An Integrated Multicenter Amplicon Sequencing Data Reveals New Evidence of the Interaction Between the Gut Microbiota and Irritable Bowel Syndrome

doi:10.21203/rs.3.rs-2208039/v1

Download PDF

Research Article

An Integrated Multicenter Amplicon Sequencing Data Reveals New Evidence of the Interaction Between the Gut Microbiota and Irritable Bowel Syndrome

https://doi.org/10.21203/rs.3.rs-2208039/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 11 Feb, 2023

Read the published version in Journal of Translational Medicine →

You are reading this latest preprint version

Background

Gut dysbacteriosis has been reported as one of the etiologies for irritable bowel syndrome (IBS). However, the association between gut microbiota and IBS is still inconclusive. The study aims to provide new evidence of the interaction between the gut Microbiota and IBS.

Method

A paired-sample study was designed by retrieving original multicenter 16s-rRNA data of IBS patients and healthy controls from the GMrepo database. The propensity score matching (PSM) algorithm was applied to reduce confounding bias. The differential analysis of microbiota composition was performed at different taxonomic levels. The co-occurrence network was established. Subgroup analysis was performed to identify specific microbial compositions in different IBS subtypes.

Results

A total of 1522 amplicon samples were initially enrolled. After PSM, 708 samples (354 IBS and 354 healthy individuals) were eligible for further analysis. A total of 1,160 genera were identified. We identified significantly changed taxa in IBS groups (IBS-enriched: the families Enterobacteriaceae, Moraxellaceae, and Sphingobacteriaceae; the genera Streptococcus, Bacillus, Enterocloster, Sphingobacterium, Holdemania, and Acinetobacter. IBS-depleted: the phyla Firmicutes, Euryarchaeota, Cyanobacteria, Acidobacteria, and Lentisphaerae; the families Bifidobacteriaceae, Ruminococcaceae, Methanobacteriaceae, and the other 25 families; the genera Faecalibacterium, Bifidobacterium, and other 68 genera). In subgroup analysis, we profiled microbial compositions in IBS with predominant diarrhea and constipation. We further identified the genera Bilophila and Enterocloster that may be involved in linking IBS with psychiatric disorders. The co-occurrence network identified three hub genera and six hub species (including Faecalibacterium prausnitzii) that may be important in IBS pathophysiology. Strong positive interactions were identified among the Bifidobacterium longum, Bifidobacterium breve, and Bifidobacterium adolescentis in the Bifidobacterium community.

Conclusion

This study provides updated evidence in identifying specific microbes that may involve in IBS pathogenesis. Future modalities may be further validated by targeting these microorganisms.

Irritable Bowel Syndrome

Gut Microbiota

Amplicon Sequencing Analysis

propensity score matching

GMrepo Database

Irritable bowel syndrome (IBS) is a commonly encountered functional gastrointestinal disorder with a relatively high prevalence of 7–16% population and an estimated annual medical cost of more than $1 billion in the United States[1]. For decades, the pathophysiology of IBS has not been well explained. Various factors may be involved in the pathogenesis of IBS, including impaired gut-brain interactions, gut microbiome dysbiosis, altered GI motility, and abnormal visceral sensation [2, 3].

In recent years, advances in 16S rRNA and metagenome sequencing technologies have facilitated the exploration of the gut microbiota composition in patients with IBS [4]. Several cross-sectional or case-control studies have demonstrated that IBS patients exhibit a reduction in beneficial commensal genera compared with healthy individuals. However, microbial compositions in IBS individuals are still inconsistent in different studies, probably due to various methodologies they applied to identify microbes for distinguishing IBS and healthy controls. Thus, the taxonomic change of IBS on specific microbes is still conflicting and inconclusive [5]. Furthermore, most relevant studies were single-center with small sample sizes. Several studies [6–9] even failed to establish a convincing case-control comparison due to the lack of specific individual information, such as sex and age. Therefore, multicenter microbial data with relatively large sample sizes are still needed to clarify the microbial change in the pathophysiology of IBS. In addition, fecal microbiota transplant (FMT) may relieve the symptoms in patients with IBS by improving microbial dysbiosis. However, the therapeutic efficacy of FMT is also controversial based on the contradictory results reported from different studies [10, 11].

Therefore, we designed a paired-sample microbial analysis by integrating the 16s rRNA sequencing data of IBS and healthy controls from multicenter studies in the GMrepo database [12]. This study will clarify a more precise microbiota panel in IBS by providing an updated quantitative interpretation of microbial composition differences between IBS and healthy individuals and exhibiting a visualization of the interaction of different microbial taxa in IBS patients. It will also provide new evidence for future FMT in IBS patients.

Study Design and Data Collection

This study enrolled microbial sequence data from human stool samples with IBS (Medical Subject Headings (MeSH) Unique ID: D043183) and health controls (MeSH Unique ID: D006262). The 16s rRNA sequencing data from human stool samples were retrieved from the GMrepo database (https://gmrepo.humangut.info), an easily-accessed online database facilitating the accessibility of the rapidly growing human microbial data [13]. This database provides microbiota data from different study projects with personal information on age, sex, region, and body mass index (BMI). We included eligible sequence data of patients: (1) diagnosed with IBS; (2) aged more than 18 years and with body-mass-index (BMI) of more than 15; (3) with positive quality control (QC) status in the database; (4) with accessible amplicon sequence data (16S rRNA gene sequence). Exclusion criteria include: (1) a recent history of antibiotics use; (2) a sample with missing information of sex, region, age, or BMI. (3) IBS combined with inflammatory bowel disease (IBD), severe intestine infection, or other chronic infectious diseases. The sub-types of IBS were determined by extracting the supplemental information of patients in the database. IBS-C was defined as IBS complicated with predominant constipation, and IBS-D as IBS complicated with predominant diarrhea. IBS with psychological disorders were defined as IBS patients accompanied by at least one of the following disorders: Migraine Disorders, Autism Spectrum Disorder (ASD), Depression, Bipolar Disorder, Schizophrenia, and Attention Deficit Disorder with Hyperactivity.

The PSM procedure

To reduce confounding bias, a propensity score matching (PSM) algorithm was applied to generate cohorts of IBS and their matched healthy individuals by controlling age, sex, region, and BMI using SPSS Statistics for Windows, Version 25.0 (IBM) Corporation, Armonk, NY) [14]. The match ratio of patients in both groups was 1:1, with 0.01 match tolerance.

Statistical analysis of the baseline characteristics

Statistical analysis was performed using R software (version 4.1.0). Normality tests were applied by the Shapiro-Wilk test. Data with normal distribution were considered as p-value > 0.05 and are presented as mean and standard deviation, and data with non-normal distribution are presented as median with Interquartile Range (Q). For comparisons, the student t-test was applied for data with normal distribution, and the Wilcoxon Rank Sum Test was performed in independent data with non-normal distribution. Pearson chi-square or Fisher’s exact test was applied to compare categorical variables. The R Packages of MatchIt, optmatch, and reshape2 were used to generate matched cohorts through the PSM algorithm.

Data processing, quality control and, taxonomic classification

FastQC (version 0.11.8) was applied to estimate the overall quality of the downloaded data, followed by the use of trimmomatic to remove vector sequences and low-quality bases. Microbial sequences shorter than ⅔ of the length of the original reads were removed from subsequent analysis. The remaining sequences were considered qualified clean data and were used for subsequent analysis. As for the steps of quality control, Samples (runs) with less than 20,000 clean reads were first removed from the subsequent analysis and marked as "failed QC (QC status = 0)". Then, after taxonomy assignment, samples containing only a single taxon (i.e., a species or genus that accounts for more than 99.99 percent of total abundance) will be marked as "failed QC". For whole genome (i.e., metagenomic or mNGS) sequences, MetaPhlAn2 was used with default parameters for the taxonomic classification of the sequencing reads. Each data of relative abundance in every sample was collected and integrated into a merged taxonomic abundance table. NCBI taxonomy database was searched for classifying organisms in different levels (Kingdom, Phylum, Class, Order, Family, Genus, Species). The taxonomic composition of each sample was integrated into a final taxonomy classification table.

The alpha and beta diversity

The diversity of the microbiome in the human data was evaluated by R software (version 4.1.0, http://www.R-project.org/). Alpha diversity was evaluated by the Shannon index, Simpson index, and richness (number of observed species), using the “vegan” package in R. The difference in alpha diversity was calculated by Wilcoxon rank-sum tests. Beta-diversity was presented by unconstrained principal coordinate analyses (PCoA) scatter plots via calculating Bray-Curtis distances. Permutational multivariate analysis of variance (PERMANOVA) was then used to determine the difference between different phenotypes [15].

Differential analyses of microbial compositions

The pairwise differential analysis was completed using the Wilcox.test in R. Multiple comparisons were performed using the Kruskal.test in R. Manhattan plot was generated using packages of metagenomeSeq and ggplot2. The Spearman correlation analysis was performed and visualized in R using the package corrplot, ggcor, and ggplot2. Clustering analysis with Z-score standardized data was applied to visualize the component difference of microbial species in the IBS and control groups, using the pheatmap package in R. The plot of clustering analysis, Venn, and species distribution in different subgroups were completed using the package “ggvenn” for Venn and package “ggplot2”, “tidyverse”, and “reshape2” for species distribution. Microbial interactions with age or BMI were investigated using the Spearman correlation. The adjusted p-values generated from multiple comparisons were also calculated using the false discovery rate (FDR) method.

Co-occurrence Network

A correlation matrix was constructed by calculating the pairwise Spearman’s rank correlations in all IBS samples. A correlation between two microbes was considered statistically robust if Spearman’s correlation coefficient was > 0.8 and the p-value was < 0.01 [16]. To reduce the chances of obtaining false-positive results, the p-values were adjusted using the Benjamini–Hochberg method. The molecular ecological network analyses (MENA) were applied to construct random matrix theory (RMT) based on co-occurrence bacterial networks and presented in Cytoscape Version 3.9.1. The most densely connected node was defined as the hub microbe, and hub microbes were identified using the cytohubba module.

Identification of microbial biomarker for IBS

Finally, a random forest model [17] was built using R package randomForest. Significant microorganisms were incorporated into a panel for classifying IBS, at the genus and species levels, respectively.

General characteristics of the study samples

Following the study criteria, 1522 amplicon samples (365 IBS and 1157 health controls) were identified. Before the PSM procedure, there were significant distribution differences in region (p < 0.001), sex (p < 0.001), and BMI (p = 0.027) between IBS and control groups. After PSM, the age, sex, region, and BMI showed no statistical difference between IBS and control groups. A total of 708 samples (354 IBS and 354 health controls) were enrolled in the final analysis. The baseline characteristics are shown in Table 1 and Table S1.

Table 1

Baseline characteristics of patients before and after propensity score matching.
No. of patients	Before Match			After Match
No. of patients	IBS (n = 365)	Controls (n = 1157)	p	IBS (n = 354)	Controls (n = 354)	p
Age (median [IQR])	45 [33, 59]	45 [35, 57]	0.407	45 [36, 57]	46 [35, 59]	0.610
BMI (median [IQR])	23.67 [21.46, 25.80]	22.80 [20.94, 5.62]	0.027*	22.90 [20.97, 25.65]	23.03 [20.98, 25.48]	0.605
Country (n, %)			< 0.001*			1.000
USA	175 (47.95)	832 (71.91)		175 (49.44)	175 (49.44)
UK	169 (46.30)	274 (23.68)		160 (45.20)	160 (45.20)
Australia	11 (3.01)	19 (1.64)		9 (2.54)	9 (2.54)
Switzerland	6 (1.64)	12 (1.04)		6 (1.69)	6 (1.69)
Canada	4 (1.10)	20 (1.73)		4 (1.13)	4 (1.13)
Sex (n, %)			< 0.001*			1.000
Female	238 (65.21)	488 (42.18)		227 (64.12)	227 (64.12)
Male	127 (34.79)	669 (57.82)		127 (35.88)	127 (35.88)
Note: ^†p value was derived from the Mann-Whitney test in data of continuous variables with abnormal distribution (M, Median; IQR, Interquartile Range). p value was derived from the Chi-square test or fisher’s exact test in data of categorical variables from IBS and healthy controls (n,%). Abbreviation: IBS: irritable bowel syndrome; BMI: body weight index.

The microbial diversities between IBS and healthy controls

The final data included human stool samples from five countries (Fig. 1A). A total of 1,160 microbial genera (including 3,463 species) were identified in both IBS and healthy control groups. Within all the identified microbes, IBS and healthy control groups shared 941 genera, whereas 131 genera exclusively exited in the IBS group and 108 genera exclusively identified in the control groups (Fig. 1B, Table S2). Among the 131 IBS-exclusive genera, the genera of Muricauda (phylum Bacteroidetes), Pelagerythrobacter (phylum Proteobacteria), and Rickettsia (phylum Proteobacteria) were the three most abundant microbes (Fig. 1C).

As for the alpha diversity, the richness of species, Shannon, and Simpson index in the IBS groups presented no difference compared to healthy controls (Fig. 1D-F, Table S3). When evaluating the beta diversity, we observed a significant difference in the Bray-Curtis distances in the Axis1 of PCoA (p = 0.006), but no significant difference in the Bray-Curtis distances in the Axis2 (p = 0.088) (Fig. 1G). The PERMANOVA test also revealed a significant dispersion difference (p = 0.005).

Differential analysis of microbes between IBS and healthy controls

At the phylum level, significantly decreased amounts of Firmicutes (p = 0.049), Euryarchaeota (p = 0.002), Cyanobacteria (p = 0.026), Acidobacteria (p = 0.049), and Lentisphaerae (p < 0.001) were observed in IBS groups compared with their healthy controls. The proportions of Proteobacteria and Bacteroidetes were increased in IBS groups, whereas the phyla Bacteroidetes, Actinobacteria, Verrucomicrobia, and Fusobacteria were depleted, but none of the six researched a statistical significance (Table S4). The Firmicutes/Bacteroidetes ratio showed no difference between the IBS and control groups (p = 0.130), whereas Firmicutes/Proteobacteria ratio was significantly decreased in the IBS group (p = 0.039) (Fig. 2A, 2B).

At the family level, the IBS group had significantly increased proportions of the family Moraxellaceae (phylum Proteobacteria) (p = 0.009), Sphingobacteriaceae (phylum Bacteroidetes) (p = 0.013), and Enterobacteriaceae (phylum Proteobacteria) (p = 0.046), whereas bacteria in the family Ruminococcaceae (phylum Firmicutes) (p = 0.001) and Bifidobacteriaceae (phylum Actinobacteria) (p = 0.017) and the other 25 taxa are significantly decreased in the IBS group. The family Lactobacillaceae (phylum Firmicutes) and Erysipelotrichaceae (phylum Firmicutes) tend to enrich in IBS groups, but none of the three reached a statistical significance (Fig. 2C, Table S5).

At the genus level, the genera Enterocloster, Sphingobacterium, Holdemania, Acinetobacter, Bacillus, and Streptococcus were significantly enriched in the IBS cohort. The genera Ruminococcus, Faecalibacterium, Bifidobacterium, and other 68 genera were significantly depleted in the IBS group (Fig. 2D, Table S6).

At the species level, 219 species had significantly different proportions of abundance between the IBS and healthy groups, including 9 species significantly enriched (Bacteroides fragilis, Blautia coccoides, Eggerthella lenta, Clostridium aldenense, Clostridium bolteae, Holdemania filiformis, Streptococcus oralis, Streptococcus mitis, Streptococcus suis) and 210 species significantly depleted in the IBS groups. Of the 219 species, 12 belonged to the genera Bifidobacterium, 10 belonged to the genera Clostridium, 7 belonged to the genera Streptococcus, 6 belonged to the genera Bacteroides, and 5 belonged to the genera Lactobacillus (Table S7)

Microbial correlations with environmental factors

To further investigate whether the microbial abundance fluctuates with age or BMI, we applied the Spearman correlation test to find microbes potentially correlated with age and BMI. We observed no genera with relatively strong correlations (|ρ|>0.4 and p < 0.05) with age or BMI, indicating that IBS individuals may have a relatively stable microbial composition unaffected by age or BMI (Table S8).

Sub-regional differential analysis

As geographical location may exhibit a great influence on the gut microbiota, we stratified data into subgroups by different regions. Figure 3A shows the comparisons of Shannon and Simpson indexes among countries. The Shannon index differs significantly between the USA and UK cohorts (p < 0.001), USA and Switzerland cohorts (p = 0.002), and Switzerland and Australia cohorts (p = 0.009). The Simpson index differs significantly between the USA and Switzerland cohorts (p < 0.001), UK and Switzerland cohorts (p < 0.001), and Switzerland and Australia cohorts (p = 0.001). However, the alpha diversities of the IBS and healthy controls had no significant difference within each country (Fig. 3B). The beta diversity showed no significant difference in the USA, UK, and Switzerland cohorts, whereas the Canada and Australia cohorts showed a statistical difference in the Bray-Curtis distances of Axis 2 in the PCoA analysis (Fig. 3C). Significantly-changed genera in each country cohort were shown in Table S9. Among significantly-different genera, Faecalibacterium, Sporobacter, and Pseudoclostridium were commonly depleted in IBS groups in samples of the USA, UK, and all five countries. (Fig. 3D) The abundance change of several significant genera in the IBS group within each country was further presented in Fig. 3E. We observed that these genera might exist different tendencies of the proportion change in the IBS groups in different countries. Faecalibacterium was depleted in the IBS groups in the UK, USA, and Australia, whereas it was slightly enriched in the IBS groups in the Canada and Switzerland cohorts. Bifidobacterium was depleted in the IBS groups in the UK, Switzerland, and Australia, whereas it was slightly enriched in the USA and Canada areas.

Subgroup analysis of microbial compositions in different IBS subtypes

Considering the microbial composition in different IBS subtypes may vary, we selected the 16s rRNA data of the IBS-C and IBS-D subtypes for further analysis. The differential analysis was performed in IBS-C, IBS-D, and healthy controls to identify significantly different genera among the three groups. At the phylum levels, there was no overt difference in microbial compositions among groups (Fig. 4A). As for the microbial diversity, the Shannon and Simpson indices were not significantly different among groups, whereas the Bray-Curtis distances of Axis 2 in PCoA analysis showed a significant difference between Health and IBS-C (p < 0.001) and between Health and IBS-D (p = 0.002) (Fig. 4B). We identified 24 genera (Kruskal-Wallis test, p < 0.05) with significantly different abundance among the three groups. Compared with healthy individuals, the following genera had increased abundance proportions in both the IBS-C and IBS-D subgroups: Streptococcus, Bacillus, Enterocloster, Sphingobacterium, and Holdemania. The genera Faecalibacterium, Ruminococcus, Oscillibacter, Coprococcus, and the other 11 genera were depleted in both IBS subgroups. (Fig. 4C) When compared with healthy individuals, four genera showed a completely different trend of abundance change between the IBS-C and IBS-D. The genera of Haemophilus, Peptoniphilus, and Roseburia were enriched in the IBS-D but depleted in IBS-C, whereas Anaerofilum was enriched in the IBS-C but depleted in IBS-D.

The microbial composition of IBS patients with psychiatric disorders

Of all the 354 patients with IBS, 136 were diagnosed with psychiatric disorders. We specifically investigated the difference in microbial compositions among IBS patients with or without psychiatric disorders. At the phylum levels, there was no overt difference in microbial compositions among groups (Fig. 5A). The Shannon and Simpson indices were not significantly different among groups, whereas the Bray-Curtis distances of Axis 2 in PCoA analysis showed a significant difference between Health and IBS with psychiatric disorders (IBS-PD) (p = 0.032) and between Health and IBS without psychiatric disorders (IBS-NPD) (p < 0.001) (Fig. 5B). The Kruskal-Wallis test identified 21 genera with significantly different abundance among the three groups. The genera Bilophila, Acidaminococcus, and Pseudoclostridium and the other five genera presented a step descent proportions of abundance in healthy, IBS-NPD, and IBS-PD, whereas the genera Enterocloster had a stepped growth. (Fig. 5C).

The co-occurrence network

The co-occurrence patterns among the IBS samples were explored using network inference based on strong (|ρ|>0.8) and significant (p < 0.01) correlations. The network was composed of 102 nodes (microbes) and 171 edges. The entire network contained 11 modules, with 81 of the 102 genera occupied by the top five modules (modules 1–5). The top 3 ranked hub microbes were the genera Butyricimonas, Christensenella, and Pseudoclostridium using the maximal clique centrality (MCC) method within the cytohubba (Fig. 6A).

To further explore the specific interactions of the microbial communities, we constructed three different networks at the species level in IBS subtypes of IBS-C, IBS-D, and IBS with psychiatric disorders. We observed similar interaction modules in the three groups (Fig. 6B-D). By calculating the MCC, the top 10 hub species were identified in the three groups, respectively (Table S10). We observed that six hub species were commonly shared by the three subgroups: Faecalibacterium prausnitzii, Blautia lutim, Roseburia inulinivorans, Bacteroides acidifaciens, Bifidobacterium merycicum, and Bacteroides luti. We also observed a strongly positive interactions among species of the genera Bifidobacterium in all three networks.

The interaction of the Bifidobacterium app.

To further investigate the interaction of species belonging to the genus Bifidobacterium (B), we specifically analyzed the correlations of 12 Bifidobacterium species identified as significantly changed species in the IBS groups. These 12 species were all significantly depleted in IBS individuals (Fig. 6E). The Spearman analysis revealed that B.longum, B.breve, and B.adolescentis had strong correlations (correlation coefficients (ρ) > 0.95, p < 0.001). Some relatively lower abundant species were also strongly correlated with other Bifidobacterium app. In these groups, B.merycicum was strongly correlated with B.catenulatum (ρ = 0.98, p < 0.001). B.callitrichos was strongly correlated with B.coryneforme (ρ = 0.97, p < 0.001) and B.lemurum (ρ = 0.97, p < 0.001). B.coryneforme was strongly correlated with B.lemurum (ρ = 0.96, p < 0.001) (Fig. 6F).

Identification of microbial biomarker for IBS

Finally, a random forest model was built and microorganisms with the top 20 values of the mean decrease accuracy were identified for classifying IBS phenotype at the genus and species levels, respectively. The Faecalibacterium, Pseudoclostridium, and Bifidobacterium were the top three genera (Fig. 7A) and the Holdemania filiformis, Bifidobacterium breve, and Bifidobacterium gallicum were the top 3 species (Fig. 7B).

To assess the difference in microbial compositions of IBS individuals versus healthy controls, we performed the integrated analysis using the multicenter amplicon data. Our study has the following characteristics: 1) the sample size is relatively large; 2) all data were from the fecal 16s rRNA sequence, which allows comparable results for bacterial profiling; 3) we applied the PSM methodology to reduce potential confounding bias; 4) we specifically compared the microbial compositions in IBS patients of different subtypes, including IBS patients with psychiatric disorders; 5) we established co-occurrence networks to investigate microorganism correlations and identified hub microbes that may play a key role in the whole bacterial community.

In general, IBS patients, compared with healthy controls, had distinguished microbial compositions at different taxonomic levels. Although the alpha diversity showed no difference, the beta diversity evaluated by the Bray-Curtis distances and PERMANOVA test revealed a dispersion difference, indicating heterogeneous microbiota distributions between IBS and control groups.

At the phylum level, the Firmicutes/Bacteroidetes ratio in the IBS group presented no significant change. However, we found that the Firmicutes/Proteobacteria ratio was significantly decreased in the IBS group. The phylum Proteobacteria has been previously reported as one of the significantly enriched phyla in IBS patients[18, 19]. In our study, we observed an increasing trend of Proteobacteria in IBS patients. Although this trend did not reach statistical significance, we found that the family Enterobacteriaceae belonging to Proteobacteria was significantly enriched in IBS individuals. The Enterobacteriaceae family has been reported as a potentially harmful taxon because it contains several pathogenic bacteria (Escherichia, Shigella, Campylobacter, Salmonella, etc.) and thus might be associated with enteric infections [20]. This family has also been reported to have positive correlations with some inflammatory markers such as interleukin-6 and interleukin-8 [21]. We further observed that four genera (Acinetobacter, Pseudomonas, Escherichia, Parasutterella) belonging to Proteobacteria were enriched in the IBS groups. This indicates that IBS patients may have enteric dysbacteriosis and potential inflammation, which promotes the growth of facultative, non-fastidious bacteria like Enterobacteriaceae. In addition to Enterobacteriaceae, we observed significant enrichment of another aerobe group: the genus Streptococcus. This is consistent with the study conducted by Rajilić-Stojanović et al [22] We also observed that Streptococcus was significantly enriched in both IBS-C and IBS-D subgroups, indicating that both diarrhea and constipation symptoms may be associated with underlying intestinal dysbacteriosis. At the species level, we noticed three significantly elevated species (Streptococcus oralis, Streptococcus mitis, and Streptococcus suis) in the IBS groups. The three species have been reported as the most frequently identified oral colonized pathogens [23]. This indicates that patients with IBS may also have altered oral microbiota, possibly due to different dietary intake and oral hygiene habits between IBS and healthy individuals. These results may also reflect the potential association between diet and IBS pathogenesis.

Our study also confirmed some potentially protective bacteria in IBS patients. The genus Faecalibacterium was significantly depleted in patients with IBS regardless of the subtype. Faecalibacterium sp. MC-41 and Faecalibacterium prausnitzii were the only two species detected in this classification, and both were significantly depleted in IBS groups. Faecalibacterium prausnitzii is a butyrate-producing microbe with anti-inflammation efficacy by mediating the expression of interleukin-17 [24], and it can also enhance the integrity of gastrointestinal barriers [25, 26]. We also identified that Faecalibacterium prausnitzii was one of the top 3 hub microbes with strong microbial interactions in the co-occurrence network. This finding has not been reported previously and it potentially indicates the vital role of Faecalibacterium prausnitzii in the signal pathways associated with IBS pathophysiology. Another group of protective commensal bacteria we noticed is the genus Bifidobacterium. Besides the significant depletion of Bifidobacterium in IBS, we found that species belonging to Bifidobacterium had strong positive correlations with each other, especially the positive interactions of B.longum, B.breve, B.adolescentis, and B.gallicum in all subtype networks. The strong positive correlation indicates that supplying a certain Bifidobacterium species may promote the growth of other species in this category. Thus, future FMT may be more targeted and effective by transplanting dominant Bifidobacterium species as it could stimulate the enrichment of the whole Bifidobacterium community.

Considering that some mental illnesses may accompany IBS patients, we further investigated the microbial composition in IBS patients with psychiatric disorders. The genera Bilophila presented the lowest abundance in IBS with psychiatric disease. This finding corresponds with several previous studies identifying the association between this genus and mental disorders. Bilophila is a bile-resistant organism. A decrease in the Bilophila abundance has been reported to be associated with ASD [27]. In our study, Bilophila wadsworthia was the only significantly depleted species identified in the Bilophila groups. The lower abundance of this species was also reported in patients with Alzheimer's disease [28]. Another group we noticed is the genera Enterocloster which had a stepped growth of abundance in groups of health, IBS without- and with psychiatric disorders. Enterocloster has been reported to increase in individuals suffering from GI symptoms and autism [29]. Interestingly, we observed that Enterocloster was also significantly enriched in IBS-D patients compared with IBS-C and healthy controls. This genus has been reported to improve the production of bile acids in the colon [30], and excess bile acids could increase stool water and result in diarrhea [31]. Thus, the genera Enterocloster may link the pathogenesis of IBS and mental illness.

Our study has also raised some concerns that should be further addressed. First, we observed some conflicting results during microbial differential analysis. For instance, several studies reported a significant increase in the family Lactobacillaceae, but we observed no statistical difference in this family in IBS and controls. Conflicting results were also seen in the phyla Firmicutes, Bacteroidetes, and the genus Ruminococcus, etc. This issue has been mentioned in several previous studies [5, 22]. The possible explanation is that microbial compositions may vary in different IBS subtypes: we found that the genera of Haemophilus, Peptoniphilus, Roseburia, and Anaerofilum presented completely different trends of abundance change between IBS-C and IBS-D. Besides, other environmental factors may also influence microbial distributions, such as diet structure, IBS symptom scores, and a history of previous intestinal infections. However, these data were unavailable in the GMrepo database, and we were unable to further analysis. In addition, similar to most previous studies, we did not use adjusted p-values (FDR) after multiple comparisons because many significantly-differentiated species presented increased false positive rates (> 0.1) after p correction. Based on the p-values reported previously, we speculate that it may also be the reason for most studies without performing p corrections. Furthermore, in the random forest model, we observed relatively low MDA values in all microbial biomarkers. All these findings indicate that the gut microbiota alone could not precisely distinguish IBS patients from healthy controls. It is necessary to establish a comprehensive predictive model for IBS by integrating multiple variables, including the intestinal microorganisms, essential metabolites, and clinicopathological characteristics, rather than using IBS-specific microbes alone. Further studies are still needed to investigate the causal relationships of specific microbes and IBS pathogenesis.

This study provides updated evidence in identifying specific microbes that may involve in IBS pathogenesis. Future modalities may be further validated by targeting these microorganisms.

Ethnic approval and consent to participate

The original data in this study were retrieved from the public GMrepo database with an open license for data use. Exception Privileges of ethical approval are granted.

Consent for publication

This manuscript has not been previously published. All authors have consented to the publication of the manuscript in this journal.

Acknowledgments

This work was supported by the grant from the National Natural Science Foundation of China (No. 82100594).

Competing interests

The authors have declared that no competing interest exists.

Funding

This study was funded by the National Natural Science Foundation of China (No. 82100594).

Authors’ contributions

Han Chen: study concept and design, analysis and interpretation of data; drafting of the manuscript, authorship; Rong Ou: study concept and design, analysis and interpretation of data, authorship. Nana Tang: data extraction, design and order the figures and tables, assessment of study quality, authorship. Wei Su, Ruoyun Yang, Xin Yu and Guoxin Zhang: design and order the figures and tables, assessment of study quality. Jianhua Jiao and Xiaoying Zhou: critical revision of the manuscript for important intellectual content; obtain funding; study supervision, authorship.

Acknowledgments

We thank that funding source: the National Natural Science Foundation of China (82100594) and public database (GMrepo).

Camilleri M. Diagnosis and Treatment of Irritable Bowel Syndrome: A Review. JAMA. 2021 Mar 2;325(9):865-877.
Ford AC, Sperber AD, Corsetti M, Camilleri M. Irritable bowel syndrome. Lancet. 2020 Nov 21;396(10263):1675-1688.
Holtmann GJ, Ford AC, Talley NJ. Pathophysiology of irritable bowel syndrome. Lancet Gastroenterol Hepatol. 2016 Oct;1(2):133-146.
Lynch SV, Pedersen O. The human intestinal microbiome in health anddisease. N Engl J Med 2016; 375:2369–2379.
Pittayanon R, Lau JT, Yuan Y, Leontiadis GI, Tse F, Surette M, Moayyedi P. Gut Microbiota in Patients With Irritable Bowel Syndrome-A Systematic Review. Gastroenterology. 2019 Jul;157(1):97-108.
Durban A, Abellan JJ, Jimenez-Hernandez N, et al. Structural alterations of faecal and mucosa-associated bacterial communities in irritable bowel syndrome. Environ Microbiol Rep 2012;4:242–247.
Rigsbee L, Agans R, Shankar V, Kenche H, Khamis HJ, Michail S, Paliy O. Quantitative profiling of gut microbiota of children with diarrhea-predominant irritable bowel syndrome. Am J Gastroenterol. 2012 Nov;107(11):1740-51.
Ringel-Kulka T, Benson AK, Carroll IM, et al. Molecular characterization of the intestinal microbiota in patients with and without abdominal bloating. Am J Physiol Gastrointest Liver Physiol 2016;310:G417–G426.
. Rangel I, Sundin J, Fuentes S, et al. The relationship between faecal-associated and mucosal-associated microbiota in irritable bowel syndrome patients and healthy subjects. Aliment Pharmacol Ther 2015;42:1211–1221.
Johnsen PH, Hilpüsch F, Valle PC, Goll R. The effect of fecal microbiota transplantation on IBS related quality of life and fatigue in moderate to severe non-constipated irritable bowel: Secondary endpoints of a double blind, randomized, placebo-controlled trial. EBioMedicine. 2020 Jan;51:102562.
Canakis A, Haroon M, Weber HC. Irritable bowel syndrome and gut microbiota. Curr Opin Endocrinol Diabetes Obes. 2020 Feb;27(1):28-35.
Wu S, Sun C, Li Y, Wang T, Jia L, Lai S, Yang Y, Luo P, Dai D, Yang YQ, Luo Q, Gao NL, Ning K, He LJ, Zhao XM, Chen WH. GMrepo: a database of curated and consistently annotated human gut metagenomes. Nucleic Acids Res. 2020 Jan 8;48(D1):D545-D553.
Dai D, Zhu J, Sun C, Li M, Liu J, Wu S, Ning K, He LJ, Zhao XM, Chen WH. GMrepo v2: a curated human gut microbiome database with special focus on disease markers and cross-dataset comparison. Nucleic Acids Res. 2022 Jan 7;50(D1):D777-D784.
Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res. 2011 May;46(3):399-424.
Tang ZZ, Chen G, Alekseyenko AV. PERMANOVA-S: association test for microbial community composition that accommodates confounders and multiple distances. Bioinformatics. 2016 Sep 1;32(17):2618-25.
Li B, Yang Y, Ma L. et al. Metagenomic and network analysis reveal wide distribution and co-occurrence of environmental antibiotic resistance genes. ISME J 9, 2490–2502 (2015).
Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC. et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77
Saulnier DM, Riehle K, Mistretta TA, et al. Gastrointestinal microbiome signatures of pediatric patients with irritable bowel syndrome. Gastroenterology 2011;141:1782–1791.
Rangel I, Sundin J, Fuentes S, et al. The relationship between faecal-associated and mucosal-associated microbiota in irritable bowel syndrome patients and healthy subjects. Aliment Pharmacol Ther 2015;42:1211-1221.
Zhu W, Winter MG, Byndloss MX, Spiga L, Duerkop BA, Hughes ER, Büttner L, de Lima Romão E, Behrendt CL, Lopez CA, Sifuentes-Dominguez L, Huff-Hardy K, Wilson RP, Gillis CC, Tükel Ç, Koh AY, Burstein E, Hooper LV, Bäumler AJ, Winter SE. Precision editing of the gut microbiota ameliorates colitis. Nature. 2018 Jan 11;553(7687):208-211.
Biagi E, Nylund L, Candela M, et al. Through ageing, and beyond: gut microbiota and inflammatory status in seniors and centenarians. PLoS One 2010;5:e10667.
Rajilić-Stojanović M, Biagi E, Heilig HG, Kajander K, Kekkonen RA, Tims S, de Vos WM. Global and deep molecular analysis of microbiota signatures in fecal samples from patients with irritable bowel syndrome. Gastroenterology. 2011 Nov;141(5):1792-801
Khan SA, Wingard JR. Infection and mucosal injury in cancer treatment. J Natl Cancer Inst Monogr. 2001;(29):31-6.
Zhang M, Qiu X, Zhang H, et al. Faecalibacterium prausnitzii inhibits interleukin-17 to ameliorate colorectal colitis in rats. PLoS One 2014;9:e109146.
Lopez-Siles M, Duncan SH, Garcia-Gil LJ, et al. Faecalibacterium prausnitzii: from microbiology to diagnostics and prognostics. ISME J 2017;11:841–852.
Riviere A, Selak M, Lantin D, et al. Bifidobacteria and butyrate-producing colon bacteria: importance and strategies for their stimulation in the human gut. Front Microbiol 2016;7:979.
trati F, Cavalieri D, Albanese D, De Felice C, Donati C, Hayek J, Jousson O, Leoncini S, Renzi D, Calabrò A, De Filippo C. New evidences on the altered gut microbiota in autism spectrum disorders. Microbiome. 2017 Feb 22;5(1):24.
Verhaar BJH, Hendriksen HMA, de Leeuw FA, Doorduijn AS, van Leeuwenstijn M, Teunissen CE, Barkhof F, Scheltens P, Kraaij R, van Duijn CM, Nieuwdorp M, Muller M, van der Flier WM. Gut Microbiota Composition Is Related to AD Pathology. Front Immunol. 2022 Jan 31;12:794519.
Davidson J, Gauthier-Signore C, Bishop KP, Wicks C, Monteiro MA, Roy PN, Auzanneau FI. ROESY and 13C NMR to distinguish between D- and L-rhamnose in the α-D-Manp-(1 → 4)-β-Rhap-(1 → 3) repeating motif. Org Biomol Chem. 2022 Apr 6;20(14):2964-2980.
Guzior DV, Quinn RA. Review: microbial transformations of human bile acids. Microbiome. 2021 Jun 14;9(1):140.
Raskov H, Burcharth J, Pommergaard HC, Rosenberg J. Irritable bowel syndrome, the microbiota and the gut-brain axis. Gut Microbes. 2016 Sep 2;7(5):365-83.

Supplementary Tables S1-S10 are not available with this version.

Table S1. The baseline information of the 708 qualified samples from the GMrepo database.

Table S2. Classification of the IBS-exclusive genera, health-exclusive genera, and commonly identified genera.

Table S3. Data of Alpha diversity.

Table S4. Differential analysis of microbial compositions between IBS and healthy controls at the phylum level.

Table S5. Differential analysis of microbial compositions between IBS and healthy controls at the family level.

Table S6. Differential analysis of microbial compositions between IBS and healthy controls at the genera level.

Table S7. Differential analysis of microbial compositions between IBS and healthy controls at the species level.

Table S8. The Spearman correlation of gut microbiota and age/BMI in IBS individuals.

Table S9. Significantly-changed genera in each sub-regional group.

Table S10. The top 10 hub species identified in each microbial co-occurrence network.

Download PDF

Journal Publication

published 11 Feb, 2023

Read the published version in Journal of Translational Medicine →

Reviewers agreed at journal
20 Nov, 2022
Reviewers invited by journal
18 Nov, 2022
Editor assigned by journal
28 Oct, 2022
First submitted to journal
26 Oct, 2022

You are reading this latest preprint version

An Integrated Multicenter Amplicon Sequencing Data Reveals New Evidence of the Interaction Between the Gut Microbiota and Irritable Bowel Syndrome

Status:

Journal Publication

Version 1

Abstract

Background

Method

Results

Conclusion

Figures

Introduction

Methods

Study Design and Data Collection

The PSM procedure

Statistical analysis of the baseline characteristics

Data processing, quality control and, taxonomic classification

The alpha and beta diversity

Differential analyses of microbial compositions

Co-occurrence Network

Identification of microbial biomarker for IBS

Results

General characteristics of the study samples

The microbial diversities between IBS and healthy controls

Differential analysis of microbes between IBS and healthy controls

Microbial correlations with environmental factors

Sub-regional differential analysis

Subgroup analysis of microbial compositions in different IBS subtypes

The microbial composition of IBS patients with psychiatric disorders

The co-occurrence network

Identification of microbial biomarker for IBS

Discussion

Conclusion

Declarations

References

Supplementary Tables

Status:

Journal Publication

Version 1