Characterization of gut microbiomes in nonalcoholic steatohepatitis (NASH) patients: A connection between endogenous alcohol and NASH†‡
Potential conflict of interest: Nothing to report.
This work was supported by a grant from the Peter and Tommy Fund, Inc. (Buffalo, NY; to S.S.B.), a grant from the SKI4E Fund (to S.S.B.), and a departmental start-up fund (to L.Z.).
Abstract
Nonalcoholic steatohepatitis (NASH) is a serious liver disease associated with obesity. Characterized by metabolic syndrome, hepatic steatosis, and liver inflammation, NASH is believed to be under the influence of the gut microflora. Here, the composition of gut bacterial communities of NASH, obese, and healthy children was determined by 16S ribosomal RNA pyrosequencing. In addition, peripheral blood ethanol was analyzed to monitor endogenous ethanol production of patients and healthy controls. UniFrac-based principle coordinates analysis indicated that most of the microbiome samples clustered by disease status. Each group was associated with a unique pattern of enterotypes. Differences were abundant at phylum, family, and genus levels between healthy subjects and obese patients (with or without NASH), and relatively fewer differences were observed between obese and the NASH microbiomes. Among those taxa with greater than 1% representation in any of the disease groups, Proteobacteria, Enterobacteriaceae, and Escherichia were the only phylum, family and genus types exhibiting significant difference between obese and NASH microbiomes. Similar blood-ethanol concentrations were observed between healthy subjects and obese non-NASH patients, but NASH patients exhibited significantly elevated blood ethanol levels. Conclusions: The increased abundance of alcohol-producing bacteria in NASH microbiomes, elevated blood-ethanol concentration in NASH patients, and the well-established role of alcohol metabolism in oxidative stress and, consequently, liver inflammation suggest a role for alcohol-producing microbiota in the pathogenesis of NASH. We postulate that the distinct composition of the gut microbiome among NASH, obese, and healthy controls could offer a target for intervention or a marker for disease. (HEPATOLOGY 2013)
Nonalcoholic fatty liver disease (NAFLD), the hepatic manifestation of metabolic syndrome, is the most common cause of elevated liver enzymes in the United States.1 NAFLD with inflammation and fibrosis is known as nonalcoholic steatohepatitis (NASH) because it resembles alcoholic liver disease (ALD) without a history of alcohol ingestion.2 The incidence of NASH has been increasing over the past 20 years.3 In the United States, the current prevalence of NAFLD and NASH could be as high as 46% and 12%, respectively.4 Without an effective available treatment, the prognosis of NASH is not optimistic. NASH is responsible for approximately 10% of liver transplants in the United States and is projected to become the most common indication for liver transplantation in the near future.5
Approximately 25% of fatty livers progress to liver inflammation, fibrosis, and then cirrhosis, but the rest remain simple steatosis.4 Therefore, it is likely that factor(s) other than, or in addition to, liver fat deposition are required for the development of NASH. Many studies have shown that an extra source of oxidative stress (OS) could be one such factor (e.g., as reviewed elsewhere6). These studies are the basis for the “two-hit hypothesis”.7 In addition to OS, Toll-like-receptors (TLRs)–mediated signaling,8 adipose-tissue–derived signals,9 endoplasmic reticulum stress,10 and genetic factors11 may be necessary for, or contribute to, the development of NASH.
Gut microbiota are thought to play a role in the pathogenesis of NASH for several reasons. First, gut microbiota are known to have a large effect on the digestion and absorption of nutrients.12 Microbiota transplantation experiments in mice suggested that certain microbiota are capable of inducing obesity independent of other environmental factors.13 Second, gut microbiota participate in the development and homeostasis of the overall immunity of the host.14 Therefore, certain microbiota may influence the development of liver inflammation. The links between gut microbiota and the host immune system include TLRs and short-chain fatty acids.15 For example, TLR5 knockout mice have a unique composition of gut microbiota, which induces hyperphagia, obesity, hyperglycemia, insulin resistance, and elevated levels of proinflammatory cytokines, when transplanted to wild-type germ-free mice.16 Third, gut microbiota may influence the production of gut hormones, such as glucagon-like peptide 1, and, subsequently, have an effect on the overall metabolism of the host.17
Spencer et al. examined gut microbiomes of adult human subjects who had fatty livers induced by a choline-deficient diet.18 They observed changes in gut microbiome composition upon liver fat induction, suggesting that gut microbiomes and liver health are closely related. In this report, we examined the gut microbiota of NASH, obese, and healthy children and adolescents. Composition of NASH microbiomes was found to be distinct from those of healthy and obese microbiomes. Escherichia stood out as the only abundant genus that differed between NASH and obese patients. Because Escherichia are ethanol producers, this finding is in concert with our previous report that alcohol-metabolizing enzymes are up-regulated in NASH livers.19
Abbreviations
ADH, alcohol dehydrogenase; ALD, alcoholic liver disease; ANOVA, analysis of variance; BMI, body mass index; CDC, Centers for Disease Control; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; OS, oxidative stress; OTU, operational taxonomic unit; QIIME, Quantitative Insights into Microbial Ecology software; ROS, reactive oxygen species; rRNA, ribosomal RNA; TLR, toll-like receptor.
Patients and Methods
Human Subjects.
This study was approved by the Children and Youth Institutional Review Board of the State University of New York at Buffalo. Three groups of children and adolescents were recruited in this study (NASH patients, obese patients, and healthy controls). All NASH patients included in this study (Table 1) underwent percutaneous liver biopsy and fulfilled Kleiner's criteria on hepatic fat infiltration, inflammation, and fibrosis.20 Recruited in the obese group were patients whose body mass indices (BMIs) were higher than the 95th percentile, but whose liver function tests were normal. Patients with elevated transaminase, but without inflammation or fibrosis, were not included in the obese group because they may have had unidentified liver conditions that caused the elevated transaminases. Healthy controls were volunteers whose BMI was less than the 85th percentile. Enrolling only pediatric patients was an additional assurance that alcohol intake would not be a confounding factor in this study. Characteristics of human subjects are summarized in Table 1.
| Characteristics | Control | Obese | NASH |
|---|---|---|---|
| Sex (F/M) | F6/M10 | F12/M13 | F10/M12 |
| Age, years | 14.4 ± 1.8* | 12.7 ± 3.2 | 13.6 ± 3.5 |
| BMI | 20.4 ± 0.1 | 33.4 ± 0.3 | 34.0 ± 0.4 |
| BMI z-score† | 0.13 ± 0.04 | 2.36 ± 0.01 | 2.24 ± 0.02 |
| AST | ND | 28.4 ± 0.6 | 51.7 ± 1.3 |
| ALT | ND | 27.7 ± 0.6 | 66.9 ± 1.9 |
| Ethnicity (black/white/other) | 0/15/1 | 4/17/4 | 0/18/4 |
- Abbreviations: F, female; M, male; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ND, not determined.
- * Mean ± standard error.
- † A z-score of 1.6449 is equivalent to the 95th percentile.
Dietary Assessment.
Dietary intake for all individuals was assessed using three different methods. A 3-day diet history and a 24-hour dietary recall were used initially.21 All patients were contacted by phone for detailed instructions on how to fill the 3-day diet history that was mailed to them. The 24-hour recall was obtained over the phone by a trained provider. Diet history and dietary recall data were analyzed with DINE Healthy version 7.0.1. The Centers for Disease Control (CDC) food frequency questionnaire was used22 (CDC 2005 National health Interview Survey Questionnaire; available at: ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Survey_Questionnaires/NHIS/2005/English/QCANCER.pdf) as an additional assessment tool for fiber intake.
Genomic DNA Extraction, Sequencing, and Quantitative Analysis of the Microbiome Composition.
Briefly, a single stool sample was collected from each patient and healthy control (details in Supporting Materials). Genomic DNA was isolated from stool samples using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA) and mechanical lysis. 16S ribosomal RNA (rRNA) sequences were polymerase chain reaction amplified for pyrosequencing on a 454-FLX-Titanium Genome Sequencer (Roche 454 Life Sciences, Branford, CT). All raw 454 sequencing reads and the associated metadata are available at MG-rast (available at: http://metagenomics.anl.gov/linkin.cgi?project=1195). Pyrosequence reads were analyzed in the Quantitative Insights into Microbial Ecology (QIIME) software23 version 1.4.0. For taxonomic assignment, sequence reads were grouped into operational taxonomic units (OTUs) at a sequence similarity level of 97%.
Enterotyping.
Enterotypes were determined for all microbiome samples based on the criteria described by Arumugam et al.24 Arumugam et al. detected three enterotypes: enterotype 1 is characterized by abundant Bacteroides (22%∼39%) and diminished Prevotella (0%∼1%); enterotype 2 is characterized by abundant Prevotella (6%∼36%) and less-abundant Bacteroides (2%∼17%); and enterotype 3 has diminished abundance in both genera (Bacteroides, 3%∼16%; Prevotella, 0%∼8%). Enterotype 3 is often also characterized by relatively more-abundant Ruminococcus (0.1%∼6.5%). Some of our microbiome samples were enriched in both Bacteroides and Prevotella and did not fit into the enterotypes described above. Therefore, we designated these samples as enterotype H, a hybrid between enterotypes 1 and 2.
Blast Search of the Escherichia OTU Sequence.
From the QIIME-generated OTU table, all OTUs assigned to Escherichia were examined. The most abundant Escherichia OTU was identified to be OTU#20341. The DNA sequence of OTU#20341 was retrieved from a QIIME-generated FASTA file containing a representative sequence for each OTU. This sequence was used as the input for Blast search (available at: http://blast.ncbi.nlm.nih.gov), against the current 16S rRNA database (search performed on April 9, 2012).
Serum Alcohol Measurement.
A single blood sample was collected from each patient at the time of liver biopsy (from July 26, 2010 through June 1, 2011). A single blood sample from each healthy and non-NASH obese subject was collected at the same period. Serum alcohol concentration was determined in subsets of obese and NASH patients in the microbiome study and a new cohort of healthy subjects. Serum samples were stored at −80 °C before ethanol concentrations were measured with an ethanol assay kit from BioVision (Milpitas, CA), following the manufacturer's instructions. Briefly, alcohol was oxidized by alcohol oxidase, and the product was subsequently measured by a colorimetric probe (at 570 nm). A linear standard curve between alcohol concentrations and OD570 was generated and used to calculate alcohol concentrations in serum samples.
Statistical Analysis.
One-way analysis of variance (ANOVA) and post-hoc Tukey's honest significant difference tests for multiple comparisons were performed to evaluate differences in taxonomic abundance, alpha diversities, and serum ethanol concentrations among three groups. Fisher's exact test was performed to examine a possible association between enterotypes and health status. ANOVA, Tukey's, and Fisher's tests were performed with R version 2.14.0. A P value less than 0.05 was considered statistically significant.
Results
Ecological Diversities of Microbiomes in Healthy Subjects, Obese Patients, and NASH Patients.
Gut microbiomes of healthy, obese, and NASH children and adolescents (Table 1) were analyzed by 16S rRNA pyrosequencing. A total of 835,591 sequencing reads were obtained from a total of 63 samples. Ecological diversity within each sample was evaluated by a phylogenetic distance metric (Supporting Fig. 1A). No significant difference was observed among three groups. However, as a measurement of species richness, Chao1 metric revealed significant differences among three groups (Supporting Fig. 1B). P(ANOVA) values were <0.05 at sampling sizes 1,268, 2,526, 3,784, and 5,042 OTUs. Post-hoc Tukey's tests indicated that differences were significant between obese and healthy controls (Tukey's P < 0.05 at sampling sizes 1,268, 2,526, 3,784, and 5,042 OTUs). Species abundance in the NASH group was marginally lower than that of the healthy group, with Tukey's P values of approximately 0.1.
Ecological diversities within each group were accessed by UniFrac analysis. The majority of the samples clustered by health status (Fig. 1), but not by age, gender, or ethnicity (Supporting Fig. 2), indicating a specific connection between health status and gut microbiomes. Exceptions from all three groups were observed, reflecting the effect from other genetic and environmental factors on these microbiomes.
UniFrac-based principle coordinates analysis: gut microbiomes clustered by health status with exceptions. The beta diversity of the study cohort was evaluated by UniFrac-based analysis. The majority of the samples clustered by health status at the PC1 versus PC3 plot, indicating that health status was a major effect factor for the phylogenetic composition of these samples. Exceptions from all three study groups were observed, reflecting the effect from other genetic and environmental factors on these microbiomes.
To investigate possible effect of dietary habits on gut-microbiome composition of patients, dietary assessments were conducted to analyze dietary intake at the time of fecal-sample collection (Supporting Table 1). No significant difference in percent energy from protein, fat, or carbohydrate was found among healthy, obese, and NASH subjects. Dietary fructose, fiber, and aspartame (a potential source of methanol) were also similar among the three study groups. No significant dietary source of alcohol was identified for any of the patients or healthy controls.
Obesity and NASH Are Associated With Specific Enterotypes of Gut Microbiomes.
Fifty-three of sixty-three microbiome samples fit into the enterotypes 1 (enriched in Bacteroides), 2 (enriched in Prevotella), and 3 (diminished in both Bacteroides and Prevotella), as described by Arumugam et al.,24 but the remaining 10 samples did not fall into a previously defined enterotype (Supporting Table 2). These 10 samples were characterized by abundant representation in both Bacteroides and Prevotella, therefore termed enterotype H (hybrid between enterotypes 1 and 2). The majority of the healthy gut microbiomes were classified into enterotypes 1 and 3, reflecting the fact that healthy microbiomes are scarcely represented by Prevotella, whereas obese and NASH microbiomes are more frequently classified into enterotype 2 (Prevotella type). NASH samples further differentiated from obese samples in that only one NASH sample was classified as enterotype 3 and seven NASH samples were classified as enterotype H. Fisher's exact test suggested that each of the three groups was associated with a specific enterotyping pattern (P < 0.01).
Comparison of Microbiomes at the Phylum Level.
Fourteen bacteria phyla were detected in gut microbiomes in this study (Fig. 2). Bacteroides and Firmicutes were the dominant phyla in these samples. Although exhibiting a broad distribution (Supporting Fig. 3), a statistically significant and drastic increase in Bacteroides and decrease in Firmicutes was apparent in the obese and NASH groups, compared to the healthy group (Figs. 2 and 3A). The abundance of Bacteroides and Firmicutes were similar between the obese and NASH groups. Another two phyla, Actinobacteria and Proteobacteria, exhibited >1% abundance in at least one of the groups. ANOVA analysis indicated that these two phyla were also significantly different among the three groups (Fig. 3B). Tukey's tests showed that Actinobacteria was significantly lower in the NASH group, compared to the healthy group. A gradually increased abundance of Proteobacteria was observed from the healthy group to the obese group and then to the NASH group. Proteobacteria was the only abundant phylum exhibiting a significant difference between the obese and NASH groups.
Average phylum distribution of gut microbiomes of healthy subjects, obese patients, and NASH patients.
Comparison of the taxonomic abundance among the healthy, obese and NASH groups. ANOVA tests indicated that these three groups were significantly different in all four abundant phyla in gut microbiomes: Bacteroides, Firmicutes (A), Actinobacteria, and Proteobacteria (B). Proteobacteria is the only phylum exhibiting significant difference between the obese and NASH groups. Within Proteobacteria, Enterobacteriaceae (C) and Escherichia (D) exhibited a significant difference between the obese and NASH groups. *P < 0.05 in Tukey's honest significance test; **P < 0.01 in Tukey's honest significance test.
Comparison of Microbiomes at Family and Genus Levels.
To gain a clear image of the taxonomic changes in obese and NASH gut microbiomes, abundant families and genera with >1% average abundance in any of the groups were examined (Table 2).
| Healthy* | Obese† | NASH‡ | P(ANOVA) | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Actinobacteria§ | 3.17 | 2.42 | 1.29 | 0.0262 | ||||||
| Bifidobacteriaceae∥ | 2.19 | 1.54 | 0.64 | 0.0446 | ||||||
| Bifidobacterium¶ | 2.16 | 1.47 | 0.63 | 0.0454 | ||||||
| Bacteroidetes | 28.65 | 50.28 | 49.11 | 0.0092 | ||||||
| Bacteroidaceae | 21.02 | 24.11 | 23.31 | 0.8960 | ||||||
| Bacteroides | 21.02 | 24.11 | 23.31 | 0.8960 | ||||||
| Porphyromonadaceae | 1.69 | 5.01 | 3.79 | 0.1920 | ||||||
| Parabacteroides | 1.40 | 1.69 | 1.34 | 0.8780 | ||||||
| Porphyromonas | 0.12 | 3.17 | 2.25 | 0.2210 | ||||||
| Prevotellaceae | 3.32 | 19.92 | 20.65 | 0.0295 | ||||||
| Prevotella | 3.05 | 19.71 | 20.62 | 0.0282 | ||||||
| Rikenellaceae | 1.97 | 0.74 | 0.41 | 0.0046 | ||||||
| Alistipes | 1.92 | 0.70 | 0.40 | 0.0048 | ||||||
| Firmicutes | 66.78 | 42.62 | 42.39 | 0.0022 | ||||||
| Clostridiales family XI | 4.21 | 12.87 | 13.59 | 0.2110 | ||||||
| Anaerococcus | 0.35 | 1.82 | 2.21 | 0.1240 | ||||||
| Finegoldia | 0.03 | 2.00 | 1.96 | 0.0882 | ||||||
| Peptoniphilus | 0.36 | 2.85 | 4.10 | 0.0372 | ||||||
| Unknown# | 3.39 | 5.94 | 4.82 | 0.8140 | ||||||
| Lachnospiraceae | 32.23 | 12.84 | 14.42 | 0.0008 | ||||||
| Blautia | 12.76 | 3.17 | 4.94 | 0.0058 | ||||||
| Clostridium | 1.13 | 1.22 | 1.11 | 0.9280 | ||||||
| Coprococcus | 3.69 | 1.32 | 1.03 | 0.0006 | ||||||
| Eubacterium | 1.18 | 0.36 | 0.29 | 0.0000 | ||||||
| Roseburia | 4.48 | 1.68 | 2.13 | 0.0155 | ||||||
| Ruminococcus | 1.74 | 1.21 | 0.77 | 0.0815 | ||||||
| Ruminococcaceae | 18.82 | 10.42 | 7.01 | 0.0007 | ||||||
| Faecalibacterium | 8.15 | 4.73 | 4.27 | 0.1040 | ||||||
| Oscillospira | 3.31 | 1.31 | 0.71 | 0.0254 | ||||||
| Ruminococcus | 1.66 | 1.39 | 0.33 | 0.0473 | ||||||
| Unknown# | 1.71 | 1.20 | 0.46 | 0.0270 | ||||||
| Veillonellaceae | 7.55 | 3.70 | 4.55 | 0.3460 | ||||||
| Acidaminococcus | 0.24 | 0.04 | 1.16 | 0.1520 | ||||||
| Dialister | 1.02 | 1.46 | 1.84 | 0.3770 | ||||||
| Megamonas | 2.72 | 0.03 | 0.27 | 0.2710 | ||||||
| Proteobacteria | 0.87 | 3.13 | 6.03 | 0.0007 | ||||||
| Alcaligenaceae | 0.21 | 0.97 | 1.11 | 0.0084 | ||||||
| Campylobacteraceae | 0.04 | 1.53 | 2.06 | 0.1470 | ||||||
| Campylobacter | 0.04 | 1.53 | 2.06 | 0.1470 | ||||||
| Enterobacteriaceae | 0.37 | 0.43 | 2.61 | 0.0075 | ||||||
| Escherichia | 0.30 | 0.37 | 2.36 | 0.0124 |
- Numbers listed under study groups are percentages.
- * Average abundance in gut microflora of healthy subjects (n = 16).
- † Average abundance in gut microflora of obese patients (n = 25).
- ‡ Average abundance in gut microflora of NASH patients (n = 22).
- § Phyla with average abundance greater than 1% in any of the disease groups are listed.
- ∥ Families with average abundance greater than 1% in any of the disease groups are listed.
- ¶ Genera with average abundance greater than 1% in any of the disease groups are listed.
- # 16S rRNA sequence distinct from any known genera in this family.
Within phylum Actinobacteria (Table 2), the only abundant family Bifidobacteriaceae and the only abundant genus Bifidobacterium were differently represented in the study groups. A progressive decreased abundance was observed from the healthy group to the obese group and then to the NASH group.
Within phylum Bacteroidetes (Table 2), family Prevotellaceae exhibited a >6-fold increase in the obese group and NASH group, compared to the healthy group, accounting for most of the increased abundance in Bacteroidetes phylum in the obese and NASH groups. Most of the Prevotellaceae sequences belonged to a single-genus Prevotella. Another noteworthy fact in this phylum was that there was a ∼20 fold increase of abundance in the genus Porphyromonas (family Porphyromonadaceae), but statistical significance was not achieved because of the large intragroup variances with the obese group and the NASH group. In contrast, a small, but significant, decrease was observed with Rikenellaceae, in which most of the sequences belonged to a single-genus Alistipes.
The decreased representation of Firmicutes in the obese group and the NASH group were mostly explained by the decreased abundance in two families: Lachanospiraceae and Ruminococcaceae (Table 2). Although most of the genera in these two families exhibited a similar trend (i.e., decreased abundance in the obese group and the NASH group, compared to the healthy group), it is noteworthy that the often pathogenic genus, Clostridium, exhibited similar representation among all groups. Also worth noting is that the most abundant genera in the Firmicutes phylum, Blautia and Faecalibacterium, showed the greatest reduction in abundance in the obese group and the NASH group.
Increased abundance of Proteobacteria in the obese and NASH groups was mainly explained by the increased abundance of Enterobacteriaceae (Table 2). Importantly, Enterobacteriaceae was the only abundant family (within the whole bacteria domain) exhibiting a significant difference between the obese group and the NASH group (Table 2; Fig. 3C). Most of the Enterobacteriaceae sequences belonged to Escherichia (Table 2; Fig. 3D), which is the only abundant genus within the whole bacteria domain exhibiting a significant difference between the obese group and the NASH group.
Furthermore, the OTUs within Escherichia were examined. A single OTU was found to dominate the sequences in Escherichia: OTU #20341 was found to account for 83%, 88%, and 90% of the Escherichia sequences in the healthy, obese, and NASH groups, respectively. The representative sequence of this OTU was then used to BLAST against the 16S rRNA sequences (Bacteria and Archaea) on the National Center for Biotechnology Information website (available at: http://blast.ncbi.nlm.nih.gov). The top four matches were sequences of Escherichia species, followed by two sequences of Cronobacter, another Enterobacteriaceae genus (Supporting Table 3; Supporting Fig. 4). Note that Shigella species have been reclassified as Escherichia coli strains based on genetic evidence.25 Similar results were obtained when the OTU #20341 sequence was searched against the Ribosomal Database Project database with the SeqMatch tool (Supporting Table 4).
Serum Alcohol Concentration in Healthy Subjects and Obese and NASH Patients.
Elevated alcohol-producing bacteria in the NASH microbiota prompted us to examine the endogenous ethanol production in patients and healthy controls. Because it is not feasible to obtain portal blood where highest ethanol concentration is expected, peripheral blood was used to determine serum ethanol concentrations of healthy subjects and obese and NASH patients (Fig. 4). A significantly elevated serum ethanol concentration was observed with NASH patients, when compared to healthy subjects and obese patients. Serum ethanol concentration was not different between healthy subjects and obese patients.
Elevated serum ethanol concentration in NASH patients. Serum ethanol of healthy subjects (healthy; n = 10), obese patients (n = 7), and NASH patients (n = 13) were measured using an ethanol assay kit from BioVision (Milpitas, CA). Data represent mean ± standard error of the mean. Significant difference was detected among three groups (P < 0.001; ANOVA). **P < 0.01 in Tukey's honest significance test.
Discussion
Here, we characterized gut microbiomes of NASH, obese, and healthy children and adolescents. Ecological diversities (alpha and beta) were different among three groups, indicating a strong connection between gut microbiomes and liver health. Each health status is associated with a unique pattern of enterotyping. Abundant differences among three groups were observed at phylum, family, and genus levels (Table 2). However, fewer differences were observed between obese and NASH microbiomes. Among taxa with greater than 1% representation, Proteobacteria, Enterobacteriaceae, and Escherichia were the only phylum, family, and genus exhibiting significant difference between obese and NASH microbiome. Proteobacteria/Enterobacteriaceae/Escherichia was similarly represented between healthy and obese microbiomes, but was significantly elevated in NASH. A strikingly similar pattern was observed with blood alcohol concentrations of healthy, obese, and NASH patients.
Liver ultrasound indicated that some obese patients had fatty liver and others did not. No significant difference was observed between these two subgroups in gut microbiomes at all taxonomic levels, possibly the result of the small sample sizes of both subgroups. Future studies with larger sample sizes may reveal differences in gut microbiomes between these two subgroups of obese patients.
Under normal conditions, alcohol is constantly produced in the human body.26 Intestinal microbiota is the major source of endogenous alcohol, as suggested by the increased blood alcohol level after intake of alcohol-free food.26-28 This endogenously produced alcohol is immediately and almost completely removed from portal blood by liver alcohol dehydrogenases (ADHs), catalases, and microsomal ethanol-oxidizing system. When ADH is inhibited, blood alcohol levels increase.26 Production of ethanol in the gut is also reflected by the fact that the liver and gastrointestinal tract have the highest activities of ADHs.29
Elevated breath alcohol levels are observed in obese mice, in which abnormal intestinal microbiota is the source for increased alcohol production and neomycin treatment decreases alcohol concentration.30 Because NASH patients are generally obese and liver histology is the same as that observed in ALD, it was hypothesized that NASH patients may also have elevated blood alcohol.30 The alcohol hypothesis of NASH could also explain the observation of increased gut permeability31 and, possibly, elevated serum lipopolysaccharide in NASH patients,32 because alcohol is known to increase gut permeability.33 The first evidence in support of this hypothesis was that the gene expression of all three major pathways for ethanol catabolism in NASH liver are highly elevated.19 Recently, elevated blood ethanol concentration was observed in NAFLD patients.34 The observation of Volynets et al.34 provides a link between blood alcohol and NAFLD. Our data further clarified that the blood ethanol concentration of obese patients without NASH is not elevated; however, obese patients with NASH had significantly elevated blood ethanol.
In anaerobic conditions, Enterobacteriaceae, including Escherichia, takes the mixed-acid fermentation pathway, a major product of which is ethanol.35-39 Because the pediatric patients had no access to frequent alcoholic beverages and no dietary source for alcohol was found, the colon microbiota is likely the major source for the elevated blood alcohol concentration in our NASH patients. Our findings of increased abundance of Escherichia in NASH microbiomes, and of the elevated blood alcohol in NASH patients, together with the well-established role for alcohol metabolism in the generation of reactive oxygen species (ROS) and, consequently, liver inflammation,40 suggest a novel mechanism for the pathogenesis of NASH is: Gut microbiota enriched in alcohol-producing bacteria (e.g., E. coli) constantly produce more alcohol than healthy microbiota and therefore supply a constant source of ROS to the liver, which, in turn, causes liver inflammation.
The most direct support for this hypothesis would come from the measurement of portal blood ethanol, which is not feasible with human subjects. Because no adequate NASH animal model bearing human microbiomes is available, we planned to examine the alcohol production by cultured endogenous Escherichia, but found that the experiment was performed many times 60 years ago. A typical result is that 1 g (wet weight) of E. coli produces 0.8 g of ethanol in 1 hour.35 Because the adolescent colon is 1-2 L,41 one would estimate that gut microbiomes of NASH patients could produce a significant ethanol insult for the liver, considering the fact that chronic intake of ∼30 g of alcohol per day will cause liver damage.42, 43 Moreover, because fat deposition sensitizes the liver to alcohol insult, even a relatively lower level of alcohol is sufficient to cause inflammation and fibrosis.44
A recent study using inflammasome-deficient mice supports that modulation of gut microbiomes exerts a critical influence on liver health, in particular, the progression of NAFLD.45 Nevertheless, before direct evidence is available, we cannot exclude the opposite scenario: the possibility that change in gut microbiomes is a consequence of liver disease. More specifically, could fatty liver cause increased abundance of Escherichia in the gut? This seems unlikely, because Spencer et al. demonstrated that induction of fatty liver diseases caused a decreased abundance of Proteobacteria (to which Escherichia belongs) in gut microbiomes.18
Besides Escherichia, other gut microbial genera, including Bacteroides,46 Bifidobacterium,47 and Clostridium,48 are capable of producing alcohol, and, collectively, these genera may generate a significant burden for liver-alcohol–scavenging mechanisms. These bacteria and, perhaps, yeast not assessed in this study may explain why some NASH patients had blood ethanol concentrations higher than healthy subjects, even though their microbiomes did not exhibit an increased abundance in Escherichia.
In conclusion, our study revealed unique characters in the composition, ecological diversity, and enterotyping patterns of gut microbiomes of NASH patients, in comparison to those of healthy and obese patients. The most outstanding character was the elevated representation of alcohol-producing bacteria in NASH microbiomes. Increased blood alcohol concentration was also observed with NASH patients. Our data suggest that microbiomes rich in ethanol-producing Escherichia may be a risk factor in driving the disease progression from obesity to NASH. An immediate future task is to characterize and quantitate the activities of the alcohol-producing genes in microbiomes. Another important future direction is to identify the cause(s) for the increased level of Escherichia in NASH gut microbiomes. Possible causes for elevated Escherichia include special dietary components, as suggested by a recent report that subtherapeutic doses of antibiotics leads to increased Escherichia in swine gut microbiomes.49 Microbiomes in patients with NASH may offer a unique opportunity for interrupting disease progression.
Acknowledgements
The authors thank Dr. Pearay Ogra (UB) and Dr. W. Allan Walker (Harvard) for their critical comments and invaluable suggestions and Jonathan E. Bard (UB Next-Generation Sequencing and Expression Analysis Core) for his expert assistance with the QIIME.
