Gut microbiome determines therapeutic effects of OCA on NAFLD by modulating bile acid metabolism

Non-alcoholic fatty liver disease (NAFLD), the most common chronic liver disease, had no approved pharmacological agents yet. Obeticholic acid (OCA), a novel bile acid derivative, was demonstrated to ameliorate NAFLD-related manifestations. Regarding the role of gut-liver axis in liver disease development, this study aimed to explore the potential role of gut microbiota in the treatment of OCA in NAFLD mice induced by the high-fat diet (HFD). Antibiotic-induced microbiome depletion (AIMD) and fecal microbiota transplantation (FMT) confirmed the critical role of gut microbiota in OCA treatment for NAFLD by effectively alleviating histopathological lesions and restoring liver function impaired by HFD. Metagenomic analysis indicated that OCA intervention in HFD mice remarkably increased the abundance of Akkermansia muciniphila, Bifidobacterium spp., Bacteroides spp., Alistipes spp., Lactobacillus spp., Streptococcus thermophilus, and Parasutterella excrementihominis. Targeted metabolomics analysis indicated that OCA could modulate host bile acids pool by reducing levels of serum hydrophobic cholic acid (CA) and chenodeoxycholic acid (CDCA), and increasing levels of serum-conjugated bile acids, such as taurodeoxycholic acid (TDCA) and tauroursodesoxycholic acid (TUDCA) in the HFD-fed mice. Strong correlations were observed between differentially abundant microbes and the shifted bile acids. Furthermore, bacteria enriched by OCA intervention exhibited much greater potential in encoding 7alpha-hydroxysteroid dehydrogenase (7α-HSDs) producing secondary bile acids rather than bile salt hydrolases (BSHs) mainly responsible for primary bile acid deconjugation. In conclusion, this study demonstrated that OCA intervention altered gut microbiota composition with specially enriched gut microbes modulating host bile acids, thus effectively alleviating NAFLD in the mice.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is a comprehensive clinical disease characterized by lipid accumulation in hepatocytes that encompasses simple steatosis, non-alcoholic steatohepatitis (NASH), and hepatic brosis and cirrhosis. 1NAFLD is one of the most prevalent chronic liver disorders worldwide 2 , found in 10-40% of adults.In China, NAFLD is surrogating hepatitis B to become the predominant cause of chronic liver disorders. 3NAFLD not only has a close relationship with insulin resistance, but also potentially results from metabolic syndromes as well as obesity. 4It is well known that gut-liver axis plays an important role in the pathogenesis of hepatic injuries like alcoholic liver disease, NAFLD, and hepatocellular carcinoma. 5,6Despite caloric restriction, body weight reduction, and the medicine for mitigating insulin resistance and hyperlipemia being considered as effective strategies of NAFLD 7 , there are no accepted FDA-approved pharmacotherapies for treating or preventing NAFLD.
Gut microbiota composition had been discovered essential to human health 8 , and its crucial functions were con rmed by pre-clinical NAFLD/NASH models and NASH patients 9,10 .In addition, gut microbiota and their metabolites can induce fat accumulation and in ammation in the liver via gut-liver axis 11 .
However, the underlying pathogenesis of steatohepatitis mediated by gut microbiome is poorly understood.Recently, several studies have demonstrated that many species in gut microbiota are associated with the progression of NAFLD 12 .For example, gut microbiota pro le of NAFLD patients showed higher abundance of Firmicutes and Proteobacteria, but a lower abundance of Bacteroidetes. 13urprisingly, some commensal bacterial products not only mediate the impact of gut microbiota pro le on the host, but also in uence the pathogenesis of NAFLD 14,15 , like Bi dobacteria, Bacills coli, and Verrucomicrobia 16 .Besides, the in ammatory response induced and metabolites (e.g.bile acids, lipopolysaccharide) produced by gut microbiota are related to NAFLD 17 .Nowadays, the modulation of intestinal microbial community is extensively applied in the treatment of diseases.Therefore, manipulating gut microbiota composition and replenishing commensal bacterial metabolites could be promising therapeutic approaches to NAFLD.
The chemical diversi cation of bile acids is the combined effort of both host and intestinal microbiota since primary bile acids synthesized by host can be converted into secondary bile acids by intestinal microbiota.It is hypothesized that gut microbiota in uences host physiology by means of metabolites production (for example, microbial-derived secondary bile acids), thus acting as "endocrine organ" 18 .
Evidence was put forward to show varied bile acids in different gut microbiota structures, and gut microbiota can modulate bile acid pools, therefore highlighting the interaction of gut-bile acid-host axis 19 .
Obeticholic acid (OCA), the 6α-ethyl derivative of bile acids and potent activator of the farnesoid X nuclear receptor, was rstly approved as the therapeutic trial of primary biliary cholangitis 20 , also considered to treat NAFLD/NASH 21 .OCA could effectively alleviate fat accumulation in liver, hepatic in ammation, and insulin resistance of NAFLD rodents [21][22][23] .However, a great knowledge gap still exists in whether gut microbiota in uences the therapeutic process of OCA administration on NALFD in consideration of the presence of gut-liver axis.Therefore, to decipherer the dynamic complexity of gut-bile acid-host axis during OCA treatment on NALFD, could probably provide a comprehensive insight into NALFD development and treatment.
Different kinds of bile salt hydrolases (BSHs) and hydroxysteroid dehydrogenases (HSDs) were implicated in the gut microbiota-mediated bile acid transformation.Microbial-encoded BSHs were responsible for hydrolyzing conjugated bile acids in the gut 24 .As critical mediators and gatekeepers of bile acid transformation, these enzymes regulated intestinal bile acid pro les.Therefore, BSHs were regarded as promising therapeutical targets 24,25 .Hydroxysteroidal dehydrogenases (HSDs) catalyze the oxidation/reduction of steroidal hydroxyl (-OH)/oxy groups.This type of reaction is not only a basic step in the biosynthesis of bile acids or vertebrate steroid hormones, but also plays a key role in maintaining intracellular receptor ligand levels through different tissue-speci c HSDs 26 .Recently, omics technology has provided a powerful tool for the studies against human diseases.Speci cally, metagenomics sequencing offered pivotal information on the interaction between gut microbiome and host 27 , while metabolomics enabled comprehensive analyses of the small-molecular metabolites in the life system.Thus, it is progressively performed to explore the impact of various biomarkers on disease discovery and uncover the biological process as well as fundamental mechanisms 28,29 .Furthermore, the integration of metagenomics and metabolomics is extensively applied to delineate the molecular mechanisms of the disease development 30 .Hence, the present study was conducted to decipher the molecular mechanisms of NAFD in mice by integrating metagenomics and metabolomics approaches.Our study not only greatly promote the application of microbial biomarkers for diagnostic of NALFD, but also lay a theoretical foundation for the therapeutic strategy of NALFD by gut microbiota intervention in the future.

Results
Gut microbiota plays a critical role in the treatment of OCA in NAFLD mice After a 2 weeks of antibiotic-induced microbiome depletion (AIMD), the body weight of high-fat-diet (HFD group) fed mice treated with following antibiotics intervention (HFD + A group) had a signi cant decrease (P < 0.05) compared with that of HFD-fed mice (HFD group).And HFD-fed mice treated with OCA administration (i.e., HFD + OCA group) also exhibited a signi cant decrease in body weight (P < 0.05) compared to HFD group.In addition, there was a group of HFD-fed mice that were given oral gavage of antibiotics and following OCA intervention (HFD + A + OCA group).Surprisingly, the body weight of mice in HFD + A + OCA group was continuously increased like HFD group (Fig. 1A, B).Also, no difference was observed in the food intake among the four groups of mice during this experiment (Fig. 1D).Meanwhile, hematoxylin-eosin (HE) staining of liver tissues of mice in HFD, HFD + A, and HFD + A + OCA groups had hepatocellular ballooning, and dramatically increased size of hepatocytes and nuclear marginalization (Fig. 1E).In oil red O staining of mouse liver tissues (Fig. 1F), lipid droplets were obviously visible in hepatocytes of mice in HFD, HFD + A, and HFD + A + OCA groups, indicating that liver fat accumulation was serious in these three groups.Additionally, HE staining of mouse adipose tissues showed that adipocyte size in HFD + A and HFD + A + OCA groups was similar with that in HFD group but was larger than that in HFD + OCA group (Fig. 1G).This phenomenon implied that OCA intervention could effectively ameliorate HFD-induced fat in ammation depending on the presence of gut microbiota.To sum up, gut microbiota played a critical role in the therapeutic effects of OCA administration by reducing hepatic fat accumulation and pathological variations of liver and adipose tissues induced by HFD in mice.
Apart from the improvement of histopathological phenotypes of liver and adipose tissues, HFD fed mice treated with OCA could effectively mitigate glucose intolerance and insulin resistance by decreasing the levels of blood glucose, and homeostatic model assessment for insulin resistance (HOMA-IR) and insulin, respectively (Supplementary Figure S1A-C).Besides, OCA apparently reduced the levels of total triglyceride (TG), total cholesterol (TC), and LDL cholesterol (LDL-C), while increasing the levels of HDL cholesterol (HDL-C) in serum compared with HFD fed mice (Supplementary Figure S1D-G).Moreover, higher levels of alanine aminotransferase (ALT) and aspartic transaminase (AST) in the serum were observed in OCA group compared to HFD group (Supplementary Figure S1H, I).To further con rm the role of gut microbiota played in the treatment of NAFLD by OCA, we performed fecal microbiota transplantation (FMT) from mice of the HFD + OCA group (donors) to the HFD group (recipients).Both body weight and the ratio of liver weight to body weight in FMT group were signi cantly lower than those in HFD group (P < 0.05) which resembled the effects of OCA treatment (Fig. 2A-C).No signi cant difference was observed in food intake among the three experimental groups (Fig. 2D).The histopathological examinations of liver and adipose tissues exhibited signi cant difference between mice of HFD and FMT.Despite some small vacuoles appeared in mouse hepatocytes of FMT group derived from lipid droplets (Fig. 2E), the amounts of large vacuoles and lipid droplets in hepatocytes were obviously decreased in FMT group compared with HFD group (Fig. 2F).And the size of adipocytes in FMT group was smaller than that in HFD group (Fig. 2G).Noticeably, the comparisons of levels of serum TG and LDL-C between FMT and HFD groups re ected that FMT exhibited similar effects in improving liver function like OCA treatment did (Supplementary Figure S2A, B).However, FMT did not exert adequate effects on levels of serum TC, HDL-C, and in ammatory factors like OCA administration (Supplementary Figure S2C-G).These results suggested that gut microbiota could participate in alleviating liver and adipose tissues damages caused by HFD, and improving liver function, thus determine the therapeutic effect of OCA administration on NALFD.

The in uence of OCA administration on gut microbiota composition of NAFLD mice
To investigate the speci c gut microbes involved in OCA treatment on NALFD, we further implemented metagenomic analysis on the mice fecal microbiota of ND, HFD, and OCA groups.For alpha diversity, Shannon index and Simpson index of gut microbiota had only consistent trends and no signi cant difference among the three groups of mice (Fig. 3A).However, OCA administration substantially improved the richness of gut microbiota in HFD fed mice.In addition, different Pielou indexes indicated that highest gut microbiota evenness in mice of normal diet (ND) group, while lowest in mice of HFD group.The slightly elevated Pielou index suggested a higher evenness of the gut microbiota in OCA group compared with HFD group.For the gut microbial community structures, both non-metric multidimensional scaling (NMDS) and principal co-ordinates analysis (PCoA) based on Bray-Curtis dissimilarity distance matrix at species level were implemented, the results exhibited signi cant difference in microbial communities among ND, HFD and HFD + OCA groups (adonis R 2 = 0.243, P = 0.001) (Fig. 3B-C).Therefore, gut microbiota structures of mice could be affected by both 20-week HFD feeding and HFD feeding with subsequently 8-week OCA administration.
Further exploration of gut microbial composition at different taxonomic levels, signi cant differences were found between HFD group and ND group, as well as OCA group and HFD group.At the phylum level, fecal microbiota mainly consisted of four phyla of bacteria: Firmicutes, Bacteroidetes, Proteobacteria, and Verrucomicrobia (Supplementary Figure S3).The comparisons of the proportions of these four kinds of bacteria in feces of mice among three groups demonstrated that HFD substantially increased the proportion of Firmicutes in mouse feces of ND group (P < 0.001, Wilcox test), while it decreased the proportions of Proteobacteria and Verrucomicrobia (P < 0.01, Wilcox test).In contrast, OCA not only profoundly reduced the proportion of Firmicutes, but also improved the proportions of Preteobacteria and Verrucomicrobia in HFD group.
At the genus level, compared with HFD group, the enriched bacteria of mouse gut in ND group were Akkermansia, Mucispirillum, Bi dobacteria, Bacteroides, Helicobacteria, and another three genera (Fig. 3D,  3F).And bacteria enriched in OCA group compared with HFD group were seven genera including Akkermansia, Bi dobacteria, Bacteroides, Escherichia, Lactococcus, and so on (Fig. 3E, 3G).At the species level (Fig. 3H, 3I), Akkermansia miciniphila, Bi dobacterium animalis, Bi dobacterium pseudolongum, Bacteroides massiliensis, Lactobacillus helveticus, Streptococcus thermophilus, Escherichia coli, and another ten species exhibited higher abundance (P < 0.05) in OCA group compared to HFD group.Also, OCA intervention signi cantly (P < 0.05) reduced the proportions of Lactobacillus murine, Firmicutes bacterium M10-2, Lactobacillus gasseri, and Enterococcus faecium of feces of HFD fed mice.Besides, compared with HFD group, ND group had higher abundance of A. muciniphila, Mucispirillum schaedleri, B. pseudolongum, Bacteroides uniformis, Bacteroides xylanisolvens, and lower abundance of eleven species including the same four species reduced by OCA administration.Thus, OCA administration recovered the gut microbial composition that was altered by HFD, and made it approximate to that in ND group.

The bile acid pro les and correlations between bile acid and gut microbiota
It is well-known that gut microbiota is tightly associated with bile acid metabolism, shifts in gut microbiota might in uence host bile acids pool.Indeed, targeted metabolomics analysis of bile acids indicated that both HFD and HFD with following OCA treatment could effectively modi ed serum bile acid pool in mice.Compared with ND group, HFD feeding profoundly increased (P < 0.05) the level of serum conjugated bile acids, taurochenodeoxycholic acid (TCDCA), taurohyodeoxycholic acid (THDCA), and tauro ursodesoxycholic acid (TUDCA) (HFD vs. ND, 2.8 times higher) in mice (Fig. 4A).In contrast, OCA signi cantly reduced (P < 0.05) the levels of CA, CDCA, hyodeoxycholic acid (HDCA), and ursodeoxycholic acid (UDCA) (Fig. 4B).Different variations were observed in the levels of fecal bile acids among groups, HFD apparently increased (P < 0.05) the levels of most fecal bile acids in mice compared with ND group (Fig. 4C), while OCA decreased the amounts of most fecal bile acids increased by HFD feeding including the level of fecal conjugated bile acids of glycodeoxycholic acid (GDCA) and glycoursodeoxycholic acid (GUDCA) (Fig. 4D).
To gure out whether bile acids variations in serum and feces were driven by gut microbes, we performed Spearman correlation analysis on abundance of species-level of bacteria and level of bile acids among the ND, HFD and OCA groups.Generally, variations of serum bile acids were stronger correlated with gut microbiota than serum bile acids variations.Bacteria species of B. massiliensis, A. miciniphila, L. johnsonii, and L. reuteri were positively correlated with the levels of serum TCDCA, THDCA, and TUDCA, and were signi cantly negatively correlated with the level of HDCA except for A. miciniphila (Fig. 4E, F).Besides, E. coli exhibited a signi cantly negative correlation with the levels of serum HDCA, CDCA, and CA (Fig. 4F).For correlations between mouse gut microbiota and fecal bile acids.The abundance of two species of Lactobacilli (L.murinus, and L. gasseri) exhibited signi cant positive correlations with the level of fecal tauroursodeoxycholic acid (TDCA), TUDCA, and TCDCA, while A. muciniphila was negatively correlated with fecal TCDCA, THDCA, and TUDCA (Fig. 4G).In addition, B. pseudolongum showed negative correlation with both the serum and fecal TDCA and TUDCA (Fig. 4G).Both A. muciniphila and S. thermophilus displayed positive correlations with fecal THDCA, while L. gasseri was vice versa (Fig. 4H).

Potential of bacterial genomes encoding enzymes involved in bile acids metabolism
To further investigate the functional pro les of gut microbiota in uencing bile acids metabolism, we performed enzymes scanning based on SGBs enriched in different groups (HFD vs. ND, HFD vs. OCA).
Compared with HFD group, SGBs mainly enriched in ND were species of Bacteroides, Parabacteroides, and Prevotellaceae in Bacteroidetes, Bi dobacterium pseudolongum, Alistipes indistinctus, and Clostridium innocuum (Fig. 5A).And OCA administration to HFD signi cantly elevated the abundance of two species of Bacteroides, Bi dobacterium pseudolongum, Alistipes indistinctus, two species of Prevotellaceae, and Escherichia_coli and Proteus mirabilis belonging to Proteobacteria (Fig. 5B).Worthy of note was that HFD compared to OCA treatment enriched bacteria exclusively belonging to Firmicutes (Fig. 5B).Intriguingly, some bacteria species were both enriched in ND and OCA, such as Bi dobacterium pseudolongum, Alistipes indistinctus, and species of Bacteroides and Prevotellaceae.The enzymes encoded by SGBs enriched in different groups indicated that bacterial species enriched in ND and OCA groups had adequate potential to encode both the BSHs (EC:3.5.1.24)and 7α-HSDs (EC:1.1.1.159),while HFD enriched bacteria exhibited poor capacity of encoding enzymes of 7α-HSDs, and mainly encoded BSHs responsible for primary bile acids synthesis (Fig. 5A, B).Collectively, HFD enabled the accumulation of bacteria encoding BSHs which accelerated the secretion of bile into gut and the uncoupling of conjugated bile acids into primary bile acids.And OCA administration elevated the abundance bacteria with potential to encode 7α-HSDs which facilitated the degradation of primary bile acids and promote of secondary bile acids bioconversion in the gut (Fig. 5C).

Discussion
2][33] Although studies on human 34,35 and mouse models 12,36 provided evidence of a causal role of gut microbiota in NAFLD development, knowledge gap still exists in whether gut microbiota in uences the therapeutic action of NALFD medication.The present study explored the role of gut microbiota on therapeutic effects of OCA administration on NALFD induced by high fat diet.In summary, we rst con rmed the critical role of gut microbiota on NALFD therapeutic effects and identi ed the key microbes modulating host bile acids pool which thereby contributing to NALFD development and OCA therapeutic effects of NALFD.Our study could provide new insights into the bile acids metabolism regulation by the gut microbes during NALFD development and treatment and lay the theoretical foundation for the NALFD prevention and treatment by the gut microbiota interventions.
In the previous studies, OCA was found to impressively mitigate hepatic lipid accumulation, liver in ammation and insulin resistance in NAFLD mice 21,37,38 .However, the effect of gut microbiota on OCA treatment to NALFD remains unclear.The current study implemented AMID and FMT trails and con rmed the critical role of gut microbial community during NAFLD treatment by OCA.Similar results were reported in studies that treated NALFD with antibiotics 39 and probiotics 40,41 in which bacteria species were supplemented as probiotics and improved liver function of NALFD.Compared with HFD group, both ND and OCA enriched the common genera including Akkermansia, Bi dobacteria, and Bacteroides which exactly comprised varieties of bene cial microbes.Indeed, Within the three genera, two probiotic species of A. miciniphila and B. pseudolongum were both attractively enriched in OCA and ND. A. miciniphila, a gut microbial member in healthy individuals, were found to exhibit host immunoregulatory effects 42 and could reverse the metabolic disorders in high fat diet induced obesity and type 2 diabetes 43 , while B. pseudolongum was suggested to improve the lipid metabolism of obese mouse model 44 .AIMD and FMT con rmed that gut microbiota was essential in NALFD treatment, and the differential gut microbial compositions suggested that speci c microbes were involved in NALFD development and therapy.
Bile acids and their receptors have emerged as important regulators of hepatic lipid metabolism 45 , and microbial modi cation of bile acids is an important mechanism by which the microbiota can interact with the host and affect liver disease 9 .Our results showed that OCA obviously decreased the levels of primary bile acids (e.g.CA and CDCA) in mouse serum, most free bile acids (e.g.DCA, CDCA, UDCA) and conjugated bile acids (e.g.TDCA, TUDCA) in mouse feces.Our nding is consistent with the previous study 46 suggesting that OCA was able to inhibit atherosclerosis by ileum bile acids metabolism modulation.In terms of bile acids, serum metabolites in circulation are usually better in manifestation of host metabolism and determines the progression of diseases.Noticeably, in the present study, serum TUDCA was predominantly increased in OCA group (2.8 times higher than HFD group).Meanwhile, OCA group enriched species such as Akkermansia miciniphila, Bi dobacterium animalis, Bacteroides massiliensis, and Lactobacillus helveticus which were signi cantly positively correlated with serum TUDCA.Less and weaker correlations were found between gut microbes and fecal bile acids in the present study.These results suggested that serum bile acids were more sensitive to the alteration of gut microbial alterations.Chen et al. 47 reported circulating bile acids rather than fecal bile acids showing apparent variations along with liver disease development which suggested circulating bile acids as promising indicators of disease status.Our study further indicated the difference of circulating and fecal bile acids was probably associated with gut microbiota alterations.More evidence was provided by functional pro ling of metagenome-assembled genomes, those species enriched in ND and OCA groups had the potential in encoding eznymes of BSHs (Bacteroides spp., Alistipes spp.and Bi dobacterium spp.) and 7α-HSDs (e.g.Alistipes spp.and Bacteroidaceae species) which were responsible for bioconversion of primary and secondary bile acids.These results demonstrated that the therapeutic effects of NALFD by OCA could be attributed to the modi cation of host bile acids by special gut microbes.Communication between the liver and intestine can be mediated by bile acids which are ligands for the nuclear receptor farnesoid X receptor (FXR) and the G-protein-coupled receptor TGR5 46 .FXR, as a ligand-activated nuclear receptor, plays an important role in repressing lipogenesis, promoting FA oxidation, and reducing fatty acids uptake in liver 48 .And TGR5 could exert a capacity to maintain glucose homeostasis and inhibit in ammation, thus improve NALFD features 49 .The present study con rmed the pivotal role of gut microbiota in NALFD treatment by OCA and revealed differential bile acids pro les resulting from gut microbial alterations caused by the high fat diet and the additional OCA treatment.
In conclusion, gut microbes, especially Bacteroides spp., Alistipes spp.and Bi dobacterium spp., could interact with the host through bile acids modi cation by encoding enzymes of BSHs and 7α-HSDs, thus determines the therapeutic effects of NALFD treatment by OCA.Although the causality of gut microbiota to NALFD therapeutic effects were con rmed in the current study, the mechanism by which variation in speci c bile acids contribute to NALFD therapeutic effects by OCA still requires further study.Future studies were still required to speci cally decipherer the interactions between gut microbiota and host bile acid pool, as well as how bile acids affect the development and treatment of NALFD.

Animal
Six-week-old speci c-pathogen-free male C57BL/6Jmice were provided by Department of Laboratory Animals, Kunming Medical University.NALFD model mice were induced by high fat diet (HFD) feeding, and normal diet (ND) feeding mice were taken as control group.OCA administration to HFD mice with or without antibiotic-induced microbiome depletion, and fecal microbiota transplantation from mice of HFD feeding with following OCA treatment to HFD feeding group were also performed.Each group in this study consisted of eleven mice and was raised for 20 weeks of experimental period.All mice were housed in a temperature-and humidity-controlled environment at Kunming Medical University with a 12 h/12 h light/dark cycle, with free access to food and water.All animal studies and experimental procedures were approved by approved by Ethics Committee of Kunming Medical University (license No. kmmu2021154).

NALFD model establishment and Antibiotic-induced microbiome depletion
The high fat diet (HFD) induced NAFLD mouse models were established with 12 weeks of HFD feeding to mice after a normal adaptive feeding.We referred to the reported protocol for preparing antibiotic cocktail (ampicillin 1g/L + vancomycin 500mg/L + neomycin 1g/L + metronidazole 1g/L) 50 .After feeding HFD for 10 weeks, the mice received antibiotic cocktail in drinking water for 2 weeks.At week 12, the mice were given oral gavage of OCA (50mg/kg body weight) for 8 weeks once a day.The indexes for detection were body weight alternation, glucose tolerance, insulin resistance, total cholesterol, triglyceride, hepatic fat accumulation, impaired intestinal barrier function and so on.

Fecal Microbiota Transplantation (FMT)
After HFD was given for 12 weeks, FMT was initiated and performed for 8 weeks (the total period was 20 weeks).Fresh fecal samples were collected from HFD-fed mice that were given OCA intervention.These collected fecal samples were diluted with saline solution and vigorously vortexed for 1 minute.Then the homogeneous liquids were centrifuged, and the supernatants were kept for FMT administration.Fresh supernatants regarded as graft materials were prepared within 10 minutes prior to oral gavage for preventing the variations of microbiota composition.New NAFLD mouse models were given oral gavage of fresh supernatants for 8 weeks once a day.During the experiment, body weight of mice in each group were monitored weekly, and the manifestations as well as the food intake dose of mice were documented once a week.Meanwhile, the related biochemical and pathological examinations of mouse serum and liver tissue were also performed weekly.

Histological analysis
Mouse liver tissues were rinsed in general stationary liquid and xed for over 24 hours.After xation, tissues were dehydrated in 30% sucrose for over 48 hours.The dehydrated tissues were embedded at -25 ℃, and then the frozen tissues were sectioned into slices with 8-10 µm thickness.Next, these slides were incubated in Oil Red O working solution for 8-10 minutes away from light, and the speci c staining time depended on the actual situation.After the slight air dry, the dyed slides underwent differentiation in 60% isopropanol solution, followed by hematoxylin counter-staining for 3-5 minutes.The staining time also relied on the actual situation.After washing under running water and differentiation in isopropanol, the slides were rinsed in bluing buffer, and then they were rinsed in running water again.Finally, these slides were mounted with glycerine jelly and covered by coverslips.The liver tissue and adipose tissue were rst xed and embedded in para n, and sectioned (5 µm) and stained with hematoxylin and eosin by standard procedures.

Examination of biochemical and in ammatory indexes
We collected 500 µl mouse blood sample, which was followed by 2000 × g centrifugation for 10 minutes.Next the supernatant was kept for detecting biochemical indexes by automatic biochemical analyzer, such as the levels of TC, TG, ALT, AST, high density lipoprotein (HDL), low density lipoprotein (LDL), fasting blood glucose and insulin.ELISA kit was used to detect the levels of in ammatory factors (IL-6, IL-1β and TNF-α) and triglyceride in serum.The manipulations relied on procedures of manufacturer.
Fecal metagenomic sequencing and bioinformatic analysis Shotgun metagenomic sequencing for fecal samples At the end of the experiment, we separately collected fresh fecal contents from mice in normal diet (ND) group, HFD group and OCA-intervened HFD (OCA) group, group information for individual assignment can be found in Supplementary Table S1A.These samples were stored at -80℃ until DNA extraction.DNA quality and quanti cation were evaluated with a NanoDrop Spectrophotometer ND-1000 (Thermo Fisher Scienti c).Metagenomic DNA libraries were constructed according to the instructions of NEXT ex Rapid DNASeq Kit and were 150-bp paired-end sequenced on the Illumina NovaSeq 6000 platform (Shanghai Biotechnology Corporation).
Metagenomic analysis based on metagenome-assembled genomes and enzymes involved in bile acids metabolism of bacterial genomes Metagenomic sequencing with binning strategy provides a better means to study the functional characteristics of speci c gut microbes based on the genome level.Previous study 59 on human microbiome reconstructed over 150,000 metagenome-assembled genomes (MAGs) from different populations and recapitulated 4930 species-level genome bins (SGBs) by 95% similarity of average nucleotide identity of genomes, the considerable number of SGBs greatly expanded the current resource of reference genomes.To better explore the functional pro le of the speci c microbes involved in NALFD, we took the 4930 SGBs as reference genomes and performed taxonomic and functional analysis based on the bacterial genomes.First, our high-quality microbial sequences were aligned to the reference genomes using BWA MEM (v0.7.17-r1188) 60 and SAMtools (v 1.9) 61 to calculate the abundance of SGBs in each sample, only SGBs with coverage of > 40% were considered detected in the samples.The abundance of each SGB was computed as the depth of the contigs of SGBs normalized by the total length of the genome to allow for sample-to-sample comparison, the abundance of SGBs in ND, HFD, and OCA groups can be found in Supplementary S1B.The differential abundant SGBs not being classi ed in the previous study were taxonomic annotated by GTDB-TK (v 2.1.0) 62against the Genome Taxonomy Database (GTDB release207_v2) 63 , and SGB-encoded proteins predicted by Prodigal (v.2.6.3) 64were annotated against KEGG database 65 using DIAMOND (v 0.9.7.108, "--evalue 1e-5") to screen the enzymes involved in bile acids metabolism.

Targeted metabolomic analysis of bile in serum and fecal samples
The serum sample was mixed with the extract (methanol: acetonitrile = 5:3) in a ratio of 1:4.The supernatant was collected by centrifugation after full shaking and standing.The organic reagent was removed, and the supernatant was redissolved in 50% methanol for inspection; A 50 mg fecal sample was accurately weighed, and 6ul of ice-cold methanol was added to each 1 mg sample.After grinding, the sample was crushed by shaking, standing and centrifugation.A 1:6 solution of frozen methanol was added to the residue a second time and the procedure was repeated.Extract was combined twice, organic reagents were removed, and 200 µL 50% methanol was redissolved for later use.The instrument parameters used for sample loading (Instruments: Waters Acquity UPLC, mass spectrometry AB SCIEX 5500 QQQ-MS; Chromatographic column: Acquity UPLC BEH C18 (1.7µm, 2.1 mm*100 mm)).The bile acid reference was weighed and a series of solutions containing the nal concentration of the reference were prepared.The standard curve was drawn according to the peak area of different concentrations and the substandard of the corresponding concentrations.MultiQuant software (v 3.0.2,Sciex) was used for integration, and the content was calculated using the standard curve.The concentration of serum and fecal bile acids can be found in Supplementary Table S2.

Statistical Analysis
GraphPad Prism8.0 was used to cope with data acquired from experiment, and these data were represented as mean ± SEM.According to circumstances, independent-samples T test or One-way ANOVAs was performed for statistically comparative analysis of inter-group differences.Kruskal-Wallis test and Mann-Whitney U test were conducted for inter-group comparison of the bile acid level.LEfSe 66 was applied to the microbial composition analysis on different taxonomic levels, and the cut-off of LDA score was set to 2, and signi cant features were considered with P-values lower than 0.05.The exploration of association between bile acid and gut microbiota was implemented by Spearman correlation analysis based on the Centered log-ratio (CLR) Normalized abundance.All the P-values were corrected using the Benjamini-Hochberg FDR method.Signi cant differences were recognized when adjusted-P values were less than 0.05.

Figure 4 Levels
Figure 4