Specific Strains of Faecalibacterium prausnitzii Ameliorate Nonalcoholic Fatty Liver Disease in Mice in Association with Gut Microbiota Regulation

Evidence linking Faecalibacterium prausnitzii abundance to nonalcoholic fatty liver disease (NAFLD) is accumulating; however, the causal relationship remains obscure. In this study, 12 F. prausnitzii strains were orally administered to high fat diet fed C57BL/6J mice for 12 weeks to evaluate the protective effects of F. prausnitzii on NAFLD. We found that five F. prausnitzii strains, A2-165, LB8, ZF21, PL45, and LC49, significantly restored serum lipid profiles and ameliorated glucose intolerance, adipose tissue dysfunction, hepatic steatosis, inflammation, and oxidative stress in a mouse model of NAFLD. Moreover, two strains, LC49 and LB8, significantly enhanced short-chain fatty acid (SCFA) production and modulated the gut microbiota. Based on the combined analysis of linear discriminant analysis effect size and microbial communities, the core microbiome related to NAFLD comprised Odoribacter, Roseburia, Erysipelatoclostridium, Tyzzerella, Faecalibaculum, Blautia, and Acetatifactor, and the last five genera can be reversed by treatment with the LC49 and LB8 strains. Additionally, the LC49 and LB8 strains enriched Lactobacillus, Ileibacterium, Faecalibacterium, Dubosiella, and Bifidobacterium and downregulated pathways involving carbohydrate metabolism, amino acid metabolism, and fatty acid biosynthesis. Interestingly, LC49 supplementation also upregulated tryptophan metabolism, glutathione metabolism, and valine, leucine, and isoleucine degradation, which might be related to NAFLD prevention. Collectively, F. prausnitzii LC49 and LB8 exerted considerable anti-NAFLD and microbiota-regulating effects, indicating their potential as probiotic agents for NAFLD treatment.


Introduction
NAFLD, the most common liver disorder characterized by fat accumulation in the liver without alcohol consumption, affects approximately a quarter of the world's population [1]. NAFLD is considered a growing indicator for liver transplantation by 2030, substantially increasing the public health burden worldwide [2]. NAFLD is asymptomatic in most affected patients and is closely linked with an increased risk of hepatocellular carcinoma (HCC), type II diabetes, and cardiovascular disease [3,4]. To date, there are no approved drugs for NAFLD treatment, and the main clinical recommendation is weight loss with dietary changes and exercise [5].

F. prausnitzii Strains and Culture Condition
Except for F. prausnitzii A2-165 (obtained from DSMZ), all strains were isolated from healthy human feces and preserved at the Jiangnan University Culture Collection of Food Microorganisms. All the participants signed an informed consent form. Detailed information on the F. prausnitzii strains is listed in Table S1. The bacterial suspensions used for oral administration were obtained as previously described [23].

Animal Experiments
Seven-week-old male C57BL/6J mice (20-22 g, SPF grade) were provided by Beijing Vital River Laboratory Animal Technology (Beijing, China). The mice were kept at a controlled temperature (23 ± 2 • C) and humidity (55 ± 10%) under a regular 12 h lightdark cycle. All experimental procedures were approved by the Animal Care and Use Committee of the Jiangnan University (JN. No. 2020930c1120130[234]).

Oral Glucose Tolerance Test (OGTT)
Before the OGTT, the mice fasted overnight. Blood samples collected from the tail vein's tip were tested for glucose levels at baseline (t = 0), just prior to oral gavage of 20% (w/v) glucose (2 g/kg body weight), and then after 30, 60, 90, and 120 min. Blood glucose levels were measured using a glucometer (Accu Check; Roche, Basel, Switzerland). The insulin content was determined using an ELISA kit (CUSABIO, Wuhan, China).

SCFA Metabolism
Feces were freeze-dried and weighed, followed by SCFA analysis according to Mao et al. [24].

Metagenomic Analysis
Total DNA from stool samples was obtained using a Fast DNA Stool Kit (MP Biomedicals, CA, USA) and applied to amplify the 16s rDNA V3-V4 region. The amplicons were purified and sequenced using the Illumina Miseq platform (Illumina, Santiago, CA, USA). The detailed method of gut microbiota analysis is described in the Supplementary Materials.

Statistical Analysis
SPSS 22.0 (IBM Corporation, Somers, NY, USA) was applied to perform all the statistical analyses. All values are expressed as means ± SD. Unpaired Student's t-tests were performed between the NC and HFD groups, and a one-way ANOVA was performed to compare the effects of F. prausnitzii in HFD-treated animals, followed by Dunnett's multiple comparison test against the HFD group. p < 0.05 was considered statistically significant.

F. prausnitzii Supplementation Decreased HFD-Induced Weight Gain and Hyperlipidemia in HFD-Fed Mice
To examine the causality between NAFLD and F. prausnitzii, 12 F. prausnitzii strains (including the reference strain A2-165 and 11 F. prausnitzii isolates) were orally administered to HFD-fed mice for 12 weeks. At the first week, the body weight gain of mice in the HFD group began to be remarkedly higher than that of the mice in the NC group ( Figure 1B; p < 0.01). However, F. prausnitzii treatment, except for the JZ10, JZ27, SM10, and HW29 strains, significantly reduced the body weight gain at 12 weeks of feeding compared to that of the HFD-fed mice without treatment ( Figure 1C; p < 0.05-0.01). No statistically significant differences were observed in the energy intake per mouse between the HFD and F. prausnitzii-treated groups ( Figure 1D; p > 0.05). Moreover, supplementation with F. prausnitzii strains, including A2-165, LB8, ZF21, PL45, and LC49, remarkably reduced serum FFAs, TC, TG, and LDL-C concentrations and elevated HDL-C content compared to the mice in the HFD group ( Figure 1E-I; p < 0.01-0.05).

F. prausnitzii Supplementation Ameliorated Adipose Tissue Dysfunction and Glucose Intolerance in HFD-Fed Mice
Adipocyte hypertrophy and inflammation are hallmarks of adipose tissue dysfunction. Therefore, the epididymal fat index, adipocyte size, and inflammatory cytokine-related gene expression were determined in this study. As shown in Figure 2A, the epididymal fat index in the HFD group was notably increased compared to the NC group (p < 0.01). However, the F. prausnitzii-treated groups, except for the JZ10, JZ27, QL33, and HW29 groups, showed a remarkably lower fat index than the HFD group (p < 0.01-0.05). Moreover, the HFDinduced adipocyte expansion and inflammation were also reversed following treatment with the F. prausnitzii strains, except for JZ10, JZ27, QL33, and HW29 ( Figure 2B, C; p < 0.01). Similarly, mice treated with F. prausnitzi showed lower mRNA expression of TNF-α and IL-6 than those of the HFD group, which was in accordance with the results of the adipose tissue dysfunction (Table S2; Figure S2). Furthermore, we evaluated the effects of F. prausnitzii treatment on glucose intolerance in HFD-fed mice. Long-term HFD-fed mice showed significantly higher fasting blood glucose (FBG), fasting insulin, and HOMA-IR values than NCD-fed mice (Figure 2D-F; p < 0.01). However, the values of these three indicators were reversed by the F. prausnitzii treatment (p < 0.01-0.05), except for the JZ10, JZ27, QL33, and HW29 strains. Likewise, impaired glucose tolerance was also restored in mice treated with these F. prausnitzii strains ( Figure 2G,H; p < 0.01) compared with the mice in the HFD group.

F. prausnitzii Supplementation Prevented Liver Injury and Hepatic Steatosis in HFD-Fed Mice
Excess dietary fat promotes excessive fat deposition in the liver, which causes hepatic steatosis, leading to oxidative stress and inflammatory reactions in the hepatocytes [25]. The serum AST and ALT levels are commonly considered sensitive indicators of liver injury. The HFD remarkedly elevated the serum ALT and AST levels in the mice, whereas treatment with F. prausnitzii effectively reversed the levels of these two indicators ( Figure 3A,B; p < 0.01-0.05), except for the QL13, QL33, SM10, and HW29 strains. Moreover, the liver indices in the A2-165, LB8, PL45, and LC49 groups were markedly decreased compared with the HFD group ( Figure 3C; p < 0.01-0.05). Regarding the liver pathology of the HFDfed mice, the structure of hepatic lobules was distorted, and the area of intracytoplasmic lipid droplets accounted for >50% of the field of view, accompanied by inflammatory cell infiltration in the hepatic lobules ( Figure 3D-F). Strikingly, F. prausnitzii treatment, except for the HW29 strain, significantly improved the hepatic pathology, which was confirmed by the quantification of the fibrotic area and lipid area per total area (Figures 3G and S3; p < 0.01). Data are shown as means with SD; ## p < 0.01 vs. NC group, unpaired Student's t-test; * p < 0.05, ** p < 0.01 vs. HFD group, one-way ANOVA followed by Dunnett's test.
To evaluate the hepatic oxidative stress and inflammatory reactions, the SOD and GSH-PX activities and MDA, TNF-α, IL-1β, and IL-6 levels were determined. As shown in Figure 3H-M, the HFD group presented a high activation of hepatic oxidative stress and inflammatory processes, which was reflected by the significantly increased levels of TNF-α, IL-1β, IL-6, and MDA as well as decreased SOD and GSH-PX activities. Among the liver indices in the A2-165, LB8, PL45, and LC49 groups were markedly decreased com pared with the HFD group ( Figure 3C; p < 0.01-0.05). Regarding the liver pathology of th HFD-fed mice, the structure of hepatic lobules was distorted, and the area of intracyto plasmic lipid droplets accounted for >50% of the field of view, accompanied by inflamma tory cell infiltration in the hepatic lobules ( Figure 3D-F). Strikingly, F. prausnitzii treat ment, except for the HW29 strain, significantly improved the hepatic pathology, which was confirmed by the quantification of the fibrotic area and lipid area per total area (Fig ures 3G and S3; p < 0.01).

F. prausnitzii Supplementation Regulated the Gut Microbial Composition in HFD-Fed Mice
Furthermore, we determined the effect of F. prausnitzii strains on the gut microbial composition based on the 16s-RNA-amplicon sequencing. F. prausnitzii treatment did not change the α-diversity among the NC, HFD, and F. prausnitzii-treated groups, with no significant differences observed in the Shannon index and observed species ( Figure 4A,B). However, the distinction in β-diversity between the NC and HFD groups was evident in the PCA analysis ( Figure 4C). Among the 12 F. prausnitzii strains, only the LC49 group was close to the NC group, indicating the restorative effect of LC49 on HFD-induced gut microbial dysbiosis. The dominant phyla in all groups were Firmicutes, Bacteroides, and Actinobacteria ( Figure 4D). The HFD-fed group showed a significantly higher ratio of Firmicutes/Bacteroides compared to the NCD-fed mice ( Figure 4E; p < 0.01), whereas LC49 treatment normalized this ratio (p < 0.05).

F. prausnitzii Supplementation Changed Metagenomic Functions and SCFA Production in HFD-Fed Mice
To investigate the impact of F. prausnitzii supplementation on the metagenomic functions of gut microbiota, a PICRUSt analysis was performed. Based on Welch's test, 17 significantly different pathways were observed between the NC and HFD groups ( Figure 5A; p < 0.01-0.05). The comparisons among all groups showed that pathways involving riboflavin metabolism; glutathione metabolism; valine, leucine, and isoleucine (BCAAs) degradation; tryptophan metabolism; the citrate cycle; glycine, serine, and threonine metabolism; folate biosynthesis; alanine, aspartate, and glutamate metabolism; and one carbon pool by folate were remarkedly downregulated in the HFD group, and the first one was reversed by the SM10, LB8, ZF21, and LC49 treatments. Moreover, glutathione metabolism, BCAA degradation, tryptophan metabolism, and the citrate cycle were also restored in the LC49 group, and the first three were even higher than those in the NC group. The rest of the 17 differential pathways were upregulated in the HFD group; the changes in pentose and glucuronate interconversions and glycerolipid metabolism were reversed by the QL33, LB8, ZF21, and LC49 strains. Additionally, LC49 treatment downregulated the pathways involved into the pentose phosphate pathway; phenylalanine, tyrosine, and tryptophan biosynthesis; carbohydrate metabolism; and cysteine and methionine metabolism.
Next, alterations in SCFA production were determined to explore the impact of F. prausnitzii treatment on gut microbial metabolism. The HFD group showed remarkably lower levels of acetate, propionate, and butyrate than the NC group (Figure 5C-E; p < 0.01). However, the LC49 and LB8 strain treatments notably elevated the content of acetate, propionate, and butyrate compared with the HFD group. Furthermore, no statistically significant difference was observed in iso-butyrate levels among the NC, HFD, and F. prausnitzii-treated groups ( Figure 5F; p > 0.05). The right panel presents the differential taxa labeled with various tags; (G) The relative abundance of taxa remarkedly changed by HFD feeding and F. prausnitzii treatment. The size and color of the circles display the average abundance and p value, respectively; (H) Interaction network analysis among microbial biomarkers. The circle size indicates the contribution, and the line color represents a positive (red) or negative correlation (blue), respectively (p < 0.05, r > 0.3 or r < −0.3). Data are shown as means with SD; # p < 0.05, ## p < 0.01 vs. NC group, unpaired Student's t-test; * p < 0.05, ** p < 0.01 vs. HFD group, one-way ANOVA followed by Dunnett's test. The right panel presents the differential taxa labeled with various tags; (G) The relative abundance of taxa remarkedly changed by HFD feeding and F. prausnitzii treatment. The size and color of the circles display the average abundance and p value, respectively; (H) Interaction network analysis among microbial biomarkers. The circle size indicates the contribution, and the line color represents a positive (red) or negative correlation (blue), respectively (p < 0.05, r > 0.3 or r < −0.3). Data are shown as means with SD; # p < 0.05, ## p < 0.01 vs. NC group, unpaired Student's t-test; * p < 0.05, ** p < 0.01 vs. HFD group, one-way ANOVA followed by Dunnett's test. Differential pathways between the NC and HFD groups (two-sided Welch's t-test at p < 0.05); (B) Heatmap of the differential pathways in (A) among all groups; (C-F) Colonic acetate, propionate, butyrate, and iso-butyrate concentrations. Data are shown as means ± SD; # p < 0.05, ## p < 0.01 vs. NC group, unpaired Student's t-test; * p < 0.05, ** p < 0.01 vs. HFD group, one-way ANOVA followed by Dunnett's test.
Next, alterations in SCFA production were determined to explore the impact of F. prausnitzii treatment on gut microbial metabolism. The HFD group showed remarkably lower levels of acetate, propionate, and butyrate than the NC group (Figure 5C-E; p < 0.01). However, the LC49 and LB8 strain treatments notably elevated the content of acetate, propionate, and butyrate compared with the HFD group. Furthermore, no statistically significant difference was observed in iso-butyrate levels among the NC, HFD, and F. prausnitzii-treated groups ( Figure 5F; p > 0.05).

Correlation Analysis between the Phenotypes of NAFLD and Multiple Indicators in the Gut
A correlation analysis was carried out to investigate the relation between the phenotypes of NAFLD (HOMA-IR and biopsy score), microbial biomarkers, and three highly predicted functional pathways in the LC49 group. As presented in Figure 6, the HOMA-IR and biopsy score were significantly positively related to the relative abundance of Figure 5. Prediction of microbial functional pathways and short-chain fatty acid production. (A) Differential pathways between the NC and HFD groups (two-sided Welch's t-test at p < 0.05); (B) Heatmap of the differential pathways in (A) among all groups; (C-F) Colonic acetate, propionate, butyrate, and iso-butyrate concentrations. Data are shown as means ± SD; # p < 0.05, ## p < 0.01 vs. NC group, unpaired Student's t-test; * p < 0.05, ** p < 0.01 vs. HFD group, one-way ANOVA followed by Dunnett's test.

Correlation Analysis between the Phenotypes of NAFLD and Multiple Indicators in the Gut
A correlation analysis was carried out to investigate the relation between the phenotypes of NAFLD (HOMA-IR and biopsy score), microbial biomarkers, and three highly predicted functional pathways in the LC49 group. As presented in Figure 6, the HOMA-IR and biopsy score were significantly positively related to the relative abundance of Tyzzerella, Faecalibaculum, Erysipelatoclostridium, Blautia, Bilophila, Anaerotruncus, and Acetatifacto and negatively correlated with the relative abundance of Lactobacillus, Dubosiella, Faecalibacterium, and Enterorhabdus as well as the fecal levels of acetate, propionate, and butyrate. Additionally, the HOMA-IR and biopsy scores were negatively associated with glutathione metabolism, BCAA degradation, and tryptophan metabolism.
Tyzzerella, Faecalibaculum, Erysipelatoclostridium, Blautia, Bilophila, Anaerotruncus, and Acetatifacto and negatively correlated with the relative abundance of Lactobacillus, Dubosiella, Faecalibacterium, and Enterorhabdus as well as the fecal levels of acetate, propionate, and butyrate. Additionally, the HOMA-IR and biopsy scores were negatively associated with glutathione metabolism, BCAA degradation, and tryptophan metabolism. The color and size of the circles in the matrix code indicate the correlation level. The thickness, color, and type (solid or dot) of the lines represent the correlation level, correlation type, and p value level, respectively. The correlation analyses in intra-matrix or inter-matrix are based on Pearson's correlation coefficient and Mantel's test, respectively. BCAAs, valine, leucine, and isoleucine degradation; HOMA-IR, homeostatic model assessment for insulin resistance; Biopsy score indicates the sum of the hepatic steatosis grade and the inflammation grade.

Discussion
The human gastrointestinal tract harbors trillions of microbes, which is approximately 10 times the number of all cells of the human body [26]. Undoubtfully, the gut microbiota is crucial for the host's energy metabolism, thus affecting metabolic syndrome [27]. Numerous studies reported an apparently decreased F. prausnitzii abundance in NAFLD patients compared with healthy individuals [28][29][30]. Since F. prausnitzii is a dominant commensal bacterium in the human gut, we proposed a hypothesis on a potential causative link between F. prausnitzii and NAFLD. Thus, we examined the hepatoprotective effects of 12 human-derived F. prausnitzii strains on HFD-induced NAFLD. Our results revealed that F. prausnitzii strain-dependently ameliorated NAFLD phenotypes. The color and size of the circles in the matrix code indicate the correlation level. The thickness, color, and type (solid or dot) of the lines represent the correlation level, correlation type, and p value level, respectively. The correlation analyses in intra-matrix or inter-matrix are based on Pearson's correlation coefficient and Mantel's test, respectively. BCAAs, valine, leucine, and isoleucine degradation; HOMA-IR, homeostatic model assessment for insulin resistance; Biopsy score indicates the sum of the hepatic steatosis grade and the inflammation grade.

Discussion
The human gastrointestinal tract harbors trillions of microbes, which is approximately 10 times the number of all cells of the human body [26]. Undoubtfully, the gut microbiota is crucial for the host's energy metabolism, thus affecting metabolic syndrome [27]. Numerous studies reported an apparently decreased F. prausnitzii abundance in NAFLD patients compared with healthy individuals [28][29][30]. Since F. prausnitzii is a dominant commensal bacterium in the human gut, we proposed a hypothesis on a potential causative link between F. prausnitzii and NAFLD. Thus, we examined the hepatoprotective effects of 12 human-derived F. prausnitzii strains on HFD-induced NAFLD. Our results revealed that F. prausnitzii strain-dependently ameliorated NAFLD phenotypes.
Dyslipidemia raises the risk of NAFLD and is closely related to cardiovascular mortality, which is the primary cause of death in NAFLD cases [31]. Approximately 20-80% of NAFLD patients have dyslipidemia, which is manifested by increased serum TC, TG, and LDL-C, and decreased HDL-C [32]. Here, we found that treatment with F. prausnitzii strains reversed dyslipidemia symptoms in the NAFLD mice, but strain-specific differences were observed ( Figure 1E-I). Furthermore, F. prausnitzii supplementation reduced adipose tissue inflammation and expansion (Figures 2A-C and S2). The adipose tissue expansion is concomitant with excessive lipolysis and the subsequently increased FFAs levels, causing dyslipidemia [33]. Therefore, the restoration of dyslipidemia and adipose tissue dysfunction indicate the beneficial effect of specific F. prausnitzii strains (A2-165, LB8, ZF21, PL45, and LC49) on NAFLD.
Earlier, the traditional "two-hit" model was commonly accepted for illustrating the pathogenesis of NAFLD [34]. Steatosis, mainly caused by insulin resistance, indicates the "first hit", sensitizing the liver to the steatohepatitis and fibrosis resulting from "second hits", such as oxidative stress, adipokines, inflammatory cytokines, and mitochondrial dysfunction [35]. Along with the experimental and theoretical developments, a "multiple-hit" theory was recently proposed. According to the "multi-hit" theory, NAFLD is induced by a combination of insults, including adipose-tissue-secreted hormones, insulin resistance, dietary factors, gut flora, and genetic and epigenetic factors [36]. Of note, liver fat accumulation still seems to represent the "first hit". Our results indicate that F. prausnitzii supplementation, including A2-165, LB8, ZF21, PL45, and LC49, notably alleviated liver injury and hepatic steatosis, inflammation, and oxidative stress (Figures 3 and S3). Furthermore, we demonstrated that F. prausnitzii treatments, except the JZ10, JZ27, QL33, and HW29 strains, significantly ameliorated the glucose intolerance caused by long-term feeding with an HFD (Figure 2D-H). The PCA analysis based on all NAFLD phenotypes suggested the beneficial effects of F. prausnitzii on NAFLD depend on the specific strain, with LC49 being the best and HW29 being the least beneficial strain ( Figure S1).
The gut microbiota, as a major node for the gut-liver axis, provides a direct target for probiotics to exert beneficial health effects [37,38]. Here we used 16s rRNA-amplicon sequencing to perform a comprehensive analysis of the gut microbiota. We observed no statistically significant differences in α-diversity between the NC, HFD, and F. prausnitziitreated groups, indicating that an HFD might affect the overall composition of intestinal flora ( Figure 4A,B) but not the richness. The β-diversity analysis indicated an apparent structural separation between the HFD and NC groups, and only the LC49 group was close to the NC group ( Figure 4C). Moreover, long-term intervention with an HFD notably impacted the gut microbial composition at the phylum level, resulting in a higher Firmicutes/Bacteroides ratio compared to mice fed an NCD ( Figure 4D,E; p < 0.01). This alteration in the Firmicutes/Bacteroides ratio caused by HFD feeding was reversed by F. prausnitzii LC49 but was aggravated by the JZ27, LA8, ZF21, and HW29 strains. Based on the results of the LEfSe analysis, Tyzzerella, Roseburia, Odoribacter, Erysipelatoclostridium, and Faecalibaculum were found to be the core microbiome in the HFD groups ( Figure 4F), which is largely in agreement with previous studies [39][40][41][42]. The HOMA-IR and biopsy scores were significantly positively associated with the relative abundance of Tyzzerella (r = 0.39, p < 0.01; r = 0.45, p < 0.01), Erysipelatoclostridium (r = 0.42, p <0.01; r = 0.40, p < 0.05), and Faecalibaculum (r = 0.34, p <0.01; r = 0.27, p < 0.05), indicating that these three HFDdependent taxa might be crucial for the pathogenesis of NAFLD ( Figure 6). Faecalibaculum ingestion might cause depression-like phenotypes [43], whereas an increased abundance of Erysipelatoclostridium is strongly related to metabolic disorders [44]. A previous cohort study addressed the negative effects of Tyzzeralla on CVD [40]. F. prausnitzii strains specifically normalized these HFD-dependent microbial biomarkers, with the LC49 and LB8 strains performing the best. In addition, LC49 and LB8 promoted the bloom of Lactobacillus, which was significantly negatively correlated with Tyzzerella and Faecalibaculum ( Figure 4H; r < −0.3, p < 0.01). Lactobacillus is reported to be beneficial for the prevention of NAFLD [17,45]. Interestingly, the Bifidobacterium abundance in F. prausnitzii-treated groups was remarkedly increased compared with the NC and HFD groups ( Figure 4G), which can be explained by the potential cross-feeding relationship between F. prausnitzii and Bifidobacterium [46]. Only LC49 treatment significantly enriched Dubosiella and Intestinimonas ( Figure 4G, p < 0.01), which are reported to be potential therapeutics for metabolic syndrome and diabetes mellitus [47]. Unexpectedly, four F. prausnitzii-treated groups exhibited a comparable abundance of Faecalibacterium with the HFD group, suggesting that human-derived F. prausnitzii strains might adapt the mouse gut in a strain-specific manner (Figures 4G and S4). The different colonization efficiencies of F. prausnitzii strains might be attributed to the strain-dependent competition with the host microbiota for nutrients and adhesion sites [48]. These results from the gut microbiota analysis demonstrated that F. prausnitzii LB8 and LC49 restored the HFD-induced taxa to normal status, along with the increased abundance of genera that is beneficial for NAFLD prevention.
To further investigate the effects of F. prausnitzii treatment on the functional capabilities of the microbial community, a PICRUSt analysis was performed. Differential pathways between the NC and HFD groups were mainly involved in carbohydrate, lipid, and amino acid metabolism, and most abnormal alterations in NAFLD mice were reversed by LC49 and LB8 treatments. Interestingly, pathways involving glutathione metabolism, BCAA degradation, and tryptophan metabolism were more highly upregulated in the LC49 group than in the NC group. Glutathione is an important cellular free radical scavenger, whereas BCAAs act as important mediators of metabolic health [49,50]. It has been demonstrated that microbiota-derived tryptophan metabolites, such as indole-3-acetate, tryptamine, and indole-3-carbinol, can reduce the pathology of NAFLD through the aryl hydrocarbon receptor (AHR) pathway [51,52]. The results of the functional pathways suggested that the strain-specific effect of F. prausnitzii on NAFLD might be due to certain metabolites derived from the gut microbiota. Auger et al. demonstrated that the different anti-inflammatory properties of F. prausnitzii are attributed to the intraspecific diversity of its metabolite, microbial anti-inflammatory molecule (MAM) [53]. Therefore, an alteration of gut microbial functional pathways might be associated with the beneficial effect of the LB8 and LC49 strains on NAFLD.
Microbiota-derived SCFAs play crucial roles in maintaining hepatic energy homeostasis. SCFAs can affect appetite and calorie consumption by stimulating the secretion of peptide YY (PYY), glucagon-like peptide (GLP-1), and leptin [3]. Moreover, SCFAs may have a positive impact on NAFLD by affecting energy expenditure and lipid metabolism. An oral treatment with propionate or butyrate increases energy expenditure and lipid oxidation [54,55], which is potentially beneficial for fat accumulation in the liver. Moreover, acetate supplementation prevents adiposity by promoting adipose tissue browning [56]. We found that treatment with F. prausnitzii LC49 and LB8 notably increased the content of acetate, propionate, and butyrate compared with the HFD group ( Figure 5C-F), indicating that gut microbiota-derived SCFAs were potential mediators for LB8 and LC49 strains exerting an anti-NAFLD effect.
Based on a comprehensive analysis of NAFLD phenotype hallmarks, SCFA metabolism, and gut microbial analysis, the capacity of F. prausnitzii strains to prevent HFD-induced NAFLD were found to be varied. Given the different capabilities of producing butyrate among these F. prausnitzii strains and the significant negative correlation between butyrate and two representative NAFLD indicators (HOMA-IR and biopsy score, Figure 6), we considered the butyrate-producing ability as the main reason for the strain-specific beneficial effect of F. prausnitzii on NAFLD. However, our previous work showed that the HW29 strain produced higher butyrate levels than ZF21 [57], but the protective effect on NAFLD was not comparable to that of ZF21 ( Figure S1). In addition, LC49, the most effective strain, showed a significantly higher enrichment in pathways involving tryptophan metabolism, glutathione metabolism, and BBCA degradation than the HFD group ( Figure 5B). Accordingly, the combined effect of SCFAs and other potential microbiota-derived metabolites might be responsible for the protective effect of F. prausnitzii strains on NAFLD, and the involvement of the gut microbiota is critical, as indicated by the results from the correlation analysis ( Figure 6).
In summary, our results demonstrated that F. prausnitzii LB8 and LC49 significantly ameliorated the symptoms associated with a mouse model of NAFLD, restored the gut microbial dysbiosis, and modulated the gut microbial functional pathways and SCFA production. Nevertheless, the detailed mechanisms underlying the microbiota-derived SCFAs, tryptophan metabolites, or other substances as key mediators of the anti-NAFLD effect of the LB8 and LC49 strains require further investigation. Collectively, the findings of this study indicate the potential of the LB8 and LC49 strains as probiotic agents for NAFLD prevention.