Sulforaphane Ameliorates Nonalcoholic Fatty Liver Disease Induced by High-Fat and High-Fructose Diet via LPS/TLR4 in the Gut–Liver Axis

The gut–liver axis has emerged as a key player in the progression of non-alcoholic fatty liver disease (NAFLD). Sulforaphane (SFN) is a bioactive compound found in cruciferous vegetables; however, it has not been reported whether SFN improves NAFLD via the gut–liver axis. C57BL/6 mice were fed a high-fat and high-fructose (HFHFr) diet, with or without SFN gavage at doses of 15 and 30 mg·kg−1 body weight for 12 weeks. The results showed that SFN reduced weight gain, hepatic inflammation, and steatosis in HFHFr mice. SFN altered the composition of gut microbes. Moreover, SFN enhanced the intestinal tight junction protein ZO-1, reduced serum LPS, and inhibited LPS/TLR4 and ERS pathways to reduce intestinal inflammation. As a result, SFN protected the intestinal integrity and declined the gut-derived LPS translocations to the liver in HFHFr diet-induced mice. SFN decreased the liver LPS levels and inhibited the LPS/TLR4 pathway activations, thus inhibiting the pro-inflammatory cytokines. Notably, Spearman correlation analysis showed that the protective effect of SFN on intestinal barrier integrity and its anti-inflammatory effect on the liver was associated with improved intestinal dysbiosis. Above all, dietary intervention with SFN attenuates NAFLD through the gut–liver axis.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is one of the most common liver diseases worldwide. It is characterized by excessive hepatic steatosis, lobular inflammation, and hepatocyte ballooning changes. Moreover, NAFLD may progress to non-alcoholic steatohepatitis (NASH) and cirrhosis, potentially leading to hepatic carcinoma [1]. NAFLD pathogenesis may be linked to abnormal glucose and lipid metabolism, inflammation, endoplasmic reticulum stress (ERS), oxidative stress, and imbalanced gut microorganisms, according to existing research [2,3]. However, safe and effective NAFLD therapies have yet to be discovered. Therefore, uncovering the pathogenesis of NAFLD and developing prevention and treatment strategies for NAFLD remains challenging for fatty liver research. The hepatic portal system anatomically connects the liver and gut and enables gut microbiota and their metabolites to be involved in the pathogenesis of NAFLD. Hence, the gut-liver axis is increasingly recognized as a critical factor in NAFLD [4]. Existing studies have observed intestinal flora disturbances and increased intestinal permeability in NAFLD animal models and NAFLD patients [3,[5][6][7]. The disturbance of gut microbiota may increase serum lipopolysaccharide (LPS), the Gram-negative bacteria's main outer SFN used in the animal was purchased from Aladdin (S111997, HPLC ≥ 95%, Shanghai, China). Fructose (C6H12O6, HPLC ≥ 99%) was purchased from Xiwang Sugar (Shangdong, China). A high-fat diet (HFD) (60% of energy derived from fat) was purchased from Research Diets (New Brunswick, NJ, USA). The maintenance diet was purchased from Keao Xieli (Beijing, China).

Animals and Treatments
Six-week-old male C57BL/6J mice were purchased from Beijing Charles River Laboratories (Beijing, China) and housed in an SPF animal house (22 ± 2 • C, 40~70% relative humidity, 12 h light/12 h dark cycles). The animal experiment design was shown in Figure 1A. After one-week acclimatization, they were randomly divided into four groups (8 mice per group): (1) mice fed the control diet (pure water and maintenance diet) as a Nutrients 2023, 15, 743 3 of 19 control (CON) group; (2) mice fed the high fructose (30% (W/V) fructose water and HFD as the HFHFr group; (3) mice fed the HFHFr diet supplemented with a low dose of SFN (15 mg/kg) as the SFN-L group; and (4) mice fed the HFHFr diet supplemented with a high dose of SFN (30 mg/kg) as the SFN-H group. SFN was given to mice every two days by gavage in a sterile saline solution. The CON group and HFHFr group were given sterile saline by gavage at a dose of 0.1 mL·10 g −1 ·body weight every two days to minimize the effects of the gavage procedure. The experiment lasted for 12 weeks during which the body weight of each mouse was monitored weekly. Fecal samples from each mouse were collected during the last week before the animals were euthanized. After 12 weeks, all animals were fasted for 12 h and then sacrificed. The organs were dissected immediately. The livers were weighed. Some of the liver and colon tissues of mice were fixed with 4% paraformaldehyde, and the rest were used for molecular and biochemical tests. The ratio of IL-10/IL-6 in the liver was measured by ELISA (n = 6). (H) The protein expression of TNF-α, and IL-1β in the liver was detected by Western blot (n = 3). Data are shown as mean ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001 compared with HFHFr, # p < 0.05, ### p < 0.001 compared with CON.
All the experimental procedures were performed according to the Guide for the Care The ratio of IL-10/IL-6 in the liver was measured by ELISA (n = 6). (H) The protein expression of TNF-α, and IL-1β in the liver was detected by Western blot (n = 3). Data are shown as mean ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001 compared with HFHFr, # p < 0.05, ### p < 0.001 compared with CON.
The contents of liver IL-6 and IL-10 were determined by enzyme-linked immune sorbent assay (ELISA) kits (R&D Systems, Minneapolis, MN, USA) by the manufacturer's protocols. The final results of IL-6 and IL-10 were normalized to total protein concentration.

Determination of Lipopolysaccharide in the Liver and Serum of Mice
The quantification of liver and serum LPS was performed following the instructions of the commercial kits (CUSABIO, Wuhan, China). The final result of the liver LPS was normalized to total protein concentration.

Histological Analysis
Briefly, frozen liver sections (10 µm) were stained with 0.5% Oil Red O (Sigma-Aldrich, St. Louis, MO, USA) for 15 min and then washed. Three fields from three sections of each mouse were viewed under a Leica microscope, and digital photographs were captured.
The liver and colon tissues were fixed in 10% (v/v) formalin overnight before being subjected to gradient ethanol dehydration. Then, the tissues were embedded, sliced, and stained with hematoxylin and eosin (H&E). The sections were scanned by 3DHISTECH Pannoramic SCAN (Budapest, Hungary). Microphotographs were taken by CaseViewer 2.4.
For immunohistochemistry (IHC), the sections of the colon were incubated in primary antibodies including ZO-1(Abcam, Waltham, MA, USA), TLR4, MyD88, and NF-κB (Proteintech Group, Inc. (Wuhan, China)). Secondary antibodies labeled with horseradish peroxidase were used. Detection was conducted using a horseradish peroxidase-based commercial detection system, disclosure with diaminobenzidine chromogen, and nuclear counterstaining with hematoxylin. The sections were observed under an Olympus BX60 microscope (Tokyo, Japan). The intensity of brown staining was analyzed using Image Pro Plus (version 6.0).

Real-Time Quantitative PCR
RNA was reverse-transcribed into cDNA with a One-Step gDNA Removal and cDNA Synthesis SuperMix (AT311, TransGen Biotech, Beijing, China). Real-time quantitative PCR (qPCR) was performed with a SuperReal PreMix Plus (FP 205-03, TIANGEN Biotech, Beijing, China) by using a real-time PCR system (C1000, Bio-Rad, Hercules, CA, USA). Gene expression levels were normalized to β-actin. Primer sequences are listed in Supplementary  Table S1.

Gut Microbiota Analysis
Microbial genomic DNA was extracted using a previously described method [30]. The bacterial V3 + V4 regions of the 16S rRNA were amplified and sequenced on a Nova-Seq platform (Illumina, San Diego, CA, USA). The raw data was filtered and analyzed by QIIME (v1.9.1). UPARSE v7.0.1001 was used to cluster the OTUs at an identity threshold of 97%. Nonmetric multidimensional scaling (NMDS) plots were analyzed using PAST v2.17 based on Bray-Curtis distance. LEfSe analysis was performed to find the differentially abundant biomarker between the experimental groups. Correlations between the experimental result parameters and the microbiota profiles were analyzed by Spearman correlational analysis.

Statistical Analysis
The results are presented as means ± SEMs. All data differences were analyzed by one-way ANOVA (Tukey's multiple comparison tests) on GraphPad Prism (version 8.0), which was considered significant at p < 0.05.

SFN Improves Weight Gain and Hepatic Steatosis in HFHFr-Diet Mice
As shown in Figure 1A, NAFLD mice models were generated with a 12-weeks HFHFrdiet. SFN-L treatment slowed the mice's weight gains from the eighth week and kept their body weight steady in the last three weeks ( Figure 1B). Likewise, the weight gain of the SFN-H-treated mice decreased. Although the decrease levels were weaker than in the SFN-L group, no statistical differences were observed between the two groups. These data indicated that SFN significantly inhibited the weight gain of HFHFr mice. Additionally, the liver weight of SFN-treated mice was lower than the HFHFr group ( Figure 1C). Serum ALT/AST ratio levels commonly used in liver injury assessments [31] were significantly decreased in the SFN-treated mice ( Figure 1D).
Oil Red O staining revealed that ( Figure 1E) the number and size of liver lipid droplets increased significantly in HFHFr mice and were improved by SFN-L or SFN-H treatments. The H&E staining showed clear and complete hepatic lobules in CON mice, with no evidence of steatosis and focal inflammatory infiltrations. In contrast to the CON group, the hepatic lobules of HFHFr mice were less complicated, with a variety of vacuolar lipid droplets, hepatocyte disorders, and severe inflammatory infiltration. The SFN supplement improved the mice's vacuole numbers and inflammatory cell infiltrations ( Figure 1E). Moreover, the intervention of low-dose SFN significantly reduced the content of TG in the liver ( Figure 1F). We assessed the regulatory effect of SFN on hepatic pro-and anti-inflammatory factors by using Il-10/Il-6. The results showed that SFN treatment reversed the decreased IL10/IL6 ratio, suggesting that it alleviated inflammation in the liver ( Figure 1G). Additionally, pro-inflammatory factors TNF-α and IL-1β were decreased in the SFN group compared with the HFHFr group ( Figure 1H). Although the SFN-L treatment showed more robust anti-inflammatory effects than the SFN-H treatment, there was no statistical difference between the two groups. These findings suggest that SFN intervention improves HFHFr diet-induced hepatic lipid accumulation, hepatic steatosis, and inflammatory infiltration in NAFLD mice.

SFN Improves Gut Dysbiosis Induced by the HFHFr Diet
An increasing number of studies have reported gut microorganism disorders in NAFLD individuals, indicating their predominant role in NAFLD development [32,33]. We used 16S RNA gene sequencing on mice feces to investigate SFN's role in NAFLD by regulating gut microbiota. NMDS analysis is a nonlinear model based on the Bray-Curtis distance. An NMDS stress of less than 0.2 means that this model accurately reflects the sample differences. Figure 2A shows the differences among samples from the experimental groups. Despite some overlaps, distinctive clustering between the HFHFr and SFN-treated mice can be observed, revealing specific differences in gut microbiota compositions after SFN intervention. We found that the differential bacteria at the phylum level were mainly Bacteroidota, Firmicutes, unidentified_Bacteria, Proteobacteria, Deferribacteres, Desulfobacterota, Campylobacterota, Verrucomicrobiota, and Cyanobacteria ( Figure 2B). We found that HFHFr-diet increased the Firmicutes/Bacteroidota ratio (F/B ratio) and significantly altered the composition of gut microbiota ( Figure 2C). SFN reduced the F/B ratio, and a statistical difference was observed between the HFHFr group and the SFN-L group.
A LEfSe (LDA > 4) analysis was performed to identify the significantly regulated specific phylotypes among the groups. Overall, 27 OTUs with significant abundance changes were screened as key phylotypes ( Figure 2D). Based on the LDA score, we found that the HFHFr treatment enriched the phylum Firmicutes, such as the family Lachnospiraceae, family Streptococcacea, and genus Lactococcus. The SFN-L treatment, however, enriched the family Atopobiaceae, phylum Actinobacteriota, and genus Coriobacteriaceae_UCG_002 ( Figure 2D). Different levels of marker taxa in experimental groups obtained from LEfSe were also supported by branch graph analysis ( Figure 2E).
Furthermore, we analyzed the dominant bacteria under SFN treatment. We noticed that HFHFr-diet increased the relative abundance of the family Streptococcaceae, Desulfovibrionaceae, genera Lactococcus, Desulfovibrio, Blautia, Dubosiella, Alistipe, and Lactobacillus, which are significantly declined by SFN treatment, especially in the low-dose treated mice ( Figure 2F-M). Existing studies have reported that the family Streptococcaceae is enriched in colonic inflammation mice [34]. The family Desulfovibrionaceae and the genus Desulfovibrio can produce LPS, resulting in increased LPS in the blood [35]. Lactococcus and Lactobacillus are positively related to body weight and liver weight [36]. Additionally, Blautia, Dubosiella, and Alistipe were raised in HFD-diet mice. These findings indicated that SFN reduces harmful gut bacteria. In contrast to the HFHFr group, SFN-L treatment enriched the family Butyricicoccaceae, genera Coriobacteriaceae_UCG-002, Alloprevotella, Lach-nospiraceae_NK4A136_group, Butyricicoccus and Bifidobacterium ( Figure 2N-S). Notably, the SFN-L supplement significantly increased the relative abundance of Coriobacteriaceae_UCG-002 compared with the other three groups ( Figure 2O). Butyricicoccaceae, Alloprevotella, Lachnospiraceae_NK4A136_group, Butyricicoccus, and Bifidobacterium were decreased in the HFHFr group. These short-chain fatty acid-producing bacteria improve gut permeability and protect the intestinal barrier [37][38][39]. The relative abundance of these bacteria increased significantly after SFN intervention, and the lower dose treatment showed a better effect. Overall, these results confirm that SFN intervention improves the gut microbiota of the HFHFr-diet mice.

SFN Ameliorates Intestinal Histopathology and Improves Intestinal Permeability in HFHFr Mice
The disruption of intercellular tight junctions contributes to increased intestinal permeability and nerve function in the pathogenesis of hepatic steatosis. So, we next assessed whether SFN intervention ameliorated intestinal damage in HFHFr mice. As displayed in Figure 3A, the intestinal tissue structure of the mice in the HFHFr group was severely abnormal, the local crypt structure in the mucosal layer of the colon tissue disappeared, and a large number of inflammatory cells was infiltrated by H&E staining, while these phenomena were improved in the SFN group. This result indicated that SFN intervention improved intestinal histomorphology and ameliorated intestinal damage.
Since tight junctions maintain the intestinal epithelial barrier, defects that may lead to intestinal damage, we assessed the integrity of the intestinal barrier. The transcript levels of intestinal epithelial proteins ZO-1 and Claudin-4 were decreased in the HFHFr group, while the mRNA levels were significantly upregulated in the SFN-L group ( Figure 3B). We further performed IHC to evaluate the expression of the intestinal tight junction protein ZO-1, and an elevated level of ZO-1 protein expression was observed in SFN groups compared with the HFHFr group ( Figure 3C,D). Additionally, impaired intestinal barrier function and intestinal flora disruption can result in increased production of LPS, a biomarker of intestinal permeability. As shown in Figure 3E, serum LPS in the HFHFr group was significantly higher, in contrast, SFN significantly reduced the level of serum LPS. These results suggest that SFN can restore the integrity of the intestinal mucosal barrier disrupted and lower intestinal permeability by upregulating tight junction proteins in the HFHFr diet.

Intervention of SFN Alleviates Intestinal Inflammation in HHFFr Mice by Inhibiting TLR4/NF-κB and ERS Pathway
Loss of barrier integrity may promote intestinal inflammation [40]. We found that the mRNA levels of pro-inflammatory cytokines TNF-α, IL-1β, and IL-6 and chemokines CCL2 and CCL4 were significantly increased in the HFHFr group, but these inflammatory mediators were decreased in SFN groups ( Figure 4A). Prior studies have largely focused on that, as LPS is a major pathogenic factor in the process of intestinal inflammation, it binds to TLR4 and activates the NF-κB pathway, causing an inflammatory response [28].

Intervention of SFN Alleviates Intestinal Inflammation in HHFFr Mice by Inhibiting TLR4/NF-κB and ERS Pathway
Loss of barrier integrity may promote intestinal inflammation [40]. We found that the mRNA levels of pro-inflammatory cytokines TNF-α, IL-1β, and IL-6 and chemokines CCL2 and CCL4 were significantly increased in the HFHFr group, but these inflammatory mediators were decreased in SFN groups ( Figure 4A). Prior studies have largely focused on that, as LPS is a major pathogenic factor in the process of intestinal inflammation, it binds to TLR4 and activates the NF-κB pathway, causing an inflammatory response [28]. Intriguingly, the mRNA expressions of TLR4, MyD88, and NF-κB were significantly increased in the HFHFr group, and the intervention of SFN significantly reversed the elevation of these genes ( Figure 4B). As demonstrated in Figure 4C-F, the intervention of SFN inhibited the activation of the TLR4/NF-κB signaling pathway in colon tissue. SFN may help to regulate inflammation and immune function by inhibiting the activation of the colonic TLR4/NF-κB pathway to reduce intestinal mucosal inflammation and colonic mucosal damage.
ERS in intestinal epithelial cells is associated with the activation of the host immune response and is the main factor in the pathogenesis of intestinal diseases [41]. We next investigated the effect of SFN on ERS-related marker genes in colon tissue. The mRNA expressions of GPR78 and IRE1α were increased in the HFHFr group, while the SFN intervention group significantly prevented the expression of ERS genes ( Figure 4G). Additionally, we detected the downstream pathway genes of IRE1α and observed that the SFN group reduced the mRNA expression of tumor necrosis factor-receptor-associated factor 2 (TRAF2), c-Jun N-terminal kinase (JNK), and CHOP ( Figure 4G). Taken together, the above results suggest that SFN intervention alleviates intestinal inflammation induced by the HFHFr diet. tervention group significantly prevented the expression of ERS genes ( Figure 4G). Additionally, we detected the downstream pathway genes of IRE1α and observed that the SFN group reduced the mRNA expression of tumor necrosis factor-receptor-associated factor 2 (TRAF2), c-Jun N-terminal kinase (JNK), and CHOP ( Figure 4G). Taken together, the above results suggest that SFN intervention alleviates intestinal inflammation induced by the HFHFr diet.

SFN Improves Inflammation by Inhibiting the TLR4/NF-κB Signaling Pathway in the Liver of HFHFr Mice
Disrupted gut barriers lead to LPS leakage into the circulation through gut enterocytes towards the liver through the portal vein due to increased gut permeability. Hence, we detected the LPS levels in the mice livers by ELISA. As shown in Figure 5A, SFN treatment significantly reduced the LPS level in the liver. The ability to recognize the increased LPS, TLR4 and their downstream pathway has shown dominance in the mechanisms of the microbiota-gut-liver axis [42]. LPS is a key factor in inducing an inflammatory response in liver tissue and plays an important role in liver injury through the LPS-TLR4 signaling pathway [43]. Interestingly, SFN inhibited LPS-triggered TLR4/NF-κB signal pathway activation, as evidenced by decreased protein expressions of TLR4, MyD88, and NF-κB ( Figure 5B-E). The activation of NF-κB is responsible for downstream inflammatory cytokines synthesis, and we further found that SFN inhibited phosphorylation of NF-κB ( Figure 5B,F). As a downstream transcription factor of the TLR4 pathway, NF-κB is responsible for stimulating the production of major pro-inflammatory cytokines [44], and the generation of TNF-α, IL-1β, and IL-6 in the livers was confirmed in Figure 1G,H. Therefore, these results suggest that SFN intervention can improve the hepatic inflammation of NAFLD by reducing LPS translocation and inhibiting the activation of its downstream TLR4/NF-κB signaling pathway.

of HFHFr Mice
Disrupted gut barriers lead to LPS leakage into the circulation through gut enterocytes towards the liver through the portal vein due to increased gut permeability. Hence, we detected the LPS levels in the mice livers by ELISA. As shown in Figure 5A, SFN treatment significantly reduced the LPS level in the liver. The ability to recognize the increased LPS, TLR4 and their downstream pathway has shown dominance in the mechanisms of the microbiota-gut-liver axis [42]. LPS is a key factor in inducing an inflammatory response in liver tissue and plays an important role in liver injury through the LPS-TLR4 signaling pathway [43]. Interestingly, SFN inhibited LPS-triggered TLR4/NF-κB signal pathway activation, as evidenced by decreased protein expressions of TLR4, MyD88, and NF-κB ( Figure 5B-E). The activation of NF-κB is responsible for downstream inflammatory cytokines synthesis, and we further found that SFN inhibited phosphorylation of NF-κB ( Figure 5B,F). As a downstream transcription factor of the TLR4 pathway, NF-κB is responsible for stimulating the production of major pro-inflammatory cytokines [44], and the generation of TNF-α, IL-1β, and IL-6 in the livers was confirmed in Figure 1G,H. Therefore, these results suggest that SFN intervention can improve the hepatic inflammation of NAFLD by reducing LPS translocation and inhibiting the activation of its downstream TLR4/NF-κB signaling pathway.

Correlation Supports That SFN Improves the Microbe-Gut-Liver Axis of HFHFr Mice
We performed a Spearman correlation heatmap to analyze the pathogenesis of the microbe-gut-liver axis in the HFHFr-induced NAFLD. Figure 6A shows the F/B ratio and the abundance of Streptococcaceae, Lactococcus, Blautia, Desulfovibrionaceae, Desulfovibrio, and Lactobacillus were positively related to colonic inflammatory genes, colonic endoplasmic reticulum stress genes, liver LPS levels, and serum LPS levels. Furthermore, the relative abundance of these gut bacteria was negatively correlated with the colonic tight junction protein genes ZO-1, Claudin-4, and the ratio of IL-10/IL-6 cytokines in the liver, while the relative abundances of Lachnospiraceae_NK4A136_group, Bifidobacterium, Akkermansia, Allprevotella, Butyricicoccaceae, and Butyricicoccus showed opposite trends to these experimental results. The results demonstrate that SFN intervention attenuates gut microbial dysregulation and protects mice from HFHFr diet-induced NAFLD.
Interestingly, we observed that the relative abundance of Butyricicoccus was significantly negatively correlated with the serum LPS levels (r = −0.6139, p = 0.0195) ( Figure 6B) and was positively correlated with Claudin-4 (r = 0.6598, p = 0.0054) ( Figure 6C) in the colon and the cytokine IL-10/IL-6 in the liver (r = 0.5998, p = 0.0066) ( Figure 6D). These findings indicated that Butyricicoccus was the key bacterium most able to exert beneficial effects through the gut-liver axis after SFN intervention. Additionally, serum LPS levels positively correlated with the F/B ratio (r = 0.6484, p = 0.0196) ( Figure 6E). Furthermore, the F/B ratio positively related to the immune-activated pathway genes, inflammatory genes, and endoplasmic reticulum stress marker genes in the colon, and was negatively correlated with the colon Claudin-4 and liver IL-10/IL-6 cytokines ratios ( Figure 6A). These results showed that HFHFr diet-induced gut dysbiosis led to increased gut permeability and plasma LPS, thereby promoting low-grade inflammation [45]. The expression of Claudin-4 in the colon was positively correlated with the relative abundance of Allprevotella and Akkermansia (r = 0.7909, p = 0.0060; r = 0.6519, p = 0.0118) ( Figure 6F-G). In addition, the expression of IL-1β in the colon was positively correlated with the relative abundance of Streptococcaceae (r = 0.7684, p = 0.00018) ( Figure 6H). The liver cytokine IL-10/IL-6 ratio was negatively correlated with the relative abundance of Blautia (r = −0.7064, p = 0.00034) ( Figure 6I) and positively correlated with the relative abundance of Bifidobacterium (r = 0.6753, p = 0.0021) ( Figure 6J). Hence, we conclude that the microbe-gut-liver axis is involved in the development of HFHFr diet-induced NAFLD and can be improved by the SFN intervention.

Discussion
Studies have shown that HFHFr diets most closely recapitulate the human phenotype of NAFLD [46], and drinking water with excessive fructose can lead to barrier deterioration [47]. Therefore, we adopted an HFHFr diet to construct the NAFLD model in mice. Existing studies on SFN against NAFLD focused on inhibiting hepatic liposynthesis and steatosis [23,48], improving insulin resistance [21], antioxidant effects after NRF2 activation [49], and protecting mitochondrial functions [22]. Nevertheless, the protective effect of SFN against NAFLD in the gut-liver axis remains unknown. In this article, we primarily revealed that SFN protected against NAFLD by enhancing the gut barrier, modulating gut microbiota, and LPS-mediated gut-liver axis.
Studies on animals and humans have reported altered gut microbes in NAFLD individuals. Gut microbiota dysbiosis was observed in HFHFr-induced obese rats [42]. We observed that SFN decreased the F/B ratio and the abundance of Lactobacillus, Lactococcus, Blautia, Dubosiella, and Alistipes at the phylum level in HFHFr mice. These bacteria are genera associated with obesity-induced metabolic diseases, and their relative abundances significantly increased in the gut of mice following a high-fat diet or a high-fat and highfructose diet [36,[50][51][52]. Among them, the Blautia abundance positively correlated with fecal deoxycholic acid (DCA) levels in cirrhotic patients [50], and damage to the intestinal barrier [53]. Desulfovibrio is a sulfate-reducing bacterium that produces LPS, induces inflammation, and is associated with the development of metabolic syndrome [54]. Our findings suggested that SFN treatment significantly decreased the relative abundance of Desulfovibrio, indicating that SFN may modulate the gut microbiota in NAFLD mice by reducing harmful bacteria, inhibiting LPS production, and reducing inflammation.
More importantly, some beneficial bacterial genera (e.g., Alloprevotella, Lachnospiraceae_ NK4A136_group, Butyricicoccus, Bifidobacterium, and Akkermansia) were increased by SFN in HFHFr-fed mice. Alloprevotella is a short-chain fatty acid-producing (SCFA) bacterium and improved NAFLD by dietary intervention in rodents [55]. The Lachnospiraceae _NK4A136_ group are butyrate-producing bacteria, which protect the intestinal mucosa and reduce host inflammation [56]. Bifidobacterium is an important health-promoting gut bacteria. It has been reported to improve liver inflammation in NAFLD and upregulate colon-tight junction genes to improve intestinal inflammation [57]. Butyricicoccus, a microorganism that primarily colonizes the associated surfaces of the colonic mucosa, contributes to the host's metabolite butyrate production, which is essential for maintaining cellular homeostasis [58]. Consistent with existing studies on SFN improving Butyricicoccus in the gut microbiota in DSS-induced colitis [59], our results showed that SFN improved gut epithelium. The correlation analysis showed that Butyricicoccus was the key beneficial bacterium in the gutliver axis regulated by the SFN intervention, and we hypothesized that its improvement potential relates to its butyrate-producing ability. Butyrates act as a colonic energy source for cells protecting the integrity of the intestinal epithelial barrier and therefore has antiinflammatory effects. Subsequently, the beneficial role of Butyricicoccus in NAFLD deserves further exploration. Therefore, SFN may protect the gut microbiota of NAFLD mice induced by the HFHFr diet by increasing the abundance of beneficial genera.
Most of the existing studies on the protective effect of SFN on intestinal inflammation have focused on DSS-induced colitis. These studies claimed that SFN protects against colitis by reducing the expression of intestinal mucosal inflammatory biomarkers and increasing the expression of NRF2-dependent genes [60]. Additionally, SFN significantly ameliorated the BBN-induced increase in intestinal permeability and disruption of the intestinal barrier in bladder cancer by increasing tight junction protein expressions [61]. Impaired intestinal epithelial barrier function leads to intestinal inflammation and increased intestinal permeability [62]. For the first time, we figured out that the SFN intervention protects intestinal barrier functions in NAFLD mice. The expression of GPR78, an ERS marker, and pro-apoptotic proteins in the jejunum and ileum was increased by the HFD diet [63]. Previous studies have shown that IRE1α directly interacts with TRAF2 in the cytoplasm, thereby activating the JNK pathway, and integrating ER protein folding interference with inflammatory and apoptosis signaling pathways [64]. Our results showed that these ERS-related genes were upregulated in the colon of HFHFr mice and were decreased by SFN treatment, indicating that SFN protected intestinal barrier integrity by regulating ERS-related signal transduction.
In this study, we highlighted that the accumulation of gut-derived LPS in the blood may lead to liver injury and gut inflammation by triggering an inflammatory response. Existing studies have reported that abnormally translocated LPS from enteric pathogens binds to TLR4 and activates the transcription factor NF-κB, which is responsible for inflammatory cytokine synthesis, ultimately leading to liver injury and hepatic steatosis [65]. We found that TLR4 (a cell surface pattern recognition receptor) was significantly increased in the livers and colons of HFHFr-diet mice. The expression of TLR4 increased in hepatic steatosis individuals and is associated with raising serum LPS levels and intestinal barrier disruption [66], suggesting that the LPS/TLR4 pathway is one of the major mechanisms acting on the gut-liver axis. For example, luteolin inhibits the TLR4 signaling pathway in the liver, thereby reducing the secretion of pro-inflammatory factors and improving HFD-induced NAFLD [67]. Likewise, the anti-inflammatory activity of green tea extract against NAFLD is mediated in a TLR4-dependent manner and prevents NASH-induced inflammation by limiting gut-derived LPS translocation and TLR4 activation [68]. TLR4, MyD88, and NF-κB p65 protein levels were found to be elevated in the livers and colons of NAFLD mice and improved after SFN treatment. In conclusion, we observed that SFN dietary intervention reduced lipid accumulation and improved hepatic steatosis in the HFHFr mice. More precisely, it reduced inflammatory cytokines by inhibiting the hepatic TLR4/NF-κB pathway and increased the anti-inflammatory factors production.
Additionally, the main function of SFN in the body is that it is easy to interact with the thiol group of the protein to form a stable covalent bond, but this bond is not dosedependent [69]. This may be an important reason why high-dose SFN did not improve hepatic and intestinal inflammation in the NAFLD model as well as low-dose SFN in this study.

Conclusions
In conclusion, this study discovers that long-term SFN supplementation improves gut microbial composition in the HFHFr diet-induced NAFLD mice. Notably, dietary intervention with SFN reduces the F/B ratio; improved Alloprevotella, Lachnospiraceae_NK4A136_group, Butyricicoccus, and Bifidobacterium (important for maintaining gut barrier integrity); and inhibits the proliferation and growth of Lactobacillus, Lactococcus, Blautia, Dubosiella, and Alistipes. Additionally, SFN enhances the tight junction proteins in the colon, inhibits the LPS/TLR4 signaling pathway and endoplasmic reticulum stress in the intestine, and improves intestinal inflammation. As a consequence, it maintains the intestinal barrier integrity, reduces the translocation of intestinal-derived LPS, and inhibits the liver LPS/TLR4 signaling pathway to improve hepatic steatosis and steatohepatitis (Figure 7). Nonetheless, this study has the limitation of not validating the effect of SFN against NAFLD in vitro. In summary, we highlight that SFN prevents the occurrence of NAFLD by modulating the gut microbiota in mice and inhibiting the LPS-TLR4-NF-κB inflammatory pathway, implying that SFN is a potential natural product for the prevention and treatment of NAFLD. inflammatory pathway, implying that SFN is a potential natural product for the prevention and treatment of NAFLD.

Data Availability Statement:
The data supporting the findings of this study are available from the corresponding author upon reasonable request.