Da-Chai-Hu-Tang (Dai-saiko-to in Japanese) (Lot: M25161) was purchased from Tsumura & Co. (Tokyo, Japan). The dried decoction of DCHT formula with eight traditional Chinese herbs contains 6 g Root of Bupleurum falcatum L [Umbelliferae; Bupleuri Radix], 4 g Tuber of Pinellia ternata (Thunb.) Makino [Araceae; Pinelliae Rhizoma], 3 g Root of Scutellaria baicalensis Georgi [Labiatae; Scutellariae Radix], 3 g Root of Paeonia lactiflora Pall [Paeoniaceae; Paeoniae Radix Alba], 3 g Fructus of Ziziphus jujuba Mill [Ramnaceae; Jujubae Fructus], 2 g Fructus Immaturus of Citrus × aurantium L [Rutaceae; Aurantii Fructus Immaturus], 1 g Rhizoma of Zingiber officinale Roscoe [Zingiberaceae; Zingiberis Rhizoma Recens], and 1 g Rhizoma of Rheum palmatum L [Polygonaceae; Rhei Radix et Rhizoma], which were added to 700 ml of water, boiled for 1 h, filtered and then concentrated to 300 ml. This decoction was spray-dried to yield 4.5 g of a powdered extract, which represents a 1-day dosage as previously described (He et al., 2014).

Narirutin, baicalein, rhein, wogonin, chrysophanol, naringenin, kaempferol, saikosaponin A, emodin-3-methyl ether, paeoniflorin and emodin were purchased from Shanghai R&D Centre for Standardization of Chinese Medicines (Shanghai, China). These structures were determined using ¹H-NMR and ¹³C-NMR spectral analysis, and its purity was more than 98% as determined by high pressure liquid chromatography analysis. All compounds were dissolved in absolute dimethyl sulfoxide (DMSO) as 100 mM, and, on the experimental day, was further diluted with culture medium.

Please see Supplementary Table S1 for a complete list of reagents and antibodies used in this study.

DCHT (0.5 g) was extracted with 2 ml of methanol under ultrasonication for 30 min, and then centrifuged at 10,000 × g, at 4°C for 15 min. High-resolution liquid-chromatography-mass spectrometry (LC-MS) was performed on a Fisher Orbi-Trap Elite instrument (Thermo, Waltham, MA, United States) to detect and analyze the main constituents in DCHT.

Pathogen-free C57BL/6 male mice (8 weeks old, 18–20 g) were purchased from the Experimental Animal Center of Chinese Academy of Science (Shanghai, China). Experimental procedures were approved by the Shanghai University of Traditional Chinese Medicine Committee on the Use of Live Animals for Teaching and Research (Animal License: No. SYXK(HU)2014-0008; Registration number: PZSHUTCM190531014). To study intrahepatic cholestasis in response to α-naphthylisothiocyanate (ANIT), mice from administration groups were treated with DCHT [0.21, 0.625, and 1.875 g/kg dissolved in water, intragastric administration (i.g.)], or OCA (20 mg/kg dissolved in 0.5% CMC-Na, i. g., a mainstay treatment for cholestasis, as a positive control (Ding et al., 2018)), or PPARα antagonist GW6471 [5 mg/kg dissolved in 4% Tween 80 in saline (v/v), intraperitoneal administration (i.p.)] (Hu et al., 2019) or PPARα agonist fenofibrate (25 mg/kg dissolved in 0.5% CMC-Na, i. g. bid) (Dai et al., 2017) for 3 consecutive days. After treatment for 3 days, mice from each of the treatment groups ANIT, ANIT + DCHT, ANIT + OCA, ANIT + DCHT + GW6471, ANIT + GW6471, and ANIT + fenofibrate were given a single dose of ANIT (80 mg/kg dissolved in olive oil, i. g.), and mice from the control group were given the same volume of olive oil (i.g.) for two consecutive days (Figure 1A and Figure 6A). For extrahepatic cholestasis induced by bile duct ligation (BDL) as previous described (Tag et al., 2015), ligated mice were divided randomly into four groups (n = 6). The sham-operated control group (water, i. g.), BDL group (water, i. g.), and DCHT groups (0.21, 0.625, and 1.875 g/kg dissolved in water, i. g.) were given for 14 consecutive days (Supplementary Figure S2A). Animal body weight was recorded every day. After the treatment, blood samples were obtained by cardiac puncture, coagulated for 2 h at 4°C, and centrifuged at 3,000 × g, 4°C for 20 min. The activities of alanine aminotransferase (ALT), aspartate aminotransferase (AST), direct bilirubin (DBiL), total bilirubin (TBiL) and total bile acid (TBA) in the serum (supernatant) were measured using an automatic biochemistry analyzer Accute (TBA-40FR, TOSHIBA, Japan). Livers with gallbladders were excised from every mouse, and then either fixed in 4% paraformaldehyde (PFA) for analysis by Hematoxylin-Eosin (HE) staining or immediately frozen in liquid nitrogen and kept at −80°C until use for further analysis by Realtime PCR or SDS-PAGE/Western blotting.

Hepatic reduced glutathione (GSH) and oxidized glutathione (GSSG) were measured using the GSH and GSSG assay kit (Cat. #: S0053, Beyotime Inst. Biotech, Shanghai, China). The levels of GSH were calculated according to the difference between concentrations of total glutathione (GSH + GSSG) and GSSG. Hepatic levels of GSH were normalized to protein concentrations in the corresponding samples.

Hepatic activities of glutathione reductase (GR), glutamate-cysteine ligase (GCL), glutathione peroxidase (GSH-Px, GPX) and glutathione S-transferase (GST) were determined using the GR assay kit (Cat. #: A062), the GCL assay kit (Cat. #: A120-1-1), the GPX assay kit (Cat. #: A005), and the GST assay kit (Cat. #: A004), respectively. These assay kits were purchased from Nanjing Jiancheng Bioengineering Institute (Nanjing, China).

Total RNA from the murine liver was extracted using the Trizol reagent (TaKaRa, Dalian, China). RNA concentrations were determined by nanodrop and then normalized. cDNA was synthesized from 1 µg of total RNA using the PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa). qPCR reactions were performed using TB Green kit (TaKaRa) on an ABI 7500 real-time PCR system (Thermo Fisher, Quantstudio 3). Please see Supplementary Table S2 for a complete list of primers used in this study.

Liver tissues were lysed in RIPA buffer containing 1 mM phenylmethylsulfonyl fluoride and protease inhibitor cocktail. Lysates were spun at 10,000 × g for 10 min and the supernatants kept at −20°C until use. Proteins were separated by SDS-PAGE and transferred to nitrocellulose filter membranes; blots were probed with appropriate combinations of primary and HRP-conjugated secondary antibodies. For repeated immunoblotting, membranes were stripped in 62.5 mM Tris (pH 6.7), 20% SDS and 0.1 M 2-mercaptoethanol for 30 min at 50°C prior to reprobing.

Putative targets of main constituents from DCHT based on the LC-MS based chemoprofile were obtained from Traditional Chinese Medicine Systems Pharmacology database (TCMSP) (http://tcmspw.com/tcmsp.php) (Ru et al., 2014) and SwissTargetPrediction (http://www.swisstargetprediction.ch/) (Gfeller et al., 2014). Known differentially expressed targets related to intrahepatic cholestasis and extrahepatic cholestasis were acquired from the GeneCards database (https://www.genecards.org/) (Safran et al., 2010) and DisGeNET database (https://www.disgenet.org/search) (Piñero et al., 2017). The visual network of the “Herb-Compound-Target-Disease” was established using Cytoscape software (version 3.7.2, Boston, MA, United States). The signaling pathways with enriched genes (FDR<0.05) were identified using database for Annotation, Visualization and Integrated Discovery (DAVID) assigning the Kyoto Encyclopedia of Genes and Genomes (KEGG) database.

HEK293T cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Invitrogen) with 10% (v/v) fetal bovine serum (FBS, Invitrogen) without antibiotics in 96-wells plates (2 × 10⁴ cells/well) overnight, before co-transfection with a plasmid encoding PPARα (pCMX-Gal-mPPARα, 0.1 μg), report plasmid (pGL4-MH100 × 4-TK-Luc, 0.1 μg), and Renilla luciferase plasmid (pREP7, 0.01 μg) using Lipofectamine 3,000 (Invitrogen, Carlsbad, CA) as previous described (Huang et al., 2006). After treatment with DCHT or compounds, cells were collected and lysed. Firefly luciferase activity or Renilla luciferase activity (as normalization control) were determined using the Firefly & Renilla luciferase Reporter Assay Kit (Dalian Meilun Co., Ltd., China) and the BioTek Synergy™ 4 microplate reader (BioTek). Data were presented as firefly/renilla luciferase ratios. This allows the data to be normalized to control for factors such as transfection efficiency or overall translational activity.

Data are presented as means ± SEM. Statistical analysis was performed by LSD-test following one- or two-way ANOVA for multiple comparisons. p values less than 0.05 were considered as statistically significant differences.

LC-MS based chemoprofile of the methanol solution of DCHT, including positive ion mode and negative ion mode were shown in Supplementary Figure S1. The presence of the following constituents in DCHT of detection mode, formula, molecular weight and Area (Max.) (Supplementary Table S3): (+)-Catechin, Lactiflorin, Albiflorin, Paeoniflorin, Oxypaeoniflorin and Benzoylpaeoniflorin (from Bai Shao); Succinic acid (from Ban Xia); Kaempferol, Sainfuran, Quercetin, Isorhamnetin, Saikosaponin A and Saikosaponin B1 (from Chai Hu); Chrysophanol, Aloe-emodin, Emodin, Rhein, Emodin-3-methyl ether, Chrysophanols 1-O-glucoside, Chrysophanols 8-O-glucoside, Emodin 8-O-glucoside, Emodin-1-O-beta-d-glucopyranoside and Rhein 8-O-glucoside (from Da Huang); (S)-Coclaurine (from Da Zao); Baicalein, Wogonin, Oroxylin A, Dihydrooroxylin A, Panicolin, Rivularin, Skullcapflavone II, Wogonin-7-O-glucuronoside and Oroxylin A-7-O-glucuronoside (from Huang Qin); 6-Gingerol (from Sheng Jiang); Synephrine, Naringenin, Luteolin, Hesperetin, Tetramethoxyluteolin, Sinensetin, Isosinensetin, Nobiletin, Naringin, Narirutin, Hesperidin and Neohesperidin (from Zhi Shi), and Baicalin is shared by Chai Hu, Huang Qin and Ban Xia, and some major constituents from DCHT were the same as previously described (Yoshie et al., 2004; He et al., 2014).

ANIT is metabolized by cytochrome P450 and is subjected to GSH conjugation, and thus induces intrahepatic cholestasis, biliary epithelial cell necrosis, bile duct obstruction and hepatocellular injury, such characteristics mimic the drug-induced cholestasis and hepatic injury in humans (Rodríguez-Garay, 2003). We first established an acute cholestatic murine model by treating mice with ANIT for 48 h, and used OCA (20 mg/kg, a mainstay treatment for cholestasis) as a positive control (Figure 1A) (Ding et al., 2018). At the end of this treatment period, we observed an increase in the levels of indicators of cholestasis, namely DBiL and TBiL, in response to ANIT (Figures 1B,C). In addition, TBA, a typical indicator of intrahepatic cholestasis, was elevated after ANIT treatment (Figure 1D). The induction of serum DBiL, TBiL and TBA by ANIT was effectively reversed by DCHT treatment in a dose-dependent manner (Figures 1B–D). After the treatment of ANIT, serum levels of ALT and AST, which serve as direct indicators of liver injury, were elevated (Figures 1E,F), accompanied by large periportal hemorrhage, diffuse vacuolization, inflammatory infiltration, and parenchymal necrosis in the liver (Figure 1G). DCHT treatment was able to alleviate ANIT-induced liver injury in a dose-dependent manner, which is similar to the effect of OCA (20 mg/kg) (Figures 1E–G). ANIT also caused a dramatic loss of body weight, which was attenuated upon treatment with DCHT (1.875 g/kg) or OCA (Figure 1H). However, DCHT treatment never alleviated BDL-induced extrahepatic cholestasis with liver injury (Supplementary Figure S2). These data indicate that DCHT can protect mice from ANIT-induced acute intrahepatic cholestasis with liver injury.

To investigate whether DCHT modulates bile acid homeostasis during acute intrahepatic cholestasis, we treated the mice with 1.875 g/kg DHCT and then examined the expression of genes implicated in hepatic bile acid biosynthesis, uptake and efflux. As shown in Figures 2A,B, the expression of the mRNA encoding the cytochromes cholesterol 7α-hydroxylase (Cyp7a1) and sterol 12α-hydroxylase (Cyp8b1) was down-regulated in response to ANIT treatment and restored by DCHT. Bile acids are first taken up by the hepatobiliary transporters NTCP and OATP2 from the plasma and are then subsequently exported into the bile by the canalicular transporters BSEP and MRP2. ANIT treatment led to a decrease in the mRNA levels for Ntcp and Oatp2, and this reduction was prevented by DCHT (Figures 2C,D). ANIT treatment also increased the mRNA levels for Bsep and decreased the level of Mrp2, which encode for bile acid efflux transporters BSEP and MRP2, respectively; both of these effects were reversed by DCHT (Figures 2E,F). Multidrug resistance-associated protein 3 (MRP3) and multidrug resistance-associated protein 4 (MRP4) are transporters of the hepatocyte’s basolateral membrane with a compensatory role. Both MRP3 and MRP4 transporters’ increased mRNA expression plays an essential role in the protective and adaptive responses of bile acid overload, and the induction of mRNA level for Mrp3 and Mrp4 was effectively reversed by DCHT treatment (Figures 2G,H). These results suggest that DCHT protects against acute intrahepatic cholestasis by restoring bile acid homeostasis.

Accumulated toxic bile acids led to oxidative stress and then caused liver injury, which can be alleviated by glutathione via eliminating reactive oxygen species (ROS) (Orozco-Aguilar et al., 2021). Indeed, GSH (reduced glutathione) levels were decreased upon ANIT treatment whereas GSSG (oxidized glutathione) was elevated, leading to a marked decrease in the ratio of GSH/GSSG, an effect which was also prevented by DCHT (Figures 3A,B). DCHT treatment led to a remarkable restoration of GR activity, whereas had little effect on GCL activity in the presence of ANIT (Figures 3C,D). The inhibition of GPX activity by ANIT was abolished by DCHT in vivo (Figure 3E). Moreover, while the activity of the detoxifying enzyme GST was increased in response to ANIT, it was reduced by DCHT in combination with ANIT treatment (Figure 3F). Interleukin-6 (IL-6), Interleukin-1ß (IL-1ß) and tumor necrosis factor alpha (TNF-α) participate in the progression in cholestatic liver injury (Bode et al., 2012). We detected an increase in Il6, Il1b and Tnfa mRNA expression, in liver tissue derived from the ANIT-treated mice compared to the control ones, and this effect was reversed in the ANIT + DCHT group of mice (Figures 3G–I), thus indicating that DCHT treatment was able to suppress the mRNA expression of hepatic inflammatory markers induced by ANIT. Taken together, these results indicate that DCHT is capable of alleviating acute cholestasis-induced liver damage by inhibiting hepatic inflammation and restoring glutathione homeostasis.

To determine the molecular mechanisms by which DCHT exerts its protective effect on acute intrahepatic cholestasis induced-liver injury, we exploited the GeneCards and DisGeNET databases which identified 42 putative targets involved in DCHT treatment for intrahepatic cholestasis (Figure 4A and Supplementary Table S4) and applied KEGG pathway enrichment analysis to the DCHT target data (Figure 4B). Notably, the expression of genes implicated in the PI3K-Akt signaling pathway was the most affected by ANIT treatment (Figure 4B). As a validation for our pathway enrichment analysis, we did indeed observe a slight increase in the level of AKT phosphorylation at Thr308 but with no change at Ser473 (Figures 4C–E); a modest decrease in the level of ERK1/2 phosphorylation (Figures 4F,G); an increase in the phosphorylation of p38 MAP kinase (Figures 4F,H) and JNK/SAPK (Figures 4F,I) in the livers of mice treated with ANIT. These effects were reversed by DCHT co-treatment. We also observed a decrease in Bcl-2 expression and an increase in the expression of Bax in ANIT-treated mice, leading to an increase in the proapoptotic/antiapoptotic Bax/Bcl-2 ratio. The proapoptotic effect of ANIT in the liver tissues was attenuated upon DCHT treatment (Figures 4J,K), consistent with the protective effect of DCHT on liver injury (Figure 1G). Western blot analysis demonstrated that STAT3 was strongly phosphorylated in the livers of mice treated with ANIT, and this was prevented by DCHT co-treatment (Figures 4L,M). Similarly, DCHT treatment effectively antagonized the phosphorylation of NF-ĸB in response to ANIT (Figures 4L–N). Additionally, DCHT treatment prevented the ANIT-induced decrease in expression of the SOCS3 protein, a feedback inhibitor of the STAT3 signaling pathway which also reduces NF-ĸB activity (Gu et al., 2011) (Figures 4L–O). This implies that the suppression of MAPK/NF-ĸB/STAT3 signaling axes may contribute to the preventative effect of DCHT on ANIT-induced liver injury.

Among these genes affected by DCHT as discovered by network pharmacology, 22 were predicted to be associated with intrahepatic but not extrahepatic cholestasis (Figure 5A and Supplementary Table S4). Among those 22 genes, 7 genes (PPARα, CXCL2, PXR, CAR, AHR, ESR1 and ESR2) were identified from both the GeneCards and the DisGeNET databases (Figure 5A and Supplementary Table S5). We validated the changes in 5 out of 7 core putative targets in the treatment of DCHT for ANIT-induced intrahepatic cholestasis, with the exception of estrogen receptor (ESR) 1 and 2 (ESR1 and ESR2) because they are more associated with intrahepatic cholestasis during pregnancy (Song et al., 2014). As shown in Figure 5B, the expression of the mRNA encoding the cytochromes C-X-C motif chemokine ligand 2 (Cxcl2, encoded by CXCL2) was upregulated in response to ANIT treatment and suppressed by DCHT. The expression of the mRNA for Pregnane X receptor (Pxr, encoded by NR1I2) was upregulated in response to ANIT treatment and further enhanced by DCHT (Figure 5C). DCHT treatment had no effect on the large increases in the expression of the mRNAs for both the aryl hydrocarbon (Ahr, encoded by AHR) and constitutive androstane receptors (Car, encoded by NR1I3) in response to ANIT (Figures 5D,E). Of note, ANIT treatment led to a decrease in the mRNA expression of Ppara (encoded by NR1C1) compared to the vehicle-treated group; and this effect was ameliorated in response to DCHT (Figure 5F). We also noted that there was a dose-dependent increase in the activity of PPARα after DCHT treatment for 24 h as determined by a dual-luciferase reporter gene assay (Figure 5G). In addition, as evidenced by the dual-luciferase reporter gene assay, any of these 11 compounds was able to increase PPARα activity individually in a dose-dependent manner (Figure 5H), which is similar to the effect of fenofibrate, a classic agonist of PPARα (Ghonem et al., 2015) (Supplementary Figure S3).

To gain insight into the role of PPARα in DCHT treatment for cholestatic liver injury, we assessed the effect of DCHT, in combination with PPARα antagonist GW6471 or PPARα agonist fenofibrate, in response to ANIT. Fenofibrate protected against ANIT-induced cholestasis with liver injury, as demonstrated by the disappearance of tissue necrosis. Serum levels of DBiL, TBiL, TBA, ALT and AST were reduced in fenofibrate-treated mice (Dai et al., 2017), a similar effect was also observed in the DCHT treatment group (Figures 6A–I). In contrast, GW6471 treatment accelerated cholestasis and had no protective effect on liver injury in response to ANIT (Figures 6A–I). Of note, pre-treatment with GW6471 attenuated the protective effect of DCHT on cholestatic liver injury evoked by ANIT (Figures 6A–I). The protective effects of fenofibrate and DCHT on ANIT-induced cholestasis and liver injury coincident with the inhibition of mRNA expression of inflammation-related genes (Il6, Il1b, and Tnfa), as well as the suppression of JNK/SAPK, NF-ĸB and STAT3 pathways. Such effects could be reversed upon GW6471 treatment (Figures 6J–M). Taken together, these data provide compelling evidence that PPARα is a potential target of DCHT in the treatment of intrahepatic cholestasis with liver injury.