2.1 Materials and reagents

DNT formulae samples (batch number: 170105, 231225, 231224) were purchased from Shanghai Hutchison Pharmaceuticals (Shanghai, China). Kaempferol (purity ≥ 98 %, batch number: 01–2007) and emodin (purity ≥ 98 %, batch number: 05–2004) were purchased from Shanghai Standard Technical Service Co., Ltd. (Shanghai, China). Luteolin (purity ≥ 98 %, batch number: C29N10Q104574) and apigenin (purity ≥ 98 %, batch number: M29GB150104) were obtained from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China). Ursodeoxycholic acid (UDCA) was purchased from Sigma-Aldrich (Merck, Darmstadt, Germany). ANIT was purchased from Sigma-Aldrich (Merck, Darmstadt, Germany).

High-performance liquid chromatography (HPLC) grade methanol, acetonitrile, and formic acid were purchased from Fisher Scientific Co. (Santa Clara, CA, USA). Deionized water was purified using a Milli-Q Academic System made by Millipore (Billerica, MA, USA). Sodium carboxymethylcellulose (CMC-Na) was obtained from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China).

2.2 Animal experiment

All animal studies were conducted by regulations of experimental animal administration issued by the State Committee of Science and Technology of the People's Re-public of China as well as the Ethics Committee of Shanghai University of Traditional Chinese Medicine (PZSHUTCM201023007). Male C57BL/6 J mice (20 ± 2 g) were purchased from Shanghai Slac Laboratory Animal Co. Ltd. (Shanghai, China). The mice were maintained on a standard chow diet and water with ad libitum access and housed in a 12-h light/12-h dark cycle environment. All animals were allowed for a 1-week quarantine and acclimation period before the experiment.

The mice were randomly divided into 4 groups, namely the vehicle control group (VEH), ANIT-induced cholestasis group (ANIT), DNT (3 g/kg) treatment group (ANIT + DNT), and UDCA (90 mg/kg) treatment group (ANIT + UDCA), with 10 mice per group. DNT formulae sample was ground into powder (100 mesh, 165 mm) and suspended in 0.5 % CMC-Na aqueous solution to obtain a final solution of 0.3 g/mL of DNT. Similarly, UDCA was dissolved in 0.5 % CMC-Na aqueous solution to obtain a final solution of 9 mg/mL. ANIT was dissolved in olive oil to obtain a final solution of 10 mg/mL. Mice in ANIT + DNT and ANIT + UDCA groups were respectively treated with DNT (3 g/kg) and UDCA (90 mg/kg) by intragastric administration for seven consecutive days, while mice in the VEH and ANIT groups received an equal volume of 0.5 % CMC-Na aqueous solution each day. On the fifth day (12 h after DNT or UDCA treatment), mice in the ANIT, ANIT + DNT, and ANIT + UDCA groups received a single oral administration of ANIT (100 mg/kg), while mice in the VEH group received an equal volume of olive oil.

At 48 h after the ANIT treatment, mice were euthanized under anesthesia for the collection of blood and liver samples. Blood samples were kept undisturbed for 1 h at room temperature, then the sera were collected by centrifugation at 4,000 g for 15 min at 4°C. An aliquot of the left lobe of the liver was immersed in 10 % PBS-buffered formalin at room temperature, and all other samples were stored at −80°C for future use.

2.3 Serum biochemical analysis and the liver histological examination

Serum liver function indexes, including alanine transaminase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total bilirubin (TBIL), direct bilirubin (DBIL), and total bile acid (TBA), were determined with an automatic biochemical analyzer (Hitachi 7080, Japan).

After fixing in 10 % PBS-buffered formalin for 24 h, the liver tissue was embedded in paraffin. Samples were subsequently sectioned (3 μm) and applied for hematoxylin & eosin (H&E) staining. Images were captured from six randomly selected microscopic fields on each slide using a Leica microsystem (Wetzlar, Germany).

2.4 LC-MS analysis

2.5 Identification of DNT-derived xenobiotic chemicals in mice

The samples, including DNT formulae samples, serum and liver samples from VEH, ANIT, and ANIT + DNT groups, were applied to the LC-MS analysis as described previously. High-resolution LC-MS data, including retention time, accurate mass, peak intensity, and MS/MS spectra, were obtained from PeakView with the MasterView plugin. The DNT-derived xenobiotic chemicals in mice were fast identified by comparing these data with available chemical reference standards and those reported in public references.

2.6 Identification of potential endogenous markers in mice responsible for the anti-cholestatic effect of DNT

Raw LC-MS data in the “(dot) wiff” format was preprocessed using the PeakView with MasterView plugin. The program XCMS was used for peak alignment, peak picking, and data normalization. The missing values of metabolic data were removed according to the modified 80% rule. Principal component analysis (PCA), an unsupervised pattern recognition approach, was applied to obtain trends in the overall metabolic profile across the groups. Metabolic data matrices between any two groups were analyzed by orthogonal projections to latent structures-discriminant analysis (OPLS-DA), a supervised pattern recognition approach. A preliminary variable of importance in the projection (VIP) value was selected for primary screening. Peak intensities with a significant fold change of > 1.5 and a P-value (by Student’s t-test) of < 0.05 indicated potential endogenous markers. An online biochemical database service METLIN and Chemspider were applied to identify and verify those potential endogenous markers. The MS² fragmentation information was matched by MetFrag and MS-DIAL, combining the network MS/MS information databases, including MassBank.

2.7 Pearson’s correlation analysis

Pearson’s correlation analysis was performed to explore the Q-markers, i.e., the xenobiotics derived from DNT responsible for the anti-cholestatic effect of DNT in ANIT-induced cholestasis mice. Pearson’s correlation analysis was performed using the DNT-derived xenobiotics and potential endogenous markers in rodent samples by MetaboAnalyst. The correlation coefficient (r), reflecting the intrinsic relations between different variable groups, was obtained and used as the indicator to establish the correlation between the endogenous biomarkers and the chemical components in DNT from serum and liver samples. In this study, a value of |r| > 0.8, which suggests a high correlation, was set as the criteria.

2.8 Network pharmacology study

The target prediction of the Q-markers was performed by the Swiss Target Pre-diction database (https://www.swisstargetprediction.ch/). The GeneCards (https://www.genecards.org/) and DisGeNET database (https://www.disgenet.org/) with “NASH”, “Cholestasis”, and “Cholecystitis” as keywords were used to search for disease targets. The intersection of data revealed the targets related to the efficacy of DNT. Then, a compound-target-disease network was constructed using the Cytoscape software to suggest the linkage of Q-markers to pharmacodynamic-related targets. Through the protein–protein interaction (PPI) analysis of the String database (https://string-db.org/), the PPI network diagram of the therapeutic effect of DNT was obtained to explore any therapy-associated protein factors. Metascape (https://metascape.org/), a platform for genetic function annotation, was applied for Gene Ontology (GO) as well as the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. The top 10 levels of biological process (BP), cellular component (CC), and molecular function (MF) were described for GO enrichment analysis. KEGG is a database that integrates genome, chemical, and system function information, which can associate gene catalogs obtained from fully sequenced genomes with higher-level system functions at the cellular, species, and ecosystem levels. A target-pathway network was constructed using the top 20 enriched KEGG pathway.

2.9 Quantitative real-time PCR

Total RNA was extracted from liver tissue by using an RNA Faster200 reagent kit (Fastagen, Shanghai, China), and cDNA was synthesized by applying the PrimeScript™ RT Master Mix Kit (Takara, Shiga, Japan) in accordance with the manufacturer’s instructions. The Real-time PCR analysis was performed on a 6000 Real-Time PCR System (Thermo Fisher Scientific) using the SYBR Premix Ex Taq™ Kit (Takara, Shiga, Japan). And the relative expression of target genes was quantified using the 2^−ΔΔCt method, normalized to that of Gapdh. The primer sequences are shown in Supplementary Table S1.

2.10 Statistical analysis

Experimental data were expressed as mean ± standard deviation (SD). Statistical analysis was performed with the two-tailed Student’s t-test or one-way analysis of variance (ANOVA) using Prism 7.0 (GraphPad Software, Inc., USA). The normality test was performed first in data analysis, and when the normality test failed, the nonparametric test was selected. The F-test was used to judge the homogeneity of variance, and different test methods were selected. A p-value < 0.05 was considered statistically significant.

3.1 DNT attenuates ANIT-induced liver injury in mice

ANIT is a well-recognized hepatotoxin that can efficiently induce cholestasis in rodents (Sundaram et al., 2017). According to our previous study (Ding et al., 2012), DNT showed a dose-dependent protective effect on liver injury in rats. Therefore, DNT (3 g/kg) was chosen to test the anti-cholestatic effect of DNT in present study. The two central veins were located in the center of the liver slices of the VEH group animals, and the structure of the portal area was typical, without any pathological changes in morphology (Fig. 2A). In our present study, mice treated with ANIT (100 mg/kg) presented focal hepatic necrosis and acute infiltration of polymorphonuclear neutrophils in the portal area, as well as enlarged hepatocytes and loose cytoplasm with balloon-like or feather-like changes, which were consistent with previous reports. No apparent necrosis of hepatocytes and infiltration of immune-active cells were observed in the ANIT + DNT (3 g/kg) treated group compared with the VEH group, suggesting that DNT can ameliorate the ANIT-induced damage to liver tissues. Moreover, UDCA (90 mg/kg), an effective clinical drug for treating cholestasis, also decreased the extent of damage to hepatocytes as well as neutrophil infiltration induced by ANIT but with occasional hepatic necrosis.

Close

Fig. 2

DNT attenuates ANIT-induced liver injury in mice. (A) Representative H&E staining (magnification × 400) images of liver sections of mice. (B) Serum levels of ALT, AST, ALP, TBIL, DBIL, and TBA. Compared with the VEH (control) group, *P < 0.05, **P < 0.01, ***P < 0.001; Compared with the ANIT (model) group, #P < 0.05, ##P < 0.01, ###P < 0.001; Compared with the ANIT + UDCA (positive) group, &P < 0.05, &&P < 0.01.

Export to PPT

In addition, a dose of 100 mg/kg of ANIT drastically elevated the serum levels of ALT, AST, ALP, TBIL, DBIL, and TBA in mice (Fig. 2B; P < 0.05). These results suggested that ANIT might cause hepatotoxicity both by direct damage to hepatocytes and by inducing cholestasis, which was consistent with the histological changes observed in our present study. Conversely, the serum levels of these indicators were significantly reduced (P < 0.05) in the DNT (3 g/kg) treated group. UDCA (90 mg/kg) treatment also reduced the levels of ALT, ALP, TBIL, DBIL, and TBA (P < 0.05) in vivo. Notably, ALT, AST, and ALP levels were significantly lower in the DNT group compared to that in the UDCA group, implying that 3 g/kg of DNT could exert a better effect than 90 mg/kg of UDCA in recovering hepatocytes injuries caused by ANIT. Taken together, these results indicate that DNT can protect mice from ANIT-induced liver injury.

3.2 Identification of DNT-derived xenobiotics in mice

DNT consists of 7 herb ingredients according to TCM formulation theory, among which RRR and PCRR are regarded as the Emperor (Jun) drugs. An in-house database of DNT containing 732 chemical compounds was established by a wide-scale literature mining to help identify compounds. Our previous study has reported various components in the formula (Zhan et al., 2016), including terpenoids, flavonoids, aliphatic compounds, anthraquinone derivatives, and stilbenes, etc. In present study, serum and liver samples of mice from both VEH, ANIT and ANIT + DNT groups were analyzed both in positive and negative ion modes (Supplementary Fig. S1), respectively, to screen the DNT-derived xenobiotic chemicals.

The candidates should meet the following criteria: 1) be present in the samples from ANIT + DNT mice; 2) be present in the DNT formulae sample; and 3) be absent in the samples from VEH and ANIT mice. The total ion chromatography (TIC) was obtained using the PeakView with MasterView plugin. And the DNT-derived xenobiotic chemicals were identified by comparing the reference information (including accurate molecular weight, retention time, and MS² fragmentation, etc.) with the chemical references available and those reported in references. As result, a total of 45 xenobiotic chemicals derived from DNT formulae were identified in mice. A summary of these compounds and the source herbal ingredient were shown in Supplementary Table S2. Here, isomer constituents with m/z 271.0599 Da in ESI⁺ mode were taken as an example to illustrate the identification process (Fig. 3). Three peaks with m/z 271.0599 Da were eluted at 11.3 min, 12.4 min, and 14.5 min, respectively. First, the exact mass of the components, i.e., m/z 271.0599, was automatic searched in the compound database of DNT. As results, a list of possible candidates was generated, including apigenin, aloe emodin, and emodin. Second, the MS² fragments of these components were extracted and compared with reported data and with those generated by their corresponding standard chemical references. Third, t_R (retention time) values of candidates and standard chemical references were matched. Finally, the three compounds of m/z 271.0599 were identified as apigenin (t_R 11.3 min), aloe emodin (t_R 12.4 min), and emodin (t_R 14.5 min), respectively.

Close

Fig. 3

Identification of DNT-derived xenobiotic compounds in mice. (A) Total ion chromatogram (upper panel) and MS² fragmentaion (lower panel) of apigenin, aloe emodin, and emodin in ESI⁺ mode. (B) Total ion chromatogram (upper panel) and MS² fragmentation (lower panel) of kaempferol in ESI⁺ mode.

Export to PPT

3.3 DNT recovers alteration in the metabolic profiling in ANIT-treated mice

Untargeted metabolomic study was performed by using serum and liver samples from mice in the VEH group, ANIT group, and ANIT + DNT groups. The LC-MS spectrum was processed using XCMS and pre-treated before performing the multivariate data analysis, including the unsupervised principal components analysis (PCA) and the supervised orthogonal partial least squares discrimination analysis (OPLS-DA). The results for serum and liver metabolic profiles are shown in Fig. 4.

Close

Fig. 4

DNT recovers alteration in the metabolic profiling in ANIT-treated mice. The PCA score plots of serum samples (A, B) and liver samples (G, H) both in the positive and negative modes. The OPLS-DA score plots of serum samples (C-F) and liver samples (I-L) both in the positive and negative modes.

Export to PPT

As for serum, the three groups formed a well-clustered inner group and were significantly separated in the PCA score plot from each other as indicated by LC-MS data acquired in the positive (Fig. 4A) and negative (Fig. 4B) modes. An OPLS-DA was further performed to mine out the markers that might contribute to the between-group differences. The values for R2Y and Q2 were all above 0.8 (Supplementary Table S3), suggesting satisfactory modeling and predictive abilities of the model. The score plots of OPLS-DA exhibited a distinguishable separation between the VEH and ANIT groups (Fig. 4C and 4D), like between the ANIT and ANIT + DNT groups (Fig. 4E and 4F). Additionally, a volcano plot of endogenous compounds also high-lighted alterations in the metabolic profiling in serum samples of mice from the VEH versus ANIT, and ANIT versus ANIT + DNT (Supplementary Fig. S2) pairs of groups. Volcano plots of differential compounds in the liver of cholestatic mice were presented in Supplementary Fig. S2E-H. While the three groups of liver samples were separated from each other based on their PCA score, acquired in both positive (Fig. 4G) and negative (Fig. 4H) modes. Besides, the score plots of OPLS-DA suggested distinguishable separation between the VEH and ANIT groups (Fig. 4I and 4 J), as well as the ANIT and ANIT + DNT groups (Fig. 4K and 4L). Following were the criteria of fold change > 1.5, P < 0.05 (by Student’s t-test), and a VIP value (in OPLS-DA) of > 1, differentially annotated peaks were screened and identified. Fragments matching of three bile acids compounds, namely taurocholic acid (TCA), glycocholic acid (GCA), and taurochenodeoxycholic acid (TCDCA), with the METLIN databases are shown in Fig. 5. The detailed endogenous metabolite identification information is supplied in Supplementary Table S4.

Close

Fig. 5

Identification of three potential endogenous biomarkers associated with the anti-cholestatic effect of DNT in mice. (A) Taurocholic acid, (B) Glycocholic acid, and (C) Taurochenodeoxycholic acid.

Export to PPT

3.4 Correlations between the DNT-derived xenobiotics and endogenous biomarkers in mice

Correlation analysis is a statistical method to study the correlation between two or more random variables. The correlation coefficient obtained from Pearson’s correlation analysis is the covariance ratio of the standard deviations. It can reflect the intrinsic relationships between the variable groups. Thus, to further discover the pharmacodynamic chemicals responsible for the anti-cholestatic effect of DNT against ANIT-induced liver injuries, Pearson’s correlation analysis was performed between the DNT-derived xenobiotics and potential endogenous markers in rodent samples by using their intensities.

A cutoff value of the correlation coefficient at 0.8 was set to screen the DNT-derived xenobiotics and potential endogenous markers those were highly correlated. As results, TCA, GCA, and TCDCA were found to be highly correlated with the xenobiotics from DNT and identified as potential biomarkers associated with the anticholestatic effect of DNT in serum samples (Supplementary Table S5) as well as in liver samples (Supplementary Table S6). Briefly, in serum samples, TCA was positively correlated with luteolin (r = 0.85) (Fig. 6A) and luteolin-7-glucoside (r = 0.97) (Fig. 6B). While GCA and TCDCA were positively correlated with luteolin-7-glucoside (r = 0.92 and 0.95, respectively) (Fig. 6B). We also identified that in the liver, TCA was positively correlated with kaempferol (r = 0.81), luteolin (r = 0.83), and luteolin-7-glucoside (r = 0.95) but negatively correlated with apigenin (r = -0.81) and emodin (r = -0.81) (Fig. 6C and 6D). The TCDCA was positively correlated with 5,4′-dihydroxy-3,6,7,8,3′-pentamethoxyflavone (DHPF) (r = 0.84) (Fig. 6D). In addition, hepatic mRNA expression of genes involved in bile acid synthesis and bile acid transport were further determined by q-PCR (Supplementary Figure S3). ANIT exposure decreased the levels of Farnesoid X receptor (Fxr), a key nuclear receptor modulating bile acid homeostasis via a feedback mechanism (Song et al., 2022; Liu et al., 2022). As those have been reported (Shi et al., 2022), ANIT exposure also decreased several genes involved in bile acid biosynthesis (including Fxr, Shp and Cyp8b1) and increased several genes involved in bile acid transport (including Ntcp and Bsep). DNT treatment reversed the changes of these genes, supporting that DNT modulates ANIT-induced bile acids impairment and bile acids are important endogenous biomarkers responsible for the anti-cholestasis effect of DNT.

Close

Fig. 6

Correlation analyses of xenobiotic compounds and biomarkers associated with the anti-cholestatic effect of DNT in mice with ANIT-induced cholestasis. Correlation coefficients of them in serum samples in positive (A) and negative (B) ion modes. Correlation coefficients of them in liver samples in positive (C) and negative (D) ion modes.

Export to PPT

Therefore, 6 xenobiotics from DNT formulae, namely, luteolin (source herb: CR), kaempferol (source herb: RRR, PCRR, CR, and CF), apigenin (source herb: PCRR), emodin (source herb: RRR, and PCRR), luteolin-7-glucoside (source herb: PCRR) and 5,4′-dihydroxy-3,6,7,8,3′-pentamethoxyflavone (source herb: CRP), were selected as the potential Q-markers for the anti-cholestatic effect of DNT. Q-markers of TCM herbs and formulas should be stably existed in the herbs and formulas, and have readily available references. Therefore, the contents of 6 active Q-markers (luteolin, kaempferol, apigenin, emodin, luteolin-7-glucoside, and 5, 4′-dihydroxy-3,6,7,8,3′-pentamethoxyflavone) in three different batches of DNT were further determined. As results, the contents of luteolin, kaempferol, apigenin, and emodin in different batches of DNT preparations were stable, but the relative contents of different ingredients were quite different. In three batches of DNT, the average contents of luteolin (source herb: CR), kaempferol (source herb: RRR, PCRR, CR, and CF), apigenin (source herb: PCRR), emodin (source herb: RRR, and PCRR), luteolin-7-glucoside (source herb: PCRR), and 5, 4′-dihydroxy-3,6,7,8,3′-pentamethoxyflavone (source herb: CRP) were calculated as 0.0505 ± 0.0080, 0.0896 ± 0.0209, 1.0184 ± 0.1301, 3.7336 ± 0.3608, 1.3540 ± 1.1354, and 0.0148 ± 0.0137 mg/g, respectively. These results suggested that these six components could serve as the Q-markers for establishing the content determination method of DNT.

3.5 Network pharmacology suggests the possible mechanism and targets for the anti-cholestatic effect of the potential Q-markers from DNT formulae

A total of 142 predicted drug targets of the 6 potential Q-markers (including luteolin, kaempferol, apigenin, emodin, luteolin-7-glucoside, and 5,4′-dihydroxy-3,6,7,8,3′-pentamethoxyflavone) were obtained. On the other hand, a total of 732 predicted disease targets associated with cholestasis-related diseases (including “NASH”, “Cholestasis”, and “Cholecystitis”) were acquired. The drug targets of the 6 potential Q-markers were matched with the disease targets, and a total of 28 targets were identified (Fig. 7A) and subjected to KEGG and GO pathway enrichment analyses. As results, a total of 60 KEGG pathways were obtained (Supplementary Fig. S4A), related to various cancer pathways, including the PI3K-Akt, AGE-RAGE, and NF-κB signaling pathways. Natural chemicals from herbs have been shown to protect against liver damage by modulating these signaling pathways (Wang et al., 2021, Wu et al., 2017). Furthermore, a total of 536 GO enrichment analysis items, including 478 biological process (BP), 15 cellular component (CC), and 43 molecular function (MF), were found. The top 10 items of each enrichment process are shown in Supplementary Fig. S4B. BP included the regulation of cellular responses to chemical stress, oxidative stress, trans-membrane transport, and ion transport. CC mainly involved the chromosomal telomeric regions, nuclear envelope, and endocytic vesicle. MF was focused on the regulation of nitric-oxide synthase activity, the binding of transcriptional coactivators, the response of NAD + ADP-ribosyltransferase activity, and ABC transporter activity. In addition, protein–protein interaction (PPI) analysis was further performed to screen the essential targets of the 6 Q-markers responsible for their cholestasis-related effects. According to the connectivity scores, four core regulatory genes, including TNF, AKT1, EGFR and ESR1, were highlighted as the key targets involved in the anti-cholestatic effect of the 6 potential Q-markers in DNT (Fig. 7B). The q-PCR analysis of the liver tissue suggested that hepatic mRNA expression of these genes was upregulated in mice after ANIT exposure and their expression was attenuated after DNT treatment (Fig. 7C).

Close

Fig. 7

Confirmation of Q-markers of DNT by network pharmacology. (A) The number of potential targets for DNT Q-markers in the treatment of cholestasis-related disease. (B)The interaction network of the compound-target-disease. The light blue diamonds represent compounds, the green rotundities represent intersection targets, and the purple blue hexagons represent diseases. The grey edges indicate the link be-tween the compounds and the target. (C) Changes of hepatic mRNA expression of Tnfa, Akt1, Egfr and Esr1 in mice after DNT treatment. Compared with the VEH (control) group, *P < 0.05, **P < 0.01, ***P < 0.001; Compared with the ANIT (model) group, #P < 0.05, ##P < 0.01, ###P < 0.001.

Export to PPT