Baseline gut microbial profiles are associated with the efficacy of Bacillus subtilis and Enterococcus faecium in IBS-D

Abstract Objective Little is known about association between the efficacy of probiotics and baseline gut microbiota in irritable bowel syndrome (IBS). We aimed to explore gut microbiota in diarrhea-predominant IBS (IBS-D) and whether baseline gut microbiota was related to the efficacy of Bacillus subtilis and Enterococcus faecium (BE). Methods This study recruited 19 healthy controls (HC) and 50 IBS-D patients, among whom 19 patients were administrated 500 mg BE orally three times daily for 2 weeks. Clinical data and fecal samples were collected from patients before and after treatment. 16S rRNA sequencing was performed to obtain fecal bacterial data. Results There was no significant difference of alpha diversity, beta diversity, profiles of microbial phyla and genera between HC and IBS. BE improved IBS-SSS (IBS severity scoring system) and stool consistency, and altered Enterococcus, Blautia, Lachnoclostridium and Fusobacterium without significant impact on microbial structure in IBS-D. Notably, baseline fecal bacterial composition differed between non-responders and responders to BE concerning abdominal pain and bloating, with Atopobium, Pyramidobacter, Ruminococcus gnavus and Peptostreptococcus enriched in responders in terms of abdominal pain. There was reduced abundance of Prevotella, Ruminococcaceae UCG, Eubacterium eligens, Faecalibacterium and Eubacterium coprostanoligenes in responders compared with non-responders. Furthermore, BE increased beneficial bacteria including Faecalibacterium, Blautia and Butyricicoccus, decreased Lachnoclostridium and Bilophila, and influenced some microbial metabolic pathways in responders, such as mineral absorption, metabolism of arachidonic acid, d-arginine, D-ornithine, phenylalanine and vitamin B6. Conclusion Baseline fecal microbiome is associated with the efficacy of BE in attenuating abdominal pain and bloating in IBS-D.


Introduction
Irritable bowel syndrome (IBS) is one of the most common functional gastrointestinal disorders characterized by recurrent abdominal pain associated with disturbed stool frequency or consistency [1], bothering approximately 7-21% of the world population [2]. Although it is found that the etiology of IBS includes altered gut motility, visceral hypersensitivity, mild inflammation, increased gut permeability and dysfunction of gut-brain axis [3], the underlying mechanism is not completely elucidated.
In recent years, there is growing evidence revealing gut microbiota play a crucial role in the pathophysiology of IBS, such as post-infectious IBS induced by acute enteritis [4], alleviation of IBS symptoms after using antibiotics [5]. IBS-like manifestation in mice fed with IBS-derived fecal microbiota [6] and different gut microbial structure between healthy people and IBS patients [7]. Moreover, improving gut microbial composition or host-microbe interaction is proved to relieve IBS symptoms [8], for example of probiotics which were defined as live microorganisms conferring a health benefit on the host when administered in adequate amounts [9] and reported to have a positive effect on abdominal symptoms in IBS patients [10,11]. Combination of Bacillus subtilis and Enterococcus faecium (BE) is a popular probiotic mixture widely used in enteritis, functional dyspepsia, functional diarrhea, constipation and IBS [12] so that overuse of it inevitably happens, which increases medical expense.
Inspiringly, a previous study demonstrated diarrhea-predominant IBS (IBS-D) patients with a more amenable baseline gut microbiome may contribute to better treatment response to probiotic, indicating gut microbiota can help recognize IBS patients with a higher potential to respond to probiotics [13].
In order to identify IBS-D patients who have a better response to BE based on gut microbiota, we aimed to evaluate the efficacy of BE in relieving IBS symptoms and the influence of BE on gut microbial profiles in IBS-D, and determine whether baseline gut microbial composition was related to treatment response to BE in IBS-D.

Ethics statement
The trial was approved by the Institutional Ethical Review Committee of Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, and registered in Chinese Clinical Trials Registry (clinical trial ID: ChiCTR2000034041).

Study design and participants
This open, single-center, fixed-dose and prospective trial was conducted at the Division of Gastroenterology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, where healthy controls (HC) and IBS-D patients aged 20-60 years old diagnosed according to Rome III criteria [1] were recruited. Participants were excluded if they had coeliac disease, inflammatory bowel disease, history of abdominal surgery, metabolic diseases, infection with pathogen, psychiatric disorder, cardiac, hepatic or renal diseases, or used proton-pump inhibitors, probiotics, antibiotics, IBS prescription medications or colon-cleansing products within a month prior to the study. Written informed consents were obtained from participants. IBS-D patients were administered 500 mg live combined B. subtilis and E. faecium (BE) orally three times daily for 2 weeks, and asked to complete questionnaires which contained IBS Symptom Scoring System [14] (IBS-SSS), IBS Quality of Life [15] (IBS-QOL), frequency of excretion daily, stool consistency according to Bristol Stool Form [16] (BSF), and evaluation of mental state based on Hospital Anxiety and Depression (HAD) scale [17] before and after treatment. IBS-D patients who had IBS-SSS decreased by !50 [18], IBS-QOL increased by !10, score of abdominal pain decreased by !20, score of bloating decreased by !20, stool frequency reduced, stool consistency improved, score of anxiety and depression reduced after treatment were considered as responders to BE in terms of IBS-SSS, IBS-QOL, abdominal pain, bloating, stool frequency, stool consistency, anxiety and depression, respectively.

Sample collection and DNA extraction
Fresh stool samples were obtained from participants before and after treatment, and immediately frozen at À80 C. DNA was extracted from stool samples using FastDNA SPIN Kit (Tiangen, Beijing, China) according to the manufacturer's instruction.
16s rRNA sequencing and analytic methods DNA samples were quantified using Qubit 2.0 Fluorometer (Invitrogen, Carlsbad, CA, USA). DNA (30-50 ng) was used to generate amplicons using a MetaVx TM Library Preparation kit. V3, V4 and V5 hypervariable regions of prokaryotic 16S rDNA were selected for generating amplicons and taxonomy analysis. A panel of proprietary primers were designed aiming at relatively conserved regions bordering the V3, V4 and V5 hypervariable regions of 16S rDNA, which were amplified using forward primers (5 0 -CCTACGGRRBGCASCAGKVRVGAAT-3 0 ) and reverse primers (5 0 -GGACTACNVGGGTWTCTAATCC-3 0 ). First round PCR products were used as templates for second round amplicon enrichment PCR. At the same time, indexed adapters were added to the ends of the 16S rDNA amplicons to generate indexed libraries ready for downstream sequencing on Illumina Miseq. DNA libraries were validated by Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA), quantified by Qubit 2.0 Fluorometer, multiplexed and loaded into Illumina MiSeq to perform sequencing using a 2 Â 300/250 paired-end (PE) configuration. QIIME [19] was used for 16S rDNA data analysis. The forward and reverse reads were joined and assigned to samples based on barcode, and truncated by cutting off the barcode and primer sequence. Quality filtering on joined sequences was performed and sequences which did not fulfill the following criteria were discarded: (i) sequence length >200 bp; (ii) no ambiguous bases; (iii) mean quality score !20. Then chimeric sequences were detected and removed using UCHIME [20] algorithm. Effective sequences were clustered into operational taxonomic units (OTUs) using the clustering program VSEARCH [21] (1.9.6) against the Silva database [22] at 97% sequence identity. The Ribosomal Database Project [23] (RDP) classifier was used to assign OTUs taxonomically at confidence threshold of 0.8. Sequences were rarefied prior to calculation of alpha and beta diversity, which represent diversity within samples and between samples, respectively. Alpha diversity was determined by Ace, Shannon and Simpson index, and beta diversity was evaluated by unweighted and weighted unifrac distance. Principle coordinate analysis (PCoA) plot and Adonis were used to visualize and test unweighted unifrac, respectively.
Significantly different taxa between two groups were identified using Linear discriminant analysis (LDA) effect size [24] (LEfSe). It was considered statistically significant when LDA value >2 and p < .05 in LEfSe. GraPhlAn [25] was used to build phylogenetic trees marked with significantly abundant taxa recognized by LEfSe.

Predicted metagenomics analysis
PICRUSt, [26] a software for predicting functional abundance of microbiota based on 16S rRNA sequence, was used to infer predictive metagenomic content of fecal samples. Specifically, OTU tables were loaded into PICRUSt to generate predictive metagenomes, which were assigned taxonomically according to KEGG [26] (KyotoEncyclopedia of Genes and Genome orthologies) database. Predictive functional profiles of microbiota from PICRUSt were analyzed using STAMP [27].

Statistical analysis
Parametric t test was used to compare clinical characteristics between HC and IBS patients, and paired t test was used to test difference of clinical data between two related groups. Difference of data from 16S rRNA sequence between two related groups was recognized by Wilcoxon test. p Values were adjusted using Benjamini-Hochberg procedure to control false discovery rate (FDR) when performing multiple tests. p Value or adjusted p value <.05 was considered statistically significant.

Data availability
All the sequence data in this study are available in the Genome Sequence Archive (GSA) database (GSA accession number: PRJCA002896).

Results
Clinical characteristics of the study population 19 HC and 50 IBS-D patients were enrolled into the study, and age, gender and BMI were matched between these two groups ( Table 1). IBS patients showed increased score of abdominal pain and bloating, frequency of defecation, stool consistency, anxiety and depression scores compared with HC (p < .001, Table 1). 19 patients received treatment with BE, and only IBS-SSS and stool consistency were improved significantly after treatment (p ¼ .005 and p ¼ .015, respectively; Table 2). Responders and non-responders according to improvement of different clinical indicators were shown in Table 3. BE altered some specific microbial genera without dramatic impact on microbial structure in IBS-D We found Ace and Shannon index were not affected significantly after treatment (p ¼ .495 and p ¼ .124, Figure 2(A) and Supplementary Figure S1A), and the community structure (b-diversity) was not separated between IBS patients before and after treatment (Figure 2(B)). Major phyla were not altered significantly after treatment, including Firmicutes, Bacteroidetes, Proteobacteria, Fusobacteria, Actinobacteria and Verrucomicrobia (Figure 2(C) and Supplementary Figure S1B). Moreover, profiles of predominant genera were comparable between IBS-D before and after treatment (Figure 2(D)), although Enterococcus and Blautia were increased (p ¼ .0017 and p ¼ .071, respectively; Figure  2(E, F)), and Lachnoclostridium and Fusobacterium were reduced after treatment (p ¼ .03 and p ¼ .061, respectively; Figure 2(G, H)).

Fecal microbial composition did not differ between HC and IBS-D
PICRUSt analysis showed that microbial functional community could not be distinguished between IBS-D patients before and after treatment (Supplementary Figure S2). Collectively, BE altered some specific genera without impacting microbial community or functional profiles in IBS-D.  Baseline gut microbial profiles were discriminated between responders and non-responders to BE in terms of abdominal pain and bloating IBS-D patients who had score of abdominal pain decreased by !20 and score of bloating decreased by !20 after treatment were identified as responders to BE in terms of abdominal pain and bloating, respectively. It was found that baseline gut microbial profiles differed significantly between responders and non-responders to BE concerning abdominal pain (p ¼ .039, Figure 3(C)) and bloating (p ¼ .046, Figure  3 Figure 4(C)) and bloating (p ¼ .068, Figure 4(D)), and no significant difference of baseline Ace index was detected between responders and non-responders regarding other indicators (Figure 4(A, B) and (E, H)).
As far as abdominal pain was concerned, LEfSe analysis showed increased abundance of Atopobium, Pyramidobacter, Ruminococcus gnavus and Peptostreptococcus, and reduced abundance of Prevotella, Ruminococcaceae UCG, Eubacterium eligens, Faecalibacterium and Eubacterium coprostanoligenes in responders compared with non-responders ( Figure 5(A, B)), indicating these genera could be used to distinguish Table 3. Clinical characteristics of non-responders and responders in terms of different indicators before and after treatment.  responders from non-responders. Additionally, Alistipes, Eisenbergiella, Ruminiclostridium and Bifidobacterium were reduced in responders compared with non-responders in terms of bloating (Supplementary Figure S3).

Particular microbial genera and metabolic pathways were altered in responders concerning abdominal pain and bloating after treatment
As far as abdominal pain was concerned, Ace index was not influenced significantly in both non-responders and responders after treatment (p ¼ .893 and p ¼ .313, respectively; Supplementary Figure S4A). PCoA plot based on Unweighted Unifrac revealed overlap between responders before and after treatment, as well as between non-responders before and after treatment (Supplementary Figure S4B). Significant changes of Firmicutes to Bacteroidetes ratio (F-B ratio) were not observed in non-responders and responders after treatment (p ¼ .839 and p ¼ .563, respectively; Supplementary Figure S4D). Although profiles of predominant phyla, families and genera did not show great alteration in non-responders and responders after treatment (Supplementary Figure S4C, S4E, S4F, respectively), BE still influenced some important genera which were reported to be associated with gut immunity or diseases. For instance, Enterococcus was increased significantly in both non-responders and responders after treatment (p ¼ .024 and p ¼ .042, respectively; Figure 6(D)). Moreover, Faecalibacterium, Blautia and Butyricicoccus were increased in responders after treatment (p ¼ .043, p ¼ .08 and p ¼ .046, respectively; Figure 6(A-C)), while Faecalibacterium and Blautia were not altered significantly and Butyricicoccus was decreased in non-responders after treatment (p ¼ .695, p ¼ .701 and p ¼ .041, respectively; Figure 6(A-C)). Abundance of Lachnoclostridium and Bilophila was reduced  in responders (p ¼ .075 and p ¼ .042, Figure 6(E, F)) rather than in non-responders (p ¼ .249 and p ¼ .722, Figure 6(E, F)) after treatment.
In terms of bloating, Ace index was not altered significantly in both non-responders and responders after treatment (p ¼ .999 and p ¼ .219, respectively; Supplementary Figure S5A). PCoA plot revealed gut microbial structure was not discriminated between responders before and after treatment, as well as between non-responders before and after treatment (Supplementary Figure S5B). Profiles of predominant phyla, families and genera did not change dramatically in non-responders and responders after treatment (Supplementary Figure S5C, S5E, S5F, respectively). F-B ratio was not influenced significantly in non-responders and responders after treatment (p ¼ .969 and p ¼ .999, respectively; Supplementary Figure S5D). BE altered some important genera in responders. Enterococcus was increased significantly in both non-responders and responders after treatment (p ¼ .018 and p ¼ .045, respectively; Supplementary Figure S6D). Blautia and Bifidobacterium were increased in responders after treatment (p ¼ .046 and p ¼ .058, respectively; Supplementary Figure S6A and S6C), while they were not altered     Bacterial motility proteins, vitamin B6 metabolism and flagellar assembly were decreased significantly in responders after treatment (p ¼ .018, p ¼ .018 and p ¼ .043, respectively; Supplementary Figure S7A-C), whereas these pathways were not influenced in non-responders (Supplementary Figure  S7A-C). In brief, although BE did not have a great impact on gut microbial structure in responders and non-responders concerning abdominal pain and bloating, some specific genera and pathways were modulated by BE in responders rather than non-responders.

Discussion
In the present study, we explored gut microbial profiles in IBS-D and the effect of probiotics containing B. subtilis and E. faecium on symptoms and gut microbial composition in IBS-D. As a result, fecal microbial profiles were not discriminated between HC and IBS-D, and BE improved IBS-SSS as well as stool consistency and altered some particular microbial genera without impacting microbial structure in IBS-D. Notably, as far as abdominal pain and bloating were concerned, baseline gut microbial profiles differed between nonresponders and responders, in whom some particular genera and metabolic pathways were altered rather than in non-responders after treatment. Although a previous study has evaluated the efficacy of BE in non-diarrhea-type IBS patients [12], this is the first study investigating the influence of BE on symptoms and gut microbiota in IBS-D and suggesting response to BE is associated with baseline gut microbiome in IBS-D.
It's reported that live combined B. subtilis and E. faecium could relieve oral and intestinal inflammation in immunosuppressed rats with experimentally induced mucositis [28], and attenuate experimental colitis in mice by downregulation of TLRs, macrophages, Th1, and Th2 but upregulation of Tregs [29]. In our study, BE reduced IBS-SSS and stool consistency in IBS patients in whom low-grade intestinal mucosal inflammation were often detected [30][31][32], which probably resulted from mitigation of mucosal inflammation caused by BE. Additionally, in spite of having no significant impact on microbial structure in IBS-D, BE upregulated Blautia which releases butyrate and relieves intestinal inflammation [33] and downregulated both Lachnoclostridium and Fusobacterium which were found to be increased in patients with colorectal cancer [34,35] and induce colorectal tumorigenesis by activating autophagy signaling [36] and Toll-Like receptor 4 signaling [37].
A previous study reported baseline gut microbiome may be indicative of response to a particular probiotic in IBS [13]. Similarly, we found baseline fecal bacterial profiles were discriminated between non-responders and responders to BE concerning abdominal pain in IBS-D, indicating baseline fecal microbial composition was related to treatment response to BE in terms of abdominal pain in IBS-D. Responders showed increased abundance of Ruminococcus gnavus and Peptostreptococcus and reduced abundance of Prevotella, Eubacterium eligens and Faecalibacterium compared with non-responders. Peptostreptococcus was reported to contribute to colorectal cancer via a PCWBR2-integrin a2/b1-PI3K-Akt-NF-jB signal axis [38] and correlate positively with UC activity [39], and Ruminococcus gnavus could produce an inflammatory polysaccharide to aggravate colitis [40]. Moreover, E. eligens is able to strongly promote the production of IL-10 to inhibit inflammation [41] and Faecalibacterium was decreased in IBS patients [42] and could ameliorate colitis in mice by releasing butyrate to suppress histone deacetylase 1 [43]. Alteration of these microbial genera in gastrointestinal tract may probably shape a dysbiotic gut microenvironment lacking health-promoting metabolites and filled with harmful metabolites in responders to enable BE to exert beneficial impact on abdominal pain. Furthermore, BE increased beneficial genera including Faecalibacterium, Blautia and Butyricicoccus [44] all of which can produce butyrate to enhance gut immunity, and reduced pathogenic genera including Lachnoclostridium and Bilophila [45] in responders rather than non-responders, further elucidating why BE could relieve abdominal pain in responders instead of non-responders. Vitamin B6 metabolism was suppressed in responders after treatment, not in non-responders. Solveig C et al. demonstrated low intake of vitamin B6 was associated with more severity of IBS symptoms [46], and Leonilde et al. showed vitamin B6 improved abdominal symptoms in IBS [47]. In the present study, vitamin B6 metabolism was found to be decreased in responders after treatment, which may lead to more vitamin B6 remained in gastrointestinal tract and thus result in relief of abdominal pain in responders.
There are many limitations in our study. First of all, the study cohort was too small to draw a definite conclusion about BE, and we cannot observe the prolonged impact of BE on abdominal symptoms and gut microbiota in IBS-D due to the short follow-up time. Second, we are supposed to figure out whether and how BE affect mucosal microbial composition in IBS-D in the future since intestinal mucosal dysbiosis was considered to play an important part in pathophysiology in IBS [3]. Third, shotgun metagenomics sequencing is able to provide distribution of microbial species which 16S rRNA sequencing cannot offer, and we need to carry out shotgun metagenomics sequencing to determine the effect of BE on gut microbiota at the species level in IBS-D in the future. Finally, metabolomics should be performed to investigate the influence of BE on metabolism of gut microbiota in IBS-D.
In conclusion, our study provides insights into gut microbial profiles in IBS-D and the impact of B. subtilis and E. faecium on symptoms and gut microbiota in IBS-D. We found fecal microbial composition did not differ between HC and IBS-D, and BE improved IBS-SSS and stool consistency and altered some particular microbial genera without significant influence on microbial structure. Notably, baseline fecal microbiota was associated with treatment response to BE concerning abdominal pain and bloating. In other word, baseline gut microbiota could be used to recognize IBS-D patients who have better treatment response to BE. Some microbial genera and metabolic pathways were improved in responders instead of non-responders after treatment. Although more studies with larger cohorts and longer follow-up time are required to test the robustness of our conclusion, our finding still suggests the feasibility of personalized probiotic therapy based on gut microbiota, which can contribute to reduction of inappropriate use of probiotics.