www.nature.com/npjbioﬁlms BRIEF COMMUNICATION OPEN Identiﬁcation of donor microbe species that colonize and persist long term in the recipient after fecal transplant for recurrent Clostridium difﬁcile 1 2 2 3 4 5 4 4 Ranjit Kumar , Nengjun Yi , Degui Zhi , Peter Eipers , Kelly T. Goldsmith , Paula Dixon , David K. Crossman , Michael R. Crowley , 1,6 5 3 Elliot J. Lefkowitz , J. Martin Rodriguez and Casey D. Morrow Fecal microbiota transplantation has been shown to be an effective treatment for patients with recurrent C. difﬁcile colitis. Although fecal microbiota transplantation helps to re-establish a normal gut function in patients, the extent of the repopulation of the recipient microbial community varies. To further understand this variation, it is important to determine the fate of donor microbes in the patients following fecal microbiota transplantation. We have developed a new method that utilizes the unique single nucleotide variants of gut microbes to accurately identify microbes in paired fecal samples from the same individual taken at different times. Using this method, we identiﬁed transplant donor microbes in seven recipients 3–6 months after fecal microbiota transplantation; in two of these fecal microbiota transplantation, we were able to identify donor microbes that persist in recipients up to 2 years post-fecal microbiota transplantation. Our study provides new insights into the dynamics of the reconstitution of the gastrointestinal microbe community structure following fecal microbiota transplantation. npj Bioﬁlms and Microbiomes (2017) 3:12 ; doi:10.1038/s41522-017-0020-7 Clostridium difﬁcile (C. difﬁcile) is the major causative agent for presence of donor microbes in patients post-FMT. Recently, 1, 2 infective antibiotic associated diarrhea. Infections acquired in Schloissnig et al., used metagenomic sequence analysis of healthcare settings have estimated health care costs in the microbe species in the human microbiome to demonstrate that 3, 4 billions. Standard treatments for C. difﬁcile infection consist of individuals have their own distinct SNV that were stable for up to 1 metronidazole, vancomycin, or ﬁdaxomicin, which result in a 20% year. In this study, we have exploited the unique SNVs of gut rate of recurrence; after a third recurrence, the risk of further microbes of individuals and developed a method to establish episodes is even higher. whether or not paired donor and recipient post-FMT samples When antibiotic therapy fails for the patients, fecal microbiota share the same unique SNVs across the genome. Our ﬁndings transplantation (FMT) has had remarkable success rates of greater demonstrate the colonization and persistence of certain donor than 90% for alleviation of symptoms and restoration of health. microbial species in the recipient post-FMT in C. difﬁcile patients. Numerous studies have characterized the microbial composition We ﬁrst constructed a reference sequence of 93 microbial 7–11 of the recipient following transplant. Although the microbe species, commonly found in healthy and FMT samples (Supple- composition of the recipient post transplant was different after mentary Table 1): 71 most abundant species from the healthy FMT, the similarity of the reconstituted microbe community to microbiome were selected from Schloissnig et al. (accounting for 7–11 that of the donor varied between different patients. The 99% of the aligned HMP data) and the remaining 22 were signiﬁcance of these differences as they relate to the long-term microbes found to be abundant in recipients. To calculate the stability of the FMT re-constructed gut microbe community is SNV, the metagenome is ﬁrst mapped on to the reference unknown. As a ﬁrst step to understanding the microbial ecology of sequences. Raw data obtained either from the NIH Human the reconstituted community, it is necessary to determine the fate Microbiome Project or our study was aligned to the reference of the donor microbes after FMT. A recent study by Li et al. has sequence using Burrows–Wheeler Aligner, and multi-sample SNV shown coexistence of donor microbes in recipient post FMT in calling was performed using Genome Analysis Toolkit (version 12 14, 15 metabolic syndrome patients. Their method uses a comparison 3.6). Then, for a given species, a pairwise comparison was of single nucleotide variants (SNV) of donor, pre-FMT and post performed between two samples to measure their genome-wide FMT samples to show the presence of donor microbes in post FMT SNV similarity. Regions of low read coverage, sequence repeats, samples. However, the pre-FMT samples of patients with C. difﬁcile indels, and structural variants cause alignment difﬁculties that may infection have been exposed to several rounds of antibiotics also produce increased numbers of (false) SNVs. Therefore, to 7–11 resulting in a near depletion of commensal gut microbes. Thus, minimize the effect of these clustered SNVs on overall genome it was not practical to use the method by Li et al. to determine the wide similarity, we developed a window-based SNV comparison 1 2 Biomedical Informatics, Center for Clinical and Translational Sciences, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA; Department of Biostatistics, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA; Department of Cell, Developmental and Integrative Biology, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA; Department of Genetics and Heﬂin Center for Genomic Science, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA; 5 6 Department of Medicine, Division of Infectious Diseases, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA and Department of Microbiology, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA Correspondence: Ranjit Kumar (email@example.com) or Casey D. Morrow (firstname.lastname@example.org) Received: 4 October 2016 Revised: 4 April 2017 Accepted: 15 May 2017 Published in partnership with Nanyang Technological University Coexistence of donor microbes in recipients post FMT R Kumar et al. Fig. 1 Window-based SNV similarity for pair wise sample comparisons. a Schemtatic of window-based SNV similarity (WSS). WSS method used to determine the similarity of two samples (a and b) for two different species (Species 1 and 2). The metagenome DNA sequences were aligned to reference genomes to call SNVs. INDELS were not included in the analysis. To calculate the WSS score, the reference genome is divided into sequential, non-overlapping 1 KB windows. A window is deﬁned as similar if the SNV pattern (denoted as orange bars) is exactly the same between two samples for that region or if no SNV (S in the equation) is present in both samples; otherwise, the window is called dissimilar (denoted as D). The WSS score (as a percentage) is calculated for each pairwise comparison for every microbial species. Both sample pairs are required to have minimum coverage ≥ 20% and average depth ≥ 5 to be included for comparison. Low coverage windows with more than 50% of the bases having a read depth < 5 were ignored. The proportion of similar windows across the genome deﬁnes the genome-wide SNV similarity between two samples for a given species, and is referred to as the WSS score (see Supplementary Data for further explanation). b Scatter plot of the WSS score for all possible pairwise sample comparisons of selected genomes from the HMP data set. WSS scores for all pairwise sample comparisons of species detected in the HMP dataset were determined. The WSS scores (percent similarity) from unrelated HMP sample pairs are presented as blue points. The sample pairs from the same individual at different times (temporally linked) are displayed as red dots. The number of dots for a given species was proportional to the number of samples where the species is present (at depth > 5X). Although 93 species were analyzed, the results from the 21 species that were present in the FMT samples are shown. For each species, we modeled the WSS of related and non-related samples and constructed a simple binary classiﬁer using logistic regression (Supplementary Data). From the classiﬁer, we identiﬁed the WSS cutoff value that differentiates a related sample from a non-related sample (Supplementary Table 7) approach, window-based SNV similarity (WSS), which measures similarity) than non-related samples at the species level. SNV variation between two samples (Fig. 1a). The resolution of Statistically, the related and unrelated pairs show a bimodal WSS is not at the level of single nucleotide variants, but is based distribution with very little overlap (p-values range 0.007–4.7E-27; on the chosen window size. This window-based comparison is less Mann–Whitney U Test). We applied a non-parametric test because sensitive to genomic regions that may have artiﬁcially large even after applying a log transformation, the data still did not pass numbers of SNVs, and therefore provides a more accurate the Shapiro-Wilk test for normal distribution (Supplementary determination of the true variability that exists between species Table 4). Thus, the WSS for a sample pair can be used to predict (i.e., representing potentially strain differences) of two samples whether two different samples are related based on their SNV (see Supplementary Data for further explanation). similarity. For each species, we modeled the WSS of related and We applied the WSS method to metagenomic data from a set of non-related samples and constructed a simple binary classiﬁer 136 metagenomes (51 individuals sampled once, 41 individuals using logistic regression (Supplementary Data). From the classiﬁer, sampled twice, and one individual sampled three times) from the we identiﬁed the WSS cutoff value that differentiates a related NIH Human Microbiome Project (Supplementary Table 2). The raw sample from a non-related sample (Supplementary Table 7). data consisted of 1.3 terabases (13.6 billion raw sequence reads) of We applied the window-based SNV method to samples which 5.5 billion reads (41.5% of raw reads) aligned to the obtained from six donors and seven FMT recipients treated for reference sequence allowing SNVs to be called. The WSS was C. difﬁcile infection with the FMT samples obtained early post-FMT calculated for all possible pairwise comparisons of samples for (1–6 months following transplant) when colonization of the gut each microbial species, of which 21 selected species are shown in with donor microbes should have occurred (one donor was used Fig. 1b and Supplementary Table 3. We found that samples from for two transplants, FMT F and G). Samples at 2 years post-FMT the same individual taken at two separate times (temporally were available for two of the seven transplants and the recipient related samples) have a distinctively higher WSS (i.e., SNV pre-FMT was available for four transplants. Five of the seven npj Bioﬁlms and Microbiomes (2017) 12 Published in partnership with Nanyang Technological University Coexistence of donor microbes in recipients post FMT R Kumar et al. A B Fig. 2 WSS Analysis for FMT donor–recipients. a Scatter plot of the pairwise sample comparisons of species from the FMT. WSS scores were calculated for all possible sample pairs (donor–donor, donor–recipient post-FMT, and recipient post FMT-recipient post FMT). The number of dots for species varies because not all samples contain every species (at depth cutoff > 5X). The blue dots represent FMT-unrelated sample pairs, whereas red and orange points represent FMT-related sample pairs (where both samples were derived from a particular FMT transplant). Orange dots represent related sample pairs with WSS scores below the boundary cutoff, and red dots represent related sample pairs with WSS scores above the boundary cutoff (see Supplementary Data). Grey horizontal bars mark the WSS score boundary cutoff derived from the HMP dataset (see Supplementary Table 7). b The WSS for different FMT donor–recipients post transplant. Donor and recipient post transplant samples from seven different FMT were analyzed (FMT A-F). DA-T1A refers to donor A compared to recipient FMT A at time 1; DA-T2A refers to donor A compared to recipient FMT at time 2. (These 2-year samples were only available for FMT-A and FMT-B). The transplant FMT-FG refers to two FMTs that used the same donor (DF, DG) for two separate recipients (F and G) FMT. The T1A-T2A and T1B-T2B are recipient-recipient comparisons at the two different time points. Red shaded boxes display WSS scores above the WSS boundary cutoff as determined by the classiﬁer, suggesting shared microbial species between the two samples. The orange shaded boxes depict sample pairs where the WSS score was below the boundary cutoff. Empty boxes denote missing data that means sample pairs with sequence coverage or read depth too low to compare transplants were with un-related donors while two of the For all pairwise comparisons of FMT-related samples that were transplants were done with donors from spouses (Supplementary predicted as related, the WSS were above the WSS classiﬁcation Table 5). The DNA from fecal samples was prepared, processed boundary cutoff (all 97–100%; Fig. 2b), implying that the microbial and sequenced, using the Illumina HiSeq2500 platform, with 100 species of the donor and recipient post transplant were related. Bacteroides spp. were identiﬁed as the most common donor base paired-end reads with an average of 38 million reads per microbe found in the recipients post transplant (in all except FMT- sample (Supplementary Table 5). The species abundance was G) which may be explained by the ﬁnding that Bacteroides have estimated using the computational tool MetaPhlan2 that relies on evolved receptors for intestinal cells that facilitate retention in the mapping whole genome sequence read data to a clade speciﬁc gastrointestinal tract. B. ovatus, B. stercoris, B. massiliensis, marker database. The microbe composition of the recipient’s pre B. celluloslyticus, and B. vulgatus were identiﬁed as having been transplant was dominated with microbes such as Escherichia coli, transferred from the donor in multiple transplants while Lactobacillus salivarius, and Klebsiella oxytoca that were not found B. eggerthii and B. uniformis were each only found in one different or were found in very low relative abundance, in the donors or in transplant. Analysis of individual FMT highlighted the complexity the recipients post transplant and were not used for subsequent of the microbial ecology of the gastrointestinal tract environment. pairwise comparisons (Supplementary Table 8). Using the For example, even though B. stercoris was transferred from the MetaPhelan2 data at the species level, we generated a PCoA plot donor to recipient in FMT-E, the B. ovatus detected in the same (Bray-Curtis) that show the relationship of the microbe commu- donor and recipient post-FMT comparison was not similar. In the nities of the donor and pre and post-FMT samples (Supplementary FMT-D transplant, the B. vulgatus of the donor and recipient was Data). not similar; however, in this transplant B. ergerthii, B. sp2116, B. The WSS was calculated for all possible pairwise comparisons of stercoris, and B. celluloslyticus were transferred from donor to FMT samples (all donor–donor, donor–recipient FMT and recipient recipient. Interestingly, in the study where the same donor was FMT-recipient FMT combinations where there was sufﬁcient transplanted into separate recipients (FMT-FG), four Bacteroides genome coverage; Supplementary Table 6). The sample pairs spp. (B. copracola, B. massillienis, B. stercoris and B. vulgatus) in FMT- were grouped in two categories. First is the FMT-related where F were identiﬁed as similar to the donor (DF-T1F) while no there is a possibility of shared species (e.g., the donor–recipient Bacteroides spp. had sufﬁcient sequence coverage for WSS score pairs, DA and T1A in FMTA). The second category is FMT-unrelated calculation in the FMT-G recipient (DG-T1G). This result could not where two samples should not share any common microbial be explained by a mechanical failure of the FMT-G transplant since species (e.g., the unrelated pairs, DB (in FMTB) and DA (in FMTA)). P. merdae, Acidaminococcus spD21, and A. putredinis from the Using the classiﬁer based on full HMP as a training dataset, we donor were identiﬁed in both FMT-F and in FMT-G. classiﬁed the FMT-related sample pairs into related (represent An effective long-term stable transfer would require the same SNV pattern) and unrelated (the sample pair does not share microbes in the FMT to access, occupy, and possibly out- same SNV pattern), implying that sample pair might have different compete the resident recipient microbes for niche space in the microbial species (Supplementary Table 7, Fig. 2a). The classiﬁer gastrointestinal tract following transplantation. Therefore, we correctly predicted that all FMT-unrelated samples (second next examined the long-term persistence of the transplanted category) were unrelated. microbial communities after FMT through analysis of samples 2 Published in partnership with Nanyang Technological University npj Bioﬁlms and Microbiomes (2017) 12 Coexistence of donor microbes in recipients post FMT R Kumar et al. years after transplant. We found identity between the donor Competing interests: The authors declare that they have no competing ﬁnancial interests. and recipient post transplant (both early and 2 year samples) for B. vugatus and B. ovatus in the FMT-A transplant, and Publisher’s note: Springer Nature remains neutral with regard to jurisdictional several Bacteroiodes spp, A. onderkondii, P. merdae, and B. claims in published maps and institutional afﬁliations. intestinihominis for the FMT-B transplant. The demonstration that certain transplanted microbes can persist for up to 2 years Change history: A correction to this article has been published and is linked from demonstrates the potential of using FMT for long-term changes in the HTML version of this article. the composition of the gastrointestinal tract microbe communities. Finally, analysis of early and late samples from FMT-A revealed REFERENCES that the F. prausnitzii that was present at early and 2 years in FMT- 1. He, M. et al. Emergence and global spread of epidemic healthcare-associated A shared identity with each other but did not share identity with Clostridium difﬁcile. Nat. Genet. 45, 109–113, doi:10.1038/ng.2478 (2013). the transplant donor, even though the abundance of F. prausnitzii 2. Jagai, J. & Naumova, E. Clostridium difﬁcile-associated disease in the elderly, in the donor was high (Fig. 2b and Supplementary Table 8). It is United States. Emerg. Infect. Dis. 15, 343–344 (2009). 3. Kyne, L., Hamel, M. B., Polavaram, R. & Kelly, C. P. Health care costs and mortality possible these F. prausnitzii strains came from dormant microbes associated with nosocomial diarrhea due to Clostridium difﬁcile. Clin. Infect. Dis. that had been sequestered in niches of the recipient (and thus not 34, 346–353 (2002). in the fecal samples), possibly as a result of the extensive 19 4. Seekatz, A. M. & Young, V. B. Clostridium difﬁcile and the microbiota. J. Clin. Invest. antibiotics used in the C. difﬁcile recipients. Alternatively, it is also 124, 4182–4189 (2014). possible that the new strain in the recipient was a minority strain 5. Louie, T. J. et al. Fidaxomicin versus vancomycin for Clostridium difﬁcile infection. in the donor sample and thus not detected in the donor, but was N. Engl. J. Med. 364, 422–431, doi:10.1056/NEJMoa0910812 (2011). at a competitive advantage following in the recipient post FMT. 6. Kelly, C. P. Fecal microbiota transplantation—an old therapy comes of age. N. Thus, as suggested by Li et al. where FMT was used in patients Engl. J. Med. 368, 474–475 (2013). with metabolic syndrome, as yet undeﬁned recipient differences, 7. Kumar, R. et al. Colonization potential to reconstitute a microbe community in patients detected early after fecal microbe transplant for recurrent C. difﬁcile. such as niche competition or availability, could impact the BMC. Microbiol. 16, 5 (2016). microbe composition of the transplant recipient. 8. Petrof, E. O. et al. Stool substitute transplant therapy for the eradication of In summary, using a method that exploits the unique SNV Clostridium difﬁcile infection: ‘RePOOPulating’ the gut. Microbiome 1, 3 (2013). patterns of gut microbes to distinguish individual strain SNV 9. Shahinas, D. et al. Toward an understanding of changes in diversity associated patterns, we demonstrate the presence and persistence of certain with fecal microbiome transplantation based on 16S rRNA gene deep sequen- donor microbes in FMT recipients. The clinical success of FMT in cing. MBio 3,1–7 (2012). treating C. difﬁcile has prompted the consideration of FMT to be 10. Shankar, V. et al. Species and genus level resolution analysis of gut microbiota in 12, 20 used for treatment of other gastrointestinal diseases. The Clostridium difﬁcile patients following fecal microbiota transplantation. Micro- results of our analysis provides new insights into the ecology of biome 2, 13 (2014). 11. Song, Y. et al. Microbiota dynamics in patients treated with fecal microbiota trans- the human gastrointestinal tract niches that are essential for plantation for recurrent Clostridium difﬁcile infection. PloS ONE 8, e81330 (2013). development of new approaches to improve health via manipula- 12. Li, S. S. et al. Durable coexistence of donor and ricipient strains after fecal tion of this complex microbial community. microbiota transplantation. Science 352, 586–589 (2016). 13. Schloissnig, S. et al. Genomic variation landscape of the human gut microbiome. Nature 493,45–50 (2013). DATA AVAILABILITY 14. Van der Auwera, G. et al. From FastQ data to high conﬁdence variant calls: the All relevant data are available from the authors upon request. Genome Analysis Toolkit best practices pipeline. Curr. Protoc. Bioinformatics. 11, The HMP metagenomics data is obtained from HMP repository 1–33 (2013). hosted at Amazon (https://aws.amazon.com/datasets/human- 15. Li, H. & Durbin, R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 26, 589–595 (2010). microbiome-project/). All gut samples were downloaded from 16. Segata, N. et al. Metagenomic microbial community proﬁling using unique clade- amazon S3 (s3://human-microbiome-project/HHS/HMASM/WGS/ speciﬁc marker genes. Nat. Methods 9, 811–814, doi:10.1038/nmeth.2066 (2012). stool/). The sequence data for the FMT are deposited in the 17. Lee, S. M. et al. Bacterial colonization factors control speciﬁcity and stability of the sequence read archive under accession number SRP082182. The gut microbiota. Nature 501, 426–429 (2013). code repository used for analysis and processing of metagenomics 18. Dunne, C. Adaptation of bacteria to the intestinal niche: probiotics and gut dis- is available at https://github.com/ranjit58/mgSNP. order. Inﬂamm. Bowel. Dis. 7, 136–145 (2001). 19. Zhang, Y. Persisters, persistent infections and the Yin-Yang model. Emerg. Microbes. Infect 3, e3 (2014). ACKNOWLEDGEMENTS 20. Landy, J. et al. Review article: faecal transplantation therapy for gastrointestinal We thank Dr. Etty N. Benveniste for helpful comments. We thank Margaret Pike, disease. Aliment. Pharmacol. Ther. 34, 409–415 (2011). Chanel Mason, and Mary E. Love for sample acquisition from FMT patients and 21. Lemon, K. P., Armitage, G. C., Relman, D. A. & Fischbach, M. A. Microbiota-targeted Adrienne Ellis for preparation of the manuscript. The following are acknowledged for therapies: an ecological perspective. Sci. Transl. Med, 4, 137rv135 (2012). their support of the Microbiome Resource at the University of Alabama at Birmingham: School of Medicine, Comprehensive Cancer Center (P30 CA013148), Center for AIDS Research (5P30AI027767), Center for Clinical Translational Science Open Access This article is licensed under a Creative Commons (UL1TR001417), and Heﬂin Center for Genomic Sciences. Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative AUTHOR CONTRIBUTIONS Commons license, and indicate if changes were made. The images or other third party R.K., N.Y., D.Z., D.K.C., and E.J.L. did the bioinformatics and statistics; P.E., K.T.G., and M. material in this article are included in the article’s Creative Commons license, unless C. carried out sample preparation and DNA sequencing; P.D. and J.M.R. did the FMT indicated otherwise in a credit line to the material. If material is not included in the and sample collection and R.K., N.Y., D.Z., J.M.R., and C.D.M. conceived the study and article’s Creative Commons license and your intended use is not permitted by statutory wrote the manuscript. All authors read and approved the ﬁnal manuscript. regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/. ADDITIONAL INFORMATION Supplementary Information accompanies the paper on the npj Bioﬁlms and Microbiomes website (doi:10.1038/s41522-017-0020-7). © The Author(s) 2017 npj Bioﬁlms and Microbiomes (2017) 12 Published in partnership with Nanyang Technological University
npj Biofilms and Microbiomes – Springer Journals
Published: Jun 7, 2017
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera