2019 Poster Session

The HCMPH Symposium poster session will take place on Wednesday, May 15, 2019, 5:00 – 6:30 pm. If you have any questions, please contact Nicole Levesque.

1. 1. Emma Accorsi
      Determinants of S. aureus carriage in the developing infant nasal microbiome
      Emma Accorsi¹, Eric A. Franzosa^1,2, Tiffany Hsu^1,2, Regina J. Cordy³, Ayala Maayan-Metzger^4,5, Hanaa Jaber⁵, Aylana Reiss-Mandel⁵, Casey DuLong¹, Marc Lipsitch¹, Gili Regev-Yochay^4,5, Curtis Huttenhower^{1,2
      1}Harvard T. H. Chan School of Public Health; ²Broad Institute; ³Wake Forest University; ⁴Sackler School of Medicine; ⁵Sheba Medical Center
      Staphylococcus aureus is a leading cause of healthcare-associated infections in infants and adults, and is increasingly difficult to treat due to methicillin resistance. About 30% of individuals carry this species asymptomatically in their nares, where factors including interactions with other species in the nasal microbiome modulate S. aureus acquisition and carriage. It is thus important to identify which ecological or functional genetic elements within the maternal or infant nasal microbiome influence S. aureus acquisition and retention in early life. We recruited 36 mother-infant pairs and profiled a subset of monthly longitudinal nasal samples from the first year after birth (total n=284) using shotgun metagenomic sequencing. We found the infant nasal microbiome to be highly variable over the first year of life, particularly within the first 1-2 months. It was weakly influenced by maternal nasal microbiome composition, but primarily shaped by developmental and external factors (e.g. environmental exposures at daycare). Infants displayed distinctive patterns of S. aureus carriage, and S. aureus acquisition was positively associated with the presence of Acinetobacter species, Streptococcus parasanguinis, Streptococcus salivarius, and Veillonella species in the infant microbiome. In gene-content based strain profiling, infants carried strains of S. aureus that were more similar to maternal strains, and S. aureus strains differed in their propensities to be lost by infants. The improved understanding of S. aureus colonization provided by these results is thus an important first step toward development of urgently needed novel, ecological therapies for controlling S. aureus carriage.

1. 1. Kalpana Acharya
      Distinct Changes in Gut Microbiota Are Associated with Estradiol-Mediated Protection from Diet-Induced Obesity in Female Mice
      Acharya, K.D.¹, Graham, M.E. ¹, Noh, H.L.², Suk, S.², Friedline, R.H.², Chen, J.⁴, Kim, J.K.^2,3, Tetel, M.J.^{1
      1}Neuroscience Program, Wellesley College, Wellesley, MA, 02481, ²Program in Molecular Medicine, and ³Division of Endocrinology, Metabolism, and Diabetes, Department of Medicine, University of Massachusetts Medical School, Worcester MA 01605, and ⁴Division of Biomedical Statistics & Informatics & Center for Individualized Medicine, Mayo Clinic, Rochester, MN 55905
      Due to an aging-associated decrease in ovarian estrogens, postmenopausal women tend to gain weight, which predisposes them to developing cardiovascular disease, type 2 diabetes, chronic inflammation, and cancer. Estradiol replacement has been shown to protect postmenopausal women and ovariectomized rodents from obesity and metabolic disorders, but the underlying mechanisms are not fully understood. In addition to estrogens, gut microbiota are also known to regulate energy homeostasis. In this study, we investigated the effects of estrogens on gut microbiota and energy metabolism in female mice on standard diet (SD) or high-fat diet (HFD). Ovariectomized adult C57BL6/J mice received implants containing estradiol (E2) or vehicle (Veh) (n=6/group). Mice were fed SD for the first two weeks and then switched to HFD for the remaining four weeks of the study. To investigate the effects of E2 and diet on gut microbiota, fresh fecal samples were collected at various times during SD and HFD for 16S rRNA gene sequencing. Using metabolic cages, the effects of E2 and HFD on energy balance were assessed longitudinally by measuring food intake, energy expenditure, and physical activity. In addition, plasma glucose, insulin, leptin, resistin and cytokines were measured during SD and HFD. Insulin sensitivity and glucose metabolism were assessed during a 2-hr hyperinsulinemic-euglycemic clamp in awake mice at the end of HFD. Our results indicate that chronic E2 treatment protected the ovariectomized mice from HFD-induced obesity, mostly due to increases in energy expenditure and physical activity. These effects were associated with altered bacterial communities; E2 increased the relative abundance of the mucin-producing microbes Akkermansia spp. (phylum Verrucomicrobia) that are responsible for the integrity of the gut epithelial barrier. Veh mice had increased the relative abundances of Erysipleotrichaceae and Streptococcaceae (family) and their lower taxa. HFD profoundly altered microbial diversity by reducing species richness, increasing evenness and altering relative abundances of multiple taxa. HFD was associated with increased Bacteriodes (phylum Bacteriodetes) and Clostridia (phylum Firmicutes), which are responsible for HFD-induced weight gain in humans. The relative abundance of Erysipleotrichi was positively correlated with HFD, but negatively correlated with E2 treatment, suggesting a potential causal relationship between this taxon and obesity. E2-treated mice were more insulin-sensitive with marked increases in whole body glucose turnover and glycogen synthesis after HFD. Taken together, these results suggest that changes in gut microbiota contribute to E2-mediated protection against diet-induced obesity and metabolic dysregulation. These findings may provide important insights for potential microbial targets for the treatment of metabolic disorders in women.
      Keywords: microbiome, estrogen, glucose metabolism

1. 1. Emrah Altindis
      The Role of Microbial Insulin Mimics in Type 1 Diabetes Autoimmunity
      Type 1 diabetes (T1D) is an autoimmune disease characterized by selective destruction of pancreatic β-cells, which is increasing in incidence around the world. Genome-wide association studies have identified ~50 genetic regions which can affect the risk of developing T1D2, but genetics alone cannot explain the increasing incidence of T1D. Various environmental factors have been considered, but the roles of these factors remain unclear. An early marker of T1D autoimmunity is the development insulin autoantibodies (IAA), the only autoantibodies specific to β-cells. In humans, IAA are usually the first to be detected and they develop months to years before the onset of overt diabetes. Moreover, there is a significant correlation between IAA concentrations and rate of progression to T1D.
      Molecular mimicry is a potential cause of autoimmunity, when the immune response to a foreign antigen also targets a similar epitope on a host protein. This mechanism is based on the high degeneracy of T cell recognition and can be either pathogenic or protective. In this study, we hypothesized that T1D autoimmunity is caused by molecular mimicry. Specifically, an immune response to microbes expressing an insulin-like peptide and/or insulin epitope stimulates the immune system to respond to native insulin. Based on this hypothesis, we previously discovered for the very first time that viruses mimic our hormones and characterized the viral insulin/IGF-1 like peptides (VILPs).
      Using a bioinformatics approach, we also identified over 100 potential insulin B:9-23 epitope mimics in different microbial proteins and have investigated the top ranking 17 of the these for their capacity to stimulate insulin-specific T cells obtained from T1D patients and NOD mice. One peptide identified in a commensal microbe in the human gut microbiome, Parabacteroides distasonis, strongly stimulated both human and murine insulin specific T-cells. Conversely, lymphocytes from mice immunized with this P. distasonis mimic showed increased secretion of cytokines when stimulated with the insulin B:9-23 peptide confirming bi-drectional cross-reactivity. Analysis of T1D metagenomic gut microbiome data (DIABIMMUNE) suggests that this bacterial epitope mimics is identified only in healthy children at the first year of age, and never in seroconverting islet antibody-positive children at risk for T1D. Moreover, this peptide was not identified in the gut microbiome of T1D progressors in Finland which has the highest T1D incidence rates, suggesting a potential protective role. We hypothesize that such cross-reactive insulin epitope mimics may become accessible to intestinal phagocytes for their presentation to T cells thus modulating T1D in genetically susceptible individuals.
      Authors: EMRAH ALTINDIS, ANTHONY N. VOMUND, I-TING CHOW, QIAN HUANG, WILLIAM KWOK, EMIL R. UNANUE and C. RONALD KAHN

1. 1. Kjetil Bjornevik
      The gut microbiota in people with amyotrophic lateral sclerosis
      Katharine Nicholson, Kjetil Bjornevik, Galeb Abu-Ali, James Chan, Brixhilda Dedi, Boyu Ren, Ramnik Xavier, Curtis Huttenhower, Alberto Ascherio, James D. Berry.
      Objective: To assess whether amyotrophic laterals sclerosis (ALS) is associated with alterations in the gut microbiota.
      Methods: We conducted a case-control study at the Massachusetts General Hospital multidisciplinary ALS clinic to compare the gut microbiota in people with ALS to that in controls. Metagenomic shotgun sequencing was performed on DNA extracted from stool samples of 68 participants with ALS, 61 healthy controls (HC), and 12 neurodegenerative controls (NDC). Taxonomic metagenomic profiles were analyzed for shifts in the microbial community structure between the comparator groups using per-feature univariate and multivariate association tests.
      Results: The relative abundance of the dominant butyrate-producing bacteria Eubacterium rectale and Roseburia intestinalis was significantly lower in ALS patients compared to HC. Adjustment for age, sex, and constipation did not materially change the results. The total abundance of ten dominant species capable of producing butyrate was also markedly lower in ALS compared to HC (p = 0.001).
      Conclusion: The levels of several butyrate-producing bacteria, which are important for gut integrity and regulation of inflammation, were lower in ALS patients compared to controls. These findings lend support to the inference that the gut microbiota could be a risk factor for ALS and warrant further investigations, preferably with a longitudinal design.

1. 1. Shrish Budree
      Evaluating the Gut Microbiome in Children with Stunting: Findings from a South African Birth Cohort
      Shrish Budree, Majdi Osman, Mamadou Kaba, Shantelle Claasens, Tommi Vatanen, Mark Nicol, Ramnik Xavier, Heather Zar
      Background: Stunting contributes a significant burden to child health globally. A large proportion of children in low- and middle-income countries develop stunting despite preventative nutritional and environmental interventions. Recent evidence suggests that the intestinal microbiome contributes to the development and persistence of stunting, however, few studies have characterized the microbiome in stunted African children. This study aims to describe the microbiome of African children with stunting compared to those with normal linear growth.
      Methods: Children enrolled into the Drakenstein Child Health Study (DCHS), a birth cohort study, had stool collected at regular study visits during the first year of life. The DCHS was situated in a low income, peri-urban setting in South Africa. Anthropometry was measure during routine study visits by trained study staff according to standard procedures. Stool samples underwent standard 16S rRNA sequencing and analysis was conducted using QIIME2 and the R statistical package, Phyloseq. Alpha (Shannon index) and beta diversity (Unifrac) between the stunted and control groups were compared. A linear mixed effects model was used for statistical comparison of alpha diversity and PEMANOVA for beta diversity. Significant differences in microbial relative abundance between the 2 groups were assessed using MaAsLiN, accounting for the longitudinal nature of stool samples and potential confounders.
      Results: We analyzed a subset of 138 children with complete clinical and microbial data. The majority of participants had approximately 3 stool samples collected at 3 different time points. The mean age of participants was 8.7 months (SD =6.77) with 45% being female. Stunting (height-for-age z score [HAZ] < -2) was found in 34 (25%) children of which 19 were severe stunting (HAZ < -3). A total of 401 stool samples were analyzed. There were significant differences in alpha (p=0.01) and beta diversity (p=0.01) between stunted and normal height children, after adjusting for method of stool collection and age. At phylum level, children with severe stunting appear to be depleted in Bacteroidetes. On MaAsLiN analysis, the genus Ruminococcus was significantly reduced in children with severe stunting (q=0.03) and more specifically Ruminococcus gnavus on species level analysis (q=0.13).
      Conclusion: These findings demonstrate a microbial signature associated with stunting in African children.

1. 1. Demetrius DiMucci
      Machine Learning Reveals Edges and Putative Mechanisms in Microbial Ecosystem Networks
      Pairwise co-culture experiments of all microbial species in a community are emerging as powerful tools for understanding the interactions that determine function and dynamics in microbial communities. These studies are performed with the goals of predicting the outcomes of new unobserved interactions and developing plausible explanations regarding mechanism but are often limited in scope by the experimental logistics of handling large communities. We have combined a trait-based representation of microbial interactions with machine learning to develop a model for predicting yet to be seen interactions and the guided generation of mechanistic hypotheses.
      All pairwise combinations between 100 genome-scale metabolic models of human gut associated bacteria [1] were performed in silico and the relative performance of each model in co-culture with each other model was recorded. Models were negatively affected by their partner 5563 times and unaffected by the presence of an interaction partner 4337 times. Interactions between microbial models were represented as a binary vector where each element corresponds to one of 388 unique metabolic reactions; vectors were subsequently used as predictors for machine learning.
      Evaluation on the full experimental space with five fold cross validation yielded a balanced accuracy of ~90% when predicting the qualitative outcome of a held out interaction. Furthermore, a learning curve analysis revealed that a balanced accuracy of 78.56% could be attained using as few as 5% (248) of possible interactions to predict the remaining 95% (4702) pairwise interactions. Finally, the application of feature selection methods to generate ranked lists of potential mechanisms results in the discovery of mechanisms of competition in interactions with negative interactions at rates as high as 97 times faster than what would be expected by chance for this data set.
      Our results indicate that the goals of predictive microbial ecology can be greatly facilitated by combining trait-level representations of microbes with machine learning. Feature selection methods allow us to generate ranked lists of potential mechanisms that are specific to an interaction of interest and expedite the discovery of molecular mechanisms of interaction. This approach should prove useful as a guide for the construction of synthetic microbial communities and for lightening the experimental burden associated with mechanistic inquiries.

1. 1. Eric Franzosa
      Linking gut microbiome structure and metabolic activity in inflammatory bowel disease
      Eric A. Franzosa1,2, H. Mallick1,2, A. Sirota-Madi1, J. Avila-Pacheco1, N. Fornelos1, H.J. Haiser3, T. Vatanen1, A.B. Hall1, L.J. McIver1,2, B.W. Stevens5, J.S. Sauk5, R.G. Wilson5, J. Scott1, K. Peirce1, A. Deik1, K. Bullock1, F. Imhann4, J. Porter3, J. Fu4, A. Zhernakova4, R.K. Weersma4, C. Wijmenga4, C.B. Clish1, H. Vlamakis1, R.J. Xavier1,5, C. Huttenhower1,2
      1Broad Institute; 2Harvard School of Public Health; 3Novartis; 4Univ. Med. Center Groningen; 5Mass. General Hospital
      The inflammatory bowel diseases (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC), are associated with major changes in the gut microbiota. However, the relationship between IBD and the gut metabolome is less-well understood. Here, we performed untargeted metabolomic and shotgun metagenomic profiling of cross-sectional cohorts of CD, UC, and non-IBD subjects. Metabolomic and metagenomic profiles were broadly correlated with gut inflammation, and discriminated cases from controls with 87% accuracy. Across >8K metabolites, we identified major chemical classes that were differentially abundant (DA) in IBD, including enrichment for sphingolipids and bile acids and depletion for triacylglycerols and tetrapyrroles. Covariation analyses identified 122 putative species-microbe relationships that were perturbed in IBD, as well as potential roles for >50% of uncharacterized DA metabolites. Our findings thus improve understanding of the microbiome-metabolome interface in IBD.

1. 1. Michaela Gold
      Mucins reduce biofilm formation of Klebsiella pneumoniae
      Authors: Michaela Gold, Katharina Ribbeck
      Antimicrobial resistance is a growing threat, with an estimated 10 million people per year dying of related infections by 2050. Our approach to antimicrobial resistance is inspired by a natural component of the human body that can inhibit infections: mucus, a hydrogel that serves as a protective barrier and houses our microbiome. Mucins, large glycoproteins that give mucus its gel-forming structure, decrease biofilm formation and overall virulence in certain opportunistic pathogens. Here, we demonstrate that mucins reduce biofilm formation by the often multi-drug resistant and hypervirulent opportunistic pathogen Klebsiella pneumoniae and present our ongoing work on identifying its mechanism of influence. By understanding the mechanism of this protection, we hope to engineer new natural virulence-attenuating therapeutics against this bacterium.

1. 1. Mathieu Groussin
      Industrialization is associated with elevated rates of horizontal gene transfer in the human microbiome
      Previous studies have shown the importance of horizontal gene transfer (HGT) in shaping human gut bacteria. However, whether genomes acquire new genes daily, or whether millennia separate transfer events, remains unknown despite its importance on gut microbiome function and human health. Here, we generate thousands of gut bacterial genomes across hundreds of species, from different human populations worldwide, to show that the timescale of these HGTs is short, and that they frequently occur in each person’s gut microbiome. We further show that the gut bacteria of individuals with industrialized lifestyle exchange genes more frequently than in non-industrialized populations, suggesting that high rates of transfer may be a recent development in human history, linked to industrialization and urbanization. Finally, we find that the genes involved in transfer reflect the lifestyle of the human hosts, with elevated transfer of carbohydrate metabolism genes in hunter gatherer populations, and transfer of antibiotic resistance genes among pastoralists who live in close contact with livestock. Our study sheds new light on the factors influencing how gut bacteria acquire new functions during our lifetime. It also shows the importance of sampling and characterizing the gut microbiome biodiversity of underrepresented traditional populations. I will present our new initiative, the Global Microbiome Conservancy, a non-profit initiative that aims to preserve the full biodiversity of the human gut microbiome in a biobank of bacterial isolates and genomes to advance human health.

1. 1. Matthew Henke
      Discovery of an inflammatory polysaccharide from a Crohn’s disease-associated gut microbe
      Matthew T Henke1, Douglas Kenny2, Hera Vlamakis2, Ramnik J Xavier2, Jon Clardy1
      1Harvard Medical School, Boston, MA 02115 2Broad Institute, Cambridge, MA 02142
      Crohn’s disease is an inflammatory disease of the human digestive tract. While there are implications of genetics and diet, imbalances in members of the gut microbial community are strongly correlated with disease status. The gram-positive commensal organism Ruminococcus gnavus is among those most strongly, and consistently associated with disease. To establish a molecular understanding of the relationship between R. gnavus and Crohn’s disease, R. gnavus was grown in monoculture in a defined medium and screened for molecules that activate host immune response in dendritic cells.
      From this screen, a potent inflammatory complex polysaccharide has been discovered which induces secretion of inflammatory cytokines in a TLR4-dependent fashion. The structure of this polysaccharide was determined by NMR on the native material and derivatives, and while not previously reported, it resembles cell-wall polysaccharides of other pathobiont species in various human diseases. A potential biosynthetic gene cluster has been identified, which will facilitate further investigations of biosynthesis and clinical expression of this inflammatory polysaccharide. Additionally, a mouse model of inflammatory bowel disease is being used to determine the effects of R. gnavus and its purified immunomodulator on disease initiation, severity, and progression. This work will expand our knowledge of how a member of the gut microbiomemodulates the immune system and its relevance for IBD, and may also identify potential avenues for therapeutic intervention.
      Funding: NIH R01AT009708 (JC and RX) and F32GM126650 (MH).

1. 1. Jemila Kester
      C. difficile-associated antibiotics prime the host for infection by a commensal-independent mechanism
      Jemila C. Kester, Jason Velazquez, Douglas K. Brubaker, Douglas A. Lauffenburger, and Linda G. Griffith
      Biological Engineering, MIT, 77 Massachusetts Avenue, Cambridge, MA 02139
      Clostridioides difficile infection (CDI) is a major public health threat; the CDC estimates CDI accounts for 29,000 deaths and over 4 billion dollars in related healthcare costs annually. The most clinically relevant risk factor for CDI is recent antibiotic treatment. Though many broad-spectrum antibiotics significantly disrupt the structure of the gut microbiota, only certain ones increase CDI risk. This discrepancy led us to hypothesize that host-dependent factors may also contribute to CDI following antibiotic treatment.
      To test this, we used our established in vitro human gut epithelial system to treat mature mucosal barriers with antibiotics having CDI odds ratios ranging from one to 17. To gain an unbiased understanding of the system’s response, we quantified transcriptional changes using RNA-seq. A Self-Organizing Map was employed to discover shared patterns of gene expression, which suggested CDI-associated antibiotics might have decreased innate immune and mucosal barrier functions.
      To validate these predictions, we used Luminex, phagocytosis and bacterial killing, qPCR, and Alcian blue assays, which confirmed a marked loss of barrier and immune function with CDI-associated antibiotic pretreatment. This was distinct from pretreatment with an antibiotic unassociated with CDI, which did not alter innate immune or mucosal barrier function. Finally, we determined that these changes have consequences for CDI: pretreatment with CDI-associated antibiotics sensitized mucosal barriers to C. difficile toxin activity compared to an antibiotic unassociated with CDI and controls.
      These data implicate commensal-independent host changes in the increased risk of CDI with specific antibiotics. Future efforts will address the molecular mechanism of antibiotic-induced host sensitization to C. difficile toxin and explore host-directed therapies for CDI.
      JCK is supported by an administrative supplement to NIH R01-EB021908. This work was supported in part by the Koch Institute Support (core) Grant P30-CA14501 from the National Cancer Institute.

1. 1. Jacqueline M Lauer
      Markers of systemic inflammation and environmental enteric dysfunction are not reduced by zinc or multivitamins in Tanzanian infants: a randomized placebo-controlled trial
      Objective: To examine whether daily zinc and/or multivitamin (MV) supplementation reduce biomarkers of environmental enteric dysfunction (EED), systemic inflammation (SI), or markers of growth in a sample of infants from Dar es Salaam, Tanzania.
      Study design: Infants participating in a randomized, double-blind, placebo-controlled trial received daily oral supplementation of zinc, MVs, zinc + MVs, or placebo for 18 months starting at 6 weeks of age. EED (anti-flagellin and anti-LPS immunoglobulins), SI (C-reactive protein (CRP) and alpha-1-acid glycoprotein (AGP), and growth biomarkers (insulin-like growth factor-1 (IGF-1) and insulin-like growth factor binding protein-3 (IGFBP-3) were measured via ELISA in a subsample of 590 infants at 6 weeks and 6 months of age. EED biomarkers were also measured in 162 infants at 12 months of age.
      Results: With the exception of anti-LPS IgG concentrations, which were significantly higher in infants who received MVs compared with those that did not (1.41 ± 0.61 vs. 1.26 ± 0.65, p = 0.006), and IGFBP-3 concentrations, which were significantly lower in children who received zinc compared to those that did not (981.13 297.59 vs. 1019.10 333.01, p = 0.03), at 6 months of age, we did not observe any significant treatment effects of zinc or MV’s on EED, SI, or growth biomarkers.
      Conclusion: Neither zinc nor MV supplementation ameliorated markers of EED or SI during infancy. Other interventions should be prioritized for future trials. This trial was registered at clinicaltrials.gov as NCT00421668.
      Keywords: anti-flagellin, anti-LPS, alpha-1-acid glycoprotein, C-reactive protein, insulin-like growth factor-1, insulin-like growth factor binding protein-3

1. 1. Sydney Lavoie
      Crohn’s Disease Pathogenesis: Unraveling Genetic Risk, the Gut Microbiota, and the Immune System using Metagenomics and Gnotobiotics
      Inflammatory bowel disease (IBD) is driven by dysfunction between host genetics, the microbiota, and immune system. Knowledge gaps remain regarding how IBD genetic risk loci drive gut microbiota changes. The Crohn’s disease risk allele ATG16L1 T300A results in abnormal Paneth cells due to decreased selective autophagy, increased cytokine release, and decreased intracellular bacterial clearance. To unravel the effects of ATG16L1 T300A on the microbiota and immune system, we employed a gnotobiotic model using human fecal transfers into ATG16L1 T300A knock-in mice. We observed increases in Bacteroides ovatus and Th1 and Th17 cells in ATG16L1 T300A mice. Association of altered Schaedler flora mice with B. ovatus specifically increased Th17 cells selectively in ATG16L1 T300A knock-in mice. Changes occur before disease onset, suggesting that ATG16L1 T300A contributes to dysbiosis and immune infiltration prior to disease symptoms. Our work provides insight for future studies on IBD subtypes, IBD patient treatment and diagnostics.

1. 1. Sanghun Lee
      Type-2 Diabetics Reduces Spatial Variation of Microbiome Based on Extracellur Vesicles from Gut Microbes across Human Body
      Geumkyung Nah1, Sangchul Park1, Sanghun Lee2 and Sungho Won1
      1Department of public health sciences, Seoul National university, Seoul, Korea
      2Department of medical consilience, Graduate school, Dankook university, Korea
      Due to the advance in sequencing technology, the role of gut microbiota has been unveiled in underlying mechanism of type-2 diabetes mellitus (T2DM). Recently, intestine hyper-permeability linked to T2DM has been focused in the interaction between gut microbes and leaky gut epithelium which increases the uptake of macromolecules like lipopolysaccharide from membrane of microbes. However, it is unclear whether either dysbiotic gut microbiota or T2DM itself is causal on dysfunction of intestinal permeability. Moreover, the assessment of intestine permeability is also not easy. Therefore, we investigate co-occurrence of microbes targeting stool microbes and microbe-derived extracellular vesicles (EVs) across serum and urine in human which is convenient of intestine permeability. The co-occurrence of microbes among those were compared between patients with T2DM and matched healthy subjects. Our results showed substantially higher correlations in T2DM compared to healthy subjects, and no difference of microbial composition at phylum and genus levels between two groups. Therefore, it provides more clear insight into the pathophysiological mechanisms of T2DM that dysfunction of intestinal permeability in mainly due to T2DM itself rather than gut microbiota. The significant correlation of EVs in T2DM patients also suggests the role of EVs as a biomarker on the intestinal permeability named as leaky gut in T2DM.
      KEYWORDS: Bacteria-derived extracellular vesicles, microbial correlation analysis, gut microbiome, urine microbiome, serum microbiome, type 2 diabetes.
      Corresponding Authors: Sanghun Lee, (Email) slee@hsph.harvard.edu / Sungho Won, (Email) swon@hsph.harvard.edu

1. 1. Yanping Li
      Dietary lignan and cardio-metabolic risk: exploring the role of the gut microbiome
      Yanping Li,1 Jun Li,1 Kerry L. Ivey,1 Jeremy Wilkinson,2 Dong Wang,1 Andrew T. Chan, Curtis Huttenhower,2,3 Frank Hu, 1,5 Qi Sun, 1,5 Eric B. Rimm1,5
      1Department of Nutrition and 2Biostatistics, Harvard T. H. Chan School of Public Health, Boston, MA;3Broad Institute, Cambridge, MA; 4Gastrointestinal Unit, Massachusetts General Hospital and Harvard Medical School, Boston, MA; 5 Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA.
      Background and study purpose: The conversion of plant lignans to to enterolignans for absorption takes place in the colon and is employed by the gut microbiota. We aim to explore the interconnections between dietary lignans, plasma enterolignans, gut microbiome, and cardio-metabolic risk (LDL-cholesterol, HDL-cholesterol, & triglycerides).
      Design and method: the study was a repeated cross-sectional survey. Dietary lignan data was collected by 7 days dietary records. DNA was extracted from 916 samples of 301 participants and RNA was extracted from 369 samples of a subgroup of 96 participants. Taxonomic profiling was performed with MetaPhlAn2 and functional profiling of genes and transcripts were performed using HUMAnN2. Differentially abundant species were identified using MaAsLin2.
      Results: A high ratio of enterolactone to lignans was significantly associated with the relative abundance of 35 species (24 enriched and 11 depleted species, Q<0.25), including 8 bacteroidetes species and 13 Firmicutes species. Increasing dietary lignans intake was only significantly associated with a lower plasma ratio of triglyceride to HDL when plasma enterolactone was high. The association between dietary lignans and ratio of TG to HDL was mainly mediated by plasma enterolactone level (mediation effect: 74.3%, P=0.002), which could partially captured by the weighted species score (mediation effect: 28.6%, P=0.02). We observed several significant interactions between dietary lignans and presence of species on ratio of TG to HDL implying that the protective effect of dietary lignans on CVD risk factor depends on the presence of certain microbiota. I-isoleucine biosynthesis and glutaryl-coa degradation pathways were significantly enriched in participants with a high level of plasma enterolactone.
      Conclusion: Health benefit of dietary lignans might, in some degree, depend on enrichment of certain species.

1. 1. Siyuan Ma
      Population Structure Discovery in Meta-Analyzed Microbial Communities
      Siyuan Ma1, Dmitry Shungin2, Himel Mallick1,2, Melanie Schirmer1,2, Raivo Kolde3, Eric Franzosa1,2, Hera Vlamakis2, Ramnik Xavier2, Curtis Huttenhower1,2
      1Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
      2Broad Institute of MIT and Harvard, Cambridge, MA, USA.
      3Philips Research, Cambridge, MA, USA
      The inflammatory bowel disease (IBD) is one of the best-studied, microbiome-linked complex diseases. However, it remains unclear whether there exist consistent population structures to characterize the large, unexplained heterogeneity within the disease population’s gut microbiome. We propose a statistical framework to identify and validate reproducible varying patterns across microbiome cohorts, taking specialized efforts to correct for the strong effects of cohort and batch differences that tend to occur for this type of data. By applying our method to a large collection of 10 16S IBD microbiome studies, we identified consistent microbial links of the disease, including both previously documented and novel findings on treatment effects, disease phenotypic severity, and interactions with sample types and disease subtypes. We found no evidence supporting discrete clustering structure in the IBD microbiome, but instead identified two reproducible, continuously varying scores representing disease-associated dysbiosis and dominant phyla trade-off, that characterized patient microbial heterogeneity across cohorts. Our statistical framework provides functionalities to perform the various tasks in microbiome meta-analysis that requires consideration for batch and cohort effects; our characterization of the IBD gut microbiome represents a first effort to consistently address the disease’s microbial links and large heterogeneity.

1. 1. Rene Niehus
      Quantifying probiotic effects on microbiome resistance in newborns of a neonatal ICU
      Antibiotics can disrupt the human gut microbiota and cause amplification of pre-existing resistance genes. High within-patient resistance is associated with higher resistance in infections, and may also contribute to increased population-level spread of resistance. Probiotic bacterial strains have been used successfully in mice to supress antibiotic-resistant pathogens. However, little is known about the effects of probiotics on the human microbiome. We analyse metagenomic shotgun data from a prospective follow-up study of eleven newborns staying in a neonatal ICU in Siem Reap, Cambodia. In addition to varying antibiotic treatments, six of these patients recieved daily probiotic treatment with Lactobacillus acidophilus. We build a dynamic model of the within-host resistance dynamics that explicitly models an internal metagenomics standard, enabling us to quantify effects of probiotics on absolute abundance of different resistance markers. Our model allows us to simulate the microbiome dynamics of resistance under varying conditions, and to design an efficient trial to infer probiotic microbiome effects.

1. 1. Jason Nomburg
      Microbiome Analysis of Tanzanian Esophageal Cancer
      Jason Nomburg 1,2,3, Susan Bullman 4, Katherine Van Loon 5, Eric Collisson 5, Charlie Vaske 6, Yulia Newton 6, Amie Radenbuagh 6, Beatrice Mushi 7, Msiba Selekwa 7, Larry Akoko 7, Elia J. Mmbaga 7, James A. DeCaprio 1,2 Matthew Meyerson 2,3
      1 Program in Virology, Division of Medical Sciences, Harvard Medical School, Boston, Massachusetts, USA
      2 Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA
      3 The Broad Institute at MIT and Harvard, Cambridge, MA 02142, USA
      4 Fred Hutchinson Cancer Research Center, Seattle, Washington, USA
      5 University of California, San Francisco, San Francisco, CA, USA.
      6 NantOmics , 2919 Mission Street , Santa Cruz , California 95060 , USA.
      7 Muhimbili University of Health and Allied Sciences, Dar es Salaam, Tanzania
      Esophageal cancer is a leading cause of worldwide cancer-associated deaths. Notably, there is a high incidence of esophageal squamous cell carcinoma (ESCC) throughout the eastern corridor of Africa. We sought to investigate a possible microbial etiology behind this epidemiological pattern. To address this question, we conducted next generation sequencing, including RNA sequencing (RNAseq) and whole-genome sequencing (WGS), of samples from 61 Tanzanian ESCC patients. We used the computational microbial identification and classification software PathSeq to conduct a preliminary microbial abundance analysis of these samples. Our initial results indicate an extremely high abundance of Fusobacterium, Selenomonas, and Prevotella in the tumors of these patients. These bacteria are notable for their association with colon adenocarcinoma (COAD) and are correlated with distinct clinical and molecular COAD characteristics. We have found that these bacteria are significantly enriched in these Tanzanian ESCC samples compared to their abundance in esophageal adenocarcinoma (ESCA) and ESCC samples from nearly 200 North American patients available through The Cancer Genome Atlas (TCGA). We seek to determine the functional role of these bacterial species in Tanzanian ESCC. We hypothesize that Fusobacterium and associated oral microorganisms are etiologic agents of Tanzanian ESCC.

1. 1. Scott Olesen
      OpenBiome: a non-profit stool bank supporting basic microbiome science and clinical FMT research
      OpenBiome is a public, non-profit stool bank whose mission is to expand safe access to fecal microbiota transplant (FMT) and to catalyze microbiome research. OpenBiome recruits and thoroughly screens stool donors, collecting and processing their donated material into capsule or liquid formulations for use in FMT. We provide FMT preparations for treatment of Clostridioides difficile infection and for research in FMT clinical trials. OpenBiome also keeps aliquots and metadata from all our donors for use in basic research.

1. 1. Ali Rahnavard
      Netome: a computational framework for metabolite profiling and omics network analysis
      Ali Rahnavard, Daniel Hitchcock, Julian Avila Pacheco, Amy Deik, Courtney Dennis, Sarah Jeanfavre, Kerry Pierce, Kevin Bullock, Zach Costliow, Clary B. Clish
      Broad Institute of MIT and Harvard, Cambridge, MA 02142, USA
      Summary: Advances in metabolomics technologies have enabled comprehensive analyses of associations between metabolites and human disease and have provided a means to study biochemical pathways and processes in detail using model systems. Liquid chromatography tandem mass spectrometry (LC-MS) is an analytical technique commonly used by metabolomics labs to measure hundreds of metabolites of known identity and thousands of “peaks” from yet to be identified compounds that are tracked by their measured masses and chromatographic retention times. netome is a computational framework that provides tools for analyzing processed LC-MS data. In this framework, we develop and provide various computational resources including individual software modules to inspect and adjust trends in raw data, align unknown peaks between separately acquired data sets, and to remove redundancies in nontargeted LCMS data arising from multiple ionization products of a single metabolite. netome is publicly available with extensive documentation and support via issue tracker at https://broadinstitute.github.io/netome under the MIT license. netome includes a set of computational methods that have been designed to execute quality control and post-raw data processing tasks for metabolomics data (e.g. scaling and clustering metabolite abundances), as well as statistical association testing in a network manner (e.g., testing relationship between metabolites and microbes). Each individual tool is available with source code, workshop-oriented documentation which includes instructions for installation and using tools with demonstration examples, and a web server with all services.

1. 1. Stephanie D Song
      Vaginal microbial diversity changes across the menstrual cycle in healthy young women
      Stephanie D Song1, Kalpana D Acharya, PhD1, Jade Zhu1, Marina RS Walther-António, PhD2,3, Marc J Tetel, PhD1, Nicholas Chia, PhD2,4
      1Neuroscience Program, Wellesley College, Wellesley, MA
      2Department of Surgery, Division of Surgical Research, Mayo Clinic, Rochester, MN
      3Department of Obstetrics and Gynecology, Mayo Clinic, Rochester, MN
      4Microbiome Program, Center for Individualized Medicine, Mayo Clinic, Rochester, MN
      The vaginal microbiome, the collection of microorganisms and their metabolites residing in the vaginal tract, is critical for maintaining women’s health. For example, the vaginal microbiome protects against pathogens by maintaining a low-pH environment. In particular, Lactobacillus species predominate most vaginal microbiomes and produce lactic acid, which has been cited as a key protector of vaginal health. Meanwhile, disruptions to the finely balanced vaginal microbiome can cause infections, such as bacterial vaginosis, and may contribute to the development of endometrial and ovarian cancer. While the crucial and diverse role of the vaginal microbiome is becoming more apparent, very little is known about the endogenous factors that influence its composition and stability. In the present study, we tested the hypothesis that the vaginal microbiota composition changes over the menstrual cycle. Volunteer participants from a women’s college self-swabbed vaginal samples daily for 2 months, while recording timing of menstruation and contraceptive use with a mobile application. High-throughput 16S rRNA sequencing was done to identify the microbial compositions of each sample. The high temporal resolution of this study allows the capture of time-dependent effects otherwise lacking in previous cohort studies. As measured by the Shannon index, alpha diversity increases during menstruation (p < 0.001). Moreover, the increase in diversity during menses is even more pronounced among individuals without hormonal contraceptive use. We are currently analyzing which microbes contribute to these differences in diversity. In addition, we are analyzing these vaginal microbiota data for effects of diet, exercise and mood. Our goal is to establish a foundation for understanding the dynamics of the vaginal microbiome in healthy women. Moreover, by tracking these participants in the long-term, these findings could provide insights into how the vaginal microbiome is involved in reproductive tract cancers in women.
      Sources of Research Support: Arnold and Mabel Beckman Foundation Beckman Scholars Program (SDS), Mayo Clinic Center for Individualized Medicine (NC), Wellesley College Research Funds (MT)

1. 1. Jenna Swarthout
      Characterizing antibiotic resistomes in human and animal feces from Bangladesh using long-read metagenomic sequencing
      *withdrawn*

1. 1. Kelsey Thompson
      Dysbiosis of the gut microbiome in patients with inflammatory arthritis
      Kelsey N Thompson1, Kevin S Bonham1,2, Nicholas E Ilott3, Lilian H Lam3, Paula Colmenero3, Andrew Filer4, Sam J Bullers3, Matthew A Jackson3, India Brough3,5, Stephen P Young5, Arthur G Pratt7, Paul Bowness5, Dan R Littman8, Karim Raza4, 5, Fiona Powrie3 and Curtis Huttenhower1,2
      1 Department of Biostatstics, T.H. Chan School of Public Health, Harvard University, Boston, MA, USA
      2 Broad Institute of MIT and Harvard, Cambridge, MA, USA
      3 Kennedy Institute of Rheumatology, Oxford University, Oxford, UK
      4 Department of Rheumatology, Sandwell & West Birmingham Hospitals NHS Trust, Birmingham, UK
      5 Nuffield Department of Orthopedics, Rheumatology and Musculoskeletal Sciences, Oxford University, Oxford UK
      6 Institute of Inflammation and Ageing, College of Medical and Dental Sciences, University of Birmingham, Queen Elizabeth Hospital, Birmingham, UK
      7 Institute of Cellular Medicine, Newcastle University, Newcastle, UK
      8 Department of Pathology and Department of Microbiology, Skirball Institute, NYU Langone Health, New York, NY, USA
      Approximately 23% or 54 million adults in the United States have been diagnosed with varying forms of arthritis. The gut microbiome has been implicated not only in gastrointestinal inflammatory conditions, but in systemic diseases such as arthritis. To identify potential structural or functional drivers of this association, we investigated the gut microbiome using metagenomic shotgun sequencing in 327 adults diagnosed with rheumatoid arthritis, ankylosing spondylitis, psoriatic arthritis, and other forms of arthritis, as well as 69 control individuals. Approximately 2.9% of the variability in gut microbial taxonomic profiles was explained by patient diagnosis, comparable to that in gastrointestinal conditions such as inflammatory bowel disease (IBD). Similar amounts of variation (2.7%) were explained by diagnosis in the functional profiles, indicating a reasonably close link between perturbed microbial community structure and function in the disease. For the purposes of identifying specific microbiome correlates with arthritis phenotypes, we accounted for additional biometric and environmental covariates including, age, gender, ethnicity, and others. This identified significant increases in taxa characteristic of the oral cavity among patients with inflammatory arthritis, including several species from the genus Streptococcus. Again, this strikingly mirrors similar findings regarding oral microbial enrichment in conditions such as IBD. Additional analyses including strain and pangenome differences between arthritis types and healthy controls, random forest modeling, and per-feature tests remain to be performed within the cohort. These will be further integrated with gut microbial metabolomic and transcriptomic profiles and host immunoprofiling. Together, this represents the first in-depth study of the gut microbiome across different etiologies of arthritis using culture-independent techniques.

1. 1. Matthew Volpe
      In vitro characterization of the colibactin-activating peptidase ClbP enables design of a fluorogenic activity probe
      Authors: Matthew R. Volpe, Matthew R. Wilson, Carolyn A. Brotherton, Ethan S. Winter, Sheila E. Johnson, and Emily P. Balskus
      The small-molecule genotoxin colibactin is produced by both commensal and pathogenic bacteria and has been linked to the development of colorectal cancer in both human patient cohorts and mouse models. In the final stages of colibactin’s biosynthesis, an inactive precursor (precolibactin) undergoes proteolytic cleavage by ClbP, an unusual inner membrane-bound periplasmic peptidase, to generate the active genotoxin. This enzyme presents an opportunity to monitor and modulate colibactin biosynthesis, but its active form has not yet been studied in vitro and limited tools exist to measure its activity. Here, we describe the in vitro biochemical characterization of catalytically active, full-length ClbP. We elucidate its substrate preferences and use this information to develop a fluorogenic activity probe. This tool can streamline the identification of pks+ bacterial isolates and accelerate the discovery of small molecule inhibitors of ClbP. These tool compounds present an opportunity to dissect the biological effects of this uncharacterized genotoxin in the context of a complex microbial community with a level of control and precision that is inaccessible with current genetic approaches.

1. 1. Dong D. Wang
      Mediterranean Dietary Pattern, Gut Microbiome and Cardiometabolic Markers in the Men’s Lifestyle Validation Study
      Dong D. Wang 1, Long H Nguyen 2,3, Yanping Li 1, Yan Yan 4, Wenjie Ma 2,3, Walter C. Willett 1,5,6, Frank B. Hu 1,5,6, Eric B. Rimm 1,5,6, Meir J. Stampfer 1,5,6, Andrew T. Chan 2,3,6, Curtis Huttenhower 4,7
      Departments of 1 Nutrition, 4 Biostatistics and 5 Epidemiology, Harvard T.H. Chan School of Public Health
      2 Division of Gastroenterology, Department of Medicine, and 3 Clinical and Translational Epidemiology Unit, Massachusetts General Hospital
      6 Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School
      7 Broad Institute of MIT and Harvard
      Objectives: To identify gut microbial signatures of the Mediterranean diet (MedDiet) and investigate the role of gut microbiome in modulating the association between the MedDiet and cardiometabolic markers.
      Design and method: We collected fecal samples from 307 participants in the Men’s Lifestyle Validation Study at up to four time points, resulting in 923 shotgun metagenomes and 347 shotgun metatranscriptomes. Dietary information was collected using food frequency questionnaires and 7-day dietary records. Each participant’s adherence to the MedDiet was evaluated by a 9-dimension MedDiet index. Plasma cardiometabolic biomarkers, including C-reactive protein (CRP), triglyceride, total cholesterol (TC) and high-density lipoprotein cholesterol (HDL-C) were measured using standard methods. Taxonomic profiling was performed with MetaPhlAn2 and functional profiling of genes and transcripts were performed using HUMAnN2. We applied MaAsLin2 to quantify the associations between the MedDiet index and gut microbial features with adjustment for potential confounding factors.
      Results: The MedDiet index was significantly associated with the relative abundance of 36 gut bacterial species (FDR P < 0.25). A majority of the identified species belongs to the Firmicutes, while several species within the Actinobacteria, Bacteroides, Proteobacteria and Verrucomicrobia were associated with the MedDiet index. Species that showed strong positive association with the MedDiet index included Eubacterium Eligens, Roseburia Hominis, Faecalibacterium Prausnitzii and Bacteroides Cellulosilyticus, whereas the relative abundance of Ruminococcus Torques, R. Gnavus, Blautia Hydrogenotrophica and Clostridium Leptum was associated strong inverse associations with the MedDiet index. In general, the associations of animal-sourced foods, e.g., red meat, with the identified species were in opposition direction of the associations of plant-sourced foods, e.g., whole grains, fruits and vegetables. The DNA and RNA enzyme commissions positively associated with the MedDiet index largely belonged to the functional categories of carbohydrate metabolism, particularly those related to complex carbohydrate degradation, and amino acid metabolism. There is a suggestive evidence that the identified species may modify the inverse associations of the MedDiet index with plasma CRP (P interaction =0.05) and the ratio of TC and HDL-C (P interaction =0.03).
      Conclusions: A greater adherence to a Mediterranean-style dietary pattern was associated with enrichment of species related to degradation of complex carbohydrates. The association between the MedDiet and cardiometabolic risk may vary by individuals’ gut microbial profiles.

1. 1. Lea Wang
      The Epidemic Plasmid Carrying blaNDM in Enterobacteriaceae Isolates in Pigs
      Background: Carbapenemase-producing Enterobacteriaceae (CPE) has spread worldwide these years, posing a severe threat to public health. Historically, the dissemination of CPE is a problem restricted to human. Nonetheless, reports of CPE in non-human sources are increasing recently. This study identified NDM-producing isolates from a surveillance program in Hong Kong on food animals. The epidemic plasmid mediating the dissemination of blaNDM is characterized.
      Materials/methods: From Sep 2008 to Feb 2017, intestinal swabs were obtained from 856 pigs. Of these, 742 were imported from 10 provinces in mainland China and 114 were reared locally in Hong Kong. Enrichment broth containing 10 mg/L vancomycin and 0.5 mg/L meropenem was used to screen for carbapenem-resistance. Strains were identified to species by MALDI-TOF and susceptibility was tested according to CLSI instruction. Common carbapenemase genes (blaNDM, blaIMP, blaVIM, blaKPC and blaOXA-48) were detected by PCR. Six representative carbapenem-resistant strains were subjected to whole genome sequencing under Illumina Hiseq platform. Raw data was trimmed, followed by assembly in CLC workbench and RAST annotation. Complete genomes were compared through SNP typing. Epidemic plasmids were studied through comparative analysis.
      Results: A total of nine carbapenem-resistant E. coli were obtained from six pigs originating from six different farms in China. All these isolates carry carbapenemase gene blaNDM-5, mostly (in 8/9) carried by an IncX3 plasmid. SNP analysis showed that the strains were in different cluster, and the correlation between geographic origins and phylogenetic relations are weak, suggesting a diverse genome-wide evolution background among the NDM-producing isolates. However, the IncX3 plasmids harboring blaNDM-5 share high similarity. The insertion sites of blaNDM-5 were identical, indicating an evolution within these IncX3 plasmids, and their possibly descendants from a human-originated plasmid, pNDM-HN380, after deletion of genetic contents.
      Conclusions: This study identified an epidemic IncX3 plasmid carrying blaNDM-5, disseminated among E. coli originating from pigs. The similarity between animal-sourced IncX3 plasmids and those from humans suggested the human-animal circulation of CPE in China.

1. 1. Caroline Werlang
      Salivary mucins reduce virulence traits of cavity causing Streptococcus mutans
      Salivary mucins reduce biofilm formation in cavity-causing Streptococcus mutans, a phenotype that is often linked with other virulence traits such as horizontal gene transfer. We show that purified mucin gels decrease mRNA levels of XIP, a putative quorum sensing peptide, and reduce natural transformation. This suggests that the production of mucins is a method the hostuses to suppress microbial virulence and gene transfer.

1. 1. Jeremy Wilkinson
      The Harvard T. H. Chan School of Public Health Microbiome Analysis Core (HMAC): Advancing Microbiome Research
      Jeremy E. Wilkinson¹, Lauren J. McIver¹, Anahit Mailyan¹, Kelsey N. Thompson¹, Curtis Huttenhower^1,21Harvard T. H. Chan School of Public Health, ²Broad Institute of MIT and Harvard
      The Microbiome Analysis Core at the Harvard T. H. Chan School of Public Health (HMAC) was established in response to the rapidly emerging field of microbiome research and its potential to affect studies across the biomedical sciences. HMAC’s goal is to aid researchers with human microbiome study design and interpretation, reducing the gap between primary data and translatable biology. The mission of the HMAC is to provide formalized support, at the highest quality standards, for human microbiome-related studies and to foster collaborative initiatives in microbiome research. The HMAC provides a single point of contact for integrative microbiome informatics and analysis of data from amplicon (16S/18S rRNA gene and internal transcribed spacer [ITS] DNA), shotgun metagenomic, and metatranscriptomic sequencing studies, as well as metabolomics and other molecular analyses of the human microbiota. The HMAC has extensive experience with microbiome profiles in diverse populations, including bioinformatics processing of such data from large cohorts, quantitative ecology, and subsequent microbial systems and human epidemiological analysis. By integrating microbial gene function and taxonomic profiles with host information, we enable researchers to interpret molecular activities of the microbiota and assess its impact on human health.

1. 1. Sungho Won
      Phylogenetic Tree-based Microbiome Association Test
      Kang Jin Kim1, Jaehyun Park2, Sang-Chul Park3 and Sungho Won1, 2, 3*
      1Department of Public Health Sciences, Graduate School of Public Health, Seoul National University, Seoul, South Korea, 2Interdisciplinary Program for Bioinformatics, College of Natural Science, Seoul National University, Seoul, South Korea, 3Institute of Health and Environment, Seoul National University, Seoul, South Korea
      Motivation: Microbial ecological patterns exhibit high inter-subject variation, with few operational taxonomic units (OTUs) for each species. To overcome these issues, non-parametric approaches, such as the Mann-Whitney U-test and Wilcoxon rank-sum test, have often been used to identify OTUs associated with host diseases. However, these approaches only utilize the ranks of observed relative abundances, leading to information loss, and are associated with high false-negative rates. In this article, we propose a phylogenetic tree-based microbiome association test (TMAT) to analyze the associations between microbiome OTU abundances and disease phenotypes. Phylogenetic trees illustrate patterns of similarity among different OTUs, and TMAT provides an efficient method for utilizing such information for association analyses. The proposed TMAT provides test statistics for each node, which are combined to identify mutations associated with host diseases.
      Results: Power estimates of TMAT were compared with existing methods using extensive simulations based on real absolute abundances. Simulation studies showed that TMAT preserves the nominal type-1 error rate, and estimates of its statistical power generally outperformed existing methods with regard to the considered scenarios. Furthermore, TMAT can be used to detect phylogenetic mutations associated with host diseases, providing more in-depth insight into bacterial pathology.
      Availability: The 16S rRNA amplicon sequencing data from the CRC and ME/CFS are available from the European Nucleotide Archive (ENA) database under project accession number PRJEB6070 and PRJEB13092, respectively. TMAT was implemented in the R package. Detailed information is available at http://healthstat.snu.ac.kr/software/tmat.

1. 1. Colin J. Worby
      The “gut-bladder axis”: gut microbiome dynamics and E. coli transmission in women with recurrent urinary tract infections
      Colin J. Worby1; Henry L. Schreiber IV1,2,3; Timothy J. Straub1; Lucas van Dijk1; Alexandra E. Paharik2,3; Wen-Chi Chou1; Christopher A. Desjardins1; Abigail L. Manson1; Sinéad B. Chapman1; Karla Bergeron4; Aleksandra Klim4; H. Henry Lai4; Karen W. Dodson2,3; Bruce J. Walker1,5; Scott J. Hultgren2,3*; Ashlee M. Earl1*
      1 Infectious Disease and Microbiome Program, Broad Institute, Cambridge, MA
      2 Molecular Microbiology, Washington University School of Medicine, St. Louis, MO
      3 Center for Women’s Infectious Disease Research, Washington University School of Medicine, St. Louis, MO
      4 Division of Urologic Surgery, Washington University School of Medicine, St. Louis, MO
      5 Applied Invention, LLC, Cambridge, MA
      Urinary tract infections (UTIs) are common, costly, and frequently recurrent. The majority of UTIs are caused by uropathogenic Escherichia coli (UPEC), which can reside asymptomatically in the gut of otherwise healthy individuals, but may ascend into the bladder through the urethra. This induces an over-exuberant inflammatory response resulting in pain, increased frequency and urgency in urination, and sometimes recurrent disease (rUTI). Despite the importance of this gut-bladder axis, the population dynamics of UPEC resident in the gut around the time of UTIs remain poorly understood. Further, it is unknown to what extent the structure of the gut microbiome influences UPEC colonization and UTI susceptibility. To address these questions, we conducted a longitudinal study of women with and without a history of rUTI using metagenomic techniques on host fecal and urine samples. We identified differences in the structure and diversity of the gut microbiota that typified rUTI history, indicating that the gut environments of women with rUTI are fundamentally different than healthy controls. Further, E. coli strain-level analyses indicated that rUTIs are frequently caused by strains from specific phylogroups which are also better adapted to stable colonization of the gut, indicating that fitness in the gut is linked to infectivity of the bladder in humans. Overall, our work suggests that an atypical microbiome composition with less strain diversity is associated with a history rUTI in women, and that the gut environments of women with frequent rUTI may provide a rich reservoir enabling UPEC transmission to the bladder.

1. 1. Yan Yan
      Structure of the mucosal and stool microbiome in Lynch syndrome
      Yan Yan1,2, David A. Drew4,5, Arnold Markowitz3, Jason Lloyd-Price1,2, Galeb Abu-Ali1,2, Long H. Nguyen1,4,5, Daniel C. Chung4, Christina Tran3, Katherine K. Gilpin4,5Dana Meixell4, Melanie Parziale4, Madeline Schuck4, Zalak Patel3, James M. Richter4, Peter B. Kelsey4, Wendy S. Garrett1,2, Andrew T. Chan4,5*, Zsofia K. Stadler3*, Curtis Huttenhower1,2*
      1 Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
      2 Broad Institute of MIT and Harvard, Cambridge, MA, USA.
      3 Memorial Sloan Kettering Cancer Center, NY, USA.
      4 Gastroenterology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA.
      5 Epidemiology Unit, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA.
      Changes in the gut microbiota have been associated with colorectal cancer (CRC), but the causal alterations that may precede CRC development have not been clearly elucidated. An investigation among patients with Lynch syndrome, the most common hereditary form of CRC, provides a unique opportunity to prospectively assess microbiome changes linked to colorectal neoplasia within a relatively short time frame. We recruited 100 individuals from two large clinical genetics centers, who provided a stool sample and biopsies from up to two colon locations at baseline, plus clinical profiles at both baseline and one-year follow-up. This yielded 87 stool specimens with 187 matched biopsies. Stool samples underwent metagenomic and metatranscriptomic profiling, while biopsies were evaluated by 16S rRNA gene sequencing. Colectomy (segmental or subtotal i.e. rectal-sparing colectomy), in tandem with a personal history of CRC (n=42), accounted for the largest effects on species composition (13.7% of variation) and metagenomic functional profiles (15.6%), as well as transcriptomic functional activity profiles (10.8%). Colectomy and age had stronger effects on metatranscriptomic functional activity than on metagenomic functional potential (e.g. 18 colectomy related transcripts and 12 age associated transcripts, whereas 8 colectomy related genes and 6 age associated functional genes). Transcripts related to flagellin, a potent immunomodulatory protein involved in colorectal carcinogenesis, were enriched in subjects without colectomy, both among individually transcribed gene families and across contributing species such as Roseburia spp. A random forest classifier leveraging stool metatranscriptomes resulted in modest additive power to predict the interval development of preneoplastic colonic polyps, suggesting a potentially nuanced link between microbially-mediated processes and progression along the adenoma-carcinoma pathway. This study demonstrates the feasibility of strategies to monitor Lynch patients using non-invasive, microbiome-focused sequencing and analytic methods and to disaggregate early-stage, potentially changes that are associated causally with the development of neoplasia from microbiome responses that correlate with CRC and its associated treatment.

1. 1. An-Ni Zhang
      Environmental Selection of Butyrate Producers Drives Genetic But Not Species Convergence
      An-Ni Zhang, Eric Alm*
      Alm Lab, Biological Engineering, MIT
      Butyrate produced by gut microbiota provides the major energy for colonocytes and indicates the health status of our gut, yet fundamental questions remain on how the environmental selection drives the butyrate producers. To study it, we first design a bioinformatic tool to predict butyrate producers using 16S, and apply this tool to 125 16S studies across 19 habitats.
      Our tool can identify and predict butyrate producers with 99.7% specificity and 91.6% coverage at the species level. We present a new tool, buty_phyl (https://github.com/caozhichongchong/buty_phyl), to predict butyrate producers using phylogenetic topology of 16S. We identify a wide lineage of butyrate producers in all currently available genomes from NCBI Genome database. Based on it, we can infer the taxonomy and abundance of butyrate producers (species and pathogenicity) in 16S data. We test and tune this tool following two criteria: 1) all genomes within one species should be either butyrate producers or non-producers; 2) if most genes within one species are divergent from the database, this species should be a true negative. We used 3,120 human gut isolates from Global Microbiome Conservancy (http://microbiomeconservancy.org/) to test and tune this tool, resulting in 99.7% specificity and 91.6% coverage at the species level.
      Here, we show that the environmental selection on butyrate producing drives genetic but not species convergence. We apply this tool to all 26,000 16S samples from the MicrobiomeHD database and the Earth Microbiome Project (EMP). We show that the urbanization and current diets have shifted our gut microbiota into a lower diversity and abundance of butyrate producers. Diseases also shape the gut microbiota from healthy status, and the response of butyrate producers is disease-dependent. Of 13 diseases we test, patients of Clostridium difficile infection, colorectal cancer, HIV, inflammatory bowel disease, and obesity show significant differences from the healthy gut. We find that butyrate is mainly contributed by beneficial commensals (averagely 98%), and diseases do not select pathogenic butyrate producers. Finally, we compare the butyrate-producing species across different habitats. Human gut harbors a much higher diversity (averagely 20 species) and abundance (a majority of 5-20%), compared to other host-associated habitats (animal and plant) and free-living habitats (sediment, soil, air, and water). The results highlight a few butyrate producers wildly distributed across all habitats. However, they form a functional cohort instead of a monophyletic group, covering five phyla. Further evidence shows that genes in butyrate-producing pathways are clustered more ecologically over phylogenetically. These observations indicate that the habitats tend to select specific genotypes but not specific taxa.

1. 1. Yancong Zhang
      Identifying Novel Bioactive Microbial Gene Products in Inflammatory Bowel Disease
      Yancong Zhang1,2, Eric A. Franzosa1,2, Sena Bae2, Cesar Arze2, Damian R. Polichita1, Ayshwarya Subramanian1,2, Andy Krueger5, Hera Vlamakis1, Ramnik J. Xavier1,3,4, Curtis Huttenhower1,2,*
      1Infectious Disease and Microbiome Program, Broad Institute of MIT and Harvard, Cambridge, MA 02142, USA
      2Department of Biostatistics, Harvard T. H. Chan School of Public Health, Boston, MA 02115, USA
      3Gastrointestinal Unit and Center for the Study of Inflammatory Bowel Disease, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114, USA
      4Center for Microbiome Informatics and Therapeutics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
      5Takeda Pharmaceutical Company Limited
      *Correspondence to: chuttenh@hsph.harvard.edu
      Gut microbial organisms and associated bioactive compounds are often disrupted in gastrointestinal conditions such as inflammatory bowel disease (IBD). Since more than one thirds of all proteins in the microbiome are uncharacterized, we developed a method to computationally prioritize all potentially bioactive gene products from metagenomes and find new immunomodulatory gene families in IBD, specifically targeting secreted/extracellular proteins to enrich for likely host interactors. When applied to joint IBD microbiome datasets containing over 3000 total metagenomes and metatranscriptomes, approximately 30% prioritized gene products were uncharacterized with little homology in UniRef90. While 90% of prioritized novel genes expanded the pangenomes of common gut taxa, about 10% were likely both functionally novel and associated with uncharacterized species. Both sets of newly prioritized bioactives will require future experiments validating the mechanisms by which they interact with host immunity in IBD and gut inflammation.

1. 1. Xiaobin Zheng
      Yogurt Consumption and Risk of Conventional Colorectal Adenoma and Serrated Lesion
      Xiaobin Zheng, MD1,2,3, Kana Wu, MD, MPH, PhD, Shuji Ogino, MD, MS, PhD, Charles S. Fuchs, MD, MPH, Andrew T. Chan, MD, MPH, Edward L. Giovannucci, MD, ScD, Yin Cao, MPH, ScD1,4*, Xuehong Zhang, MD, ScD5*
      1Division of Public Health Sciences, Department of Surgery, Washington University School of Medicine, St. Louis, MO, USA
      2Department of Colorectal Surgery, the Sixth Affiliated Hospital, Sun Yat-sen University, Guangzhou, P.R. China
      3Guangdong Provincial Key Laboratory of Colorectal and Pelvic Floor Diseases, the Sixth Affiliated Hospital, Sun Yat-sen University, Guangzhou, P.R. China
      4Siteman Cancer Center, Washington University School of Medicine, St Louis, MO, USA
      5Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
      Background & Aims: Yogurt may reduce risk of colorectal cancer (CRC) through maintaining gut homeostasis. However, the association between yogurt intake and risk of precursors of CRC, including both conventional adenoma and serrated lesion, has not been investigated.
      Methods: We prospectively evaluated the association between yogurt intake and risk of colorectal adenoma (conventional adenomas/serrated lesions) among 31,307 men in the Health Professionals Follow-up Study (HPFS) and 55,756 women in the Nurses’ Health Study (NHS) who underwent at least one lower endoscopy (1986-2010). We used multivariate logistic regressions for clustered data to calculate odds ratios (ORs) and 95% confidence intervals (CIs).
      Results: We documented 5,561 conventional adenomas and serrated lesions in HPFS and 8,116 cases in NHS. Compared with never intake, men consumed ≥2 servings/week of yogurt had a multivariable OR of 0.82 (95% CI: 0.71-0.94, Ptrend=0.01) for conventional adenoma, after adjustment for a list of putative risk factors for CRC including non-yogurt dairy intake and total calcium. This inverse association was not observed for serrated lesion (Ptrend=0.41). Additionally, the inverse associations were more pronounced for high-risk conventional adenoma (OR=0.74, 95% CI: 0.60-0.93, Ptrend=0.01) than that for low-risk adenoma (Ptrend=0.21), and appeared stronger for colon (OR=0.82, 95% CI=0.70-0.95, Ptrend=0.01) than for rectal adenoma (Ptrend=0.94). Such benefit was more evident among men aged ≥65 years, who exercised less, or had lower dietary fiber intake, although the interactions were not significant. No association was observed between yogurt intake and risk of conventional adenoma or serrated lesion among women.
      Conclusions: Higher consumption of yogurt was associated with decreased risk of conventional adenoma in men, especially for high-risk adenoma. Future studies are warranted to confirm these findings in different racial/ethnical groups, investigate possible gender or anatomic subsite differences, and elucidate potential underlying mechanisms.

1. Qi Sun
   Diet, Gut Microbiota, and Circulating Trimethylamine N-oxide
   Jun Li^1,2, Yanping Li¹, Kerry L. Ivey¹, Jeremy Wilkinson³, Andrew T. Chan⁴, Curtis Huttenhower³, Frank B. Hu^1,2,5, Eric B. Rimm¹,Qi Sun^{1
   1}Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA, USA
   ²Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA
   ³Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA
   ⁴Department of Immunology and Infectious Diseases, Harvard T.H. Chan School of Public Health, Boston, MA, USA
   ⁵Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
   Corresponding Authors: Dr. Qi Sun, MD, PhD (Department of Nutrition, Harvard T.H. Chan School of Public Health. 665 Huntington Avenue, Boston, MA 02115. Tel: 617-432-7490; Fax: 617-432-2435; Email: qisun@hsph.harvard.edu)
   Objectives: Trimethylamine N-oxide (TMAO), a metabolite may be produced from dietary choline and L-carnitine intake by gut microbiota, has been associated with higher a risk of atherosclerosis. However, microbial taxa and enzymes associated with TMAO production in free living individuals are less studied.
   Design and method: We performed analyses among 307 healthy men who participated in the Men’s Lifestyle Validation Study. Information on dietary consumptions of choline/L-carnitine containing foods was collected by food frequency questionnaires and plasma biomarkers of choline and L-carnitine, assessed twice, six months apart. Circulating levels of TMAO, blood lipids, C-reactive protein, and HbA1c, were assayed in two repeated plasma samples. Two pairs of fecal samples were collected, also 6 months apart (totaling 925 samples), followed by metagenomic profiling with shotgun genomic sequencing. Taxonomy was determined using MetaPhlAn2 and functional enzymes were identified with HUMAnN2. Association between microbial features and TMAO was examined using a random-effects linear regression in MaAsLin2.
   Results: Higher intakes of L-carnitine, phosphatidylcholine, liver, egg, and fish, were associated with higher circulating TMAO levels, which were associated with an adverse lipid profiles and higher HbA1c levels, especially among overweight men. The relative abundance of one Bacteroidetes species (Alistipes shahii) and 6 Firmicutes species (namely Clostridium citroniae, Clostridium clostridioforme, Clostridium nexile, Clostridiales bacterium, Eubacterium hallii, Roseburia hominis) was significantly associated with circulating TMAO levels (FDR<0.05). The presence/abundance of these species, together with some specific enzymes (e.g., transaldolase, histidine kinase, glycerophosphodiester phosphodiesterase, nadh dehydrogenase) and genetic pathways (e.g., adenosine ribonucleotides de novo biosynthesis, 5-aminoimidazole ribonucleotide biosynthesis, coenzyme a biosynthesis II) inside these species, were associated with higher production of TMAO from different dietary choline/L-carnitine precursors.Conclusions: In a group of healthy men, the production of the atherogenic TMAO from dietary intakes of choline/L-carnitine-containing foods, may partially depend on the abundance of specific enzymes within certain microbial taxa, especially Bacteroidetes and Firmicutes bacteria such as Alistipes shahii, Clostridium, and Eubacterium hallii.

We gratefully acknowledge the support of the National Institutes of Health NIDDK grant R24DK110499 for this inaugural meeting through the Human Microbiome Bioactives Resource.