AS was purchased from MCE (Shanghai, China), and comfrey powder was purchased from Fsinuote (Baoji, Shanxi, China). For the animal experiments, comfrey powder was dissolved in saline and used to generate an AS diet. For cellular experiments, AS was dissolved in DMSO, prepared as a 1 mM storage solution, and stored in aliquots at −80 °C. The PPARG antagonist GW9962 was purchased from MCE (Shanghai, China).

According to relevant TCM guidelines and professional websites, this study selected four recommended formulas for the treatment of non-alcoholic fatty liver disease: Zaozhu Yinchen decoction (ZZYCF), Qinggan Huatan Huoxue decoction (QGHTHXF), Yuganlong granules (YGLKL), and Ganning tablets (GNP). To identify their active ingredients, potential components with oral bioavailability (OB) ≥30% and drug-likeness (DL) ≥0.18 were first screened using the TCMSP (https://old.tcmsp-e.com/tcmsp.php, accessed on 5 June 2024) and SymMap (http://www.symmap.org/, accessed on 10 June 2024) databases. Subsequently, a Venn diagram was used to analyze the common active ingredients among the four formulas, ultimately identifying seven shared components, including AS. Furthermore, target prediction for these common ingredients was performed using the SwissTargetPrediction database (http://swisstargetprediction.ch/, accessed on 17 June 2024). The resulting protein targets were submitted to the UniProt database (http://www.uniprot.org, accessed on 25 June 2024) and standardized by applying a “Homo sapiens” filter. Finally, Cytoscape 3.7.2 was used to construct a target gene interaction network, systematically revealing the potential mechanisms of action.

Target information related to MASLD was collected from three databases: the DisGeNET database (https://www.disgenet.org/, accessed on 19 June 2024), the GeneCards database (http://www.genecards.org/, accessed on 19 June 2024) and the CTD database (https://ctdbase.org/, accessed on 19 June 2024). Relevant targets were retrieved using the search term “nonalcoholic fatty liver disease.” The disease targets in the CTD database were screened with an inference score greater than 20, and the disease targets in the GeneCards database were screened with a relevance score greater than the median.

The UniProt database was used to unify disease targets into gene symbols. Targets that are common in each database are screened out as therapeutic MASLD targets. Subsequently, the targets of MASLD were intersected with the targets of seven key active components, and the overlapping sites were selected as potential targets for MASLD intervention.

The shared drug–disease targets were imported into the STRING database (https://cn.string-db.org/, accessed on 27 June 2024) to construct a PPI network. Results were visualized using Cytoscape 3.7.2, and the network topology parameters (degree, betweenness, and closeness) for these targets were obtained using the NetworkAnalyst tool in Cytoscape.

Core targets were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses using the DAVID online platform (https://david.ncifcrf.gov/, accessed on 7 July 2024). The species was set to H. sapiens, and a significance threshold of p < 0.05 was applied. The results were sorted in a descending order based on p-values, and the top 10 entries for GO analysis and the top 20 entries for KEGG analysis were selected for visualization. We aimed to explore the potential roles and mechanisms of the related genes in biological processes, molecular functions, cellular components, and signaling pathways, providing an intuitive representation of the potential mechanisms by which seven shared active ingredients, including AS may treat MASLD.

Eight-week-old male C57BL/6J mice were purchased from SPF (Beijing) Biotechnology Co., Ltd. and allowed to acclimate for 1 week.

The MASLD model was constructed using a high-fat and high-cholesterol (HFHC) diet. After 1 week of adaptive feeding, C57BL/6 mice were randomly divided into three groups (n = 6). The grouping and treatment regimens were as follows: (1) NCD group: normal chow diet + intragastric administration of normal saline for 6 weeks; (2) HFHC group: high-fat and high-cholesterol diet (fat content: 20% and cholesterol content: 2%) + intragastric administration of normal saline for 6 weeks; and (3) HFHC + AS group: high-fat and high-cholesterol diet + intragastric administration of AS (600 mg/kg/day) for 6 weeks.

A liver fibrosis model was established by intraperitoneal injection of carbon tetrachloride (CCl₄). After 1 week of adaptive feeding, C57BL/6 mice were randomly divided into three groups (n = 6 per group), and all mice were fed a normal diet throughout the experiment. The grouping and treatment protocols were as follows: 1) normal chow diet (NCD) group: intraperitoneal injection of corn oil (twice a week) + intragastric administration of normal saline for 4 consecutive weeks; 2) carbon tetrachloride (CCl₄) group: intraperitoneal injection of CCl₄ (0.5 mL/kg per injection, twice a week) + intragastric administration of normal saline for 4 consecutive weeks; and 3) carbon tetrachloride + AS (CCl₄ + AS) group: intraperitoneal injection of CCl₄ (0.5 mL/kg per injection, twice a week) + intragastric administration of AS (600 mg/kg per day) for 4 consecutive weeks. For the convenience of injection, the CCl₄ solution for intraperitoneal injection was prepared at a volume ratio of CCl₄: corn oil = 1:9.

At the end of the treatment period, all mice were sacrificed. Liver tissues were harvested for morphological observation of hepatic injury and subsequent experimental assays. All mice were housed in an SPF-grade animal facility at the School of Medicine, Guizhou University, under standard conditions (25 °C, 12-h light/dark cycle, with free access to water). All experimental procedures involving animals were approved by the Ethics Committee for Animal Use and Care of Guizhou University and complied with relevant ethical guidelines (HMEE-GUZ-2024-T025). Liver tissues from ob/ob and db/db mice were generously provided by the laboratory of Professor Bin Liang at the Center for Life Sciences, Yunnan University.

The harvested liver tissue was fixed in 10% neutral buffered formalin, embedded in paraffin, and cut into 4-μm sections. Sections were stained with hematoxylin and eosin (H&E) using a commercial staining kit (SolarBio, G1120), according to the manufacturer’s instructions for histological assessment.

Frozen sections were prepared from collected liver tissue and stained with Oil Red O using a commercial kit (G1262, SolarBio, China), according to the manufacturer’s instructions. The cellular Oil Red O staining procedure was also carried out according to the instructions.

Cells were cultured in DMEM (E600033, Sangon Biotech Co., Ltd., Shanghai, China) supplemented with 10% fetal bovine serum (164210-50, Procell Life Science and Technology Co., Ltd., Wuhan, China) and 1% penicillin–streptomycin (S110JV, Yuanpei Biotech Co., Ltd., Shanghai, China) and routinely incubated in a humidified incubator at 37 °C with 5% CO₂; Hepa1-6 cells, HCCLM3 cells, and LX2 cells were kindly provided by the research group of Professor Bin Liang from the Center for Life Sciences, Yunnan University, China. The MASLD cell model was established by inducing Hepa1-6 cells or HCCLM3 cells with a mixture of palmitic acid (PA, H8780, SolarBio Science and Technology Co., Ltd., Beijing, China) and oleic acid (OA, 0108485, Aladdin Biochemical Technology Co., Ltd., Shanghai, China) at a final concentration of 0.6 mM (OA: PA = 2:1); the hepatocyte fibrosis model was constructed by incubating LX2 cells with 10 ng/mL transforming growth factor-β1 (TGF-β₁) for 24 h; during modeling, different concentrations of AS were added to the administration groups for co-incubation.

Hepa1-6 and HCCLM3 cells were seeded in 96-well plates at a density of 2.5 × 10³ cells per well and allowed to adhere overnight. The following day, the cells were treated with varying concentrations of AS (5, 10, 20, 30, 40, 50, 60, and 70 μM) for 24 h. A control group, treated with 0.1% DMSO (vehicle), was included to correspond to the 0 μM AS concentration. The CCK-8 Assay Kit (K1018, APExBIO, United States) was used to assess cell viability, according to the manufacturer’s instructions.

The contents of total cholesterol (TC) and triglyceride (TG) in hepatocytes were quantified using commercial detection kits (TC kit: Nanjing Jiancheng Bioengineering Institute, China, Cat. No. A111-1-1; TG kit: Nanjing Jiancheng Bioengineering Institute, China, Cat. No. A110-1-1), following the manufacturer’s standard protocols. In accordance with the kit instructions, the detection wavelengths for TC and TG were set to 450 nm and 510 nm, respectively.

Total RNA was extracted using RNAiso Plus (108-95-2, TaKaRa, Japan), and RNA quality was assessed using a Thermo NanoDrop 2000 Spectrophotometer. cDNA synthesis was performed using the PrimeScript RT Reagent Kit (RR036A, TaKaRa, Japan). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the internal control to normalize target gene expression. Relative expression levels were calculated using the 2^−ΔΔCt method. All analyses were performed independently in triplicate. Primer sequences are listed in Supplementary Table S1.

First, hepa1-6 or HCCLM3 cells were seeded onto coverslips placed in 24-well plates and treated with palmitic acid/oleic acid (PA/OA) and AS or other drugs 24 h later. After treatment, the cells were fixed with 4% paraformaldehyde in PBS solution for 1 h at room temperature or overnight at 4 °C. The coverslips were then washed with PBS. The coverslips were then washed three times with PBS for 5 min each time. Cells were stained with 10 μM BODIPY (121207-31-6, MCE, United States) for 30 min, followed by three washes with PBS. Finally, the coverslips were sealed with a sealer containing DAPI.

Molecular docking was performed to further validate the interactions between AS and key target genes. The 2D structure of AS was downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/, accessed on 9 March 2025) and converted to PDBQT format using ObGUI software. The structures of the target proteins were retrieved from the Protein Data Bank (PDB) (https://www.rcsb.org/, accessed on 9 March 2025). After removing water molecules, adding polar hydrogen, and calculating Kollman charges, simulation docking of bioactive compounds and target proteins was performed using AutoDock Vina 1.5.6 software. Docking scores lower than −5.0 kcal/mol were considered indicative of strong binding interactions between the compound and its targets. The best docking poses were visualized using PyMOL software.

Hepa1-6 cells were cultured under standard conditions (37 °C and 5% CO₂) until reaching approximately 80% confluence and were then randomly divided into two groups: the control group and the AS-treated group. For the AS-treated group, cells were incubated with 4 μM AS at 37 °C for 2 h; the control group was treated with an equal volume of vehicle (dimethyl sulfoxide, DMSO) without AS. After treatment, cells were harvested and resuspended in phosphate-buffered saline (PBS) containing 1% protease inhibitor cocktail to prevent protein degradation. Cell lysis was performed using three consecutive freeze–thaw cycles (1 min of freezing in liquid nitrogen followed by 1 min of thawing in a 37 °C water bath per cycle). The cell lysates were centrifuged at 12,000 × g and 4 °C for 15 min, and the supernatants were collected.

Each supernatant sample was aliquoted into seven equal portions (50 μL per portion) and subjected to different temperature treatments (37, 40, 43, 46, 49, 52, and 55 °C) in a thermal cycler for 3 min. Immediately after heating, the samples were placed on ice for 5 min to terminate thermal denaturation, followed by centrifugation at 12,000 × g and 4 °C for 15 min to remove heat-aggregated insoluble proteins. The resulting supernatants (containing thermostable proteins) were mixed with 2× sodium dodecyl sulfate (SDS) loading buffer at a 1:1 volume ratio and boiled at 95 °C for 10 min to denature the proteins. The relative expression level of PPARγ in each sample was detected by Western blot analysis as described in Section 2.16.

Proteins were extracted from liver tissues (ob/ob mice and db/db mice liver tissues were generously provided by Prof. Bin Liang’s laboratory, School of Life Sciences, Yunnan University, China) or HCCLM3 cells using a high-potency RIPA solution, and the manufacturer’s procedure was strictly followed. Protein concentration was then quantified using the BCA Protein Assay Kit. A protein sample of 30 μg was added to each well of a 12% SDS-PAGE gel for electrophoresis. After separation, proteins were efficiently transferred to PVDF membranes (Millipore Corp, United States). To block non-specific binding, the membranes were incubated in 5% skimmed milk powder for 2 h. The membranes were incubated overnight at 4 °C with primary antibodies against PPARγ protein (16643-1-AP, 1:1000, Proteintech) and GAPDH (60004-1-Ig, 1:10,000, Proteintech). After washing the membrane three times, the membrane was incubated with the HRP-coupled secondary antibody (1:10,000, Azyme) for 1 h. Images were acquired using an automated chemiluminescence imaging system and analyzed using ImageJ 1.8.0.

Statistical analysis and graphing were performed using GraphPad Prism 8.0.2 software. Data are presented as the mean ± standard deviation (SD). A one-way analysis of variance (ANOVA) was used for comparisons among multiple groups, while the Student’s t-test was applied for comparisons between two groups. For statistical significance, p < 0.05 was considered statistically significant.

To discover natural products targeting MASLD with lipid-lowering effects, we analyzed classical TCM formulas from MASLD clinical guidelines. We identified small molecules consistently present across multiple formulae, specifically Ganning tablets, Zaozhu Yinchen decoction, Qinggan Huatan Huoxue decoction, and Yuganlong granules. Screening for DL and OB yielded 430 initial candidate compounds (Supplementary Table S2). Among these, seven constituents (aloe-emodin, AS, rhein, phellopterin, sitosterol, alloisoimperatorin, and ethyl oleate) were identified as common components shared across all four independent formulas (Figures 1A,B).

Furthermore, target prediction and analysis were performed for the seven compounds to elucidate their mechanisms of action. PPI network/hub gene analysis (Figures 1C,D) was followed by GO and KEGG enrichment analyses (Supplementary Table S3). GO analysis revealed significant associations with protein phosphorylation, inflammatory response, hypoxia response, and amyloid-beta processing, while molecular function enrichment highlighted protein kinase activities and ATP binding (Figures 1E–G). KEGG analysis identified key pathways, including pathways in cancer, chemical carcinogenesis, lipid and atherosclerosis, insulin resistance, NF-κB signaling, and apoptosis (Figure 1H). These findings indicate that the compounds likely ameliorate MASLD by modulating targets involved in NF-κB signaling, inflammation, apoptosis, and insulin resistance. Among the seven shared components, AS stood out due to its superior drug-likeness and oral bioavailability. Given its favorable pharmacokinetic properties and the unclear role of its mechanism in MASLD, we prioritized AS for in-depth mechanistic investigation.

To explore the potential of acetylshikonin (AS) in improving MASLD, we established a liver injury and metabolic disorder model in male C57BL/6J mice using an HFHC diet, followed by 6 weeks of AS intervention (Figure 2A). The results showed that compared with the HFHC control group, AS treatment markedly inhibited the increases in body weight and liver weight induced by an HFHC diet (Figures 2B,C) and improved HFHC-induced hepatic steatosis, as manifested by a decrease in the liver-to-body weight ratio (LW/BW) (Figure 2D). Meanwhile, AS treatment could reduce blood glucose levels, alleviate insulin resistance (Figures 2E,F), and decrease hepatic triglyceride and total cholesterol contents (Figure 2G). H&E and Oil Red O staining of liver tissues further confirmed that AS could significantly reduce intrahepatic lipid deposition (Figure 2H). At the molecular mechanism level, AS could significantly regulate the mRNA expression levels of key hepatic lipid metabolism regulators (Pparγ, Pparα, Cd36, Srebp2, Srebp1c, and Hmgcr) (Figures 2I,J), and significantly downregulate the mRNA expression of inflammation-related genes (Tnfα, Mcp1, and Il10) (Figure 2K).

Liver fibrosis is a critical progressive stage of fatty liver disease. To clarify the role of AS in liver fibrosis, we established a CCl₄-induced in vivo liver fibrosis model (Supplementary Figure S1A). Sirius Red staining confirmed that AS could significantly improve CCl₄-induced liver fibrosis (Supplementary Figure S1B). In in vitro experiments, LX-2 hepatic stellate cells (HSCs) were stimulated with TGF-β₁ to establish an activation model. Quantitative real-time polymerase chain reaction (qPCR) results showed that AS could markedly inhibit the upregulation of mRNA expression of key activation markers of hepatic stellate cells induced by TGF-β₁ (Col1a1, Col3a1, Acta2, and Tgfb ₁ ) (Supplementary Figure S2). In summary, AS can effectively improve HFHC diet-induced hepatic steatosis and alleviate CCl₄-induced liver fibrosis.

Building on these findings, we further explored the effects of AS on hepatic cells by delineating its impact on Hepa1-6 hepatocytes, focusing on cell viability, lipid accumulation, and transcriptional regulation. AS demonstrated dose-dependent cytotoxicity over 24 h (IC₅₀ = 26.65 μM; Figure 3A). Meanwhile, the planned experimental concentrations of AS, when combined with PA/OA, did not exert cytotoxic effects on cells (Figure 3B). Treatment with AS (2–4 μM) markedly attenuated PA/OA-induced elevations in TG and TC levels (Figures 3B,C). Oil Red O staining corroborated these findings, revealing a marked reduction in the relative area of lipid-positive deposits following AS administration (Figures 3D–G). Complementary BODIPY staining visualized a substantial decrease in both the number and size of intracellular lipid droplets upon AS treatment, with quantitative fluorescence analysis confirming significantly lower intensity compared to PA/OA controls (Figure 3H). Furthermore, AS elicited coordinated downregulation of key pathological genes: lipid metabolism regulators Cd36 and Pparγ (Figure 3I); cholesterol biosynthesis mediators Srebp2 and Srebp1c (Figure 3J); and inflammatory cytokines Il-8 Ccl2 and Tnfα (Figure 3K). Collectively, these data demonstrate that AS suppresses lipid accumulation and inflammation in hepatocytes, demonstrating therapeutic potential for MASLD.

To comprehensively evaluate the lipid-lowering potential of AS for MASLD, we extended our investigation to human cellular models. In particular, AS exhibited dose-dependent cytotoxic effects in human HCCLM3 cells, with a 24-h IC₅₀ value of 14.45 μM (Figure 4A). Meanwhile, the planned experimental concentrations of AS, when combined with PA/OA, did not exert cytotoxic effects on cells (Figure 4B). Following PA/OA induction, AS treatment (2–4 μM) markedly attenuated intracellular TG and TC contents (Figure 4C). Oil Red O staining confirmed markedly reduced lipid deposition, with AS treatment decreasing the relative stained area by 58.3% at 4 μM (Figures 4D–G). Complementary fluorescence imaging demonstrated that AS substantially reduced both the number and size of lipid droplets, as visualized using BODIPY staining. Quantitative analysis revealed lower BODIPY fluorescence intensity in AS-treated cells versus PA/OA controls (Figure 4H). Collectively, these results demonstrate that AS effectively reduces lipid accumulation in PA/OA-treated human HCCLM3 cells, further supporting its potential role in MASLD.

Building on the above findings, we further investigated the potential mechanisms by which AS exerts its therapeutic effects in MASLD. To this end, we systematically integrated data from multiple databases (CTD, DisGeNET, and GeneCards) to analyze the overlap between MASLD-related genes and AS targets. As shown in Figure 5A, a total of 245 genes were commonly identified across all three databases, highlighting their potential significance in MASLD pathogenesis (Supplementary Table S4). Further analysis revealed that 245 genes were associated with MASLD, while 282 were related to AS, with 31 overlapping genes implicated in both disease progression and AS pharmacological activity (Figure 5B; Supplementary Table S5). To better understand the functional relevance of these intersecting genes, we constructed a core gene interaction network, identifying central nodes such as Pparγ, Pparα, mTOR, Fasn, Hif1a, Esr1, and Bcl2, which are key regulators of lipid metabolism, inflammation, and cell proliferation (Figure 5C). Functional enrichment analysis further demonstrated that these genes participate in critical biological processes and signaling pathways: for instance, Pparγ and Fasn are linked to fatty acid metabolism, Hif1a to hypoxia responses, and Esr1 to nuclear receptor activity (Figures 5D–G; Supplementary Table S6). Additionally, Figure 5H highlights the top-ranked genes based on the network degree, underscoring the pivotal roles of Pparγ, Pparα, and Hif1a. Collectively, these integrative analyses provide mechanistic insights into the multi-target actions of AS in MASLD and support its therapeutic potential.

To validate these network predictions, we performed molecular docking analysis between AS and core regulators (Pparγ, Pparα, mTOR, Fasn, Hif1a, and Esr1) identified in the gene interaction network. Figure 6A illustrates the binding modes of AS with these targets, demonstrating favorable binding affinity to all six targets. Molecular docking results demonstrated strong binding affinities between AS and all target proteins, validating the rationality of our screening strategy (Supplementary Figure S3). Based on the interaction networks identified in Figures 5C, 6A, we next directed our attention to the AS–PPAR signaling axis as a potential mechanistic pathway. In two widely used MASLD models, ob/ob and db/db mice, Western blot analysis revealed upregulation of PPARγ expression compared to wild-type controls (Supplementary Figure S4A,B), highlighting its pathological relevance.

To further elucidate the mechanism of action of AS, we investigated its modulatory effects on PPARγ. Figure 6C further revealed that AS significantly enhanced the thermal stability of PPARγ across various temperatures. To validate the regulatory effect of AS on PPARγ expression, we first assessed Pparγ mRNA levels in the livers of HFHC-fed mice treated with AS. The results showed that AS significantly downregulated Pparγ expression compared to the HFHC group (Figure 5C). Furthermore, in cellular models, Western blot analysis demonstrated that AS dose-dependently reduced PPARγ protein expression (Figure 5D). Consistently, co-administration of the PPARγ antagonist GW9662 potentiated the lipid-lowering efficacy of AS, manifested as markedly reduced lipid droplet accumulation in vitro (Figures 6E,F). These results definitively establish the involvement of PPARγ in the mechanism of action of AS, indicating that its therapeutic efficacy against MASLD operates through suppression of the PPARγ pathway.