The chemicals and reagents are described in Supplementary Material M1. Crude drugs Artemisia capillaris Thunb. (210218), Atractylodes macrocephala Koidz. (210430), Aconitum carmichaelii Debx. (210309), Zingiber officinale Rosc. (210403), Glycyrrhiza uralensis Fisch. (2106034), and Cinnamomum cassia Presl. (210309) were purchased from Shanghai Kangqiao Chinese Medicine Tablet Co., Ltd. and authenticated using morphological and microscopic identification in accordance with the Chinese Pharmacopoeia by Dr. Jinrong Wu from Shanghai University of Traditional Chinese Medicine (Commission, 2020).

YCZFD extract was prepared according to our previously reported method (Wang et al., 2020). Briefly, the crude materials of Artemisia capillaris Thunb. (300 g), Atractylodes macrocephala Koidz. (600 g), Aconitum carmichaelii Debx. (150 g), Zingiber officinale Rosc. (150 g), Glycyrrhiza uralensis Fisch. (300 g), and Cinnamomum cassia Presl. (100 g) were immersed in 16 L of water for 30 min and boiled for 1 h to collect the first decoction. After filtration, 12.8 L of water was added and the mixture was boiled for 1 h to collect the second decoction. The two decoctions were mixed and concentrated with a rotary evaporator. The mixture was freeze-dried to obtain powdered YCZFD extract with 19% yield.

Qualitative analysis using an ultra-high performance liquid chromatography coupled to a linear trap quadrupole-Orbitrap (UHPLC-LTQ-Orbitrap) elite MS system (Thermo Fisher Scientific, Bremen, Germany) (Wang et al., 2020) revealed 58 plant metabolites in YCZFD extract (Supplementary Table S1). The quality control of YCZFD was performed as the methods described previously (Li et al., 2023). A total of 18 chemical plant metabolites (chlorogenic acid, scoparone, 4-hydroxyacetophenone, atractylenolide I, atractylenolide Ⅱ, atractylenolide Ⅲ, benzoylaconine, benzoylhypaconine, benzoylmesaconitine, formononetin, glycyrrhitinic acid, glycyrrhizic acid, isoliquiritigenin, isoliquiritin, liquiritigenin, liquiritin, ononin, and cinnamic acid) were quantified and shown in Supplementary Table S2. Among these plant metabolites, glycyrrhizic acid (1275.88 μg/g), liquiritin (901.26 μg/g), chlorogenic acid (782.2 μg/g), isoliquiritin (91.1 μg/g), and atractylenolide I (39.11 μg/g) had the highest contents.

Male C57BL/6J mice weighing 20 ± 2 g were obtained from Shanghai SLAC Laboratory Animal Co., Ltd. The mice were housed and acclimatized for 1 week at the Animal Experimental Center of Shanghai University of Traditional Chinese Medicine (SHUTCM). The temperature and humidity in the breeding room were maintained at 20°C–26°C and 40%–60%, respectively. Pharmaceutical experiments were performed after 1 week. The mice were divided into four groups (n = 8): control (CON), DDC model (DDC), DDC+YCZFD-L (1.14 g/kg YCZFD extract, equivalent to 6 g crude drug/kg), and DDC+YCZFD-H (2.28 g/kg YCZFD extract, equivalent to 12 g crude drug/kg). The mice in the CON group were fed a normal diet, while those in the three other groups were fed a diet containing 0.025% DDC. After 1 week of diet feeding, the mice in the CON and DDC groups were orally administered normal saline, while those in the DDC+YCZFD-L and DDC+YCZFD-H groups were orally treated with YCZFD extract once a day for the following 4 weeks. During the drug treatment period, the mice in the DDC, DDC+YCZFD-L, and DDC+YCZFD-H groups were still fed a diet containing 0.025% DDC. The experimental design is shown in Figure 1A. All mice were sacrificed at the experimental endpoint, and the serum and liver tissues were collected for further studies. Animal experiments were approved by the Experimental Animal Welfare and Ethics Committee of SHUTCM (Approval number Pzshutcm190823002).

Details regarding serum biochemistry and liver histopathological morphology analysis can be found in Supplementary Material M2.

Mouse livers fixed by formalin were paraffin embedded, sectioned, and then stained with an anti-alpha-smooth muscle actin (anti-α-SMA) or collagen type 1 alpha 1 chain (COL1A1) antibody sourced from Abcam (Cambridge, MA, United States). To facilitate histological examination, the sections were ultimately affixed using the DPX Mountant (Sigma).

The human HSC line (LX-2) was cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS and incubated at 37°C in an atmosphere of 5% CO₂. Cells were treated with 5 ng/mL transforming growth factor-β (TGF-β) or TGF-β (5 ng/mL) and increasing concentrations (5, 10, 25, and 50 μg/mL) of YCZFD for 48 h.

Details regarding real-time PCR are provided in Supplementary Material M4 and Supplementary Table S2.

Details regarding Western blot analysis are presented in Supplementary Material M5. The primary antibodies were as follows: anti-FN1 (151613-1-AP, Proteintech), anti-COL1A1 (NBP1-30054, Novus), anti-α-SMA (ab124964, Abcam), anti-TIMP2 (5738, CST), anti-TGF-β (ab92486, Abcam), anti-GAPDH (G8795, Sigma-Aldrich), anti-PDGFRβ (13449-1-AP, Proteintech), anti-p-PDGFRβ (AP0815, ABclonal), anti-PDGF-B (ab178409, Abcam), anti-AKT1(2967S, CST), anti-p-AKT1 (15116, CST), anti-PI3K (19H8, CST), anti-p-PI3K (AT-3241, Affinity), and anti-proliferating cell nuclear antigen (anti-PCNA; 61079, Active motif) antibodies.

LX-2 were cultured in DMEM with TGF-β (5 ng/mL), or with TGF-β (5 ng/mL) and YCZFD extract (5 and 25 μg/mL) for 48 h. Subsequently, the cells were harvested in cold PBS at 4°C and then fixed using 70% ethanol at the same temperature overnight. Following fixation, the cells were washed with cold PBS to remove excess fixative and then stained with propidium iodide containing RNase A. FlowJo (Tree Star) was employed to evaluate the DNA content of the stained cells.

After the injection of 25% pentobarbital, incision of abdominal wall, and separation of the portal vein, a cannula connected to an infusion tube was inserted into the portal vein, the inferior vena cava was cut, and a peristaltic pump was turned on. Perfusion solution was perfused for 40 min. The liver was placed in culture medium and then agitated to release hepatocytes. The extracted cells were passed through a 200-mesh cell strainer, centrifuged, and then resuspended in DMEM. The cells were transferred to the upper layer of the medium containing Percoll and then centrifuged. The viable cells were resuspended in DMEM, centrifuged, resuspended in Williams’ MediumE (WME) containing 10% FBS, 10 nM insulin, and 10 nM dexamethasone, and counted.

On the day before the sandwich culture and treatment of primary cells, the lower gelatin layer [100 mL of ultrapure water + 114 μL of ice-cold acetic acid (0.02 M) + 1.6 mL of type I mouse tail collagen] was prepared, and the culture dishes were coated. On the day of primary cell extraction, the plates were seeded. On the second day, the upper gelatin layer (50 mL system: 2.5 mL of serum, 0.5 mL of glutamine, 0.5 mL of dexamethasone, 0.5 mL of ITS, and 46.45 mL of WME) was prepared. On the third day, drug treatment was administered. The CCK-8 assay was used to screen the safe dosage range of plant metabolites of YCZFD (Supplementary Figure S2). Glycyrrhetinic acid at 10 μM and the other plant metabolites at 30 μM did not affect the viability of primary cells and thus were used in subsequent experiments. Taurocholate acid (TCA) and TCA combined with individual plant metabolites were administered simultaneously. After 24 h of incubation, the mRNA expression of PDGFRβ was detected.

The data were statistically analyzed using GraphPad Prism version 8.0 (GraphPad Software, La Jolla, CA, United States). ANOVA followed by Tukey post hoc tests was performed. Data are presented as mean ± SD. p < 0.05 was considered statistically significant.

After treatment with YCZFD, the DDC-induced increases in ALT, AST, TBA, DBIL, and TBIL were significantly attenuated (Figure 1B). Histopathological assessments revealed severe hepatic necrosis, inflammatory cell infiltration, and bile duct proliferation in the liver tissues of the mice subjected to DDC intervention. Treatment with YCZFD decreased the liver damage (Figure 1C). Overall, these findings demonstrated that treatment with YCZFD ameliorated CLI in the DDC-induced mice.

Sirius red staining showed greater parenchymal matrix deposition within the liver tissues in the DDC group than that in the CON group (Figures 2A, B), and this increase in collagen deposition was attenuated in the livers of the YCZFD treatment group. Immunohistochemical staining revealed that the increased expression of liver fibrosis markers (α-SMA and COL1A1) in the fibrotic septa of the DDC group was decreased in the YCZFD treatment group (Figures 2C, D). Consistent with the immunohistochemical staining results, the protein expression of COL1A1, FN1, and α-SMA and the mRNA expression of COL1A1 were increased in the DDC group and attenuated in the YCZFD group (Figures 2E–G). Additionally, the gene expression levels of SMAD2 and SMAD3 and the protein levels of TGF-β in the cholestatic mice were significantly decreased by YCZFD treatment. Overall, these findings demonstrated that treatment with YCZFD ameliorated CLF.

DIA technology was used for the quantitative proteomic analysis of liver tissues from the CON, DDC, and DDC+YCZFD (DDC+YCZFD-H) groups to elucidate the mechanism by which YCZFD ameliorates cholestatic liver fibrosis. The experimental workflow is shown in Figure 3A. A total of 4,746 proteins were identified, and 4,669 proteins had quantitative information in 60% of the samples (Figure 3B). The proteomic data were subjected to median normalization for each sample to eliminate variation in sample injection amounts (Figure 3C). The distribution of protein abundance in each sample was similar, indicating good parallelism among samples. Furthermore, the median coefficients of variation among the eight replicate samples from the CON, DDC, and YCZFD groups were all less than 0.1 (Figure 3D), indicating good quantitative reproducibility of the data. Overall, the proteomic data were of high quality. ANOVA was used to identify DEPs among the three groups. Based on BH-corrected p-values (p < 0.05) and fold change (>1.2 or <0.83), 420 DEPs were identified (Supplementary Table S3).

The DEPs were clustered into four subclusters (Figure 4A, left panel) by using the mfuzz clustering algorithm, and a pheatmap was used to visualize the expression levels (Figure 4A, right panel). Proteins in Cluster 2 and Cluster 4 were differentially expressed between the DDC and CON groups, with the attenuation of expression changes after YCZFD intervention. Cluster 2 consisted of 127 proteins, with downregulated expression in the DDC group and restored expression after YCZFD treatment (Figure 4A). Cluster 4 consisted of 164 proteins, with upregulated expression in the DDC group and restored expression after YCZFD treatment.

Proteins in the four clusters were evaluated through GO biological process and KEGG pathway enrichment analyses (Figures 4B, C). Proteins in Cluster 2 were mainly enriched in glutathione metabolism, proximal tubule bicarbonate reabsorption pathways, and lipid metabolic process. Proteins in Cluster 4 were mainly enriched in the sterol biosynthesis pathway, spliceosome signaling pathways, ECM–receptor interaction, PI3K-AKT pathway, and cell adhesion process.

The PPI network and pathway enrichment results were integrated (Figure 5). Pathways related to sterol metabolism were significantly enriched, including sterol synthesis-related proteins (e.g., Sqle, Hmgcs1, and Nsdh1), which belong to Cluster 4 (Supplementary Figure S1). Importantly, PI3K-AKT and focal adhesion-related proteins, such as PDGFRβ, PIKR3R1, ITGB5, COL1A1, COL6A1, COL6A3, and LAMA5, are mainly associated with liver fibrosis. PDGFRβ was identified as a hub protein, suggesting that YCZFD exerts a protective effect against cholestasis by alleviating fibrosis through the regulation of PDGFRβ.

PDGFRβ is a key protein involved in the PI3K/AKT signaling and focal adhesion pathways (Figure 5). As revealed from the proteomic analysis, the protein levels of PDGFRβ were significantly upregulated in the DDC group (fold change > 4) and downregulated after YCZFD administration (fold change > 2) (Figure 6A). Western blot and real-time qPCR results further confirmed these changes (Figures 6B–D). Treatment with YCZFD also downregulated PDGFRβ protein expression in the liver tissues. The expression of the ligand of PDGFRβ, PDGF-B, was also downregulated after YCZFD intervention (Figures 6B–D). The expression of PIK3R1, a regulatory subunit of phosphoinositide 3-kinases (PI3Ks), was significantly upregulated in the DDC group and restored in the YCZFD treatment group (Figure 6A). The activation of PI3K and AKT1 in the DDC group and attenuation in the YCZFD group were validated through Western blot (Figures 6C, D).

We validated the impact of YCZFD on the PDGFRβ-PI3K-AKT pathway in HSCs (Figure 7A). The expression of α-SMA, which reflects the activated state of HSCs, was increased after TGF-β intervention (Figures 7B, C). The phosphorylation levels of PDGFRβ (Tyr1021), PI3K (Tyr607), and AKT (Ser473) and the protein levels of PDGFRβ and PDGF-B increased in response to TGF-β (Figures 7B, C). Treatment with YCZFD attenuated the changes in levels of these factors. The PCNA expression and cell cycle progression were further investigated. The findings indicated that PCNA expression was downregulated following YCZFD intervention. The composition and ratio of cells in various phases varied with the concentration of YCZFD (0, 5, and 25 μg/mL). Compared to the distribution in TGF-β-treated cells, the percentage of cells in the G2/M phase increased from 25.8% to 36.9% (5 μg/mL YCZFD) and 46.1% (25 μg/mL YCZFD) (Figures 7D, E). These results demonstrated that YCZFD could effectively induce cell cycle arrest in the G2/M phase.

To find the effective plant metabolites of YCZFD in primary cells, we utilized a sandwich culture of primary cells incubated them with TCA (known to accumulate in cholestasis) (Figure 7F). PDGFRβ expression was upregulated in the TCA model group compared with the control group (Figure 7G). However, treatment with YCZFD plant metabolites such as chlorogenic acid, scoparone, isoliquiritigenin, glycyrrhetinic acid, formononetin, atractylenolide I, and benzoylaconitine downregulated the mRNA expression of PDGFRβ (Figure 7G). These results revealed that multiple plant metabolites of YCZFD may target PDGFRβ.