The Microbial Composition of Bacteroidetes Species in Ulcerative

Original Research Article—Basic Science

The Microbial Composition of Bacteroidetes Species in Ulcerative Colitis Is Effectively Improved by Combination Therapy With Fecal Microbiota Transplantation and Antibiotics

Dai Ishikawa, MD, PhD,* Takashi Sasaki, PhD,†,‡ Masahito Takahashi, MD,* Kyoko Kuwahara-Arai, PhD,§

Keiichi Haga, MD, PhD,* Shoko Ito, MD,* Koki Okahara, MD,* Akihito Nakajima, MD, PhD,* Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 Tomoyoshi Shibuya, MD, PhD,* Taro Osada, MD, PhD,¶ Keiichi Hiramatsu, MD, PhD,† Sumio Watanabe, MD, PhD,* and Akihito Nagahara, MD, PhD*

Background: We previously reported that fresh fecal microbiota transplantation (FMT) after triple-antibiotic therapy (amoxicillin, fosfomycin, and metronidazole [AFM]; A-FMT) synergistically contributed to the recovery of phylum Bacteroidetes composition associated with the endoscopic severity and treatment efficacy of ulcerative colitis (UC). Here, we performed further microbial analyses using a higher-resolution method to identify the key bacterial species in UC and determine whether viable Bacteroidetes species from donor feces were successfully colonized by A-FMT. Methods: The taxonomic composition of Bacteroidetes in 25 healthy donors and 27 UC patients at baseline was compared at the species level using a heat-shock protein (hsp) 60–based microbiome method. Microbiota alterations before and after treatment of UC patients were also analyzed in 24 cases (n = 17 A-FMT; n = 3 mono-AFM; n = 4 mono-FMT). Results: We found species-level dysbiosis within the phylum Bacteroidetes in UC samples, which was associated with reduced species diversity, resulting from hyperproliferation and hypoproliferation of particular species. Moreover, in responders treated with A-FMT, diversity was significantly recovered at 4 weeks after a fresh round of FMT, after which high degrees of similarity in Bacteroidetes species composition among recipients and donors were observed. Conclusions: A-FMT alleviated intestinal dysbiosis, which is caused by the loss of Bacteroidetes species diversity in patients with UC. Eradication of dysbiotic indigenous Bacteroidetes species by AFM pretreatment might promote the colonization of viable Bacteroidetes cells, thereby improving the intestinal microbiota dysbiosis induced by UC. Our findings serve as a basis for further investigations into the mechanisms of FMT. Key Words: fecal-microbiota transplantation (FMT), triple-antibiotic therapy using amoxicillin, Fosfomycin, and metronidazole (AFM), FMT after AFM (A-FMT), ulcerative colitis, Bacteroidetes

INTRODUCTION intestinal microbiota of patients with UC exhibits decreased Inflammatory bowel diseases (IBDs), including Crohn’s diversity and richness, resulting in dysbiosis.3, 4 disease and ulcerative colitis (UC), are highly prevalent life- Recently, fecal microbiota transplantation (FMT), a long chronic disorders caused by complex interactions between therapeutic approach used to restore normal intestinal microbiota genetic, immunological, and environmental factors.1, 2 The functions by transplanting bacterial microbiota from feces derived from a healthy donor, has been evaluated in the context of treating patients with UC.5, 6 The therapeutic potential of FMT Received for publications May 7, 2018; Editorial Decision July 18, 2018. for treating various diseases has been extensively investigated; From the *Department of Gastroenterology, Juntendo University School of recent studies have suggested that FMT has great potential for Medicine, Tokyo Japan; †Center of Excellence for Infection Control Science, Graduate 7, 8 School of Medicine, Juntendo University, Tokyo, Japan; ‡Animal Research Center, treating constipation because of its efficacy and safety. The high Sapporo Medical University School of Medicine, Sapporo, Japan; §Department of degree of FMT efficacy for therapy-refractory UC and recurrent Microbiology, Juntendo University School of Medicine, Tokyo, Japan; ¶Department Clostridium difficile infection, another intestinal disease linked to of Gastroenterology, Juntendo University Urayasu Hospital, Chiba, Japan dysbiosis, has been demonstrated by van Nood et al.9 Although Conflicts of interest: The authors have no conflicts of interest to declare. Supported by: This work was supported by JSPS KAKENHI (grant number 2 recent randomized placebo-controlled trials of FMT in UC 10, 11 JP16K09328). patients have been reported, the efficacy of FMT treatment in Address correspondence to: Dai Ishikawa, MD, PhD, Department of UC patients remains controversial. A third randomized placebo- Gastroenterology, Juntendo University School of Medicine, 2-1-1 Hongo, controlled trial, conducted in Australia in 2017,12 involved Bunkyo-ku, Tokyo 113–8421, Japan ([email protected]). multidonor, intensive FMT performed 5 times weekly for 8

© 2018 Crohn’s & Colitis Foundation. Published by Oxford University Press. weeks, resulting in remission or positive responses in UC patients. All rights reserved. For permissions, please e-mail: [email protected]. Although the method was efficacious, this approach would be doi: 10.1093/ibd/izy266 time-consuming and complicated. Therefore, more effective and Published online 14 August 2018 simpler strategies are desirable for validating FMT in UC patients.

Inflamm Bowel Dis • Volume 24, Number 12, December 2018 2590 Inflamm Bowel Dis• Volume 24, Number 12, December 2018 The Composition of Bacteroidetes Is Improved by A-FMT in UC

Previously, we reported that FMT following triple-an- endoscopic, and histologic criteria. Eligible participants were tibiotic therapy with amoxicillin, fosfomycin, and metroni- ≥20 years old and had active UC based on a Lichtiger clinical dazole (AFM) synergistically contributed to the recovery of activity index (CAI) ≥4 or an endoscopic Mayo Clinic score ≥1. Bacteroidetes composition, which was associated with endo- Patients with intestinal superinfections due to cytomegalovirus scopic severity of UC and a high clinical improvement rate.13 (determined by blood testing) were excluded from the study. Clinical responses were observed in 14 of the 17 (82.3%) patients Further exclusion criteria included pregnancy, current serious who completed A-FMT, and 9 of the 14 (53.0%) respond- diseases, and participation in other clinical studies. The clinical ers achieved clinical remission at 4 weeks post-treatment. We features of UC were judged using CAI scores before treatment referred to this combination of FMT and AFM therapy as and 4 weeks after A-FMT. Clinical responses were defined as a

A-FMT, indicating both “antibiotics plus FMT” and “AFM CAI score <10 points and a decrease of ≥3 points. Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 plus FMT.” In our previous study, we observed bacterial alter- Spouses or relatives of patients were selected as donor ations in intestinal microbiota during A-FMT therapy in UC candidates and were asked about their medical history, current patients using 16S rRNA-based microbiome analysis, which diseases, travel history, sexual behavior, recent gastrointestinal has been used to provide insight into correlations between infections, and defecation habits. Donor candidates who had microbial ecology and diseases.14, 15 However, despite its high used antibiotics within 3 months before the study were excluded. taxonomic universality, this method exhibits poor resolving Blood and fecal samples of donors were screened for pathogens, power below the genus level.16–18 Further species-level analyses as described in the Amsterdam protocol for FMT.24, 25 Chronic using a higher-resolution microbiome method are needed to hepatitis B and C, HIV, cytomegalovirus, and syphilis were excluded identify key bacterial species associated with diseases. serologically, and donor stool samples were tested for C. difficile, In this study, we used a microbiome method targeting enterohemorrhagic Escherichia coli, Salmonella, Shigella, Yersinia, heat-shock protein (hsp) 60, utilizing partial sequences widely and Campylobacter and for parasites and helminths. used in phylogenic analysis and species identification using Sanger sequencing because of its higher diversity than that of A-FMT Protocol 16S rRNA19–21; this hsp60-based microbiome method has been A-FMT therapy was performed as previously reported.13 reported in recent studies.22, 23 Here, we conducted a comparative Briefly, patients received a combination antibiotic regimen con- analysis of intestinal Bacteroidetes species composition sisting of oral amoxicillin (1500 mg/d), fosfomycin (3000 mg/d), between healthy individuals and UC patients using a higher- and metronidazole (750 mg/d), followed by fresh FMT. resolution microbiome method. In addition, using a similarity Combination antibiotic therapy was administered to patients analysis method at the species level, we evaluated whether the with UC for 2 weeks until 2 days before fresh FMT. After bowel species composition in recipient intestinal microbiota post-A- lavage, patients underwent a total colonoscopy. A diluted and FMT resembled that in the donors. filtered bacterial suspension derived from donor feces was transferred into the patient’s colon, preferably within 6 hours METHODS of stool sample collection. Ethical Considerations Microbial Analysis The study protocol was reviewed and approved by the ethics committee of the Juntendo Institutional Review Board, Sample collection Juntendo University School of Medicine (Tokyo, Japan), and Fecal samples for microbial analysis were collected from the clinical study committee of Juntendo University Hospital UC patients before and 4 weeks after treatment, and from individ- (approval Nos. 14–017 and 15–059). This trial was also registered ual FMT donors. Once patients completed AFM therapy, fecal at http://www.umin.ac.jp/ctr/index-j.htm:UMIN000014152 and samples were collected before bowel preparation to avoid any http://www.umin.ac.jp/ctr/index-j.htm:UMIN000018642. All microbial bias that may have been introduced by the bowel prepa- participants provided written informed consent. All studies were ration. All fecal specimens were transported to our laboratory performed at Juntendo University Hospital (Tokyo, Japan). within 6 hours of collection, diluted 10-fold in TE buffer (10 mM Tris, 1 mM EDTA [pH 8.0]), and frozen at –80°C until use. Patients and Donors We collected fecal samples from 27 UC patients: 17 UC DNA extraction patients who completed A-FMT, 3 UC patients who completed We used 500 µL of each diluted stool sample for DNA AFM monotherapy (whose microbiome data were analyzed in extraction, as previously reported.13 Following pretreatment with a previous study13), 4 UC patients who completed FMT mono- 50 U of achromopeptidase (Wako Chemical Co., Ltd., Osaka, therapy, and 3 patients who withdrew during treatment. All fecal Japan) in TE buffer at 37°C for 10 minutes, total DNA was isolated specimens were collected between July 2014 and March 2016. using a PowerFecal DNA isolation kit (Mo Bio Laboratories, Ulcerative colitis diagnosis was established by standard clinical, Carlsbad, CA, USA) according to the manufacturer’s protocol.

2591 Ishikawa et al Inflamm Bowel Dis• Volume 24, Number 12, December 2018

Amplification of Polymerase Chain Reaction cgi?PAGE_TYPE=BlastSearch), with respective E-value cutoff Products and Preparation of DNA Libraries for values of e−10 and e−20. The bacterial hsp60 reference sequences Sequencing were obtained from the NCBI database and used to generate a multi-FASTA file for BLAST analysis (Supplementary Table 1 In this study, we used universal hsp60 primers without and File 1). Taxonomic classifications ofhsp60 sequences were adapter sequences. H729 (5′-CGC CAG GGT TTT CCC AGT manually processed, and those exhibiting <95% nucleotide CAC GAC GAI III GCI GGI GAY GGI ACI AC-3′) and H730 identity relative to any Bacteroidetes species were omitted from (5′-AGC GGA TAA CAA TTT CAC ACA GGA YKI TCI CCR species-level identification. AAI CCI GGI GCY TT-3′)19, 26 have been widely used for species identification and taxonomic studies in Bacteroidetes bacteria.21 In these sequences, “I” represents inosine, “K” represents guano- Statistical analysis Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 sine or thymidine, “R” represents adenosine or guanosine, and The diversity and evenness of Bacteroidetes species dis- 28 “Y” represents cytidine or thymidine. We used KOD Multi & Epi tributions were calculated using Simpson’s diversity index (Toyobo Co., Ltd., Osaka, Japan) as the polymerase for microbi- and were compared between healthy donors and patients ome analysis by polymerase chain reaction (PCR) because of its with UC. Similarities between Bacteroidetes species compo- 29 acceptable high fidelity, low amplification bias, and compatibility nents in stool samples were calculated using Morisita’s Cλ 30 with inosine-containing primers (http://www.toyobo-global.com/ and Kimoto’s Cπ values, and a dendrogram was constructed 31 seihin/xr/lifescience/products/pcr_019.html). Polymerase chain from Horn’s R0 values. Statistical differences between pair- reaction was performed in 25-µL reaction mixtures containing wise comparisons were calculated using Mann-Whitney U 70 ng of DNA extract, 0.5 μL of KOD polymerase, 50 pmol of tests. Differences with P < 0.05 were considered statistically each primer, 12.5 μL of the 2× KOD buffer, and molecular-grade significant. water. Reaction conditions were as follows: 1 cycle at 94°C for 2 minutes, followed by 25 to 30 cycles of 98°C for 20 seconds, 50°C RESULTS for 30 seconds, and 68°C for 1 minute, and a final cycle at 68°C To identify key Bacteroidetes species and genera related for 2 minutes. DNA fragments (558 bp) of hsp60 were analyzed to UC, we performed microbiome-wide analysis of partial by electrophoresis in 1× Tris–acetate–EDTA buffer on a 1% agar- sequences of hsp60, recently reported as a target sequence use- ose gel stained with ethidium bromide. Polymerase chain reaction ful for bacterial metagenome analysis.13 From the hsp60 PCR products were purified using AMpure beads (Beckman Coulter, product DNA libraries, we obtained an average of 229,405, Inc., Brea, CA, USA) according to the manufacturer’s protocol. 126,460, and 101,311 valid Bacteroidetes reads per sample from DNA libraries for sequencing were prepared by trans- the 25 healthy donors group, the 27 UC patients at baseline poson-based fragmentation of the hsp60 PCR product using group, and the 24 UC patients at 4 weeks post-treatment group, a Nextera XT DNA sample prep kit (Illumina, San Diego, respectively. A total of 76 samples were analyzed in this study CA, USA). DNA libraries obtained in this manner were highly (Supplementary Table 2). heterogeneous, resulting in the generation of next-generation sequencing (NGS) reads with sufficient quality using the Bacteroidetes hsp60-Based Microbiome Analysis Illumina MiSeq platform. at the Order or Family Level Bacteroidales was the predominant Bacteroidetes order NGS and quality filtering of sequencing reads in the intestinal microbiotas of the 25 healthy donors (aver- Sequencing was performed using the MiSeq reagent kit age of relative abundance ± SD, 0.996 ± 0.008), followed by (version 3; 600-cycle format; Illumina) and a paired-end 2 × 300- Cytophagales (0.003 ± 0.008). In contrast, Bacteroidales bp cycle run on an Illumina MiSeq sequencing system. After abundance in the 27 UC patients was considerably lower sequencing, MiSeq-reads 1 and 2 were stitched using FLASh (0.854); however, it was relatively higher than the abundance (http://ccb.jhu.edu/software/FLASH/),27 resulting in significantly of other Bacteroidetes orders such as Flavobacteriales and improved taxonomic clustering because of the longer reads. Sphingobacteriales. Therefore, we focused on the taxonomic Merged reads were filtered and trimmed by removing bases with composition of species, genera, and families belonging to the quality value scores ≤20 and read lengths <150 bases, followed order Bacteroidales in the intestinal microbiota. by conversion from FASTQ to FASTA format using the FASTX Of the families belonging to this order, Bacteroidaceae toolkit (version 0.0.14; http://hannonlab.cshl.edu/fastx_toolkit). (average relative abundance, 0.791), Porphyromonadaceae (0.118), and Prevotellaceae (0.065) were the most frequently Taxonomic analysis based on bacteroidetes hsp60 identified taxa in the intestinal microbiotas of the 25 healthy sequences donors. Of these 3 families, only the relative abundance of Taxonomic assignments of the hsp60 sequences were per- Prevotellaceae was elevated in the 27 UC patients before ther- formed using BLASTN (https://blast.ncbi.nlm.nih.gov/Blast. apy (0.112).

2592 Inflamm Bowel Dis• Volume 24, Number 12, December 2018 The Composition of Bacteroidetes Is Improved by A-FMT in UC

Moreover, members of the order Sphingobacteriales Odoribacter, Porphyromonas, and Hallella) accounted for exhibited significantly higher abundance in UC patients before ~99.7% of all Bacteroidetes reads. The average relative abun- treatment than healthy donors (Table 1). dance of the top 10 genera decreased considerably from 0.996 in the 25 healthy donors to 0.852 in the 27 UC patients before Bacteroidetes hsp60-Based Microbiome at the therapy. Moreover, the genera Leeuwenhoekiella, Aquimarina, Genus Level and Dyadobacter exhibited significantly higher abundance in At the genus level, Bacteroides was the most frequent UC patients before treatment than in healthy donors (Table 1). taxon in the hsp60-based microbiomes of the 25 healthy donors (average relative abundance, 0.757), followed by Parabacteroides Bacteroidetes hsp60-Based Microbiome at the (0.122) and Prevotella (0.084). The average taxonomic composi- Species Level Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 tion of the overall top 10 genera (Bacteroides, Parabacteroides, Bacteroides vulgatus and Bacteroides uniformis were the Prevotella, Alistipes, Barnesiella, Paraprevotella, Microscilla, predominant species in both the 25 healthy donors (average

TABLE 1. Differences in Phylum Bacteroidetes Composition in the Gut Microbiota Between UC Patients and Healthy Donors

Relative Abundance, Positive % (Standard Deviation) Number, % Taxonomic Description Healthy UC Significance (Taxonomic Level) Healthy UC (n = 25) (n = 27) Level t Test

Sphingobacteriales 0.045 0.83 17 20 0.04 (order) (0.19) (1.9) (68) (74.1) Leeuwenhoekiella 0.000015 0.00059 1 5 < 0.05 (genus) (0.000077) (0.0014) (4) (18.5) Aquimarina 0 0.0049 0 6 0.04 (genus) (0) (0.012) (0) (22.2) Dyadobacter 0.0033 1.58 17 19 < 0.05 (genus) (0.0057) (3.91) (68) (70.1) Bacteroides caecigallinarum 0.016 0.00013 12 5 0.004 (species) (0.026) (0.00037) (48) (18.5) Bacteroides sartorii 0.046 0.0066 17 12 0.02 (species) (0.078) (0.026) (68) (44.4) Bacteroides reticulotermitis 0.0013 0.0001 9 5 0.02 (species) (0.0026) (0.00032) (36) (18.5) Bacteroides faecichinchillae 0.0096 0.00028 12 6 0.03 (species) (0.021) (0.00062) (48) (22.2) Bacteroides cellulosilyticus 0.75 0.0097 19 12 < 0.05 (species) (1.87) (0.038) (76) (44.4) Bacteroides intestinalis 0.19 0.0016 14 8 0.02 (species) (0.40) (0.0036) (56) (29.6) Bacteroides stercorirosoris 0.031 0.0019 17 9 0.03 (species) (0.067) (0.0039) (68) (33.3) Bacteroides zoogleoformans 0.0023 0.000013 8 1 0.03 (species) (0.0054) (0.000068) (32) (3.70) Bacteroides gallinarum 0.016 0.00082 12 9 < 0.05 (species) (0.037) (0.0018) (48) (33.3) Parabacteroides goldsteinii 0.026 0.0016 12 4 0.02 (species) (0.0501) (0.0055) (48) (14.8) Parabacteroides gordonii 0.021 0.0012 13 2 0.01 (species) (0.039) (0.0040) (52) (7.41) Alistipes putredinis 1.26 0.037 18 10 0.01 (species) (2.26) (0.12) (72) (37.0) Alistipes obesi 0.00079 0.000069 9 2 0.02 (species) (0.00149) (0.00028) (36) (7.41)

2593 Ishikawa et al Inflamm Bowel Dis• Volume 24, Number 12, December 2018

relative abundance, 0.141 and 0.139, respectively) and the To confirm the differences in Bacteroidetes species com- 27 UC patients (0.178 and 0.094, respectively), followed by position between healthy donors and UC patients, we compared Parabacteroides distasonis. The average number of species the diversity. As shown in Figure 2A, Simpson’s diversity was in the 27 UC patients before therapy was 27.0, which was significantly lower in UC patients (0.56 ± 0.26) than in healthy lower than that observed in healthy donors (41.0). Moreover, donors (0.78 ± 0.11; P = 0.0003). Additionally, the average 12 species (Bacteroides caecigallinarum, Bacteroides sartorii, number of Bacteroidetes species in treated UC patients was sig- Bacteroides reticulotermitis, Bacteroides faecichinchillae, nificantly lower than that in healthy donors (25.9 ± 11.1 in UC Bacteroides cellulosilyticus, Bacteroides intestinalis, Bacteroides patients vs 41.0 ± 14.3 in healthy donors; P < 0.0001) (Fig. 2B). stercorirosoris, Bacteroides zoogleoformans, Bacteroides

gallinarum, Parabacteroides goldsteinii, Parabacteroides Changes in Microbial Composition at the Genus Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 gordonii, Alistipes putredinis, and Alistipes obesi) exhibited and Species Levels After A-FMT significantly lower abundance in UC patients before treatment After therapeutic intervention, the average relative abun- than in healthy donors (Table 1). dances of the 10 genera were successfully recovered in the 14 The heatmap of the relative abundance of Bacteroidetes responder UC patients after A-FMT (0.999), whereas those in species (Fig. 1) demonstrates highly aggregated distributions the 3 nonresponder UC patients after A-FMT (0.272) and the 3 (red or orange color, >0.5000 relative abundance) of particular UC patients after mono-AFM (0.095) showed further decreases species in 37.0% (10 of 27) of UC patients compared with (Fig. 3). Therefore, the decreased proportions of the top 10 4.0% (1 of 25) of healthy donors. B. vulgatus (n = 2), B. fragilis Bacteroidetes genera (especially the genus Bacteroides) were (n = 2), and Prevotella copri (n = 2) were found in multiple significantly associated with UC patients and antimicrobial patients with UC, whereas Bacteroides dorei, Prevotella buccae, therapy and exhibited no self-recovery without FMT therapy. Prevotella denticola, and P. distasonis were identified only in The average number of Bacteroidetes species increased a single patient. The average number of species in the 27 UC significantly in the 14 responders treated with A-FMT patients (26.9 ± 11.1) was significantly lower than in the 25 (25.8 ± 8.3 at baseline vs 33.6 ± 10.6 post A-FMT; P < 0.05), healthy donors (41.0 ± 14.3; P = 0.0002). whereas those in 3 nonresponders treated with A-FMT (6.3) and

FIGURE 1. Heatmap visualization of Bacteroidetes species distributions (relative abundance) in UC patients at baseline and post-treatment and in individual donors. Log10 scales are visualized using a red–yellow–green gradient, with red representing the highest score (0) and green representing the lowest score (<–6) in terms of the abundance ratios among 106 Bacteroidetes species.

2594 Inflamm Bowel Dis• Volume 24, Number 12, December 2018 The Composition of Bacteroidetes Is Improved by A-FMT in UC Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021

FIGURE 2. A, Microbial diversity (Simpson’s diversity index). B, Phylum Bacteroidetes OTUs in UC patients vs healthy donors. Fecal samples obtained from 27 UC patients were compared with those from 25 healthy donors (diversity: P = 0.0001; evenness: P = 0.0009; OTUs: P < 0.0001). in the 3 UC patients treated with mono-AFM (13.7) decreased. As shown in Figure 5, the species components in 10 of In the 4 UC patients treated with mono-FMT, the number of the 14 clinical responders treated with A-FMT were most or species increased from 31.8 to 40.0; however, this difference was second-most similar to those of their donors, in contrast to the not significant. Furthermore, the average Simpson’s diversity other therapeutic intervention groups (nonresponders, n = 3; index in the 14 responders following A-FMT treatment also FMT monotherapy, n = 4; and AFM monotherapy, n = 3). recovered successfully, becoming indistinguishable from that of These results indicated that FMT therapy after AFM treat- healthy donors (Fig. 4). ment allowed the Bacteroidetes cells existing in fecal specimens Therefore, dysbiosis occurring in the intestinal microbi- to transplant effectively from the donor into the intestinal envi- otas of UC patients resulted from highly aggregated distribu- ronment of the recipient, in parallel to clinical improvement tions of particular species and reduced operational taxonomic by A-FMT. units (OTUs) and diversity, which were successfully reversed by A-FMT. DISCUSSION Two main findings emerged from this study using a high- Similarity in Bacteroidetes Species Components er-resolution method: (1) dysbiosis in intestinal microbiota Between Donors and Recipients resulting from UC is primarily involved in reduced Bacteroidetes To evaluate the influence of bacterial transplantation OTU number and species diversity, resulting in hyperprolif- from donors to patients by A-FMT, we calculated the simi- eration and hypoproliferation of particular species, which do larities between the Bacteroidetes species compositions in 76 not easily self-recover without FMT therapy; and (2) A-FMT samples. combination therapy contributed to the microbiological

FIGURE 3. Bacteroidetes hsp60-based microbiome analysis at the genus level. The relative abundances of Bacteroidetes genera in the indicated patient groups are shown.

2595 Ishikawa et al Inflamm Bowel Dis• Volume 24, Number 12, December 2018

improvement of intestinal dysbiosis in UC patients via successful transplantation of live Bacteroidetes cells from donors. We previously demonstrated that A-FMT induced dramatic changes in the phylum Bacteroidetes, which constitutes a criti- cal factor correlated with clinical responses.13 The main findings of the present study strengthen our previous results. We observed a number of tendencies in the intestinal dysbiosis of microbiotas resulting from UC in this study. Order-level analysis using hsp60-based microbiome interro-

gation suggested that Bacteroidales was a beneficial bacterial Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 taxon; this was supported by our previous report showing that the abundance of the phylum Bacteroidetes in the intestinal microbiota of UC patients correlated with the endoscopic severity of UC.13 However, based on our analyses at the genus and species levels, it is likely that some unfavorable bacterial FIGURE 4. Microbial diversity (Simpson’s diversity index) of phylum taxa exist among the order Bacteroidales. For example, highly Bacteroidetes in responders before and after FMT compared with aggregated distributions of Prevotella species were observed in healthy donors. Fecal samples from 14 responders before and 4 weeks after FMT were compared with those from 14 healthy donors (healthy 4 UC patients (14.8%) in the present study. Overgrowth of the donors vs baseline of responders: P = 0.03; baseline of responders vs genus Prevotella has been shown to lead to osteomyelitis due responders after FMT: P < 0.05). to a reduction in the protective effect against this disease by

FIGURE 5. Similarities in patients at (A) baseline, (B) post-treatment, and (C) between individual donors. Similarities between the Bacteroidetes species composition of 76 samples were calculated by dendrogram analysis of Bacteroidetes species profiles. The most similar and second-most similar relationships are indicated with red or orange lines.

2596 Inflamm Bowel Dis• Volume 24, Number 12, December 2018 The Composition of Bacteroidetes Is Improved by A-FMT in UC

Lactobacillus in a mouse model.32 Scher et al.33 reported that Bacteroidales adhere to the mucosal surface, which might enhanced susceptibility to arthritis was associated with an be associated with their ability to utilize mucin as a nutrient expansion of intestinal P. copri, which was also highly distrib- source.46, 47 In this study, we analyzed microbial composition uted in multiple UC patients in the present study. Therefore, as only in fecal samples, which may not reflect the mucosa-asso- previously reported, hyperproliferation of the genus Prevotella ciated microbiota. Hence, it may be more relevant to analyze in intestinal microbiota might be closely related to various the mucosa-associated microbiota (ie, from bioptic samples) to chronic inflammatory diseases including UC. identify bacterial species potentially involved in A-FMT effi- “Colonization resistance” is associated with stable and cacy. In our previous study, all nonresponders exhibited exten- diverse microbiota, together with controlled lack of inflam- sive UC and had at least 9 points on the endoscopic sum score mation. This involves specific interactions between the muco- that were refractory to A-FMT both in terms of the clinical set- Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 sal immune system and the microbiota.34 Healthy microbiome ting and improvement in the gut microbiota. This suggests that and mycobiome can effectively inhibit colonization and over- Bacteroidetes genera might be unable to adhere to the niche growth by invasive pathogenic species such as C. difficile and of extensively damaged mucosa without sufficient mucin pro- Candida albicans.35–37 This native protection may also prevent duction. Therefore, supplementation of insufficient mucin pro- effective intestinal colonization of the transplanted micro- duction in UC might constitute an effective prebiotic method biota derived from donors by FMT. Intestinal Bacteroides for improving A-FMT. These species could be also affected by members have developed the commensal colonization factor specific dietary features, as previously reported.48–50 Elucidation (CCF) system, which is involved in microbial homeostasis of the nutrients and pathways that control UC-related of the gut microbiota.38 Moreover, intestinal bacteria exhibit Bacteroidetes species could aid the development of prebiotics antagonistic abilities against exogenous bacteria to maintain for controlling UC pathogenesis. the bacterial composition of the intestinal microbiota.39 In the This study is limited by a lack of long-term analysis. present study, eradication of indigenous bacterial microbiota Our data suggest that studying the diversity and similarity of by AFM pretreatment might have promoted the effective colo- Bacteroidetes species is useful for evaluating the clinical state nization of Bacteroidetes cells supplied by FMT, as supported of UC patients and the therapeutic effect of A-FMT. Further by the high similarity between the bacterial composition of follow-up studies are required to evaluate (1) the composi- UC patients and their donors. Thus, it is likely that tempo- tion of Bacteroidetes members at the species level, (2) whether ral suppression of the indigenous bacteria with antagonistic they correlate with recurrence and remission maintenance, abilities by antibiotic pretreatment contributes to successful and (3) the long-term efficacy of our A-FMT protocol. Our intestinal colonization by A-FMT. Specific Bacteroidetes spe- Bacteroidetes taxa analysis methods could provide a reasonable cies correlate with clinical activity in IBD and the control of biomarker for predicting recurrence and therapeutic effects in inflammation associated with zwitterionic capsular polysac- UC treatment. charides, which are bacterial products that modulate T cells, In conclusion, we demonstrate that intestinal dysbiosis in including induction of anti-inflammatory interleukin-10-se- UC patients was involved in the loss of Bacteroidetes species creting T regulatory cells.40, 41 In the present study, distinct diversity, resulting in hyperproliferation and hypoproliferation Bacteroidetes species components resulting from UC were of particular species. A-FMT can effectively improve the dysbi- successfully improved by microbial colonization from healthy otic gut microbiota via temporal suppression of the indigenous donor samples containing significantly high proportions of bacteria with homeostatic maintenance of intestinal microbiota particular species (ie, key bacteria). If the recovery of diver- function. To the best of our knowledge, this is the first report sity among viable Bacteroidetes species through the introduc- showing that intestinal dysbiosis related to UC was improved tion of rich key bacteria has therapeutic potential, then fresh by FMT based on microbial analyses at the Bacteroidetes spe- FMT could constitute an amenable combination for AFM cies level. The strategy presented in this study might serve as therapy as a treatment methodology for transplanting viable the basis for further investigations of the mechanisms under- Bacteroidetes cells (not freeze-tolerant bacteria). Therefore, lying alterations in the intestinal microbiota following therapy. the use of frozen and lyophilized FMT42–45 needs to be consid- Moreover, this protocol might serve as a useful strategy for ered; we are currently evaluating frozen FMT methods in our managing UC patients. ongoing basic and clinical studies. In this study, we recruited relatives of the patients as donors because the aim was to SUPPLEMENTARY DATA improve the safety of the FMT process and to reduce the psy- Supplementary data are available at Inflammatory Bowel chological stress of the patients; however, this strategy using Diseases online. feces from relatives may complicate the evaluation of donor effects. Currently, we are examining selected nonrelative vol- unteer donors to elucidate the crucial factors that determine REFERENCES 1. Cosnes J, Gower-Rousseau C, Seksik P, et al. Epidemiology and natural history of matching for individual recipients and FMT efficacy. inflammatory bowel diseases.Gastroenterology . 2011;140:1785–1794.

2597 Ishikawa et al Inflamm Bowel Dis• Volume 24, Number 12, December 2018

2. Liu JZ, van Sommeren S, Huang H, et al; International Multiple 24. Rossen NG, MacDonald JK, de Vries EM, et al. Fecal microbiota transplantation Sclerosis Genetics Consortium; International IBD Genetics Consortium. as novel therapy in gastroenterology: a systematic review. World J Gastroenterol. Association analyses identify 38 susceptibility loci for inflammatory bowel 2015;21:5359–5371. disease and highlight shared genetic risk across populations. Nat Genet. 25. Owens C, Broussard E, Surawicz C. Fecal microbiota transplantation and donor 2015;47:979–986. standardization. Trends Microbiol. 2013;21:443–445. 3. Kostic AD, Xavier RJ, Gevers D. The microbiome in inflammatory bowel disease: 26. Brousseau R, Hill JE, Préfontaine G, et al. Streptococcus suis serotypes char- current status and the future ahead. Gastroenterology. 2014;146:1489–1499. acterized by analysis of chaperonin 60 gene sequences. Appl Environ Microbiol. 4. Shen ZH, Zhu CX, Quan YS, et al. Relationship between intestinal microbiota 2001;67:4828–4833. and ulcerative colitis: mechanisms and clinical application of probiotics and fecal 27. Magoč T, Salzberg SL. FLASH: fast length adjustment of short reads to improve microbiota transplantation. World J Gastroenterol. 2018;24:5–14. genome assemblies. Bioinformatics. 2011;27:2957–2963. 5. Smits LP, Bouter KE, de Vos WM, et al. Therapeutic potential of fecal microbi- 28. Simpson EH. Measurement of diversity. Nature. 1949;163:688. ota transplantation. Gastroenterology. 2013;145:946–953. 29. Morisita M. Measuring of interspecific association and similarity between com- 6. Narula N, Kassam Z, Yuan Y, et al. Systematic review and meta-analysis: fecal munities. Mem Fac Sci Kyushu Univ Series E. 1959;3:65–80. microbiota transplantation for treatment of active ulcerative colitis. Inflamm 30. Kimoto S. Some quantitative analysis on the Chrysomelid fauna of the Ryukyu Bowel Dis. 2017;23:1702–1709. Archipelago. Esakia. 1967;6:27–54. Downloaded from https://academic.oup.com/ibdjournal/article/24/12/2590/5074053 by guest on 29 September 2021 7. Brandt LJ, Aroniadis OC. An overview of fecal microbiota transplantation: tech- 31. Horn HS. Measurement of overlap in comparative ecological studies. Am Nat. niques, indications, and outcomes. Gastrointest Endosc. 2013;78:240–249. 1966;100:419–424. 8. Johnsen PH, Hilpüsch F, Cavanagh JP, et al. Faecal microbiota transplantation 32. Lukens JR, Gurung P, Vogel P, et al. Dietary modulation of the microbiome versus placebo for moderate-to-severe irritable bowel syndrome: a double-blind, affects autoinflammatory disease.Nature . 2014;516:246–249. randomised, placebo-controlled, parallel-group, single-centre trial. Lancet 33. Scher JU, Sczesnak A, Longman RS, et al. Expansion of intestinal correlates with Gastroenterol Hepatol. 2018;3:17–24. enhanced susceptibility to arthritis. Elife. 2013;2:e01202. 9. van Nood E, Vrieze A, Nieuwdorp M, et al. Duodenal infusion of donor feces for 34. Lawley TD, Walker AW. Intestinal colonization resistance. Immunology. recurrent Clostridium difficile. N Engl J Med. 2013;368:407–415. 2013;138:1–11. 10. Rossen NG, Fuentes S, van der Spek MJ, et al. Findings from a randomized 35. Schubert AM, Sinani H, Schloss PD. Antibiotic-induced alterations of the controlled trial of fecal transplantation for patients with ulcerative colitis. murine gut microbiota and subsequent effects on colonization resistance against Gastroenterology. 2015;149:110–118.e4. Clostridium difficile. Mbio. 2015;6:e00974. 11. Moayyedi P, Surette MG, Kim PT, et al. Fecal microbiota transplantation induces 36. Markey L, Shaban L, Green ER, et al. Pre-colonization with the commensal fun- remission in patients with active ulcerative colitis in a randomized controlled trial. gus Candida albicans reduces murine susceptibility to Clostridium difficile infec- Gastroenterology. 2015;149:102–109.e6. tion. Gut Microbes. 2018; 18:1–13. 12. Paramsothy S, Kamm MA, Kaakoush NO, et al. Multidonor intensive faecal 37. Sam QH, Chang MW, Chai LY. The fungal mycobiome and its interaction with microbiota transplantation for active ulcerative colitis: a randomised place- gut bacteria in the host. Int J Mol Sci. In press. bo-controlled trial. Lancet. 2017;389:1218–1228. 38. Lee SM, Donaldson GP, Mikulski Z, et al. Bacterial colonization factors control 13. Ishikawa D, Sasaki T, Osada T, et al. Changes in intestinal microbiota follow- specificity and stability of the gut microbiota.Nature . 2013;501:426–429. ing combination therapy with fecal microbial transplantation and antibiotics for 39. Crost EH, Pujol A, Ladiré M, et al. Production of an antibacterial substance in ulcerative colitis. Inflamm Bowel Dis. 2017;23:116–125. the digestive tract involved in colonization-resistance against Clostridium perfrin- 14. Mayer EA, Tillisch K, Gupta A. Gut/brain axis and the microbiota. J Clin Invest. gens. Anaerobe. 2010;16:597–603. 2015;125:926–938. 40. Neff CP, Rhodes ME, Arnolds KL, et al. Diverse intestinal bacteria contain puta- 15. Karlsson FH, Fåk F, Nookaew I, et al. Symptomatic atherosclerosis is associated tive zwitterionic capsular polysaccharides with anti-inflammatory properties.Cell with an altered gut metagenome. Nat Commun. 2012;3:1245. Host Microbe. 2016;20:535–547. 16. Claesson MJ, O’Sullivan O, Wang Q, et al. Comparative analysis of pyrosequenc- 41. Darnaud M, Dos Santos A, Gonzalez P, et al. Enteric delivery of regenerating ing and a phylogenetic microarray for exploring microbial community structures family member 3 alpha alters the intestinal microbiota and controls inflammation in the human distal intestine. PLoS One. 2009;4:e6669. in mice with colitis. Gastroenterology. 2018;154:1009–1023.e14. 17. Razzauti M, Galan M, Bernard M, et al. A comparison between transcriptome 42. Lee CH, Steiner T, Petrof EO, et al. Frozen vs fresh fecal microbiota transplan- sequencing and 16S metagenomics for detection of bacterial pathogens in wild- tation and clinical resolution of diarrhea in patients with recurrent Clostridium life. Plos Negl Trop Dis. 2015;9:e0003929. difficile infection: a randomized clinical trial. JAMA. 2016;315:142–149. 18. Kump PK, Gröchenig HP, Lackner S, et al. Alteration of intestinal dysbiosis 43. Hamilton MJ, Weingarden AR, Sadowsky MJ, et al. Standardized frozen prepa- by fecal microbiota transplantation does not induce remission in patients with ration for transplantation of fecal microbiota for recurrent Clostridium difficile chronic active ulcerative colitis. Inflamm Bowel Dis. 2013;19:2155–2165. infection. Am J Gastroenterol. 2012;107:761–767. 19. Goh SH, Potter S, Wood JO, et al. HSP60 gene sequences as universal targets for 44. Jiang ZD, Alexander A, Ke S, et al. Stability and efficacy of frozen and lyophilized microbial species identification: studies with coagulase-negative staphylococci.J fecal microbiota transplant (FMT) product in a mouse model of Clostridium dif- Clin Microbiol. 1996;34:818–823. ficile infection (CDI). Anaerobe. 2017;48:110–114. 20. Ghebremedhin B, Layer F, König W, et al. Genetic classification and dis- 45. Staley C, Hamilton MJ, Vaughn BP, et al. Successful resolution of recurrent tinguishing of Staphylococcus species based on different partial gap, Clostridium difficile infection using freeze-dried, encapsulated fecal microbiota; 16S rRNA, hsp60, rpoB, sodA, and tuf gene sequences. J Clin Microbiol. pragmatic cohort study. Am J Gastroenterol. 2017;112:940–947. 2008;46:1019–1025. 46. McGuckin MA, Lindén SK, Sutton P, et al. Mucin dynamics and enteric patho- 21. Sakamoto M, Ohkuma M. Usefulness of the hsp60 gene for the identifica- gens. Nat Rev Microbiol. 2011;9:265–278. tion and classification of gram-negative anaerobic rods.J Med Microbiol. 47. Sheng YH, Hasnain SZ, Florin TH, et al. Mucins in inflammatory bowel diseases 2010;59:1293–1302. and colorectal cancer. J Gastroenterol Hepatol. 2012;27:28–38. 22. Uyaguari-Diaz MI, Chan M, Chaban BL, et al. A comprehensive method for 48. Zimmer J, Lange B, Frick JS, et al. A vegan or vegetarian diet substantially alters amplicon-based and metagenomic characterization of viruses, bacteria, and the human colonic faecal microbiota. Eur J Clin Nutr. 2012;66:53–60. eukaryotes in freshwater samples. Microbiome. 2016;4:20. 49. Wu GD, Chen J, Hoffmann C, et al. Linking long-term dietary patterns with gut 23. Links MG, Dumonceaux TJ, Hemmingsen SM, et al. The chaperonin-60 univer- microbial enterotypes. Science. 2011;334:105–108. sal target is a barcode for bacteria that enables de novo assembly of metagenomic 50. Schirmer M, Franzosa EA, Lloyd-Price J, et al. Dynamics of metatranscription in sequence data. PLoS One. 2012;7:e49755. the inflammatory bowel disease gut microbiome.Nat Microbiol. 2018;3:337–346.

2598