Schisantherin A (Sin A) powder was obtained from Nanjing Goren Bio-Technology Co., Ltd. (Nanjing, China, CAS 58546-56-8). The high-fat diet (HFD) was purchased from Research Diets Co., Ltd. (D12492, USA). The toll-like receptor (Tlr4) inhibitor, CLI-095, was provided by Med Chem Express Co., Ltd. Shanghai, China (HY-11109; Shanghai, China). Ampicillin Sodium Crystalline (A9518-25G-9), Neomycin Trisulfate Salt Hydrate Biorea& (N6386-5G), Vancomycin HCL (V2002-250MG) and Metronidazole Bioxtra (M1547-5G) were purchased from Sigma-Aldrich Co., Ltd. (USA). Primary antibody Occludin (ab216327) was purchased from Abcam (Cambridge, UK) and Goat anti-Rat IgG(H+L) Secondary Antibody, Alexa Fluor 488 Conjugate(A-11006), was provided by Thermo Fisher Scientific Inc. (NYSE: TMO, USA). In all cases, cell culture wells and reagents were obtained from Corning Incorporated (Corning, NY, USA) or Gibco (Carlsbad, CA, USA).

Animal experiments were conducted in accordance with the Guidelines for Animal Experimentation of Shang Hai University of Traditional Chinese Medicine (China), and the protocols were approved by the Animal Ethics Committee of this institution.

To investigate the potential protection of Sin A in HFD-induced NAFLD mice, C57Bl/6J male mice (aged 6 weeks) were randomly divided into three experimental groups (n=8–10 mice per group) as follows: chow diet group (Chow); HFD group (HFD); and HFD+Sin A group (Sin A). Treatment was started concomitantly with the introduction of HFD and consisted of daily oral doses of Sin A (80 mg/kg) (Nanjing Goren Bio-Technology Co., Ltd, Nanjing, China) for a total of 6 weeks.

To elucidate the function of LPS-TLR4 signaling in the HFD-fed mice, C57Bl/6J male mice (aged 6 weeks) were randomly divided into four experimental groups (n=5–8 mice per group) as follows: HFD group (HFD); HFD+Sin A group (Sin A), HFD + CLI-095 group (CLI-095), and HFD + CLI-095 + Sin A group (CLI-095 + Sin A). A TLR4 inhibitor (CLI-095, 2 mg/kg, i.p. injection) was administered every day for 6 weeks. Controls were injected daily with the vehicle (10% DMSO in saline). Parallel experiments using the HFD model and Sin A administration were performed over 6 weeks.

To confirm the mediation of the gut microbiota, C57Bl/6J male mice (aged 6 weeks) were randomly divided into four experimental groups (n=5–8 mice per group) as follows: HFD group (HFD), HFD + Sin A group (Sin A), HFD + ABX group (ABX), and HFD + ABX + Sin A group (ABX + Sin A). Antibiotics (Ampicillin Sodium Crystalline, Neomycin Trisulfate Salt Hydrate rea, Vancomycin HCL and Metronidazole Bioxtra were purchased from Sigma-Aldrich, Shanghai, China) were administered in drinking water ad libitum and replaced with freshly prepared cocktails every second day. Parallel experiments using the HFD model and Sin A administration were performed over 6 weeks.

Serum or hepatic levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), free fatty acids (FFAs), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglyceride (TG), and total cholesterol (TC) were measured using corresponding assay kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, Jiangsu, China) according to the manufacturer’s instruction.

For hematoxylin–eosin (H&E) staining, liver and ileum tissues were fixed in a 10% phosphate-buffered formalin, and paraffin was embedded. A tissue section of 5-μm thickness was cut and then stained with H&E according to a standard protocol. Immunofluorescence detection of Occludin was performed on the paraffin section using a standard technique as follows: blocking with 5% BSA at RT for 1 h; overnight incubation with primary rabbit anti-Occludin antibody (ab216327, Abcam, Cambridge, UK) at 4°C; and incubation with secondary antibody (Goat anti-Rat IgG(H+L), Alexa Fluor 488 conjugate, Thermo Fisher Scientific, A-11006, 1:500) at RT for 45 min.

The hepatic lipid accumulation was evaluated by Oil Red O (ORO) staining assay. Briefly, a liver fraction was washed 3 times with PBS buffer and then was transferred to ice incubation with 10% formalin. After 30 min of fixation, the hepatic tissues were stained with 100 μl of 0.5% Oil Red O suspended in a water solution for 20 min, then transferred to 75% ethanol for 2 min. Lastly, after being rinsed in 100 μl of distilled water for 3 times, the lipid accumulation of tissues was visualized by imaging.

At week 5, mice were individually placed in metabolic cages to determine oxygen consumption and physical activity using CLAMS (Columbus Instruments, Columbus, OH, USA) according to the manufacturer’s instructions. All metabolic studies required a room temperature of 24°C under a 12-h light/dark cycle. Heat production and Respiratory Exchange Ratio were calculated as described previously (Solon-Biet et al., 2015).

At week 6, mice were subjected to a cold room (4–6°C) without access to food, but with normal water. The rectal temperature was measured every 20 min within 2 h after exposure to the cold.

The glucose tolerance test was performed in diet-induced obese mice after overnight fasting. After a bolus i.p. injection of glucose at 2 g/kg, tail vein blood was immediately collected to measure glucose concentration at the indicated time point. In parallel, an insulin tolerance test was performed in 5-h fasted mice, and the glucose levels were measured at the indicated time point after i.p. injection of human insulin at 0.75 U/kg.

Total RNA was extracted using TRIzol reagent (Takara Bio, Inc., Shiga, Japan; #3732) according to the manufacturer’s instructions. RNA concentrations were equalized and converted to cDNA using a kit (Takara; #RR037). Gene expression was measured by qPCR (Applied Biosystems, Thermo Fisher Scientific, Carlsbad, CA, USA) using the SYBR Green system (Takara Bio, Inc.). Expression was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or β-Actin. Primer information is provided in Supplementary Table S1 . All gene expression procedures, including the design of primers, quantitation methods, and validation of PCR environment, were conducted according to previous studies (Johnson et al., 2014).

At week 6, mice were subjected to an administration of Dextran-FITC in the morning. Blood (30 μl) was collected from the tail in 4-h fasted mice. After 15 min of centrifugation, 15 μl of serum from each sample was obtained, then diluted with water to 100 μl. Using a black bottom 96-well plate, the concentrations of Dextran were determined at 525 μm according to the standard curve method.

The changes in endotoxin in serum and feces were measured using an LPS enzyme-linked immunosorbent assay (ELISA) kit. In all cases, samples were performed under pyrogen-free conditions and inactivated for 10 min at 70°C. The preparation of fecal samples was performed following a standard protocol (Kim et al., 2012). Finally, endotoxin was measured with a modified kit protocol.

Using fresh and frozen stool samples, gut microbiota was characterized. Total DNA for the gut microbiome analysis was extracted from 200 mg of feces following a standard protocol (Chen et al., 2018). The 16S rRNA gene high-throughput sequencing was performed using a the Illumina HiSeq PE250 platform, according to standard protocols (Magoč and Salzberg, 2011) by Majorbio Bio-Pharm Technology Co. Ltd. Shanghai, China. To reach the 97% similarity of operational taxonomic units (OTUs), using USEARCH, chimeras were filtered and the remaining sequences were clustered (Edgar, 2013). The RDP classifier was required for assigning a representative sequence of each OTU in the Ribosomal Database Project database. The process of 16S rRNA gene sequence data analysis was performed according to linear discriminant analysis effect size (LEfSe), whose differences among biological groups were determined for significance following nonparametric factorial Kruskal-Wallis rank-sum and Wilcoxon rank-sum tests (Wang et al., 2007).

The collagenase perfusion method was performed to obtain primary hepatocytes in this study. Briefly, the wide type C57 BL/6J male mice of 6–8 weeks were used to isolate primary hepatocytes. The liver tissues were digested by a perfusion of collagenase type IV solution, then dissected, minced, and transferred to a culture medium through a 70 μm strainer. Finally, the hepatocytes were collected by a centrifugation method with a 500 g speed. The isolated hepatocytes were resuspended and cultured in William’s E medium including 10% FBS and 1% penicillin-streptomycin (PS).

Human epithelial colorectal adenocarcinoma (Caco-2) cells were obtained from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin at 37°C with 5% CO₂ in a cell incubator. During the experiment, cells were pretreated with Sin A (50 μm) for 2 h and then were stimulated with LPS (2.0 μg/ml) for 22 h.

The Caco-2 cells were seeded into Corning Costar™ Transwell™ Permeable (Corning, NY, USA) Supports in 24-well plates at a cell density of 2×10⁵ cells/cm². In all cases, the incubation was 14 days until a monolayer of confluency was reached. Using an EVOM meter, all transepithelial electrical resistance (TEER) measurements were determined at 37°C, following standard protocols (Li et al., 2010).

The results of the biological assay are presented as means ± SD. The differences between the two groups were analyzed by Student’s t-test. Results that involved more than two groups were assessed by one-way or two-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test. All statistical significance was calculated using GraphPad Prism 7 (LaJolla, CA, USA) as follows: a p-value of less than 0.05 (*p < 0.05, **p < 0.01, and ***p < 0.001) was considered statistically significant. The statistical methods and corresponding p values for the data shown in each panel are included in the figure legends.

To address whether Sin A displayed protective effects for a liver metabolic phenotype and inflammation, we used HFD-fed mice as a NAFLD model (Kakimoto and Kowaltowski, 2016) ( Supplementary Figure 1A ). In HFD-fed mice, a 16-week treatment of HFD significantly increased liver size, liver weight, and the ratio of liver weight to body weight. This effect was significantly reversed with a 6-week administration of Sin A ( Figures 1A, B ). According to a previous study, NAFLD is characterized by excessive intrahepatic triglyceride (IHTG) content, which is universally associated with obesity and can lead to inflammation and fibrosis (Lu et al., 2018; Smith et al., 2020; Bril et al., 2020). We found that Sin A significantly decreased the levels of triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), free fatty acid (FFA), and increased high-density lipoprotein cholesterol (HDL-C) compared with the vehicle controls ( Figures 1C–F ). Consistent with these alterations on lipid profiles, the results of H&E and Oil Red O staining (ORO) also suggested that more lipid droplets occurred in the HFD-fed mice than the control mice, whereas Sin A treatment improved the phenotype of liver in obese mice. (i.e., reduced macrosteatosis, hepatocyte ballooning, and IHTG contents) ( Figure 1G and Supplementary Figure 1B ). In summary, these results showed that Sin A performed robust metabolic protection against NAFLD in HFD-fed mice.

To further examine the relationship between NAFLD and lipid metabolism, the key genes associated with lipid profiles were evaluated. As shown in Figure 2A ; the expression levels of genes related to fatty acid synthesis were significantly suppressed after Sin A administration, especially the sterol regulatory element binding protein-1c (Srebp1c), acetyl-CoA carboxylases alpha (Acaca), monoacylglycerol O-acyltransferase 1 (Mogat1), and peroxisome proliferative activated receptor, gamma (Pparg). This alteration of lipid synthesis genes was consistent with that in the primary hepatocytes ( Figure 2B ). Furthermore, fatty acid transport-related genes cluster of differentiation 36 (Cd36) and fatty acid transport protein 5 (Fatp5) were also significantly reduced in Sin A-treated mice when compared with vehicle- treated NAFLD mice ( Figure 2C ). The changes in very low-density lipoprotein export mRNA levels implied that more LDL-C accumulated in the liver of the HFD group compared with the control group and was significantly reversed by Sin A treatment ( Figure 2D ). To understand the close relationship between NAFLD and chronic inflammation (Furman et al., 2019; Farzanegi et al., 2019), the expression levels of specific inflammatory mediators were analyzed by quantitative real-time PCR. The result showed that Sin A was efficacious in reducing the liver inflammatory response as indicated by markedly suppressed expression of tumor necrosis factor α (Tnfα), interleukin 6 (Il6), nitric oxide synthase2 (Nos2), and Lps-binding protein (Lbp) ( Figure 2E ). At the same time, Sin A also exerted protective effects against HFD-driven liver damage, as observed by significantly reduced serum levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) ( Figure 2F ). Finally, we also evaluated the effects of Sin A on hepatic macrophage infiltration using F4/80 staining. We found that Sin A could remarkably suppress macrophage recruitment in the liver when compared with HFD mice ( Figure 2G and Supplementary Figure 1C ). The results above underlined metabolic improvements of Sin A in NAFLD, as validated by greatly ameliorated liver steatosis and reduced inflammatory response.

Increased gut inflammation and compromised intestinal barrier function are increasingly emerging as the primary cause of NAFLD progression, which is commonly associated with enhanced translocation of gut microbiota metabolites (Canfora et al., 2019; Ji et al., 2019). As expected, the result of ileum tissue H&E staining showed that vehicle-treated NAFLD mice displayed a pronounced epithelial disruption compared to control mice, which could be modestly restored by Sin A treatment ( Supplementary Figure 1D ). It is well-established that gut macrophage infiltration is the leading cause of intestinal inflammation by secreting inflammatory cytokines and recruiting other inflammation-driven cells. After a 6-week treatment with Sin A, we observed a significant decrease in macrophage infiltration compared with vehicle controls ( Figure 3A ). Consistently, Sin A administration significantly decreased mRNA levels of inflammatory cytokines (Il6; Il1β, interleukin 1 beta) ( Figure 3B ). We further evaluated the impact of Sin A on intestinal barrier function by utilizing an intestinal permeability assay. Sin A significantly reduced the concentration of Dextran-FITC in the HFD groups ( Figure 3C ). To assess whether Sin A could influence the release of LPS into circulation, which is closely related to epithelial integrity and gut permeability (Stephens and von der Weid, 2020; Longo et al., 2020), we evaluated the levels of fecal and serum endotoxin. As expected, Sin A treatment significantly reduced the production and release of endotoxin in HFD mice compared to vehicle controls ( Figures 3D, E ). Consistently, the restoration of gut barrier integrity in Sin A-treated mice was confirmed by an obviously increased expression of ileum tight junction molecular Occludin ( Figure 3F and Supplementary Figure 1E ) and increased mRNA levels of key intestinal barrier mediators and anti-barrier peptides ( Figure 3G ).

To further elucidate the effect of Sin A on the integrity of -cellular barriers, we used human epithelial colorectal adenocarcinoma cells (Caco-2 cell line) to perform in vitro experiments. Caco-2 cells were pretreated with Sin A using pre-determined concentrations for 2 h, then stimulated by LPS (2 μg/ml) for 22 h. As expected, the changes in gene expression indicated that Sin A administration (50 μm), co-incubated with LPS treatment, significantly attenuated the inflammatory response and restored intestinal permeability compared to vehicle controls ( Figure 3H ). We also found that the TEER of the LPS group was significantly decreased compared with that in the control group. However, combination with Sin A remarkably reversed this effect ( Figure 3I ). The heatmap generated by Spearman’s correlation analysis described the correlations between different parameters of NAFLD, including endotoxin, blood lipids, hepatic lipids, dextran concentration, inflammation, and intestinal integrity-related genes in ileum tissue ( Supplementary Figure 2A ). This heatmap demonstrated that the intestinal barrier function was negatively correlated with parameters promoting NAFLD and positively correlated with preventing NAFLD.

As described here, we demonstrated that a circulating increase of LPS impaired the maintenance of intestinal integrity and disrupted metabolic features in NAFLD mice (Carpino et al., 2020). To further investigate the role of LPS-TLR4 signaling in regulating the effect of Sin A on metabolic physiology in NAFLD, we exposed mice to HFD and treated them with the TLR4 inhibitor, CLI-095. HFD mice were evaluated in the following groups: (1) control, (2) Sin A treatment, (3) CLI-095 treatment alone, and (4) CLI-095 + Sin A treatment ( Supplementary Figure 2B ). The CLI-095-treated mice displayed lower lipid accumulation when compared with HFD-fed control mice, while comparable changes were observed in both the CLI-095 and CLI-095 + Sin A treatment groups ( Figures 4A, B ). We also found that Sin A inhibited the LPS-TLR4 signaling pathway to ameliorate hepatic steatosis of NAFLD mice using liver histological staining ( Figure 4C ). To assess whether the loss of TRL4 signaling can influence the lipid profiles, we assessed the changes in specific gene expression in liver sections. Compared to HFD-fed NAFLD mice, Sin A-fed mice displayed a significant decrease in fatty acid synthesis and intake while displaying a significant increase in β-oxidation. These effects could be diminished with the loss of TLR4 signaling ( Figures 4D, E ). qPCR analysis demonstrated that mRNA expression of genes associated with inflammation (Tnfα, Il6, Il10, and Nos2) were not changed in mice with inhibited TLR signaling ( Figure 4F ). However, this expression could be significantly suppressed by Sin A in control groups ( Figure 4F ). Strikingly, when the TLR4 signaling was pharmacologically suppressed, Sin A administration neither restored intestinal barrier function nor improved gut inflammation in HFD mice ( Figures 4G, H ). These results suggest that mediation of LPS-TLR4 signaling was required for the beneficial improvements of Sin A on steatosis and inflammation in HFD-fed mice.

Given that NAFLD is accompanied by obesity, and considering that the known impacts of TLR4 signaling in high-fat-induced metabolic disarrangements (Liu et al., 2019; Zhang et al., 2020b), we therefore examined obesity-related parameters in HFD-fed mice in the presence of CLI-095 treatment. Sin A treatment significantly inhibited body weight gain in NAFLD mice, whereas CLI-095 administration abolished this improvement ( Figure 5A ). The effects of Sin A on serum free fatty acids (FFAs) and free glycerol were diminished when TLR4 signaling was blocked in NAFLD mice ( Figure 5B ). In addition, the effect of Sin A on fasting blood glucose (FBG) was remarkably reversed with the loss of TLR4 signaling in mice ( Figure 5C ). At the same time, we observed a potential role for Sin A on LPS-TLR4 signaling with respect to glucose homeostasis, as supported by no significant changes in glucose tolerance and insulin tolerance during CLI-095 treatment ( Figures 5D, E ). The metabolic cage test indicated that Sin A administration exhibited no apparent impact on heat production in CLI-095-treated mice ( Figure 5F ). To confirm this, we also subjected four groups to cold exposure: (1) HFD control, (2) Sin A treatment alone, (3) CLI-095 treatment alone, and (4) CLI-095 + Sin A treatment. We observed that TLR4 signaling is required for the effect of Sin A in the process of energy expenditure, as revealed by no significant changes in fecal temperature observed in both the CLI-095 and CLI-095 + Sin A groups when compared with the control groups ( Supplementary Figure 2C ). Taken together, these data suggested that Sin A mediates LPS-TLR4 signaling pathway to protect against obesity in HFD-induced NAFLD mice.

It is widely believed that the interplay between host components and the gut microbiome is far from our understanding of their mechanisms (Peng et al., 2019; Huang et al., 2019; Zeng et al., 2020). A great deal of evidence demonstrates that the gut microbiota is the key regulator in the development of NAFLD (Kolodziejczyk et al., 2019; Safari and Gérard, 2019), and targeting the gut microbiome could be a promising therapy for NAFLD. To explore whether Sin A treatment changed the composition of gut microbiota, we performed a 16s rRNA sequence analysis for normal chow-fed, HFD-fed, and Sin A treatment groups. Based on principle coordinates analysis (Pocan) of the weighted Unirac distance, we observed an apparent separation into color-coded clusters among the above three groups ( Figure 6A ), and we also observed an apparent difference in gut microbiota composition after Sin A treatment in the family and genus levels ( Supplementary Figure 3A ). To evaluate the effect of Sin A on the overall composition of the bacterial community, we then evaluated the changes in the degree of bacterial taxonomic similarity at the phylum level. As shown in Figure 6B ; there were more Firmicutes and fewer Bacteroidetes observed in the HFD-fed mice as compared with the chow-fed mice, whereas Sin A treatment exhibited obvious protection against these effects. It is possible that HFD-driven gut dysbiosis may cause more production of LPS and then lead to disrupted epithelial integrity, as well as impaired gut permeability, resulting in the increased release of LPS into the circulation system. Therefore, we conducted an antibiotics (ABX) test to investigate whether the mediation of the gut microbiome is responsible for Sin A-induced protection against NAFLD. First, we observed that there was a significant decrease in the LPS level both in the serum and feces of ABX-treated mice compared with vehicle controls ( Supplementary Figures 4A, B ), suggesting that antibiotic treatment may lead to microbiota depletion. ABX-treated mice displayed no significant alterations in lipid accumulation compared with control mice, as revealed by the TG and TC levels in serum and the liver ( Figures 6C, D ). Consistently, there was no apparent effect on the level of gene expression associated with fatty acid synthesis (Srebp1c, Acaca, Fasn, Pparg, and Mogat1) and uptake (Cd36, Fatp5, and Fabp1) in ABX-treated groups ( Figures 6E, F ). These results demonstrated that ABX treatment can diminish or even abolish the effects of Sin A on lipid metabolism. Second, we explored the roles of gut microbiota in Sin A-induced amelioration of liver inflammation using gene expression performance. We determined that the beneficial effects of Sin A on hepatic inflammation were also dismissed in ABX groups ( Supplementary Figure 4C ). As expected, ABX-treated mice displayed comparable changes in gene expression associated with gut inflammation and intestinal barriers, compared with control mice ( Figures 6G, H ). These data imply that the metabolic benefits of Sin A in NAFLD mice, including improved hepatic lipid accumulation and inflammation along with restored gut barrier function, were partly dependent on the mediation of gut microbiota.