12, 20, 30, 60, 120, 180, 240, 360, 540, and 660 min after K50 and K100 treatment, respectively. Then continuous gavage was implemented daily for 2 weeks. When the second blood samples were collected (15-day), the time without Kae treatment was more than 48 h since the last Kae administration. The plasma was separated from blood sample by centrifuging at 8000 rpm for 5 min. Feces were collected at 15-day. Liver, plasma, gallbladder, and fecal samples were secured in a state of deep freeze at −80°C, immediately after sacrifice.

2.4 Pharmacokinetic Studies

The separated plasma (5 μL) and genistein (5 μL internal standard) were mixed, and 190 μL methanol was added. The mixture was swirled and centrifuged at 20,817 g for 30 min at 4°C. 160 μL supernatant was taken and dried in a nitrogen dryer, then redissolved with 50% methanol (100 μL) and centrifuged for LC–MS/MS analysis. The detailed experimental procedure of LC–MS/MS analysis was in accordance with our previously described method (Zheng et al. 2016; Liu et al. 2022).

2.5 BA Analysis via LC–MS/MS Method

The plasma, gallbladder, liver, and fecal samples were meticulously prepared for the quantification of BAs. We analyzed a comprehensive panel of 12 CBAs, encompassing T-CA, G-CA, T-CDCA, G-CDCA, T-βMCA, T-αMCA, T-DCA, G-DCA, T-UDCA, G-UDCA, T-HDCA, T-LCA, and 10 UBAs, including LCA, HDCA, ωMCA, CA, βMCA, HCA, αMCA, CDCA, UDCA, and DCA. These BAs were detected using a sensitive LC–MS/MS approach. The concentrations of BAs were measured in the gallbladder, plasma, feces, liver, and intestinal contents. The sample processing techniques and analytical conditions were meticulously aligned with the well-established methods previously reported in the literature (Li et al. 2022).

2.6 LC–MS/MS Analysis of DMEs

Mouse tissues (liver, intestine, colon, and kidney) were isolated, shredded, and washed with cold saline, respectively. A cold buffer saline solution with 0.28 mM phenylmethyl sulfonyl fluoride (PMSF) as a protease inhibitor was gently mixed with the minced tissues to facilitate homogenization. Then, the resulting suspension was subjected to centrifugation at a high speed of 9000 g for 20 min at 4°C. The suspension was collected and further hydrolysis to polypeptide for LC–MS/MS analysis. The meticulous processing methodologies and stringent testing conditions employed in this study were in strict accordance with those detailed in our previous research (Chen et al. 2017, Li et al. 2020).

2.7 Molecular Docking

The protein structure of human FXR (6HL1) was retrieved from the Protein Data Bank (PDB) (https://www.rcsb.org/). For the chemical structure of Kae (PubChem ID: 5280863), we got its three-dimensional structure from PubChem (https://pubchem.ncbi.nlm.nih.gov/). We used AutoDockTools to optimize the receptor protein for docking, adding hydrogen atoms, and balancing charges. AutoDock Vina 1.1.2 software (Trott and Olson 2010) was used to perform molecular docking of the receptor protein (6HL1) and small ligand (Kae) molecule. The complex of 6HL1 and endogenous agonist CDCA were further analyzed through Protein Plus debase (https://proteins.plus/).

2.8 The Assay of BSH Activity

Fecal samples (50 mg/mouse) were mixed with 500 μL of cold phosphate-buffered saline solution containing 1 mM PMSF using a motorized homogenizer. The mixture was then sonicated in an ice bath for 30 min and centrifuged at 12,000 g for 15 min at 4°C. Protein was obtained from the supernatant and the protein concentration was determined using the BCA protein assay kit (Bio-Rad, Beyotime, China). For the incubation process, the samples were incubated in a 37°C environment with a 3 mM sodium acetate buffer at pH 5.2, along with 0.1 μg of fecal protein and varying concentrations of the substrate T-CDCA-d5 (1 μM, 5 μM, and 10 μM). After a 30-min incubation period, 100 μL of the internal standard (LCA-d4) was added to the reaction mixture. Following centrifugation, the supernatant was carefully collected and prepared for LC–MS/MS analysis.

2.9 Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR)

The qRT-PCR was conducted according to our previous study (Li et al. 2022). Seventeen genes were selected as target detectors: Cyp7a1, Cyp27a1, Fxr, Shp, Fgr15, Baat, Bacs, Cdo1, Csad, and Taut. A house keeping gene, Gapdh, was applied as the internal control. The levels of gene expression were determined by qPCR using the Power SYBR Green PCR master mix (Applied Biosystems, CA, USA) in QuantStudio5 system (Applied Biosystems). Primer sequences are detailed in Table S1. mRNA expression levels were standardized to Gapdh for consistency, with data analyzed using the 2^−ΔΔCt formula.

2.10 Western Blot

Mouse liver samples were collected and two samples per group were combined for protein analysis. Three replicates of these pooled samples were maintained for each experimental set. Total protein was isolated using RIPA buffer from Sigma-Aldrich. The proteins were then examined with antibodies specific to β-ACTIN, CYP27A1, CYP7A1, and FXR (all Santa Cruz, diluted 1:1000). Detection was facilitated by secondary antibodies against mouse and rabbit IgG (SAB, diluted accordingly). The methodology followed was in line with our previous reports, ensuring consistency in processing (Li et al. 2022).

2.11 Fecal DNA Extraction and ERIC-PCR Analysis

Total genomic DNA was extracted from fecal samples using QIAamp DNA Stool Mini Kit (QIAGEN) according to the manufacturer's protocol. The extracted DNA samples were analyzed for the similarity of GM composition using ERIC-PCR as described (Huang et al. 2017). The resulting DNA banding patterns on the gel were digitized by Image Lab 3.0 system (Bio-Rad). Based on the distance and the intensity of each DNA bands, SIMCA-P 14.0 tool (Umetrics, Umea, Sweden) with 95% (p < 0.05) confidence level was applied to obtain the PLS-DA score plots (Huang et al. 2017).

2.12 16S rRNA Amplicon Sequencing

The methods used for processing were in accordance with our established protocols (Li et al. 2022).

2.13 Data Analysis

Statistical differences were assessed using one-way ANOVA for normally distributed data and Kruskal-Wallis H for non-normally distributed data, both conducted in SPSS 19.0. Alpha and beta diversity analysis and partial least squares discriminant analysis (PLS-DA) were analyzed using R and SIMCA-P 14.0 tool, respectively. Correlation analyses were performed using SPSS 19.0, according to the Pearson product–moment correlation for normal related data and Spearman's rank correlation for non-normally related data. Significance levels were set at p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***).

3.1 Pharmacokinetic Characteristics of Kae, K-7G, K-3G, and K-7S in C57BL/6 Mice

15-day Kae treatment showed no significant differences in food intake, water consumption, and body weight among the Ctr, K50, and K100 groups (Figure S1a). Additionally, none of the treated mice showed variations in the organ index (Figure S1b). At 0-day and 15-day, the concentrations of Kae, K-7G, K-3G, and K-7S were determined and their mean plasma concentration-time curves are displayed in Figure 1, respectively. The corresponding pharmacokinetic parameters are shown in Table S2. Once orally administered, Kae was quickly absorbed and metabolized into K-7G, K-3G, and K-7S. For Kae, from the parameters of the time to peak drug concentration (T_max) and the maximum concentration (C_max), it can be inferred that long-term intervention with Kae can accelerate the absorption of Kae in body; for K-3G, C_max was 1.9-fold higher than the 0-day groups, p = 0.07; for K-7G, C_max was increased compared with the 0-day groups (1.9- and 1.3-fold, p = 0.23 and 0.51, respectively); for K-7S, no significant changes were observed. For K-3G and K-7G, the area under the plasma concentration-time curve from 0 to last time (AUC_0–t) and the area under the plasma concentration-time curve from 0 to infinite time (AUC_0–∞) display similar increased trend after Kae treatment for 15 days (1.1~1.8 folds).

3.2 Effect of Kae on the Expression of DMEs in Liver Tissue

After 15-day Kae treatment with a dosage of K50 and K100, we observed the up-regulation of protein amounts in liver tissue (Figure 2a), including CYP2C29 (1.7- and 1.9-fold), 2D22 (1.6-fold), 2E1 (3.5- and 2.8-fold), and 3A11 (1.8-fold), compared with the control group; the UGT isoforms were observed higher expression, including UGT1A1 (2.5- and 2.6-fold), 1A6a (1.8- and 1.9-fold), 1A9 (1.8- and 1.7-fold), 2A3 (1.7- and 1.6-fold), 2B5 (1.8- and 1.9-fold), and 2B34 (1.8- and 1.9-fold); there was no significant change in SULT1A1 and SULT1D1. Among them, the amounts of CYP7A1 and CYP27A1 display an opposite result, an increased and a decreased expression after K50 and K100 treatment (1.3~1.7-fold, p < 0.05). There are no significant changes in intestine, colon, and kidney tissue (except for UGT2A3 in intestine, UGT1A5 and 2B5 in kidney tissue, Figure 2b–d).

3.3 Effect of Kae on the Expression Levels of CYP7A1 and CYP27A1

Based on the results from Figure 2, we further confirmed the expression of CYP7A1 and CYP27A1. Figure 3a demonstrates that the Cyp7a1 mRNA levels were downregulated by 1.7-fold in the K50 group and 2.8-fold in the K100 group compared to the Ctr, both with statistical significance (p < 0.05). And the mRNA expression of Cyp27a1 was increased by 1.7- and 1.4-fold in Kae-treated groups (p > 0.05). We further analyzed the relationship between mRNA expression and protein amounts of CYP7A1 and CYP27A1, and we found that there was a closed relationship between the mRNA expression and the protein amounts (r = 0.55, p = 0.01 and r = 0.463, p = 0.035, Figure 3b). Moreover, the western blotting results also showed that the CYP7A1 was downregulated in K100 group. Whereas the expression of CYP27A1 was upregulated in both K50 and K100 groups compared to the Ctr group (Figure 3c,d).

3.4 Effect of Kae on BAs Pool in C57BL/6 Mice

There were 12 CBAs and 10 UBAs, which were detected using LC–MS/MS method (Figure 4 and Figure S2). In liver tissues, T-CA and T-DCA markedly rose by 2.5 to 7.5-fold in the K50 and K100 groups versus the Ctr (p < 0.05); in the gallbladder, the amounts of T-DCA and T-LCA were observably upregulated in K100 group than in the Ctr group (1.4 folds, p < 0.05); in duodenal contents, the K50 group exhibited significant upregulation of T-βMCA, T-CDCA, and G-CA by 2.8-, 1.9- and 1.7-fold respectively compared to the Ctr group (p < 0.05); in plasma, the levels of T-αMCA, T-βMCA, and T-LCA were significantly upregulated in the Kae treatment groups compared with the Ctr group (1.4–37.5 folds, p < 0.05).

3.5 Kae Activates the Liver FXR Expression and Reduced the BSH Activity

The Fxr mRNA level was significantly increased in the K100 group compared with the Ctr group (1.4-fold, p < 0.05, Figure 5a). Shp and Fgf15 display an increasing tendency without statistical significance. Especially for the Fgf15, there is significant individual variation among the groups (Figure S3). Western blotting results showed that the expression of FXR was significantly increased in the K50 and K100 groups when compared to the Ctr group (1.9- and 1.4-fold, respectively, Figure 5b,c). The molecular docking results showed that the predicted binding sites between Kae and human FXR with the amino acid residues TYR361, TYR369, HIS447, SER332 by hydrogen bonding, and MET290 through hydrophobic interactions (Figure 5d). The docking score reaches −7.93 kcal/J (Figure 5e). Then we further analyzed the complex of 6HL1 and endogenous agonist CDCA through Protein Plus debase and found that the main pharmacodynamic groups of the agonist CDCA are similar to Kae's. Moreover, the activation site (the binding amino acid residues) of CDCA on the protein is identical to that of Kae (Figure 5f,g). Figure S4 also indicates that FXR (gene name Nr1h4) acts as an important regulatory molecule in the process of Kae regulating BA metabolism.

As shown in Figure 5h, there are no significant differences in liver enzymes responsible for the synthesis of conjugated BA following Kae treatment. While, Kae slightly reduced the BSH activity in mouse feces after Kae treatment, especially in the low-dose group (p = 0.044, Figure 5i). Consequently, we further investigated the abundance of bacteria capable of secreting BSH hydrolases. As shown in Figure 5j, Lactobacillaceae significantly decreased in the K50 and K100 groups when compared to the Ctr group (3.1- and 3.7-fold, respectively, p < 0.05); Clostridiaceae decreased in the K50 and K100 groups than in the Ctr group (1.5 and 1.6 folds, respectively, p < 0.05); Enterobacteriaceae also showed a decrease in the Kae treatment groups compared with the Ctr group (4.1 and 2.1 folds, respectively, p < 0.05).

3.6 The Change of GM Composition and the Association With Differential BA

In this study, we first investigated the GM profile of the fecal DNA samples from experimental mice using the ERIC-PCR technique. The PCA plot showed a clear separation among the K50-, K100-, and the control groups (Figure 6a). To further analyze the alteration of GM composition after Kae treatment, 16S amplicon sequencing was performed. Alpha diversity analysis showed that K50 and K100 increased the observed, Chao 1, and Shannon index compared to the control group (Figure 6b). Furthermore, weighted UniFrac distance analysis showed a difference between the mice treated with Kae and control mice (Figure 6c). At the phylum level, K50 enhanced Firmicutes' prevalence while reducing Tenericutes. K100 significantly increases the relative abundance of Bacteroidetes but decreases Tenericutes (Figure 6d). Moreover, the difference also existed among the groups in the family taxon (Figure 6e). The Bacteroidaceae, Lachnospiraceae, and Porphyromonadaceae were significantly increased, while the Clostridiaceae markedly decreased after Kae treatment. The previous reports showed that changes in GM composition, including Bacteroides, Clostridium, Lactobacillus, Bifidobacterium, and Listeris, were correlated to the de-conjugation of BSH (Jia, Xie, and Jia 2017). Here, our results (Figure 7) also showed that the metabolic products of BA, i.e., T-CA, T-DCA, and the total CBA in liver tissue were negatively related to the species of Bacteroides uniformis (B. uniformis), Parabacteroides chinchilla (P. chinchilla), Ruminiclostridium clostridium methylpentosum (R. clostridium methylpentosum), Ruminiclostridium eubacterium siraeum (R. eubacterium siraeum), Lactobacillus reuteri (L. reuteri), Bacteroides sartorii (B. sartorii), Bacteroides massiliensis (B. massiliensis), Clostridium sp., and Lachnoclostridium clostridium bolteae (L. clostridium bolteae) (p < 0.05). However, the concentration of T-CA, T-DCA, and the total CBA were positively related to the species of Bacteroides acidifaciens (B. acidifaciens) and Bifidobacterium choerinum (B. choerinum) (p < 0.05).

3.7 Kae Elevated the Beneficial GM in C57BL/6 Mice

Figure 8 and Figure S4 indicate that Kae can act as a prebiotic. Notably, several beneficial bacteria, including B. acidifaciens, B. choerinum, Eubacterium coprostanoligenes (E. coprostanoligenes), and Lachnoclostridium saccharolyticum (L. saccharolyticum) were increased in the mice treated with Kae (Figure 8a). Comparatively, K100 effectively suppressed the growth of certain pathogenic bacteria such as Alistipes, Citrobacter spp., and Helicobacter spp. (Figure 8b). Additionally, we further conducted a correlation analysis between these bacteria and key liver proteins related to BA synthesis and metabolism. The results (Figure S5) indicated that Cyp7a1 showed a negative correlation with probiotic Butyricicoccus pullicaecorum (B. pullicaecorum) and a positive correlation with harmful bacterium Alistipes putredinis (A. putredinis); FXR was significantly negatively correlated with harmful bacterium Alistipes massiliensis (A. massiliensis); Bacs and Taut exhibited significant positive correlations with some probiotics, such as B. acidifaciens and Bamesiella viscericola (B. viscericola).