Abnormal Fecal Microbiota in Patients With Hepatitis B
Abnormal Fecal Microbiota in Patients With Hepatitis B
The proportion of high-quality reads in all 40 of the fecal DNA samples was about 91%, and the actual insert sizes ranged from 327–384bp. We obtained an average of 18,852,250 paired-end reads and 8,657,932 single-end reads for each sample, making up a total 1.3 billion high-quality reads that were free of human DNA and adaptor contaminants. We constructed non-redundant gene catalogues that contained 1,603,579 genes from the patient samples (an average of 80,179 genes per sample) and 2,118,215 genes from the normal samples (an average of 105,911 genes per sample).
Analysis on the composition of fecal microbiota (Figure 1) clearly demonstrated that the samples from the cirrhotic patients contained reduced numbers of Bacteroidetes and increased numbers of Proteobacteria compared with the normal samples. Specifically, our data showed that Bacteroidetes made up 53% of the normal fecal microbiota but only 4% of the HBLC fecal microbiota, whereas Proteobacteria, which contains most of the opportunistic pathogens, made up only 4% of the normal fecal microbiota but increased to 43% of the HBLC fecal microbiota. These trends observed at the phylum level were also seen at the family level. For example, the Enterobacteriaceae, Veillonellaceae, and Streptococcaceae families made up less than 1% in normal fecal microbiota but were relatively dominant, with proportions ranging from 18–39% in HBLC fecal microbiota, which is consistent with the results of Chen and Zhao for intestinal microbiota from cirrhosis patients.
(Enlarge Image)
Figure 1.
Composition of fecal microbiota from HBLC patients and healthy individuals. Data are represented as the average percentage of each individual profile. Fecal microbiota community analysis at the phylum, class, order, family levels demonstrated that the fecal microbiota from the HBLC patients contained a significant absence of Bacteroidetes (4% compared with 53% in the microbiota from the controls) and enrichment of Proteobacteria (43% compared with 4% in the microbiota from the controls), which included most of the pathogens. The microbiota in the HBLC patients showed a significant enrichment of Gammaproteobac (42%), Negativicutes (21%), and Bacilli (20%) at the class level, Enterobacteriales (41%), Selenomonadales(20%), and Lactobacillales (19%) at the order level, and Enterobacteriaceae (39%), Veillonellaceae (19%), and Streptococcaceae (18%) at the family level.
To further investigate the differences in fecal microbiota between the HBLC and normal samples, PCA was employed. We found that the first and second principal components clearly separated the normal and HBLC fecal microbiota structures (Figure 2A). Some of the species with significantly different abundance between the two groups were Escherichia coli, Veillonella dispar, Veillonella parvula, which were significantly enriched, and Bacteroides species, which were reduced, in the HBLC samples compared with the normal samples.
(Enlarge Image)
Figure 2.
Comparison of healthy and HBLC microbial structure. (A) Bacterial species abundance differentiates HBLC patients and healthy individuals. The first five principal components (P value in Tracy-Widom test < 0.05 and contribution > 3%) were examined: PC1 = 33.38%, PC2 = 21.81%, PC3 = 12.74%, PC4 = 10.31%, and PC5 = 6.17%. The first two components (PC1 and PC2) are plotted. (B) Species annotation of the significantly differential genes. Blue bars represent gene coverage of the significantly differential species, red bars represent gene coverage of the significantly differential species in the HBLC samples, and green bars represent gene coverage of the significantly differential species in the control samples. (C) Bacterial groups quantified using real-time qPCR. Blue bars represent the control samples, and red bars represent the HBLC samples. The Student t test was used to evaluate the statistical difference between the two groups. *P < 0.05; ** P < 0.01.
To find which species were significantly absent or enriched in patients, we used the Wilcoxon rank-sum test method to analyze the genes that were in significantly differential abundance between the HBLC and normal fecal microbiota. We found that 49,442 genes and 72,584 genes were significantly enriched (P < 0.01) in patients and normal samples, respectively. Of these, a total of 103,099 genes could be matched to known species with a large ratio of coverage (Figure 2B). We found that the fecal microbiota from the cirrhotic patients contained a remarkable absence of Bacteroides cellulosilyticus, Bacteroides intestinalis, Bacteroides uniformis, Bacteroides ovatus, Bacteroides_fragilis, Bacteroides thetaitaomicron, Bacteroides sp.D1, Bacteroides eggerthii, Bacteroides stercoris, and Bacteroides vulgates, all of which belong to the Bacteroides genus (P < 0.01) and were negatively correlated with the CTP scores (R < −0.7). In contrast, the fecal microbiota from the cirrhotic patients contained high abundances of Escherichia coli, Klebsiella pneumonia, Enterobacter cloaca, Veillonella_parvula, Shigella dysenteriae, Veillonella dispar, Shigella flexneri, Salmonella enteric, Enterobacter cancerogenus, and Escherichia albertii, all of which belong to the Veillonella genus or Enterobacteriaceae family (P < 0.01) and were positively correlated with the CTP scores (R > 0.7).
A comparison of the dominant and subdominant bacteria genera between the HBLC and normal fecal microbiota was carried out by real-time qPCR (Figure 2C). Compared with the normal samples, Enterobacteriaceae and Veillonella were significantly increased, and Bacteroides and Clostridium were significantly decreased in the HBLC samples, confirming the results obtained from the high-throughput Illumina/Solexa sequencing.
A functional analysis of our data revealed functions that were enriched or decreased in the HBLC fecal microbiota compared with the normal fecal microbiota (Figure 3A, B). At the highest hierarchical levels, fecal microbiota from HBLC patients revealed an enrichment in amino acid transport and metabolism (P < 0.01), secondary metabolites biosynthesis, transport and catabolism (P = 0.005), inorganic ion transport and metabolism (P = 0.001), extracellular structures (P = 0.001), energy production and conversion (P = 0.014), and intracellular trafficking and secretion (P = 0.019), and a decrease in cell wall/membrane biogenesis (P = 0.047), signal transduction metabolism (P = 0.02), replication, recombination and repair (P = 0.018), and general function prediction only (P = 0.012). Interestingly, for most of these strikingly different functions, the relative abundance of genes in each category was statistically lower in the HBLC microbiota samples compared with their abundances in the normal microbiota samples, indicating a reduction in functional diversity in the HBLC microbiota.
Bile salt hydrolases (BSHs) are members of the choloylglycine hydrolase family (EC 3.5.1.24) and are important in bile acid metabolism. BSHs have been isolated and/or characterized from several species of intestinal bacteria. We found that genes annotated as BSH related to primary and secondary bile acid biosynthesis [KEGG:K01442] were in much higher abundance in the normal microbiota than in the HBLC microbiota (P = 0.013) (Figure 3C).
(Enlarge Image)
Figure 3.
Function and metabolism analysis of fecal microbiota from HBLC patients and control samples. Blue bars represent the control samples, and red bars represent the HBLC samples. The Student t test was used to evaluate the statistical difference between the two groups. *P < 0.05; ** P < 0.01 (A) Genes annotated using the eggNOG database. The significantly differential genes were annotated as: E, amino acid transport and metabolism; P, inorganic ion transport and metabolism; Q, secondary metabolites biosynthesis, transport and catabolism; and W, extracellular structures. (B) Genes annotated using the KEGG database. The significantly differential genes were annotated with pathways: Cellular, cell growth and death; Environmental, membrane transport; Environmental, signal transduction; Metabolism, biosynthesis of other secondary metabolites; Metabolism, energy metabolism; Metabolism, enzyme families; Metabolism, metabolism of other amino acids; and Metabolism, xenobiotics biodegradation and metabolism. (C) Relative abundance of genes annotated as BSH related to primary and secondary bile acid biosynthesis [KEGG: K01442]. Green bars represent the control samples, and red bars represent the HBLC samples. The Student t test was used to evaluate statistical difference between the two groups, P = 0.013.
Human gut bacteria are expected to encounter a broad spectrum of carbohydrate substrates. The fecal microbiota from the cirrhotic patients showed enhanced metabolic ability for carbohydrate transportation because of the high abundance of genes involved in phosphotransferase systems (P = 0.006) and ABC transporters (P = 0.004). Furthermore, the HBLC microbiota showed an enhanced ability to transform non-carbohydrate carbon substrates into glucose because of the high abundance of genes involved in the metabolism of gluconeogenesis (GNG) and pyruvate (pyruvate is one of the main gluconeogenic precursors) (P < 0.05).
In the lipid metabolism KEGG pathway, the fecal microbiota from the HBLC patients was significantly enriched in genes associated with fatty acid metabolism (R = 0.84, P = 0.025), unsaturated fatty acids biosynthesis (R = 0.59, P = 0.018), glycerophospholipid (R = 0.82, P = 0.004), alpha-linolenic acid (R = 0.77, P = 0.007), and butanoate (R = 0.78, P = 0.011), which may provide additional energy resources in these patients.
In the amino acid metabolism pathway, the fecal microbiota from HBLC patients was significantly enriched in genes associated with the metabolic ability of branched-chain amino acid (BCAA) involved in the valine, leucine and isoleucine biosynthesis (P = 0.030) and degration (P = 0.002). Conversely, genes associated with the metabolism of aromatic amino acids, including tyrosine (P = 0.0007), phenylalanine (P = 0.011), were less abundant in the microbiota from HBLC patients than in the microbiota from the healthy controls.
In addition, the fecal microbiota of HBLC patients were enriched for glutathione metabolism (P = 0.008), glutathione synthase (EC 6.3.2.3) and glutathione reductase (NADPH) (EC 1.8.1.7) (P < 0.01).
In brief, our data showed that the fecal microbiota structure, as well as a variety of functions were different in HBLC patients compared with in the normal controls, suggesting that the microbiota had changed to adjust to the cirrhosis-related intestinal microenvironment.
Results
High-throughput Illumina/Solexa Sequencing Data
The proportion of high-quality reads in all 40 of the fecal DNA samples was about 91%, and the actual insert sizes ranged from 327–384bp. We obtained an average of 18,852,250 paired-end reads and 8,657,932 single-end reads for each sample, making up a total 1.3 billion high-quality reads that were free of human DNA and adaptor contaminants. We constructed non-redundant gene catalogues that contained 1,603,579 genes from the patient samples (an average of 80,179 genes per sample) and 2,118,215 genes from the normal samples (an average of 105,911 genes per sample).
Comparison of the Microbial Structure in the Healthy and HBLC Samples
Analysis on the composition of fecal microbiota (Figure 1) clearly demonstrated that the samples from the cirrhotic patients contained reduced numbers of Bacteroidetes and increased numbers of Proteobacteria compared with the normal samples. Specifically, our data showed that Bacteroidetes made up 53% of the normal fecal microbiota but only 4% of the HBLC fecal microbiota, whereas Proteobacteria, which contains most of the opportunistic pathogens, made up only 4% of the normal fecal microbiota but increased to 43% of the HBLC fecal microbiota. These trends observed at the phylum level were also seen at the family level. For example, the Enterobacteriaceae, Veillonellaceae, and Streptococcaceae families made up less than 1% in normal fecal microbiota but were relatively dominant, with proportions ranging from 18–39% in HBLC fecal microbiota, which is consistent with the results of Chen and Zhao for intestinal microbiota from cirrhosis patients.
(Enlarge Image)
Figure 1.
Composition of fecal microbiota from HBLC patients and healthy individuals. Data are represented as the average percentage of each individual profile. Fecal microbiota community analysis at the phylum, class, order, family levels demonstrated that the fecal microbiota from the HBLC patients contained a significant absence of Bacteroidetes (4% compared with 53% in the microbiota from the controls) and enrichment of Proteobacteria (43% compared with 4% in the microbiota from the controls), which included most of the pathogens. The microbiota in the HBLC patients showed a significant enrichment of Gammaproteobac (42%), Negativicutes (21%), and Bacilli (20%) at the class level, Enterobacteriales (41%), Selenomonadales(20%), and Lactobacillales (19%) at the order level, and Enterobacteriaceae (39%), Veillonellaceae (19%), and Streptococcaceae (18%) at the family level.
To further investigate the differences in fecal microbiota between the HBLC and normal samples, PCA was employed. We found that the first and second principal components clearly separated the normal and HBLC fecal microbiota structures (Figure 2A). Some of the species with significantly different abundance between the two groups were Escherichia coli, Veillonella dispar, Veillonella parvula, which were significantly enriched, and Bacteroides species, which were reduced, in the HBLC samples compared with the normal samples.
(Enlarge Image)
Figure 2.
Comparison of healthy and HBLC microbial structure. (A) Bacterial species abundance differentiates HBLC patients and healthy individuals. The first five principal components (P value in Tracy-Widom test < 0.05 and contribution > 3%) were examined: PC1 = 33.38%, PC2 = 21.81%, PC3 = 12.74%, PC4 = 10.31%, and PC5 = 6.17%. The first two components (PC1 and PC2) are plotted. (B) Species annotation of the significantly differential genes. Blue bars represent gene coverage of the significantly differential species, red bars represent gene coverage of the significantly differential species in the HBLC samples, and green bars represent gene coverage of the significantly differential species in the control samples. (C) Bacterial groups quantified using real-time qPCR. Blue bars represent the control samples, and red bars represent the HBLC samples. The Student t test was used to evaluate the statistical difference between the two groups. *P < 0.05; ** P < 0.01.
To find which species were significantly absent or enriched in patients, we used the Wilcoxon rank-sum test method to analyze the genes that were in significantly differential abundance between the HBLC and normal fecal microbiota. We found that 49,442 genes and 72,584 genes were significantly enriched (P < 0.01) in patients and normal samples, respectively. Of these, a total of 103,099 genes could be matched to known species with a large ratio of coverage (Figure 2B). We found that the fecal microbiota from the cirrhotic patients contained a remarkable absence of Bacteroides cellulosilyticus, Bacteroides intestinalis, Bacteroides uniformis, Bacteroides ovatus, Bacteroides_fragilis, Bacteroides thetaitaomicron, Bacteroides sp.D1, Bacteroides eggerthii, Bacteroides stercoris, and Bacteroides vulgates, all of which belong to the Bacteroides genus (P < 0.01) and were negatively correlated with the CTP scores (R < −0.7). In contrast, the fecal microbiota from the cirrhotic patients contained high abundances of Escherichia coli, Klebsiella pneumonia, Enterobacter cloaca, Veillonella_parvula, Shigella dysenteriae, Veillonella dispar, Shigella flexneri, Salmonella enteric, Enterobacter cancerogenus, and Escherichia albertii, all of which belong to the Veillonella genus or Enterobacteriaceae family (P < 0.01) and were positively correlated with the CTP scores (R > 0.7).
A comparison of the dominant and subdominant bacteria genera between the HBLC and normal fecal microbiota was carried out by real-time qPCR (Figure 2C). Compared with the normal samples, Enterobacteriaceae and Veillonella were significantly increased, and Bacteroides and Clostridium were significantly decreased in the HBLC samples, confirming the results obtained from the high-throughput Illumina/Solexa sequencing.
Functional Characterization of HBLC Fecal Microbiota
A functional analysis of our data revealed functions that were enriched or decreased in the HBLC fecal microbiota compared with the normal fecal microbiota (Figure 3A, B). At the highest hierarchical levels, fecal microbiota from HBLC patients revealed an enrichment in amino acid transport and metabolism (P < 0.01), secondary metabolites biosynthesis, transport and catabolism (P = 0.005), inorganic ion transport and metabolism (P = 0.001), extracellular structures (P = 0.001), energy production and conversion (P = 0.014), and intracellular trafficking and secretion (P = 0.019), and a decrease in cell wall/membrane biogenesis (P = 0.047), signal transduction metabolism (P = 0.02), replication, recombination and repair (P = 0.018), and general function prediction only (P = 0.012). Interestingly, for most of these strikingly different functions, the relative abundance of genes in each category was statistically lower in the HBLC microbiota samples compared with their abundances in the normal microbiota samples, indicating a reduction in functional diversity in the HBLC microbiota.
Bile salt hydrolases (BSHs) are members of the choloylglycine hydrolase family (EC 3.5.1.24) and are important in bile acid metabolism. BSHs have been isolated and/or characterized from several species of intestinal bacteria. We found that genes annotated as BSH related to primary and secondary bile acid biosynthesis [KEGG:K01442] were in much higher abundance in the normal microbiota than in the HBLC microbiota (P = 0.013) (Figure 3C).
(Enlarge Image)
Figure 3.
Function and metabolism analysis of fecal microbiota from HBLC patients and control samples. Blue bars represent the control samples, and red bars represent the HBLC samples. The Student t test was used to evaluate the statistical difference between the two groups. *P < 0.05; ** P < 0.01 (A) Genes annotated using the eggNOG database. The significantly differential genes were annotated as: E, amino acid transport and metabolism; P, inorganic ion transport and metabolism; Q, secondary metabolites biosynthesis, transport and catabolism; and W, extracellular structures. (B) Genes annotated using the KEGG database. The significantly differential genes were annotated with pathways: Cellular, cell growth and death; Environmental, membrane transport; Environmental, signal transduction; Metabolism, biosynthesis of other secondary metabolites; Metabolism, energy metabolism; Metabolism, enzyme families; Metabolism, metabolism of other amino acids; and Metabolism, xenobiotics biodegradation and metabolism. (C) Relative abundance of genes annotated as BSH related to primary and secondary bile acid biosynthesis [KEGG: K01442]. Green bars represent the control samples, and red bars represent the HBLC samples. The Student t test was used to evaluate statistical difference between the two groups, P = 0.013.
Human gut bacteria are expected to encounter a broad spectrum of carbohydrate substrates. The fecal microbiota from the cirrhotic patients showed enhanced metabolic ability for carbohydrate transportation because of the high abundance of genes involved in phosphotransferase systems (P = 0.006) and ABC transporters (P = 0.004). Furthermore, the HBLC microbiota showed an enhanced ability to transform non-carbohydrate carbon substrates into glucose because of the high abundance of genes involved in the metabolism of gluconeogenesis (GNG) and pyruvate (pyruvate is one of the main gluconeogenic precursors) (P < 0.05).
In the lipid metabolism KEGG pathway, the fecal microbiota from the HBLC patients was significantly enriched in genes associated with fatty acid metabolism (R = 0.84, P = 0.025), unsaturated fatty acids biosynthesis (R = 0.59, P = 0.018), glycerophospholipid (R = 0.82, P = 0.004), alpha-linolenic acid (R = 0.77, P = 0.007), and butanoate (R = 0.78, P = 0.011), which may provide additional energy resources in these patients.
In the amino acid metabolism pathway, the fecal microbiota from HBLC patients was significantly enriched in genes associated with the metabolic ability of branched-chain amino acid (BCAA) involved in the valine, leucine and isoleucine biosynthesis (P = 0.030) and degration (P = 0.002). Conversely, genes associated with the metabolism of aromatic amino acids, including tyrosine (P = 0.0007), phenylalanine (P = 0.011), were less abundant in the microbiota from HBLC patients than in the microbiota from the healthy controls.
In addition, the fecal microbiota of HBLC patients were enriched for glutathione metabolism (P = 0.008), glutathione synthase (EC 6.3.2.3) and glutathione reductase (NADPH) (EC 1.8.1.7) (P < 0.01).
In brief, our data showed that the fecal microbiota structure, as well as a variety of functions were different in HBLC patients compared with in the normal controls, suggesting that the microbiota had changed to adjust to the cirrhosis-related intestinal microenvironment.
Source...