Coptis chinensis Franch. and T. ruticarpum (A. Jussieu) T. G. Hartley were purchased from Beijing Tongrentang Co., Ltd. (China). Ursodeoxycholic acid (UDCA), α-naphthylisothiocyanate (ANIT), and sodium pentobarbital were obtained from Merck Pharmaceuticals, Inc. (Germany); and olive oil was from Shanghai Yi En Chemical Technology Co.

The characteristics of the botanical drugs used in these four combinations are shown in Table 1. The dosage of each medicinal plant was based on the dosage range specified in the "Pharmacopoeia of the People’s Republic of China.” Each group of botanical drugs was prepared according to the decoction method stipulated in the "14th Five-Year Plan” of the National Higher Education Materials on TCM. Briefly, each group of botanical drugs was mixed separately and immersed in 20 times the volume of purified water. Heating was started after 30 min and continued for an additional 30 min. The liquid was filtered and collected; the above procedure was repeated; and the twice-collected liquid was mixed and stored at −20°C.

The active metabolites in these herbal decoctions were identified by Q-Orbitrap high-resolution LC-MS. Information on the equipment used, sample handling procedures, mass spectrometry conditions, and chromatographic conditions is included in Supplementary Material 1. The decoctions were subjected to high-resolution liquid chromatography, with the results collected using CD 3.3 (Compound Discoverer 3.3, Thermo Fisher Scientific) and compared with the database (mzCloud, https://www.mzcloud.org/) to identify the compounds. Determination of the activity of orally administered Chinese botanical drugs requires overcoming barriers due to absorption, distribution, metabolism, and excretion processes. The active metabolites obtained from the comparison were further screened by assessing their oral bioavailability (OB) (Xu et al., 2012) and drug-likeness (DL), with substances having a DL index ≥0.18 considered highly drug-like (Guo et al., 2019). The OB and DL of each active metabolite was assessed using the TCMSP database (https://old.tcmsp-e.com/molecule.php?qn=2649). Active metabolites were screened based on their OB, DL, and comparison with the database (mzCloud, https://www.mzcloud.org/), with active metabolites regarded as molecules with an OB ≥ 30%, a DL ≥ 0.18, and score of an mzCloud best match ≥80.

Thie animal protocol was approved by the Animal Ethics Committee of Chengdu University of Traditional Chinese Medicine (Experimental Animal Welfare Ethics Review Certificate No. 2023024), and the animals were handled strictly according to internationally recognized animal management and rules.

The rat model, the administration of drugs by gavage, and the procedures for anesthesia, blood collection, and extraction of liver tissues have been described (Yu et al., 2022a). Briefly, 42 healthy male Sprague Dawley rats weighing 250 ± 20 g were purchased from Chengdu Da Shuo Experimental Animal Co., Ltd., production license number SCXK (CHUAN) 2020-0030. After 6 days of acclimatization feeding, the rats were randomly divided into seven groups of six rats each: the control group (Control), the model group (Model), the CF low dose group (CF_L), the CF high dose group (CF_H), the CF low dose CF plus TR group (CT_L), the CF high dose plus TR group (CT_H), and the positive control group (UDCA). Figure 1. shows the entire process of model construction and drug administration by gavage. Cholestasis in rats was induced by intragastric administration of ANIT, at a dose of 50 mg/kg, dissolved in olive oil. Intragastric administration of CF and CT was determined relative to body surface area. After gavage administration on day 17, all rats were fasted for 16 h. After collecting fresh feces from each rat on day 18, the rats were anesthetized by intraperitoneal injection of sodium pentobarbital 30 mg/kg. Blood samples were collected, the rats were sacrificed, and liver tissue samples were collected. Feces and liver tissues were frozen and stored at −80°C.

Rat blood samples were centrifuged at 3,500 × g for 15 min at 4°C to obtain serum. ALT, AST, ALP, γ-GT, DBIL, TBIL, and TBA concentrations in serum were determined using a fully automated biochemical analyzer (BS-240VET). Liver tissue samples were fixed with 4% paraformaldehyde, washed with water, dehydrated, embedded, and sectioned. The sections were stained with HE and examined under a microscope for pathological determination.

The anti-inflammatory effects of treatment were evaluated by determining the levels of expression of TNF-α, IL-1β, IL-6, and IL-10 mRNAs in rat liver tissue by real-time quantitative PCR (qRT-PCR), using qRT-PCR primers (Table 2) based on the mRNA sequences in the NCBI database.

The effect of regulating lipid metabolism was evaluated by staining liver tissue with oil red O, as well as by measuring TC and TG concentrations in blood samples. The serum concentrations of GSH, ROS, NO, MDA, and SOD were determined by micro-enzyme assay (cat. no. A006-2), enzyme-linked immunosorbent assay (cat. no. YJ206302), colorimetry (cat. no. E-BC-K035-S), tribarbituric acid assay (cat. no. A003-1), and hydroxylamine assay (cat. no. A001-1), respectively.

Metabolites in each group of rats were analyzed by gas chromatography-mass spectrometry (GC-MS). The analytical process included metabolite extraction and derivatization, detection by GC-MS, data pre-processing, and statistical analysis. Additionally, 16S rRNA amplicons were sequenced to determine the fecal microbial characteristics of rats in each group. This analysis included DNA extraction, PCR amplification, library construction and sequencing, and bioinformatics analysis. Both tests were performed by Shanghai OE Biomedical Technology Co., Ltd., and their methodology and procedure have been described (Yu et al., 2021). Further details can be found in Supplementary Material 2.

Differences among multiple groups were assessed by one-way ANOVA or the Kruskal Wallis test. All statistical analyses were performed using GraphPad Prism 9, with p < 0.05 considered statistically significant.

The active metabolites of medicinal plants observed in each group are shown in Figure 2. The primary active metabolites in the CF_L and CF_H groups were berberine, glycitein, and arachidonic acid, whereas the primary active metabolites in the CT_L and CT_H groups were these three metabolites, along with evodiamine, catechin, isorhamnetin, obacunone, and quercetin (Table 3).

Low and high doses of CF, as well as their combinations with TR, exhibited significant anti-cholestatic activity, with a therapeutic efficacy similar to that of UDCA (Figures 3A–G). Pathohistological examination showed that all groups of botanical drugs reduced the inflammation and liver cell necrosis induced by ANIT (Figures 3H–O). Taken together, these findings indicated that CF and CT had significant therapeutic efficacy against cholestasis.

CF_L significantly reduced the concentrations of inflammatory factors IL-1β and IL-6 in liver tissues of cholestatic rats, showing some anti-inflammatory efficacy, whereas CF_H significantly reduced IL-1β, TNF-α, IL-6, and increased IL-10 (Figure 4). CF had significant anti-inflammatory effects, which were more pronounced at high than low CF doses, indicating a quantitative relationship between CF dose and anti-inflammatory activity. The addition of TR, however, did not enhance the anti-inflammatory effects of both low and high doses of CF. Interestingly, CT_H was less effective than CF_L in reducing IL-6 and increasing IL-10 concentrations, suggesting an antagonism between the anti-inflammatory effects of the two drugs in the treatment of cholestasis.

The effects of treatment on lipid metabolism were determined by measuring the levels of lipids in serum and liver samples. Using an automatic biochemical analyzer, lipid-related indices (Figures 5A–D) were detected in the serum of each group of rats. Steatosis was assessed by oil red O staining of liver tissue (Figure 5E). The results indicated that either CF or CT could significantly reduce the accumulation of lipid droplets in the liver tissue of cholestatic rats and improve lipid metabolism disorder. CT_L had a significantly more substantial hypolipidemic effect than CF_L, suggesting that CF and TR had a synergistic hypolipidemic effect in the treatment of cholestasis.

The anti-oxidative stress effects of treatment were determined by measuring serum concentrations of GSH, ROS, SOD, NO, and MDA in these rats (Figure 6). CF_L did not show a significant anti-oxidative stress effect, whereas CF_H had a highly significant anti-oxidant effect, perhaps because only high doses of CF have anti-oxidant stress efficacy. The addition of TR, however, did not significantly enhance the anti-oxidative stress effect of CF.

The overall distribution among the samples and the stability of the whole analysis process were initially determined by unsupervised principal component analysis (PCA) (Figures 7B–F). Subsequently, supervised orthogonal partial least squares analysis (OPLS-DA) was used to differentiate the overall differences in metabolic profiles among the groups and to identify the differentially expressed metabolites in these groups (Figures 7G–J). The quality of the model was evaluated by using seven cycles of interactive validation and 200 response ordering tests to prevent model overfitting (Figures 7K–N), with the model parameters shown in Figure 7A. The above analyses showed significant metabolic differences between the CF_H and CF_L, the CT_L and CF_L, and the CT_H and CF_H groups.

Metabolites differentially expressed by pairs of groups were screened by unidimensional and multidimensional analyses. In the OPLS-DA analysis, variables important in projection (VIP) were used to measure the strength of the influence of the metabolite expression pattern on the classification discrimination of each group of samples and to screen for significant differences in metabolites. Significant differences in the levels of different metabolites between groups were verified by t-tests. The criteria for screening were VIP >1 and p < 0.05, with volcano plots (Figures 8A–D) showing the expression trends and numbers of differential metabolites in between group comparisons. Comparisons of the Model and Control groups showed that 617 metabolites were downregulated and 479 upregulated in the former. Comparisons of the CF_H and CF_L groups identified 222 metabolites that were downregulated and 184 that were upregulated in the former. Furthermore, comparisons of the CT_L and CF_L groups found 255 metabolites downregulated and 307 upregulated in the former, whereas comparisons of the CT_H and CF_H groups identified 269 metabolites downregulated and 288 upregulated in the former. Cluster statistical analyses of between group comparisons screened the top 50 differentially expressed metabolites (Figures 8E–H). Differences in the therapeutic effects on cholestasis produced by different doses of CF and by combination with TR may be associated with the differential expression of these metabolites.

The presence of differentially expressed metabolites was likely due to differences in the anti-cholestatic efficacy of various doses of CF and their combinations with TR. Pathway enrichment analysis of these differentially expressed metabolites may reveal the mechanisms underlying these differences in anti-cholestatic metabolic pathways. KEGG enrichment analysis identified overlapping pathways in comparisons of the Model and Control groups (Figure 8I) and the CF_H and CF_L groups. These pathways included those involving ABC transporters, steroid biosynthesis, and linoleic acid, arachidonic acid, and tyrosine metabolism (Figure 8J). These pathways may be responsible for the differences in the anti-inflammatory, hypolipidemic, and anti-oxidative stress effects of high and low doses of CF. Differences in the efficacy of CT_L and CF_L in treating cholestasis were found to be related to differences in the biosynthesis of unsaturated fatty acids, in ABC transporters and in CAMP signaling pathway. The difference in efficacy of CT_H and CF_H was associated with the modulation of linoleic acid metabolism (Figures 8K, L).

Based on the results of amplicon sequence variant (ASV) clustering analysis, the number of shared and unique ASVs in the rat groups were analyzed and plotted as a flower plot (Figure 9A). Three ASVs were shared by the Control, Model, CF_L, CF_H, CT_L and CT_H groups. Figure 9B summarizes the relative abundances of each sample at different taxonomic levels, including of phylum, class, order, family, genus, and species.

The reasonableness of the sequencing volume of the samples was evaluated by alpha diversity analysis, with this evaluation including the adequacy of the sampling volume, and the richness and homogeneity of the species in each sample (Figures 9C–E). Between-group differences in diversity were evaluated by beta diversity analysis, including PCoA and NMDS analyses (Figures 9F–N). Fecal microbial communities were found to differ significantly in the CF_H and CF_L, CT_L and CF_L, and CT_H and CF_H groups. These effects on the composition of the gut microbial community and its function may also be responsible for the differences in the efficacy of CF in treating cholestasis at different doses and after combination with TR.

The species of differentially present microorganisms and their functions were further clarified by LEfSe and PICRUSt2 analyses. Figure 10A shows the significant species with relatively high abundance at the phylum, class, order, family, and genus levels, as determined by LEfSe analyses. Figures 10B–J show the top 10 species that differed at the genus level in between-group comparisons. Comparisons of the microbiomes obtained from the Model and Control groups identified Alloprevotella and Prevotella as being differentially expressed in the CF_H and CF_L groups, Prevotella and UCG-005 as being differentially expressed in the CT_L and CF_L groups, and Lactobacillus and Prevotella being differentially expressed in the CT_H and CT_L groups. 16S-based prediction of KEGG function revealed that, in treating cholestasis, the difference between CF_H and CF_L may be related to the biosynthesis of amino acids, the difference between CT_L and CF_L may be related to 2-oxocarboxylic acid metabolism, and the difference between CT_H and CF_H may be related to the biosynthesis of amino acids.

Based on fecal metabolomics and analyses of microbial diversity analyses, differentially expressed metabolites and microorganisms were subjected to Spearman correlation analyses, thereby assessing the relationship between fecal metabolites and genera of fecal microbiota (Figure 11). Fecal metabolites were found to strongly correlate with fecal microbiota. Differences in the therapeutic effects of different doses of CF and different doses of CF paired with TR on cholestasis may be due to differences in the regulation of both fecal metabolites and fecal microbiota.