Microbial Dysbiosis and Nonalcoholic Fatty Liver Disease
Microbial Dysbiosis and Nonalcoholic Fatty Liver Disease
We identified nine original research studies (five in humans and four in animal models) that characterised microbial communities in the setting of NAFLD. This review will initially discuss important background concepts of dysbiosis, then we will give a detailed overview of the nine original research studies that directly characterise the microbiome in the setting of NAFLD, and finally, we will discuss current mechanistic concepts linking the intestinal dybiosis with NAFLD.
Molecular Assessment of the Microbiome: Technical Aspects. Advances in culture-independent microbiologic technology over the last decade have facilitated the characterisation of both the composition and diversity of the intestinal microbiota. Previously, assessment for particular intestinal inhabitants required direct culture, which was slow and labour intensive. Consequently, only a small fraction of the intestinal inhabitants could be directly cultured under routine conditions. Now, with sequencing of highly conserved genes or entire genomes, the vast array of microbes can be more incisively profiled, even in complex environments, such as the distal bowel. As ribosomes are essential to all cellular life, the genes encoding ribosomal components, most notably the small-subunit ribosomal RNA (rRNA) gene, have become the gold-standard for delineating the phylogenetic relationships among cellular life. Specifically, the 16S rRNA gene sequences for archaea and bacteria are used as stable phylogenetic markers to define which lineages are present in a sample. Figure 1 illustrates the general work-flow leading from sample collection, extraction of bulk genomic DNA, pan-microbial 16S rRNA gene PCR amplification, next-generation sequencing, to the bioinformatic output that describes the proportional distribution of microbial taxa in the sample set.
(Enlarge Image)
Figure 1.
Workflow for 16S rRNA sequence-based profiling of microbial communities. Following specimen collection, bulk genomic DNA is prepared using robust extraction protocols. This DNA is subjected to broad range PCR amplification using primers that amplify all bacterial 16S rRNA genes (other primer sets can be used to profile archaea, fungi or all organisms). PCR amplicons are sequenced using next-generation sequencing platforms. The resulting sequence datasets are quality-filtered and each sequence assigned a taxonomic classification using phylogenetic analysis software. The results provide an assessment of the types and relative abundances of microorganisms in the sample set.
The Challenges of Determining a 'Normal' Human Core Microbiome. In the process of trying to investigate dysbiosis and disease associations, it is necessary to understand the types and distributions of microbes that colonise ostensibly healthy individuals. However, composition and dynamics of this 'normal' microbiome remains incompletely defined. A pivotal study by Turnbaugh et al. investigated the gut microbiota of 154 adult individuals including monozygotic twins, dizygotic twins, and their mothers. A key question that was addressed was whether a 'core' intestinal microbiome of specific microbial lineages are commonly shared. The investigators found that no single bacterial species was detectable at an abundant frequency (defined as >0.5% abundance in the community) in the guts of all 154 individuals. Some similarities were found among family members, but each individual's microbial community displayed a unique pattern at the species level. However, there did appear to be a core microbiome at the functional level based on deep metagenomic sequencing.
Further complexity is apparent when one considers the temporal and spatial distributions of microorganisms within an individual's own intestinal ecosystem. Eckburg et al. studied the microbiota in detail in three individuals with samples taken by stool collection and then also by direct mucosal biopsy at six sites throughout their colonic tract. The investigators found significant intra-subject differences between the stool samples and more directly obtained mucosal community samples. At the species- and strain-levels, the profound variability between and within individuals' gastrointestinal tract complicates attempts to define what is 'normal' microbiota. This diversity remains a significant challenge in studies of the microbiome in human diseases.
Given the diversity at the individual species level, some studies have focused on assessing broader patterns at higher taxonomic levels, such as genus and phylum, when assessing for dysbiosis linkage with disease. The intestinal microbiota of adult humans is composed predominantly of two bacterial phyla, Bacteroidetes and Firmicutes. As discussed below, changes in the relative abundances of Bacteroidetes and Firmicutes have been observed in obesity and the metabolic syndrome, with an increased abundance of Firmicutes being associated with obesity in some studies. Defining these broader patterns of phylum composition have been important initial steps in trying to understand alterations in the microbiome that may be associated with disease states. However, how the diversity of the vast array of specific bacterial species may play a role in disease processes in humans is an area of future research.
Dysbiosis and Obesity: A Brief Overview. Studies of intestinal dysbiosis have highlighted an association with obesity. The first evidence for alterations in microbiome ecology in the setting of obesity was reported in mouse models. Support for the microbiome affecting body weight initially was produced through experimental transfer of the intestinal microbiota from normal, conventionally raised mice into germ-free mice, which led to an increase in body fat with no change in food consumption. Ley et al. subsequently reported that leptin-deficient obese mice (ob/ob), in comparison with lean wild-type mice, exhibited a shift towards higher relative abundance of the bacterial phylum Firmicutes, with concomitantly lower abundance of the phylum Bacteroidetes. Although dysbiosis has also been observed in multiple studies of human obesity and weight change, study results have been conflicting with respect to the microbial phylogenetic groups that are altered in abundance, and several studies have found no significant correlations between body mass index and the ratio of abundance of Firmicutes to Bacteroidetes. Furthermore, understanding whether obesity alters the microbiome, or if the microbiome alters risk for obesity in humans is a question that requires further longitudinal study.
Nonalcoholic fatty liver disease has been linked unequivocally with obesity and is widely considered the hepatic manifestation of the metabolic syndrome. And more recently, NAFLD has been associated with dysbiosis. At a basic level, NAFLD might be influenced by the microbiome simply via the increased risk for obesity mediated by increased caloric intake. However, the relationship between NAFLD and intestinal dysbiosis is likely much more complex and driven by multiple molecular pathways. Below, we review nine original research studies (five human, four animal models) that characterise the microbiome in the setting of NAFLD.
NAFLD and Dysbiosis in Animal Models. Animal studies have been the first step in critically evaluating the role of intestinal dysbiosis in the risk for NAFLD. Early animal studies examined changes in the microbiome occurring with disease processes such as obesity, and interventions such as high fat diet feeding. Subsequent investigations have more directly assessed the effects of transmitting dysbiotic microbiomes to recipients and then measuring the effect on the recipient phenotype in the context of risk for NAFLD. Zeng et al. studied a murine model in which mice were fed a high fat diet for 10 weeks. The high fat diet fed mice increased their body weight by 34% compared to the low fat fed mice and the livers of the mice fed the high fat diet also exhibited dramatic increases in the number of lipid droplets, inflammatory cell infiltration and inducible nitric oxide synthase protein concentration. The authors also reported that a high fat diet increased the levels of lactobacillus species in mouse faecal material when compared to the group of mice fed a low fat diet for 10 weeks. A positive correlation was noted between the amount of lactobacillus DNA in the faecal samples and the severity of steatosis within mouse livers. The authors postulated that the lactobacillus species could have an effect on lipid metabolism through effects on bile acid metabolism, and thus contributed to the risk for fatty liver.
Other animal studies have assessed the effects of altering the microbiota using probiotics on metabolic risk factors linked with NAFLD. Cano et al. studied both wild-type mice and high fat diet-induced obese mice. Both groups were given either placebo or a probiotic consisting of Bifidobacterium pseudocatenulatum for 7 weeks. Changes in metabolic parameters, including insulin resistance, were measured between the mouse groups and samples were obtained for liver histology to determine the degree of steatosis pre- and post-intervention with the probiotic or placebo. A high fat diet supplemented with bifidobacteria was associated with lower food intake and also lower body weight compared with the placebo group. The authors also found that bifidobacteria supplementation resulted in decreased insulin resistance, decreased hepatic steatosis, and a reduction in serum inflammatory markers compared with the high fat diet fed mice that were not supplemented. In contrast, the wild-type mice fed a control diet showed no difference in these metabolic and inflammatory parameters, regardless of the presence or absence of probiotic supplementation. The authors also reported differences in gut microbial composition between the mouse groups; specifically, the addition of the probiotic increased the number of bifidobacteria in the high fat diet group and reduced the number of Enterobacteriaceae spp. The authors concluded that the probiotic was able to ameliorate the metabolic and immunological dysfunction, including hepatic steatosis, related to obesity in the high fat diet fed mouse model.
Henao-Mejia et al. studied mouse models deficient in the pro-inflammatory multi-protein complexes, termed inflammasomes, to explore the potential role of the microbiome in NAFLD. In a prior study, the authors had found alterations in the faecal microbiome in inflammasome-deficient mice characterised by expanded representation of the bacterial phyla Bacteroidetes. To explore how these changes in the microbiome possibly influenced NAFLD, they utilised several mouse models of NAFLD (methionine choline-deficient diet model, the genetic leptin receptor deficiency steatosis model, and the high fat diet model) within the inflammasome deficient mice. All NAFLD models exhibited an exacerbated NAFLD phenotype in the inflammasome-deficient genetically modified mouse strains compared with wild type. They found activation of pro-inflammatory pathways through influx of TLR4 and TLR9 agonists into the portal circulation, which then led to enhanced hepatic tumour-necrosis factor (TNF) within the liver. The authors further observed that co-housing and thus transferring the dominant microbiome from the inflammasome-deficient mice to wild type exacerbated the phenotype of NAFLD and NASH in the co-housed wild-type NAFLD mice models. The authors concluded that transfer of a dysbiotic microbiome worsened hepatic steatosis and thus potentially affected NAFLD progression, thus causally linking dysbiosis with NAFLD pathogenesis.
Another murine investigation conducted by Le Roy et al. demonstrated that the risk of developing NAFLD was transmissible by faecal transplant to recipient mice. The investigators exploited an initial observation that conventionally raised mice display variable degrees of weight gain, hyperglycaemia and insulin resistance when exposed to a high fat diet. The investigators selected two types of donor mice: one donor mouse that demonstrated weight gain with increased systemic inflammation (increased serum levels of MCP-1 and TNF-α) and increased insulin resistance on a high fat diet; and a second donor mouse that had weight gain on the high fat diet, but did not display increased inflammation or insulin resistance. They transferred faecal samples from each of the donors into sets of germ-free mice, which subsequently took on features of the phenotypes of their faecal donors. In this experiment, the effect of the transfer of the inflammatory and insulin resistance phenotypes via the donor microbiome was independent of obesity induced by the high fat diet. The mice that developed the inflammatory and insulin resistance phenotype were also characterised by increases in hepatic macrovesicular steatosis. Interestingly, no increase in inflammatory infiltrate was noted within the livers of these mice, which might suggest that this faecal transfer led to elevated risk for NAFLD, but not necessarily NASH. The short follow-up duration may also have been inadequate to detect the eventual inflammatory damage that would signify NASH. The investigators also assessed specific compositional differences in the microbiota between the recipients of the inflammatory/insulin resistant faecal donor sample vs. the recipients of the non-inflammatory/non-insulin resistant faecal donor sample. They found distinct differences at the phylum, genus, and species levels. At the phylum level, increased Firmicutes were found in association with the inflammatory NAFLD prone phenotype, which is in concordance with increases in this phylum previously reported for other metabolic disorders, such as obesity. These initial mouse studies are of great interest given their implication that the intestinal microbiota may directly affect risk for NAFLD and NASH within the host.
NAFLD and Dysbiosis in Humans. To better understand how findings from animal studies might relate to human disease, several investigators have begun to assess whether gastrointestinal dysbiosis is associated with NAFLD risk in humans. Five original studies within the published literature have explored the intestinal microbiome among cohorts of human subjects. Table 1 summarises the overall characteristics of these studies and provides details regarding the specific microbiologic results. These published human studies have great variability in study design, population selection, and reporting of results, all of which complicate meta-analysis. Thus, the current review focuses on a qualitative summary of these studies.
Spencer et al. performed one of the first studies that explored the microbiome in relation to human NAFLD. They utilised a population of 15 normal healthy adult females within a larger longitudinal metabolic study that entailed an in-patient intervention of a choline depletion diet. Choline is an essential nutrient and low-choline diets are associated with health problems in humans, including hepatic steatosis. The authors assessed the faecal microbiome at baseline, after 10 days on a normal controlled diet, up two times during a 42 day period on a choline-depleted diet (which lasted only 42 days if they did not develop evidence of end organ damage), and then twice during a 10 day choline repletion diet. Serial faecal samples from study subjects were analysed by 16S rRNA gene sequencing. The longitudinal design allowed the authors to prospectively characterise alterations in microbiota that were associated with the fatty liver changes that developed in some individuals on the choline-depletion diet. The microbiome of each individual remained distinct, as evidenced by continued clustering by individual on serial samples, but shifts in microbial community profiles were observed in response to the choline-depleted diet. Most notably, higher baseline levels of the phylum Proteobacteria, specifically the class Gamma Proteobacteria, correlated with lower risk for developing fatty liver, whereas higher baseline levels of the class Erysipelotrichia (a member of the phylum Firmicutes) correlated with higher risk for fatty liver. The authors concluded that the baseline levels of the bacteria on the ad libitum normal diet modified the degree of the subject's susceptibility to fatty liver when on a diet deficient in choline. It is important to note that the changes in fatty liver composition within this study were measured by changes assessed by MRI, and not by serial histology. Thus, this study design did not assess the risks for inflammation with NASH.
Zhu et al. characterised the intestinal microbiome composition within a group of 63 children (mean age approximately 13 years) by 16S rRNA gene sequencing. The authors assessed cross-sectional differences between faecal samples of 22 children with NASH found on biopsy, 25 clinically obese children with no clinical suspicion of NASH, and 16 healthy normal weight control children. When compositional differences were measured at the phylum level, decreased levels of Firmicutes and increased Bacteroidetes were observed within NASH and obese individuals compared with the controls, but the abundance of Bacteroidetes and Firmicutes were similar between patients with NASH and those with obesity. To attempt to address possible mechanisms of microbiota influence, the authors also measured serum ethanol levels in study participants and found increased ethanol levels in the NASH children. The authors postulated that increased Escherichia coli within the NASH group might have elevated ethanol levels, given that this group of microbes is able to produce ethanol. The authors also hypothesised that increased ethanol production by the microbiota could lead to chronic, low-level exposure to this hepatotoxin, thus putting the individuals at risk for steatohepatitis. These investigators reported in a separate publication from the data on this patient population that no differences in the levels of endotoxemia were observed between these paediatric groups of NASH, obese, and controls. This finding led the authors to postulate that endotoxemia may not be required for the pathogenesis of paediatric NASH.
In another study, Mouzaki et al. recruited a group of 50 adults: 11 with biopsy-proven 'simple' steatosis, 22 with biopsy-proven NASH, and 17 healthy controls with no steatosis by biopsy. They utilised quantitative polymerase chain reaction with custom-made primer/probes that were specific for total bacterial counts and also identified the following five specific groups of bacteria: Bacteroides/Prevotella, Ruminococcaceae (Group IV clostridia), Lachnospiraceae (Group XIVa clostridia), E. coli, and Archaea. The authors found significantly decreased levels of Bacteroidetes in patients with NASH compared to individuals with simple steatosis and healthy controls, even after adjusting for body mass index. They also reported that NASH patients had an increased number of Lachnospiraceae species compared with the simple steatosis group, but the significance of this difference was lost after adjusting for BMI and fat intake. The strengths of this study included the use of liver biopsies to confirm the histological diagnosis for each of the patient groups. However, the cross-sectional nature of the study, which is common for human microbiome studies, was a limitation as far as drawing conclusions regarding causality.
Raman et al. investigated faecal microbiomes in a cross-sectional study of 60 adults: 30 obese individuals with clinically defined NAFLD (no biopsy available) and 30 non-obese controls. The authors reported that selected members of the phylum Firmicutes were increased in NAFLD, compared with non-obese controls. A strength of this study was that it measured levels of hepatotoxic volatile organic compounds (VOC), which have been hypothesised to promote the development of NAFLD, in parallel with the microbiome profiling. Indeed, elevated volatile compound levels were detected in the stools of the NAFLD patients, leading the authors to propose that these compounds increased as a consequence of dysbiosis. However, a limitation acknowledged by the authors was that their NAFLD group was clinically diagnosed and thus the findings could not be linked to the risk of NASH. An additional limitation was the lack of an obese control group without NAFLD. As the control group was composed of non-obese individuals, it was not possible for the investigators to determine whether the links between dysbiosis and liver disease were confounded by increased BMI.
A longitudinal microbiome study by Wong et al. investigated 20 adult patients with biopsy-proven NASH who were randomised to receive either a probiotic supplement consisting of a combination of Lactobacillus plantarum,Lactobacillus delbrueckii ssp bulgaricus,Lactobacillus acidophilus,Lactobacillus rhamnosus and Bifidobacterium bifidum (n = 10), or no probiotic supplementation (n = 10) for a total of 6 months. Both groups received standard of care counselling from an investigator to lose weight, reduce fat intake, and exercise at least three times per week. In the proof-of-concept assessment of treatment of NASH with probiotics, the investigators found a greater reduction in hepatic fat as measured by magnetic resonance spectroscopy (MRS) and also decrease in aspartate aminotransferase in the group supplemented with probiotics compared to the counselling only group at 6 months. Hepatic fat content decreased from 22.6 ± 8.2% to 14.9 ± 6.6% (P = 0.034) in the probiotic group and remained relatively stagnant in the counselling only group (16.9 ± 6.1% to 16.0 ± 6.6%, P = 0.55). To assess the microbiome changes over time in these populations, the faecal microbiota were profiled both at baseline and at 6 months in both groups. A total of seven patients from the probiotic group and nine from the no probiotic group (counselling only) had adequate samples to allow inclusion in the final analysis. The authors also enrolled a group of healthy controls (n = 22) to compare to the NASH patients. This design allowed the authors to identify baseline differences between NASH and controls, as well as changes in the microbiota over the 6-month study period in the longitudinally sampled group. At baseline, no differences in Bacteroidetes were apparent between groups, but reduced levels of Firmicutes were found in the NASH subjects, compared with the control subjects without NASH. After 6 months, reduced hepatic fat as measured by MRS was associated with an increase in Bacteroidetes and a decrease in Firmicutes, among the overall group of NASH patients.
Taken together, these human studies demonstrate measureable differences in the microbiome between clinical states of NAFLD and NASH, but due to the great variability study design, methods, and clinical endpoints, it remains challenging to interpret the clinical significance of intestinal dysbiosis in the context of liver disease. Further work is needed, particularly longitudinal studies, to delve further into the pathologic significance of these results in human NAFLD.
The aforementioned animal and human studies provide primarily descriptive characterisations of the microbiota within diseased populations. These and other investigators have also focused their attention on elucidating the mechanisms by which microbes may affect their hosts' metabolic processes and thereby modulate the risks for NAFLD and NASH. Several mechanisms have been postulated to explain the associations between NAFLD and dysbiosis. As noted above, Spencer et al. assessed the possible role of altered choline metabolism, Zhu et al. measured endotoxin and ethanol levels as possible hepatotoxins, and Raman et al. measured increased VOC s in stool as potential hepatotoxins. These studies are important early steps towards determining the mechanisms by which dysbiosis affects risk for NAFLD and NASH, beyond simply characterising the composition of the gut microbiome in the setting of NAFLD. Understanding underlying mechanistic relationships should open new avenues of research into novel, readily modifiable targets to reduce disease risk and progression.
Several mechanisms have been proposed within the literature, and highlighted by prior reviews, that may link the microbiome to NAFLD including the impact of microbiota on energy utilisation and the association with obesity; influences of microbiota on gut permeability that induce low grade inflammation; modulation of dietary choline metabolism by the microbiota; regulation of bile acid metabolism; and endogenous production of hepatotoxic substances such as ethanol and other VOCs by the microbiota. These postulated mechanisms, discussed further below, may potentiate each other through shared molecular pathways of fat deposition and activation of inflammation within the liver. Molecular pathways that are directly linked to the microbiome and have been shown to contribute to models of NAFLD include lipopolysaccharide (LPS) endotoxemia, induction of immune toll-like receptor (TLR) signalling, and inflammasome mediated pathways. Figure 2 highlights these potential mechanisms and pathways in NASH.
(Enlarge Image)
Figure 2.
Mechanistic pathyways in NAFLD pathogenesis. Highlighted pathways within the enterohepatic circulation that influcence progression of NAFLD and NASH. Dysbiosis influences several pathways that lead to the delivery of metabolic and inflammatory factors through the portal circulation with resulting activation of molecular pathways that promote lipid accumulation and inflammation. EtOH, ethanol; Fiaf, fasting induced adipocyte factor; FFA, free fatty acids; FXR, farnisoid X receptor; LPL, lipoprotein lipase; LPS, lipopolysaccharide; SCFA, short-chain fatty acids; TG, triglycerides; TLR, toll-like receptor; TMA, toxic methylamine; VLDL, very low density lipoprotein; VOC, volatile organic compounds.
Facilitation of Host Energy Harvest and Utilisation by the Microbiota. Several investigators have linked the microbiome with obesity. Much of this research has proposed that the microbiome affects energy extraction, absorption, utilisation, and storage, processes that also could have links with the pathogenesis of NAFLD. Further research has also suggested that dysbiosis may directly increase the absorption of short-chain fatty acids (SCFA), free fatty acids (FFA), and carbohydrates, with an associated up-regulation of ChREBP and SREBP-1c, along with inhibition of Fiaf [fating induced adipose factor, which is responsible for encoding the inhibitor of lipoprotein lipase (LPL)]. This leads to LPL activation resulting in increased lipogenesis and the development of the milieu for NAFLD. However, the direct effect of the microbiota on dietary energy absorption through increased SCFAs, FFAs and carbohydrates is just one link in the risk for NAFLD and NASH, as it is likely that other mechanisms are involved, particularly given that not all obese individuals develop NASH and simply increasing caloric intake is not sufficient for NASH pathogenesis. The SCFAs include: acetate (C2), propionate (C3) and butyrate (C4) and are the main metabolic products of anaerobic bacteria fermentation in the intestine. It is notable that the SFCAs not only serve as fuel for the intestinal epithelial cells but also act on leucocytes and endothelial cells to regulate several leucocyte functions including production of cytokines (TNF-α, IL-2, IL-6 and IL-10), eicosanoids and chemokines (e.g. MCP-1 and CINC-2). In general, SCFAs, such as propionate and butyrate, inhibit adhesion molecules, chemokine production and monocyte, macrophage and neutrophil recruitment, suggesting an anti-inflammatory action; however, there is also evidence favouring a pro-inflammatory action of SCFAs in some conditions. Thus, SCFAs may have a complex effect on the inflammatory process in NASH.
Altered Dietary Choline Metabolism by the Microbiota. Choline is an essential nutrient that has important roles in human metabolism, including: neurotransmitter synthesis, cell-membrane signalling and lipid metabolism. Because of this wide range of effects, dietary choline-deficiency has been linked with a variety of conditions including neurological developmental disorders, heart disease and liver disease. Diets deficient in choline can lead to increased hepatic steatosis, which can be reversed with choline supplementation. The intestinal microbiota plays a role in the conversion of dietary choline to toxic methylamine, a substance that both mimics a choline-deficient diet by decreasing effective choline levels and exposes the host to an inflammatory toxic metabolite. Spencer et al. showed that, in individuals on a choline-deficient diet, a particular baseline microbial composition correlated with the risk for the induction of hepatic steatosis, as noted on serial MRI scans. They found increased Gammaproteobacteria abundance and decreased Erysipelotrichi abundance at baseline were protective against developing steatosis. The risk for NASH was not assessed in this study. The theory that microbial dysbiosis can affect risk for NAFLD and NASH by altering the metabolism of an essential nutrient such as choline is intriguing and will require further study to better understand.
Impact of the Microbiota on Bile Acid Metabolism. Bile acids are synthesised from cholesterol through enzymatic pathways within hepatocytes and are then conjugated within the hepatocytes with either glycine or taurine before secretion into bile and released into the small intestine. In the small intestine conjugated bile acids not only assist in lipid absorption and transport but have also been increasingly recognised to function as nuclear receptor binders and to have a putative role in altering the microbiome. Bacteria within the intestine can also chemically modify bile acids and thereby alter the composition of the bile acid pool. The effects of the microbiome on the size and composition of the bile acid pool have been demonstrated by Swann et al., who used a germ-free mouse model to compare bile acid species in different mouse tissues. The authors found that, when compared to conventional mice, germ-free mice had significant differences in the distribution of bile acid species in several different tissues (liver, heart and kidney). Specifically, germ-free mice had increased taurine conjugated bile acids in liver, kidney, heart and plasma compared to conventional mice and comparable results were found using an antibiotic treated mouse model. Overall bile acid diversity was also found to be lower in both germ-free and antibiotic treated mice when compared with controls. Sayin et al., further observed that conventionally raised mice had smaller bile acid pools than did their germ-free counterparts. Other animal studies indicate that the host bile acid pool in turn influences the composition of the microbiome. Alterations in bile acid composition may be a mechanism by which the microbiome might affect risk for NAFLD and NASH.
Although bile acids have long been known to play a role as detergents to facilitate fat absorption, their role in human physiology is more complex and bile acids are now recognised as important cell signalling molecules that activate multiple pathways regulating lipid metabolism, carbohydrate metabolism and inflammatory response. These effects are mediated through their binding and activation of the nuclear hormone receptor, farnesoid X receptor (FXR), and the G protein coupled cell surface receptor TGR5. The signalling pathways triggered by FXR and TGR5 activation modulate bile acid synthesis, as well as the metabolism of glucose, triglycerides and cholesterol. Specifically, FXR activation results in inhibition of gluconeogenesis and glycogenolysis in the liver (improving glucose metabolism) and results in increased insulin sensitivity in adipose and skeletal muscle. Lipogenesis is also attenuated by FXR activation through down-regulation of SREBP-1c and an increase in beta fatty acid oxidation. It is notable that FXR knockout mice develop features of NASH, including hepatic steatosis and hepatic inflammation. FXR has also been associated with inhibition of NFkB activation suggesting an anti-inflammatory role for FXR, as well. Figure 3 illustrates associations between the bile acid pool, the microbiota and FXR activation.
(Enlarge Image)
Figure 3.
Bile acid modulation of FXR receptor alteration in the bile acid composition leads to activation of the FXR nuclear receptor, which stimulates transcription of FGF19 that is transported to the liver through the portal circulation and activates signalling pathways that downregulate bile acid synthesis through Cyp7A1 inhibition. This altered pool of bile acids may ultimately feedback and also favourably or unfavourably influence the microbiota composition within the intestine, though this purported association needs to be further defined with experimental data. The FXR and FGF19 triggered pathways also influence glucose and lipid metabolism through a decrease in gluconeogenesis and decrease in lipolysis.
Further insights into the complex relationships between dysbiosis and bile acid pathways relative to NAFLD and NASH have come from experimental data in mouse models. McMahan et. al., showed that activation of bile acid receptors with a receptor agonist was able to improve NAFLD histology in an obese mouse model. Jiang et al., altered mouse microbiota by treatment with either antibiotics or tempol, a potent antioxidant, and noted an increase in the conjugated bile acid metabolites in the treated animals compared to controls. Both the control mice and the treated mice with altered microbiota were then exposed to a high fat diet. The increased conjugated bile acid metabolite seen in the treated mice (tauro-β-muricholic acid) inhibited intestinal FXR signalling, leading to a downstream decrease in the amount of triglyceride accumulation in response to the high fat diet compared with the control mice. Further studies are needed in humans as the majority of the research in this area has been undertaken in murine models, and there are differences in the bile acid composition between mice and humans. Furthermore, the complex interplay between the microbiome and the human bile acid pool requires further elucidation in the context of risk for NAFLD and NASH. Bile acid receptor agonists are under active investigation as possible therapeutic targets for NAFLD and NASH, specifically FXR agonists that have anti-inflammatory, immunomodulatory and anti-fibrotic properties. Specifically, obeticholic acid, an FXR receptor ligand, has been shown to improve features of NASH histology among individuals with biopsy-proven NASH in a Phase II clinical trial.
Gut Permeability Alterations. The intestinal microbiota plays a critical role in maintaining the integrity of the intestinal barrier. Miele et al. found evidence of a disruption in the intestinal barrier of biopsy-proven NAFLD patients by measuring increased levels of urinary excretion of Cr-ethylenediaminetetra-acetate as well as by demonstrating disruption of tight junctions in duodenal biopsy using immunohistochemical analysis of the zona occludens. The authors also found an increased rate of small bowel overgrowth within these patients, as measured by glucose breath testing, suggesting that alterations in the microbiome may have contributed to disruption of gut barrier integrity. Similarly, studies of mouse models have found correlations between hepatic inflammation and measures of loss of intestinal barrier integrity, which further supports the hypothesis that loss of intestinal barrier integrity influences the pathogenesis of NAFLD and possibly NASH. Loss of this barrier function theoretically could expose the liver to increased levels of bacterial pro-inflammatory products such as LPS, as well as toxic bacterial metabolic by-products, including ethanol and other VOC such as ethanol, acetone and butanoic acid, among others.
Future studies will continue to assess microbial composition with respect to liver pathology, but will also more precisely define the functional roles of microbial groups in either promoting or attenuating NAFLD pathogenesis. The expanding fields of metagenomics, metatranscriptiomics and metabolomics, which examine the genomic capacity, gene expression profiles and small metabolites, respectively, of microbial communities, will be key tools to advancing the science in this field. By better determining how gut microbial communities function in health and fail to function in liver disease, we can begin to unravel the mechanisms by which the microbiome regulates NASH. Knowledge of the microbial functional pathways that are most closely associated with NAFLD and NASH will undoubtedly provide new targets for diagnostic testing and treatment in the future. A major challenge, however, will be to discover novel means to stably modify the microbiome without interfering with the health-promoting symbiotic relationships between the human host and our microbiota.
Results
We identified nine original research studies (five in humans and four in animal models) that characterised microbial communities in the setting of NAFLD. This review will initially discuss important background concepts of dysbiosis, then we will give a detailed overview of the nine original research studies that directly characterise the microbiome in the setting of NAFLD, and finally, we will discuss current mechanistic concepts linking the intestinal dybiosis with NAFLD.
Important Concepts in the Study of Dysbiosis
Molecular Assessment of the Microbiome: Technical Aspects. Advances in culture-independent microbiologic technology over the last decade have facilitated the characterisation of both the composition and diversity of the intestinal microbiota. Previously, assessment for particular intestinal inhabitants required direct culture, which was slow and labour intensive. Consequently, only a small fraction of the intestinal inhabitants could be directly cultured under routine conditions. Now, with sequencing of highly conserved genes or entire genomes, the vast array of microbes can be more incisively profiled, even in complex environments, such as the distal bowel. As ribosomes are essential to all cellular life, the genes encoding ribosomal components, most notably the small-subunit ribosomal RNA (rRNA) gene, have become the gold-standard for delineating the phylogenetic relationships among cellular life. Specifically, the 16S rRNA gene sequences for archaea and bacteria are used as stable phylogenetic markers to define which lineages are present in a sample. Figure 1 illustrates the general work-flow leading from sample collection, extraction of bulk genomic DNA, pan-microbial 16S rRNA gene PCR amplification, next-generation sequencing, to the bioinformatic output that describes the proportional distribution of microbial taxa in the sample set.
(Enlarge Image)
Figure 1.
Workflow for 16S rRNA sequence-based profiling of microbial communities. Following specimen collection, bulk genomic DNA is prepared using robust extraction protocols. This DNA is subjected to broad range PCR amplification using primers that amplify all bacterial 16S rRNA genes (other primer sets can be used to profile archaea, fungi or all organisms). PCR amplicons are sequenced using next-generation sequencing platforms. The resulting sequence datasets are quality-filtered and each sequence assigned a taxonomic classification using phylogenetic analysis software. The results provide an assessment of the types and relative abundances of microorganisms in the sample set.
The Challenges of Determining a 'Normal' Human Core Microbiome. In the process of trying to investigate dysbiosis and disease associations, it is necessary to understand the types and distributions of microbes that colonise ostensibly healthy individuals. However, composition and dynamics of this 'normal' microbiome remains incompletely defined. A pivotal study by Turnbaugh et al. investigated the gut microbiota of 154 adult individuals including monozygotic twins, dizygotic twins, and their mothers. A key question that was addressed was whether a 'core' intestinal microbiome of specific microbial lineages are commonly shared. The investigators found that no single bacterial species was detectable at an abundant frequency (defined as >0.5% abundance in the community) in the guts of all 154 individuals. Some similarities were found among family members, but each individual's microbial community displayed a unique pattern at the species level. However, there did appear to be a core microbiome at the functional level based on deep metagenomic sequencing.
Further complexity is apparent when one considers the temporal and spatial distributions of microorganisms within an individual's own intestinal ecosystem. Eckburg et al. studied the microbiota in detail in three individuals with samples taken by stool collection and then also by direct mucosal biopsy at six sites throughout their colonic tract. The investigators found significant intra-subject differences between the stool samples and more directly obtained mucosal community samples. At the species- and strain-levels, the profound variability between and within individuals' gastrointestinal tract complicates attempts to define what is 'normal' microbiota. This diversity remains a significant challenge in studies of the microbiome in human diseases.
Given the diversity at the individual species level, some studies have focused on assessing broader patterns at higher taxonomic levels, such as genus and phylum, when assessing for dysbiosis linkage with disease. The intestinal microbiota of adult humans is composed predominantly of two bacterial phyla, Bacteroidetes and Firmicutes. As discussed below, changes in the relative abundances of Bacteroidetes and Firmicutes have been observed in obesity and the metabolic syndrome, with an increased abundance of Firmicutes being associated with obesity in some studies. Defining these broader patterns of phylum composition have been important initial steps in trying to understand alterations in the microbiome that may be associated with disease states. However, how the diversity of the vast array of specific bacterial species may play a role in disease processes in humans is an area of future research.
Dysbiosis and Obesity: A Brief Overview. Studies of intestinal dysbiosis have highlighted an association with obesity. The first evidence for alterations in microbiome ecology in the setting of obesity was reported in mouse models. Support for the microbiome affecting body weight initially was produced through experimental transfer of the intestinal microbiota from normal, conventionally raised mice into germ-free mice, which led to an increase in body fat with no change in food consumption. Ley et al. subsequently reported that leptin-deficient obese mice (ob/ob), in comparison with lean wild-type mice, exhibited a shift towards higher relative abundance of the bacterial phylum Firmicutes, with concomitantly lower abundance of the phylum Bacteroidetes. Although dysbiosis has also been observed in multiple studies of human obesity and weight change, study results have been conflicting with respect to the microbial phylogenetic groups that are altered in abundance, and several studies have found no significant correlations between body mass index and the ratio of abundance of Firmicutes to Bacteroidetes. Furthermore, understanding whether obesity alters the microbiome, or if the microbiome alters risk for obesity in humans is a question that requires further longitudinal study.
Review of Original Research Characterising Dysbiosis in the Setting of NAFLD
Nonalcoholic fatty liver disease has been linked unequivocally with obesity and is widely considered the hepatic manifestation of the metabolic syndrome. And more recently, NAFLD has been associated with dysbiosis. At a basic level, NAFLD might be influenced by the microbiome simply via the increased risk for obesity mediated by increased caloric intake. However, the relationship between NAFLD and intestinal dysbiosis is likely much more complex and driven by multiple molecular pathways. Below, we review nine original research studies (five human, four animal models) that characterise the microbiome in the setting of NAFLD.
NAFLD and Dysbiosis in Animal Models. Animal studies have been the first step in critically evaluating the role of intestinal dysbiosis in the risk for NAFLD. Early animal studies examined changes in the microbiome occurring with disease processes such as obesity, and interventions such as high fat diet feeding. Subsequent investigations have more directly assessed the effects of transmitting dysbiotic microbiomes to recipients and then measuring the effect on the recipient phenotype in the context of risk for NAFLD. Zeng et al. studied a murine model in which mice were fed a high fat diet for 10 weeks. The high fat diet fed mice increased their body weight by 34% compared to the low fat fed mice and the livers of the mice fed the high fat diet also exhibited dramatic increases in the number of lipid droplets, inflammatory cell infiltration and inducible nitric oxide synthase protein concentration. The authors also reported that a high fat diet increased the levels of lactobacillus species in mouse faecal material when compared to the group of mice fed a low fat diet for 10 weeks. A positive correlation was noted between the amount of lactobacillus DNA in the faecal samples and the severity of steatosis within mouse livers. The authors postulated that the lactobacillus species could have an effect on lipid metabolism through effects on bile acid metabolism, and thus contributed to the risk for fatty liver.
Other animal studies have assessed the effects of altering the microbiota using probiotics on metabolic risk factors linked with NAFLD. Cano et al. studied both wild-type mice and high fat diet-induced obese mice. Both groups were given either placebo or a probiotic consisting of Bifidobacterium pseudocatenulatum for 7 weeks. Changes in metabolic parameters, including insulin resistance, were measured between the mouse groups and samples were obtained for liver histology to determine the degree of steatosis pre- and post-intervention with the probiotic or placebo. A high fat diet supplemented with bifidobacteria was associated with lower food intake and also lower body weight compared with the placebo group. The authors also found that bifidobacteria supplementation resulted in decreased insulin resistance, decreased hepatic steatosis, and a reduction in serum inflammatory markers compared with the high fat diet fed mice that were not supplemented. In contrast, the wild-type mice fed a control diet showed no difference in these metabolic and inflammatory parameters, regardless of the presence or absence of probiotic supplementation. The authors also reported differences in gut microbial composition between the mouse groups; specifically, the addition of the probiotic increased the number of bifidobacteria in the high fat diet group and reduced the number of Enterobacteriaceae spp. The authors concluded that the probiotic was able to ameliorate the metabolic and immunological dysfunction, including hepatic steatosis, related to obesity in the high fat diet fed mouse model.
Henao-Mejia et al. studied mouse models deficient in the pro-inflammatory multi-protein complexes, termed inflammasomes, to explore the potential role of the microbiome in NAFLD. In a prior study, the authors had found alterations in the faecal microbiome in inflammasome-deficient mice characterised by expanded representation of the bacterial phyla Bacteroidetes. To explore how these changes in the microbiome possibly influenced NAFLD, they utilised several mouse models of NAFLD (methionine choline-deficient diet model, the genetic leptin receptor deficiency steatosis model, and the high fat diet model) within the inflammasome deficient mice. All NAFLD models exhibited an exacerbated NAFLD phenotype in the inflammasome-deficient genetically modified mouse strains compared with wild type. They found activation of pro-inflammatory pathways through influx of TLR4 and TLR9 agonists into the portal circulation, which then led to enhanced hepatic tumour-necrosis factor (TNF) within the liver. The authors further observed that co-housing and thus transferring the dominant microbiome from the inflammasome-deficient mice to wild type exacerbated the phenotype of NAFLD and NASH in the co-housed wild-type NAFLD mice models. The authors concluded that transfer of a dysbiotic microbiome worsened hepatic steatosis and thus potentially affected NAFLD progression, thus causally linking dysbiosis with NAFLD pathogenesis.
Another murine investigation conducted by Le Roy et al. demonstrated that the risk of developing NAFLD was transmissible by faecal transplant to recipient mice. The investigators exploited an initial observation that conventionally raised mice display variable degrees of weight gain, hyperglycaemia and insulin resistance when exposed to a high fat diet. The investigators selected two types of donor mice: one donor mouse that demonstrated weight gain with increased systemic inflammation (increased serum levels of MCP-1 and TNF-α) and increased insulin resistance on a high fat diet; and a second donor mouse that had weight gain on the high fat diet, but did not display increased inflammation or insulin resistance. They transferred faecal samples from each of the donors into sets of germ-free mice, which subsequently took on features of the phenotypes of their faecal donors. In this experiment, the effect of the transfer of the inflammatory and insulin resistance phenotypes via the donor microbiome was independent of obesity induced by the high fat diet. The mice that developed the inflammatory and insulin resistance phenotype were also characterised by increases in hepatic macrovesicular steatosis. Interestingly, no increase in inflammatory infiltrate was noted within the livers of these mice, which might suggest that this faecal transfer led to elevated risk for NAFLD, but not necessarily NASH. The short follow-up duration may also have been inadequate to detect the eventual inflammatory damage that would signify NASH. The investigators also assessed specific compositional differences in the microbiota between the recipients of the inflammatory/insulin resistant faecal donor sample vs. the recipients of the non-inflammatory/non-insulin resistant faecal donor sample. They found distinct differences at the phylum, genus, and species levels. At the phylum level, increased Firmicutes were found in association with the inflammatory NAFLD prone phenotype, which is in concordance with increases in this phylum previously reported for other metabolic disorders, such as obesity. These initial mouse studies are of great interest given their implication that the intestinal microbiota may directly affect risk for NAFLD and NASH within the host.
NAFLD and Dysbiosis in Humans. To better understand how findings from animal studies might relate to human disease, several investigators have begun to assess whether gastrointestinal dysbiosis is associated with NAFLD risk in humans. Five original studies within the published literature have explored the intestinal microbiome among cohorts of human subjects. Table 1 summarises the overall characteristics of these studies and provides details regarding the specific microbiologic results. These published human studies have great variability in study design, population selection, and reporting of results, all of which complicate meta-analysis. Thus, the current review focuses on a qualitative summary of these studies.
Spencer et al. performed one of the first studies that explored the microbiome in relation to human NAFLD. They utilised a population of 15 normal healthy adult females within a larger longitudinal metabolic study that entailed an in-patient intervention of a choline depletion diet. Choline is an essential nutrient and low-choline diets are associated with health problems in humans, including hepatic steatosis. The authors assessed the faecal microbiome at baseline, after 10 days on a normal controlled diet, up two times during a 42 day period on a choline-depleted diet (which lasted only 42 days if they did not develop evidence of end organ damage), and then twice during a 10 day choline repletion diet. Serial faecal samples from study subjects were analysed by 16S rRNA gene sequencing. The longitudinal design allowed the authors to prospectively characterise alterations in microbiota that were associated with the fatty liver changes that developed in some individuals on the choline-depletion diet. The microbiome of each individual remained distinct, as evidenced by continued clustering by individual on serial samples, but shifts in microbial community profiles were observed in response to the choline-depleted diet. Most notably, higher baseline levels of the phylum Proteobacteria, specifically the class Gamma Proteobacteria, correlated with lower risk for developing fatty liver, whereas higher baseline levels of the class Erysipelotrichia (a member of the phylum Firmicutes) correlated with higher risk for fatty liver. The authors concluded that the baseline levels of the bacteria on the ad libitum normal diet modified the degree of the subject's susceptibility to fatty liver when on a diet deficient in choline. It is important to note that the changes in fatty liver composition within this study were measured by changes assessed by MRI, and not by serial histology. Thus, this study design did not assess the risks for inflammation with NASH.
Zhu et al. characterised the intestinal microbiome composition within a group of 63 children (mean age approximately 13 years) by 16S rRNA gene sequencing. The authors assessed cross-sectional differences between faecal samples of 22 children with NASH found on biopsy, 25 clinically obese children with no clinical suspicion of NASH, and 16 healthy normal weight control children. When compositional differences were measured at the phylum level, decreased levels of Firmicutes and increased Bacteroidetes were observed within NASH and obese individuals compared with the controls, but the abundance of Bacteroidetes and Firmicutes were similar between patients with NASH and those with obesity. To attempt to address possible mechanisms of microbiota influence, the authors also measured serum ethanol levels in study participants and found increased ethanol levels in the NASH children. The authors postulated that increased Escherichia coli within the NASH group might have elevated ethanol levels, given that this group of microbes is able to produce ethanol. The authors also hypothesised that increased ethanol production by the microbiota could lead to chronic, low-level exposure to this hepatotoxin, thus putting the individuals at risk for steatohepatitis. These investigators reported in a separate publication from the data on this patient population that no differences in the levels of endotoxemia were observed between these paediatric groups of NASH, obese, and controls. This finding led the authors to postulate that endotoxemia may not be required for the pathogenesis of paediatric NASH.
In another study, Mouzaki et al. recruited a group of 50 adults: 11 with biopsy-proven 'simple' steatosis, 22 with biopsy-proven NASH, and 17 healthy controls with no steatosis by biopsy. They utilised quantitative polymerase chain reaction with custom-made primer/probes that were specific for total bacterial counts and also identified the following five specific groups of bacteria: Bacteroides/Prevotella, Ruminococcaceae (Group IV clostridia), Lachnospiraceae (Group XIVa clostridia), E. coli, and Archaea. The authors found significantly decreased levels of Bacteroidetes in patients with NASH compared to individuals with simple steatosis and healthy controls, even after adjusting for body mass index. They also reported that NASH patients had an increased number of Lachnospiraceae species compared with the simple steatosis group, but the significance of this difference was lost after adjusting for BMI and fat intake. The strengths of this study included the use of liver biopsies to confirm the histological diagnosis for each of the patient groups. However, the cross-sectional nature of the study, which is common for human microbiome studies, was a limitation as far as drawing conclusions regarding causality.
Raman et al. investigated faecal microbiomes in a cross-sectional study of 60 adults: 30 obese individuals with clinically defined NAFLD (no biopsy available) and 30 non-obese controls. The authors reported that selected members of the phylum Firmicutes were increased in NAFLD, compared with non-obese controls. A strength of this study was that it measured levels of hepatotoxic volatile organic compounds (VOC), which have been hypothesised to promote the development of NAFLD, in parallel with the microbiome profiling. Indeed, elevated volatile compound levels were detected in the stools of the NAFLD patients, leading the authors to propose that these compounds increased as a consequence of dysbiosis. However, a limitation acknowledged by the authors was that their NAFLD group was clinically diagnosed and thus the findings could not be linked to the risk of NASH. An additional limitation was the lack of an obese control group without NAFLD. As the control group was composed of non-obese individuals, it was not possible for the investigators to determine whether the links between dysbiosis and liver disease were confounded by increased BMI.
A longitudinal microbiome study by Wong et al. investigated 20 adult patients with biopsy-proven NASH who were randomised to receive either a probiotic supplement consisting of a combination of Lactobacillus plantarum,Lactobacillus delbrueckii ssp bulgaricus,Lactobacillus acidophilus,Lactobacillus rhamnosus and Bifidobacterium bifidum (n = 10), or no probiotic supplementation (n = 10) for a total of 6 months. Both groups received standard of care counselling from an investigator to lose weight, reduce fat intake, and exercise at least three times per week. In the proof-of-concept assessment of treatment of NASH with probiotics, the investigators found a greater reduction in hepatic fat as measured by magnetic resonance spectroscopy (MRS) and also decrease in aspartate aminotransferase in the group supplemented with probiotics compared to the counselling only group at 6 months. Hepatic fat content decreased from 22.6 ± 8.2% to 14.9 ± 6.6% (P = 0.034) in the probiotic group and remained relatively stagnant in the counselling only group (16.9 ± 6.1% to 16.0 ± 6.6%, P = 0.55). To assess the microbiome changes over time in these populations, the faecal microbiota were profiled both at baseline and at 6 months in both groups. A total of seven patients from the probiotic group and nine from the no probiotic group (counselling only) had adequate samples to allow inclusion in the final analysis. The authors also enrolled a group of healthy controls (n = 22) to compare to the NASH patients. This design allowed the authors to identify baseline differences between NASH and controls, as well as changes in the microbiota over the 6-month study period in the longitudinally sampled group. At baseline, no differences in Bacteroidetes were apparent between groups, but reduced levels of Firmicutes were found in the NASH subjects, compared with the control subjects without NASH. After 6 months, reduced hepatic fat as measured by MRS was associated with an increase in Bacteroidetes and a decrease in Firmicutes, among the overall group of NASH patients.
Taken together, these human studies demonstrate measureable differences in the microbiome between clinical states of NAFLD and NASH, but due to the great variability study design, methods, and clinical endpoints, it remains challenging to interpret the clinical significance of intestinal dysbiosis in the context of liver disease. Further work is needed, particularly longitudinal studies, to delve further into the pathologic significance of these results in human NAFLD.
Mechanistic Pathways in NAFLD Pathogenesis
The aforementioned animal and human studies provide primarily descriptive characterisations of the microbiota within diseased populations. These and other investigators have also focused their attention on elucidating the mechanisms by which microbes may affect their hosts' metabolic processes and thereby modulate the risks for NAFLD and NASH. Several mechanisms have been postulated to explain the associations between NAFLD and dysbiosis. As noted above, Spencer et al. assessed the possible role of altered choline metabolism, Zhu et al. measured endotoxin and ethanol levels as possible hepatotoxins, and Raman et al. measured increased VOC s in stool as potential hepatotoxins. These studies are important early steps towards determining the mechanisms by which dysbiosis affects risk for NAFLD and NASH, beyond simply characterising the composition of the gut microbiome in the setting of NAFLD. Understanding underlying mechanistic relationships should open new avenues of research into novel, readily modifiable targets to reduce disease risk and progression.
Several mechanisms have been proposed within the literature, and highlighted by prior reviews, that may link the microbiome to NAFLD including the impact of microbiota on energy utilisation and the association with obesity; influences of microbiota on gut permeability that induce low grade inflammation; modulation of dietary choline metabolism by the microbiota; regulation of bile acid metabolism; and endogenous production of hepatotoxic substances such as ethanol and other VOCs by the microbiota. These postulated mechanisms, discussed further below, may potentiate each other through shared molecular pathways of fat deposition and activation of inflammation within the liver. Molecular pathways that are directly linked to the microbiome and have been shown to contribute to models of NAFLD include lipopolysaccharide (LPS) endotoxemia, induction of immune toll-like receptor (TLR) signalling, and inflammasome mediated pathways. Figure 2 highlights these potential mechanisms and pathways in NASH.
(Enlarge Image)
Figure 2.
Mechanistic pathyways in NAFLD pathogenesis. Highlighted pathways within the enterohepatic circulation that influcence progression of NAFLD and NASH. Dysbiosis influences several pathways that lead to the delivery of metabolic and inflammatory factors through the portal circulation with resulting activation of molecular pathways that promote lipid accumulation and inflammation. EtOH, ethanol; Fiaf, fasting induced adipocyte factor; FFA, free fatty acids; FXR, farnisoid X receptor; LPL, lipoprotein lipase; LPS, lipopolysaccharide; SCFA, short-chain fatty acids; TG, triglycerides; TLR, toll-like receptor; TMA, toxic methylamine; VLDL, very low density lipoprotein; VOC, volatile organic compounds.
Facilitation of Host Energy Harvest and Utilisation by the Microbiota. Several investigators have linked the microbiome with obesity. Much of this research has proposed that the microbiome affects energy extraction, absorption, utilisation, and storage, processes that also could have links with the pathogenesis of NAFLD. Further research has also suggested that dysbiosis may directly increase the absorption of short-chain fatty acids (SCFA), free fatty acids (FFA), and carbohydrates, with an associated up-regulation of ChREBP and SREBP-1c, along with inhibition of Fiaf [fating induced adipose factor, which is responsible for encoding the inhibitor of lipoprotein lipase (LPL)]. This leads to LPL activation resulting in increased lipogenesis and the development of the milieu for NAFLD. However, the direct effect of the microbiota on dietary energy absorption through increased SCFAs, FFAs and carbohydrates is just one link in the risk for NAFLD and NASH, as it is likely that other mechanisms are involved, particularly given that not all obese individuals develop NASH and simply increasing caloric intake is not sufficient for NASH pathogenesis. The SCFAs include: acetate (C2), propionate (C3) and butyrate (C4) and are the main metabolic products of anaerobic bacteria fermentation in the intestine. It is notable that the SFCAs not only serve as fuel for the intestinal epithelial cells but also act on leucocytes and endothelial cells to regulate several leucocyte functions including production of cytokines (TNF-α, IL-2, IL-6 and IL-10), eicosanoids and chemokines (e.g. MCP-1 and CINC-2). In general, SCFAs, such as propionate and butyrate, inhibit adhesion molecules, chemokine production and monocyte, macrophage and neutrophil recruitment, suggesting an anti-inflammatory action; however, there is also evidence favouring a pro-inflammatory action of SCFAs in some conditions. Thus, SCFAs may have a complex effect on the inflammatory process in NASH.
Altered Dietary Choline Metabolism by the Microbiota. Choline is an essential nutrient that has important roles in human metabolism, including: neurotransmitter synthesis, cell-membrane signalling and lipid metabolism. Because of this wide range of effects, dietary choline-deficiency has been linked with a variety of conditions including neurological developmental disorders, heart disease and liver disease. Diets deficient in choline can lead to increased hepatic steatosis, which can be reversed with choline supplementation. The intestinal microbiota plays a role in the conversion of dietary choline to toxic methylamine, a substance that both mimics a choline-deficient diet by decreasing effective choline levels and exposes the host to an inflammatory toxic metabolite. Spencer et al. showed that, in individuals on a choline-deficient diet, a particular baseline microbial composition correlated with the risk for the induction of hepatic steatosis, as noted on serial MRI scans. They found increased Gammaproteobacteria abundance and decreased Erysipelotrichi abundance at baseline were protective against developing steatosis. The risk for NASH was not assessed in this study. The theory that microbial dysbiosis can affect risk for NAFLD and NASH by altering the metabolism of an essential nutrient such as choline is intriguing and will require further study to better understand.
Impact of the Microbiota on Bile Acid Metabolism. Bile acids are synthesised from cholesterol through enzymatic pathways within hepatocytes and are then conjugated within the hepatocytes with either glycine or taurine before secretion into bile and released into the small intestine. In the small intestine conjugated bile acids not only assist in lipid absorption and transport but have also been increasingly recognised to function as nuclear receptor binders and to have a putative role in altering the microbiome. Bacteria within the intestine can also chemically modify bile acids and thereby alter the composition of the bile acid pool. The effects of the microbiome on the size and composition of the bile acid pool have been demonstrated by Swann et al., who used a germ-free mouse model to compare bile acid species in different mouse tissues. The authors found that, when compared to conventional mice, germ-free mice had significant differences in the distribution of bile acid species in several different tissues (liver, heart and kidney). Specifically, germ-free mice had increased taurine conjugated bile acids in liver, kidney, heart and plasma compared to conventional mice and comparable results were found using an antibiotic treated mouse model. Overall bile acid diversity was also found to be lower in both germ-free and antibiotic treated mice when compared with controls. Sayin et al., further observed that conventionally raised mice had smaller bile acid pools than did their germ-free counterparts. Other animal studies indicate that the host bile acid pool in turn influences the composition of the microbiome. Alterations in bile acid composition may be a mechanism by which the microbiome might affect risk for NAFLD and NASH.
Although bile acids have long been known to play a role as detergents to facilitate fat absorption, their role in human physiology is more complex and bile acids are now recognised as important cell signalling molecules that activate multiple pathways regulating lipid metabolism, carbohydrate metabolism and inflammatory response. These effects are mediated through their binding and activation of the nuclear hormone receptor, farnesoid X receptor (FXR), and the G protein coupled cell surface receptor TGR5. The signalling pathways triggered by FXR and TGR5 activation modulate bile acid synthesis, as well as the metabolism of glucose, triglycerides and cholesterol. Specifically, FXR activation results in inhibition of gluconeogenesis and glycogenolysis in the liver (improving glucose metabolism) and results in increased insulin sensitivity in adipose and skeletal muscle. Lipogenesis is also attenuated by FXR activation through down-regulation of SREBP-1c and an increase in beta fatty acid oxidation. It is notable that FXR knockout mice develop features of NASH, including hepatic steatosis and hepatic inflammation. FXR has also been associated with inhibition of NFkB activation suggesting an anti-inflammatory role for FXR, as well. Figure 3 illustrates associations between the bile acid pool, the microbiota and FXR activation.
(Enlarge Image)
Figure 3.
Bile acid modulation of FXR receptor alteration in the bile acid composition leads to activation of the FXR nuclear receptor, which stimulates transcription of FGF19 that is transported to the liver through the portal circulation and activates signalling pathways that downregulate bile acid synthesis through Cyp7A1 inhibition. This altered pool of bile acids may ultimately feedback and also favourably or unfavourably influence the microbiota composition within the intestine, though this purported association needs to be further defined with experimental data. The FXR and FGF19 triggered pathways also influence glucose and lipid metabolism through a decrease in gluconeogenesis and decrease in lipolysis.
Further insights into the complex relationships between dysbiosis and bile acid pathways relative to NAFLD and NASH have come from experimental data in mouse models. McMahan et. al., showed that activation of bile acid receptors with a receptor agonist was able to improve NAFLD histology in an obese mouse model. Jiang et al., altered mouse microbiota by treatment with either antibiotics or tempol, a potent antioxidant, and noted an increase in the conjugated bile acid metabolites in the treated animals compared to controls. Both the control mice and the treated mice with altered microbiota were then exposed to a high fat diet. The increased conjugated bile acid metabolite seen in the treated mice (tauro-β-muricholic acid) inhibited intestinal FXR signalling, leading to a downstream decrease in the amount of triglyceride accumulation in response to the high fat diet compared with the control mice. Further studies are needed in humans as the majority of the research in this area has been undertaken in murine models, and there are differences in the bile acid composition between mice and humans. Furthermore, the complex interplay between the microbiome and the human bile acid pool requires further elucidation in the context of risk for NAFLD and NASH. Bile acid receptor agonists are under active investigation as possible therapeutic targets for NAFLD and NASH, specifically FXR agonists that have anti-inflammatory, immunomodulatory and anti-fibrotic properties. Specifically, obeticholic acid, an FXR receptor ligand, has been shown to improve features of NASH histology among individuals with biopsy-proven NASH in a Phase II clinical trial.
Gut Permeability Alterations. The intestinal microbiota plays a critical role in maintaining the integrity of the intestinal barrier. Miele et al. found evidence of a disruption in the intestinal barrier of biopsy-proven NAFLD patients by measuring increased levels of urinary excretion of Cr-ethylenediaminetetra-acetate as well as by demonstrating disruption of tight junctions in duodenal biopsy using immunohistochemical analysis of the zona occludens. The authors also found an increased rate of small bowel overgrowth within these patients, as measured by glucose breath testing, suggesting that alterations in the microbiome may have contributed to disruption of gut barrier integrity. Similarly, studies of mouse models have found correlations between hepatic inflammation and measures of loss of intestinal barrier integrity, which further supports the hypothesis that loss of intestinal barrier integrity influences the pathogenesis of NAFLD and possibly NASH. Loss of this barrier function theoretically could expose the liver to increased levels of bacterial pro-inflammatory products such as LPS, as well as toxic bacterial metabolic by-products, including ethanol and other VOC such as ethanol, acetone and butanoic acid, among others.
Future Investigations Into NAFLD and Dysbiosis: Role of Metagenomics
Future studies will continue to assess microbial composition with respect to liver pathology, but will also more precisely define the functional roles of microbial groups in either promoting or attenuating NAFLD pathogenesis. The expanding fields of metagenomics, metatranscriptiomics and metabolomics, which examine the genomic capacity, gene expression profiles and small metabolites, respectively, of microbial communities, will be key tools to advancing the science in this field. By better determining how gut microbial communities function in health and fail to function in liver disease, we can begin to unravel the mechanisms by which the microbiome regulates NASH. Knowledge of the microbial functional pathways that are most closely associated with NAFLD and NASH will undoubtedly provide new targets for diagnostic testing and treatment in the future. A major challenge, however, will be to discover novel means to stably modify the microbiome without interfering with the health-promoting symbiotic relationships between the human host and our microbiota.
Source...