SCFAs, such as acetic, propionic and butyric acid, are generated through the fermentation of polysaccharides by gut microbes in the large bowel. The diet and composition of the gut microbiome impacts the quantity and type of SCFAs synthesized in the gut. Diets high in fibre, particularly from plant-based foods or a Mediterranean diet, are associated with increased levels of faecal SCFAs [37,38]. Altering the gut microbiome with prebiotics and probiotics that promote the growth of beneficial microbiota can induce changes in SCFA production [39,40]. SCFAs have a role in inflammation, lipid and glucose metabolism and regulation of energy harvested from the diet [41]. As the effects of SCFAs are so diverse and widespread, elucidating the overall impact has been difficult. While animal studies have shown diets high in SCFAs to have favorable metabolic affects (reduced hepatic cholesterol and fatty acid synthesis and increased lipid oxidation), human studies have not been so clear [42,43].

SCFAs are a major source of energy, although most SCFAs are utilized in the gut, some are transported through the portal vein and channelled into the tricarboxylic acid cycle or utilised for hepatic gluconeogenesis or lipogenesis. Thus, changes in the gut microbiome that favor SCFA production can increase energy delivery to the liver and reduce faecal energy loss. The first evidence to support this was provided by Turnbaugh et al. [44] who found that, compared to their lean littermates, obese mice had more carbohydrate metabolising genes in their gut microbiome, increased concentration of SCFAs in their caecum, and less energy in their stool. Furthermore, faecal microbial transplant (FMT) from obese mice to germ free mice caused greater fat gain than FMT from lean animals. In humans, increased production of SCFAs by the gut microbiome was also observed in obese people, compared to lean subjects [45].

SCFAs can alter lipid and glucose metabolic pathways through activation of G-coupled receptors. Kimura et al. [46] showed that through activation of G-protein receptor-43, SCFAs suppress insulin signalling in adipocytes, which inhibits fat accumulation in adipose tissue and promotes the metabolism of unincorporated lipids and glucose in the liver and other tissues. The intestinotrophic effects of SCFA were first proposed by Koruda et al. [47] who found that SCFA supplementation to rats having parenteral nutrition prevented associated mucosal atrophy. Beneficial effects of SCFA on the intestinal mucosa are thought to be mediated through glucagon-like-peptide-2 (GLP-2), as SCFA supplementation induces the expression of ileal proglucagon mRNA and plasma GLP-2 [48]. In mice, increasing GLP-2 levels, by microbial intervention or subcutaneous GLP-2 administration, reduces intestinal permeability, which leads to lower plasma lipopolysaccharide (LPS) and cytokines levels, and consequently reduced hepatic oxidative stress and inflammation [49].

SCFA supplementation has shown beneficial effects on several inflammatory conditions including asthma, arthritis and colitis [50,51]. The anti-inflammatory effects of SCFA are thought to be mediated through activation of G-protein coupled receptor-43. Maslowski et al. [50] induced colitis in mice and demonstrated that inflammation is more severe in germ-free mice than in those conventionally raised. The colitis was ameliorated by acetate supplementation, a finding that was absent in G-protein coupled receptor-43 knock out mice [50]. By suppressing colitis, SCFAs can improve gut permeability and therefore reduce hepatic delivery of harmful microbial cell components and metabolites. Conversely, Rau et al. [52] found that patients with NAFLD had higher concentrations of intestinal acetate and propionate, which correlated with peripheral levels of pro-inflammatory T-cells. Therefore the role on SCFAs in inflammatory pathways within NAFLD is still controversial.

Histologically, NAFLD and alcohol-induced liver injury are remarkably similar, and are therefore likely to share common pathogenic pathways [53]. Some microbiota harbour genes that can ferment dietary sugars into ethanol. The amount of alcohol produced depends on the availability of carbohydrates from the diet and the composition of the gut microbiome, particularly the presence of Proteobacteria (especially Klebsiella pneumoniae and Escherichia coli) [28,54]. Animal studies have demonstrated that obese mice with NASH have higher early-morning breath alcohol content compared with lean mice without NASH; which is eliminated by neomycin treatment [55]. In human studies, patients with NAFLD, particularly children, have been shown to have increased blood ethanol levels [28,56,57]. Yuan et al. [54] found high-alcohol-producing Klebsiella pneumoniae in the gut microbiome of up to 60% of individuals with NAFLD in a Chinese cohort. The study went on to show that transfer of high-alcohol-producing Klebsiella pneumoniae, by oral gavage or FMT, into healthy mice induced NAFLD [54]. Gut-derived ethanol is not only directly hepatotoxic but it may alter the gut-liver axis, by increasing intestinal permeability and endotoxemia, to compound hepatic damage [58].

Choline is an essential nutrient sourced from foods such as meat and eggs. Choline deficiency induces many features of NAFLD and is often utilized in studies to create animal models of NAFLD. Choline deficiency leads to impaired synthesis of phosphatidylcholine resulting in diminished very low density lipoprotein assembly and secretion and consequently reduced hepatic triglyceride clearance [59]. In addition to hepatic steatosis, choline deficient mice also develop hepatic inflammation and fibrosis, which is thought to be caused by impaired mitochondrial β-oxidation and increased oxidative stress [60,61].

The gut microbiome plays an important role in choline metabolism and therefore regulates the concentration that is delivered to the liver via the portal circulation. Dysbiosis, and in particular an excess of the class Erysipelotrichia, has been associated with choline depletion in both animal and human studies [25,62,63]. Choline is metabolised by the gut microbiome to generate methylamines such as trimethylamine (TMA), which is converted in the liver to TMA-N-oxide (TMAO) by flavin-containing monooxygenase-3 (FMO3). TMAO has adverse effects on glucose homeostasis and is implicated in atherosclerosis [62]. NAFLD is associated with lower levels of choline and higher levels of TMA in the blood, implicating the role of gut microbiota in the imbalance of choline metabolism [63]. Around 10–15% of bacterial species require choline to synthesize phosphatidylcholine, a component of their membrane; therefore in the context of bacterial overgrowth, the demand for choline may increase and contribute to choline deficiency [64,65].

LPS, a constituent of gram negative bacteria, are found in the systemic circulation of individuals with NAFLD, and correlate with the severity of steatohepatitis in both human and animal studies [66-69]. The concentration of LPS in the gut microbiome and plasma is increased by the consumption of a western diet, high in saturated fat and low in fibre [70,71]. It has been hypothesised that a high-fat diet may facilitate LPS uptake through elevated chylomicron production in intestinal epithelial cells [72]. In mice on a standard diet, continuous subcutaneous infusion of low-dose LPS results in hepatic steatosis and hepatic insulin resistance [73,74]. LPS interact with toll-like receptors (TLRs), particularly TLR-4, on hepatic Kupffer cells and stellate cells to stimulate pro-inflammatory and profibrotic pathways via a range of cytokines, including interleukin (IL)-1, IL-6, and tumour necrosis factor (TNF) [75-78]. TLR-4‐deficient mice display decreased liver injury, inflammation, and lipid accumulation in comparison with wild‐type mice in NAFLD models [79,80]. TLR signalling in the mucosa also lead to the production of inflammasomes (cytosolic multiprotein oligomers of the innate immune system) which initiate a variety of pathways to produce proinflammatory and pro-fibrotic mediators such as caspase-1, IL-1β, and IL-18 [81,82]. In support of a role for inflammasomes in the development of more severe liver disease in NAFLD, Wree et al. [83] found significantly higher levels of the inflammasome NLRP3 (NOD-, LRR- and pyrin domain-containing protein 3) in the livers of individuals with NASH compared to those with simple steatosis. Conversely, other studies have shown that the absence of inflammasomes is associated with more aggressive disease. Pierantonelli et al. [84] demonstrated that, in a Western lifestyle model, the combination between high-fat and high-carbohydrate diet and the lack of the NLRP3-inflammasome increased the degree of liver injury and was associated with an abundance of gram-negative Proteobacteria and Verrucomicrobia (mucus-degrading bacteria that promote bacterial translocation), higher intestinal bacterial translocation, increased TLR activation and a more severe degree of liver injury.

BAs have been implicated in the pathogenesis of NAFLD through their role in lipid metabolism and as signalling molecules via the farnesoid X receptor (FXR). By binding to the FXR, BAs can increase insulin sensitivity and decrease hepatic gluconeogenesis and circulating triglyceride concentrations [85]. Obeticholic acid, an FXR agonist, has been shown to improve hepatic histology in ‘The Farnesoid X nuclear receptor ligand obeticholic acid for non-cirrhotic, non-alcoholic steatohepatitis (FLINT)’ trial [86]. However, results with this FXR agonist were less promising in a planned interim analysis of a further large trial in patients with NASH, where treatment with obeticholic acid failed to produce resolution of NASH (compared to placebo), although treatment with obeticholic acid 25 mg/day significantly improved liver fibrosis [87].

BAs and the gut microbiome have a bi-directional relationship and are both highly influenced by the diet. Dietary fat content regulates BA synthesis; in particular a high-fat diet increases production, while a low-fat diet reduces production [88,89]. The gut microbiome regulates BA homeostasis through a number of mechanisms including dihydroxylation of primary BAs to secondary BAs. BAs reciprocally regulate the gut microbiome, both directly via antibacterial activity (particularly deoxycholic acid) and indirectly via FXR induced antimicrobial peptides. The gut microbiome and BAs indirectly interact through TMAO, a product of gut microbiome choline metabolism. TMAO inhibits two key enzymes involved in BA metabolism: CYP7A1 and CYP27A1, therefore reducing the overall BA pool size [90]. Furthermore, FXR activation can regulate FMO3 activity and therefore TMAO production. Bennet et al. [91] demonstrated that FXR ligands administered to wild-type mice could induce FMO3 expression and increase TMAO levels, but this effect was abrogated in FXR knock out mice.

Several studies have demonstrated that the interaction of BAs with the gut microbiome plays an important role in NAFLD pathogenesis. Parséus et al. [92] fed germ-free and conventionally raised wild-type and FXR knock-out mice a high-fat diet for 10 weeks. The gut microbiome promoted weight gain and hepatic steatosis in an FXR-dependent manner, and the BA profiles and gut microbiome composition differed between FXR knock-out and wild-type mice [92]. These findings were supported by Jiang et al. [93] who found that improvements in hepatic steatosis related to antibiotic treatment were dependent on FXR signalling.

The portal vein exposes the liver to potentially harmful substances derived from the gut, including translocated bacteria and toxic bacterial products such as LPS. Dysbiosis can disrupt the integrity of the mucosal wall, increasing mucosal permeability. Dietary factors have an important role in the maintenance of the intestinal mucosal barrier, and not just through their role in altering the gut microbiome composition. Dietary depletion of glutamine, tryptophan and zinc or excess ingestion of fat, alcohol and food additives, have been directly associated with increased intestinal permeability [94-98].

Animal studies have shown convincing evidence that increased intestinal permeability leads to translocation of bacteria and bacterial toxins, such as LPS, that worsen the severity of NASH [66,73]. Miele et al. [99] found that patients with NAFLD had significantly increased gut permeability (as measured by urinary excretion of 51 chromium-radiolabeled ethylenediaminetetraacetic acid) compared to healthy volunteers. Furthermore, in patients with NAFLD both gut permeability and the prevalence of small intestinal bacterial overgrowth correlated with severity of steatosis, although not with steatohepatitis [99]. Verdam et al. [100] found plasma immunoglobulin G levels against endotoxin were increased in biopsyproven human NASH and positively correlated with the severity of inflammation. Conversely, Yuan et al. [101] found that in paediatric patients with NASH, serum endotoxin levels were not correlated with disease severity, however peripheral endotoxin levels may not represent the concentration in the portal system, and the pathogenesis of NAFLD in children may be different to adults.

Adipose tissue expansion, dysfunction, and inflammation are hallmarks of obesity and play a critical role in the development of NAFLD. Through several mechanisms the composition of the gut microbiome is thought to alter the metabolism and function of adipose tissue. LPS and other gut microbiome derived TLR ligands have been shown to contribute to adipose tissue inflammation. Caesar et al. [102] showed that mice fed a lard diet had increased TLR signalling and white adipose tissue inflammation compared with mice fed an isocaloric fish oil diet. In addition, mice genetically deficient in various components of the TLR signalling pathway were protected against white adipose tissue inflammation. de Groot et al. [103] demonstrated that FMT from a donor after Rouxen-Y bariatric surgery, to a host with metabolic syndrome led to a reduction in the adipocyte inflammatory marker chemokine ligand-2 gene expression in adipose tissue and circulating levels in plasma.

SCFAs, in particular propionate, have been demonstrated to regulate adipose tissue metabolism, both directly through modulation of the transcription factor peroxisome proliferator-activated receptor-γ, and indirectly through activation of the sympathetic nervous system [43,104]. The gut microbiome composition alters adipose tissue thermogenesis, including brown adipose tissue activity and browning of white adipose tissue. Schugar et al. [105] demonstrated that in mice the genetic deletion of the TMAO-producing enzyme FMO3 protected against high-fat diet induced obesity, in part by stimulating the beiging and enhanced thermogenesis of white adipose tissue.

Increased dietary fat intake, particularly saturated fat, is associated with the development of hepatic steatosis [107-109]. A high fat diet can alter the gut microbiome rapidly, decreasing the microbial diversity and favoring gut bacteria associated with the development of NAFLD [110]. Changes associated with a high fat diet include reduced levels of the Bifidobacterium genus and an increase in the ratio of Firmicutes to Bacteroidetes phyla [73]. Ingestion of long-chained saturated fats, compared to short-chained unsaturated fats, are associated with a more pronounced reduction in phylogenetic diversity and beneficial bacteria [102]. The mechanisms by which dietary fatty acids affect the gut microbiome are not well defined. Only a minority of fatty acids will pass through the gastrointestinal tract and directly modulate gut microbiome composition. Fatty acids have a broad spectrum of antibacterial activity including lysis of bacterial cell membranes and inhibition of bacterial adenosine triphosphate (ATP) production [111,112], and the antibacterial action of fatty acids is affected by carbon chain length, saturation and double bond position [113].

Changes in the gut microbiome caused by a high fat diet are associated with; increased energy harvest from the gut, upregulation of genes related to lipid metabolism in the distal small bowel and the production of SCFAs favoring the development of NASH [73,102]. High fat diets in animals, particularly those involving long-chained saturated fatty acids, are accompanied by increased intestinal permeability, resulting in bacterial translocation and elevated LPS levels [73]. Inflammatory processes are also upregulated, with increased hepatic TLR-4 activation and a rise in inflammatory mediators including TNF-α, IL-1, and plasminogen activator inhibitor-1 [73,102].

Despite the association between a high fat diet and the development of NAFLD and dysbiosis, some fatty acids have demonstrated favorable effects on the gut microbiome and the development of hepatic steatosis. In a randomized controlled trial (RCT), Scorletti et al. [114] demonstrated that consumption of the omega-3 polyunsaturated fatty acid, docosahexaenoic acid, was independently associated with a decrease in liver fat percentage in patients with NAFLD. The ingestion of omega-3 polyunsaturated fatty acids and oleic acid have been shown to protect against high-fat diet induced dysbiosis and improve the composition of the gut microbiome; including increased abundance of butyrateproducing bacteria and the Bifidobacterium genus [115-117].

Dietary fructose has been strongly implicated in the pathogenesis of NAFLD. A large association study demonstrated an increased risk of NAFLD in those consuming regular sugary drinks, especially if overweight [118]. Fructose is both a substrate and an inducer of hepatic de novo lipogenesis. In individuals with NAFLD, 26% of hepatic triglycerides are produced by de novo lipogenesis using fructose and other dietary sugars as substrates [119]. Fructose increases hepatic lipogenesis by activating several key transcription factors such as sterol response element-binding protein-1c and carbohydrate-responsive element-binding protein [120]. Dietary fructose can induce rapid and harmful changes in the gut microbiome composition, including a reduction in the phylogenetic diversity and a lower concentration of Bifidobacterium genus [121]. Fructose induced dysbiosis has been demonstrated to increase intestinal macrophage counts and lower tight junction occludin protein expression, associated with worse endotoxaemia, more bacterial translocation and increased hepatic TLR expression [122].

During hepatic fructose metabolism by fructokinase, ATP is rapidly consumed, which results in the breakdown of adenosine monophosphate (AMP) to inosine monophosphate and the generation of uric acid. Uric acid therefore increases in the plasma and liver with fructose consumption, and may mediate some of the adverse effects associated with fructose [123]. Uric acid can exacerbate hepatic insulin resistance through activation of retinol binding protein-4 [123], impair hepatic fatty acid oxidation through inhibition of AMP-activated protein kinase, and induce hepatic oxidative stress through NADPH oxidase activation [124]. Xanthine oxidase inhibitors that block uric acid generation, have been shown to inhibit fructose induced hepatic steatosis [125], and serum uric acid concentrations and a high dietary fructose consumption are both independently associated with NASH in children [126].

AGEs are formed in food when reducing sugars react non-enzymatically with the amino groups on proteins. The concentration of AGEs is high in western diets and contributes to tissue injury via activation of receptor for AGEs (RAGEs) and generation of reactive oxygen species. RAGEs have been found on hepatic stellate cells and stimulation has been shown to exacerbated liver inflammation and increase hepatic proliferation and expression of collagen in animal models of NAFLD [127]. The absorption of dietary AGEs is limited, the majority pass through the gastrointestinal tract to the colon, where they can act as substrates for the gut microbiome [128]. Several animal and human studies have shown that the consumption of AGEs is associated with a compositional change in the gut microbiome and altered production of SCFAs, however there is a lack of agreement between studies on the specific microbial changes, which may be due to the different glycated substrates used [128].

A Mediterranean diet is characterised by increased consumption of vegetables, legumes, fruits, nuts, olive oil and fish and low consumption of red meat, dairy products and saturated fats. A weight neutral Mediterranean diet can reduce liver steatosis and may improve insulin sensitivity [129,130]. A Mediterranean diet contains a high concentration of monounsaturated fatty acids and a balanced polyunsaturated fatty acid omega-6 to omega-3 ratio, which has favorable effects on hepatic lipid metabolism [131]. The high levels of polyphenols, carotenoids, vitamin E and vitamin D are protective against the development of NAFLD through their antioxidant and immunomodulatory properties [132-135]. Some of the hepatoprotective effects of a Mediterranean diet may be mediated through beneficial changes in the gut microbiome composition, including; increased levels of the Bifidobacterium genus, reduced levels of gram-negative LPS containing bacteria, altered SCFA production and reduced toxic metabolites such as TMAO [37,136].

NCS are widely used by patients with the metabolic syndrome; however their safety and side effect profile remains a topic of controversy. The use of NCS have been linked to the pathogenesis of obesity, T2DM and NAFLD through a number of postulated mechanisms. In an animal study the use of NCS weakened the ability of sweet taste to predict energy and evoke autonomic and endocrine learned responses, such as the cephalic response, that prepares the digestive tract to optimally deal with ingested food [137]. Other animal studies have shown that NCS interact with sweet taste receptors expressed in enteroendocrine cells and increase intestinal glucose absorption (through sodium-dependent glucose transporter isoform 1 and GLUT-2) leading to obesity, hyperinsulinemia and insulin resistance [138,139].

Many NCS are associated with changes in the composition of the gut microbiome, while others have shown no effect [140]. Suez et al. [141] found that saccharin intake in mice was associated with reductions in intestinal Akkermansia muciniphila, a commensal bacterium that exhibits probiotic properties. These mice developed impaired glucose tolerance, an effect that was abrogated by antibiotic treatment, and fully transferrable by FMT. Trocho et al. [142] found that aspartame accumulates in the liver of both healthy and cirrhotic rats and might increase the risk of NAFLD via mitochondrial dysfunction and ATP depletion in the liver. Conversely, other animal studies have shown that NCS are not associated with hepatic steatosis, and some, such as xylooligosaccaride, may actually be protective [143].

Green tea and its polyphenols, such as epigallocatechin gallate, have demonstrated protective effects against the development of NAFLD. The consumption of green tea in mouse models of NAFLD has led to reductions in; hepatic steatosis, hepatic inflammation, hepatic fibrosis and insulin resistance [144-147]. There are various mechanisms by which green tea may exert these beneficial effects. Consumption of green tea has been found to restore the changes in gut microbiome composition, such as the Firmicutes to Bacteroidetes ratio, which are associated with the development of obesity, hepatic steatosis and insulin resistance [148]. Several studies have also demonstrated that green tea has anti-inflammatory effects, possibly through interaction with the 67-kDa laminin receptor [149]. Green tea can also act as an anti-oxidant, directly by scavenging reactive oxygen and nitrogen species [150], and indirectly by upregulating the transcription of genes related to the cellular antioxidant defence [151]. In a placebo-controlled RCT, green tea consumption was associated with significantly reduced alanine transaminase and aspartate transaminase levels after 12 weeks, in individuals with NAFLD [152]. In contrast, a cross-sectional study of 1,024 Japanese men did not find an association between green tea consumption (≥3 cups of green tea a day) and hepatic steatosis diagnosed by ultrasonography [153].

There is evidence suggesting that caffeine and coffee may be protective against the development of several elements of the metabolic syndrome, including NAFLD [154,155]. A large meta-analysis found both caffeinated and decaffeinated coffee consumption to be inversely associated with the risk of T2DM in a dose-response manner [156]. The mechanism of action is thought to be largely mediated by an ester of caffeic acid named chlorogenic acid, which improves glucose metabolism by inhibiting gut absorption of glucose and hepatic gluconeogenesis [157]. Although chlorogenic acid is not specific to coffee and occurs in other food stuffs another meta-analysis found a significantly decreased risk of both developing NAFLD and it progressing to fibrosis in individuals who drank coffee on a regular basis [158]. Caffeine is thought to have antifibrotic properties, through antagonism of adenosine receptor A2a, which inhibits hepatic stellate cells [159]. Caffeine also leads to more rapid gut transit and decreased energy harvest from the diet, in part through reducing the expression of aquaportin-8, a water channel protein expressed in the intestinal mucosa [160]. Consumption of coffee can alter the gut microbiome composition, in healthy volunteers and consumption of three cups of instant coffee per day was associated with an increase in the quantity and metabolic activity of Bifidobacterium, a genus of bacteria thought to be protective against the development of NALFD [161].