Obesity is a complex and multifactorial disorder, involving the balance between energy intake and expenditure(Reference Maruvada, Leone and Kaplan1). Long-term persistence of obesity will increase the prevalence of diabetes but also of non-alcoholic fatty liver disease (NAFLD) that is becoming the most important cause of chronic liver disease in the western countries(Reference Perumpail, Khan and Yoo2). The biological events driving the progression of NAFLD are not clearly elucidated. The hundreds of billions of micro-organisms present in our gut, constituting what is called microbiota, are now considered as potential new therapeutic targets(Reference Maruvada, Leone and Kaplan1, Reference Torres-Fuentes, Schellekens and Dinan3). An interesting study reported that the gut microbiota composition (including bacterial diversity) explains the differential response to dieting in terms of improvement of metabolic disorders and inflammation in obese individuals(Reference Cotillard, Kennedy and Kong4).
The dysfunction of the microbial ecosystem in obesity contributes to alterations of gut barrier, thereby promoting systemic inflammation, namely through the translocation of lipopolysaccharides (LPS)(Reference Cani, Bibiloni and Knauf5). This process is called metabolic endotoxaemia. The portal vein directly carries gut-derived products to the liver. This organ functions as a secondary ‘firewall’ and protects the body from intestinal pathogens and other microbial products that have crossed the primary barrier of the intestinal tract(Reference Brandl, Kumar and Eckmann6). The liver is also exposed to bacterial metabolites that may have beneficial effects on metabolic and inflammatory processes, as explained later. Indeed, the intestinal bacteria are able to metabolise nutrients into a wide range of co-metabolites (so-called since they can be further metabolised by host tissues) with metabolic, immune and/or neuroactive properties(Reference Husted, Trauelsen and Rudenko7–Reference Krishnan, Ding and Saedi9). Although efforts have recently been dedicated to the identification of bacterial by-products, the mechanism behind the beneficial or detrimental effects of key microbial metabolites on host health in the context of NAFLD remains to be studied. In this review, we describe how bacterial metabolites may impact on both the metabolic and inflammatory status of the liver. We discuss novel approaches based on microbiome–nutrient–host interactions (e.g. probiotics, prebiotics) and the need for research and adequate intervention studies to evaluate the feasibility and relevance of these new therapies for NAFLD.
Gut microbial dysbiosis in obesity and non-alcoholic fatty liver disease
Obesity is associated with metabolic alterations related to glucose and lipid homoeostasis (e.g. glucose intolerance, type-2 diabetes, insulin resistance, dyslipidaemia) (WHO, 2016, Obesity and overweight; http://www.who.int/mediacentre/factsheets/fs311/en/). During the past few decades, the implication of the gut microbiota in the progression of metabolic alterations has been mostly described in rodent models of obesity(Reference Cani, Osto and Geurts10). Indeed, it has been well established that gut microbiota controls the gut barrier function and, therefore, the progression of metabolic endotoxaemia characterised by the translocation of specific microbe-associated molecular patterns such as LPS into the systemic circulation(Reference Cani, Amar and Iglesias11). In mouse models of obesity (high-fat diet, ob/ob or db/db mice), the changes in gut microbiota composition are often associated with an increased gut permeability marked by an alteration of the tight junction proteins, occludin and zonula occludens −1(Reference Cani, Possemiers and Van de Wiele12, Reference Neyrinck, Van Hee and Piront13). The disrupted gut barrier function observed during obesity induces metabolic endotoxaemia leading to a low-grade inflammation(Reference Cani, Bibiloni and Knauf5, Reference Cani, Possemiers and Van de Wiele12). This mechanism has been confirmed in human subjects, and more specifically in overweight patients with type-2 diabetes(Reference Al-Attas, Al-Daghri and Al-Rubeaan14). Interestingly, a chronic low grade inflammatory state has been linked to the development of insulin resistance in peripheral tissues during obesity and diabetes(Reference Galic, Oakhill and Steinberg15). This suggests a major role for the gut microbiota in the control of insulin sensitivity in the whole organism.
Data are accumulating in animal models and human subjects suggesting that obesity and type-2 diabetes are associated with bacterial dysbiosis(Reference Tilg and Moschen16). Subjects with a low bacterial richness are characterised by a higher adiposity, insulin resistance, dyslipidaemia and inflammatory phenotype compared with high bacterial richness individuals(Reference Le Chatelier, Nielsen and Qin17). Faecalibacterium prausnitzii was less abundant in obese subjects with diabetes and associated negatively with inflammatory markers. Interestingly, the relative abundance of F. prausnitzii significantly increased in these diabetic patients after by-pass surgery, suggesting that this bacterium could be associated with the reduction in low-grade inflammation(Reference Furet, Kong and Tap18).
In addition, butyrate-producing bacteria such as Roseburia intestinalis were lower in overweight subjects with type-2 diabetes(Reference Tilg and Moschen16). In human subjects, a study demonstrated that the ratio Bacteroides:Prevotella was lower in obese patients compared with control subjects and that this ratio was negatively correlated with corpulence(Reference Furet, Kong and Tap18).
Some studies suggest that gut microbiota composition during childhood could influence the development of obesity. Indeed, early differences in faecal microbiota composition in children, such as the level of bifidobacteria, may predict the appearance of overweight in adolescence. Consistent with this result, a negative correlation between BMI and Bifidobacterium spp. was reported in several studies(Reference Castaner, Goday and Park19). It has been reported that the relative abundance of Akkermansia muciniphila that contributes to the homoeostasis of the protective mucus layer, is lower in the gut microbiota from obese and diabetic patients(Reference Le Chatelier, Nielsen and Qin17), together with rodent models of obesity(Reference Everard, Belzer and Geurts20). It could also be of therapeutic interest as A. muciniphila administration prevents the fat mass accumulation and metabolic endotoxaemia in high-fat diet fed mice(Reference Everard, Belzer and Geurts20, Reference Plovier, Everard and Druart21).
NAFLD is a clinical syndrome almost systematically associated with obesity(Reference Haas, Francque, Staels and Julius22), and gut microbial dysbiosis is in a similar manner associated with this pathology. The syndrome is caused by an abnormal accumulation of TAG in hepatocytes (liver parenchymal cells) and can eventually evolve in some patients to non-alcoholic steatohepatitis (NASH), characterised by hepatic inflammation, and in worst cases, to cirrhosis and hepatocellular carcinoma. The implications of the microbiota in this pathology have long been hypothesised(Reference Tilg, Cani and Mayer23) and are gradually being explored thanks to next-generation sequencing techniques. In the past few years, it has been demonstrated that the digestive microbial composition of NAFLD and NASH patients differs from healthy individuals to variable degrees(Reference Abdou, Zhu and Baker24), regardless of the energy intake and the BMI(Reference Duarte, Stefano and Miele25). In a pioneer study, Le Roy et al.(Reference Le Roy, Llopis and Lepage26) demonstrated that mice fed a similar high fat diet with similar genetic backgrounds develop varying levels of steatosis that can be correlated with differences in the intestinal microbial composition. The susceptibility to the development of NAFLD is transmissible via gut microbiota transplants in mice, highlighting the direct metabolic impact of the microbiota upon hepatic metabolism and health.
The mechanisms by which the microbiota impacts on hepatic metabolism and health are however poorly understood. The liver is directly connected to the gastrointestinal tract through the gut–liver axis. The portal vein enables the transport of nutrients and bacterial compounds and metabolites from the intestinal lumen through the gut barrier to the liver, contributing to homoeostasis under healthy physiological conditions. Under dysbiosis, several mechanisms contribute to the development of hepatic steatosis. As observed during obesity, intestinal barrier integrity can be compromised with an increased gut permeability associated with increased bacterial translocation(Reference Miele, Valenza and La Torre27) and endotoxaemia that directly contributes to hepatic lipid metabolism disruption(Reference Cani, Amar and Iglesias11). Even if endotoxaemia is not characterising all NAFLD patients, it has been observed that endotoxin levels are higher in NASH patients compared with NAFLD patients(Reference Abdou, Zhu and Baker24) causing an increase in the release of pro-inflammatory cytokines. Intestinal inflammation caused by dysbiosis also favours NAFLD progression as demonstrated by Henao-Mejia et al.(Reference Henao-Mejia, Elinav and Jin28). Finally, changes in gut microbial composition alter its function, inducing changes in microbial metabolite production such as SCFA, trimethylamine, ethanol and secondary bile acids (BA) that can modify lipid metabolism and inflammatory processes in the liver(Reference Bashiardes, Shapiro and Rozin29).
Despite the pathophysiological association between obesity and steatosis, the microbial shift in NAFLD distinguishes itself from the one occurring with obesity. NAFLD is not systematically associated with a reduction in bacterial diversity compared with control(Reference Da Silva, Teterina and Comelli30–Reference Zhu, Baker and Gill33). Boursier et al.(Reference Boursier, Mueller and Barret34) did not evidence any association between microbial diversity and NAFLD severity whereas Hoyles et al.(Reference Hoyles, Fernandez-Real and Federici35) observed a decrease in diversity with the increase in NAFLD severity. The β diversity represents the species diversity among habitats or sites. Various studies have used β-diversity metrics as tools to assess whether the gut microbiota composition is different between patients with steatosis and healthy controls and contradictory results have been obtained(Reference Da Silva, Teterina and Comelli30–Reference Zhu, Baker and Gill33, Reference Shen, Zheng and Sun36). An increase in the abundance of the Prevotella genus has been observed in NASH and NAFLD children(Reference Zhu, Baker and Gill33, Reference Michail, Lin and Frey37) while a decrease in the abundance of this genus is reported in adults(Reference Jiang, Wu and Wang31, Reference Boursier, Mueller and Barret34, Reference Shen, Zheng and Sun36). The relevance of Prevotella genus as a risk factor or a marker of steatosis would depend on the age of the patient. Data concerning the abundance of Roseburia are also inconsistent: some authors reporting decreases in adult(Reference Duarte, Stefano and Miele25) and child(Reference Zhu, Baker and Gill33) NASH patients, while others have observed an increase in NAFLD children(Reference Raman, Ahmed and Gillevet38), indicating a possible variation in the abundance of this taxon with the severity of the steatosis. Decreases in the Coprococcus (Reference Da Silva, Teterina and Comelli30, Reference Zhu, Baker and Gill33) and Faecalibacterium (Reference Duarte, Stefano and Miele25, Reference Da Silva, Teterina and Comelli30, Reference Wong, Tse and Lam32, Reference Shen, Zheng and Sun36) genera, in particular the F. prausnitzii, an anti-inflammatory species(Reference Da Silva, Teterina and Comelli30), have been observed in NAFLD and NASH patients. Such modifications are also observed in obese patients, independent of the presence of steatosis(Reference Furet, Kong and Tap18). Decreases in the abundance of the Ruminococcaceae family(Reference Da Silva, Teterina and Comelli30, Reference Jiang, Wu and Wang31, Reference Zhu, Baker and Gill33, Reference Shen, Zheng and Sun36, Reference Raman, Ahmed and Gillevet38) in particular the Ruminococcus genus(Reference Duarte, Stefano and Miele25, Reference Da Silva, Teterina and Comelli30) have also been reported in several studies while increases in the Lactobacillus genus have been observed in adult NASH and NAFLD patients(Reference Duarte, Stefano and Miele25, Reference Jiang, Wu and Wang31, Reference Raman, Ahmed and Gillevet38). These two taxa comprise a large variety of species with very different functional implications. Many bacteria from the Ruminococcaceae family are butyrate producers; the modulation of their abundance could thus impact on SCFA production and subsequently on liver metabolism. Likewise, the Lactobacillus genus comprises different species with important immunological implications that influence liver health, but conclusions on the functional implications of this genus are impossible to make at that taxonomic rank. The lack of consistency in the microbial changes between the data is possibly due to the high variability in the populations studied (differing in diet, medical treatments, age, environment, sex, severity of the steatosis) and the variability in analytical methods (sequencing, bioinformatics treatment, databases used, statistics). Studies on larger well-characterised cohorts are needed in order to have robust associations between specific bacterial taxa and liver steatosis. The diet is among the main triggers of NAFLD and the gut microbiota is modified accordingly, underlining the importance of the concomitant study of the nutrients and microbial impact on liver health and metabolism.
How do nutrients influence the gut–liver axis?
The gut microbiota has the capacity to produce a diverse range of compounds that play a major role in regulating the activity of distal organs such as the liver through portal circulation and thus modulate hepatic metabolism and health in a broader manner, which will be discussed in detail (Fig. 1). Metabolites are produced through the metabolism of food components by the gut microbiota, such as SCFA, indole- or phenol- derivatives(Reference Postler and Ghosh39). Secondary BA are also key metabolites produced from the gut microbiota from primary BA synthesised from cholesterol by host liver(Reference Ridlon, Kang and Hylemon40).
Contributions of bacterial metabolites derived from amino acids or prebiotics to the gut–liver axis. Gut bacteria catabolise tryptophan into indole and indole-3-acetic acid, phenylalanine into phenylacetic acid and tyrosine into p-cresol (among others). SCFA such as acetic, propionic and butyric acids are produced by inulin-type fructans (ITF) fermentation. ITF treatment may affect bile acid (BA) profiles. Some of these metabolites cross the intestinal epithelium and are transported to the liver through the portal circulation whereas others act on intestinal permeability and/or intestinal L cells to produce glucagon-like peptides (GLP)-1 and -2, that improve liver metabolism (insulin sensitivity) and gut barrier (translocation of lipopolysaccharides (LPS)), respectively. Hepatic enzymes produce host-microbiota co-metabolites such as indoxyl-3-sulphate, hippuric acid and phenylacetylglutamine. (Host-)microbiota (co-)metabolites influence various physio(patho)logical processes, namely inflammation and/or lipid accumulation (steatosis) in the liver. For more details, please refer to the main text. Most of the findings illustrated in the figure have been obtained using mouse models of obesity or non-alcoholic fatty liver disease. The figure was produced using Servier MedicalArt (http://www.servier.com). GPR, G protein-coupled receptor; TGR5, Takeda G protein-coupled receptor 5.
SCFA
Among SCFA, acetic, propionic and butyric acids are the major products of carbohydrate fermentation by the gut micro-organisms, the gut microbiome producing 50–100 mm daily of these compounds(Reference Leung, Rivera and Furness41). Branched SCFA can be produced from amino acid fermentation by the gut microbes(Reference Nyangale, Mottram and Gibson42). The types and amounts of SCFA synthesised in the gut are influenced by the amount of dietary non-digestible carbohydrates, and by the saccharolytic and metabolic characteristics of the gut microbiota. SCFA derived from the gut provide an energy source for different cell types. Butyrate is mostly used as an energy substrate in the colonocyte, whereas acetate and propionate may be used as substrates for glucose and fatty acid synthesis, respectively. Data obtained with labelled SCFA introduced in the colon of healthy subjects using colon delivery capsules and followed by the measurement of plasma levels of 13C-SCFA, 13C-glucose, 13C-cholesterol and 13C-fatty acids allowed the quantification of the respective contributions of SCFA as metabolic substrates in human subjects(Reference Boets, Deroover and Houben43). Systemic availability of colonic-administered acetate, propionate and butyrate was 36, 9 and 2 %, respectively. Conversion of acetate into butyrate (24 %) was the most prevalent interconversion by the colonic microbiota and was not related to the butyrate-producing capacity in the faecal samples. Less than 1 % of administered acetate was incorporated into cholesterol and <15 % in fatty acids. On average, 6 % of colonic propionate was incorporated into glucose. In rodents, it has been shown that SCFA reduced hepatic cholesterol synthesis and lowered hepatic fatty acid synthase activity and hepatic lipid synthesis whereas there was an increase in hepatic lipid oxidation, shifting hepatic lipid metabolism towards a more oxidative state(Reference Wong, de Souza and Kendall44–Reference Demigne, Morand and Levrat47). SCFA act on the G protein-coupled receptors (GPR) GPR41 and GPR43 of gut enteroendocrine L cells to produce several effects that might impact liver metabolism. These L cells release glucagon-like peptide (GLP)-1, which can act directly on hepatocytes, by activating genes involved in fatty acid β-oxidation and insulin sensitivity(Reference Svegliati-Baroni, Saccomanno and Rychlicki48). SCFA have also been known to modulate the production of pro- and anti-inflammatory mediators by immune cells, namely through the binding to GPR43 receptors(Reference Bindels, Dewulf and Delzenne49). GPR43 has been implicated in mice in the control of glucose homoeostasis(Reference Bjursell, Admyre and Goransson50) and lipid metabolism(Reference Kimura, Ozawa and Inoue51). Altogether, these data reinforced the hypothesis that GPR43 is involved in metabolic control and could therefore participate in some metabolic disturbances in the liver tissue(Reference Bindels, Dewulf and Delzenne49). Interestingly, a recent study provided the novel insight that butyrate regulated PPARα to stimulate hepatic fatty acid β-oxidation and inhibited inflammation in high-fat-diet-induced NAFLD(Reference Sun, Jia and Hong52).
Bile acids
BA are important signalling molecules and metabolic regulators that control glucose and lipid homoeostasis (for review, see(Reference Leung, Rivera and Furness41, Reference Fu and Cui53)). The majority of BA that are secreted into the intestinal lumen are reabsorbed from the terminal end of the small intestine and return to the liver through the portal blood. The BA that are synthesised in the liver from cholesterol and conjugated with taurine or glycine on the side chain are called primary BA, which further undergo deconjugation, dehydroxylation, epimerisation and oxidation into secondary BA by intestinal bacteria, in the large intestine. BA are signalling molecules that coordinately regulate metabolism and inflammation via the nuclear farnesoid X receptor (FXR) and the Takeda G protein-coupled receptor 5(Reference Chavez-Talavera, Tailleux and Lefebvre54). These receptors control the expression of genes involved in BA, lipid and carbohydrate metabolism, energy expenditure and inflammation. In rodents, FXR activation may reduce NAFLD, as it lessens steatosis by inhibiting lipogenesis, decreases chemically induced hepatic inflammation and fibrosis; it may help maintaining intestinal barrier integrity, thus protecting the liver from bacteria-derived inflammatory signals(Reference Chavez-Talavera, Tailleux and Lefebvre54). It has been recently shown that in obese patients, the hepatic necro-inflammatory lesions observed in NASH are not associated with alterations in BA metabolism and signalling. BA alterations rather reflect the metabolic phenotype associated with NASH(Reference Legry, Francque and Haas55).
Aromatic amino acid derivatives
Recently, bacterial metabolites derived from aromatic amino acids (tryptophan, phenylalanine and tyrosine) emerged as a new class of microbial molecules influencing liver functions. Interestingly, these bacterial metabolites are further metabolised by host intestinal and hepatic enzymes into host-microbiota co-metabolites that can be detected in the plasma or urine of conventional mice but not of germ-free or antibiotic-treated mice(Reference Wikoff, Anfora and Liu56, Reference Zheng, Xie and Zhao57). Gut bacteria produce numerous tryptophan catabolites, including indole, indole-3-propionic acid, indole-3-acetic acid, indole-3-aldehyde, tryptamine and 3-methylindole (skatole)(Reference Agus, Planchais and Sokol58, Reference Roager and Licht59). Among them, indole is the most abundant (low millimolar concentration in the gut) while the concentrations of the other metabolites are much lower (<10 µm)(Reference Jin, Lee and Sridharan60). Since many tryptophan-derived bacterial metabolites beneficially control intestinal permeability and mucosal immunity(Reference Agus, Planchais and Sokol58, Reference Roager and Licht59), these compounds could limit the translocation of harmful microbiota-derived components (LPS, peptidoglycan, etc.) from the gut to the liver. Moreover, indole regulates GLP-1 secretion from L-cells in vitro and thus could indirectly impact hepatic metabolism(Reference Chimerel, Emery and Summers61). Alternatively, tryptophan catabolites could directly regulate hepatic physiology after transport through the portal blood. Unfortunately, for most of these bacterial metabolites, their concentrations in the portal circulation or in the hepatic tissue are not known. Bacterial indole is known to reach the liver since the hepatic enzymes Cyp2e1 and Sult1a1 convert it into indoxyl-3-sulphate, the main tryptophan derived co-metabolite(Reference Banoglu, Jha and King62, Reference Banoglu and King63). We recently demonstrated that indole (100 µm) alleviates hepatic inflammation in precision-cut liver slices treated ex vivo with bacterial endotoxin or prepared from genetically obese mice, through mechanisms partly involving Kupffer cells(Reference Beaumont, Neyrinck and Olivares64). Indole oral administration in mice also prevented inflammation in association with a reduction of LPS-induced alterations of cholesterol metabolism and a regulation of liver X receptor target gene expression(Reference Beaumont, Neyrinck and Olivares64). In vitro, indole-3-acetic acid reduced pro-inflammatory cytokine expression in macrophages, fatty acid accumulation in hepatocytes and mRNA levels of the key lipogenesis proteins fatty acid synthase and sterol regulatory element-binding protein-1c(Reference Krishnan, Ding and Saedi9).
Most of the bacterial metabolites derived from tryptophan activate the aryl hydrocarbon receptor (AhR) which could mediate their hepatic effects(Reference Hubbard, Murray and Perdew8). For instance, the reduction of lipid accumulation in hepatocytes by indole-3-acetic acid involves the AhR(Reference Krishnan, Ding and Saedi9). Intriguingly, both indole and the related co-metabolite indoxy-3-sulphate induced the expression of AhR target genes but only indole reduced hepatic inflammation, suggesting a dissociation between AhR activation and the anti-inflammatory effects of indole(Reference Beaumont, Neyrinck and Olivares64). Obese individuals have low faecal concentrations of tryptophan-derived metabolites (indole, indole-3-acetic acid, 3-methylindole and tryptamine) and accordingly low AhR agonist activity(Reference Natividad, Agus and Planchais65). High-fat feeding in mice also reduced the intestinal concentration of indole, indole-3-acetic acid and tryptamine(Reference Natividad, Agus and Planchais65). Another study similarly reported low caecal and hepatic concentrations of indole-3-acetic acid in mice fed a high-fat diet(Reference Krishnan, Ding and Saedi9). Interestingly, supplementation of diet-induced obese mice with a Lactobacillus reuteri strain producing tryptophan catabolites reversed hepatic metabolic alterations through improvement of gut barrier function and incretin secretion in an AhR-dependent manner(Reference Natividad, Agus and Planchais65).
The main bacterial catabolites of phenylalanine are phenylacetic acid, phenylpropionic acid and benzoic acids(Reference Russell, Duncan and Scobbie66). Phenylacetic acid and benzoic acid are transported to the liver and metabolised into the co-metabolites phenylacetylglutamine (phenylacetylglycine in mice) and hippuric acid, respectively(Reference Wikoff, Anfora and Liu56). In obese women, plasma phenylacetic acid concentration was positively associated with steatosis severity(Reference Hoyles, Fernandez-Real and Federici35). In primary human hepatocytes, treatment with a high concentration of this bacterial metabolite (10 mm) induced lipid accumulation and altered the expression of genes involved in lipid and glucose metabolism(Reference Hoyles, Fernandez-Real and Federici35). Moreover, phenylacetic acid supplementation (0·8 %) in a standard diet for 2 weeks increased hepatic TAG levels in mice(Reference Hoyles, Fernandez-Real and Federici35). Regarding hepatic inflammation, phenylacetic acid (1 mm) and benzoic acid (1 mm) did not reduce LPS-induced pro-inflammatory gene expression in mouse precision-cut liver slices(Reference Beaumont, Neyrinck and Olivares64).
Tyrosine is degraded by the gut microbiota into p-cresol, phenol, 4-hydroxyphenylacetic acid, 4-hydroxybenzoic acid, 4-hydroxyphenylpropionic acid and 4-hydroxyphenyllactic acid(Reference Russell, Duncan and Scobbie66, Reference Windey, De Preter and Verbeke67). After absorption by the intestinal epithelium, p-cresol and phenol are metabolised by host enzymes into the co-metabolites p-cresyl sulphate and phenyl sulphate, respectively(Reference Wikoff, Anfora and Liu56). p-Cresol (1 mm) prevented LPS-induced proinflammatory gene expression ex vivo in mouse precision-cut liver slices(Reference Beaumont, Neyrinck and Olivares64). However, since this bacterial metabolite has toxic effects on the intestinal epithelium(Reference Andriamihaja, Lan and Beaumont68), more studies are needed to decipher its impact on hepatic tissue.
Altogether, recent studies show that bacterial metabolites derived from aromatic amino acids play a pivotal role in the microbiota-host communication through the gut–liver axis, some of them being protective for hepatic metabolism and inflammation (indole, indole-3-acetic acid) while others (phenylacetic acid) potentially contribute to steatosis progression (Fig. 1). Importantly, the hepatic effects of most aromatic amino acid bacterial catabolites have not been characterised yet. Moreover, in vivo, these metabolites reach the liver in combination, which could induce antagonist or synergistic effects between them but this hypothesis remains to be tested. In overweight or obese individuals, high-protein diets increased the faecal concentration of phenylacetic acid and urinary excretion of the co-metabolites phenylacetylglutamine, p-cresyl sulphate and indoxyl sulphate, these effects being modulated by the protein source(Reference Beaumont, Portune and Steuer69, Reference Russell, Gratz and Duncan70). However, the hepatic consequences of this upregulation of aromatic amino acid catabolism by the gut microbiota during high-protein diets were not investigated. Since all bacterial metabolites increased by high-protein diets are not protective for the liver, more specific nutritional strategies (e.g. targeting only indole) would be highly desirable. Increasing the substrate availability for the gut microbiota is the most straightforward approach but remains challenging since free amino acids are efficiently absorbed in the upper part of the gut. A promising approach would be to encapsulate aromatic amino acids to ensure their targeted release in the distal intestine where the gut microbiota could degrade them into bioactive metabolites.
Intervention studies with probiotics and prebiotics in non-alcoholic fatty liver disease
Probiotics are selected micro-organisms which, when given in adequate amount, have a beneficial effect on host health(Reference Hill, Guarner and Reid71). Prebiotics are nutrients selectively utilised by host micro-organisms that confer health benefit to the host(Reference Gibson, Hutkins and Sanders72, Reference Gibson and Roberfroid73). A recent systematic review and meta-analysis has been published on prebiotic, probiotic and synbiotic (combination of probiotics and prebiotics) therapies for patients with NAFLD in randomised controlled trials supporting the potential use of microbial therapies in the treatment of NAFLD(Reference Loman, Hernandez-Saavedra and An74). Meta-analysis indicated that microbial therapies significantly reduced BMI, hepatic enzymes (aspartate aminotransferase (AST), alanine aminotransferase (ALT), γ-glutamyl transferase), serum cholesterol, LDL-cholesterol and TAG, but not inflammation (based on TNF-α and C-reactive protein determination). Subgroup analysis by treatment category indicated similar effects of prebiotics and probiotics on BMI and liver enzymes but not total cholesterol, HDL-c and LDL-c(Reference Loman, Hernandez-Saavedra and An74). One meta-analysis study highlighted that the use of probiotics significantly reduced AST, ALT and ultrasonographic grade of fatty liver(Reference S Lavekar, V Raje and Manohar75). Another study showed that administration during 3 months of Lactobacillus bulgaricus and Streptococcus thermophilus decreased ALT, AST and γ-glutamyl transferase in patients with NAFLD(Reference Aller, De Luis and Izaola76). Moreover, 6 months of treatment with a probiotic mixture containing Lactobacillus and Bifidobacterium strains reduced liver fat accumulation and AST in patients with NASH(Reference Wong, Won and Chim77). A 4-month supplementation with VSL#3 probiotics also significantly improves NAFLD in children(Reference Alisi, Bedogni and Baviera78).
Prebiotic definition now includes a large panel of non-digestible nutrients(Reference Gibson, Hutkins and Sanders72). Inulin-type fructans (ITF) were the first to be recognised as prebiotics, together with (galacto-)oligosaccharides. Dietary ITF, which are present in various fruit and vegetables, are fermentable carbohydrates that display prebiotic properties, as their metabolisation by gut micro-organisms modulates the composition and/or activity of the gut microbiota, thus conferring a beneficial physiological effect on the host(Reference Bindels, Delzenne and Cani79). The beneficial effect of ITF prebiotics on cardio-metabolic risk have been demonstrated in many mouse models of obesity, an effect namely linked to gut peptides (GLP-1)(Reference Roberfroid, Gibson and Hoyles80, Reference Delzenne, Cani and Everard81). Our recent study suggested that changing the microbial composition using ITF impacted largely on the production of secondary BA and could contribute to the improvement of the host's health(Reference Catry, Bindels and Tailleux82). Changes in BA profiles by ITF treatment may also result from the modulation of BA metabolism, including hepatic synthesis (in favour of Cyp7a1) and intestinal reuptake. FXR stimulation seems to suppress NF-κB and in doing so decreases hepatic inflammation. Of particular interest, muricholic acids, primary BA only present in rodents, with FXR antagonistic properties can be induced by ITF in a dietary model of n-3 PUFA depletion inducing hepatic steatosis and endothelial dysfunction in mice(Reference Catry, Bindels and Tailleux82).
Animal and human research has demonstrated that ITF improve several NAFLD-associated metabolic risk factors including gut microbiota dysbiosis, intestinal permeability, endotoxaemia, inflammation, glycaemia and hepatic lipogenesis(Reference Mokhtari, Gibson and Hekmatdoost83–Reference Kellow, Coughlan and Reid87). ITF induce secretion of GLP-2 (co-secreted with GLP-1 by L cells) that is implicated in the lower systemic inflammation occurring in obese mice. The decrease in LPS absorption through an improvement of the expression and activity of proteins involved in gut barrier function (zonula occludens-1 and occludin), occurs in prebiotic-treated animals. In an exploratory, double blind intervention study with ITF in obese women, we demonstrated that the changes in gut microbiota induced by ITF prebiotics are correlated with serum LPS levels, despite a lack of significant effect on body weight(Reference Dewulf, Cani and Claus85). A more recent study demonstrated that ITF decreases serum LPS in adults with overweight/obesity(Reference Parnell, Klancic and Reimer88).
Inulin-type prebiotics could have a beneficial impact on hepatic lipogenesis in healthy human subjects(Reference Letexier, Diraison and Beylot89). Following this observation, some research studies have focused on the potential interest of prebiotics to regulate hepatic metabolic functions in the context of obesity or associated metabolic disorders such as NAFLD in patients. Our team has previously shown that 8 weeks ITF supplementation significantly decreased the serum AST in patients with NASH(Reference Daubioul, Horsmans and Lambert86). In this context, important human intervention studies are currently ongoing in obese and NAFLD patients. The design of a first clinical trial performed in overweight patients with confirmed NAFLD has been published recently(Reference Lambert, Parnell and Eksteen90). In this study, patients were asked to consume prebiotics (16 g/d) during 24 weeks. Recently, a multicentre intervention trial in obese patients presenting co-morbidities has been conducted during the FOOD4GUT project devoted to better understand how inulin-type prebiotics present in food play a role on gut microbiota homoeostasis and health (https://sites.uclouvain.be/FOOD4GUT/). It consists in a single blind randomised control trial in which patients were asked to consume 16 g native inulin daily combined with recipes based on vegetables naturally rich in ITF for 10–16 weeks. The placebo consists in eating 16 g maltodextrin daily, together with recipes favouring vegetables that are not rich in ITF. The primary outcome is to relate the changes in gut microbiome with the metabolic alterations. Such a protocol will allow evaluation of the potential effect on liver disorders, since liver fat accumulation and fibroses will be assessed by ultrasonography (Fibroscan) as well as through the measurement of biomarkers.
Synbiotic approaches, combining prebiotics and probiotics, have also been tested in the NAFLD context. First, in a randomised controlled clinical trial, patients with NAFLD were asked to consume daily 300 g synbiotic yoghurt containing Bifidobacterium animalis strain combined with 1·5 g inulin(Reference Bakhshimoghaddam, Shateri and Sina91). At the end of the protocol, compared with the placebo group, synbiotic formula significantly decreased the grade of NAFLD determined with ultrasonography and improved the serum concentrations of hepatic enzymes in NAFLD patients(Reference Bakhshimoghaddam, Shateri and Sina91). Likewise, in NASH patients, Bifidobacterium longum strain combined with ITF also reduced hepatic steatosis and serum concentrations of hepatic enzymes(Reference Malaguarnera, Vacante and Antic92). In addition, in patients with NAFLD, synbiotic supplementation based on several strains of Lactobacillus, Streptococcus and Bifidobacterium combined with ITF ameliorated hepatic fibrosis, improved liver enzymes and attenuated inflammation in the systemic circulation, compared with the placebo group. Finally, another intervention based on the consumption of L. reuteri with guar gum and inulin, during 3 months in patients with NASH, significantly reduced hepatic steatosis in the treated group but not in the control(Reference Ferolla, Couto and Costa-Silva93). However, this amelioration of hepatic steatosis was not associated with a regulation of hepatic enzymes (AST, ALT, alkaline phosphatase and γ-glutamyl transferase). Another important clinical trial is currently in progress in NAFLD patients consuming 4 g ITF twice daily combined with B. animalis subsp. lactis BB-12 (INSYTE study)(Reference Scorletti, Afolabi and Miles94). The primary outcomes of this study will be to observe the liver fat accumulation by magnetic resonance spectroscopy, to measure a composite liver fibrosis score generated from blood concentrations of three analytes (serum hyaluronic acid, serum amino-terminal pro-peptide of type-III collagen and tissue inhibitor matrix metalloproteinase 1). The gut microbiota composition will also be determined by 16S rRNA sequencing.
In addition to the pre-, pro- and synbiotic approaches, it has been proposed that the replacement of the gut microbiota of ill patients by the microbiome of ‘healthy’ volunteers, could be helpful if the gut microbiota is involved in the evolution of the disease. Faecal transplantation from lean donors into obese/diabetic patients was linked to a marked increase in the proportion of the butyrate producer R. intestinalis, and improved insulin sensitivity(Reference Vrieze, Van Nood and Holleman95). The improvement of insulin sensitivity upon faecal microbiota transplantation from lean donor in metabolic syndrome is driven by baseline intestinal microbiota composition since it is more effective in patients with bacterial low diversity(Reference Kootte, Levin and Salojarvi96). To date, no data are available regarding this kind of approach in NAFLD patients, with a primary outcome being the effect on hepatic disease.
Conclusions
In conclusion, few studies are available concerning the beneficial impact of dietary approaches modulating the gut microbiota composition for the treatment of NAFLD. The available data suggest that modulation of gut microbiota by pre-, pro- and synbiotics could reduce the liver fat accumulation and decrease the serum concentrations of hepatic enzymes. Several clinical trials are ongoing in obese and NAFLD patients (FOOD4GUT, INSYTE studies) in larger cohorts. They could bring mechanistic insight into the role of the gut microbiota in the management of liver disorders by prebiotics and/or probiotics. Further research should consider the limitations of biomarkers currently used for the diagnosis and progression of NAFLD, in addition to the inherent challenges of personalised microbial-based therapies. The identification of new classes of bioactive bacterial metabolites paves the way to development of nutritional strategies aiming to control their production by the microbiota and ultimately tackle hepatic liver diseases and associated metabolic dysfunctions. Among these metabolites, phenylacetic acid already appears as a strong candidate. Such a metabolite was identified using metabolomics, reinforcing the interest and the relevance of non-targeted metabolomic approaches to think ‘outside the box’ the relationship between the gut microbiota and host health.
Acknowledgements
This work is supported by Wallonia (FOOD4GUT project, convention 1318148). N. M. D. is a recipient of an EU grant (613979 MyNewGut, Project) and Belgium National Scientific Research Fund (FRS-FNRS, Belgium) grants. C. K. is a beneficiary of the AgreenSkills+ fellowship program which has received funding from the EU Seventh Framework Programme under grant agreement No. FP7-609398.
Financial support
None.
Conflict of Interest
None.
Authorship
N. M. D. conceived, drafted and supervised the manuscript. C. K., M. B., J. R., A. N. and L. B. B. drafted sections of the manuscript.
